You are on page 1of 356

CONTENTS

THE

NA

T U RE o r INFEBE NCE
1
Ps ych ological and L ogi cal Elem ents in Inference p age 1 ;
Obj e ctive and S ubj ective Ne cess i ty 4 ; Data of Presen
tation 5 ; S y s tem as Gro und of Inference 6 ; Th e Im
Cnar I
.

li i

di

d ho h
l

Exp c t, 11 ; In erence me ate


t r ug
pl icit an
th e Un versal , 12 ; C nce ptual Pr cess es, 13 ; Exp ana
t n, 14

io

CRA P II
.

IND UC T IO N

A ND

D E DUC TIO N

16

Various Op inions

concerning th ei r R el ati ve Imp ortance


p age 16 ; R egarde d as D ifferent Phases of One and
th e S ame Proce ss 17 ; Th ei r R el at ion to th e G roun d
of Inference regarde d as a System 17 Th eir Relation
to th e Universal 18 ; Dierenc e b etween T ruth and
Fact 19 ; Mutual D ep endence of In duct ion and De
duction 20
,

CHA P

Tu n ES SE N

III

The Induct ive Hazard ,

TI AL S

or

INDU C TIO N

l o

24

of

d
d io
tion 25 ; Its
regard ed as an Inverse Process 27 Law and Rule 30 ;
Law in Terms of an Hyp oth etical Univers al 3 1 Induc
Th e Se ien
tion in th e Conduct of Human A ffairs 3 2
tic S p ir it 33

24 ; Bas a P stu ate


In uc
Episte melo gical Nature , 26 ; In uct n

page

CRAP IV
.

Th e

Tr1 Es
>

or

IND UC

M ethod of Enumeration
on

ti

36

Incompl e te

IV E

INF E RE NCE

pag e 3 5 :
Enumerat

ion

34

Perfect Induc
37 ;

Pr ba

vi

CON ENT

M ethod
of Scientic A naly sis or Causal D eterm ination 40 ;
Th e Causal Postulate und erlyi ng A ll th e M e thod s 43
R elation of Mental Habit to Choice of M ethod 47 ;
G eneralization 48
b ility , 38 ; Th e

Method of Analogy

39 ; Th e

'

AP V CA U SA TIO N
60
L og ical S ignicance of th e Causal Conc pt page 50 ; Its
Ph eno menal S ig n ican ce Th e Cons ervat ion of E m
ergy 5 1
Its Philo sop hi c al S ignicance 53 Its L og ical
S ignicanc e 54 ; It Epistemelog ic ai G round 58 P op
ular and Sci entic Idea of Cause 58 ; Causal A nalysis
60 L i mitations of K nowl ed ge 6 2
CH

A U S AL A N ALY SIS N D D TE RM IN A T O N
64
S equence page 64 ; Concurrence 66 ; Co existence 66 ;
V ital Growth and D evelopment 68 ; Collocation 6 8 ;
D ifferent Modes of T ransfer of Energy 7 1 ; Quantita
tive D etermination 7 2 ; O b s ervat ion and Ex p eri m nt
73 ; Negative D etermination 7 8 ; Pseudo causal Co n

CR

AP

VI C

nection ,

AP

CH

82

V II
.

MI LL s INDU C T I VE

ME T HO D S THE M ETHO D

A G RE E M E N T
84
The F ive M ethods page 84 ; Th e M e thod of A greement
8 6 ; S y m b olic R epresentat ion 8 7 ; V ariation of In
stances 9 0 ; Th e M ethod of A gre em ent and O bserva
tion 9 1 ; R el ation t o S imp l e Enum eration
9 1 ; Se
qu ence an d Co exi stence 92 ; D e fects of thi s M ethod
9 3 ; Its Chief Function th at of S ugg estion 9 6 ; Illus
OF

trations, 9 7

AP V III TR M THO D O F D IFFE R N CE


101
R elation to M ethod of A greement page 10 1 ; Its Charac
teris tics
Symb olic Representation 103 S imila
CR

CONTENT S

a ive

to Neg t
o ry o f C

Determination, 104 ;

R elation

ii

to the Th e

omb inations 105 ; Cri ticisms of this Method


106 ; Practical D ifficul ti es 109 ; Illustrations
113 ;
Blind Exp erim ents 115

AP

CH

I"

TRE

JO IN

T ME THO D

OF

A G REEMEN T

D I FFE REN C E

A ND

117

R elation to M ethod of

page

Dieren ce,

S y m bolic Rep

117

D ifcul ty of Elimination 12 1 ; Illus


tratio ns
Advantage of thi s M ethod over th e
S imple Me thod of A greement 128
resen tat io n ,

118 ;
124

CH

AP X
.

THE

M E THO D O F CO N CO M I T AN T V A RI A TIO N S

130

h aracteristics page 130 ; Its Sy mb olic R epresentation


13 1 Quant itative Det e rm i nation 13 1 G raphic R ep t e
scu tatio n 133 ; Psycholog ical Impressions 133 ; Illus
tratio ns 134 ; Th e Comp reh ension of th e Intens ity of
Unknown Forces fac ilitated b y this M ethod 141 Lim
itations of thi s M e thod 142

Its C

AP

CH

M E T HO D

" L TRE

OF

R E S ID U E S

146

A Method of Elimination page 146 ; S ymb olic Representa


tion 146
A D educ tiv e M ethod 14 7
Th e Comp l ex i ty
of th e R es idual E l ement 148 ; Illustrations 149 ; Re
s id u al Error in Exp erim ents
153 ; Th e M ental Hab i t
of insp ecting A ll Remaind ers 154
,

'

CR

AP

" II
.

VE R1FI C A T IO N

A ND

PR E D I C

T ON

156

deduc tive Method page 156 Verication 157


P rediction 159 ; Ill ustrations 160 ; Bacon s A nti c ipa
t ions of Nature 163
S c ientic Though t 164 Indi rect
M ethod of Prediction 166 Exception Ph eno mena 170 ;
Generalizat ion 17 1 M ath emati cal M ethod

Th e Ind u cto-

viii

CON EN

TS

17 4
AP X III HYPO THE S S
Hyp othesis as Preli minary t o Exp eri ment page 174 ; Hy
po th es is in p l ace of Exp eri m ent 176 ; Illustrations
Function of th e Imagi nat ion in Hy poth esi s 184 ;
17 7
A nalysis and Synth es is 186 R e quirem ents of a L egit
Postul ate and Hy p othes is 189 ;
imate Hy p oth es is 18 7
F ictions 196 ; S ugges tions through F ailure of Hy poth
ese s 19 7
Cons ili enc e of Inductions 198 ; Experimen
W h ewell and Mill 20 1
tum Cruc is 199

CR

CR

AP

" IV

A NALO G Y

204

A nalogy as Suggestive of Inductive Inquiry page 204 ;


Analogy in G eneralization 204 Formation of Co n
Class i
c epts an d Analog y 2 0 5 ; Natural K inds 205
cation 207 T eleology 208 ; Fal se A nalogies 220
,

CR

AP

"V P
.

RO BA BIL I TY

226

l exity of th e Causal nexus page 226 ; Relation to


Enum erative Induc tion 2 28 Calcul ation of th e Prob a
b ility of a Particular Event 230 ; Adverb ial Pro b abil
ity 2 3 2 ; E sti mate of A ggregates 234 ; Ch ance and
Coincidence 243 ; C ircumstantial Evidence 247 R el a
tion to th e M ethod of R es idue s 2 5 1

C mp

CR

AP

" VI

E M PI R I C AL

L AW S

252

Three Classes of Laws of Varying D egree of Probab ility


page 252 ; E mpiri cal Law as Expression of Causal
R el ation in Process of D eterm ination 2 53 ; Colloca
t ions g iving R i se to Emp i rical Laws 25 4 ; G enerali za
tions e xpres s ing an A ggregate of Quali ti es in th e S am e
Individual 2 56 Probab ility and E mp i ri cal L aws 258 ;
Th e M e thod of A greem ent 2 59
Th e E m p iri cal Nature
of th e Causal R elation 2 60
,

C ON EN

CRAP " VII


.

TS

ix

FA LLA CIE S

262

Failure to compreh end the


Entire F ield of V i s ion 263 ; b Failure to concentrate
A ttention 26 5 ; 0 Errors d ue to A pp erce ptive Proj ec
tion 266 ; Of Judg ment 266
a Fals e Asso ciat ions
267 ; b Emot ional Pe rturb at ion 267 c G eneral F rail
tie s of Human N ature Bac on s Idols 269 ; O f Imag i
nation 2 7 1 0 f th e Conceptual Proces s es 27 5 a Hasty
General i ation 2 7 6 ; b Interp olation in a Seri es 2 77 ;
c Provinc iali sms 2 78 ; d F als e A nalog i es 278 ; c In
correct Clas sication 2 79 ; Psy chological Character of
th ese F all acies 2 7 9

Of Perception, page 26 3

h E IN D U C TIV E ME TRO DS
VA R I O U S S C IE N C E S

CRAP " VIII


.

TRE

As

APPLIED

To

2 81

of M ethod will vary w ith Nature of the Ph enomena


page 281 Complication of th e D octrine of the Conser
Th e Ph enom ena of O ne S ci ence
v at ion of Energ y 2 8 7
to be interp rete d in th e Lig h t of th e R e sults of A noth er
S cience 2 90 G rowing T endency to supp lement Ded uc
tive Method by Inductive 292

Nature

CR A P " I"
.

HI S TO RIC AL

S TCH
KE

OF

IND U C

TI O N

S o crates , 297 ; Plato , 297 ; Ar is totl e , 298 ;

29 7

R oger Bacon
300 ; L eonardo da V inc i 3 0 1 ; Teles ius 3 02 ; Camp a
nella, 303 ; Caasalpinus C opernicus G ilbert K epler
,

Brah, Galil eo , 304 ; Fran ci s Bac on, 3 04 ; L ocke , 3 07 ;


Newton, 3 07 ; He rs ch el , 308 W h ewell 3 10 ; Mill, 3 11
,

CRAP " "


.

Lo o I C A L

E" E

RC I S E S

313

PREFACE
IT

h as been my aim, in th e follo wing pa ges , t o


presen t th e essent ial features o f ind uct ive lo g ic,
in th e h ope th at th is w ork may prove a tt in g
supplement t o th e element ary courses in formal or
deduc tive lo g ic T h e impression is to o oft en left
in th e m in d s of th ose wh o h ave pursue d th e
s t udy o f d e duct ive lo gic exclusively th at th e fo r
m al law s of th e syll o g ism c o ns t i t ut e th e ent ire
bo dy o f lo g ical doct rine, an d th at reasonin g con
sis t s solely in d raw in g c onclusions fr o m g iven
premises
T h ere is d an g er h ere lest reasoning
become associat e d w ith an art i cial proce du re
th at seems to n d its proper sph ere in the solu
tion of verbal qui bbles an d lo g ical puzzles In
th e act ual experiences o f life, w e d o no t nd
o u r p r em ises rea dy ma d e
Th ey are th e resul t
o f w i d e O bservat ion an d pat ien t inves t i g at ion an d
e xperiment
We c h allen g e premises th at are
g iven, and w ei gh th eir si gnicance
We meet
par t icul ar fac t s before w e d o th e g enera" Iaws
.

xi

R FACE

xii

P E

Th e former mus t be t est e d an d int erpret ed before


w e can rise t o th e g eneral law s wh ic h un derlie
th em, an d wh ic h may s t an d as th e maj or prem
Th us w ith in th e very
ises o f ou r syllo g isms
sph ere of d e duction it self th ere nat urally O pens
a w i de el d fo r in duct ive inquiry
T h erefo re I
h ave emph asize d th e necessity o f a th orough
kno wle dg e o f th e principl e s o f in duct ive lo g ic
in order to c o mpreh en d th e mat erial as w ell as
th e fo rmal elem ent s in infere nce, an d with out
wh ich n o rm g rasp o f th e g eneral process of
reasoning is possible I h ave also insis t e d upon
re g ard in g in duct ion an d d e duc t ion as mutually
d epen d ent ; no t as s e parat e mo d es o f inference,
but rathe r as different ph ases o f o ne an d th e
sam e lo g ical proce dure
I h ave en d eav o re d als o , t o in d icat e in s o me
m easure at l e as t, th e salient ch arac t eris t ics o f
th e mo d ern lo g ic e specially as present e d in th e
w orks o f L otze, S igwart J ev o ns Gre en Bosan
quet, an d V enn
In th e illus t rat i o ns o f th e
various in duct ive meth o d s I h ave sou ght fres h
mat erial as far as possible, w ith th e vie w o f
represent in g th e act ual m o d es of reas onin g an d
meth o d s of inves ti gat i o n e mpl oy e d by th ose wh o
h ave beco m e em inent in th eir s e veral sph e res o f
research, such as F arad ay, Tyn dall, Darwin, an d
,

R FACE

xiii

P E

L ubbock ;

and

especially th e different meth o d s


wh ic h h ave le d t o import ant discoveries in th e
various sciences
T his applies no t only t o th e
illus t rat ions in th e t ext proper, but also t o th ose
wh ic h I h ave collect e d in Ch apt er " " un d er th e
h ea d o f L o g ical Exercises
It seems t o me,
moreover, th at in asmuc h as th e principles o f
in duc t ive inves ti g at ion are in suc h accor d w ith
th e s cienti c S pirit o f our a g e, th eir im port ance
as a lo g ical d iscipline canno t be too h i gh ly value d
.

J G H
.

RIN CE T O N

NJ
.

March

2, 1896

INDUCTIVE LO GIC
CHAP TER I
TRE N A T U RE

OF

IN FE REN CE

is a par t icul ar mo d e o f inference in


g eneral ; an d th erefore before it s nature and sc o pe
c an be a d e q ua t ely d e ne d , it w ill be n e ce s
sary t o
g ive some accoun t o f th e th eory o f infer ence an d
it s precise lo g ical si gni ca tion M oreover, it is n o t
possible t o appreciat e th e dis tinc t ion be tw een th e
processes o f in duction an d d e duct ion, un til w e h ave
rs t examine d th e charac t eris t ic featu r es wh ich
are com m on t o th e two, an d wh ich cons t itut e th e
essen t ial element s of inf erence it s elf Th e n atur e
We
o f infe re nc e m a
y be un fol d e d in two w ays
m ay consi de r w
hat i t is in i t s outwar d aspect
that is, through its ph enomenal manif e st ation in
what it eects ; o r it m ay be m o re s t ric t ly d ened
in t er m s o f it s w arrant o rgroun d F rom the rs t
p o int of vie w w e examine inf erence as re g ar d s its
psych olo gical S i gnicance ; th at is, wh at is inf er
ence consi dere d as a psyc h ical experience, it s n a
ture an d c h ar ac t eris tics
But we mus t consi der
als o th e secon d ques t ion wh ethe r th ere is any
IN D UCT I O N

'

'

TV

O GIC

IN UC I E L

necessity limit in g an d de t erminin g th e subj e ct ive


e xperienc e wh ic h pr e sent s th e ch arac t e r o f a la w
h avin g un iversal vali dity Wh at g oes on in th e
A lso , wh at
m in d d urin g th e process of inference ?
is th e rat ionale of suc h a process ? Th ese ques
t ions w e w ill examine more closely, in o r der t o
S h o w th e nat ure o f inference un d e r th e tw o aspec
th e o ne psych olo g ical , an d the oth er lo g ical
It is a w ell reco gnize d fact in psych olo gy , th at
in o ur S imples t as w ell as th e m o re complex per
ceptio ns, th e int erpre t at ion o f th e d at a o f presen ta
t ion al ways g oes bey o n d th e st ric t cont ent o f th e
d at a th emselve s We s ee more th an is g iven in
The
th e el d of vision im m e d iat ely before us
min d supplies h ere an d th ere th e necessary part s
th at are lackin g in th e actual e l em e nt s o f present a
t i on an d yet wh ich are n e cessit at e d by th e kno w n
nat ure of that wh ich is act ually g iven We fo rm
o u r j u dg ment o f di st ance in d ir e c t ly , an d no t throu gh
dire ct present at i o n S o, als o, o u r i d ea of a th ir d
dim ension is acquire d by a pr o cess marvellously
com plex in wh ich th e d at a both in d icat e an d yet
are t ranscen de d by th e re sul t s Wh eth er th e nat iv
is t o r empiricis t h ol d s th e t rue posit ion concern
ing o ri gi nal psych ical experi e nce, i t s t ill mus t be
conc e de d acc o r din g t o eithe r th eory th at th e d evel
o pm ent o f o ur percep t ions c o rrespon d s t o a la w of
gr o wth bas e d up o n accum ul at e d inferences Infer i
enc e h as been de ne d as th e indi rect referenc e of
a c o nte nt to realit y an d as such, w e see th e be
g inn in g s of infe rence in th e m o s t simple o f our
Every percep t ion cont ains a d irec t
per cept ions
,

THE N

AT URE OF

F R

IN E EN CE

re ference t o reality , but als o som e thin g wh ic h in


a g reat er or less de g ree is referre d in directly t o
reality T h e fac t th at o ur kn o wle dg e as g iven in
th e com plet e percept i o n cont ains m o re th an is act u
al ly me diat e d th rou gh th e avenues o f th e s enses
,
is due t o th e app e rcep tive processes o f conscious
ness M in d is ac tive in percept i on an d n ot a m ere
passiv e recep t acle T h at wh ich is g iven, the raw
mat e rial o f th e senses is elaborat e d an d ext en d e d
as it is c o mbin e d w ith th e w ealth o f repres ent at ive
an d concep tual mat erial whi c h the min d brin g s t o
every ne w p e rc e pt ion To th is ext ent at leas t th e
min d possesses a creat ive func ti o n
A
app e arance of sky c o mbine d w ith peculiar
t ions o f m in d an d t em perat ure l e ad s o n e t o as
w ith some d e g ree o f cert itu de, th at it w ill
bef o re mornin g Th e pre dict ion is an infere nce
base d up o n an d g r owing ou t o f th e ac tual d at a
o f percep t i o n an d ye t far o u t runnin g th e m
We
r e co gniz e a frien d from h is st ep o r voic e T h e
mere pre sent ati o n is o nl y a s o un d T h at it is
as socia t e d w i th a pers o n an d n o t an anim al o r a
thin g, is an inference ; th at th is is th e part icular
person wh om we rec o gnize as a frien d an d can call
by name even befo re w e t urn ar o un d t o conrm
th e opinion by d irect t es timony o f vision th is is
An d even wh en we o pen
a s t ill fur th e r i nfe rence
w e ll up m any
o u r eyes in s imple visi o n i t self
a g ap in o u r m in d s, an d g ive depth an d dis t ance,
an d int erpret th e contr ast s o f li ght an d s h a de an d
th e play o f colors , th rou gh th e pr o cess o f inference,
al th ou gh w e may no t be aw are of th e pr o cess it self,
.

TV

O GIC

INDUC I E L

wh ich is auto m at ically operat ive th rou gh lon g


co n
t inn e d h abit Wh en w e th us reg ar d infer ence as
a psyc h ol o gical ph e nomenon, it may be rea dily
plaine d by th e law s of comparison, association,
reco gniti o n g eneralization et c A n d as such, in
fe rence h as a subj ective force at least, an d lead s to
th e h abit o f pred ict i o n an d e xpec t at ion T h e w ill ,
inuenc e d by th e resul t in g belief lead s t o activ i
ti e s c onsis t ent w ith such expect at ion
Her e, h o w ever th e question arises, wh ich is
ur ge d w ith such force by Hum e , Is th ere obj ec t ive
vali dity as w ell as subj ect ive necessity
T h is
lead s t o a consi de rat ion of infe rence, fro m th e sec
o nd p o int o f vie w above ment ione d
We may be
const raine d to belie ve cert ain th ing s c o ncernin g th e
great w orl d lying beyon d th e sph er e of imm e diat e
c onsci ousness ; but wh at w arrant h ave w e in so
d oin g o r wh at assurance th at our conclusions are
May w e not be d eceive d, aft er all , an d by
c o rrect
s om e psych olo g ical t rick be le d t o re g ar d th e p h e
n o men a o f c o nsci o usness as quit e o th er w ise th an
th at wh ich obt ains in reality ? We m ay h ave a
s t ron g aversion t o sitt in g do w n at a t able wh ere
th e num be r o f p e rsons w ill be th irt een But h as
th e subj ec t iv e convic t ion th at o n e o f th e th irt een
w ill die in th e course of th e year, any value wh en
w e c o me to re fer it t o reality an d ask ourselve s th e
nat ure of th e g roun d upon wh ic h th e convic t io n is
base d ?
O n th e oth er h an d, h o w ever i t is qui t e a d iffer
ent kin d o f necessity wh ic h cons t rain s us t o j u dg e
th at if a p e rs o n j um ps o ff of th e ro of of a house,
.

THE N

AT URE

F R

O F IN E ENCE

h e mus t surely fall t o th e g roun d belo w


S ome
g rossly supers titious an d i g norant people may be
lieve th e former w ith as obs t inat e a convic tion as
th e latt er, so th at a purely psych olo g ical crit erion
o f th e s t ren gth o f c o nvict i o n is no t at all a de quat e
o r sa t isfac t o ry
I s th ere any o th er crit e rion
In
wh at ins t ances d oes th is subj ect ive c o ns t raint pr o
c e e d f o m th e necessi t ies o f reali t y
o r in o th er
w or d s , in wh at cases are w e able to discover a lo g i
cally g roun d e d w arran t wh ic h c o mpels th e infer
enc e in d is t inct ion fro m th e mere psych olo g ical
co m pulsi o n wh ich is occasione d by th e psyc h ical
t end encies o f ass o ciat i o n an d g eneralizat i o n 9
T h is lea d s us t o c onsi d er th e lo g ical , in d is t inction
from th e psyc h o l o g ical nat ure o f inference Inas
m uch as th e ch arac t eris t ic feat ure o f infe re nce c o n
sis t s in th is th at wh ile d ep e n din g upon cert ain d at a
o f presen t at ion, it n e ver th el e ss wh olly t ransc e n d s
th em th e que s ti on n at urally su gg e s t s it s e lf, wh eth e r
i t is some th in g w i thin th e d at a th em selv e s , or wi th
o u t, by vir t ue o f wh ic h th e min d th us g oes b e yo n d
th e m in th e proces s o f inference If it lies wh olly
w ith ou t th e d at a it m us t be s o m e th in g im p o se d
up o n th e m by th e m in d an d as such can h ave only
a psych ol o g ical force an d value Fo r ins t ance, th e
belief th at if th irt een sit do wn t o g eth e r at a t able
o n e w ill d ie in th e course of th e year, can h ave only
a subj ect ive value an d si g ni cance T his is t rue in
all cases wh ere th e nec e ssity o f convic t ion nd s i t s
o ri gin in preju d ice or in supers t it ion o r it may be
in th e force o f au th o rity . In all suc h ins t ances w e
feel th e lack o f a satisfac tory lo g ical groun d How
.

TV

O GIC

IN UC I E L

ever, o n th e oth er h an d if th e d at a o f c o nscious n ess


cont ain w ith in th em s e lve s th at wh ich enables us t o
t ranscen d th e m at th e sam e tim e th at w e int erpre t
th em th ere is ext e rnal vali dity for our infe re nc e
th at h as a l o g ical w orth T h is s eem s at th e rs t
glance t o be a para dox H o w can any c ont ent
enabl e us t o s t at e c onc e rnin g it m o re th an is co n
Th e ans w er t o th e se em in g par a
tain ed w ith in it
d o x is th at e ve ry c o nce p t an d ev e ry percep t i o n as
w ell h ave b oth an explicit an d im plicit cont en t W e
never att ain c om pl et e vision o r perfe ct appreh e nsion
T h ere are m ore o ver m any p o int s of vie w eac h
g i v in g addit i o nal kno wl e dge c o ncernin g any p h e
n o m en o n present in consciousness
We see th ere
fo r e only in part an d yet th at wh ich is seen c o nt ains
c e rt ain n e ce ssary implicat ions concernin g th at wh ich
is not seen In th e pro g ress o f kn o wl e dge ,
quent o bservat i ons diffe rent p o int s of vie w
c onrming an d am plifyin g our inferences ,
us to p e rc e iv e imm e diately wh at form erly
Th e process by whi c h
inferre d
bec o m in g e xplicit in dicat es a necessary
e xis t i n g b e tw e e n th at wh ic h is kn o w n In
an d th at wh ich is kn o wn imm e d iate ly Mo re over,
c o nsci o usn e ss has been repres ent e d as a s t ream or an
int ricate ly inte r woven web
somethin g extremely
c o mpl e x Every part is relat e d b oth pr o ximat ely
an d rem ot ely T he re is n o such th in g as an is o lat e d
p e rcep tion ; eve ry p e rce ption h as it s c o m pl e x re l a
t i ons an d c o nne ct i o ns S O also e very c o nc e p t wh ic h
is f o rm e d by g eneralizat i o n th rou gh co m parison an d
abs tr act ion, o f o ur pre sent at ions as int erpret e d by
,

THE N

AT URE OF

F R

IN E ENCE

us, possesses this ch aract erist ic of g r eat er o r less


complexi t y In th is manner the w orl d of conscious
ness is c o ns t ruc t e d ; th at is th e w orl d as i t is for us
T h is f o rms a c o mplex wh ol e m a de up of part s wh ich
in the ms e lves m ay be re gar d e d as wh o les an d yet
whi ch may be s t ill f urth er divi d e d an d sub divi de d
S uch an int errelat e d wh ole w e may s tyle a sys
tem o r, in o th er w o r d s a c o m plex wh ole wh o s e par t s
are con g ruently arran g e d T h e i de a o f sys t e m nd s

e xpression in th e
L aw o f T otality, th at o u r
kno wle dg e is capable o f arran g ement in a self c o n
sist e nt an d h armoni ous sys t em an d wh ic h m o re o v e r
in it s c o nte nt an d fo rm faith fully represent s o bj ec
1
ti ve r e ality
W e nd th erefore th at in th e f o cus
of consci o usness at any o n e t im e, wh eth er in th e
S p h er e o f presen t ati o n o r in th e r e g ion o f re pres e nt a
t ive o r th e c o ncep t ual pr o ce sses wh at e ve r is g iv en
carries with it al ways cert ain i m plicat i o ns an d th e re
fo re cert ain necessary relat i o ns T h is is specially
e mp h asize d in Bo s an qu et s d e nit i o n of sys t em :

S ys t em is a g roup o f relat i ons o r prop e rti e s o r


th in g s , so h e l d t o g eth e r by a c o mm o n nat ure th at
you can j udg e fr o m som e o f th em wh a t th e oth e rs
mus t be 2 Two fact s re g ard e d as in d epen d ent an d
c onsi de re d separat ely may give n o info rm ation be
y o n d th eir explicit cont ent s ; but wh e n conj o ined
th ey imply more th an th e su m of th eir part s Ho w
oft en tw o i d eas in separat e m in d s yie l d n o re sul t ;
but brou ght t o g e th er th ey g ive li ght I s o lat ion
.

Ueb erweg , A S ys tem of Log ic

trine ,
2

pp

54 0

an d

His tory of Log ical Do c

Bo san que t , Th e Essen tials of Log ic ,

140

TV

O G IC

IND UC I E L

ne g at ives inferenc e T o unfol d any d at a in all


th eir m anifol d implicat i ons is th e process o f infe r
ence Its w arran t li e s in th e fun d am en t al p o s tulat e
o f kn o wle dg e wh ic h w e are cons t rai ne d t o assume ;
nam ely th at o u r consci ousness m us t be self con
sis t ent th rou gh o ut Wh ate ver is a dmitt e d as t rue
mus t n d a c o ng ruent plac e in th e sys t e m t o wh ich
it is possible t o refe r it T h e necessi ty o f ttin g i t
in it s pr o per place g iv e s rise t o certain i mplicati ons
which n ec e ssit at e c o rresp o n d in g r elat i o ns and attri
but es A n d if it c oul d n ot b e put int o such a place,
w e w o ul d fee l th at w e s h o ul d h ave t o surren d er th e
i dea o f s elf c o nsis tency in th e vari o usly r elat e d ele
m ent s o f our c o nsci o usness Th e very in t e g rity o f
o u r m e nt al lif e n e cessi t at es th is c o nvic t i o n
h erefore a part being g iven w e supply in o u r
d s o th er part s o r th e wh o l e t o wh ic h th e g iv e n
ust nec e ssarily bel o n g T o ach ieve th is w ith
l o gical w arrant o u r kn o wl e dg e o f th e part m us t be
a d equat e to th e ex t en t th at w e kn o w th at th e ele
m en t un d er consi d erat ion canno t be c o mpl et e in
it self, but mus t be suppl e m ent e d by it s appr o pri
ate ly r elat e d e l em e nt s wh ic h w ith it go t o m ak e up
th e c o mpl et e sys tem
W e infe r th e nat ure o f th e
o we r n o t yet in bu d by th e S pr o ut in g l eaf
Th e
o n e nec e ssi t at es th e o th e r by vir t ue o f th eir com
m o n inh erence in th e sam e plant sys t em
We
kno w th at g s d o n ot com e from th o rns n o r g rapes
fr o m th is tles Colum bus , no tin g th e s e aw ee d, an d
bird s, and th e drift o f th e sea inferr e d a s h ore
bey o n d t o wh ich h e w as c ons t rain e d by th e ne ce s
sit ie s of th o ught t o refer th e m It is sai d o f Cuv ie r
.

NATURE

THE

O F INF ERENCE

th at h e w as able t o recons t ruc t part fo r part th e


ent ire frame an d org anism o f an animal wh ose
fossil t oo th alone forme d th e o ri g inal d atum He
knew th e sys t e m to wh ic h it mus t h ave bel o n g e d
an d to wh ic h it alone coul d possibly be re ferre d
A n int eres t in g qu o t at ion from Cuvier h imself illus
t rat es mos t appro priat ely th is func t i on o f inference
He says in h is Ossemens Foss iles I d oub t if any
o n e w o ul d h ave d ivine d, if un t au ght by observat i o n
th at all rum inant s h ave th e f o o t cleft an d tha t th ey
alone h ave it I d oubt if any o ne w oul d h ave
divine d th at th ere are front al h orns only in th is
cl as s ; th at th ose am on g th em wh ic h h ave sh arp
canines fo r th e m os t part lack h orn s Ho we ve r,
since th ese relat ions are cons t ant th ey mus t h ave
som e sufcient cause but since w e are i gnorant of
i t w e mus t make g oo d th e d efe ct o f th e th eory by
means o f observation : it enables us t o est ablis h
em pirical law s wh ic h be c o me almos t as cert ain
as rat i o nal la w s wh en th ey res t o n sufciently
repeat e d o bservat ions ; s o th at no w wh oso s e es
merely th e prin t o f a cleft fo ot may c o nclu d e th at
th e anim al wh ich left th is impression ruminat e d
an d th is conclusi o n is as cert ain as any o the r in
ph ysics o r m orals
T h is foo tprint alone th en,
yiel d s t o h im who obs erves i t th e form of th e
t eeth , th e f o rm of th e j aw s th e form o f th e ver t e
brae th e form of all th e bo n es o f th e le g s , o f th e
th i gh s , o f th e sh oul d ers an d of th e pelvis o f th e
1
animal wh ic h h as passe d by
In th e common c o n d uc t o f e very d ay life w e infer
.

Q uo te d

by Jev

o ns ,

P rinciples

f S cience, 2d ed

683

D UCTIVE

10

IN

O GIC

beyon d th e im m e diat e pr e s en t experi ence t o fut ure


h appenin g s an d in a sim ilar manner My t rain is
h alf an h our lat e I kno w I mus t miss m y connec
t i o ns at th e s t at ion ah e ad ; for th e t rain I am h o p
in g t o cat ch at th at place is sch e dul e d t o leave v e
m inut es aft e r th e t ime o f arri v al of th e t rain I am
n o w on
T h e t ime relations h ere necessit at e my
m issin g my connect ions
T h is is ren d ere d s till
m o re ce rtain if th ey ar e rival roa d s ; o n n o ac count
will one w ai t for th e oth e r Mo r eover th e t rain I
h ope t o make is mad e up an d leaves th e s t at ion in
ques t ion, an d s o I cannot fall back upon th e favorin g
ch ance th at it also may be d et aine d en rout e, an d
s o enable me aft er all , t o reac h it in t ime
with every a dd it ional kn o wl e dg e of th e sys t em
fo rms th e g r o un d of my i nferen ce, an d th e v
c o nd i tions wh ic h affect it, th e vali dity o f my inf
ence is th ereby increas e d
Infer ence re g ar d e d as
th e analysis o f a syst em o f int erre late d par t s is
illus t rat e d in th e follo win g par ag rap h o f Profess o r
J ames : T h e resul t s o f reasonin g may be h i t up o n
by acci d en t Cat s h ave be en kno wn t o open d o o rs
by pull in g latch es , e tc Bu t n o eat if th e lat ch g ot
o ut o f or d er c oul d open th e d oor a g ain unless som e
,
ne w accide nt at ran dom fumblin g t aught h e r t o ass o
ciat e some n e w t ot al m o vement w ith th e t o t al p h e
n o meno n o f th e clos e d d o or
A r e as o nin g m an,
h o w ever, w o ul d open th e do or by rs t analyzin g th e
He w oul d ascert ain wh at pa rt icular
hin d rance
feature o f th e door w as w ron g Th e lever,
d oes not raise th e latch sufciently fr o m it s slot
cas e of insufficient elevati on
raise d o or bo d ily o n
.

THE N

ATURE OF

F R

11

IN E ENCE

d oor s t ick s at t o p by frict i o n a gains t


linte l
press i t b o dily d o wn " I h ave a s tud en t s
lam p of wh ich th e am e v ibrat es m o s t unpleasant ly
un less th e c o llar wh ic h bears th e c h imney be raise d
ab o ut a sixt e enth o f an inc h I learne d th e re m e dy
aft er m uch t o rm ent by acciden t and no w al w ays
k eep th e collar up w ith a small w e dge But my
proce d ure is a m ere associat i on o f tw o t ot als d is
e ase d o bj e c t an d r em e dy
O n e learn e d in pn e umat
ics c oul d h av e nam e d th e cause o f the d is e ase an d
th en ce inferre d th e re me dy i mm e diat ely 1
Inference th er e f o r e may be r e g ar d e d as a d eep
penet rat in g insi ght
Th e e xplicit is th at wh ic h
lies upon th e surface wh ich th e min d i mm e diat ely
g rasps for i t lies d irect ly in th e f o cus o f consci o us
ness Wh ereas th e i m plicit is ben e ath th e surface
an d is reveale d o nly th rou gh a search in g analysis
T h is d ifference m ay be ex h ibit e d th r o ugh th e dis
t inct i o n be tw een th e act ual an d th e pot e nt ial A
ch il d re g ard s gunpo wder m e rely as a pil e o f coarse
g raine d san d Th e m an s ee s wh at the ch il d sees ,
but also th e e xist in g pos sibilities un d er c e rt ain con
He appreh e n d s th e
d itio n s o f explosive forc e
p ot ent ial as w ell as th e ac t ual ; an d h is inferenc e
as t o th e possibl e r e sul t s is bas e d upon h is superi o r
It is th e r efore th e w ell furnis h e d m in d
insi ght
wh ich sees th in g s as m o s t wi de ly r elat e d, an d dis
c e rns th e po t e ntial as we ll as th e actual m anifes ta
t ion wh ich w ill prove th e m o s t fert ile in accurate
inference in pr o ph et ic sugge s tion an d in invent ive
resource

h in g es " O r

,
.

James , P s ych o lo g y , V o l II
.

pp

339 , 340

12

TV

O GIC

IN UC I E L

Th e wh ole w orl d o f reality as well as th at of


kn o wl e d g e m ay be consi d ere d as o ne sys t em em
brac in g w ith in th e unity o f it s t o t ality all th e vari
ous sys te m s w ith th eir c o mplicat e d p art s F ro m
th is point o f vie w e verythin g bears relations t o
every th in g else in th e universe T h e ori gi nal si g
nicatio n of th e t erm u niverse is th us e m p h asize d
T his th o ught no d oubt, Tennyson h a d in min d in
th e follo w in g verse :
,

Flower in th e crannie d wall


I p l uck you o u t of th e crannie s
I hold y o u h e re root an d all in my h and
L i ttl e o we r bu t if I could und erstand
W h at y o u a e root and all and all in all
I S hould know wh at Go d and man is
,

We can in this connect i o n bes t exh ibit th e pre


cis e nature an d function o f th e universal in infer
e nc e
Th e possibili t y o f unf o l d in g th e pr o p e rt ies
o r relat ions o f any th in g in all it s im plicat i ons
d epen ds up o n o ur kn o wle dg e o f th e universal co n
cept to wh ic h th e prope rt ies o r relat ions in ques
t ion are naturally referre d
Wh il e a sin g ular
pr o p o sit ion is th e st at e m en t o f th e m ere occur
r ence o f a ph en o m en on th e un iversal al w ays
implie s a kn o wl e dge o f th e con d it i o ns an d re l a
l
f
e
m
t ions o th e ph no en on
In si ght is o nly po s
sibl e wh ere th e re is a w eal th o f unive rsal concep t s
We see an animal wh ich w e o bs e rve t o be clov e n
fo ot e d We infer th at it also ch e w s it s en d W e
d o n o t obser ve thi s
T h e assert ion does not aris e
,

S ee

Green

Philosophi cal

Wo rks

Vo l II
.

pp

284 , 285

THE N

AT URE O F

IN

FERENCE

di rec t ly

13

from observe d reality but indi rectly


th rou gh th e g eneric concept th at h as g raspe d t o
g e th er th e tw o att ribut es o f ch e win g th e end an d
cloven feet as al ways and necessarily coexis tin g
in one an d th e same animal I nfe re nc e, in th is
sense, m ay be re g ar d e d as th e in direc t reference
o f kno wle dg e t o reali ty , an d th is is al w ays me d i
at e d th rou gh th e uni v ersal
T h e universal h as
th is c h arac t eris tic feat ure , th at i t preserves an
i d ent ity in th e mi d s t o f manif o l d die re nc es T h e
sam e th o ught may be expre sse d by sayin g th at th e
un iversal manifes t s a uni ty in th e mi d s t of d iv e r
H o wever w i d ely d ifferent in many respec t s,
sity
th e animals may appear th at ch e w th e cu d as
th e co w, d eer, s h eep et c
th ere is al w ays th e
con s t ant c h arac t erist ic th at th ey are cloven foot ed
S uch a point o f i d ent ity furnis h es th e cons tant
fac t o r wh ic h d et ermines th e nat ure an d th e vali dity
Were it n ot for th is concep tual
o f th e inference
p o we r o f th e min d th is ability t o g ras p ph enom
an d c o nsi de r th em
e n a in th eir universal essenc e
as int e rrela t e d an d connec t e d w e coul d ne v e r pass
beyon d in divi dual an d part icul ar experience s wh ic h
w o ul d form a series o f wh olly d isconnec t e d ev ent s
K n o wle dg e co ul d n o t th en form a self c o nsis t e nt
sys t e m or infe rence possess any h i gh er w o rth th an
a h aph azar d guess
A S Green says , A mere
fact, a fact apart fro m relat ions wh ich are not s en
sible, w o ul d be no fac t, w o ul d h ave no nature w oul d
n o t adm i t of any th in g bein g kno w n or sai d about
,

it l
.

Green

P hi lo sop h ic al

Works

Vo l II
.

301

14

IN

DUCTIVE

O GIC

Mo reover, infer enc e is no t m ere ly employe d t o


e xt en d th e el d o f c o nsciousne ss in unf o l d in g sup
plem en tary ele m e n t s lyin g bey o n d th e sph e r e o f
dire c t co g niti o n ; th e e l e m en ts m ay all be g iven
imm e diat ely, an d inference em ploy e d t o d iscover
th eir connect i o n an d int errelations, by virtue of
wh at b o n d th ey bel o ng in one or th e same sys
t em Infe rence h ere func t ions as e xplanation A
man is found d ea d ; ther e are many w o un d s up o n
h is pers o n, evi denc e s o f a s t ru ggl e in an out o fth e
w ay plac e up o n a l o nely road S uc h a c o m bin a
ti o n o f fac t s calls for an explanation wh ic h s h all be
consis t ent with the m Th e fact s mus t all be cor
relat e d in a sys t em wh o se relat e d fac t s an d th e
unity of th e wh o le w ill co m plet ely sat isfy th e
min d Th e min d is sat is e d o nly wh e n all h an g
t o g eth er in wh at s e em s th e o nly possible s e lf
c o nsis t ent c o o r dinat e d sys t e m Th e fact s bein g
g iven, th ey m us t be rea d back war d t o th eir ori gin
Th e o th e r asp e c t of infe r e nce is th e rea d in g o f
fact s for war d s , o r unfo l din g th e m in th e ir n ec e s
sary c ons e quenc e s I nference is th e re ply t o th e
natural ques t ions o f th e m in d, wh ence an d
whith e r
A n d th e process is e ss e nt ially th e sam e,
wh ethe r it s pecul iar m o d e c o nsis t s in th e ev o lu
t i on o r th e inv olut i o n o f th at wh ic h is gi ven in
c onsci ousness
M o re o ver, th e m e re psych ol o g ical inference, th e
subj e ct ive e xt ensi o n o f th e d ata o f c o nsciousne ss
w ith o ut any o bj ective g r o un d o r w arrant, S h oul d
e ver be correc t e d , or e v e n at ti m es wh o lly se t asi d e
by m e ans o f th e t ruly l o g ical inferenc e Wh ere
.

"

THE N

ATURE OF

F R

IN E ENCE

15

th e psych olo g ical experience, in t ranscen din g simple


present at ion, procee d s upon s t rict ly lo g ical g r o un d s,
an d h as obj ec t ive vali dity as w ell as subj ect ive
necessity , w e posses s a w arrant of th e h igh es t pos
sible w orth
.

CHAPT ER II

IN DU C T I O N

AN D

DE D U CT IO N

T HERE h ave been d iver g ent t en d encies in th e


h is tory of l o g ic to m ake eithe r de duc tion or in
duc t ion alone th e wh ol e o f l o g ical pr o c e d ur e in th e
process o f infe rence T h e fac t th at th e A ris t ot e lian
lo g ic whic h is ess e nt ially d e duct ive h as been f o r
cent uries exclusiv ely ass o ciat e d w ith lo g ic as a
wh o l e h as l e ft th e im pressi o n up o n m any m in d s
th at it is th e be g inning an d e n d of th e l o gical en
cycl o p aedi a O n th e o th er h an d, J S Mill an d his
f o llo w ers h ave at tem pt e d t o analyze th e syllo gism
to pr o ve it s e ss e nt ially in d uct ive c h arac t er ; an d
th ey h ave maint aine d th at all re asonin g is ind uc
t ive T h is is th e posit ion in the main o f Bac on
L ocke an d Rain L ocke for ins t ance insis t s th at
th e syll o gis m is o f l e ss value th an ext e rnal an d
int e rnal e xp e rienc e induct i o n, an d co mm on sense l
S o als o in a sim ilar vein S c hl e i e rm ach er says :

T h e syll o g is t ic proce dure is o f n o value fo r th e


re al c onst ruction o f ju dgm e nt s f o r th e substi tute d
j udgm ents can only be h i gh e r an d l o w er ; n oth in g
is expre sse d in th e c o nclusi o n but th e re lat ion of
two t er m s to e ac h o th er wh ich h ave a c om m o n
,

Ess ay

on

Hu man Un d ers tand ing Bo o k IV


,

16

DEDUCTION

INDUCTIO N AND

17

membe r, an d ar e not with out, but wi th in, eac h


oth er Ad vance in th inkin g, a ne w co g nit ion, can
n o t o r i g inat e by th e syllo g ism ; i t is merely th e re
ectio n upon th e w ay in wh ich w e h ave att aine d, or
co ul d attain, t o a j u dgm ent, th e conclusion ; no
1
new ins i ght is ever reac h e d
vi e w s thus in dicat e d d o no t necessit at e conictin g
or mut ually exclusive processes It is bett e r to
re g ar d th em, n o t as r ad ically differe nt ty pes of
inference, but r ath er as diffe rent ph ases of o ne an d
th e s ame infer ential p roces s We h ave seen th at
infe rence consis t s in int e rpr et in g th e implicat ions"
of th e sys t em t o wh ic h th e give n in c o nsciousness ;
In th e li ght of thi s d e ni t ion w e can be s t ?
belon g s
indi cat e th e relative func tions of in d uct i o n an d d e
duct ion in th e pr o cess o f inference Wh en th e
system can be consi dere d as a W h ole, an d is appro
h en d e d in i t s ent ire ty , th en it may become th e
grou n d upon wh ic h th e inf erence is bas e d, resultin g
in unf ol di n g th e necessar y natur e or re lat ions o f
any o f th e part s consi d ere d in th emselves , o r in
r eference to th e sys tem as a wh ole Th e p r oce dure
in such a cas e is fr om th e natur e o f th e W h ole
sys t em , t o th e nat ure o f th e several part s , an d th eir
exis t ent r elations, an d th is is d ed uc t ive in it s os
sential feat ures
O n th e o th er h an d, wh en w e kno w th e various
p art s, and procee d fr om th em as dat a to cons truc t
th e sys t em wh ic h th ei r kno w n nature an d rela
t ions necessi tat e, it is in duc t ion, o r p roce dure from
elemen tary part s to th e wh ole th us n ecessit at e d
'

'

S ee Ueberw eg , S ystem
C

f Log ic

etc . ,

345 .

18

IN

DUCT IVE

O GIC

F ro m a kn o wle dge o f th e planet ary sys te m , w e


can infe r th e n ecessary posit i o ns o f sun m oon, an d
earth at any require d t ime as fo r ins t anc e in th e
calcul at ion o f an e clips e T h is is d e d ucti o n But
wh en we be g in w ith inves ti g at in g th e several m o ve
m e nt s o f th e different pl anet s an d fro m th em infe r
th e n e cessary nat ure o f th e sys t em o f wh ic h th ey
S uc h
arepar t s w e h ave th e process o f in d uc t i o n
pr o cesses w e s e e mus t be c o mplem ent ary an d mu
A s L avat er says He only see s
tu ally d e p en d ent
well wh o sees th e wh ole in th e part s, an d th e part
in th e wh ol e
M o r e o v e r th e d is t inct i o n b etw e e n d e d uct ion an d
in duc ti o n m ay be sh o wn th r o ugh th e ir respect ive
re lat i o ns to th e universal , wh ich w e h ave seen is th e
gr o un d o f infer ence Th e quest i o n wh o s e ans w er
lea d s to th e d e duct ive pr o cess in reas o nin g is , Wh at
d o e s th e un iversal necessit at e
In in d u ct ion, th e
que s ti o n wh ich s t art s th e inves t i g at i o n is Int o
what sys t em m ay I c o ns t ruc t th e g iven mate rial
pr o p e rt i e s o r relat i ons s o as t o re ac h a universal
c o ncept th at will be consis t ent w ith it self an d wi th
th e wh o le o f kno w le dg e wh ic h f o rms th e w orl d of
c o nsci o usne ss ? In th is th ere is an analy t ical d is
crim inat i o n o f th e e ss ent ial an d acci d ent al elem e nt s
an d th e gath erin g t o g eth e r of th e f o rm e r int o t
c om plex wh o le wh ic h is th e univ e rsal In duct
th erefo re, is infere nce vie w e d fro m th e si de o f
diffe rences ; d e duc ti o n is inference vie w e d fro m
o f th e univ e rsal
F o r ins t ance, we m ay inves t i
th e ch aract erist ic fe at ure s of a diam o n d an d n d
th at a cert ain specic g ravity,
as com pared wit h
,

IN

D UCT ION

A ND

DEDUCT ION

19

w at er, is a cons t ant an d d e te rm inin g a tt ribut e an d


as such mus t be inc o rp o rat e d as an essent ial element
o f th e gene ral conc ep t d iam on d We can th en form
th e universal j udgm e nt W h at ever s t ones possess
th is sp e c ic g ravity are diamo n d s Th eir di fferences
re g ard in g siz e brill iancy, et c , m ay all be set asi d e
as acci de nt al but th e one const ant d et erm inin g
feature in d icat es a oneness in wh ich th ey all a gree
An d s o wi th th e o th er essen tial att ribut es A fte r
possessin g suc h kno wle dg e g aine d in ductively w e
m ay u s e i t prac t ically in a d e duc t ive manner ;
an d i t is s o use d in d iscrim inat in g betw een t rue
an d imi t at ion s t ones as d escribe d in th e foll o w in g
process
Diamon d s rubies an d sapp h ires are
n o w t es t e d by oat in g t o prove th eir g enuineness
T h e liqui d use d h as ve t im e s th e d ensity o f w at er
an d is co m pose d of d o uble nit rat e o f silv e r an d
th allium T h e t es t s are rapi dl y m a d e as all stones
o f th e same nat ure h ave th e sam e speci c g ravi ty
wh ile none o f th e bo gus one s h ave th e same w ei ght

e
as th o se th ey are mad t o imit at e
An oth er V iew o f th e relat i o n o f in duct ion to de
d uc ti on may be g ain e d by callin g att ent ion to th e
d ifference o f si g ni cance betw een th e t erm s a t ruth
A fac t carries with it onl y th e special
an d a fac t
an d in d iv i dual c h arac t er o f th e part icular o ccur
rence in wh ic h it is manifest e d A truth, h o w ever
is al w ays universal in it s very nat ur e, adm ittin g o f
uni versal applicat ion an d capabl e o f illus t rat ion
in an in d e n it e num ber of d ifferent fact s wh ich
e m bo dy it s e ssence In d e duct ion w e h ave g iv e n
some t ruth o f un iversal nat ure th at 1e
,

20

IN

DUCTIVE

O GIC

vi dual fact s th at m ay be subsume d und er it In


induc tion w e int erpret a fact or a number o f fact s
in th e li ght of th eir universal implication on
th e g ro un d th at th ere can be no such th in g as
an is o late d fact bu t eve ry fact must h ave so m e
re lation t o a unive rsal to wh ich it mus t be referre d
Wh ile considerin g th e d is tincti o ns betw een in
d uct ion an d de duct ion w e m us t not overlook their
We cannot procee d in d e
m ut ual d epen d ence
duct i o n irre spe c tiv e o f in d uc t i on because th e uni
v ersal up o n which th e d e duc tive pr o cess is base d
aris e s in th e m aj o rity o f cases from a previ o us
induct i on It is t rue th at the unive rsal t erm may
be in a proposition th at is kn o wn a p rio ri as th e
axio m s of g e om et ry an d c e rt ain space an d t im e
pos tulat e s ; bu t a v ery sm all prop o rt ion o f maj o r
premises can be sai d t o h ave such an o ri g in a
nd
th eir resul t in g conclusi o ns h ave very sli ght m a
terial si g ni cance
De duction th at reach e s oth er
th an purely abs t ract and fo rmal c onclusi o ns mus t
res t upon in duct i o n f o r th e m at erial t o form it s
pr e mises W e nd th is eve n in th e t ech nical c o n
s truct ion of th e syll o g ism wh e re f o r ins tance, th e
ques t i o n of th e d is t ribut i on of th e t erm s is raise d
W e may insis t th at a cert ain mi ddl e t erm is d is
t ribut e d as it is th e subj e c t of an universal afrm a
t ive propositi on ; but th e n the furth er ques ti o n
nat urally sugg es t s it self Ho w d o w e kno w th at
th e pr o p o si t i o n in qu e s t i o n is really a universal ?
Its mat erial si gni cance al o ne t e lls us th at w e
m ay write it as an A o r I pr o p o si t ion as th e case
m ay be
T h e matt er is a fun c tion of the fo rm ,
.

TO

DEDUCT ION

IN UC I N AND

21

an d th e form a func t ion o f th e matt er T h ey can


not be separat e d in fac t unless w e c o nceive reason
in g as a purely form al pr o cess o f d e t erminin g a
conclusion irrespect ive o f th e t ruth or falsity o f
If w e re g ar d th e prem ises as
th e prem ises
g iven, an d w e accept th em w ith unquest ionin g
cre d enc e th e d e d uct ion is purely formal ; s o
als o if th e vari o us t erm s are expresse d by l ett ers
A B 0 etc an d d evoi d o f any mat erial sig ni
cance A ny process o f reas o nin g base d upon a
slavis h accept ance of prem ises can only re ac h
art i ci al an d e ve n fal se r e sul t s
In th e act ual ex
perien ces of life o u r pr e mis e s are n o t m ad e fo r
us T h ey m us t be cons t ruct e d by u s th r o ugh o ur
int erpret at ion o f reality Disre g ar d o f th is h as
brought formal lo g ic int o m uch disreput e an d i t
h as o ft e n d e g enerat e d int o th e barren d iscussion of
lo g ical puzzles an d quibbles Grant a p e rson any
premises h e m ay ch oose t o assum e, irrespe ct ive o f
an in duc t ive t es t o f th eir vali d it y h e can pr o ve
black wh i te an d wh it e black
O n th e o th er h an d, ind uct ion is d epen d en t upon
d e d uc t ion ; for w e cann o t re as o n from particul ar
ins t ances t o a universal pr o p o siti on unl e ss w e as
sum e as basis o f th e wh o le in d uc t ive process some
pos tu late whic h h as real universal si g ni cance
O th erw is e w e reach only a h i gh d e g ree o f prob
ability , but n o t necessi ty ; a ru d e g eneralizat i o n,
but not universality Wh en w e assert som e suc h
g eneral st at ement as th is, th at arsenic al w ays ac t s
as a poison, we h ave th e universal ch aract er o f th e
proposit i o n upon an un d erlyin g post ul at e th at is
.

'

22

IN

D UCT IVE

O GIC

un d erst o o d e ven th o u gh it is no t e xpresse d such as


th e uniformity o f nature, th at un d er i de ntical con
T h is
d itio n s w e al w ays l ook f o r i d en tical effe c t s
w ill be d iscusse d lat er m ore in d e t ail ; it is re
ferre d t o at th is p o in t m er ely t o illus t rat e th e d e
Bra dley insis t s th at
d u c tiv e basis of in duc t i o n
th e re can be n o such th in g as in duct ion be cause it
al ways res t s up on an im plie d unive rsal wh ic h g ive s
t o th e process as a wh ole a de duct ive ch aract e r l
His cri t icis m h as th e f o rc e o nly o f provin g th at
in ducti o n canno t be in depen d ent o f d e ducti o n
T h is depend ence d oes not h o wever n e c e ssarily
vit iat e th e int e g rity o f in ducti o n as a mo deo f th e
inferent ial process L ot ze h as pl ace d sp e cial em
p h asis u pon th is depen dence of in duct ion upon
He says : It is th e cus t o m in o u r
de duct ion
d ay t o c oll e ct int o o n e bo dy th e numerous opera
ti ons wh ich assis t us in asc e n din g fr o m part iculars
t o g en e rals o r t o call th is in d uc t ive lo g ic , an d t o
set it agains t th e d e d uct ive o r d e m ons t rat ive lo g ic
alon g with much d isparag e ment of th e latt er S uch
disparag ement res t s o n a m is t ake T h e in d uct ive
m eth o d s i t is ce r tain, are th e m o s t e ffec t ual h elps
t o th e att ainm ent o f new t ruth, but it is no less
certain th at they re st ent irely o n th e resul t s o f
de duc t iv e l o g ic 2
Mo reover in in d uct ion th e resul t s ob t aine d an d
f o rmulat e d in g ene ral pr o p o sit ions may be ex t en d e d
an d o fte n m o die d by a de duc t io n wh ic h is base d
,

dl ey

Bra

P rinciples

Lo tz e , Log i c,
119
.

f Log ic

2 88

S ee

a so

332

Bo sanqu et, Log ic , V o l II


.

TO

IN UC I N AN D

DEDUCTIO N

23

upon th em as maj or prem ises ; for th e d e duct i o n


thus pr o c ee din g fr o m th em reveals ne w ins t ance s
whic h confo rm o r p e rh aps mo dify th e si m ple in d uc
t ive result s th e m selves Wh at is popularly call e d
a h as t y g en e ralizat i o n if mad e a m aj or prem ise o f
a syll o g ism , w ill o ft en lead us ast ray th rough th e
d educt ions drawn fro m it A s s oo n as we are aw are
o f th is
w e r e t urn t o ques t ion th e vali d it y o f th e
g eneral izati on wh o se w eakne ss is not appreciat e d
un til thus t est e d an d rev e ale d Th us de duc t ion
s e rves t o ex t en d an d c o rre c t th e res ul t s o f i nduc
t ion, an d at th e same t im e it it s e lf is depen d en t
upon th e r e sul t s o f in duct ive g ene ralizat i o n fo r th e
mat erial t o fo rm i t s pre m is e s We c o m e t o see,
th erefore h o w int imat ely ass o ciat e d th e s e tw o pr o c
esses are in ac tual reasonin g F o r c o nvenience o f
illus tratin g th eir in divi dual c h aract e rist ics th ey
may be consi de r e d as separat e, an d each inves t i gat e d
as an in depen dent m o de of inferenc e But th ey are
in reality m ut ually relat e d an d de pen d ent an d are
al w ays f o un d manife s t in g th eir funct i o ns t o g eth er
In any c o urse of reas o nin g conce r nin g th e c
o f o u r every d ay affairs or in scien t i c inves t i g at ion,
any wh ere, in de e d o ut si de of th e art i cial e xamples
o f lo g ical t ex t books w e reas o n b o th in d uct ively an d
d e duct ively in o ne complex pr o cess
.

CHAPT ER III
THE ESS E N T I A L S

OF

IN D U CT I O N

W E n o w pr o cee d t o a m o re precise d et e rm inat i o n


of th e nature of in duct i o n Its p o int o f vi e w in all
re as onin g re g ar d s c o ncret e ins t anc e s Th ey are th e
dat a an d fr o m th em g en e ral propositi ons are t o
r e sul t T h e proce dure is from g iv e n fact s t o law s
wh ich are th e groun d and e xplanat ion o f th ese
fact s We are h e re h o w e ve r at onc e st ru ck w ith th e
Pro
e vi d e nt br e ak in th e c o urs e o f o ur reasonin g
ce d u re fro m th e par ticular t o th e un iversal canno t
T h e re is a g ap s o me wh ere
be a c o nt inu o us proc e ss
T h e conclusi o n c o nt ains m ore th an th e premises
In de d uc ti o n w e are procee d in g fr o m th e g r e at e r
t o th e l e ss an d w e exp e ri ence no violat i o n o f o ur
l o gical s e ns e ; but at o nc e w e appreciat e th e dith
cul ty wh ic h att en d s th e reverse pr o c e ss , from th e
l e ss t o th e g re at e r He re w e soon reac h a point
wh ere w e pass beyon d th e sph ere o f our exp e ri e nce
eneralizatien which m e c e ss arily embrac e s
tF t
f iho r
th an o ur expe rie nce
T h i s is t he so call e d
fa
I n d uct ive leap ;
referre d t o as th e
in duct iv e h azard But is th is a l e ap in th e d ark
a w il d g uess concernin g all th at lie S be on d t h e
y
sensu o us sph ere of our imm e diat e exp e rie nce T h is
.

'

'

'

24

THE ES ENT IAL

S OF

TO

25

IN UC I N

w oul d be th e case w ere we c ompelle d t o use th e


mere d at a of experie nce as sole g roun d fo r our
inferenc e s J o h n Stuart Mill insis t s th at n o thin g
wh at ever is g iven in consciousness but par tic ular
sen sat i o ns , an d th ese are but subj ect ive s t at es of
fe elin g an d w ith no assurance o f any d enit e c o r
resp o n d ence wi th th e e x te rnal w o rl d W i th such
purely empirical d at a it is im p o ssible to proc ee d t o
t ruth s of universal vali dity It is n e cessary t o p o s t
ul ate some universal t ruth wh ich th e m in d th r o u gh
s t ric tly a p rio ri consi d erati o ns is cons traine d t o
formulat e an d wh ich w ill s e rve t o bri dg e th e gulf
be tw een th e part icular an d th e universal
Th is p o s tul at e h as be en vari ou sly express e d by
d ifferent auth o rs , yet w ith subst ant ially th e sam e
si gni cance in all In th e o l de r l o gic it is put
un der th e c o nvenient fo rm ula o f th e unifor m ity o f
nat ure ; th at is th at b eyon d th e sph ere o f e x peri
ence ph eno m ena will b eh ave in th e sam e m anne r,
un de r like c o n diti o ns as in th e sp h e re of im m e diat e
o bservation an d e xperim ent In th e m o de rn l o gic
th is is s o me wh at d ifferently e xpress e d The p hrase

uniform ity o f nat ure bein g s o m e wh at i n d e nit e


an d im plyin g a p o in t o f vie w purely o bj e cti v e is
n o t us e d
M o d ern w ri t ers h ave o m i tt e d i t lar g ely
fr o m th eir t e rmin o lo gy L o t z e says : T h e lo g ical
i de a upon wh ic h in duction res t s is by n o means
mere ly probable but cert ain an d irre frag able It
consis t s in the c o nvic tion base d up o n th e principle
o f i d ent i ty , th at every d e t er m inat e p h enomenon M
can depen d up o n o nly o n e d et erm inat e con d iti on
an d accord in g ly th at, wh ere un d er apparent ly dif
,

TV

26

O GIC

IND UC I E L

fo rent circum st ances o r in di ffer ent subj ec t s P S


T U th e sam e M occurs th ere m us t in e vi t ably be
in th em s o m e c omm o n elem ent 2 wh ich is th e t rue
i dent ical c o n dit i o n o f III o r th e t rue subj e c t o f M 1
W e h ave a so m ewh at sim ilar d e scripti o n of th e
basis of the in duc t ive process g ive n by S i gw art :

T h e lo g ical j ust i cat ion o f th e in d uc t ive process


res t s up o n th e fac t th at i t is an inevit able p o s t ulat e
o f o u r effo r t aft er kno w l e dg e th at th e g iv e n is
necessary an d can be kno wn as proc ee d ing fr o m it s
2
Bosanquet
g r o un d s acc o r din g t o universal law s
c onsi de rs as th e basis o f in d uc t ive infe rence th at
wh ic h h e calls th e pos tul at e o f kn o wl e dg e th at th e
univers e is a rat ional sys te m t akin g rat ional t o mean
no t only of such a nat ure th at it can be kno w n by
intelli gence but furth er o f such a nat ure th at it
3
can be kn o wn an d h an dl e d by o ur int elli g ence
I h ave quo t e d th e s e passag e s from L o t ze Bosan
que t an d S i gwart th at w e may appreciate th e mo d
e rn t en d ency t o d e riv e th e in d u c t iv e pos t ulat e fr o m
an e piste m ol o gical s o urc e ; nam ely, th at our kno wl
e dg e mus t be c o nsiste nt th r o u gh out w i th it self,
p ar t to part an d part s to wh ole an d th at th e w orl d
for us is th e w orl d as c o ns truct e d by o ur kn o wl e dg e
Wh at eve r is given in c o nsci o usn e ss m us t belon g
the r efo re in th e o ne plac e wh ere i t appr o priately
an d necessarily belon g s He r e als o th ere mus t be
a plac e f o r e ve rythin g an d every th in g in it s place
T he re mus t be a unifo rmity of consciousness ; th at
,

1
2
3

Lo t ze , Log ic ,
102
g wart , Lo g ic ( Eng t rans at o n ) , Vo l II
Bo sanq ue t , Th e E s sentials of Lo g ic , p 166
.

Si

l i

289

N TIALS

THE E SS E

OF

27

I ND U C TIO N

is, th e primary pos tulat e an d th e uniformi ty o f


nat ure is secon d ary t o this an d implie d in i t T his
pos tul at e may al so be expresse d as foll o w s : Wh at
is once t rue is al ways t rue Here t rue is use d in
th e sense of th e un iversal si gnicance o f a fact
Wh enever a concret e in s tan ce is present in con
its exis t ence mus t be consi d ere d as
sc io u sn ess
necessit at e d by some ant ece d ent wh ic h can satis
fac to ril y accoun t for i t an d wh ic h c an at th e same
t ime be appropria t ely a dj ust e d t o th e wh ole o f o u r
kno wle dg e in int erpretin g it Bosanquet says th at

i d eally speakin g every concret e real tot ali ty can


be analyze d int o a complex of necessary relations 1
T h ese necessary relations of course h ave a uni versal
si gnicance, an d th erefore in every concret e in
s t ance if w e can ri ght ly int erpre t i t w e may d is
cern th e un iversal element whic h is cont aine d in it
an d gives i t a place an d meanin g in th e w orl d as
co gni ze d by us
T h ere i s a sense in wh ich induction may be re
g ar d e d as th e inverse process of d e duct ion In
d e duc tion th e problem is concerne d w ith th e ques
In
tion, Wh at does th e universal necessi t at e
in duction th e inst ance is given and th e problem
is Wh at universal can be d iscovere d wh ic h coul d
give ris e t o th e ins t ance in ques tion
Th is view
of in duct ion is especiall y associat e d wi th th e nam e
of J evons wh ose in duct ive sys t em is d escribe d as
He calls i t th e d eciph er
th e inverse of d e duc t ion
2
h
ing of th e h i dd en meanin g of natural p enomena
,

B o s anquet Log ic Vo l II p 82
2 Jev o ns Prin c i les o S cie n ce
f
p
p

124

INDU C TIVE

28

LO GI C

T h e name commonly use d t o d esi gnat e th is vie w of

in duct ion is th at of re duc tion ori ginally su g


g es t e d by Duh amel 1 T his process w as kno w n t o
th e ol d lo gicians wh o calle d it Meth o d t o d enot e
th e process of huntin g for mi ddle t erms by th e ai d
2
of whic h a g iven conclusion coul d be prove d
L ike
all inverse processes, i t is by it self an in d et erminat e
,

o ne

Given

All A is B,
A ll B

an d

is 0 ,

w e infer by th e direct process of d e duction th at


A ll A is 0

But in th e in direc t

inverse process w e h ave


given all A is
and th e problem , t o n d a mi ddle
t erm wh ich necessit at es suc h a conclusion is an
in d et erminat e one
T h ere may be a number of
mi ddle t erms T h is is analo gous t o th e me th o d
of inte g ral calculus ; wh ile differentiat ion lea d s t o
a d enite resul t th e inverse process o f int e g ration
lead s t o an in d e t erminat e result S o also w e mul
tiply tw o numbers pro ducin g on e d e t erminat e re
sul t ; bu t inversely wh en w e h ave given a cer tain
number an d ask wh at fac t ors mul tiplie d t o g e th er
coul d pro duce th is number we m ay reac h several
different solutions T h e ans w er is in d e t erminat e
P rofessor J evons in his sch eme of in d uc tive infer
ence falls back upon probability t o in dicat e wh ic h
of several possibilities is th e mos t likely o ne in th e
or

1
2

Duh amel , Meth o d es, Vo l I p 24


.

Venn

c a l L og i c

E mp iri

p 361
.

THE ESS EN

T IALS

or

29

INDU C TIO N

given case But before th e in verse operation can re


sul t in d et erminat e resul t s , th e given t erms such as
A an d 0 mus t be subj ec t e d t o some analysis in ord er
th at th eir mat erial si g ni cat ion may g ive su gg e s
tion as t o th e nature of th e mi d dle t erm F or in
s t ance a man is foun d d ea d, w ash e d ash ore by th e
t i d e ; th e natu ral supposi tion w o ul d be th at h e met
h is d eath by d ro w nin g
A n d ye t i t mi ght possibly
h appen th at th e man die d th rou gh injuries in ict e d
by blo w s , or by poison or h eart failure T h e at
t en dant circum s t ances an d bo dily indications mus t
sugg es t th e mos t probable cause to ac co unt fo r th e
g iven effect V enn criticises J evons view of induc
t ion makin g i t th e inverse process of d e duction o n
th e g roun d that i t is purely a formal process an d
th erefore can lea d only t o in det erminat e resul t s 2
I t is al ways possible h o w ever t o make some
anal ysis of th e mat erial si g ni cance of th e d at a as
h as been above indicat e d wh ich relieves th e purely
formal processes from th e in d en itenes s of th e re
Bosanque t cri t icises J evons th eory of in
s u l ts
d u ctiv e inference in th at th e h ypoth esis propose d
t o accoun t fo r th e given in reali ty can at bes t be
3
onl y h i ghl y probable
Ho w ever V enn L ot ze Bo
s an qu et S i gw ar t all allo w a place t o th e inverse
fu nc tion of all in ductive reasonin g ; th eir c o nten
t ion, h o w ever is this th at i t does not furnis h an
ad equa t e acco u n t of th e wh ole matt er
l

"

Jv
e

o ns ,

E mpiri

Prin cip les of S cien ce

c a l L og i c

p 219
.

p 359
3
o s an uet , Lag ic , V o l II p 17 5
4
II p 17 5 ; Sig wart, Vo l
o s an ue t, V o l
e nn , 3 6 1 ;
Lo tze , Ou tlines of Log i , p 93
p 203 , 289
2

II

I NDU C TIVE

30

O GIC

I t is int eres tin g t o not e th at Wh e well s th eory o f


in duc tion correspon d s in th e main t o this i d ea o f
re duction or inverse process He n d s in in duc
t ion a tw ofol d operation o f th e min d, consis tin g in
th e colli g ation of fac t s an d th e explication o f con
By th e colli g at ion of fac t s h e refers t o
ceptio n s
th at insi ght whic h is able to see th e connections
and relations wh ich necessarily exis t be tw een th e
different p h enomena present in consciousness ; an d
by explicat ion of concep tions h e refers to th e ap
pro priate ttin g in of th ese relat e d fact s t o some
conception o f th e min d which mos t rea dily ac
count s for th em 1 S uch a process is merely th e
rea ding of given fact s back w ar d t o th eir origi n ,
or s ubs t antially an inverse process wh ere th e pro
ced ure is from th e given concre t e to th e explanat ion
of th e same in t erms of th e universal to wh ich i t
can be mos t appropriat ely referre d S o also Mill s
account o f proce dure by h ypoth esis as w e S h all se e
lat er o n, presen t s ch arac t eris tics similar t o th is
process of re duc tion
Th e en d of in duction is t o discover a law h avin g
obj ective vali di ty an d universal application Th ere
is a dis t inc tion whic h mus t be no t ice d an d clearly
kept in min d ; namely th e dis tinction betw een a law
an d a r ule In duc tion seeks a law and no t a rule
A law expresses th e essent ial an d universal rela
tions subsis tin g be tw een given p h enomena elimi
natin g entirely all acci d ent al an d local colorin g A
law h as obj ective vali di ty an d preserves a cons t ant
nature T h ere can be only one law in reference
.

W h ewell , P h ilosop h y of

d c

th e I n u tiv e S c ien ces , pp 17 2 , 202


.

THE E

SS ENT IALS

IND U C TIO N

OF

31

an d th e same connec tion of fact s A rul e


h o wever is subj ec t ive d ealin g with th e indivi dual s
attitud e t o ph enomena, rath er th an an explanation
of th e essential featur es of th e ph enomena th em
selves It oft en is d et ermine d in th e concret e by
th at wh ic h is ext ernal local, an d acci dent al Th ere
may be many rul es varyin g with many min d s an d
many climes F un d ament al an d universal l aw s o f
political economy become maxims an d rules in d if
fero nt commun ities T h e law s of morali ty univer
sal an d immut able becom e rules of con d uc t in in
d ivi dual experience adm i ttin g of wi d e difference of
O pin ion an d di versity of application In th e pro c
es ses of in duc t ion th erefore th e la w is th e d esi d era
turn, an d no t th e rul e
L aw h o w ever is use d rath er loosely in o ur ordi
nary t erminolo gy L aw as use d in jurispru dence
h as a meanin g qui t e different from law as use d in
p hysical science An d so also, th e law s of biolo gy
th e law s o f poli t ical economy th e law s o f e thics
are referre d to with different sh a des of meanin g in
each sph ere Ho w ever ambi guous may be th e si g

law in or dinary th ou ght an d usag e,


n ic an ce of
neverth eless in in duc tion i t h as a const ant an d a
simple si gnicance, wh ic h if careful ly a dh ere d t o
will a v oi d con fusion an d obscuri ty as w ell in o u r
inferential processe s an d resul t s L aw in in duction
is al w ays in th e form of an h ypo th e t ical un iversal
to

one

If

is , B

is

I t d oes not as sert wh at h as h appene d, but


1

Lo tze , Log i

p 3 35
.

w h at

INDUCT IVE

32

O G IC

S h oul d h appen un d er cert ain condi tions

Given

ant eced ent A a cert ain d eterminat e consequent


Th e relation is cons t ant
B is al w ays necessi t at e d
and invariable an d th erefore h as a universal signi
cance
Induc tion h ol d s a peculiar an d import ant place
in our every d ay life, because i t h as t o d o wi th th e
analytical t reatment of ins t ances as th ey appear in
experience Th e lar g e part o f our conscious think
ing h as t o d o wi th th e concret e th e raw mat erial
in duc tion alone can h an dle
o f experience ; th is
L eonar do d a Vinci s maxim was t o be gin wi th ex
1
perien c e an d by means o f i t t o direc t th e reason
T h us th e supers tructure o f kno wle dg e is raise d day
by day Th e given is continually bein g int erpret e d
an d referre d to its appropriat e place as th e s t ones
o f th e quarry are h e wn an d tt e d in th eir proper
p osi t ion in th e buil din g for wh ic h th ey h ave been
desi gned T h ere are cert ain in divi dual experiences
wh ich i t is impossible t o d et ermine th rou gh ou r
syllo gis tic forms T h ey canno t be ju dg e d d e d uc
tiv ely
Th ere is no g en eral cat e gory un d er which
th ey can be subsume d T h ey may be formally
illo gical if thus expresse d an d yet adm it of d irec t
inves ti g ation an d experiment in an in ductive man
ner fo r th e purpose o f disclosin g th e law un d er
lyin g th em and as yet unkno wn
I t O ft en h appens th at th rou gh in difference o r
in d olence w e are cont ent t o refer many ph enomena
to lon g es t ablis h e d an d convenient cat e g ories wh ic h
if inves ti g at e d in depen d ently w e w oul d n d coul d
th e

Ueb erweg , Log i

p 42
.

THE ESS EN

TI AL S

OF

INDU CT ION

33

not possibly be so t reat e d T h e convenient pi g eon


h ole because near at h an d receives much th at d oes
n o t properly belon g th ere
I t is th e Ofce o f in duc
tion t o inves ti g at e anew th e ol d mat erial an d th en
t o recl as sify in accord ance with th e revise d g en
e ral izatio n s wh ic h suc h inves t i g at ions may n eces
S i t at e
T h e proce dure by in duction is in keepin g with
th e scien ti c spiri t of th e d ay to int erpre t th e
ph enomena of natur e as g iven an d not t o antici
pat e nat ure th rou gh preconceptions an d w res t fac t
in or d er to t th eory
It comes t o th e sources in
nature w ith empty vessels to draw an d carry away
th at which natur e alone can give
.

CH A P T ER
T Y PE S

OF

IV

I N D U CT IV E I N FE REN C E

process of in duction as w e h ave seen is a


proce dure fro m g iven ins t ances t o th e discovery of
th e law which und erlies th em an d wh ich is th e
g roun d o f th e connect ion of th e various attribu t es
an d relations th at unit e in th e o ne concre t e wh ole
Vie w e d from th e s t andpoint o f th e direction of th e
process w e h ave foun d th at it is al w ays t o war d s
so m e g eneral expression of in divi dual experiences,
an d in this respect i t is th e inverse O f d e duction,
which procee d s from th e g eneral t o th e particul ar
which is embrace d in i t T h ere is, h o wever anoth er
and import ant point of view th at S h oul d no t be over
looke d We h ave t o consi d er th e mo d e of th e pro c
e ss as w ell as its d irec t ion ; no t merely th e resul t
t o be att aine d but also th e peculiar manner of
realizin g th e same mus t be consi d ere d Difference
in me th o d h ere gives rise to various kin d s of in duc
tive inference T h e en d propose d in all is to g en
e ralize o u r experiences as th ey occur in th e concre t e
an d particul ar W h en I n d a g iven ph enomen on
A , g iven in consciousness , an d ch ar ac t erize d by
several di s t inctive feat ures amon g wh ich I not e
specially th e mark B, th e ques tion at onc e mo st
TH E

34

TYPES O F

I NDU C T IVE INF ERENC E

35

naturally su gg es t s i t self I s th ere a reasonabl e


expec t at ion th at I sh all al w ays fin d B as an insep
arable accompaniment O f A so th at I can assert
cond en tly th at wh enever A is found B also will
be foun d ? Th ere are th ree w ays o f satisfyin g
ourselves as t o th e exis t ence of any cons tant rath er
th an coinci d ental connec tion betw een ant ece den t
an d consequen t as A an d B T h ese give rise t o
thr ee dierent meth o d s of in ductive research, an d
th ey are as follo w s :
I T h e Meth o d of Enum erat ion
II T h e Meth o d of Compariso n o r A nalo gy
III T h e Me th o d o f S cientic An aly sis or
S earc h aft er Causal Connect ion
F ail ure to dis t in guish betw een th e th ree meth o d s
h as given rise t o confusion in th e d eni tion o f an d
correspondi n g reference to in ductive inference ;
s ome auth ors use in duct ion in o n e an d some in
ano th er of th ese senses I t is necessary t o dis
criminat e carefully an d to maint ain a s tric t con
sis t ency in th e usag e o f th e terms as d en e d
I The M etho d of E n umeration W e observe th e
various ins t ances in wh ic h cert ain attribut es as A
an d B are conj o in e d in o ur experience We count
th em in th e sense o f no t in g t o wh at ex t ent th ey
accum ul at e w i th out no ticin g any exception t o wh at
seems at leas t an invariable connection We d o
no t necessaril y coun t by precise enumeration re ac h
in g a nu merically d e nit e resul t
We not ice
merely t o wh at ext ent th e observe d ins t ances o f
like nat ure accumul at e ; th at is wh eth er a fe w a
consi d erable number, or a very lar ge nu mber Th e
,

INDU CT IVE

36

O GI C

mere number of ins tances pro duces a cert ain psych o


lo gical impression wh at ever may be th eir lo gical
force T his is brou ght about th rou gh th e la w s of
association an d creat es an expec t ation of a con
tinu o u s repe t i tion o f th e experience in ques t ion
T his arises from a nat ural t en d ency of th e m in d to
g eneralize We observe th at cro w s are black ; an d
th e increasin g number of conrmin g ins t ances goes
far to est ablis h a connection between th e cro w an d
it s color which seems t o h ave universal vali dit y
T h e enumeration of ins t ances may lea d us t o any
o n e of th ree resul t s :
1 We may mee t w ith n o excep t ion wh at soever,
until th e scope of o bservation complet ely embraces
th e su m of all possible ins t ances T his i s complete
enumeration an d wh en enumeration reach es th is
limit it passes o v er int o d e duc tive reasonin g by
virtue o f th e lo gi cal canon th at wh at ever is t rue of
th e part s is t rue of th e wh ol e dis t ribu tively ; th at is
provi d e d th e summation o f th e part s h as been an
exh aus tive o ne We assert th at all th e sh eep o f a
given ock are wh i t e ; for w e h ave observe d each
separat ely an d no one h as been misse d in th e coun t
S o al so th e ju dgmen t th at all planet s move aroun d
th e s u n resultin g from an enumeration of th e planet s
o n e by one
I t i s possible also t o h ave a perfect in
duction wi th an innit e enumerat ion of part s T h is
is possible in tw o cases as point e d ou t by Ben ek e l
Firs t wh en th e part s are connect e d t o g eth er contin
u o u sly in space so th at a survey of all is possibl e
in a nit e, and oft en a very S h ort time This occurs
,

'

Q u o ted by Ueb e rw eg Lo g ic , p
,

482

TYPES

OF

I ND U CT IVE IN FERENCE

37

in

g eome trical d emons t ration wh en th e inf erence,


base d upon th e S imple g ure i t refers to is ex t end e d
t o all gures fallin g un d er the like d eni tion An d
secon d wh en th e part s are n o t continuously con
n e c te d if i t can be prove d syllo g is t ically th at wh at
i s t rue of a d enite nth part mus t also be t rue for
th e (n 1) th part
P erfect in duc ti on al so embrac es arith metical
meth o d an d compu t ation Here th e wh ol e whi ch
is th e sum o f th e fac t s in eac h cas e is a to t ali ty
o r universal wh ose differences wh ic h are all sepa
rat e an d di s t in g uis h able are yet h omo g eneous an d
equal l T h ere is no quali t ative differentiation of
part s onl y a quant i tative one Th e t o t al is th e
su m of th e un i t s an d eac h uni t is like every o th er
o ne
If w e h ave o n e h un dre d uni t s makin g a
t ot ali ty th e o ne th at may be th e tw enty seventh
is precisely like th e six ty seventh I t is a cas e
wh ere each one count s fo r one an d no o ne fo r
more th an o n e in an absolut ely li t eral sense
I t h as been urg e d a g ains t perfect in duc tion th at
i t a o rd s no ne w i nf ormat ion and th erefore i t s
resul t s are n o t valuabl e Ho w ever th e sum mation
o f pa rticul ars in abbrevi at e d forms is al w ays an
a dvan t ag e I t is a labor savin g process t o th e min d
I t enables th e mind t o ret ain a lar g e num ber of
fac t s by th ro win g th em int o o ne an d th e same cat e
g ory ; an d i t fac ilit at es arith metical processes by
conveni ent compre h en din g of unit s wi th in a t o t ality
2 T h e secon d resul t th at is possible is th at in
coun tin g ins tances , o u r enum erat ion s h oul d pro v e
,

B os an quet

Lo g i

Vo l II p 54
.

I NDU C TIVE

38

O GIC

incomplet e F rom th e necessi ties of th e case, w e


are oft en not able t o Observe th e en tire sp h ere of
possible occurrences and cover th e wh ole g roun d
I t may be th at beyon d th e sph ere of our expe
th e cons tant connec tion betw een certain
rien c e
p h enomena may be dis turbe d by th e appearance
of some variable fac tor o f wh ic h w e h ave been
wh olly i gnorant I t is th e possibili tie s beyon d
th e sph ere of observation whic h ren d er uncer t ain
th e res ul t s o f our count We are sure as far as w e
h ave observed ; but w e h ave not g one far enou gh
perh aps S uch resul t s , formul at e d in g eneral prop
are t erme d empirical law s ; th at is g en
o sitio n s
e ralizatio n s from an experience necessarily limi t e d
3 We h ave s t ill a th ir d case ; wh ere in our
enumeration of posi tive inst ances w e meet wi th
excep tions to a great er or less ex t ent Here w e
cannot even su m up th e act ual experience in t erms
of a g eneralization T h ere are out s t an din g ex cep
tions whic h will invali d at e it We mus t th erefore
fall back upon th e th eory of probabili ty an d th e
calculation of ch ances presuming th at in g eneral,
w e will meet with th e same proportion of e xcep
tions t o positive ins t ances in th e future th at w e
h ave alrea dy observe d in th e pas t S o w e m ake
in our min d s at least comparative t ables o f posi
t ive c as e s over a g ains t excep t ions an d reac h a
summary o f th e resul t in th e form of a rat io,
w h ose n umerat or w ill be th e nu mber o f posi t iv e
cases observe d, an d th e denominator th e t ot al num
ber of ins t ances inclu din g posit ive ins t ances an d
th e correspondin g exceptions We observe th at
.

TYPES

I NDU C TIVE I NFERENCE

OF

39

some crypt o g amous plant s possess a purely cel


lu lar s t ruc t u re ; oth ers , h o w ever, d o n o t, bein g
par ti al ly vascu lar T h e probability th at a ne w
cryp t o g am wil l be cellular can be es timat e d only
on
th e g roun d o f th e comparative number of
know n cryp to g ams wh ic h are cellular, as over
a g ain s t th e t ot al number of cryp t o g ams , bo th ce l
lul ar an d vascular prev iously o bserv e d I
II T he M etho d Of A n alogy Here, also, w e
s t art w i th th e experience th at A is c h arac t erize d
by th e mark B But th ere is add itional kno wledg e
o f wh ic h w e may avail ourselves in th e g eneraliz a
tion of some pas t experience alrea dy effec t ed, suc h
as th e follo w in g : th a t A very closely resembles O in
th at th e two h ave m any properties o r att ribut es in
comm on T h e i nference by analo gy is th at 0 also, as
w ell as A, w ill h ave th e mark B It may be th at w e
cannot examine 0 in a number o f various ins t ances
O ur only
to s ee in h o w many th e mark B occurs
resource is th e inference wh ic h is base d upon th e
know n resemblances or analo gies Th is kin d of
inf erence fo r exam ple was employe d by S ir Isaac
N e wt on in a v ery int eres tin g manner He h a d oh
serve d th at cert ain fat sulph ureous unc tious
bo dies, such as camph or Oils spirit of t urpentine,
amber, et c , h ave refrac t ive po w ers tw o o r th ree
t imes g reat er th an mi ght be ant icipat e d from th eir
d ensities He notice d aISO th e unusually hi gh re
frac tiv e in d ex o f di amon d, an d from th is resem
blance bas e d upon similari ty in r eference t o one
att ribut e onl y h e inferre d th at diamon d also w oul d
.

Jv
e

o ns ,

Pri nc ip les of S c ien ce

pp 146 , 147
.

I NDU C T IVE

40

O GI C

prove to be com bus tible His pre d iction in this


re g ar d was veri e d by th e F lorentine A ca d emicians
in
Bre w s t er ma d e a s t rikin g commen t upon
N e wt on s inference t o th e effect th at if N e wt on
h a d d raw n a like analo gy in reference t o g reenock
it e an d oct ah e drit e as h e di d concernin g diamon d,
inasmuch as th ey, to o, h ave a very h i gh refractive
in d ex h e w oul d h ave been wh olly incorrect T h is
is an in d icat ion of th e fac t th at ar g umen t by anal
o gy is no t conclusive
Bosanquet h as very s t rikin gly expresse d th e es
sence o f th e analo gical meth o d in sayin g th at in
A nalo g y w e w ei gh th e ins t ances rath er th an coun t
2
th em
T h e dis tinc tion be tw een an alo gy an d
en umeration o f ins t ances lies in th is th at in th e
former w e count similar att ribut es in th e cont ent s
O f tw o ins t ances an d balance th em a g ains t th e d is
similar o r u nknow n In in duct ion by enumera
t ion w e count similar ins t ances consi d erin g th em
in th eir t ot ality wi th ou t examination an d compari
so n o f th eir respec t ive att ribu t es
III The M etho d of S cien tic A n alysis
The
inst ance in ques tion A , wh ic h is ch arac t erize d by
th e mark B is subj ec t e d t o a vi g orou s analytical
examinat ion to Sh o w th at A an d B are relat e d
th ro ugh a causal connec tion
T his analysis is
effect e d eith er th rou gh a minut e observation or
by means o f exact experiment Th e en d t o be
att aine d by such analysis is t o separat e a complex
ph enomenon int o i t s several element s, by wh ich
.

1
2

Jv
e

o ns ,

Princ ip les of S c ience

B o sanquet

The Essen tials

p 527
Lo g i , p 155
.

TYPES

OF

INDU C TIVE IN FERENCE

41

process a caus al co nnec t ion may be reveale d, wh ose


very exis tence is dis guis e d by th e complexity of
th e p h enomenon F or ins t ance, th e p h enomenon
o f d ea th follo win g th e t ak in g of arsenic is an
event so complex as t o evad e a precise d et ermin a
t ion o f th e causal relation Wh en analyze d int o
s impler element s, i t is foun d th at th e imme diat e
e ffec t o f arsenic upon th e bo dily t issues is t o
h ar d en th em s o as to prevent th eir normal funct ion
in g Thi s is th e causal g roun d of th e d eath d ue t o
arsenic
Moreover, th is an al y tic process , wh ich
may be appropriat ely cal le d a mat erial o n e, is sup
plem en te d by a formal process of ne g at ion, th at is,
an ins t ance in which th e suspect e d causal element
is absen t in th e complex p h enomenon un d er in
v e stigation , an d th e relat e d eff ec t, before O bserve d
n o w no lon g er appears
Th is formal process ac t s
as a c h eck an d as a v erication as w ell of th e
mat eri al anal ysis of th e p h enomenon
Fo r ex
ample, an an t i do te, as sesquioxi d e o f iron, bein g
ad minis t ere d, no d eath from arsenic occurs ; an d i t
is al so observe d th at no h ard enin g of th e tissues
h as resul t e d, th erefore th e former res ul t, h ar d enin g
of tissues pro d ucin g d eath h as been thus c o rro bo
rat e d ne g atively by th e fac t th at wh ere no h ar d en
in g of t issues h as resul t e d d eath d oes no t follo w
W e see at once th e ad vant ag e of suc h a me th o d
over th at of coun tin g all ins t ances wh ere t akin g of
T h e latte r is a ph enom
arsenic h as cause d d eath
en al ly a d ju dg e d res ul t ; th e former pene trat es wi th
analy tic insi ght to th e g roun d of th e ph en o m e
non it self T hus o ne in s t ance, if its part s an d th eir
.

42

INDUCTIVE

O GI C

manifol d relations are a dequat ely compreh en d e d,


may s ufce for a uni v ersal conclusion base d upon it
It is t rue, h o wever as remarke d by Bosanquet th at

num ber of observations d oes, as a rule, assis t an aly


S cient i c
sis an d cont ribu t e t o eliminatin g error
analysis as such, h o w ever d oes no t deal w i th in
s t ances but only with cont ent s 1
In cases wh ere th e ph enomenon d oes not reveal
i t s component element s und er observation an d i t is
impossible t o s ubj ect i t to experiment th e mos t
likely cause of th e effec t in ques tion is t ent ati v ely
j udg e d t o be th e real cause unt il i t can be veri e d
in reali ty T his is proce dure by hypo th esis, an d
is al ways resort e d t o as preliminary t o a subsequent
experimen t whic h is its t es t or else in lieu o f such
an experiment wh en i t is by th e nature of th e case
preclud e d I t is a form of id eal analysis Th e ex
T h e ph en o m e
perim ent is cons t ruc t e d ment ally
non is separat e d int o wh at w e w oul d reasonably
imag ine i t s simpler element s w oul d be We are
cons t raine d t o believe th at if th e hypoth et ical ant e
ce d ent exis t e d it w oul d be ad equat e t o pro duce th e
effect A l thou gh risin g in th e sph ere o f th e imag i
nation it is th at with wh ich th e min d is for th e t ime
at leas t, satise d as an explanation of th e fac t s which
d eman d some cause t o account for th em R e g ar d
in g in duction as a process of re duction h ypo th e
s is is th e assume d universal or mi dd le t erm wh ich
w ill necessi t at e th e p h enomenon un d er inves t i g a
t ion as it s lo gical conclusion
We w ill no w procee d t o a furth er examination
.

B o s anq uet

L og i

V o l II p 118
.

TYPES

OF

INDU C TIV E IN FERE NC E

43

of

th ese me th o d s, consi d ere d both sin gly and t o


g e th er
1 T h ey all procee d upon th e supposi tion th at
wh at is given in consciousness h as some necessary
g roun d for i t s bein g In enum erative induction
th ere is some causal conn ec tion presuppose d yet in
a very g eneral and in d e n i t e manner, an d acc o m
panie d by no analysis o f th e various concep t s ei th er
by a sys t ematic observation o r expe rii n ent I t is a
va gue sense of uniformi ty wh ich wh en observe d for
many times w e feel wil l continue in d en it ely Th at
wh ic h h as h appene d oft en an d n o t contradi c t e d car
ries wi th i t a cert ain convincin g po w er by d int of bare
repeti tion especially t o persons of narro w e xperi
en ce, an d unaccus t ome d to di scriminatin g observ a
tion U ebe rwe g h as ma d e th e follo w in g comment in
reference to th e s o called imperfec t in duc tion
The
conclusion is ma d e uni versal w i th more o r less prob
abil ity, an d th e bl an k which rem ains over in th e
given relat ions o f sph eres is le gi timat ely l le d up
partly o n th e universal presupposi tion of a causal
nexus in th e Obj ec t s of kno wled g e partly on th e
part icul ar presupposi tion th at in th e case present e d
such a causal nexus exis t s as connect s th e subj ect
T h e d eg ree o f prob
and pre di cat e of th e conclusion
ability o f th e in duc tive inference d epen d s in eac h
case on th e ad m issibili ty o f th is las t presupposi t ion,
an d th e various in duc t ive operations , th e ext ension
of th e sp h ere of observation, th e s implication of th e
observe d con di tions by successive e xh aus tion o f th e
1
unessential, e tc all t en d to secure it s admissibili ty
.

Ueb erweg , Logi

pp 483 f
.

INDU C TIVE

44

OGIC

A nalo gy

like wise procee d s upon th e assumption of


an un d erlyin g cause amon g th e observe d ph enomena,
an d this is more d eni t ely in th e fore g roun d th rou gh
out th e process than in th at of in duction by e nu mer
ation A nalo gy is base d upon th e post ulat e th at sim
ilar p h enomena h ave similar causes ; th e g reat er th e
ag reement o f th e various att ribut es o f th e d ifferent
ph enomena compared, th e great er will be th e resul t
ant probability th at causes capable o f pro ducin g
th e m as effect s will be similar Th e similari ty of
th e li ghtnin g as h t o th e elec t ric spark sugg es t e d
to Benj amin F ranklin th e possibili ty th at th ey w ere
due t o a like origin an d by experiment h is analo g ical
reasonin g w as actually conrme d as is w ell kno wn
U pon th e th eory th at th e w orl d as i t exis t s for u s
in kno wle dg e forms a sys t em t o some place in wh ich
every ph enomenon g iven in experience mus t be
appropriat ely an d necessarily referre d, it follo w s
th erefore, th at a simple experience d evoi d of any
complexity of part s may t into several possible
places in o u r w orl d o f consciousness an d remain s o
far forth ind e t erminat e Ho w ever a complex p h e
w ith many par t s intricat ely connec t e d
n o m en o n
will t int o one unique place only in th e sys t em t o
wh ic h i t mus t be referre d I t is like a key th at
T h e presu m ptio n th ere
w ill t int o only one look
fore is th at any oth er ph enomenon whic h resem
bles th e rs t th rou gh muc h Of its ent ire cont ent,
part fo r part att ribut e for att ribut e will also re
semble i t furth er as re g ar d s oth er att ribut es n ot yet
examine d so as i t will like wise t int o th e peculiar
place in th e sys t em of kno wle dg e to whic h th e
.

TYPE S

U TV

45

I NFERENC E

O F IND C I E

s t h as been foun d to belon g T h ere is al w ays a


s tron g probabili ty th at ag reement in sph eres of g reat
complexi ty is no t a m ere coinci d ence, but th e resul t
o f a causal relation
O ne c h arac t eris tic of a sys t em ,
whic h w e h ave foun d t o be th e g roun d o f inf erence
g eneral ly is th e c o or din ation of like thin g s un d er
one concep t A nalogy , th erefore, is base d upon th e
v i e w of caus al connec t ions wi th in th e sys t em wh ic h
comprises th e w orl d as given in con sciousness
In th e th ir d meth o d th e causal relation is more
prominent s t ill an d th e searc h for i t ch aract erizes
T h at, wh ic h in th e o th er
th e proce dur e employe d
meth o d s may exis t merely as a vague impression,
is h ere formulat e d an d ma d e th e direc t an d sole
O bj ec t of research
2 T h e th ree meth o d s in th e or d er h ere presen t e d
s h o w an increasin g prominence given t o th e causal
conn ect ion in th e ph enomena o f experience An d
th erefore th ey possess a relat ively increasin g sc ien
A s th e rs t h as only in direc t reference
tic value
t o th e causal connec tion o f i t s fac t s, i t is th e leas t
t rus tw orth y an d h as no claim as a scientic meth o d
I t breaks d o wn as soon as an exception is not e d ;
an d is even w eakene d by th e fac t th at it is con
stan tly menace d by th e possibil i ty at leas t of th e

appearance of an exception
Ho w d o w e kno w
says G reen th at th e ins t ances , with th e examina
tion o f wh ich w e are al w ays dispensin g on th e
s t ren gth of th e rule ( th at i s ou r g eneralization) ,
mi ght n o t be jus t wh at w oul d invali d at e i t if th ey
w ere ex amine d ? 1 We may arrive at th e conclu
r

'

Green Ph il Works
,

Vo l II p 282
.

INDU C T IVE

46

O GI C

sion, base d upon o u r observation an d consequent


recor d, th at all sh eep are whit e an d yet blac k
s h eep d o occur even in every ock as th e proverb
A ccor din g to A ris t o tle th e proposi tion th at
h as i t
all s wans are whi te w as a perfectly g eneral one,
an d yet in recent t imes black s wans h ave been
discovere d in A us t ralia Bacon s criticis m upon

In d u c tio quae
th is meth o d h as become clas sic
pro c e d it per e nu meratio n em simplicem , res pu erilis
e st e t precario co n c lu d it et peric ul o expo n itu r ab
in stan tia cont radi c t oria et pl erum qu e secun dum
pau c io ra quam par e s t et ex iis t antu mmo d o quae
1
r
n
u
n
i
a
t
pres t o sun t p o
c
T h e vali dity of this meth o d o f proce dure d epen ds
larg ely upon th e probabili ty o f o u r meetin g an d
noticin g excep tions w ere th ey to occur A s L ot ze
put s i t : A man wh o never observes a place o f
public resort but onc e in every seven days an d th at
o n a S un d ay aft ernoon
h as no ri ght to suppose
because it is cro wd e d th en th at i t is as cro wde d
2
on a w eek d ay
He is h ere in no posit ion to not e
th e exceptions even sh oul d th ey occur
A nalo gy , unl ess conrme d by experiment o r upon
th e g roun d of resemblance es t ablish e d by a veri abl e
hypoth esis h as no claim t o be consi d ere d as a scien
t ic me th o d
T h ere may be false analo gies d epen d
in g upon surface resemblances A chil d mi ght
conclu d e th at o il woul d put out re because i t
so closely resembles w at er wh ich h e kno w s can ex
tin gu ish th e ames
T h e difference betw een essen
tial an d acci d ent al ag reement betw een ph enomena
,

'

N ov um O rg anon , i 105
.

Lo tze , Log i

p 343
.

TYPES

OF

IN DUCT IVE I NFERENCE

47

can be reveale d onl y wh en th e un derlyin g cause is


ascert aine d
T h e thir d meth o d alone h as sci ent ic w orth
True in duc tion mus t be a continue d searc h t o d is
cover a causal relation
3 T h e tw o rs t processes ful l th eir fun c t ions
larg ely as t ent at i v e an d su gg es tive me th o ds In
enum erat ion o f ins t ances, w e are oft en le d to no t e
resemblances wh ic h become th e basis of analo g y
A n d analo gy su gg es t s in turn, h ypo th esis wh ic h is
capable o f verication thr ou gh subsequent experi
ment
T h e ques tion may be put Wh ic h of th e th ree

proces ses is in duc t ion proper ?


T h e fac t is th at
i t may involve all th ree but i t is no t complet e until
i t reach es th e th ir d th e experiment al meth o d
A nalo gy is especially fert il e in su gg es t ion
S eien
tic min d s mos t caref ully t raine d an d verse d in
scient ic meth o d s o f research are oft en mos t keen
in no t in g resemblances , an d d et ec tin g analo g ies
wh ich become th e basis o f th eir experiment s N e w
t on possesse d th at rare insi ght wh ich in S pit e of
th e manifes t dissimilari ty of th e tw o p h enomena
coul d ye t discern an essential likeness be tw een th e
fall of an apple an d th e g ravit ating force o f th e
moon t o w ar d s th e ear th
4 I t is also t o be observe d th at th e c h oice o f
me th o d will d epen d lar g ely upon ment al h abi t
S ome min d s nat urally or by special t rainin g an d
surroun din g s are given t o experiment T h ey h ave
a t es tin g fac ility an d inventive capaci ty O th ers
naturally are suscep tible in an unusual d e gree t o
.

I NDU C T IVE

48

O GIC

con t ras t s and resemblances O th ers a gain are ac


cust ome d t o accurat e Observat ion th at is ever push
in g beyon d an d seekin g t o ext en d i t s sph ere T hus
w e h ave a natural division o f th ese meth o d s accord
in g to psych ical proclivi ties T h e ch oice o f m eth o d
is oft en cond itione d by th e force o f circums tances
A re all cro w s
Experiment is n o t al w ay possible
black
Th ere is n o connec tion betw een th e g eneral
or g anism of th e cro w an d its color th at h as thus far
been reveale d th rou gh analysis or experimen t Th e
only recourse is t o nu mber ins t ances over th e wi d es t
possible fiel d We say moreover th at Mars may be
inh abi t e d ; for i t h as an atmosph ere similar t o th e
earth an d th erefore capable o f sus t ainin g life
A nalo gy is th e only gui d e in such a case an d i t is
impossible t o verify it ei th er by observation or
experiment
5 All th e me th o d s t en d t o o n e en d th at of e f
fectin g a g eneralizat ion o f experience
T h e g en
e ralizatio n may be ei th er a numerically g eneral o n e,
or one expresse d in t erms of a g eneric concept
(1) Th e former consis t s in th e ext ension o f several
ins t ances t o th eir repeti tion un d er like condi tions
(2) Th e secon d consis t s in th e ext ension of several
ins t ances to all co gnat e species un d er th e same
g enus
E xampl es O f th ese tw o kin d s o f g eneralizat ion
are as follo ws : T h e g eneral proposi tion th at all
sulp hur is combus t ible is of th e former kin d ; all
inst ances are subst ant ially Of th e same nature, an d
d o not differ as dis tin guish able S pecies un d er th e
same g enus, but rath er a rep etition o f lik e ph enom
.

,
.

TYP ES

OF

I NDU C T IVE I NFERENCE

49

ena T h e g eneral concept in th e above proposi tion


i s of th e nature o f an inma sp ecies O n th e o th er
h and , th e proposi tion th at all m ammal s are vert e
brat es , h as th e subj ect term in form of a g eneric
concept Many species , differin g wi d ely among th em
selves, may be embrace d un d er it 1
.

Sig wart, Log ic , V o l

II

pp 310 , 3 11
.

CH AP T ER

CAU S ATI O N

W E h ave seen th at ind uc tion as a t ruly scient i c


meth o d consis t s in th e analy tical d et erminat ion o f
th e relations o f cause t o effec t in any complex ph e
n o men o n accompanie d by a g eneralizat ion of th e
resul t ob t aine d T h e nal outcome of such a proc
e s s is an un iversal concep t whic h embo d ies a la w
expresse d in t erms of a const ant connection betw een
ant ece d ent an d consequent A s G reen h as sai d
Th e essence of in duction consis t s in th e discovery
1
A causal vie w of
o f th e causes o f ph enome n a
th e universe gives rise t o lo gical concep t s wh ereas
a my th olo gical view o f th e universe as in ancient
times resul t e d in mere empirical concept s whic h
g ave no assurance ei th er o f s t abili ty or in v ari
ability
I t w ill be necessary th erefore t o d e
t ermine more precis ely th e lo g ical si g nicance o f
th e causal i d ea wh ich seems t o un d erlie all in duc
tive inference T his is no easy t ask A ccor din g to
Cliffor d cause h as S ixty four meanin g s in P lat o, an d
fortyei ght in A rist otle ?
,

Green Ph il Works V o l II p
2 C liffo rd L ec tu res an d E ss a s
y
1

50

284

V o l I p 149
.

CA

US AT IO N

51

T h e causal i d ea h as so m etimes found expression


in th e p h rase th e uniformity of nature, o r it is oft en
referre d t o as th e doc t rine o f universal causation
T h ese two ph ras es are O ft en use d int erc h an g eably ;
this g ives rise t o confusion of th ou ght, for th eir
meani n g s are quit e dis tinc t U niformi ty of nat ure,
s t rict ly int erpret e d means th at like ant eced ent s
und er precisely th e same cond itions w ill be fo l
lo w e d by like effec t s ; th is i d ea expresses one ph ase
O f causation, viz its invariabilit y T h e d oct rine of
universal causation h o w ever, expresses th e impo s
sibility of p h enomena risin g spont aneously , w i th ou t
an ant ece d ent or ant ece d en t s, sufcient rationally
to accoun t fo r th em
T h e tw o i d eas li e at th e root
A s Tennys o n h as put i t :
o f th e causal i d ea
,

Fo r n othing is th at

e rs

fro m Law

confusion h as also arisen from th e failure t o


d iscriminat e precisely betw een th e philosoph ical an d
th e purely lo gical ques tions relative t o th e g eneral
subj ec t of causat ion Causat ion may be vie w e d
from th ree dierent point s of view
1 Wh at it is p h enomenally, th at is , as re g ar d s
i t s physical aspect s
2 Wh at i t is essent ially as re g ar d s i t s real
nature T his is a me t ap hysical ques t ion
3 Wh at i t is in respec t t o i t s c h aract eris t ic
attribut e of invariabili ty T h is is a purely lo gical
ques tion
(1) A s t o th e rs t, wh at is causation ph enome
nally ? W h at is i t s purely physic al signicanc e ?
Inves t i g at ions in th is line h ave le d t o th e d oc t rine
S ome

N UC TI VE

52

I D

LO

GIC

th e conservat ion o f ener g y


T h is is substan
tially th e asser tion th at in every event no ne w
energy is calle d forth which di d not exist before
po t entially at leas t nor can any ener gy be ul ti
mat ely los t ; no thin g new is creat e d
th ere is only
a ch an g e o r transfer from one s t at e or con d ition t o
anoth er Moreover th e sum tot al of ener gy in th e
un iverse is a cons t ant quanti ty ; it can neith er be
add e d to , nor sub t rac t e d from T h ere is an excel
lent illus t ration Of this th eory in th e a dmirable

f
c h apt er on Conservation o Ener gy by P rofessor
Tai t
I g ive i t somewh at in full : I allo w an
elect ric current to pass throu gh a g alvanic batt ery
and th ere is fo r th e m oment a cer t ain quant ity of
zinc consume d or as w e may put i t a cert ain quan
tity of potential ener gy in th e batt ery h as been
convert e d in t o th e kinet ic ener gy of a current O f
elec t ricity T ha t current of elect ricity passes roun d
some yar d s of copper wire, coile d roun d a bar o f
iron or a number o f n e iron wires wh ich are s t an d
in g vertically insi d e this apparatus T h e moment
th e current passes th ese iron w ires are convert e d
into magnet s, but in consequence of th e conserv a
tion of energy while th is is g oing on th ey w eaken
th e current T h e current of elec trici ty becomes
w eaker in th e ac t of m akin g th e ma gnet but th e
moment th e mag net S prin g s into exis t ence i t a g ain
is w eakene d because from th e necessi t ies o f its
posit ion its m ere comin g in t o exis t ence necessit at es
th e passa g e of a new current of elec t ricit y in an
o th er coil Of w ire which surroun d s this ext ernally
an d nally this last current w e can u se t o pro d uce
of

CA U S ATION

53

h eat, or li ght, o r soun d 1 In th is cycle o f ch an g es


w e see h o w closely conn ec t e d even disparat e ph enom
ena are an d h o w th e appearance o f ener gy in any
o n e d e ni t e s t at e is d epen d en t upon i t s previous
exis t ence in some oth er s t at e T h e d oc t rine o f
conservat ion of ener gy w e s h all see lat er o n may
be su gg es t ive as to th e nature of th e analytical
t reat ment o f cause an d effect
(2) T h e ph ilosoph ical quest ion as t o th e inner
nature of causation met wi th o ne ans w er g enerally
unt il th e t ime of Hum e ; namely t h at th e i d ea of
cause si g ni e d th at th e ant ece d ent was e fficient in
pro ducin g th e correspondin g consequent, implyin g
th e t ransfer O f po w er sufcient to brin g about th e
effec t
Hume, h o w ever, cont en d e d th at in th e
g reat es t possible ex t ent of o ur kno wle dg e, all th at
w e cert ainly kno w is th is , th at one event follo w s
anoth er
We h ave no g roun d fo r an asser tion
concernin g th e manner in whic h th e sequence is
effect e d nor assume any real tie be tw een th em
Hume Insis t e d th at ph enomena w ere conj oine d, but
2
never connect e d
His O pponent s as K ant an d
o th ers , d eny h im, h o w ever h is fun d am ent al posi
t ion th at th e ori gin O f th e cau sal concept comes
from experience alone T h ey ur g e d th at it h as an
a p rio ri ori g in, a concep t S imple an d unanalyzable ,
g iven th rou gh in tui tive insi ght ; d evelope d in th e
sph ere of experience but not d epen den t upon ex pe
ri c u ce for it s w arrant
It is an int eres ting fac t
th at th e i d ea o f th e conservat ion of ener gy d evel
.

1
2

Tait

Recen t A dv an c es in

Hume , E ssay

on

dea

Ph ysi ca l S c ien ce

pp 7 6 , 7 7

f N e cessary Caus ation

INDU CTIVE

54

O GI C

ope d subsequent to Hume s t ime It seems to give


evi d ence which Hume insis t e d w as no t and coul d
not be forth comin g namely concernin g th e i dea of
th e ant ece d ent as an efcient po w er Th rou gh th e
mo d ern doct rine th e impression of a t ransfer of
real pow er is pro duce d, th ough its mo d e an d man
ner s t ill re main a mys t ery
(3 ) Th e lo gical aspect concerns not th e ph eno me
nal manifes tation of cause an d effect n o r th eir inner
nature but rath er th e element of invariability in
causat ion Two q u estio n s h ere su gg es t th emselves
First Is invariabili ty a fact, a cons t ant ele
ment in causation ? S econ d Ho w d o w e account
for its exis t ence
T h e rs t only h as t ruly lo g ical
si gnicance T h e invariability of causation, th at
like ant ece d ent s un der precisely th e same con di
tions pro duce like effec t s , alone makes in duc tion
possible Mill says th at it is th e belief in th e
uniformity of nature which s t an d s as th e ul timat e
major premise in every process o f in duc tion
Hume accept e d it an d base d inferences upon it,
an d never ch allen g e d it as a w orking basis as
re gar d s th e affairs of every d ay life He ack n o wl
e dg e d th e element of invariability an d only d enie d
th e bon d of connection T his element h as peculiar
lo g ical si gni cance : wi th out it, it w oul d be im po s
sible t o ext en d our kno wle dg e beyon d th e seen an d
th e h ear d in d ee d th at wh ich is seen an d h ear d
woul d th en h ave no meanin g, an d n o basis for th eir
int erpret ation an d appreciat ion Being assume d
h o w ever in our lo gical pos tulat e, w e h ave a basis
for in duction, a cons t ant t o be sought for, an d t o

C AUSA T IO N

55

be d epen d e d upon, in explanation of th e pas t an d in


pre diction of th e fut ure
Wh en w e come to th e secon d quest ion whic h is
essentially a g enetic o n e, h o w th e belief in th e uni
formi ty of nature arose w e nd two classes which
ans w er respec tively th at th e belief arose a p riori,
an d on th e o th er h an d from experience simply
T h e former is th e O pinion especially associat e d w i th
th e S co ttis h S ch ool o f philosophy Hume h ol d s
th at i t procee d s from a psych olo gical law O f cus tom
or h abi t, an un broken line of ment al associations
in ducin g a belief w ith in, concerning th e uniformi ty
o f nature w i th out
Mill h as also a like empirical
basi s fo r a belief in th e uniformity of nature ; h e
h ol d s th at h aving observe d uniformity in many
experiences , in fac t never cont ra dict ed, w e g eneral
ize so as to cover a sp h ere beyon d o u r experience
Moreover w e possess th e consensus of t es timony ,
coex t ensive wi th th e his t ory of hum anity , of th e
in d enit ely wi d e ext ent of th e sph ere of causation
an d th e accompanyin g ch aract eris tic of uniformi ty
His posit ion is fortie d by th e fac t th at in th e
process of incomplet e in duc t ion its probabili ty is
s t rength ene d wh ere th ere h as been excep tionally
abun d an t scope for observat ion, so that th ere is th e
overwh el min g conviction th at if th ere h a d been a
time or pl ac e wh ere th e law w oul d prove unt rue it
w oul d h ave been notice d Ins t ea d o f universal
causation Mill an d h is follo w ers make a more
cautious s t at ement ; causation as coext ensive
T h is is
w i th th e sum t o t al of h uman experience
abun d antly a d equat e t o embrace all possible cir
.

,
.

IN DU CT IVE

56

mstances

OGIC

pract ical inference Th e immensely


hi gh d e gree o f probability eng end ers a subj ec t ive
certi t ud e whic h in every day conduc t of affairs, an d
e v en in th e more exact requirement s of scientic
inves ti g at ion is never ques t ione d
Preyer h as g iven an int erestin g account o f th e
ex t remely early appearance o f th e appreciation o f
th e causal relat ion in th e case of his chil d who at
th e th ree h un dre d ninet eenth day o f its life, s t ruck
several t imes with a spoon upon a plat e I t h ap
pene d acci d ent ally while h e was doin g this th at h e
t ouch e d th e plat e w i th th e h and th at w as free ; th e
sound w as dulle d an d th e Chil d notice d th e differ
ence He no w t ook th e spoon in th e o th er h an d
s t ruck wi th it on th e plat e an d dulle d th e soun d
a g ain an d S o o n In th e evenin g th e experiment
w as rene w e d wi th a like resul t Evi d ently th e
func tion o f causalit y h a d emer g e d in some s t ren gth
for i t prompt e d th e experiment T h e cause o f th e
dullin g of th e soun d by th e h an d w as it in
th e h and or in th e plat e
T h e o th er h an d h a d
th e same dullin g effect s o th e cause w as n ot lo dg e d
with th e one h an d P rett y nearly in this fash ion
th e chil d mus t h ave int erpre t e d h is sound impres
sion an d this at a time wh en h e di d no t kno w a
S in gle word o f h is lat er lan g ua g e 1
T h e th eoret ical soun dnes s of Mill s speculat ions ,
h o wever h as a a w al th ou gh th e prac tical resul t s
may not be th ereby invali d at e d T h e ind uctive proc
e ss wh ic h is suppose d t o be a t ruly scienti c me th o d
and superior to in duction by simple enumeration
cu

of

Preye r, The S e nses

and th e

Will

pp 87 , 88
.

CA US AT ION

57

mus t accor din g t o Mill at th e las t anal ysis re s t


upon a principle wh ich is it self base d upon an in
complet e in duc tion A very fair an d searchin g ori t i
l
cism of Mill is th at of V enn s in h is E mp irical Lo ic
g
W h at ely insis t s th at th e wh ole ques tion concernin g
th e nature of o ur belief in uni formi ty is irrelevant,
as i t is a purely psyc h olo g ical an d n o t a lo gical
one Mansel h ol d s a me diatin g posit ion in insis t
in g th at th e i d ea of un iversal causat ion is intu i t ive
while th at o f uniformi ty is necessarily empirical
S i g war t h as very t renc h ant ly cri t icise d Mill in th at

t akin g aw ay wi th one h an d wh at h e gives wi th


th e oth er h e sh o w s in th e uncert ainty o f his
vie w s th e h elplessness of pure empiricism, th e im
possibility Of erec tin g an e dice of universal propo
sitio n s on th e san d h eap of S h iftin g an d isolat e d
fact s or more ac curat ely sensations ; th e eu
d eav o r t o ext r ac t any necessit y from a mere sum
of fact s mus t be fruit less T h e only t rue point in
th e wh ole t reat ment is o ne in whic h Mill as a lo g i
o ian g e t s th e be tt er o f M ill as an empiricis t ; n am ely
th at every ind uctive inference cont ain s a universal
principle ; th at if it is t o be an inference an d n ot
merely an association of onl y subj ec t ive vali dity
th e t ransi tion from th e empirically universal ju dg
ment al l kno wn A s are B t o th e uncondi tionally
univers al all th at is A is B can only be ma d e by
means o f a u niversal maj or premise an d th at o nly
upon con di t ion of th is bein g t rue are w e jus tie d

in inferrin g from th e particular kno wn A s t o th e


2
s till unkno wn A s
,

,
.

Venn

E mp iri

cal Lo g ic

p 130
.

S ig wart, Log i

V o l II p 30 3
.

INDUCT IVE

58

OGI C

T h e wh ole t en d ency of th e mo dern lo gic is to


base th e causal pos t ulat e upon a g roun d wh ich is
epist emolo g ical ; namely inasmuc h as our kno wl
e dg e is one an d self consist ent th rou gh out all i t s
separat e element s , th ere mus t be a correspon din g
invariability in th e ph enomena th emselves as th ere
is in th e sys t em of kno wle dg e whic h result s from
th e i nt erpret ation of th ese ph enomena
T his is
th e g eneral view o f S i gw art, Bosanque t, L ot ze and
l
Green
T his vie w may be consi d ere d also as an expres
sion of th e L aw o f S ufcient R eason ; namely th at
th ere is an inh erent ch arac t eris tic of int elli g ence
which d eman d s th at every ele m ent of conscious
ness must be referre d t o some oth er element for i t s
explanation, an d th at it is only wh en th e lo gical
connect ion of i d eas correspon d s t o a real causal
connec tion th at th e min d discovers a reason for
its several experiences wh ich is satisfyin g
I t h as
been sai d by U eberwe g as given expression t o th is
vie w : T h e ex t ernal invariable connection amon g
sense ph enomena is, w ith lo gical correc tness ex
plaine d by an inner conf ormabili ty t o la w, accor d
ing t o th e analo gy of th e causal connec t ion perceive d
in ourselves betw een volition and it s ac tual acco m
,

plish m e nt

T h ere is a dist inction th at is of import ance to


not e between th e popular and th e scient ic i d ea of

S ig wart , Log i , V o l II pp 119 , 120 ; B o s an uet , Lo g i , V o l


II pp 220 , 22 1 ; Lo tz e , L og i , p 68 ;
reen , P h il Work s , V o l II
p 286
1

Ueb erweg , Log i

pp 2 81, 282
.

C AUS AT IO N

59

cause T h e former is th e out come o f th e supposi


t ion th at wh at ever imme d iat ely prece d es th e effect
h as evi d ent ly pro duce d i t an d th at this is sufficien t
wh olly t o account for i t S uch an i d ea of causes
lea d s, at th e best but to a loose an d su percial
de t ermination of th e relation be tw een any ant e
ce d ent an d its consequent, an d th ere is th e d an g er
moreover of a h as ty inference wh ich resul ts in th e
fallacy o f p o st ho c ergo p rop ter hoc In or der t o
att ain a t rue vie w of causation w e mus t especially
att en d t o th e ext reme complexi ty o f th e causal con
n ectio n
Th ere is no suc h th in g as a simple cause
follo w e d by a simple effect Th e cause is al ways a
combination of several element s circum s t ances and
cond i t ions ; th e effec t is alw ays manifol d This
c h aract eris tic h as been ad mirably presen te d in
Mill s ch ap t er on th e P lurality of Causes an d th e
Int ermixture o f Effect s I t is w ell kno wn th at th e
variat ion in th e h ei ght of a baromet er is due par tly
t o th e variation of th e atmosph eric pressure an d
partly t o th e variation of th e expansion of th e mer
curial column due t o h eat In exac t d et ermination
some experiment or calcul ation mus t prece d e before
th ere can be a discrimination betw een th e element s
A n d s o also a num ber o f cir
o f th e j oint effec t
cu m stan ce s may combine t o res t ore an invali d t o
h ealth no o n e o f wh ic h alone being capable o f
e ffectin g h is recovery
T h e cause of any ph enomenon h as been d en e d
by Mill as also by Bro wn an d Hersch el as th e su m
T his s t at ement h as
t o t al of al l i t s ant ece d ent s
been cri ticise d inasmuch as th e sum to t al o f all

I NDU C TIVE

60

O GIC

ant eced ent s is in d et erminat e an d th a t th ere is no


end t o th e possible ramicat ions in all direc tions
wh ich an exh aus tive analysis o f any complex cause
will yiel d Ho w ever th e problem is one of re due
tion t o simples t possible t erms with in th e ran g e o f
T h ere
o ur po w ers o f observat ion an d experiment
is muc h in th e sum t ot al of all th e ant ece d ent s o f
any g iven effec t wh ic h is irrelevant
I t is th e
peculiar func tion of lo g ical analysis t o d iscriminat e
betw een th e relevant an d irrelevant T h e t emper
ature o f th e laborat ory w ill no t affec t one w ay o r
th e oth er experiment s with fallin g bo dies ; but w ill
essentially inuence cert ain ch emical experiment s
an d mus t ent er as one o f th e d et erminin g fact ors in
th e s um t o tal of an t ece d en t s
I t may be th at cer
t ain element s o f a complex wh ole may seem t o us
ul timate an d unanalyzable, an d yet be th emselves
sys t ems of more or less complexi ty
T h ere is
al w ays a li m it t o analysis , both experiment al an d
ment al T h e analysis is t o ext en d t o th e ult imat e
part s as far as possible I t is no t an exact process,
but a process whic h t en d s t o exactnes s to th e ex
t ent wh ich th e scope o f nit e int elli g ence w ill per
mi t Th e reason is not at faul t s o much as th e
natural limit ations o f observation an d experiment al
analysis T h e en d of our researc h in causal an aly
sis is t o discover an invariable relat ion th at c an be
expresse d in th e form of an h ypo th e tical universal ,
If A th en B
In or d er t o effect th is th e complex A mus t be
separat e d in t o its part s a b c e t c and th e e ffe c
tive, an d necessary, an d in dispensable element pro
,

C AUS AT IO N
d u c in g B

61

mus t be d et ermine d S uppose i t prove s


t o be a i t may be possible t o subj ec t th is t o furth er
analysis an d re duce d t o simpler elemen t s suc h as
at y 2 etc
an d :1: foun d t o be th e si gnicant ele
men t o f th e real cause E ac h an al ysis d et ermines
a narro w er an d s till narro wer sph ere wi thin wh ich
th e cause lies A man is s h ot We say th e bul let
kille d h im ; th en th e drivin g force beh in d th e bul
le t ; th en th e explosive po w er O f th e gunpo wd er ;
th is in t urn w as occasione d by th e combine d ch emi
cal an d mec h anic al energy of its in g re dient s , wh ere
by a soli d is t ransforme d int o a g aseous subst ance
many times it s ori g inal bulk
S ooner or lat er w e mus t reac h th e en d of our an
alysis an d th e inves t i g at ion be necessarily ch ecke d
N o explanat ion is ul t ima e ; w e onl y tr an sfer our
po int of vie w from a less to a more familiar sph ere
o f int erpre t at ion
We d o not feel th e nee d of ex
plainin g th e very familiar ; th ou gh th e mos t famil
iar is h ar d es t satisfact orily t o explain because th ere
is no th in g simpler in wh ose t erms w e may para
ph rase it We feel this in g ivin g a d eni tion of
t erms wh ose meanin g w e bes t kno w an d whic h w e
mos t frequen tly use Mr Barrett a former as sis t
an t at th e R oyal Ins ti tut ion sai d of F ara d ay : I
w ell remember one d ay w h en Mr F arad ay w as by
my si d e I h appene d t o be s t eadyin g by means of
a mag net th e mot ion o f a ma gne tic nee dl e un d er a
g las s sh a de Mr F ara d ay su dd enl y looke d mos t
impressively an d earnes tly, as h e sai d : Ho w wo n
d erfu l an d mys t erious is th at po w er yo u h ave th ere
T h e more I th ink over it, th e less I seem t o kno w
.

INDUCT IVE

62

O GI C

A n d yet,

h e wh o sai d this kne w more o f it th an any


1
living man
A l th ou gh our kno wle dg e is limi t e d as in all cases
of causation h o w ever simple neverth eless as far
as i t g oes , th e several element s are relat e d lo g ically,
th at is, necessarily an d universally We may only
kno w in part, but s till w e kno w an d th e w orl d, as
int erpret e d for u s in kno wle dg e is a w orl d of
invariable sequences
T h e process o f in ductive
analysis th erefore, consis t s in re ducin g a complex
ant ece d ent t o it s ul timat e part s in or d er t o reveal
th e element o r element s in it which h ave cause d
th e given effect I t somet imes h appens th at differ
ent element s in an ant ece d ent may be re gar d e d
severally as th e cause accor din g to th e p syc h o l o gi
cal point of vie w as re gar d s th e int eres t s O f th e
inves ti gat or
It is no t al w ays th at a scientic
d et erminat ion Of th e cause i s require d ; it may be
th at all th at is d esire d is a kno wle dg e of th at part
of th e ant ece d ent whic h is mos t closely an d prom
in ently connec t e d w i th th e event in ques t ion
An
inquiry may be s tart e d in reference t o th e cause of
an epi demic in a community O ne may discover
th e cause in th e carelessness o f S anit ary en g ineers ;
anoth er may s ay th e cause lies in th e poor cons t ruo
tion of th e sew era g e ; ano th er says th at th e cause
of th e epi d emic is a cert ain kin d of bacilli E ach
o n e is lookin g at th e c h ain of e v ent s relat e d as
cause an d effec t ; but th ey all look at d i erent links
of th e same ch ain O ne element th erefore o f a
complex ant ece d ent may be brou ght in t o more or
.

Glad sto ne

Mi ch ael F arad ay

p 180
.

63

CA US ATIO N

less prominence as
efcient element o f th e
cause, accor din g as th e point o f vie w is s hift e d
If, in th e search for th e cause o f ph enomena th e
sum t o t al o f an t ece d en t s w ere al w ays g iven exh au s
tiv ely th e explanat ion mi ght become so loa d e d
d o wn w ith d e t ails as t o bur d en th e min d an d con
fuse, rath er th an clear, th e un d erst andi n g
.

CHA P TER
THE

ME THO D

OF

VI

CAU S AL AN ALY S I S
M I N A TI

DETER

A ND

ON

I T w ill be w ell t o consi d er th e various problems


wh ich will confront u s in seek in g t o analyze a com
plex ant ece dent for th e purpose o f discoverin g i t s
cause
1 T h ere are problems wh ere cause an d e ect
appear in evi d ent sequence T h ere is an antece
dent whic h is follo w e d by a consequent If A
h appens , th en B w ill h appen Ins t ances of th is
kind mos t readily yiel d th emselves to th e process
of analysis, because a ch an g e in any given ph e
n o meno n i s occasione d by th e efciency o f th e an t e
ce dent which is observe d in connect ion w ith th e
ch an g e i t self I t is easier to no t e active th an
passive relations , th e dynamic rath er th an th e
s t atic T h e att ention is att rac te d an d h el d by
ch an g e T h e bird yin g across o u r path is o h
serve d, an d th e one perch e d upon th e t ree near
at h an d, h o wever con spicuous may be i t s posi tion ,
is passe d by wi th ou t any n o tice t aken o f i t I t is
easier to connec t th e mois ture o f th e grass w ith
fallin g rain, th an wh en th e same is occasione d by
th e d ew In one case, th e causal relation is ex
.

64

C A US AL

T RMIN AT IO N

AN ALY SIS

65

AND DE E

hibited in

operat ion ; in th e oth er, th e connec t ion


is veile d We n d th e g rass w et ; wh at prece d e d
i t w e are n ot able t o see T h ere are several in
s t ances O f sequence amon g O bserve d ph enomena
wh ic h mus t be carefully discriminat e d in or d er t o
avoi d confusion of th ou ght T h ey are as follow s :
(1) Wh en w e h ave A follo we d by B an d A
ceases wh olly w hil e B endures fo r an appreciable
time aft erw ar d s , o r i t may be permanent ly A
billiar d ball s t rikes anoth er th e secon d g oes on
by virtue of th e ne wly acquire d en ergy t ransferre d
by impac t from th e rs t which h o w ever s t ops
al t o g eth er I th ro w a ball wh ich lo dg es o n th e
t op o f a buil din g ; th e effec t pro duce d las t s per
m an en tly, for th e b all h as g aine d a g ravi ty po t en
t ial due t o th e ener gy impart e d to it by th e init ial
th ro w in g
T h e O l d formula, th erefore d oes not
al w ays h ol d : Cessant e causa cess at effectus
(2) Cases wh ere A ceases an d th ereupon B
imme d iat ely ceases also If w e cut o ff th e supply
of g as wh ich feed s a ame, th e ame at once dis
appears Th ere are cases, h o w ever wh en an ap
prec iabl e t ime mus t elapse in or d er th at th e
t ransferre d ener gy in th e effect may be dissi
pat e d Wh en w e s hu t our eyes th e s t imulus caus
in g th e percept ion i s cut Off an d th e percep tion
at once i s at an en d ; h o w ever, th ere are cases
wh ere th e s timulus being very s t ron g, aft er ima g es
are in duce d wh ic h remain fo r some time in th e
d ark el d aft er th e eyes are close d
(3 ) Cases wh ere th e ant ece d en t is wh olly in
a d equat e t o pro duce th e effect but wh ose func tion
.

'

I NDU C TIVE

66

OGIC

merely t o liberat e po t ential ener gy alrea dy


s t ore d, an d w ait in g an occasion for its ac t ive
manifes t ation A sli ght blo w upon a piece of
dynamit e causes an explosion wh olly dis pro por
A S is
tion ate to th e s t rikin g force employe d
w ell kno wn h eat is Oft en an exci ting cause of
ch emical act ion In such cases th e real cause is
more o r less conceale d, wh ile that whic h is appar
ent upon th e surface is no t a cause s o much as an
occasion o f th e ph enomenon in ques tion I t ouc h
th e pen dulum an d a clock s tarts an d so continues for
many hours ; th e s win gin g pen dulum , h o wever, is
only th e occasion of liberatin g th e po t ential ener gy o f
th e woun d u p sprin g, an d th ence th e po w er wh ich
runs th e clock pen dulum , wh eels , h an d s an d all
2 We h ave also ins t ances no t so muc h o f se
Th e planet s revolve
qu ence as of concurrence
aroun d th e cent ral su n ; h ere th e cause is cons t ant
att en d e d by const ant effect T h e machine never
runs d o wn nor h as t o be woun d up, so th at it can
be seen th at th e cause ant e dat es th e effect
3 Cases of Coexis tence Th ese are more d ii
Cult t o analyze, for th e ph enomena d o no t h ere
appear as ant ece d ent an d consequent in th e mi d s t
of ch an gin g con ditions an d circums t ances
We
h ave coexis t ence o f two kin d s
(1) Coexis tin g att ribut es in o ne an d th e same
or g ani sm T h ey are al w ays foun d t o g eth er T h ey
form o ne g eneric concep t an d are calle d by one
name Co w s h ave h orns cloven feet are rumi
nant, et c Do g s h ave th eir dis tinct an d cons t an t
ch arac t eris tics T h e oran g e h as its correlat ion o f

is

CA U S AL

N ALY S IS

AND DET ERMIN A

T IO N

67

color, t ast e, smell A nd so w e h ave th e so calle d

ie
nat ural kin d s
or g anisms present in g an
unique an d c h aract eris tic appearance, d ierenti
at e d th ereby from all oth ers
T h ere are also
cer tain correlations of gro wth wh ic h present a
cons t an t relation be tw een cert ain att ribut es, as
th e fac t h o w ever w e may explain i t, th at cat s
w i th bl u e eyes are invariably deaf
Th ere are,
moreover, illus t rations O f th e same in an inor g anic
sp h ere, as th e law wh ich connect s th e at omic
w ei ght of subs t ances an d th eir specic h eat by
an inverse proportion ; o r th at o th er law wh ich
O bt ains be tw een th e speci c g ravity of subst ances
in th e g aseous s t at e, an d th eir atomic w ei ght s
th ey bein g eith er equal o r th e one a mul tiple of
th e o th er In many cases , th e bare fact o f co
exis t ence mus t be accep t e d with out bein g able t o
explain th e causal g roun d of i t T h e several ele
men t s present a cons t ant association an d th at is
all th at can be sai d about i t
In oth er cases ,
h o w ever, a cause may be foun d as re g ar d s for
inst an ce, th e corr
elation of warm bloo d e d animals
al w ays possessin g lun g s Th e connec tion betw een
respiration an d th e g enerat ion o f h eat is foun d t o
d epen d upon ch emical ac tion as its causal basis
(2 ) A relat ion of st atics rath er th an dynamics , as ,
for in s t ance, a pillar supportin g a roof o r arch , is
sai d t o be th e cause in th e sense of th e sus t ainin g
caus e o f th e supers t ructure S o also th e coh esive
force wh ic h h ol d s t o g eth er th e par ticles of a st one
In suc h cases th e ener gy inh erent in th e cause is of
th e nature of a s t ress an d s t rain
-

INDUCT IV E

68

O G IC

U n d er

th is h ead are embrace d th e ph enom


ena of vi t al g rowth or d evelopment T h ese are th e
mos t difcul t o f all th e causal problems t o d et er
mine ; for i t is require d t o discover th e inner n eces
sity of essence , an d h o w th e succee din g s t a g es o f
d e v elopment unfol d th rou gh th e play of th e central
forces inh erent in th e very nature an d bein g of th e
or g anism i t self Mill is cont ent with classifyin g
org anisms as different natural kin ds an d h e is no t
concerne d wi th th e reason wh y th ere s h oul d be
such and such kin d s no r does h e att empt t o discover
any law concernin g th ese natural correlations an d th e
mo d e of th eir g ro wth In in ductive analysis our
concept s mus t n o t merely g rasp wh at th e natural
kin d s are but also wh at h as d et ermine d th em t o be
wh at th ey are Darwin put s special emph asis upon
th e environment as affecting Ch an g es in or g anisms
an d pro ducin g different iatin g mo dicat ions amon g
species This h o w ever mus t be consi d ere d no t as
sole fact or but o n e wh ich is combine d with inner
nee d s an d necessi ties Moreover Darwin h as drawn
att ention t o th e fac t th at in divi dual differences nee d
scient ic explanation as w ell as th e common attri
but es as for ins t ance why some S h eep are black ,
an d why some pi g eons are fan t aile d an d oth ers are
not In all suc h consi d erations w e mus t no t lose
si ght O f th e fac t th at th ere are tw o d et erminin g
fac t ors
th e inner necessi ty Of development ; an d
th e ext ernal necessity of causalit y, as or g anisms
are act e d upon by th eir env ironment 1
5 Cases of collocat ion wh ere n o o n e elemen t of
4

S ig wart , Log i

V o l II pp 322 , 3 30 , 331
.

C A US AL A N ALY SIS

T RMINA T ION

69

AND DE E

th e cause is efficient, but all t o g eth er th ey combine


t o pro duce th e effec t In search in g for th e cause,
w e mus t n o t only nd a cert ain amount o f ener gy
capable of pro ducin g th e effect, but w e mus t also
di scover wh at peculiar arran g ement o f th e element s
concerne d mus t exis t before th e ener gy in ques tion
can become operative
Ch almers says th at th e
exi s tin g collocations of th e mat erial worl d are as
impor t an t as th e law s wh ich th e obj ect s obey th at
many overlook this dis tinct ion an d for g et th at
mere law s wi th out collocations w oul d h ave ao rd ed
no securi ty ag ains t a turbi d an d disor derly Ch aos 1
We w oul d naturally say th at th e sole cause of w at er
boilin g at 2 12 is th e envelopin g h eat ; i t h as h o w
ever, been observe d th at o n top of Mont Blanc
Th is in dicat es
w at er boils at 180 inst ea d of
th at, in a ddition t o t h e re , th e atmosph eric press
ure is an element in th e cause very easily over
looke d Ch arcoal an d diamon d are of th e same
subs t ance ; a di fference only in th e arran g ement of
th e molecules resul t s in such ra dically different
combinations T h ere are in th e main th ree special
kin d s of coll ocations, as follo w s :
( 1) Cases O f mo d ifyin g circums t ance A s t ron g
win d blo w s d o wn a t ree ; this w oul d no t h ave O C
Curre d h a d n o t th e t ree been h ollo w T h e h ollo w
ness o f th e t ree is h ere a c o O perat ive circums t ance
th at is combine d w ith th e efcient cause
th e force
An ins t ance wh ere arran g ement of
o f th e w in d
th e element s concerne d rath er th an th eir efcient en
e rgies is pro d uc tive o f th e eec t, is th at of c apil
.

Q u o te d by

Jevons P
,

rin

cip les of S cience

p 7 40
.

INDUCT IVE

70

O GIC

th e risin g of liqui d in a tube O f excee din gly


small bore Here form is more essential t o th e
effect th an th e expen diture o f any visible ener gy
(2) Cases in wh ic h cert ain ne g ative con dit ions
prevent th e realization of th e effec t T h e plant s
an d sh rubs die in a lon g d routh because it di d no t
rain A train colli d es wi th anoth er, because th e
re d si gnal w as no t expose d as it Sh oul d h ave been
A mat c h will i gni t e gunpo wd er g enerally , bu t i t
fails to d o S O S h oul d th e po wd er prove t o be w e t
(3 ) Th ere are also cases of count eractin g causes
wh ere th e effect of cause A is not realize d as cause
B neut ralizes th e force of cause A ; as wh en an
anch ore d boat will n o t respon d t o th e pull O f th e
o ar
S ome times th e cause is not wh olly count er
act e d o r i t may be th e count eractin g cause more
th an h ol d s th e posi tive cau se in c h eck an d is it self
O perat ive T h e rise O f a balloon in th e air is due
to th e fact th at th e force o f g ravi ty is more th an
overbalance d by th e expansive force of th e g as
within th e balloon ; one force pullin g d o wnw ar d s
th e oth er bearin g up, an d th e latt er prevailin g
Mech anical forces ac tin g in combination a d mit
o f a resolut ion of th eir joint effec t accor d in g t o
th e th eory o f th e parallelo g ram of forces
Ch emi
c al an d vit al forces canno t be t reat e d in suc h a
w ay at all F rom th e ch aract er of th e element ary
forces in mech anics, o n e can calcul at e th eir c o m
binat ion In c h emis t ry h o w ever, wh en th e ele
ment s are given, th e resul tin g co m poun d canno t
be thus d et ermine d S O, also in vi t al an d ment al
ph enomena, th e necessarily complex nature of th e
larity,

C AUS AL AN ALY SIS

AND DE ER MI NA

T ION

71

element s involve d prevent s n ot only pred ic tion of


resul t in g combinat ions , but even a dequat e explana
t ion o f th at wh ic h may be imme diat ely g iven in
consciousness
6 I t is necessary, in th e in ves t i gat ion o f causal
relations , t o un ders t an d th e different mo d es Of th e
t ransfer o f ener gy wh ich are as follo w s :
( 1) Molar or m ech anical as in th e case o f a
bil liar d ball t ransferring it s energy t o anoth er
th rough impac t
(2) Molecul ar as h eat, ch emical an d elect rical
O ne passes int o
an d ma gne t ic forces li ght et c
anoth er, as ch emical force pro ducin g elec t ric elec
t ric pro ducin g mag netic o r pro ducing h eat an d
li ght
(3 ) Cases wh ere mech anical force becomes mo
lecul ar as frict ion in ducin g h eat ; o r cases wh ere
molecular beco m es mech anical as h eat t ransferre d
int o th e drivin g po w er o f an en gine o r elect ricity
applie d as a mot or A precise d et erminat ion of
equivalent s can be mad e betw een molar an d mo l co
ular energy ; as , fo r example it h as been foun d
th at i t t akes th e same amount o f ener gy to raise
7 7 2 poun d s a di s t ance o f o ne foo t th at it d oes t o
raise th e t em perature of one pound O f w at er 1 F
or th e h eat requisit e t o boil a g allon o f freezin g
w at er w oul d lift
poun d s th rou gh a dis
t ance o f o ne foo t
A s a consequence o f th e d oc t rine of th e t ransfer
o f ener gy a causal law can be so s t at e d as t o ex
press th e fac t th at variations in th e ant ece d ent s
will call for correspon din g variat ions in th e effect,
.

I NDU CTIVE

72

LO

G IC

inst ance such a law as th e follo win g : Re


s istan c e in a w ire o f cons t ant sec t ion an d mat erial
is d irectly proport ional t o th e len gth an d inversely
proportional t o th e area o f th e cros s section 1
T h e ne glect o f quant i t at ive det erminat ion of th e
proportionat e variations o f th e ant ece d ent an d c on se
quent w as a g laring d efect in th e in ductive sys t ems
bo th o f Mill an d of Bacon
T h rou gh th e represent ation o f th e various s t ag es
o f suc h variat ion i t is also possible t o es t ablis h th e
upper an d lo w er limit s beyon d which th e cause d oes
n o t pro d uce th e correspon d in g effec t ; as in Weber s
law concernin g th e relat ion of stimulus t o sens a
tion, th at s timulus mus t increase g eome t rically in
or der th at th e sens at ions increase arith metically
Th ere is an upper and lo w er limi t beyon d wh ic h
this proportion does no t h ol d
Th e d octrine o f conservation o f en ergy creat es
th e impression of con tinuous ch an g e in causation
in which th e effect unfol d s ou t o f th e cause We
d o no t th ink of ph enomena un d er th is aspect as
discre t e events More th an ever in th e li ght o f
mo d ern science d oes th e ol d sayin g ob t ain, N at ura

n o n faci t saltum
We n o lon g er look fo r cat as
t roph ic resul t s in nature but re gar d causation
as a con t inuous t ransfer of po t ent ial ener g y in t o
kinetic or ac tual ener gy
We come now t o th e consi d erat ion o f th e meth o d
by wh ic h th e causal analysis is me diat e d
This
is effec t e d th rou gh O bservat ion an d experimen t
O bservation is somethin g more th an mere lookin g at
as,

fo r

enkin ,

c c

E le tri ity

an d

Mag netism ,

p 83
.

C AU S AL AN ALYS IS

AND DETERMI NATIO

73

ph enomena ; i t means concentration of att ention for


th e purpose of researc h ; it means discriminating
insi ght an appreciati on o f likeness an d di fference ;
it means a pene t ration beneath surface appear
an ce s an d an appreh ension of th e essen t ial features
E xperiment consis t s
o f th e obj ec t s of percept ion
in mo difyin g th e element s wh ic h form th e complex
ant ece d ent in or d er to Observe th e resul t an t effect
upon th e correspon din g consequent F orces may
be a dd e d o r sub t rac t e d ; th eir in tensity may be
varie d increase d or d ecrease d ; th e circums t ances
o r con di t ions may be al t ere d
Hersc h el speaks of
observation an d experiment as passive an d active
observat ion
Wh en w e int erfere t o Ch an g e th e
course of nature or t o brin g natural forces wi thin
th e ran g e of o u r observation w e are experimentin g
Observation is preliminary t o experiment an d su g
g es t s th e lines alon g wh ich experiment S h o ul d pro
A n obser v ation th at sees th e par t s in th e
o ec d
wh ole an d th e wh ole in th e part s, is in it self an
analysis of a p h enomenon in course of which proc
T h e scien
ess cau sal rela tions mus t be d isclose d
tic spiri t d eman d s absolu t e veraci ty in observat ion
O ne ou ght not t o be blin d t o fact s even th ou gh th ey
t en d t o cont ra dic t preconceive d th eories Bacon
h as observe d th at men mark wh en th ey h it never

mark wh en th ey miss
We mus t s t rive a g ains t a
natural t en d ency to see thin g s as w e w oul d h ave
th em an d not as th ey s t rictly are
We mus t also carefully dis tin guish bet ween oh
serve d fact s and inferences which w e inst inctively
d raw from th ese fac t s O bservation is preliminary
,

INDU CTIVE

74

O GI C

t o an in duct ive inference th erefore i t mus t no t it


self involve an inference or w e s h oul d be ar guin g
in a circle A n int eres t in g illus t ration of th e dif
ference be tw een O bservation an d inference base d
upon i t, is narrat e d in th e life of F arad ay : An
artis t was once maint ainin g th at in natural appear
auces an d in pictures , up an d d o wn, an d hi gh an d
lo w, w ere xe d in dubitable realities ; but F ara d ay
t ol d h im th at th ey w ere merely conventional acc e p
tatio ns base d on s t and ar d s O ft en arbi t rary
The
d isput ant coul d no t be convince d th at i deas wh ich
h e ha d h i th erto never d oubt ed h a d such s h ift in g
foun d ations Well , sai d F arad ay h ol d a w alkin g
s tick bet ween your chin an d great to e ; look alon g

i t an d say wh ich is th e upper end Th e experiment


w as t rie d an d th e art is t foun d h is i d ea Of perspec
tive at complet e variance with h is sense o f reality ;
eith er end o f th e s tick mi ght be calle d upper
pict orially i t w as o n e physically i t w as th e o th er 1
T his in dicat es h o w readily o u r inferences an d
observations blen d, an d h o w difcul t it is t o separat e
th em in consciousness De Mor g an h as poin t e d
o u t th at th ere are four w ays of o n e event seemin
g
to follo w ano th er or t o be connec t e d wi th i t, w i th
out really bein g s o :
(1) Ins t ea d of A causin g B o ur perception of A
may cause B A man d ies on a cert ain d ay wh ic h
h e h as al ways re g ar d e d as his las t th rou gh h is o wn
fears concernin g it
(2) T h e event A may make ou r percept ion o f
B follo w wh ic h woul d oth er wise h appen w i th out
,

Gl ad stone

Mich ael F arad ay,

pp 165 , 166
.

CA US AL AN ALYSIS

T RMIN AT IO N

75

A ND DE E

bein g perceive d I t w as th ou ght th at more comet s


appeare d in h o t th an col d sum mers ; no ac coun t
h o w ever, w as t aken of th e fact th at h o t summ ers
w oul d be comparatively cloudl ess, an d affor d be tte r
oppo rtunities fo r th e di scovery of comet s
O ur percep tion of A may make ou r percep
tion of B follo w This is illus t rat e d by th e fall ac y
of th e moon s in uence in th e di ssipation o f cloud s
W h en th e sky is d ens ely cloud e d, th e moon w oul d
not be vi sible at al l ; i t w oul d be necessary for us
to see th e full moon in or d er th at o ur att ent ion
s h ou l d be s t ron gly draw n t o th e fact an d th is w oul d
h appen mos t oft en on th ose nigh ts wh en th e sky is
cloud l es s
(4) B is really th e an t ece d ent event, bu t our
perception O f A , whi ch is a consequence of B, may
be necessary to brin g about our percep tion of B
U p ward an d d o wn w ar d curren t s are contin u ally cir
c ul ating in th e lo w es t s t ratum of th e atmosph ere ;
but th ere is no evi d ence of this, un t il w e perceive
cum ulous clouds , wh ich are th e consequence of such
current s 1
T h ere are certain natural limit ations to o bser
vation, as thi n g s to o minut e t o be seen to o s wift
to be c areful ly examin e d ; th ere are soun d s which
some ears can d e t ect, wh ile oth ers cann o t, an d
s had es th at some eyes c ann o t di scrim inat e T h ere
are effec t s proc ee din g from certain causes th at are
s o sli ght th at w e fail t o observe th em an d yet erro
P rofessor
n eo u sly in fer th at th ey d o no t exis t
Tyn dal l has given a s trikin g illus t ration of th e dif
.

Q uote d by

J vons Princip les


e

f S c ienc e

pp 409 411
.

I NDU C TIVE

76

O GI C

ference o f au dit ory po w er in two in divi duals ; h e


says : In crossin g th e Wen g ern A lp in company
wi th a frien d th e g rass at eac h si d e of th e path
s w armed wi th insect s wh ic h to me rent th e air with
th eir s h rill chirrupin g My frien d h ear d no th in g
th e insec t music lyin g quit e beyon d his
o f th is
limi t of aud ition 1 M uc h h as been d one by in
v en tiv e skill t o increase ou r po w ers of O bservat ion,
an d at th e sam e t ime t o rend er th em more accurat e
as th e t elescope microscope th e vernier fo r precise
meas urement of minu t e differences o f ma gnitu d e th e
chrono g raph for time measurement s self re gis t ering
th ermomet ers th e th ermopile g alvanomet ers et c
O ne of th e ch ief problems o f scientic meth o d is t o
overcome natural limit ations of observat ion th rou gh
mech anical d evices
Observations on a lar g e scale an d over a c o n sid
crable perio d Of time mus t sometimes be t aken in
ord er t o disclose t en d encies as seen only in th e
averag e o r th e mean of th e observe d result s T hus
met eorolo g ical vit al s t atis tics an d o th ers o f a like
kin d mus t ext en d over a lar g e area an d embrace a
lar g e number o f ins tances in or d er t o reac h resul t s
of any value It is know n th at Tyc h o Brah e m ad e
an immense number of mos t exac t recor d s o f th e
positions of th e h eavenl y bo dies w ith th e ai d of th e
bes t of as t ronomical ins t rument s an d th ese record s

afterw ard s became th e foun d ation of K epler s law s


an d o f mo d ern as tronomy ?
T h e faculty for accurat e Obs ervation can be in
,

'

1
2

Tynd all

G o re

d
c

0 u S o u n , pp 7 3 , 74
Th e A rt of S ien ti D is
,

cov ery

p 3 16
.

CAU S AL ANALYS I S

N ATIO N

AND DE ERMI

77

crease d by acquirin g th e h abit o f ex aminin g care


full y every th in g w i th in th e el d of vision
We fail
t o see many thin g s because w e fall int o th e easy way
o f passin g th em by w i th ou t not in g th eir presence
o r appreciat in g th eir si g ni cance
I t was sai d of
Ch arles Darwi n by hi s so n that h e w is h e d t o learn
as muc h as possible from every experiment, s o th a t
h e did n ot conne himself t o observin g th e sin gle
poin t to wh ich th e experiment w as di rect e d an d
h is po w er of seein g a number of oth er th in g s w as
w on d erful 1 T h e open eye d v ision is th e prime
requi si t e for scientic investi g ation
T h e limit ations of observation naturally lea d t o
experiment, wh ose special fun ction is t o so mo di fy
ph enomena as t o brin g a s uspec t e d causal element
more prominently int o not ice T his can be d one by
int ensifyin g th e force in q u es tion or by neu t ralizin g
all o th er element s in combination w ith it, s o th at
th e sole effec t of th is force in actual operat ion can
be observe d Wh en th e cause is no t a S imple ele
ment but a combination th en th e problem is to
vary th e con ditions s o th at but one possible com
bination, th en anoth er can be operative al one, an d
not e th e correspon din g effect
Given a cert ain
number of element s , th e number of possible combi
n ations is math ematically d e t erminat e, an d can be
trie d seriatim un til all possibili ties are ex h aus t e d
V enn h as given a lon g an d int eres tin g illus t ration
?
A ll combinations
of th is in h is E mp irical L ogic
nee d no t be t rie d, h o w ever ; fo r many will be seen
.

Letters

Lif e

pp 402 ff
.

an d

f Charles D arwin

Vo l I p 122
.

I NDU CT IVE

78

O GIC

t o be eith er impossible or irrelevan t T h e aim is


t o obt ain an ant ece d ent which sh all consis t eith er
of a simple elem ent, or a combination such th at with
it s presence th e effec t in ques t ion is pres ent also,
but w ith i t s disappearance th e effec t is w antin g
It is no t sufcient t o not e merely th e presence
of an ant ece d ent connect e d with a correspon d in g
consequent ; scientic d et ermination consis t s, in ad
dit ion, in provin g th e absence of th e suspec t e d cause
in cases wh ere th e given effect is n ot presen t T his
is calle d d et erminat ion by ne g ation
A proposi t ion
wh ich is h el d afrmatively h as only a va g ue s ig
n ic an ce ; i t mus t be d e t ermine d wi th in d eni t e
limi t s assi gne d to it by virtue of wh at i t is not
Denin g means t o set limit s t o a t erm ; th ese limit s
gro w out of th e nat ure of th e thin g it self T h e
ne ga tive ju dg ment marks a t ransi tion al w ay s from
th at which is in d enit e an d incoh erent t o th at
which is denit e an d coh erent F or inst ance w e
h ave a vague notion O f ch emical afni ty th at ele
ment s combine to form compoun d s
T h at is th e
nucleus of our kno wle dg e ; i t grow s in d eniten ess
th rou gh a con tinuous process o f limit ation by ne ga
t ion We nd th at not all element s combine w ith
eac h oth er th at th ey d o no t combine except in cer
t ain propor tions an d th at even th ose whic h d o in
certain d eni t e proportions will no t combine in th e
presence of o th ers h avin g g reat er afni ty as , fo r
ins t ance in th e presence of oxy g en, an d so on
E v ery ne g at ive proposi t ion es t ablis h e d ren d ers th e
ori ginal one more accurat e
T his may be illus t rat e d also in th e concret e, wh en
.

ANALY SIS

CA US AL

T RMINA T ION

A ND DE E

79

in dissec tion one is t racin g a nerve ; i t is follo w e d


th rou gh out its course by a S eries of ne g ative ju dg
ment s th ou gh th ey be u nexpresse d : T h is is no t a
nerve bu t an art ery ; th is is n o t a nerve but a vein ;
this is not a nerve but a lament, or sh re d o f
muscle, et c S o w e rise th rou gh ne g ative discrim
in atio n t o a clear appreh ension of an O bj ec t un d er
inves ti g at ion Th e ori ginal proposition mus t be
readjus t e d with every ne w ne g ative d et ermination
It somet imes h appens th at th e ori ginal proposition
is complet ely ne g ative d by th e ne g at ive d etermin
ation, some tim es ag ain i t is conrme d
A proposit ion th at h as no t been w orke d over
th rou gh such a process h as no real lo g ical worth
or scientic value T h erefore in th e analysis of
ph enomena wh en th e suspect e d cause an d effect
are combine d in a proposi tion, it can at rs t be h el d
only t ent atively I t mus t be conrme d ne gat ively
or else readjus t e d t o conform t o th e ne g ative re
S uppose w e h ave g iven th at A is
quirem en ts
follo w e d by B as far as w e h ave been able t o ob
serve We may procee d by experimen t to mul tiply
ins t ances o f A s connection with B but still th e
causal relation is not absolut ely proved We mus t
g o on t o S h o w th at in all cases of n ot A th ere is not
N e g ative
B, or in all cases o f n o t B th ere is n o t A
experiment pro duces th e cont raposit ive or th e co n
verse cont rapositive of th e proposition un d er inv es
tig atio n, wh ic h d e duct ively necessi t at es th e vali d i t y
o f th e ori ginal proposit ion

T h is is subst antially Mill s meth o d of d ifference,


th at if an ins t ance in whic h th e ph enom enon un d er
,

INDUCT IVE

80

O GIC

inves ti gation occurs an d an inst ance in wh ich it


d oes not occur h ave every circums t ance save one in
common an d th at o n e occu rrin g only in th e former ;
th e circu ms t ance in whic h alone th e tw o ins t ances
differ is th e effec t o r cause o r a necessary par t of
th e cause of th e ph enomenon T his me th o d will
be d escribe d lat er ; it is th e main ind uct ive meth o d,
th e o th ers bein g lar g ely mo di cations o f i t A
ne g ative inst ance wh ich is es tablis h e d concerning
relations of n o t A an d not B is absolut ely conclu
sive, inasmuch as no t A is th e cont radict ory of A ,
and not B is th e cont ra dict ory of B T h ey are
mutu ally exclusive N O oth er possibility can be
forth comin g an d th e experiment al analysis is ex
h au stiv e
P rofessor Tyn d all gives th e foll o win g
account of an experim ent t o d et ermin e th e cause of
resonance I h ol d a vibratin g t unin g fork over a
glass j ar ei ght een inch es d eep ; but you fail to h ear
th e soun d of th e fork P reservin g th e fork in i t s
posi tion I pour w at er w ith th e leas t possible noise
int o th e j ar T h e colum n of air und erneath th e
fork becomes sh ort er as th e w at er rises T h e soun d
au gment s in int ensity an d wh en th e w at er reac h es
a cert ain level , i t burs t s forth wi th ex t raor dinary
po w er
I con tinue t o pour in w at er th e soun d
S inks an d becom es n ally as inau dible as at rst 1
F rom th is i t is inferre d th at a cert ain colum n of
w at er of deni t e h ei ght is necessary to th e pro duc
tion of th e soun d, for above an d belo w th e limi t s
no soun d is h ear d This experiment al so in dicat es
th at which is mos t import an t in causal d et ermina
,

Tynd all

0 n S ou n

p 17 2
.

C AU SAL

N ALYS IS

AND DETERMI

N AT IO N

81

tion ,
a variat ion in cause accomp anie d by a vari
ation in effect as also a maximum an d minimum
as re g ar d s th e int ensit y of th e soun d
E xperimen t
procee d s upon th e supposi tion o f th e measurable
ness Of p h enomena an d seeks numerically expres
sible resul t s in this re gar d For ins t ance, by
different experiment s Tyn dall prove d th at th e
len gth of th e column o f air wh ich resoun d s t o th e
fork in a maxim u m d e gree of int ensity is equal t o
o n e fourth o f th e len gth of th e w ave pro d uce d by
l
th e fork
T h e ne g at ive d e t ermi nat ion o f a suspect e d con
n ectio n o f cau se an d effec t mus t be precise in or d er
t o es t ablis h th e causal relat ion w ith th at d e g ree Of
accuracy whic h is d eman d e d in a t ruly lo gical an d
scientic me th o d U pon this point Bosanquet h as
a very su gges tive passag e : T h e essence o f signi
cant ne g ation consis t s in correctin g an d conrmin g
o u r ju dg ment o f th e nat ure o f a posi t ive ph en o m e
non by s h o win g th at j ust when it s cond i tion ceases
j us t then some thin g else be gins T h e Ju st no t is
th e import ant point an d this is only gi v en by a
po si tive ne g at ion within a d en it e sys t em Y o u
want t o explain o r d ene th e case in wh ich A be
comes B Y o u w ant Observation of no t B, so th at
Wh at you mus t d o is t o
you are los t in c h aos
n d th e point w i th in A wh ere A I wh ic h is B passes
int o A 2 whic h is 0 and th at wi ll give yo u th e j u s t

n o t
wh
ic
h
is
th
e
valuable
ne
g
a
t
ive
ins
t
ance
B
F or exam ple in P rofessor Tyn dall s experiment, th e
,

T yn d all 0 n S ou n d p 174
2 B o san q u et The E ssen tials o Lo ic
f
g

p 13 4
.

INDU C TIVE

82

OGIC

si g nicant ne gative ins t ance w as th is wh en th e


w at er in th e t ube reach e d jus t th at h ei ght wh en for
th e rs t time durin g th e experimen t no soun d w as
au d ible Th e discriminatin g observation th at can
mark an d measure th e precise point of t ransi tion
from soun d t o no soun d, h as d et ermine d accu rat ely
th e con ditions necessary t o pro duce th e soun d, an d
precisely d ene th eir limi t at ions
In all observation an d experiment, th e follo w in g
possibilities sh oul d be kep t before th e min d in or der
t o avoi d a h as ty conclusion in reference to a seemin g
causal conn ection We may th ink th at w e h ave d is
covere d th e relation th at if th ere is A th en th ere
mus t be B an d th e one th erefore th e cause of th e
oth er but it may h appen th at
1 Bo th A an d B are effec t s of anoth er cause an d
are th ereby relat e d c o ordinat ely in reference t o i t
2 A may be merely a liberatin g circums t ance
or an invariable accompaniment o f B
3 A may n o t be th e cause of B but only an
el ement O f a complex collocation wh ic h is th e
cause of B
4 E ac h separat e ins t ance of B may so differ as
t o respon d t o th e ac tion of A in a manner d ierent
from th e oth ers
5 A may be relat e d t o B in a sys t em o f suc h a
nature th at th e sys t em in continuously d evelopin g
n e w effec t s causes B as th e int ro d uc t ion of me dicine
int o an org anism wh ose forces are th emselves effect
in g a h ealin g process
6 I t is O ft en very di fcul t t o t ell wh eth er A is
th e cause of B o r B th e cause of A as in dis t rict s
,

CA US A L

A NA LY SIS

T RMIN AT I O N

83

AN D DE E

wh ere d runkenness an d poverty are prevalen t, o r


cases o f moral an d int ellec t ual feebleness W h ic h
is th e cause an d wh ic h th e effec t ? In many
cases suc h as th ese th e forces reac t upon each
oth er, th e effec t t en din g to increase th e int ensity
of th e cause
7 T h e connec t ion of A an d B may be one o f
mere coinci d ence, an d no t O f a causal nature wh at
soever
N e wt on w as muc h impresse d w ith th e
apparent conn ection betw een th e seven int ervals of
th e oct ave an d th e fact th at th e colors of th e spec
t rum divi d e in t o a like series of seven int ervals
A nd ye t th ere is no causal co nnec t ion th at c a
n be
prove d t o exis t betw een th e tw o
T h e more w e dw ell upon th ese various possibilities
th e m ore are w e impresse d w ith th e ex t reme com
ple xity in wh ic h th e relat ion o f cause an d effec t is
involve d
Th e investi g at or mus t bring t o his re
search th e spirit o f pat ience an d perseverance, as
w ell as a clear vision an d discriminatin g insi ght
S ir J oh n L ubbock , in h is observat ions upon th e
h abit s o f an t s , says th at at o ne time h e wat ch e d
an an t from six in th e mornin g unt il a quart er
t o t en at ni ght, as Sh e w orked w ith out in term is
sion durin g all th at time 1 I t is t o suc h pat ien t
inv esti g at ors th at N ature reveals h er secret s
.

Sir

John Lubb

o ck ,

S cientic Lec tures, p 73


.

CH AP T ER

V II

M ILL S I N DU CT I VE METH O D S
A G R EE M E N T

THE

ME TH O D

T HE RE

OF

are various meth o d s o f causal res earc h


which h ave receive d th e name of in duc tive me th o d s
an d h ave been especially associat e d w i th th e con
tribution o f J oh n S tuart Mill t o th e h is t ory o f
lo gic T h ere are v e of th ese meth o d s or inferen
tial processes as g iven by Mill , an d formin g th e in
T h ey are
tegral part of his sys t em of in d uc t ion
as follo w s :
1 T h e Me th o d o f A g reement
2 T h e Me th o d of Difference
3 T h e J oint Me th o d o f A g reement an d Dif
erence
4 T h e Meth o d o f Concomi t ant V ariat ions
5 T h e Meth o d of R esi dues
T h e meth o d O f a greement consis t s in inferrin g
th e exis t ence of a causal relation wh en in a num
ber of varyin g ins t ances i t is observe d th at th e
suppose d cause is al ways accompanie d by th e ph e
n o m en o n in ques t ion as correspon d in g effec t
Th e
meth o d of d ifference is th e comparin g o f an in
s t ance wh ere th e suppose d cause i s presen t acc o m
pan ied by th e correspon din g e ffec t w i th an ins t ance
.

84

THE

I NDU C T I VE

METH ODS

85

h avin g precisely th e sam e se ttin g, but wh ere th e


suppose d cause is w ithd rawn, th e effect also disap
pearin g ; th e inference o f a causal relat ion is th en
permis sible Th e j oint me th o d of a g reement an d
difference is th e comparin g of ins t ances wh ere th e
suppose d cause is presen t, with S im ilar ins t ances
wh ere i t i s absen t ; if th e correspon din g effec t is
present in th e former an d absent in th e latter g roup
of ins t ances a causal relat ion may be inferre d
This differs from th e meth o d of difference, th at in
th e latt er th e same ins t ance now wi th an d ag ain
w ith out th e presence of th e suspect e d cause is th e
subj ec t o f observat ion ; in th e j oint meth o d i t is a
number of ins t ances wi th compare d w ith a number
o f similar ins t ances w i th ou t, th e presence of th e su p
pose d cause T h e meth o d of concomitant variations
consis t s in so mo difyin g any g iven ph enomenon th at
th e suppose d caus e w ill vary in int ensity ; th en a
correspondin g variation in th e accompanyin g effect
is evi d ence of a causal relat ion T h e meth o d of
resi dues consist s in th e analysis of a given complex
ph enomenon in which all element s save one o f th e
an t ece d ent are kno w n t o be relat e d in a causal
manner to all element s save one of th e c o n se
quent ; th en th e resi dual element o f th e one may be
re g ar d e d as th e cause o f th e res idual element o f th e
oth er
We will n o w examine th ese meth o d s more in
d et ail Th e brief outline above is int en d e d merely
t o give a g eneral i dea O f th e meth o d s th at i t may
lea d t o a bett er u n d ers t andin g o f th e more exact
s t at ement of th eir nature an d ch aract eris t ics
.

I NDU CTIVE

86

M etho d

O GI C

A greement

T h e more precise
s t at ement o f this meth o d is g iven in th e rs t canon
of Mill , wh ic h is subs tantially as follo w s
If tw o or more inst ances of th e p h enomenon
un d er inves ti g ation h ave only o n e circums t ance in
common th e circums t ance in which alone all th e
ins t ances ag ree is th e probable cause (or effect) O f
th e given ph enomenon, or sus tains some causal
relation t o i t
T h e above is base d upon th e causal axiom th at
th e cons t ant element s wh ich emer g e in any given
series o f si m ilar p h enomena are t o be consi d ere d as
connect e d in some m anner wi th th e caus e of th e
ph enomena ; but th at th e variable element s are no t
connect e d wi th th e ph enomena in any causal man
ner wh at soever
Th e meth o d of a g reemen t is illust rat e d in th e in
v e stig ation o f th e very common p h enomenon o f th e
t ransformation of subs t ances from th e soli d t o th e
liqui d s t at e Wh at is th e on e circumst ance wh ic h
is al ways present wh en w e consi d er th e mel tin g of
suc h wi d ely different subs tances as butt er ice, lead
iron, etc ? In all ins tances t o wh at soever ext ent
th ey may be mul tiplie d of th e c h an g e from soli d t o
liquid s t at es h eat h as been observe d t o be present,
an d is th ereby indicat e d as th e likely cause o f th e
ph enomenon in ques tion
T h e meth o d may be
represent e d th rou gh th e u se of symbols which ao
cor din g to Mill are th e capi t al lett ers to d eno t e
ant eced ent s an d th e smaller lett ers to d enot e cor
respon din g consequen t s L et th e follo win g be a
number of different ins t ances w i th th e ant ece d ent s
The

TH OD

THE ME

OF

A GREEMENT

87

an d consequent s arran g e d in or der, an d represent e d


as above in dicat e d :
A BC

a bc

A DE

ad e

A MN

amn

etc

et c

By inspect ion of such a t able of ins t ances th us

an al yze d, an d symbolically represent e d, i t w ill be


rea dily seen th at A is th e only element common t o
all th e ant ece dent s wh ile a is th e only o ne common
t o all th e consequent s T h e inference, th erefore is
th at A is th e cause of a I t h as been obj ect e d t o
th is sys t em of represent atio n th at i t articially ar
ran g es th e element s O f an t ece d ent and consequent,
as th ou gh th ere w ere a number Of dis tinc t cause
element s each connec t e d w ith a correspon dingly
d istinc t effect element, an d it pro duces th e impres
sion th at i t is quit e an easy matt er t o s e e h o w th ese
1
A s nat
causal pairs are th us separat ely relat e d
ure present s h er ph enomena to u s, h o w ever th ere
is such complexity th rou gh out, th at th e analysis
canno t rea dily dis t ribu t e part to part in appropri
at e causal relations To avoi d such an error in
not at ion I h ave ad opt e d th e follo wing symbols,
wh ic h w ill be use d h ereaft er t o d escribe th e various
meth o d s L et us t ake 0 as th e lett er t o represent
th e suppose d causal element, an d S , th e entire set
t in g o f accompanyin g circums t ances ; let e d eno t e
th e correspon d ing eec t an d s th e sum t o t al of th e
att en d ant consequences T h e causal relation will
,

'

Venn

E mp iri

cal Log ic

p 4 11
.

INDU CT I VE

be th en in dicat e d accor ding


ment, as follow s

to

88

O GIC
th e meth o d of a g ree

+ 0

8
'

"

C
+

Here th e settin g ch an g es throu gh out, as in d icat e d

'
by S , S , S et c but 0 remains cons t ant in th e
antece dent s ; also th e correspon din g settin g in th e
co nsequen t s c h an g es, as in dicat e d by s s '
etc
but e remains cons t ant th rou gh out S uc h a not a
tion d oes no t att emp t t o represent jus t wh ic h part s
of S cause corresponding part s of 3 nor by wh at

"
element s precisely S differs from S an d S e t c
I t d oes represent h o w ever, th e difference be tw een
th e variable an d cons t ant element s o f th e t able of
ins tances which are arran g e d for comparison, an d
this is th e key t o disclose th e causal relat ion
A s an example O f th is meth o d let u s t ake th e
ph ysical la w th at di fferen t bo dies t end at th e same
time t o absorb an d to emit th e same w aves of li ght
I t is kno wn th at every subs t ance in burning gives
it s o wn lines in th e spec t rum analysis so dium for
ins t ance, pro ducin g a very bri ght line in th e yello w
portion of th e spect rum in a d enit e locali ty ( L ine
If n o w ins t ea d of burnin g
D o f F raun h ofer)
so dium w e int erpose th e vapor o f so dium in th e
path o f th e ray wh ic h s h oul d give a cont inuous
S pect rum th e ph e nomenon is comple t ely reverse d ;
at th e exac t point wh ere th ere wa
s a bri ght line in
th e spec t r um a d ark li n e n o w appears Th us th e
vapor O f so dium , actin g as a screen, absorbs th e
,

THE

METH OD O F

A GREEMENT

89

rays wh ich it emit s wh en i t ac t s as th e luminous


source A similar effec t is observe d in th e case o f
vapors o f io dine of s t rontium of iron et c ; an d is
a ph enomeno n th erefore a dmi ttin g o f g eneraliza
tion by in duction 1 T h is is accor din g t o th e meth o d
o f a g reemen t ; an d w e may m ake th e foll ow in g
represent ation
.

Vapor

of

so dium ac tin g as a screen


io dine

'

"

C
0

et c
The

correspon din g con sequent s are :

R eversin g bri ght So dium line t o dark

io dine

s t rontium
et c

e
'

T h erefore w e h ave
S

111

+ 0
+ 0

et c

In th is th e cons t ant C of th e ant ece d ent s is th e


vapor o f any subst ance actin g as a screen ; th e con
s t ant e is th e reversa l in each case o f th e bri ght
1

Saig ey, Th e Unity of N atural

P henomena

pp 94 , 95
.

I NDU C T IVE

90

LO GI C

th e subs t ance in th e S pect rum t o th e corre


FrOm th is i t is
spo n d ing d ark line o f th e same
inferre d th at th e vapor of any subs t ance actin g as
a screen absorbs exactly th ose ray s whic h i t emi t s
wh en i t act s as th e luminous sou rce
I t is of g reat import ance th at th e ins tances se
lecte d for observation o r experimen t be as varie d as
possible s o th at wi d ely di fferin g p h enomena may
be g ath ere d t o g eth er T h en if runnin g th rou gh
th em all th ere is one common element observe d
amon g th e ant ece d ent s an d one common element
amon g th e consequent s th e g reat er th e variation
amon g th e inst ances th e more pronounce d will be
th e si gnicance of th e cons t ant element s In th e
illus trat ion g iven th e subst ances Wh ich are so differ
ent as iron s t rontium , so dium io dine e tc preclu de
th e possibility of th e resul t an t ph enomenon d e
scribe d bein g due t o th e peculiar propert ies o f any
S o many an d s o dif
o n e me t al or g roup o f me t als
ferent in kin d, are t aken as t o eliminat e th e peculiar
i ties att ach e d to any one in particular In th is re
spect th e meth o d is one of elimination By varyin g
th e inst ances th e n o n essential is eliminate d an d th e
essential , whic h remains as th e element common t o
all is th ereby emp h asize d an d d ifferen tiat e d from
all att end an t circums t ances
T h is meth o d also is o ne of d iscrimination o f
discernin g th e cons t ant element un d er th e various
ch an gin g forms whic h i t can assume an d as such it
is similar t o th e lo gical process o f th e format ion of
a concept Th e concep t is th e g raspin g O f th e
universal element whic h is present th rough th e

line

of

THE

AGREEMEN T

METHOD O F

particular an d concre t e manifes t ation s o f th e same


T h rou gh th em all th ere is th e like common element
wh ich is th e basis of th e concept it self S O o ut
of many particular ins t ances th e min d g rasps th e
element s which are common t o all an d consi d ers
th em as relat e d in a cons t ant an d th erefore causal
manner, wh ic h h as in i t self th e ch aract er of a uni
versal concept an d so a dmit s of bein g formulat e d in
th e form of a law universal , which is th e en d of all
in duc tion
Th is me th o d moreover, is peculiarly a dapt e d t o
observation th e collatin g of a nu mber o f inst ances,
rath er th an t o experiment Ins t ances canno t al
w ays be manufacture d an d so i t may be beyon d
th e po w er of experiment t o repro d uce th em T h ey
h ow ever, al ways be th e O bj ect s o f research
c an
an d as suc h fall n atiua lly int o th e el d of o bser
vation
T h e ques tion may properly be aske d at th is point
Ho w d oes th is metho d differ from th at o f in duct ion
by S imple enumeration 7 T h e latt er w e have seen
i s never satisfact ory because th e enumeration can
an d may be cont ra d ict e d by an
n o t be comple t e
enl ar g e d experience
T his meth o d h o w ever is
superior in th at it provi d es for more th an S imple
enumeration of ins t ances in whic h th e ph enomenon
in ques tion h as occ u rre d ; th ere mus t be a corre
spo n din g an al ysis of th e ins t ances , accompan ie d by
a discriminatin g insi ght t o dis tin guish th e essential
from th e unessential
N umber o f inst ances in
creases th e probabili ty th at th e variable element s
h ave been elim inat e d an d enables th e min d t o con
.

INDU C T IVE

92

OGI C

centrat e upon th e cons t ant element s th at remain


an d are th ereby disclose d
T h is meth o d primarily adm it s O f applic ation to
ins t ances wh ere a sequence i s observable ; th at is
wh ere ant ece d ent can be dis tin guis h e d from con se
quent by an appreciable t ime element It is h o w
ever, possible to apply th is meth o d to th e inves ti
gatio n of coexi s t ences , wh ere i t may s h o w th a t ei th er
th e coexis t in g element s are relat e d as cause an d
effec t, or th at in some causal manner th ey are th e
correlat e d eff ect o f some cause sufcient t o account
for th em both Many in s t ances may be a dduce d
o f th e prevalence o f pover t y an d crime associa te d
t o g eth er T his may indicat e a causal relation be
tw een th em an d yet a sequence cannot be Obser v e d
of sufcient d enitene ss t o in dicat e which is th e
cause an d wh ich th e effect T h e problem is th us
left in det erminat e, wi th th e su gg es tion o f some
o th er cause which may possibly account for th em
both All th at th e meth o d of a greement can at
t ain is by collectin g a n umber O f ins tances of d i
verse nature t o in dicat e th at in some w ay at least
poverty and crime are connec t e d by causal t ies
T h e cons tant coexis t ence O f attribut es in one in d i
vi dual admit s of a similar t reat ment an d similar
resul t s T h e fac t o f th e hi gh colorin g o f male but
teries in a lar g e number O f ins t ances in reference t o
a variety o f species in dicat es a cons tant relation be
tween th e fac t o f its being a male an d th e possession
of brilliant colorin g T his inseparable association
in dicat es a causal relation wh ich h ow ever cannot
be more precisely det ermine d by th is meth o d Th e
.

THE

METH OD O F

A GREEMENT

93

full explanation o f th e ph enomenon requires some


supplemen t ary hypoth esis d ependi ng upon con d i
tions no t disclose d by this meth o d an hypo th esis
suc h th at th e h i gh colorin g h as th e special function
Of attractin g th e female butt ery an d h as been
in t ensi e d an d develope d by natural selec tion
T h e m eth o d of ag reement is open to criticism at
several poin t s an d ye t i t mu s t be at th e be ginnin g
un d ers t oo d th at th is me th o d d oes n o t rank as a
nal meth o d
We s h all soon see th at i t serves
rath er as su gg es tive o f an d lea din g t o experiment s
accor din g t o th e meth o d o f difference t o corroborat e
o r d isprove th e resul t s wh ic h th e me th o d of a g ree
ment may h ave att aine d Th e c hief criticisms th at
h ave been mad e o f th is me th o d may be sum me d up
as follo w s :
1 T h e cause in dicat e d by th e me th o d O f ag ree
ment is no t th erebyprove d t o be th e sole cause of
th e ph en om en o n in ques t ion
W e may g ath er t o
g e th er a number of varie d ins tances wh ere an ex
t ensive failure of crops in th e summer h as cause d
h ar d times durin g th e wint er follo win g A n d yet
th ere may be an d as a fact th ere are many o th er
causes wh ich en g en d er perio d s o f indus t rial d epre s
sion We may s ay, th erefore, th at th is meth o d is
capable of es t ablis h in g t en t at ively at least, an uni
versal proposi t ion o f th e form All a: is y ; i t does
n o t, h o w ever a tt emp t t o g ive any i nd icat ion o n e w ay
or th e o th er re g ar di n g th e vali dity of th e converse
All 3; is a
K no w in g th e limi t ations of a me th o d
d oes not by any means d es t roy its le gitimacy as a
meth o d ; it rath er increases it s efcien cy w ithin its
,

'

,
,

'

INDUCT IVE

94

O GIC

proper sph ere by th e more exac t kno wle dg e as t o


th e precise ext ent of th at S p h ere it self
2 It is ur g e d th at wh ile i t is possible t o reco g
nize in most if no t in all cases , th e common element
in th e several effec t s o f similar ph eno mena, it is no t
so easy a matt er to d i fferent iat e th e common ele
ment in th e correspon d ing ant ece dent s by th e sim
ple meth o d of ag reem ent alone Fo r ins t ance in
Bacon s illus t ration of th e in v es t i gat ion o f th e cause
of h eat, h e ci t es su ch disparat e ph enomena as th e
sun s rays friction, combustion, et c T h e element
of h eat is readily discernible th rough th em all ;
but wh at is th e common element wh ic h operat es as
caus e in each case ? T h ere is th e difcul ty S ig
w art illus t rat es th is in th e case of th e p h enomenon
o f d eath
T h e effec t can be easily d e tect e d as sim
ilar th rou gh ou t bu t in all th e ant ece d ent s th e only
property common to th em all is life, an d th erefore
w e are le d int o th e fallacy of att ributin g t o life
l
th e cause of death
We mus t th erefore ackn o w l
e dg e th at some p h enomena may occur in suc h a
variety an d such a number of manifes tations as t o
dis guise th e nature of th e cause un der th e mask
of a g enerality t oo in d enit e t o be reco g nize d In
all such inst ances th e me th o d o f a g reem ent mus t
operat e upon su gg es t ions receive d from so m e o th er
source, as t o th e nature of th e common element in
th e ant ece d ents O r some minor circumst a nces
att en din g th e effect may indicate more precisely
th e nature of th e cause as for inst ance, th e peculiar
sympt oms associat e d w ith d eath by dro wnin g
,

S ig wart, Log i

V o l II p 341
.

THE

METH O D OF

A GREEMENT

95

Th e common element in th e ant ece d ent s may


prove t o be an unessential accompaniment of all th e
ins t ances examine d I t s presence th erefore, may
h ave no th in g wh at soever t o d o w ith th e O bserve d
A number o f d ifferen t me d icines fo r ex
effec t s
ample may pro duce a cert ain effec t alike in all
ins t ances T h e only common element th at can be
d e t ec t e d in th e various me dicines examine d may
be th e alcoh ol whic h is use d as th e common vehicle
o f th e d i fferent d ru g s
and ye t its e ffect may be
entirely inert as re g ar d s th e me dicinal qualities in
ques t ion T h e common element really efcient may
be overlooke d an d anoth er common element wh ich
is easily discernible may neverth eless remain wh olly
inoperative T h is difcul ty may be overcome by a
more th orou gh analysis of th e ph enomena observe d
wh ic h may be a tt aine d by a ju dicious variation of
th e ins t ances , so as t o reveal in turn th e precise
effec t of th e various simple element s wh ich t o g eth er
cons ti tut e th e complex wh ole of th e ph enomenon in
ques tion T h e d efect s of th e meth o d in this respect
are , in a w or d, th e d efect s of in duc tion by simple
enumeration
4 T h e cause may be present in all th e antece
d ent s , an d, notwi th s t andin g th e correspon din g ef
fect, no t appear, an d th is , no t because th e tw o are
no t relat e d in a causal manner but because th e
cause is neut ralize d by th e associat e d elem ent s
wh ic h appear in combination w ith i t in th e various
ant ece d ent s F or ins t ance diphth eria g erms are th e
cause of diphth eria, an d h ave been foun d ac co mpa
nyin g th is disease in all cases wh ich h ave been o h
3

INDU C T IV E LO G IC

96

serve d A nd ye t th eir presence is oft en no t e d wh en


th e disease it self d oes not d e v elop T h e t endency
exist in g is counterac t e d by th e con dition o f th e
organism at th e t ime s o th at th e d rea d bacilli are
inoperative and th erefore h armless : A s we h ave
seen before th e presence o f th e e ffec t necessi tat es
th e presence of th e correspond in g cause ; but by no
means is i t al w ays true th at th e presence of th e
cause necessi t at es th e effec t T h e cause al ways
pro duces th e t end ency at leas t, wh ich, h o w ever, may
be neutralize d
5 T h is meth o d is oft en applie d in a very care
less way to th e observations of persons wh o do not
possess th e po w er Of accurat e d iscrimination, an d
th erefore observed C o i nci d ences are h ast ily assume d
t o be particular inst ances of an universal law
Such proce dure lead s t o supers t i tion an d preju d ice
It not only warps th e ju dgmen t o w in g t o it s ill o gi
cal nature, but it also aects in directly th e man s
moral vie w as i t implies a w eakness in c h arac t er
T his criticism refers , h o w ever,
as w ell as in min d
t o th e abuse rath er th an th e le g itimat e use Of
this meth o d un d er suc h res t ric tions as h ave been
already in dicat e d
T h e chief funct ion of th is me th o d is th at of su g
ges tion It in dicat es oft en only th e possibili ty of
th e exist ence Of a causal relat ion ; in o th er cases i t
lead s t o an inference of h i gh probabili ty In all
cases h o w ever it marks but th e preliminary s t eps
of an inves ti g at ion which s h ou l d be follo we d up by
pains takin g experiment A s it is th e meth o d O f
observation chiey it is n ati i ral th at i t s h oul d pre
.

THE

ME TH OD O F

A GREEMENT

97

ce de experiment ; for i t is only by re ect ion upon


our observations th at w e discover th e nature and
relations o f ph enom ena, wh ic h serve as d at a for
subsequent experiment
I h ave select e d several illus t rations to in dicat e
th e v arious el d s o f researc h in which th is meth o d
of ag reement h as le d t o satisfac tory result s
T h e rs t refers t o th e relation betw een th e
occurrence o f nancial crises an d th e prevalence
of over pro duction
Guyo t, in his ETW L QL
SW
,
gi ves th e follo w in g ins t ances :
A n enormous cons um p t ion of capit al in th e U ni t e d
S t at es in th e seventies for th e cons t ruc t ion o f rail
roa d s w as follo w ed by unusual commercial d e pre s
sion T h en th e like outlay in In dia for rail w ay
cons t ruct ion by means o f loans an d t axes which
absorbe d th e wh ole circulatin g capit al Of th e In dian
population, was follo w e d by a d evast atin g famine
A g ain in G er
an d g eneral commercial paralysis
many th ere w as an en ormous con sumpt ion o f capi
t al in for t s an d armament s an d g eneral milit ary
equipment , brin gin g on th e crisis O f 187 618 7 9
En glan d at th e same t ime w as un dul y supplyin g
circulat ing capit al t o th e Uni t e d S t at es , Egyp t,
an d h er colonies , an d a nancial crisis w as th e
res ul t
T h rou gh all th ese varyin g ins t ances an d
o th ers of a like nature whic h m i ght be a dd e d th e
cons t ant relat ion o f over consumpt ion in th e an t e
ce d ent s t o th e in dus t rial d epression evi d ent in th e
effect, in dicat es th e one t o be th e cause o f th e o th er,
ei th er in wh ole or in part

A g ain , i t is narrat e d in Bre w s t er s Treatise o n


.

INDUCT IVE

98

O GI C

t ics th at h e acci d ent ally t ook an impression from


a pi ece o f mo th er O f pearl in a ce ment of resin an d
bees wax, an d ndin g th e colors repeat e d upon th e
surface o f th e w ax, h e procee d e d t o t ake o th er
impressions in balsam, fusible me tal, lea d, gum
arabic isin glass , et c , an d al w ays foun d th e iri des
cent colors th e same His inference was th at th e
form of th e surface is th e real cause of suc h color
effec t s ? T h e common element whic h appears in
all th e ant ece d ent s is evi d ently th e same form
impresse d upon each, which was ori ginally receive d
from th e mo th er Of pearl T h e cause is, moreover,
in depend ent of th e nature of th e subs t ance in each
case which receive d th e impression upon its su r
face because such a variety of subst ances was
chosen as t o eliminat e th e indivi dual nat ure o f each
as an in uencin g fac t or in th e resul t
In th is
experiment w e see th e advant ag e o f varyin g th e
ins tances as far as possible for this very purpose O f
eliminatin g all irrelevant element s S imilar experi
ment s h ave prove d like resul t s in reference t o th e
colors exhibi t e d bv thin plat es an d lms Here
th e ring s an d lines of color h ave been foun d to be
nearly th e same wh at ever may be th e nature of th e
subs t ance A sli ght variation in color is due t o
th e refractive in dex o f th e int ervenin g subs t ance
With th e excep tion of this , th e nature Of th e su b
s t ance is n ot operat ive in pro ducin g th e color effec t,
bu t th e form alone
T h e celebrat e d scientist P as t eur, in th e year
186 8 w as carryin g o n h is inves ti g at ions as t o th e
-

Q uo ted by

Jv
e

o ns ,

Pri nciples of S c ience

p 419
.

THE

METH OD O F

A GR EEMENT

99

cause o f th e bli ght th en d evas t at in g th e silkw orms


o f F rance
O ne of his experiment s consis t e d in
selec tin g thirty perfec tly h eal thy w orms from moth s
th at w ere entirely free fro m th e corpuscles wh ich
latt er are th e g erms o f disease, or at th at t ime su s
T h en rubbin g
pecte d t o be th e g erms o f d isease
a small corpusculous w orm in wat er, h e smeare d
th e mixt ure over th e mulberry leaves
A ssurin g
h imself th at th e leaves h ad been eat en h e w at ch e d
th e consequences d ay by d ay O ne aft er th e o th er
th e w orms languis h e d ; all S h o w e d evi d ences o f
bein g th e prey of th e corpusculous matt er an d
nally w ith in o ne m on th s t ime, all die d
Pas
teu r s inference nat urally w as th at th e corpuscles
h a d pro duce d th e d eath
O f course his resul t s
w ere no t foun d e d upon this experiment alone but
o th er experiment s carrie d o n in m any different
ways , serve d t o corroborat e th e causal relation
wh ich th e experimen t us t d escribe d h a d su gg es t e d
as at leas t h i gh ly probable
In me dicine al so th e meth o d of ag reement is
O ft en use d with effec t Cert ain d ru g s are ad minis
tered in a num ber o f cases an d th e resul t s no t e d
A n uniform effec t consequent upon th e a d minis t ra
tion of a g iven d ru g in dicat es a causal connection
capable o f g eneralization N ot only are s ubj ec t s in
disease, but also in h eal th , selec t e d, an d th e effec t s
upon both th e normal an d morbi d natures compare d
T hus a variat ion in inst ances is secure d If a num
ber o f different d ru g s pro duce like effec t s, th e ques
tion at once su gg es t s it self Wh at is th e property
common t o th em all ? T h e me th o d of ag reement
.

INDU CT I VE

100

LO

GIC

some in dication O f th is wh en th e elim


in atio n of th e inert propert ies c an be accomplish e d
th rou gh a sufcient variation of ins t ances
Th e
difcul ty lies h o wever in this very thin g t o s o
vary th e ins tances as t o di sclose th e efcient ele
ment present in th em all V arious me dicines pre
sent a compl ex nature of such a ch arac t er th at i t
is ext remely di fcul t to ascribe th e precise e ffec t s
which th e several component part s in divi dually
exercise
T h e meth o d o f a g reement is also u sed perh aps
unconsciously, in th e con duc t of th e every d ay
affairs of life Wh enever different ph en omena in
our experience present certain c h aract eris tics o f a
cons tant nature we are at once le d to suspec t a
causal conn ec tion an d t o s t art upon a more search
in g investi g ation of th e same TOO oft en , h o w ever
th e supplement ary inves ti g ation is omi tt e d an d th e
min d res t s cont ent w ith a fe w surface resemblances
th at lea d to a h as ty g eneralization w ith out bein g
more precisely an d a dequat ely det ermine d
oft e n

gi ves

CHAP T ER

METH O D

THE

OF

VIII
DIFFE REN CE

meth o d o f ag reement, as w e h ave seen pre


sent s a causal relation as a su gg es tion, a dmi ttin g O f
a h i gh d e gree o f probability i t may be, but requir
in g t o be t es t e d by some more scientic me th o d
Thi s is accomplish e d by th e meth o d of difference
Here a ph enomenon is O bserve d, in whic h th e
suppose d cause element an d effec t element appear ;
th en while all o th er circums t ances an d con ditions
remain unal t ere d, th e suppose d cause element is
withdra wn, or i t s force a d equat ely eliminate d ; th e
imm e diat e disappearance of th e suppos ed effec t
element consequent upon this, in dicat es a causal
relat ion exis tin g be tw een th e two O r th e e xperi
ment may be ma de in a different manner but t o
th e same en d ; th at is a ph enomenon may be ch ar
acte rized by th e absence of bo th cause element an d
th en, if th e int ro duction o f th e
eec t elemen t ;
cause
e lement does no t di s turb th e p h enomenon in
question except immediat ely t o pro duce th e eff ec t
element th e inference may be d raw n th at th e o n e
i s th e veritabl e cause O f th e o th er
If an in
Can o n of the Metho d of Dierence
st ance in whic h th e ph enomenon un d er invest i g ation
THE

'

101

INDU CTIVE

102

O GIC

occurs and an ins t ance in wh ich it d oes no t occur


h ave every circums t ance save o ne, in common th at
o n e occu rrin g only in th e former th e circums t ance
in wh ic h alone th e tw o ins t ances differ is th e e ffect
or a necessary part of th e
o r i t may be th e cause
cause of th e ph enomenon
T his meth o d h as manifol d illus t ration in o ur
every d ay inferences A person is asleep in th e
room with u s an d w e h ear th e lou d noise o f a S lam
min g door, and observe th e person at once aw akenin g
with a s t ar t an d exclamation We h ave no h esi t ancy
in ascribin g th e aw akenin g to th e noise imme diat ely
prece din g it We observe a gain some one receivin g
a lett er o r t ele g ram, an d imme diat ely u pon openin g
i t th e face grow s whi te with anxiet y an d fear th e
h an d s t remble, an d th ere are sh o wn g eneral symp
t oms o f perturbation Th e messa g e receive d w e
h as cause d th e ment al s h ock an d p h ysical
s ay
accompaniment s
O r, taking a simple experiment in qui t e anoth er
sp h ere, it w as observe d by Boyle in 167 0, th at an
ext ract of li tmus w as imme diat ely turne d re d by
th e int ro duction Of an aci d T his subsequently
became a t es t for th e presence O f aci d s th e infer
ence bein g th at an aci d h as th is capaci ty of ch an g
in g th e li t mus t o are d color from it s ori ginal blue
P rofessor Tyn dall d escribes an experiment to
prove th at w aves of eth er issuin g from a s t ron g
source such as th e sun or elect ric li ght are compe
t ent to s h ake asun der th e at oms o f g aseous mole
cules suc h as th ose of th e sulp hur an d oxy g en
wh ich cons ti tut e th e molecule of sulp hurous aci d
,

THE MET O D

OF

103

DI FFERENCE

He enclose d th e subs t ance in a vessel pl acin g i t in


a d ark room an d sen din g th rou gh i t a po w erful
beam of li ght A t rs t no th in g was seen ; th e
vessel cont ainin g th e g as seeme d as empty as a
vacuum S oon alon g th e t rack o f th e beam , a
beaut iful sky blue color w as observe d, due to th e
liberat e d part icles of sulphur F or a time th e blue
g re w more int ense ; i t th en became wh itish ; an d
from a wh itish blue i t passe d t o a more or less per
fe ct wh i t e
Cont inuin g th e ac tion th e t ube became
lle d w i th a d ense clou d of sulp h ur part icles wh ich ,
by th e application o f proper means coul d be reni
?
d ere d visible
In th is series of continuous c h an g es,
w e n d th e one ant ece d ent, gi vin g initiative causal
impul se, t o be th e beam of li ght I t w as th e one
element int ro du ce d wh ich s tart e d th e several
ch an g es lea din g to th e appearance of th e sulphur
visibly manifest e d T h e o n e th erefore, is t o be
re g ar d e d as th e cause of th e o th er
I t is possible t o represent this meth o d by mean s
of symbol s in a manner similar t o th at of th e
meth o d o f a g reement
L et C be th e suppose d
cause an d e th e effect correspon din g while S an d 3
denot e th e settin g o f an t ece dent an d consequent
respectively We h ave th erefore, th e follo win g :
,

'

T h en, w ithdrawin g

C,

w e h ave th e absence

Tynd all

T h e inference th en
1

of e

Use

and

is

Limit

th at

is th e cause of

f the Imag in atio n

in S ien

ce

p 33
.

104

INDU CT I E LO G IC

In th e meth o d o f ag reement, a number of in


s t ances w ere t aken ag reein g only in th e posses
sion of tw o circums t ances, th e cause an d effect
element s common t o th em all In this meth o d
only tw o i n s t ances are t aken, an d th ey mus t be
precisely alike wi th th e o ne excep tion, th e pres
ence o f tw o circumst ances in o ne th at is th e
cause an d th e effec t element s an d th e absence o f
th e same in th e oth er In th e meth o d o f a g ree
ment w e compare th e various ph enomena t o not e
wh erein th ey ag ree ; in th e meth o d of difference
w e compare th e tw o ph enomena to not e wh erein
th ey d iffer Th e lo gical axio m und erlyin g th e tw o
m e th o d s is subs t antially o ne an d th e same, d iffer
ing only in its special ad ap t ation in eac h case
T h e meth o d res t s o n th e assumpt ion, wh ic h mus t
be accept e d as a fun dament al pos tul at e th at wh at
ever can be eliminat e d from th e various inst ances
is no t connec t e d wi th th e p h enomenon un d er in
v e stigatio n in any causal manner ; an d th e me th o d
of difference is base d on th e pos t ulat e th at wh at
ever canno t be eliminat e d is connect e d w ith th e
ph enomenon by a cau sal law
T h e meth o d of di fference is evid ently th e meth o d
by ne g ation wh ich h as already been in dicat e d as
th e trul y scientic process in in duc tion I t is also
pro eminent ly th e meth o d of experimen t rath er
th an observation ; for th e withd raw al o r intro d u c
tion of forces can only be accomplis h e d at w ill
wh en w e brin g th e p h enomena und er experiment al
co nt rol A t t imes N ature h erself may perform
th e experiment for u s, an d w e s t an d simply as
-

METHO D

TH E

OF

DI FF ERENCE

105

observers t o no t e th e resul t s T h is is especially


th e case in th e cat as t rophic ph enomena suc h as
volcanic eruption earth quakes et c
G enerally
speakin g h o w ever th e meth o d O f d ifference is th e
proces s of man s manipulation t o secure purpose d
resul t s in wh ich a causal relation is disclose d
A ques tion nat urally su gg es t s i t self Wh at is th ere
t o de t ermine th e precise mo d e of experiment
We
may h ave given a concret e wh ole of ext reme com
In our experiment wh ich element sh all
ple xity
w e procee d t o eliminat e in or d er t o no t e th e re
sul t ? A n ans w er may be g iven us th rou gh su g
g estions receive d from th e resul t s of th e meth o d
of a greement wh ich h as been alrea dy applie d t o
th e problem If it is no t possible to avail one s
self of this cont ribution from ano th er sph ere o f in
v e stigatio n th en th e complex wh ole mus t be broken
up as far as possible int o i t s simples t component
part s an d o ne aft er anoth er th ese part s sin gly
th en in pairs and all o th er possible combinations
cause d t o be w i thdraw n, or th eir force neut ralize d,
and th e resul t s in each case not e d, as t o wh eth er th e
effec t un d er inves ti g ation disappears T h e e xh au s
tion o f all possible combinations mus t yiel d some
d enit e result S uppose for inst ance th ere is a
com plex ant ece d ent involvin g four separable ele
Withd raw severally A , B,
m en t s as A , B C D
C and D, no tin g resul t s ; th en w i thd raw in turn
A B A C A D BC BD CD tha t is th e pos
sible combinations of four element s t aken tw o at
a time ; th en withdraw A BC th en BCD A BD
an d A OB, th at is, combinations o f four e lement s
.

INDUCTIVE LOGIC

106

t aken th ree at a time By such a process th ere


will be d isclose d wh eth er one element alone o r
wh eth er a combination o f two o r more h ave pro
d u ce d th e effec t un d er inves ti g ation ; also wh e th er
more th an one element or combin ation o f element s
may h ave cause d th e effec t ? T h e pract ical d if
cu l ty in separatin g th e element s of a complex
wh ole an d wi thd rawin g th e several combinations
from th e whole ren d ers th is process in many cases
impossible T h e cause h o w ever is g enerally su s
I t may be su gg est e d by th e meth o d o f
pec te d
a g reem ent by analo gy o r by th at insi ght wh ic h at
once d eclares certain combinations t o be impossible
an d oth ers irrelevant Th ere is generally a mental
experiment in wh ic h th e j u dgm en t rej ect s unlikely
combinat ions thus narrow ing th e el d of inves ti ga
t ion as preliminary t o th e experiment s proper
T h e meth o d of difference is open t o various o ri ti
c is m s ; th e mos t impor t an t are th e follo win g :
1 In applyin g th is me th o d w e may be s o easily
misled in supposin g ou r tw o ins t ances are precisely
alike wi th th e on e excep tion o f th e presence or
ab sence of th e s u ppose d cause an d yet in reali ty
th e ins tances may differ radically and ye t w e may
be unable t o d et ec t this A pat ient may h ave
me dicine a dminis t ere d to h im , an d be gin at once
rapi dly t o recover an d yet th e very t akin g o f th e
me dicine in i t self may h ave ma de such a ment al
impression in ducin g con d ence an d h ope th at th e
real cause of th e recovery may be due wh olly to
.

Th is pro c ess

a strik ing

h as

manner by

een

d and c iti i d at l
Emp iric al L g ic pp 40 1 ff

ill ust rat e

Venn

c se

e ng th

in

THE ME

TH O D

OF DIFF ERE NCE

10 7

this ment al reaction P ersons t akin g pills c om


pose d of inert subs t ances h ave oft e n g iven evi d ence
of bo dily effe ct s wh olly impossible t o t race t o th e
me d icine i t self A nd yet th is cri ticis m is one o f
cau t ion rath er th an of cens ure fo r th e d efect s are
but difcul ties wh ich ext reme care an d insi ght may
overcome
9
I t h as been obj ect e d th at this meth o d may
point out th e cause in th e concret e inst ance before
th e experiment er but th at this furnish es n o basis
wh at soever for a wi d er g eneralization th at th e effec t
in ques tion is al ways pro duce d by th is cause S ig
w art h as illus t rat e d this obj ection by th e inst ances
in which typh us fever h as been t race d t o th e d rink
in g of impu re w at er ? T h e causal relat ion may be
fully es tablish e d in th e cases inves ti g at e d but th e
universal proposition d oes n o t follo w th at wh erever
typhus feverappears impure w at er h as been drunk
T his obj ec t ion applies especially t o cases of ex t rem e
complexity , wh ere proximat e causes alone can be
discovere d, an d th eir ul t imat e nature wh ich may
appear in various forms is no t reveale d ; for in
s t ance, th e impure w at er is not in i t self th e ultimat e
cause of th e typhus fever It cont ains th e poison
g erms, th e real cause ; th ey may be intro duce d int o
th e sys t em in some oth er w ay
Care th erefore,
s h oul d be t aken t o reveal th e cause in an d by it self
an d no t th e cause of th e cause T h e obj ec tion
To
th erefore m ay be in a measure overcome
effect a g e neralization m oreover o f lo gical vali d
i ty , i t is necessary t o supplement th e me th o d o f
.

Sig wart , Log i

Vo l

p 420
.

IN DUCTIVE LOG IC

108

difference by hypo th esis an d subsequen t v eric a


tio n s whic h will be d escribe d lat er on
3 T h is meth o d may lea d to error in cases wh ere
th e suppose d causal element is re g ar d e d as th e
cau se in its entire ty wh en i t is in reality but a part
O f t h e cause If one sh oul d plant see d in a g ar d en
an d w at er only one h alf of th e plot an d i t S h oul d
follo w th at only th e wat ered part brou ght forth th e
leaf an d o w er th en an inference accor din g t o
th e meth o d of difference mi ght be draw n th at th e
w at er cause d th e sproutin g o f th e youn g plant s
A n d ye t i t mus t be re gar d e d simply as con t ribu t ory
t o th e real cause Suc h a difcul ty may be o bv i
at e d by a careful discrimination in th e analysis of
th e p h enomenon inves ti g at e d
4 S omet imes a liberat in g cause may be reveale d
by a s trict int erpret ation o f th e meth o d of differ
ence wh en th e real cause is more Obscure an d may
be overlooke d A st one may s t rike a can of dyna
mi t e and th e explosion wh ic h occurs may be t race d
I t is th e one element
to th e impac t of th e s t one
of difference int ro duce d in th e sph ere o f th e o b
serve d ph enomena with th e consequent resul t
Th e force exis tin g as a pot ent ial is naturally o h
scure and apt t o elu d e observation T h erefore
wh enever a cause disclose d by th e meth o d o f dif
ference seems t o be o u t of all proport ion t o th e
effec t i t at once su gg es t s th e probabili ty th at a
po t ential force n o t d iscerne d by o ur po w ers o f o b
servation h as been th e real cause an d th e former
a condi tioni n g cause merely A no th er illus t rat ion
of this is th e experiment o f P ries tley which le d t o
.

THE METH O D O F D FF EREN CE

109

h is d iscovery of oxy g en in 17 7 4 He place d some


oxi d e o f mercury upon th e to p o f quicksilver in
an invert e d g lass tube lle d w i th th at me tal an d
s t an din g in mercury ; h e th en h eat e d th e oxi d e by
means of a g lass lens an d th e sun s rays an d o b
tain e d a gas wh ic h h e calle d ni t rous air aft er
w ar d s d esi g nat e d as oxy g en T h e h eat in th is case
w as th e sole element of difference be tween th e tw o
ins tances one in wh ic h th ere was no g as an d th e
second aft er application of th e h eat wh en th e gas
w as presen t
Here th e h eat mus t be re g ard e d as
th e liberatin g an d n o t in any sense th e pro ducin g
cause A g ain as L o t ze says , th e fac t th at w i th
th e d es t ruction of a sin gle part of th e brain a d e
ni t e psych ical func t ion ceases is no proof th at j u s t
th is sin gle part w as th e or gan wh ich alone pro
d u c e d th at func t ion 1
In a ddition t o th e difcul ties att en din g th is
meth o d wh ic h h ave been en umerat e d and wh ich
h ave to d o w ith th e lo gical th eory o f th e meth o d
th ere are also d i fcult ies of a practical nature which
arise in th e act ual application of th is meth o d in ex
T h ey are as follow s
perimental inquiry
1 Care mus t be t aken th at in th e two p h e
n o m ena compare d , w i th an d w i th ou t th e suppose d
cause th ere s h all no t be an int erval o f time elaps
in g, in which perio d some o th er cause mi ght be
int ro duce d unk n o wn t o th e inves ti g ator, an d yet
capable of pro ducin g th e res ult, or else of neu traliz
in g some force th at is present an d it self capable of
ro
d
ucin
g
th
e
resul
t
F
or
ins
t
ance
if
a
c
h
emical
,
p
.

Lo tze , Log i

p 322
.

INDU CTI VE

110

O GIC

compoun d be left fo r an appreciable time w e may


no tice cert ain ch an g es an d be able t o asser t posi
tiv e ly that no n e w element h as been int ro d uce d
an d yet th e ac tion O f th e air may in i t self h ave
been sufcient t o w ork th ese ch an g es W h en th e
two ph enomena to be compare d c an be present e d
for inspection simul t aneously this d i fcul ty is o bv i
at e d T his is illus t rat e d in an experiment d evise d
t o exhibit th e presence of li gh t effec t s in th e
spect rum beyon d th e violet rays ; th at is beyon d
A
th e place wh ere th e spec t rum seems to en d
sh eet o f paper is t aken th e lo w er part o f wh ic h is
moist ene d with a solution o f sulp h at e of quinine
while th e upper part remains dry L e t th e imag e
o f th e solar ray fall upon th is s h ee t ; th e spec t rum
preserves at th e to p o f th e sh ee t in th e d ry portion
o f th e paper its or d inary appearance wh ile in th e
mois t ene d portion a brilliant ph osph orescence ap
pears beyon d th e region O f th e violet rays Here
th e d ry an d w et portions are simul t aneously pre
sent e d, an d th ere is but o n e point o f difference be
tw een th e tw o Th e inference th erefore is rea dily
drawn th at th e solu tion o f sulph at e o f quinine is a
subs tance sensi t ive to th e ul tr a violet port ion of th e
sun s rays th e ph osph orescence being th e effec t of
th ese rays upon th e solution
2 E xt reme care mus t be t aken th at, in th e w i th
d raw in g O f any elemen t in th e course of th e ex peri
ment no oth er elemen t is inadvert ent ly intro duce d
an d th at in a ddin g any element no exis tin g elemen t
o r combinat ion of elemen t s is d es troye d
or th eir
effect neut ralize d Mr V enn h as a dmirably illus
,

THO D

THE ME

trate d

IFFEREN CE

111

OF D

th is difculty an d I give th e follo win g


quo t at ion in full from him : We suppose th at
wh en w e h ave put a w ei ght int o one pan of a pair
of scales w e h ave d one nothin g more th an this , o r
can at any rat e by due cau tion succee d in d oin g
no th in g more But if w e exact th e utm ost ri gi d ity
O f con d itions w e easily see th at w e h ave d one a
g reat d eal more O ur bo dies are h eavy, an d th ere
fore th e mere approac h t o th e machine h as al tere d
th e ma g nitu d e an d d irec tion of th e resul t ant attrac
tion upon th e scales Our bo dies are presumably
warmer th an th e surroun din g air ; accor din gly w e
w arm and th erefore li ght en th e air in wh ich th e
scales h an g an d if th e tw o sc al es an d th eir con
t ent s are not of th e same volume w e at once al t er
th eir w ei ght as measure d in th e air O ur breath
pro duces dis turbin g current s of air O ur approach
affect s th e surface of th e n o n ri gi d oor or g ro un d
on whi c h th e scales s tan d an d pro duces ano th er
source of di s turbance an d so on th i o u gh th e wh ol e
ran g e o f th e p hysical forces 1
In th e R eport of th e Brit is h A ssociation 18 81 an
accoun t is given o f P rofessor G H Dar win s exper
imen ts t o measure th e lunar dis t urbance of g ravi ty
at th e Caven dis h L aborat ory by means of an ex
I t w as foun d th at
trem ely d elicat e pen du lum
approac hin g th e pen dulum in ord er t o observe it s
readin g, th e surface level o f th e s tone oor on
wh ic h th e ins trument s t oo d w as d e ec t e d by th e
w ei ght o f th e O bserver A S h e s too d t o t ake th e
rea din g, th e s h ift ing o f his w ei ght from one le g t o
,

'

Venn

'

Emp iric a l Log ic , p 416


.

INDUCT IVE

112

O GI C

th e o th er was percep t ible ; s o it became necessary


t o observe th e rea din g by a t elescope from a dis
?
t ance or t o a dop t some similar plan
F arad ay w as able at will t o pro duce or remove
a magne tic force th rou gh th e reveale d properties
Many of h is experimen t s
o f th e elec t roma g ne t
woul d h ave been impossible if i t h a d been n e c e s
sary to remove a cumbersome ma gnet an d reins t at e
it ag ain an d ag ain in his experiment s Th e electro
m a g ne t h o w ever coul d pro duce or d es t roy th e
presence o f magnetic force w ith out any inci dental
perturbat ion s T hus F ara d ay w as enable d t o prove
th e ro tat ion of circularly polarize d li ght by th e
fact th at certain li ght cease d t o be visible wh en
th e elec tric c u rrent o f th e ma g ne t w as cu t o ff,
an d ins t antly re appeare d wh en th e current was
F arad ay says o f th e experiment
re es t ablis h e d
himself : T h ese p h eno m ena coul d be reverse d at
pleasure and at any ins tant o f time an d upon any
occasio n sh o win g a perfec t depen dence of cause
2
an d effec t
3 In so me cases i t is impossible t o remove an
element which is suppose d to be th e cause o f an
effec t un d er invest i g atio n Its removal m i ght cau se
th e d es t ruc tion o r th e impairin g of th e wh ole ph e
n o m en o n
T h e force th erefore th at cannot be
eliminat e d must be neu tralize d by an equal an d
opposin g force Fo r ins t ance th e force o f gravity
cannot be eli minat e d ; i t mus t th erefore be coun
terbalan ced by some d evice of th e inves ti g at or
,

Q uo te d b y Venn in E mp iric al Lo g ic , p 419


2 E x eri m en tal Research es in E le c tri ci t
p
y , Vo l III p 4

THE

METH O D O F DI FF ERENCE

113

In ch emis t ry th e removal of an element from a


compoun d may be impossible w i th ou t d est royin g
u tt erly th e compoun d i t self ; in such a case also
a neut ralizing a g ent mus t be int ro duce d Dar win
wis h e d t o prove th at th e o dor o f o w ers is att rae
t ive t o insect s irrespective o f th e att ract ion of
color He th erefore covere d cert ain o w ers w i th
a muslin net, whic h s till att ract e d th e i n sect s to
th em ?
T h e follo win g illus t rations may serve furth er
t o exhibi t th e various features of th e me tho d of
,

d ieren ce :

Mr R ober t Malle t g ives th e followin g int eres tin g

I t mus t be n o w
acco un t of his visi t t o F ara d ay :
ei ght een years a g o wh en I pai d him a visi t an d
brou ght some S lips o f exible an d to ugh Mun tz s
yello w met al t o S h o w h im th e ins t an t aneous
ch an g e t o com plet e brittleness w ith ri gi dity pro
d uced by d ippin g int o perni t rat e of mercury so
He g o t th e solution an d I sh o w e d h im
lu tio n
th e fact s ; h e obviously di d n o t d oub t w h at h e
s aw me do before an d close t o h im ; but a sor t
ins tinc t seeme d to require h e
o f experiment al
s h oul d try i t himself S o h e t ook o n e of th e
slips bent i t for war d an d back w ar d d ippe d i t
an d broke i t up in t o s h ort bi t s betw een his o wn
He h a d not before S poken
T h en h e
n g ers
sai d Ye s i t is pliable an d i t d o es becom e in

2
Here th e experiment with and
n
l
sta t y bri tt le
w ith out th e si gnicant ant ece d ent an d consequent
.

1
2

Darwin , Cros s a nd S elf Fertilizatio n , p 374


.

Glad sto ne

Mich ael F arad ay , p

175

I NDU C TIVE

114

O IC

L G

resul t in dicat es th e causal relat ion especially as


th e ins t antaneous effect preclu des th e possibility
of th e operation of any oth er cause
A noth er experiment O f F arad ay s is th at of h is
inves t i g ation of th e beh avior of L ycopo dium po w
d er on a vibrating plat e I t h ad been observe d
that th e minut e particles o f th e po wd er collect e d
t o g eth er at th e point s of g reat es t mo tion wh ereas
sand and all h eavy particles collecte d at th e no d es,
wh ere th e mo tion was leas t I t occurre d t o F ar a
d ay t o t ry th e experimen t in th e exh aus t e d re
c eiv er o f an air pump an d i t w as th en foun d th at
th e li ght po wd er beh ave d exac tly like h eavy po w
d er Th e inference w as th at th e presence of air
w as th e con di tion o f import an ce because i t w as
th row n int o e ddies by th e motion o f th e plat e
an d carrie d th e Lycopo dium po wd er to th e point s
o f g reat es t a g it at ion
S and w as t oo h eavy to be
carrie d by th e air ?
S ir J oh n L ubbock gives an accoun t o f ex peri
m en t s performe d upon insec t s to prove th at th e
se n se o f smell is in some w ay connect e d with
th eir ant enn ae
O ne experimen t was performe d
by F orel wh o remove d th e wing s from some bl ue
bottle ies an d place d th em near a d ecayin g mole
Th ey imme diat ely w alke d t o it , an d be g an lickin g
it an d layin g e gg s
He th en t ook th em away an d
remove d th e ant enn ae all oth er circ mn stances re
mainin g th e same as before aft er wh ich , even
wh en place d close t o th e mole th ey di d no t ap
pear to perceive it Anoth er experiment similar
,

Je

v o ns ,

Pri n cip les of S c ien c e

p 4 19
.

THE ME

TH OD O F

DI FF EREN C E

115

t o th is w as t rie d by P lat eau , wh o put som e foo d


o f wh ic h cockroac h es are fon d on a t able
and
surroun d e d i t with a lo w circular w all of car d
boar d
He th en put some cockroach es on th e
t able ; th ey evi d ently scent e d th e foo d, an d ma d e
s t rai ght for i t He th en remove d th eir ant enn ae,
aft er wh ich , as lon g as th ey coul d n o t see th e
foo d, th ey faile d t o nd it even th ou gh th ey
?
w an d ere d abou t quit e close to it
A n o th er experiment is th at of G raber t o prove
th e sense o f h earin g in ins ect s He place d some
w at er boat men ( Co rixa) in a d eep j ar full O f w at er
at th e bo tt om of wh ich w as a layer of mud He
droppe d a s t one on th e mu d but th e beetles wh ich
w ere reposin g quiet ly o n some w eed s, t ook no
no tice He th en put a piece o f glass o n th e mu d,
an d droppe d a s t one on t o it, thus makin g a noise,
th ou gh th e dis turbance o f th e w at er w as th e sam e
Th e
as wh en th e s t one w as d roppe d o n th e mu d
w at er boat men h o w ever th en at once t ook i ght ?
A n illus t rat ion of th e me th o d of difference occurs
in th e s o calle d blind exp eriments wh ich are O ft en
ma d e in c h emis t ry especially A s P rofessor J evons V
h as d escribe d suc h an experimen t : S uppose, for
ins t ance a ch emis t places a cer t ain suspec t e d sub
st ance in Mars h s t es t apparatus an d n d s th at it
g ives a small d eposi t o f me t al lic arsenic h e cannot
be sure th at th e arsenic really procee d s from th e
suspec t e d subs t ance ; th e impuri ty o f th e zinc or
.

bb ck

Lu

mals , p
2

45
o

0 n the S enses , I ns tin ts ,

0 n S enses , etc

bb ck

Lu

p 75
.

and

ce of Ani

In tellig en

INDU CT IVE

116

O G IC

sulphuric aci d may h ave been th e cause of its ap


I t is th erefore th e practice o f ch emis t s
pearan ce
to make wh at th ey call blin d experimen t s , th at is
t o t ry wh eth er arsenic appears in th e absence of th e
suspect e d subs t ance T h e same precau tion ou ght
t o be taken in all import ant analytical Operat ions
Ind ee d i t is no t m erely a precaution it is an essen
tial part of any experiment If th e blin d t rial be
no t ma d e th e ch emis t merely assumes th at h e
1
kno w s wh at w oul d h appen
.

Je

v o ns ,

Prin cip les of S cienc e

p 43 3
.

CH AP T ER I"
THE J O IN T

M E TH O D

OF

FE

IT

A G RE EM E N T

AND

DIF

RE N CE

already been Sh ow n th at th e meth o d of


d ifference is sometimes not available inasmuch as
it may be neith er poss ible n o r prac t icable to remove
from th e ph enomenon t o be inves ti g at e d th e su s
pected causal element wi th ou t d es t royin g th e ph e
n o m eno n i t self
S ome t imes t oo i t is i m possible
even t o neut ralize th e e ect o f th e causal element
if it is allo w e d t o remain as an int e g ral part of th e
ph enom enon T h is is especially th e case in all
vit al ph enomena an d also in many ch emical ph e
n o m en a
T h erefore anoth er m eth o d is resort e d t o
whic h is kno wn as th e j oint meth o d of a greement
an d di fference Inasmuch as th e suspec t e d causal
element canno t be remove d w e must select ano th er
ph en omenon as much like th e former as possible
which is h o w ever ch aract erize d by th e absence of
th e causal elemen t By th e simple meth o d o f dif
ference tw o ins t ances only nee d be compare d, th e
one wi th an d th e oth er w i th ou t th e causal element,
bu t ag reein g precisely in every oth er particular
In th e j oint meth o d, th e ins t ances wi th an d w ith out
h as

117

118

IN DU CTIVE

O GIC

th e causal element d ie r it may be in several par


A number of varyin g ins t ances mus t
tic ul ars
th erefore be selec t e d so as to eli m inat e th e possi
bility of any of th ese differin g Ch arac t erist ics bein g
th e cau se in ques tion T h erefore two set s of in
s t ances are collec t e d, an d compare d T h e one set
comprises all th e positive inst ances h avin g th e pres
ence of th e suppose d causal element an d th e secon d
set consis t s o f th e ne g a tive ins t ances h avi n g th e
suppose d causal element absent al t o g eth er T h e
vali dity of th e meth o d d epen d s upon th e similarity
of th e two set s o f ins tances A s th e similari ty
increases th e meth o d approximat es t o th e si mple
meth o d o f difference
The Can o n of the J o int M etho d If several
ins t ances in whic h th e ph eno m enon occurs h ave
only one circums t ance in common while several
ins tances in wh ic h it does no t occur h ave nothin g
in common save th e absence of th at circum
st ance ; th e circums t ance in wh ic h alone th e two
set s of ins t ances differ is th e effect or cause or a
necessary part o f th e cause o f th e p h enomenon
Th e symbolical represent ation of this meth o d
may be ex hibit e d as follo w s , usin g a similar nota
tion t o th at employe d in th e previous meth o d s :
,

I Table
.

S
S

'

of

+ 0
+ C

+ 0
m
S
+ 0

posit ive ins t ances

et c

ME TH OD

II

O F A GREEMENT

IFFERENCE

A ND D

Table of ne g at ive inst ances

119

St

S.
8

e tc

1it

etc

In th e tw o set s O f ins t ances , th e follo win g con


d itio n s mus t be observe d in or d er to ren d er th e
meth o d vali d :
1

S + 0,

~
m
s + C, s + O, s + 0 ,
'

etc ,
.

mus t be s o varie d th at th ey reveal but one cons tant


element, common t o th em all , as C I t may be th at
S wi ll resem ble S ' in more marks th an th e o n e,
namely C an d this may be t rue of any tw o or more
ins t ances ; h ow ever t aken all t o g eth er th ey mus t
posses s but th e o ne common element, 0
2 In th e same w ay S , may resemble S in more
marks th an merely th e absence of C and so fo r any
et c
tw o or more ins t ances in th e series S , S
Ho w ever, th e o ne charac t eris tic com m on t o th em all
mus t be th e absence of C
3 If in th e ins tances c h osen an element is com
mon to all in a dd ition t o C or in th e secon d set i t s
absence th en a ddi tional ins tances mus t be adde d t o
th e t ables both posi tive an d ne g ative in or der t o
secure th is all impor tant condi tion of elimination
th rou gh sui t able variation
4 Moreover, th e tw o series posi t ive an d ne g at ive ,
must h ave th eir settin g s similar S , S
et c ,
mus t resemble S S S
et c o th erwise th e ne ga
tive ins t ances w oul d no t be si gnicant T h ey mus t
.

INDUCT IVE

120

O GIC

be ch osen from th e same sph ere as th e positive, th at


th ey may be similar It is possible t o mul tiply ne g
ative inst ances ad innitu m, which h o wever, w oul d
furnis h no g roun d fo r any inference because th ey
woul d be wh olly irrelevant t o th e problem un d er
inves ti g ation
5 If S , i s so si milar t o S as t o be i d entical w i th
it an d also 3 , pass over int o
th en w e h ave th e
meth o d of difference in its pure form :
.

'

S + O
S'

e
3

'

Here th e setting ins t ead of bein g similar in th e tw o


cases is th e same in eac h
T h e follo win g is an experiment o f Sir J oh n Lub
bock s concernin g th e sense o f smell in insect s , wh ic h
I h a v e ch osen as illust ratin g th is meth o d of in duc
tive research He t ook a lar g e ant an d t eth ere d
h er on a boar d by a th read Wh en sh e w as qui t e
s till h e brou ght a tunin gfork int o close proximity
to h er ant enn ae, but sh e w as not d is turbe d in th e
least He th en approach e d th e feath er o f a pen
very quietly, s o as almos t t o t ouch rst one an d
th en th e oth er of th e ant enn a which h o w ever d id
no t move He th en d ippe d th e pen in th e essence
of musk an d di d th e same ; th e ant enna w as slow ly
ret ract e d an d drawn qui t e back He th en repeat e d
th e same w ith th e o th er ant enna an d with like
resul t Care w as t aken th rou gh out n ot t o t ouch
th e ant enn ae L ubbock th en repeat e d th e experi
ment with a number o f d ifferen t ant s an d usin g
various subs tances T h e resul t s in all cases w ere
,

METHO D

O F A GREEMENT

A ND D

IFFERENCE

121

th e same and th e inference w as naturally d rawn


th at th e ant ennae possesse d th e sense o f smell In
th ese experiment s , various subs t ances w ere t aken
h avin g noth in g in common save th e o dor o f musk
th at h a d been place d upon th em
In so m e cases i t is n o t possible t o discover posi
t ive ins t ances in whic h th e only common eleme nt
is th e suspec t e d cause
In such cases th e meth o d
is n ot conclusive in its resul t s al th ou gh i t may
att ain a h i gh d e g ree of probability if all th e com
mon element s save th e suspec t e d cause element
are kno wn to be irrelevan t, o r can in any oth er way
be prove d t o h ave no i n uence wh at soever upon th e
result F or ins t ance an illus t ration is oft en given
of this me th o d wh ich fails in th e manner jus t d e
scribe d A man is att emp tin g t o d iscover wh eth er
a part icular art icle of foo d d isag rees w ith him
He no tices several occasions a larg e num ber if
you please, wh en h e h as eat en this particular kin d
o f foo d , an d h as soon aft er experience d dis t ress
T h ese are th e posit ive ins t ances T his peculiar
d is t ress h as never been experience d wh en h e h as
abs t aine d from th e foo d in ques tion T h e inference
is th at th is foo d h as cause d th e dis t ress In th e
various ins t ances, h o w ever, th e sole element in
common is no t merely th e t aking o r n o t t aking th e
foo d Th e person s wh ole bo dily org anism is com
m o n t o all th e ins t ances Wi th in i t un foreseen
complications in depen d ent o f this article o f foo d
mi ght h ave cause d th e t rouble In such cases
number o f ins t ances mus t be resort e d t o in or d er t o
ren d er th e possibility o f a coinci d ence impossi ble
,

INDU C TIVE

122

LO

GI C

So

also in such cases as th e t reat ment of any given


d isease in a h ospit al A n experiment may be t rie d
in th e treatment , say, of typh oi d fever One w ard
may be subj ect e d t o a particular kin d of t reatment
an d anoth er war d not subj ec t e d t o th at t reat ment
If recovery is h ast ene d in th e o ne and retar d e d
in th e oth er case, an inference may be d rawn as
t o efcacy Of this treat ment In th ese ins t ances
ag ain while th ey are all different patient s, st ill th e
nursin g, surroun ding s etc , are common t o th em
all
I t mus t be S h o wn th at th ese are present
both in th e ne g ative an d positive ins t ances , an d
equally capable of accomplis hin g th e effect if th ey
h ad been real causes
T h ey may th erefore be
eliminate d in comparin g th e two set s o f ins t ances
because common bo th t o th e n egative an d posi tive
cases In this also resort must be h ad to th e num
ber of ins t ances in or d er t o eliminat e ch ance coin
c id en ces
T h e presence o f common element s in
excess of th e common causal element may be rep
resent e d ac cor ding t o th e symbolical not ation of
th e joint meth o d, by th e int ro duc tion of anoth er
symbol at L e t a: s t an d for th at which is common
t o all ins t ances in addit ion t o th e comm on element
C
We th en h ave :
.

,
.

S et

of positive in s t ances

+ C+ m
+ C+ w

'

il l

+ C +
1

+ 6 +

et c

HI

METHO D O F A GREEMEN T

II

S et

AND D FFERE

of ne g ative ins t ances

NCE

S,

a:

Sn

SM

etc

123

,
11

etc

We observe a in all ins t ances both posit ive an d


ne g at ive Bein g present wh en th e effect occurs
an d wh en it d oes not, in differen tly, w e can at once
infer th at x is no t th e wh ol e cause of e Ho w ever
it may h ave unit e d wi th C in th e rs t set of ins t ances
to pro duce th e effec t c so th at C w i th ou t a), or some
part o r part s of a, coul d n ot alone pro duce th e effect
e
In all suc h cases th e exact force of a: mus t be
estimat e d in some oth er w ay If a: is ex tremely
complex or subj ect t o ch an g e from forces emanatin g
from w ithin i t self, as in th e case o f or g anic ph enom
ena th en i t becomes ex t remely difcul t t o d et ermine
a ; an d consequent ly th e me th o d of a g reeme nt an d
A s lon g
d ierence d oes n ot yiel d as exac t resul t s
as th e force of a remains un know n i t becomes th e
source O f possible dis turbance, wh ic h may wh olly
vi tiat e th e resul t s att ai ne d
Mr Dar win, in h is experiment s upon cross an d
self fertilizat ion in th e v e g e t able kin gd om , place d
a net abou t one hun dre d o w er h ead s , thus prot ect
in g th em from th e bees an d from any ch ance of
fertilization by means of th e pollen conveye d t o
th em by th e bees He at th e same time place d one
hun d re d oth er o wer h ea d s o f th e same variety of
plant wh ere th ey w oul d be expose d t o th e bees an d,
as h e observe d, w ere repeat e dly visit e d by th em
.

INDU CT IVE

124

O GIC

Here w e h ave th e tw o se t s o f ins t ances, in one th e


o wers accessible to th e bees and in th e oth er, n o t
accessible He obt aine d th e follo win g resul t T h e
prot ec t e d o w ers faile d to yiel d a sin gle see d T h e
o th ers pro duce d 6 8 g rains w ei ght of see d wh ic h h e
est imat e d as numberin g 2 7 20 see d s Cross fertiliza
tion as th e cause in th is case is thus prove d T h e
comm o n element in all th ese ins t ances , h o w ever, is
not merely th e presence in o ne case an d th e absence
in th e oth er of th e bees ; th ere is also th e element
of th e common plant s t ruc ture runnin g th rou gh all of
T his element is, h o w
th e tw o h un d re d ins t ances
ever o f such an unvaryin g nature in all th e ins t ances ,
an d th e number observe d so many as t o eliminat e
th e possibility of any given plant s t ructure posses s
ing unobserve d peculiarities sufcien t t o pro duce th e
resul t in ques tion I t may th erefore be consi d ere d
as an inert elemen t as re g ar d s th e effect s no tice d in
th e one an d absent in th e oth er set of ins t ances
S ir J oh n L ubbock in h is research es concerni ng
th e different functions of th e tw o kind s of eyes in
insec t s illus t rat es th e j oint meth o d in it s g eneral
features T h e tw o kind s of eyes are th e lar g e c o m
poun d eyes si tuat e d one o n eac h S i de o f th e h ea d
an d th e ocelli o r small eyes , of wh ich th ere are
g enerally th ree arran g e d in a t riangle betw een th e
o th er two He wis h e d to d et ermine th e precise
function o f th e small eyes th e ocelli ; an d h e h as
g ath ere d t o g eth er th e follo win g fact s P lat eau h as
s h o wn th at cat erpillars wh ic h possess ocelli but no
compoun d eyes are very s h ort S i ght e d no t seein g
a bove one t o two centimet res He h as also prove d
,

METHOD

O F A GREEMENT AND DIF EREN CE

125

by experiment s th at spi d ers wh ich h ave ocelli but


n o compoun d eyes are very s h ort si ght e d ; th ey w ere
easily d eceive d by art icial ies o f mos t inar tis t ic
cons t ruc t ion and even h untin g spi d ers coul d no t
see beyond t en centimet res (four inc h es) L ubbock
experiment e d o n th is point wi th a female spi d er,
which , aft er layin g h er e gg s h a d rolle d th em int o a
ball an d h ad envelope d th e wh ole w ith a silken ba g
wh ic h s h e carrie d about w ith h er Havin g capture d
th e female an d h avin g t aken th e ba g o f e gg s from
h er , h e place d it on a t able about two inc h es in
front o f h er Sh e evi dently di d not see it He
th en push e d it g ra dually t o w ar d s h er, but s h e t ook
no not ice t ill i t nearly t ouch e d h er, wh en s h e eag erly
seize d it He th en t ook it aw ay a secon d time an d
put it in th e mi dd le of th e t able, which was two
feet four inch es by one foo t four, an d h a d noth ing
else o n it T h e spi d er w an d ere d about for an h our
an d fty minut es before sh e foun d it, apparently
by acci d ent He th en took it aw ay ag ain an d put
it d o wn as before, wh en s h e w an d ere d about fo r an
h our w ith out n din g i t L ike experiment s w ere
t rie d w i th oth er spi d ers an d w ith th e same resul t s
Plat eau also experiment e d w ith scorpions wh ich
h a d oc elli an d no compoun d eye s T h ey appeare d
scarcely to see beyon d th eir o wn pincers More
over, th e ocelli are especially d evelope d in insec t s ,
s uc h as ant s bees an d w asps, wh ic h live par tly in
th e open li ght an d part ly in th e d ark recesses of
nes t s Ag ain, th e ni ght ying mo th s all possess
ocell i O n th e oth er h and h o wever, th ey are en
ith, accor ding
tirely absent in all b utt er ie s w
,

12 6

I NDU C T IVE

LO

GI C

t o S cu dd er but one excep t ion namely th e g enus


F ore" moreover varnish e d th e co m
Pamph ila
poun d eyes o f various insec t s wh i c h h ad ocelli as
w ell Th e latt er, ho wever h e allo we d to remain in
th eir nat ural s t at e I nasm uch as th eir h abit s of
ight require d po w ers o f vi s ion in th ese inse c t s
ext en din g to a con s i derable dis t ance i t h appene d
th at wh en place d o n th e groun d th ey ma d e no at
t emp t t o rise ; while if th ro wn int o th e air, th ey
e w r s t in o ne d irec tion an d th en in ano th er,
s trik in g ag ains t any obj ect th at cam e in th eir w ay,
and bein g apparently quit e unable t o gui d e th em
selves T h ey e w repeat e dly a g ai n s t a w all fallin g
t o th e groun d an d unable t o alight a g ains t it, as
th ey d id s o cleverly wh en th ey h a d th eir compoun d
eyes t o g ui de th em A ll th e s e in st ances t aken
t o g eth er in th eir positive an d ne g ative aspec t s , le d
S ir J ohn L ubbo c k to infer th at th e ocelli w ere u s e
ful in d ark place s an d for near vision, while th e
compoun d eyes were for th e light an d more dis t ant
vi s ion 1
A noth er illus t ration of th is meth o d may be foun d
in D arwin s account of th e ex t reme t amene ss of th e
bir d s in th e G alapag os an d F alklan d islan d s I
quot e some ex t rac t s from h is narrative in wh ic h it
will be seen th at D arwin s inferenc es follo w from
h is comparis on o f th e po sit ive an d ne g at ive ins tance s
before him He s ay s : T his t amenes s of dis po
sitio n is common t o all th e t erre st rial species o f
th e s e islan ds in th e G alapag o s A rchipela g o ; namely
,

bb o ck ou
pp 175 ff

Lu

mals ,

th e S en ses , I ns tin c t, a nd I ntellig en ce of A ni

METHOD O F

GREEMENT

AND

DIFFERENCE

12 7

t o th e mockin g th rush es th e nch es w rens tyrant


-

yc atc h ers,

th e d ove an d carrion buzzar d A ll of


th em oft en approach e d suf ciently near t o be kille d
with a s w it ch an d somet imes as I myself t rie d,
w ith a cap o r h at A gun is h ere almos t super
u o u s ; for w i th th e muzzle I pu sh e d a h aw k O K
th e branc h o f a t ree I n Ch arles I slan d wh ic h h ad
been coloni ze d about six years I saw a boy sitt in g
by a w ell w ith a s wit c h in h is h and w ith wh ic h h e
kille d th e d oves an d nch es as th ey c ame to d rink
He h a d alrea dy procure d a little h eap o f th em fo r
h is d inner ; an d h e sai d th at h e h a d cons tantly been
in th e h abit of w aitin g by th is well for th e s ame
purpose T h e F alklan d I s lan d s offer ins t an c es of
bir d s w ith a similar dispo sition T h e snipe uplan d
an d l o wlan d g oose th rush bun t in g an d even s o m e
t rue h a wks are more or less t ame T h e blac k
necke d s w an is h ere w il d an d i t w as impo ss ible t o
k ill it I t h o w ever is a bir d of passag e w hi c h
probably brought w ith i t th e wis d om learne d in
forei gn coun t ries
F rom th ese several fact s w e may I th ink con
c lu d e th at th e w il d ness of bir d s w ith re g ar d t o
man is a part ic ular ins tinc t d irect e d a g ain s t h im
and no t d epen d ent on any g eneral d e gree o f c aut ion
arisin g from oth er sourc e s of d ang er ; se c on dly
th at it is no t acquire d by in dividual bir d s in a
s h ort time, even wh en mu ch persecut e d but th at
in th e course o f successive g enerat ions it bec ome s
h ere d it ary
W ith d omes ticat ed animals w e are
ac cus t ome d to see ne w ment al h abit s or ins t in c t s
acquire d an d ren d ere d h ere dit ary but w ith animals
-

INDU CTIVE

128

O GIC

in a s t at e o f nat ure i t must al w ays be mos t dif cul t


t o discover ins t anc es of a c quire d h ere dit ary kno wl
e dg e In re g ar d t o th e w il dness o f bird s t o ward s
man th ere is no w ay of acc ounting fo r it ex c ep t as
an inh erit e d h abit : co mparatively fe w youn g bir d s
in any o n e year h ave been injure d by man in
En glan d, yet almo st all even nes t lin g s are afraid
o f him ; many in d iv id uals o n th e o th er h an d, bo th
at G alapag os an d at th e F alklan d s h ave been pur
sue d and injure d by h im but yet h ave not learne d
1
a s alut ary drea d o f h im
I h ave given this quot at ion somewh at at len gth
in or d er t o sh o w th e meth o d of a g reat inve s ti g at or
in th e realm of nat ure ; an d th at it may be s een
h o w nat urally h e falls int o th e meth o d of c ompar
in g posi t ive an d ne g ative s e t s o f ins t ances relative
t o th e obj ect of research T h e animal an d v e ge
t able k in gd oms are espec ially a d apt e d t o th e appli
c at ion o f th is j o int me th o d an d th erefore i t is in
biolo gy th at it is most frequently employe d an d
wh ere it h as yiel d e d th e mos t fert ile result s
T h e advant ag e of th e j oint meth o d over th e sim
ple meth o d o f ag reement is th at it lar g ely cl imi
nat e s th e pos sibil ity of th ere bein g any oth er c au s e
of th e giv en ph enomenon than th e o n e d is c lose d
by th e operation of this meth o d T h e meth o d of
a greemen t, as w e h ave seen oft en fails o f a d e ni t e
re s ul t o win g t o th e plurality o f causes T h e j oint
meth o d t en ds t o ind ic at e th e one an d o nly cause
an d wh en th e in st ance s are rig orously s elect e d ao
c or d in g t o th e con dit ion s of th e canon th ere is a
,
.

Darwin , Voyag e of

N atu ra lis t, Vol II


.

pp

172 f

METH OD O F

A GREEMENT

AND

DIFFERENCE

129

h i gh d e g ree o f probability th at th e sole cause is


d iscovere d Mr Mill at th is point claims t oo much
for th e meth o d in insis t ing th at it gives a cert ainty
re g ar d in g th e sole cause wh en th e requiremen t s
are perfec tly realize d It is imposs ible t o realize
th e requirement s perfec tly I n select in g th e ne ga
t ive in st ances w e are never sure th at w e h ave
compas s e d th e e nt ire sph ere of s ign icant ne g at ive
ins tance s We may h ow ever att ain res ul t s h i ghly
probable in th i s re gar d th ou gh th ey may not re ac h
an absolut e cert ainty S uc h a s t at emen t is more
mo d erat e in it s expression, an d pract ic ally it assures
as sat isfact ory resul t s
.

CH AP T ER "
THE

ME T HO D

OF

CO N CO M I T AN T VA RI A T I O N S

meth o d o f c oncomit ant variat ions is a


pro c ess of det erminin g a causal relat ion wh en, as
an element in an ant ece d ent varie s in int ensi ty
g reat er or les s th ere is O bs erve d a c orres pon d in g
o r conc omi t ant variat ion in th e c onsequen t
TH E

Can o n of the M etho d of Co nco mitant Variatio ns

Wh at ever ph enomenon varies in any manner wh en


ever anoth er ph enomenon varies in so m e part icular
is eith er a c ause or an effec t of th at p h enomenon , or
is c onnect e d with it th rou gh s o m e fac t o f causat ion
T h e latt er clause of this canon provi de s for th at
circums t ance in wh ic h th e varyin g element s may
both be concom it a nt eects of a common caus e
W h en w e are assure d o f th e absence o f any possi
ble common caus e to wh ich w e c an ass ign th e tw o
p h enomena as effe c t s th en th ey mus t be relat e d
betw een th emselves as cause an d effec t A simple
illust ration of th is meth o d is th e rise o f th e mer
cury in th e th ermomet er o win g t o th e inc rease of
h eat ; i t s fall wh enever th ere is d ecrease o f h eat
O ne varies as th e o th er concomit antly , an d w e infer
a caus al relation th at w e at once procee d to gen
e ralize w i th out h esi t at ion
,

130

MET HOD OF CONCOMITA NT VARIATIONS


T h e symbolical representation
as follo w s :
S + O i d0

e tc

of

13 1

th is meth o d

e ;t

etc

is

de

T h en 0 is th e cause o f e
I h ave use d d 0 , an d d c to denot e th e increment s
o r d ecremen t s of th e cause an d effec t respec t ively
T h is me tho d is use d g enerally wh en th e meth o d o f
d i eren ce is impo s s ible, o w in g to th e fac t th at th e
suppose d causal element canno t be made t o vanish
wh olly In all suc h cases a variat ion o f th e ele
ment is resor t e d t o an d th e correspon d in g re s ul t
observe d Heat is relat ive an d not absolut e as also
th e h ei ght of mercury in th e t ube T h e relation
is d et ermined, th erefore by variations, g reat er an d
le ss Th is me th o d is al s o use d to s upplement th e
re s ul ts o f o th er me th o d s by wh ic h a causal relation
h as been d et erm ined but not in exac t qu antit ative
t erms It may be kno wn th at a c ert ain ph en om e
non C is al way s th e cau s e o f a cert ain e ffe c t e an d
th e meth o d o f c oncomit ant variat ions w ill th en be
o f use in d et erminin g precisely h Ow m uc h o f a vari
ation in 0 w ill c au se a speci e d variat ion in e A
la w n ds scien ti c expression only wh en st at e d in
t erms o f exact quantit at ive relat ion betw een varia
t ion s in ant ece dent an d consequent We may ex
pres s th e la w o f uni versal a tt rac t ion in a v a gue way
th at bo dies al w ays attrac t each oth er and th e great er
att ract ion wh en th e bo d ies are nearer to g eth er, an d
th e lar g er th ey are But th is s t at ement nee d s to
.

13 2

INDU CTIVE

O GIC

be recas t in t erms exh ibit in g th e precise quan tita


t ive variat ion Bo dies attrac t eac h o th er direc tly
as th e pro d uc t of th eir masses an d inversely as th e
square of th eir dis t ance It is evi d ent th at th e
special func tion of th is meth o d o f concom it ant vari
atio n s consis t s in j us t th is quan t it at ive d e t erm in a
t ion I n o ne respe c t th erefore, it may be re g ar d ed
an d
as a s ubs t it ut e fo r th e me th o d o f d i fference
in anoth er w ay as a supplement t o the me th o d of
differenc e in lea din g t o quanti tatively d et erminat e
resul t s
T h e quant it at ive variat ion bet ween an t ece dent
an d c onsequent may be eith er direc t o r invers e vari
ation T h e former is wh en o n e increases as th e
o th er increa s es, o r wh en o ne d ecrease s as th e o th er
d ecrease s Th e inverse is wh en o ne d ecreases as th e
o th er inc reases o r v ice v ers a T h is may be expresse d
symbolically
.

0 i d0

+e :1 de

We h ave, for ins t ance Boyle s law as re gar d s th e


variat ion of volume of g a s e s accor ding to th e press
ure ; th at is wh en w e d oubl e th e pressure, w e h alve
th e volume T h is may be pro v e d experiment ally
T h e meth o d also was use d by Ricar d o to prove his
law th at th e rat e o f prot s varies in invers e rat io t o
th e rat e of wag e s We h ave al s o th e t en dency o b
serve d in respect t o increase of crimes wh en th ere
is d ec rease o f oppor tunities for labor
T h e expres sion of a law in t erms of th e quant it a
t ive relation betw een ant ec e d ent an d consequent
may be facili t at e d by a graph ic re present at ion o f th e

METHOD OF

C O NCO

MITANT VARIATIONS

133

same, th rou gh correspon d in g abscis s a; an d or dinat es


T h e varying ant ece d en t s, for ins tance, may be lai d
O ff on th e axis of " , an d eac h several consequent
represent e d by th e correspon din g or dinat es T h e
resul t in g curve thus obt aine d w ill represent th e law
I f th e equat ion o f th e
o f th eir mu t ual relat ion
curve can be d et ermine d it w ill represent th e math
em atic ally exac t expre s sion o f th e law in ques tion
I f th is is no t po ss ible, it may prove at least su g
g estiv e o f th e law wh i c h o th er w ise mi ght h ave
remaine d conc eale d
T h is graph ical meth o d is
especially useful in d ealin g w ith phy s ical p h enom
ena I f th e absci s s ae represent int ervals of time,
an d th e or dinat es correspon ding h ei ght of th e ba
ro m eter w e may const ruc t curves wh ic h sh o w a t a
glan ce th e d epen d ence of baromet ric pre s s ure upon
th e t ime of d ay S uch curves may be accurat ely
d raw n by p h o t o g raph ic processes on a sh eet o f sen
sitiv e paper place d behin d th e mercurial c olumn
an d ma d e t o move p as t it w ith a uniform h orizont al
velocity by clock work A similar process is applie d
t o th e t emperature an d elec t ricity of th e at mosp h ere
1
an d to th e component s o f t erres t rial ma gneti s m
T h is meth o d, moreover, h as th e a dvant a g e o f th e
p s yc h olo g ical impression wh ic h it m ake s
The
min d is more suscep t ible to th e percep tion o f varia
t ion in forces wh ere th e c h ang e is apparent to th e
sense s, th an to th e percep t ion o f a cons t ant force,
wh o s e cons t an t ch arac t er th ereby conceals i t s nat
ure an d function from th e senses
S ynch ronous
.

119

mso n
.

an d

Tait, Elements

f N atural P hilo soph y

Vol

13 4

IN

DUCTIVE

O GIC

ch ang es attrac t th e att ent ion an d a d mit o f rea dy


comparison as w e follo w o u t th e variat ions from
point t o point We may rin g a bell in a vacuum ,
an d d et ec t no soun d wh at soever an d th en allow th e
We no t ice th at as th e air
air t o ent er g ra dually
ent ers more an d more freely, th e soun d gro w s
lou der an d loud er T h e relat ion o f cause an d effect
is th us d emons t rat e d t o th e senses in th e mos t vivi d
manner possible T h e variations are exh ibit e d si d e
by sid e, an d th u s pre s ent e d t o g eth er in th eir c o n
c om it ant relation pro d uce th e d eeper impression
T his meth o d is o f spec ial a dvant a g e in all experi
ment s wh ere th e int ensity of th e forces can be
varie d at w ill an d th e consequent effect s exh ibit e d
T h e d et erminatio n o f
in some palpable manner
th e h eat rays in th e solar s pect rum is accomplis h e d
by th is meth o d T h e s pec t rum may be receive d
upon a plat e pierce d with a narro w slit th rou gh
whic h th e ray s can ac t upon a th ermo elec t ric pile
wh ic h w ill in dic at e by d eec t ions o f a nee dle th e
varyin g int ensity o f th e h eat in th e s everal rays o f
th e s pec t rum N o w, move th e slit th rou gh th e
wh ole ex t ent of th e spect ru m be g innin g w ith th e
violet port ion Wh ile in th e violet th e in d i go ,
th e blue an d even th e green th e nee dle of th e th er
mo sco pic apparatus w ill be d e ect e d but sl ightly
it will in d ic at e an amount of h eat increasin g as
th e slit cro ss es th e yello w nex t th e oran g e th en
th e re d ; an d th en beyon d th e re d an d ent erin g th e
dark port ion of th e spec t rum w e n d th e great es t
d e ect ion o f all Th e maximum of h eat is th ere
fore in a re g ion beyon d th e Observat ion o f th e
,

METHOD

O F C ONCO

MITANT VARIAT IONS

135

senses wh en un ai d e d by experimental d evice ; an d


1
e
t
reveale
d
conclusively
by
th
is
me
t
h
o
d
y
P rofessor Tynd all performe d a very int eres ting
experiment t o prove th at th e clou d of darkness sur
roun din g ames o f g reat h eat w as d ue to th e fac t
th at th e h eat consume d th e oatin g mot es in th e
air wh ic h serve t o scatt er th e li ght wh ic h is visibl e
onl y wh en t hus diffuse d T h e ph enomenon wh ich
h e en d eavore d to explain w as some wh at as foll o w s :
Beneath a be a
m o f electric li ght, a re d h o t poker
w as place d, and from it black w reath s as of smoke
w ere seen to ascen d A l arg e hy dro g en ame bein g
employe d, it pro duce d wh irlin g m as ses of d arkne s s
far more copiou sly th an th e poker
Of th is P ro
fo ssor Tyn d all remarke d : S moke w as out o f th e
ques tion ; wh at th en w as th e blackness ? It w as
simply th at of st ellar space ; th at is t o say, blackness
resul tin g from th e abs ence from th e t rack o f th e
be am of all m att er compet ent t o scatt er it s li ght
W h en th e ame was place d belo w th e beam , th e o at
in g matte r was d estroye d in s itu , an d th e air free d
from th is matt er rose int o th e beam, j os tle d asi d e
th e ill uminat e d partic les, an d s ubs tit ute d fo r th eir
li ght th e darkness d ue to it s o wn perfec t tran spar
ency N o th in g coul d more forcibly illus trat e th e
invisibility o f th e a g ent wh ich ren ders all th in g s
visible T h e beam crossed , unseen, th e black ch asm
forme d by th e transparent air wh ile at both si d es
o f th e g ap th e th ick s t re wn par t icles s h one ou t like
a lum inous oli d un d er th e po w erful illumination 2
.

Saig ey, Th e Unity of N atu ral Ph enomena,


2 T n all F a m en ts o S c ie nce
280
r g
f
,
,
1

y d

61

13 6

INDU CTIVE

LO G IC

S uch bein g th e ph enomenon an d P rofessor Tyndall s


e xplanat ion i t will be seen th at h e procee d e d accor d
in g t o th e me th o d of concomit an t variations in th e
follo win g experiment o f many wh ich h e performe d
t o subs t antiat e th is th eory :
A platinum t ube w ith its plug o f plat inum
g auze was connect e d w ith an experiment al tube,
th rough whic h a po w erful beam coul d be sent
from an electric lamp place d at its en d
T he
platinum t ube was h eat e d t ill i t glo w e d feebly
but dis tinct ly in th e d ark T h e experiment al tube
was th en exh aus t e d an d lle d w ith air th at h a d
pas s e d th rou gh th e re d h o t t ube A consi d erable
amoun t of oat ing m att er wh ich h a d escape d com
bustio n was reveale d by th e elect ri c beam
T h en th e t ube w as raise d t o a bri ght er re d ness
an d th e air permitt e d t o pass slo wly through it
Th ou gh di m inis h e d in quant ity a cer t ain amount
of oat in g matt er passe d int o th e exh aus t e d ex
perimental t ube
T h e plat inum tube w as ren d ere d s till h ott er ; a
barely percept ible t race of th e oat in g matt er no w
passe d th rou gh it T h e experiment was repeat e d
with th e difference th at th e air w as sent more
slo wly th rou gh th e re d h o t tube
T h e oatin g
matt er w as t ot ally d est roye d
T h e plat inum t ube w as n o w lo w ere d unt il i t
bor dere d upon a V i sible re d h eat T h e air, sent
th rough it s t ill more slow ly th an in th e las t ex
perim ent, carrie d with it a c lou d o f oat in g m at

t er P rofessor Tyn d all s comment ary upon th is


experiment is as follo w s : If, th en, th e sus
,

MET HOD

OF

O NCOMIT ANT VARIATIO N S

13 7

pen d e d matt er is d est roye d by a bri ght re d h eat


much more is it dest roye d by a ame wh ose t em
perature is vas t ly h i gh er th an any employe d in
th i s experim ent S O th at th e blacknes s int ro
d u ce d int o a luminous beam wh ere a ame is
place d beneath it is due, as st at e d, to th e d es t ru o
t ion o f th e suspen d e d matt er 1
P rofessor Tyn d all also supplement e d th is ex peri
ment by o ne wh ich w as accor din g to th e j oint
me th o d of ag reement an d difference He prepare d
oxy g en s o as to exclud e all oat in g part icles, and
foun d th at wh en blo wn int o th e beam d arkness
w as pro d uce d ; also th at hy d ro g en, nit ro g en car
bonic aci d, an d coal gas , wh en prepare d in a similar
w ay, eac h pro duce d arkness wh en poure d or blo wn
int o th e beam Th ese ins t ances combine d with
var ious posit ive ins t ances o f illum inat ion o f mo t e
s t re wn current s o f air illus t rat e th e me th o d o f
a g reement an d d ifferenc e
A n a dd it ional experiment accor d in g to th e meth o d
o f d ifference , w as as follo w s : P ro fes s or Tyn d all
place d an ord inary glass sh a d e in th e air with it s
mouth do wn war d T h is permitt e d th e t rack o f th e
beam to be seen cross in g it L ett in g coal g as, or
hy d ro g en, ent er th e s h ad e by a t ube reach ing to
it s t op th e g as g ra dually lle d th e sh a d e from th e
A s soon as it oc c up ie d th e s pace
to p d o wn w ar d
cro s se d by th e beam th e luminous t rack was in
L iftin g th e s h a d e so as to brin g
stantly abolis h e d
th e common boun d ary o f gas an d air above th e
be am , th e t rack as h e d forth A ft er th e sh a d e w as
,

Tynd al l , Frag ments of S ci enc e,

pp

283 , 284

138

IN

DUCTIVE

OGIC

full, h e invert e d it ; th ereupon th e g as passe d u p


w ar d like a black s moke amon g th e illuminat e d
1
part icles
T h e meth o d of concomit ant variat ions is n o t
only capable of illus t rat ion by laborat ory metho d s
an d d evic es ; it n d s abund ant illus tration as w ell
in th e realm Of nat ure wh ere observation al one
becomes th e ins trument of inve st i g at ion an d wh ere
experiment is impo s sible o r imprac ticable L yell,
in h is P rincip les of Geo logy, g ive s a very int eres t
in g ac coun t of th e al t ernat e elevat ion an d suh s i
d ence o f th e t emple o f J upi t er S erapis, at P o zzuoli,
2
I t is si tuat e d in proximity
o n th e Bay of N aples
t o several volcanoe s Ve s uvius , h o w ever bein g at
I t h as been observe d th at th ere is
s ome dis t ance
a cert ain connec t ion betw een ea ch era o f up h eaval ,
an d a local d evelopment of volcanic h eat ; an d on
th e oth er h an d betw een eac h era of d epression an d
th e local quiescent condition o f volcanic ph enomena
Before th e Ch ris t ian era wh en I s c hia w as in a st at e
o f eru ptio n an d A vernus an d o th er po in t s in th e
Ph le g raean el d s w ere celebrat e d for th e ir volcanic
ch arac t er it w as Observe d th at at th at t ime th e
g roun d on which th e t emple s too d w as s everal feet
above w ater V esuvius w as th en re g ard e d as a spent
volcano Aft er th e Ch ris tian era Ves u vius bec ame
ac t ive an d th en scarcely a sin g le erupt ion occurre d
in I s chia or arou n d th e Bay o f Baiae I t was oh
serve d th at at th at t ime th e t emple was sinkin g
V esuviu s th en became quiet for ve centuries pre
.

'

'

1
2

Ty nd all , Frag men ts of S cien ce,

Chapter XXX

pp

284 , 285

METHOD OF

MITANT

CO NCO

VARIAT IONS

13 9

ce d in g th e erup tion o f 1631, an d d urin g th at perio d


th e S olfat ara w as in erup t ion in 1198 I s c h ia in 1302,
an d Mont e N uovo w as forme d in 1538 Th en th e
foun dat ions O f th e t emple w ere observe d t o be ris
ing a g ain V esuvius became ac t ive aft er th at an d
h as cont inue d so ever since an d durin g th is time
the t em ple h as been subsi d in g
Th e inference is
th at as th e subt errane an h eat increases an d lava
formin g w ith ou t O btainin g an easy vent like th at
affor d e d by V esuviu s th e surface is elevat e d, but
wh en th e ro c k s belo w are coolin g an d contrac t in g
th e pent u p re bein g withdrawn in th e erup tion of
th e g reat V esuvius, th en th ere is a correspon din g
subsi d ence
T h e observat ion o f concomit ant variat ions is
furth ermore illus t rat e d in D ar win s researc h e s con
cernin g th e formation o f coral reefs as re g ar d s th e
ques t ion wh ic h some nat uralis ts h ave raise d as to
wh ich part o f th e coral reef is most favorable t o th e
1
f
ro
wth
coral
He a dd uces th e following fac t s,
o
g
mos t of wh ich are th e d irec t resul t of h is ob s erva
t ions : T h e great mo un d s o f livin g Porit es an d o f
Millepora roun d K eelin g atoll occ ur exclusively
on th e ext reme verg e of th e reef wh ic h is was h e d
by a cons t ant succe s sion of breakers At th e Mar
s h all I sland s th e lar g er kin d s o f coral wh ic h form
rocks measuring s everal fath oms in thicknes s pre
fer th e mos t V iolent s urf Th e out er margin o f th e
Mald iv a at olls con s i s t s of livin g corals an d h ere
th e surf is s o t remen d ous th at even lar g e s h ip s
h ave been th ro wn by a single h eave o f th e sea
,

Darwin, Co ral Reefs ,

pp

87 f

I NDU CTIVE

140

OGIC

h i gh and d ry on th e reef, all o n board th us escap


in g w ith th e ir l ives I n th e Red S ea th e st rong es t
c oral s live o n th e out er reefs an d appear to love th e
surf F rom these fac t s it is certain th at th e s t ron g
e st an d mo s t mass ive corals ouris h wh ere mo s t
expose d T h e les s perfect s t ate of th e reef of mos t
at oll s on th e leew ar d and le s s expo s e d S i d e com
pare d with its st at e to th e wind w ar d an d th e anal
o go u s cas e of th e g rea t er number of breac h es on th e
rear s i d es of th ose at olls in th e Maldiv a A rc hipelag o,
wh ic h affo rd some prot ec tion t o each o th er are o b v i

T h ere s eems
o u sl y explaine d by th is c ircu m s t an c e
t o be h ere a combination of th e me th o d of a g ree
ment w ith th at of conc omitant variations
An d
s uc h a c ombinat ion may be employe d t o a d van t a g e
in ca s e s wh ere th e ph enomena un d er inves t i g ation
S h o w for c e s un d er vary in g d e g rees o f int ensi ty ;
th e cau s al relation is more apparent an d th e po s
s ibil ity of for t ui t ou s co inci d enc e is lar g ely c limi
n ated if a number of ins t ance s can be collec t e d
in wh ic h th e force s m anifes t th emselves in vary
in g d e g ree s
Accumulation of ins t ances , S h o win g
conc omit ant variat ions in th e forces observed c o r
responds t o th e actual variat ions wh ich in an e xperi
ment are effec t e d by th e inves t igat or h im s elf I n
suc h ob s erve d inst ances h o w ever w e cannot al ways
h ave before us th e variat ions expresse d continuously
Th ere are evi dent g aps th at mus t be int erpolat e d
ment ally I n th e experiment, h o we v er o f wh at ever
nat ure, th e d e grees of int ensity can be exh ibit e d
c on t inuously
one d e g ree mer gin g int o anoth er
th rou gh inapprec iable increment s T h ere is th us
.

METHOD

OF

14 1

C ONC OMITANT VARIATIONS

a g ra d at ion wh ic h h as no g aps t o be lle d an d th e


psyc h olo gical impression is th ereby h ei ght ene d
By th e me th o d o f concomi t ant variat ions it is
possible also t o represent to th e min d th e magni
t u d e o f an unkno wn forc e or unobservable force
by comparison w ith th e int ensity of a kno wn force
whi c h l ies w ith in th e sp h ere o f observation F or
ins t ance Mr D ar w in g ives an int eres t in g account
in h is narrat ive o f th e n d in g near th e s h ore s of
th e P lat a a g roup of V i tri e d S ilicious t ubes wh ic h
h ad been forme d by li ght nin g en t erin g lo o se s an d
T h e int ernal s urface of s uch tubes is complet ely v it
rie d , g lossy , an d smoo th an d th e t ubes th em s el v es
are g enerally compresse d, an d h ave d eep l o n g itu
d inal furro w s so as c losely to resemble a S h rivelle d
ve g e t able s t alk or th e bark of an elm or c ork t ree
T h eir circumference is about two inch es but in
some fra g ment s whi c h are cylin dr ical an d with out
any furro w s i t is as much as four inch es J u dg in g
from th e uncompresse d frag ment s th e measure o r
bore o f th e li ghtnin g prove d t o be about o n e inc h
an d a quart er In cont ras t with th e force of li ght
nin g as th us reveale d in it s effec t s Mr D arwin c it es
some experiment s performe d in P ari s by an arti
c ial force of g reat ma g nit u d e in d ee d an d yet w i th
resul t s th at seem insi g ni c antly s mall in compari
son He says : At P ar is M Hat ch ett e an d M
Beu d an t succee d e d in making tu bes in mos t re s pect s
similar t o th ese ful gurit es by passin g very st ron g
s h ocks o f g alvanis m th rou gh nely po wd ere d glass :
th ey faile d, h o w ever , both w ith po wd ere d felspar
an d quart z O ne t ube, forme d with poun d e d glass ,
,

142

INDUC IVE LO GIC

w as very near an inch lon g namely 9 82, an d h a d


an int ernal diamet er of 019 o f an inch Wh en w e
h earth at th e s t rong e st batt ery in P ari s was u s e d,
an d th at its po wer on a subs t ance o f su c h easy fusi
bility as glass w as to form t ubes so d iminut ive, w e
must feel g reatly as t onis h e d at th e forc e of a s h ock
of l ightn ing whic h s t rikin g th e san d in several
place s h as forme d cyl in d ers in one in s t ance at leas t
th irty feet long an d h avin g an int ernal bore wh ere
not compre s s e d, of full an in ch an d a h alf ; an d
this in a mat erial so extraordinarily refrac t ory as
quartz
T h e meth o d O f concom it ant variat ions may be
u s e d in re gar d t o ph enomena wh o s e nature is such
as s eem in gly to in dicat e a con st ant law of variation
and yet inferen ces base d th ereupon lea d t o false
res ul t s It is th erefore w ell t o not e s ome of th ese
in st ance s by w ay o f g eneral precaut ion in applyin g
this meth o d
1 It d oe s no t necessarily follo w th at h avin g
ob s erve d tw o force s varying in a c ons t ant rat io
th rough several concomit ant mo di cations th e
same ratio will be pre s erve d in d enit ely th rou gh
all s ubsequ ent c h an g es Wat er c ontrac t s as it is
c oolin g
S uppose w e beg in t o not e this c ont inue d
cont ract in g of w at er from 100 F t o
w e nat u
rally expect t o n d it c ont inuin g th rou gh 9 0 t o
A n d as w e observe w e n d o u r expect at ions con
rme d A n d so o n th rou gh t o
w e n d th at
w at er cont inues t o cont ract I t is th erefore mos t
nat ural fo r us t o expect t o n d w at er contracting
,

Darwin, Vo y ag e of

N aturalis t, V o l I
.

pp

76 f

METHOD O F C ON C O MITANT VARIAT IONS

143

Bu t ju s t at th is point in th e series , th ere is


at
a break in th e cont inuity of variat ion ; at 3 9 w at er
be g ins t o expan d an d so c ont inues un t il i t passe s into
th e soli d form at th e freezin g point T h e same also
is il lus t rat e d in Weber s law alrea dy mentione d
wh ic h expresses th e quant it at ive relat ion betw een
th e s t imulu s an d th e corre spon d in g s ensat ion Th e
law is th at th e force of th e s t im ulus m ust inc rease
g eomet rically in or d er th at th e int en s ity of th e
sensation s h oul d increa s e arithmetical ly T h is law
h o w ever, break s dow n t o w ard s th e upper or lo w er
limit s w ith a s timulus o f sli ght d e g ree o f int ensi ty
an d w ith o n e o f ex t reme in t en s ity We n d al s o
an increas e of t emperat ure as w e procee d t o w ar d s
th e cen tre o f th e earth o f about one de g ree t o
every fty th ree feet o f d es c ent T h is by no mean s
w arran t s us in inferrin g th at th is rat io cont inues
con s t ant t o th e very cent re it s elf In certain
ph eno m ena, moreover th ere are nat ural limit s as
in soun d for example wh ere th e pit c h rises as th e
number o f vibrations increas e s At a cert ain point,
varyin g ac cor d in g t o different in div i dual s increas e
of v ibrat ions give s no re s ul tin g soun d wh at soever ;
an d S O th ere is a lo wer limi t vibrat ion s may
d ecrea s e to a poin t beyon d wh ic h no soun d is
h ear d
A n illust rat ion o f th i s fallacy, th ou gh not s t rictly
of th e meth o d o f c on com it ant variation s is g iven
by Jevons He t ake s th e follow in g s erie s of prime
number s : 41, 43 , 47 53 6 1, 7 1 8 3 9 7
et c I t will be seen th at th ey all ag ree in bein g
2
$
a:
4 1,
values o f th e g eneral expression

'

144

IN DU CTI VE

O GIC

wh ere w e put for w th e successive values o f 0 1,


2
i
n
a
:
0
w
F
or
n
t
a
ce
le
t
in
4
c
s
2 3 , , et
+x+
,
1 in th e same w e g et 43 ;
4 1, w e g e t 4 1 ; le t x
wh en w 2 w e g e t 47 ; an d so o n It seem s as
th ough w e coul d keep th is up ind e nit ely, pro duc
in g an increas in g series al w ay s of prime numbers
It is foun d h o w ever th at if w e t ake m= 40, in
2
4 1 w e S h all h av e 40 x 4 0
th e formula x
x
40
4 1 wh ich equals 168 1 an d th is number is
th e square of 41 an d th erefore no t a prime number
At th is point th e law breaks d o wn 1
I n th e sph ere o f pol itical economy also w e mi gh t
be le d int o an easy yet false inferenc e S uppose
a c ert ain farm yiel d 500 bu sh el s o f corn w ith a
g iven a moun t of expen dit ure an d labor We mi ght
think th at if w e d ouble d th e expen di ture an d
labor w e will also be able t o double th e resul t s
an d obt ain a y iel d o f 10 00 bush els as over ag ain st
th e 500 of th e previous year Here h o w ever wh at
is kno w n as th e law of d ecreas in g ret urns ob t ains ;
t o double th e pro duc t it may be nece ss ary t o t riple
or qu ad ruple th e labor and expense
T hu s in th e
pro duct ion of any plo t o f lan d th ere is a point o f
equilibriu m wh ich marks an impassable li m i t no t
of cours e a limit wh ich c oul d not be passe d if it
w ere w ish e d but one th at no o n e w i sh es to pass
2
b ecau s e th ere is n o th in g t o be g aine d by s o d oin g
To kno w th at s u ch false inference s are at lea st
possible in th e applicat io n of th is me th o d Of con
comit ant variations t o th e unknow n re g ions beyon d
,

f S cience p 230
P o litic al E co nomy p 325

Jevo ns
Gid e ,

P rin cip les

MET HOD O F

C NC

OMITANT V ARIAT ION S

145

experience may serve at least t o keep us o n


g uar d und er similar circums t ances
2 T h ere are cer t ain ph eno m ena, moreover, in
wh ich an increase d int ensity o f th e force in ques
t ion may give ri s e t o incid ent al eects which t end
t o neut ralize th e ch ief effect to be att aine d F or
ins t ance an over d ose o f ars enic causes V iolent
con t ractions of th e s t omach so th at i t s cont ent s
are imme diat ely ej ect e d, an d thus th e syst em is
relieve d o f th e noxio us subs t ance
3 T w o element s in a g iven p h enomenon may
vary t o g eth er cons t antly an d yet th ey may no t be
relat e d at all as cause an d effect , but appear as c oin
I t h as
c id en tal effec t s o f o n e an d th e s ame c ause
been observe d th at th e occurrence of th e Aurora
Boreali s h as been accompanie d by pronounce d ma g
netic dis t urbances I t h o w ever, cannot be inferre d
th at th e former h as been th e cau s e of th e latt er ;
th ey are probably th e varie d effect s o f some w i d ely
operat in g mag ne t ic force
T h e precaut ion above ment ione d h as alrea dy been
referre d t o as provi d e d for in th e c anon o f th is
meth o d wh ich s t at e s th at th e observe d concomit ant
variat ion may indicat e not al w ays a direct causal
element betw een th e tw o varyin g element s, but
th at th ey are at leas t connect e d w ith th e ph e
n o m e non un d er inves t i g at ion th rou gh some fac t of
c ausat ion

o ur

CHAP T ER " I
TH E

ME T H O D

OF

RE S I D UE S

meth o d of re si due s cons is t s in th e analys is


o f a g iven p h enomenon base d upon previous in d uc
tion s th rou gh wh ic h it h as been d et erm ine d that
cert ain element s in th e ant ece d ent h ave c ause d
cert ain element s in th e consequent ; th e inference
is th en d rawn th at th e remainin g element s of th e
ant ec e d ent are necessarily th e cause of th e remai n
der of th e con s equent It is a meth o d of elimina
tion of th e know n relations so as t o Simplify th e
complex c h aract er o f th e ph enomenon and d is c lose
th e relat ions th at are unkno w n in th e li ght of a
cau s al connect ion wh ic h w e are cons t raine d to be
lieve must Obt ain
T he Can o n of the M etho d of Res id u es
S ub duct
from any p h enomenon s uc h part as is kno wn by
previous in duc tions t o be th e effect of certain ant e
c e d en t s an d th e re sid ue of th e ph enomenon is th e
effect o f th e remainin g ant e c e d en t s
T h e symbolic al represent at ion is as follo w s :
TH E

Given S
I f it

is

'

+e

known th at th ere exis t s th e causal relat ion

3:

146

THE

METHOD

OF RESIDUE S

147

w e may th en infer th at O is th e cause o f e In


th i s 0 may be simple o r complex ; if it is simple
th e caus al relat ion es t abli sh e d is expresse d in i t s
simples t t erm s an d is th erefore a d et erminat e resul t
If, h o w ever, th e resi due 0 is complex, it mus t be
re duce d by experimen t al an alyis t o it s simples t
element s, an d th eir relation t o th e el ement s int o
wh ic h e c an be analyze d furth er d et erm ine d
T h e mos t st riking illust rat ion o f th is meth o d, an d
o n e of th e mos t brilliant ach ie v ement s of s c ience
as w ell is th e discovery of th e planet N eptune by
Ad ams an d L e V errier, workin g on th e problem in
d epen d ently an d reach in g th e same resul t Th ese
as t ronomers h a d observe d cert ain pert urbat ions in
th e planet U ranus It d i d not keep in it s proper
orbit as d e t erm ine d by th eir math e mat ic al c al c ula
tions b ase d upon th e presenc e of th e kno wn s t ellar
bo d ies I t beh ave d as th ou gh beyon d its orbit was
an ou t er planet, wh ose presence alone coul d acc ount
for th e ob s erve d pert urbat ions Ad ams an d L e
V errier th en procee d e d t o calculat e th e exact posi
t ion o f suc h a d is t urbin g bo dy as d e t ermine d by
th e nature an d ma gnitu de of th e perturbat ions o f
U ranus T h e t elescope w as th en point e d t o th e
exact point in th e h eavens as th us ind icat e d, an d
th e planet N eptune w as reveale d t o th e eye accor d
in g t o th e d et erminat ion o f far reach in g proph ecy,
wh ic h cond ently assert e d th at it mus t be th ere
T h e meth o d o f resi due s is really a d e duc t ive
meth o d base d upon th e law of sufcient reason ; s o
many elem ent s o n th e o n e h an d pro ducin g so many
elements on the oth er ; if, th en, a par t of th e former
-

INDU C IV E LO G IC

14 8

is t o be ch ecke d off as cause of a part o f th e latt er,


th en th e rem ain d er on o n e h an d mus t be th e cause
T his is pure d e
of th e remain d er o n th e o th er
d uction Fo r w e ask , W hy are w e cons traine d to
accoun t for th e remain d er on o ne sid e by th e re
T h e only possi ble ans w er
m ain d e r on th e o th er
is th at it mu st be account e d for w ith in th e sys t em
t o wh ich it is referre d ; an d but one part th erein is
left whic h can possibly ac c oun t for it , because all
th e o th ers are specic ally d et ermine d in th e kno wn
effect s wh ich th ey h ave pro duce d T h is meth o d,
h ow ever, h as a proper pl ac e amon g th e in duct ive
me th o d s inasmuch as i t is base d on previous in du c
tion s , an d lead s t o investi g at ions th at can be prose
out e d only by th e various in duc t ive processe s of
experiment
Wh en th e resi due o f th e ant ece d ent is a simple
element an d no oth er po s sible causal element can
lie c onceale d from ou r observat ion, th en th e in fer
enc e is simple an d conclusive A di f cul ty h o w
ever may presen t it sel f o w in g t o th e fac t th at th e
re s i dual element is ap t t o be complex an d leave th e
ph enomenon st ill ind et erminat e, o r th ere may be a
lurk in g element unnotice d by us wh ic h is th e real
caus e in qu es tion T h e funct ion o f th is m eth o d is,
th erefore largely sugg est ive It says th e effect is
n o t wh olly ac coun t e d for by th e kno w n caus al el e
ment s ; th ere is a re si d ue unaccount e d fo r an d it s
cause is t o be sought in th e resi d ue of th e antec e
d ent an d if it is th ou ght th at th e wh ole o f th e
ant ece d ent is compreh en d e d th e ques t ion is s t art e d,
May th ere not be unobserve d circumst ance s of the
.

METHOD

THE

OF

RES IDUES

149

ant ece dent th at furth er experim ent w ill be c alcu


lat e d to reveal ? I n many cases th erefore th is
meth o d mus t be supple m ent e d by s ome oth er ex
perim ental meth o d in or d er t o secure more prec ise
d e t erminat ion g enerally th e m e th o d of d ifference
I t oft en h appens in inve s ti g ation s in ch em is t ry
as t ronomy an d p hysics th at th e ac t ual ph enomena
vary in great er Or less d e g ree frOm th eir expec te d
beh avior accord in g t o es tablis h e d th eory
This
mus t lea d eith er t o a recons t ruc tion o f th eory o r
to a searc h for some uno bs erve d force sufcien t to
account fo r th e discrepanc y Hers c h el w as th e
rs t to point o u t th e si gni cance of suc h d isc rep
an cies in s c ien t i c researc h, an d h e cal le d th em
,

res id u al

p hen o m ena

A n illus t ration of such a S ituation and th e solu


t ion of th e problem th us present e d is th at of S ir
Humph ry D avy s experiment s upon th e d ecompo s i

t ion o f w at er by g alvani sm
He foun d th at
be s i d es th e tw o compon ent s of w at er oxygen an d
h y dro g en an aci d an d alkali w ere d evelope d at th e
tw o opposit e poles of th e mach ine A s th e th eory
o f th e an alysis of w a t er d id no t g ive reason t o ex
peet th ese pro duct s th ey w ere a res id u al p heno me
The
n o n t h e cau s e of wh ic h w as s t il l to be foun d
ins ight o f D avy conj ec ture d th at th ere mi ght be
s ome h i dd en cau se o f th is por t ion o f th e effec t ; th e
glas s containin g th e w at er mi ght s uffer part ial
d ecomposit ion , or some forei gn matt er mi ght be
min gle d wi th th e w at er an d th e aci d and alkali be
dis en ga g e d from it so th at th e w at er woul d h ave
A ssumin g th is , h e
n o s h are in th eir pro d uc t ion

INDU CTIVE

150

O GIC

proceed e d t o t ry wh eth er th e t ot al removal o f th e


cause w oul d des t roy th e effec t pro duce d By th e sub
stitutio n of g ol d ve ss el s for th e g lass , w ith out any
c h an g e in th e effec t h e a t once d e t ermine d th at th e
glas s w as no t th e cause Employing dis tille d w at er
h e foun d a marke d diminution of th e quant ity o f acid
an d alkal i evolve d ; yet th ere was enou gh t o S h o w
th at th e cause wh at ever it was was s t ill in opera
t ion Th e impurity of th e w at er th en w as not th e
He no w conce ive d
s ole , bu t a concurren t c ause
th at th e perspirat ion from th e h an d s t ouch in g t h e
ins trument s might affec t th e case as it w oul d con
t ain c ommon sal t, an d an acid an d alkali w oul d
re s ul t from it s d ecomposit ion un der th e ag ency o f
ele c t ri c ity By carefully avo idin g suc h c ont act, h e
re duce d th e quant ity o f th e pro ducts s till fur th er
until no more th an s l ight traces o f th e m were per
Wh at remaine d o f th e effec t mi ght be
ce ptibl e
t raceable t o impurit ie s o f th e at mosph ere d ec o m
po s e d by cont act with th e elec t rical apparatu s
A n experiment d et ermine d this ; th e mac h ine was
place d und er an exh aus t e d receiver, and wh en th us
se c ure d from atmosp h eric inuence it no lon g er
evolve d th e ac id and alkali 1
By means of th e s u gg es t ions inci d ent upon th is
meth o d Bunsen in 186 0, d i s covere d tw o ne w alk a
l ine met al s , caesium an d rubi dium
He w as ex
am in in g alkal ies pro d uce d by th e evaporat ion o f
mineral wat er from Durk h eim Th e ame of th ese
s al t s w as examine d by th e spec t roscope
Bunsen
d isc overe d s everal bri ght lines wh ich h e h a d never
1 Go re Th e A rt o S c ie nt
f
ic Discov ery pp 432 433
.

THE

METHOD OF RESIDUES

151

no t ice d before an d wh ich h e kne w coul d no t be


pro duce d by pot as h o r so da, wh ose correspon din g
lines w ere in close proximity He th en subj ect e d
th e mixt ure to a search in g analysi s an d succee de d
in obt ainin g tw o ne w alkaline subs tances W h en
h e h a d separat e d th em , h e th en t est e d th em by th e
meth o d o f difference, by wh ic h h e foun d th at th ey
w ere capable o f pro ducin g th e lines at rs t not ice d ;
but wh en w ithd ra wn, th e lines ins t ant aneously dis
appeare d
T h omson an d Tait in th eir E lements of N atu ral
P hilo sop hy, h ave th e follo w in g
reference an d
illus t ration o f th is meth o d Wh en, in an ex
perim ent, al l kno w n c auses bein g allo w e d fo r,
th ere remain un explaine d effec t s ( excessively
sl ight i t may be) th ese mus t be carefully inv es
tig ate d an d every conceivable variat ion o f ar
ran g ement o f apparatus , et c , t rie d ; unt il if
po s s ible, w e manag e s o t o exagg erat e th e resi d ual
ph enomenon as t o be able t o d et ect its cause I t
is h ere perh aps th at in th e present s t at e o f
science w e may most reasonably look for ext en
sions o f o u r kno wle dg e ; at all event s , w e are
w arrant e d by th e recent his t ory of nat ural p h i
T h us t o t ake only a very
l o s o phy in so d oin g
fe w ins tances an d t o say no th in g of th e discovery
o f ele c t rici ty an d ma gne tism by th e ancien t s , th e
peculiar smell O bserve d in a room in wh i c h an
elec t ric al mac h ine is kept in ac tion w as lon g
a g o observe d but calle d th e smell o f elec tric ity
an d th u s left unexplaine d T h e sa g acity o f S ch en
bein l ed to th e d iscovery th at th is is d ue t o th e
,

152

IND

UCTIVE

OGIC

format ion o f ozone a mos t ex traor dinary bo dy


o f enormous c h emical ener g ies ; wh ose n a t ure is
s till uncert ain th ough th e att ention o f c h emis t s
h as for years been direct e d t o it 1
A noth er illu st ration o f th is meth o d is s een in
th e comparison o f th e observe d an d calculat e d
positions of En cke s comet
It w as foun d th at
th e come t returne d a little sooner th an i t sh oul d
h ave done th e perio d reg ularly d ec reasin g from
d ays , betw een 17 8 6 an d 17 89 t o
between 1855 an d 1858 Th e inference h as been
th at th ere is a resist in g me dium as th e eth er
llin g th e S pac e th rou gh wh i ch th e comet passes
Wh at th e resi s t in g medi um is an d it s nature, is
as far as re
o f c our s e a matt er o f c onj e c ture
veale d by th is meth o d alone T h e meth o d merely
in dicat e s s ome re s i st in g me d ium t o account for
2
i
th e observe d d screpancy
Hersch el has observe d th at all great as t ronom
ical dis c overie s h ave been di s close d in th e form
T h e pract ice w as int ro
o f re s i d ual d ifferences
d u ce d by Halley
wh en as tronomer royal, o f
c omparin g
s y s t emat ically
th e positions o f th e
h eavenly bo die s as ac tually observe d w ith wh at
might h ave been expec t e d th eoret ically His re
d uction s of th e lunar observatio n s g ave a s eries
o f resi dual errors
ext en din g from 17 2 2 t o 17 39
Th e s e w ere carefully t abul at e d, an d forme d th e
ba sis for cert ain mo dicat ion s of th e lunar th eory 3
,

,
.

pp

homso n and

113 f

Tait , E lemen ts

f N atu ral P hilo sop hy, Vo l

Jevo ns

P ri nciples

f S cien ce , p

57 0

I bid

57 2

THE

METHOD O F R ES ID UES

153

A d iscrepancy w as observe d by N ewt on be tween


th e th eore tical an d act ual velocity o f soun d ; th e
former be in g 9 68 feet per secon d, an d th e latt er
1142
N ewt on s experiment s an d calculat ion w ere
bo th inac curat e ; neverth eless, a real discrepancy
h as been prove d t o ex is t th e th eore tical bein g
9 16 an d th e real veloci t y 109 0 fee t per s econ d
I n 1816 L a P lace s h o w e d th is differenc e t o be
due t o th e h eat evolve d by th e su dd en c ompres
s ion o f
th e air d urin g th e propag at ion o f th e
soun d w ave th e h eat h avin g th e effec t of in
creasin g th e elas ticity of th e air an d th erefore
appreciably ac c elerat in g th e sou n d impulse
I t sometimes h appens th at in repeat in g an ex
peri m ent w e are confront e d w i th evi den tly different
resul t s T h en w e may be su re th e exper im ent h as
been carelessly or inaccurat ely performe d ; o r else
th ere are s ome dis t urbin g cause s no t O b s erve d by
us O n th e o th er h an d h o w ever, if th ere is n o
l ikelihoo d of coinc i d ence on repeat e d t rials yet,
neverth eless a marke d a g reement is no t ice d in th e
result s o f various t rial s th e min d Sh oul d be at
once alert t o discover th e h i dd en cause of such
a g reement and possibly may be le d to new t ruth s
Th e follo win g illu s t ration is
o f g reat import ance

iven
by
T
h
omson
an
d
Tai
t
With a very goo d
:
g
ach romatic t elescope a s t ar appears to h ave a sens i
ble disc But as it is O bserve d th at th e d iscs of all
s t ars appear t o be of equal an gular diamet er w e of
course s us pec t some common error L im it in g th e
aperture o f th e obj ec t g las s increases th e appear
ance in quest ion, wh ich, on full inves t ig ation, is

'

1
54

IN

DUCT IVE

O GIC

foun d t o h ave noth in g to d o w ith d iscs at all I t


is, in fact, a ph enomenon due to diffract ion of
1
light
It was sai d of D arwin th at in h is research es th e
resi dual ph enomena were al w ays th e special obj ect s
His son F rancis D arwin, says of
o f h is att ent ion
h im : T h ere w as one qual ity o f m in d wh ic h seem e d
t o be of S pecial an d ext reme ad vant a g e in leadin g
h im t o make discoveries It w as th e po w er o f
never lettin g exceptions pas s unno t ice d Every
bo dy not ices a fact as an ex c ept ion wh en it is s t rik
in g or frequent but h e h a d a special ins tinct for
arres tin g an except ion A point apparently s l ight
an d unc onnec t e d w ith his present w ork is passe d
over by many a man almos t unconsciously w ith
some h alf c on sidere d explanation whic h is in fact
no explanat ion It was jus t th ese th in g s th at h e
sei ze d upon t o make a s tar t from I n a cert ain
s en s e th ere is no th in g special in th is proce d ure
many d i scoveries bein g mad e by means Of it I
only ment ion it because, as I w at ch e d h im at h is
w ork th e value of th is pow er t o an experimen t er
2
w as so s t ron gly impresse d upon me
This is
st rikin g t es t imony as t o th e pract ical w orth of th is
meth o d as an ins t rument of research
T his meth o d h as al s o been applie d t o th e more
practic al usag e o f examinin g th e refuse of manu
factu red an d oth er pro d uct s in or d er t o d iscover
.

homson and

114
2

125

f N atu ral P hilo sop h y Vo l

Tait , E lements

Darwin, Lif e

a nd

Le tters

f Ch arles Darwin , V o l I

THE

MET HOD O F RESIDUE S

155

some conceal e d u tility T h e analysis o f c oal t ar


refuse h as led to th e di s covery of many valuabl e
subs t ance s th at h ave prove d o f use in th e art s , and
in me dicine as w ell G lauber, th e eminent ch emis t
an d a di scoverer o f several ch emical com pound s
sai d h e ma d e it a rul e t o examin e wh at every o th er
c h emis t thre w aw ay
.

CHAP TER " II


VE RIFICA T IO N

AND

PRE DI CT I O N

We h ave seen
Metho d
th at th e in duc t ive meth o d s are efcient in reveal in g
th e c ause of a given ph enomenon und er inve s ti g a
tion ; an d yet th ey d o not w arrant u s in g eneral
izin g th e s pecial ins t ance so as t o formulat e a
universal law
T h ere is al w ays th e pos s ibil ity
th at wh ile th e special case wh ic h w e experiment
upon may g ive us in dicat ion s of an exis t in g causal
relat ion still a wid er experienc e mi ght disprove, o r
else mo dify mat erially our c onc lusions T h e w ell
reco g nize d fac t of th e plurality of cau s e s an d th e
int ermixture o f effec t furth er embarras s e s us in
th e att empt t o ri s e t o a la w h avin g un ivers al sig
nic an ce an d vali di ty
T h e resul t s o f th e ind uc
tive meth o d s , th erefore nee d t o be supplemen t e d
by s ome corroborative obs ervations or experiment s
th at will conclu sively verify th e re s ul t s as obt aine d
This supplement ary meth o d is o ne wh ic h combines
d e d uc tion with in duction
Mr Mill c all s it th e
D e d uctive Meth o d
It is, h o w ever more ad e
qu ately d e s i gnat e d by th e name th e In d u c to d e
It con si st s of th ree s t ages
d u ctiv e Me th o d
1 O bt ainin g, by th e in d uct ive me th o d s alread y
The Ind ucto-d ed u ctiv e

156

F ATI ON

VERI IC

157

AND PREDIC IO N

d escribe d, th e evi d ence o f some exi s tin g causal


conn ec t ion, t ent at ively e xpresse d in th e form of a
universal la w
2 Re g ar d in g th is universal law as th e basis for
subsequent d e d uc t ions, by wh ich w e g ain a knowl
e dg e of th e nat ure of unkno wn p h enomena, as
necessit at e d by th e condi t ions o f th is la w
3 Verifyin g th e re s ul t s th us ob t aine d by th eir
correspon d ence w ith th e p h enomena as actually
observe d W h ere th is c orres pon dence is w antin g
th en eith er th e la w w as n o t co rrec tly expresse d, or
th ere mus t h ave been some error in o ur d e d uction
bas e d upo n it Wh en w e are assure d th at th e l at
ter is not th e case, th en a discrepancy betw een th e
th eoret ically d e duce d resul t an d th e actual fact s as
observe d, al ways discre d it s o ur ori g inal in duct ion
Thi s meth o d of veri cation serves as a ch eck upon
h as ty g eneral iza tion o n th e o ne h and ; an d o n th e
o th er, it serves t o ext end o ur kno wle dg e int o nu
kno w n re gions , an d is v al uable as a me ans o f scien
tic pre d ic t ion
In th e d evelopment of scien t i c
kno wle dg e it h as been a potent fact or in enlargin g
th e boun ds o f k no wle dg e beyon d th e sp h ere of im
m e diat e observat ion
T his com bine d proce s s o f reas onin g is th e o ne
comm only employe d by us all I n duction an d d e
d uc tion are not separat e processes , but, as before
remarke d, th ey are complement ary fac t ors in th e
proces s o f reas onin g
W e are con
o n e ac t ual
tinu ally usin g o ur in duction s as a d e d uc t ive basis ,
inferrin g h ow th in g s s h oul d be before th ey are
reall y seen ; and , wh en seen, at once ins tinct ively
.

158

INDU CTIVE

O GIC

comparin g prior inference w ith present fac t, w e


are eith er con rme d in our reasonin g pro c ess, or
compelle d t o d iscar d our previous inference as
false o r inadequa t e as th e cas e may be
O ur
w orl d, th e w orl d of kno wle dg e, is b uilt up o f th e
seen and th e unseen as w ell , because necessit at e d
by inferences g row ing out o f th e seen wh ich w e are
const raine d t o make ; th e unseen wh ich w e th us
are cont inually bu il din g int o th e seen an d re g ar d
in g it as th ou gh th e kno wn, w e are h o w ever,
from time t o t ime compelle d t o al t er, an d h ere an d
th ere t ear d o w n wh at w e h ave t oo ra sh ly buil d ed
up as th e s t ruc ture is put t o th e t es t of verifyin g
fac t
This me th o d o f vericat ion was use d t o d eci d e
betw een inferences drawn by N e wt on an d Huy
gh en s re s pec t ively , re g ar ding th e nat ure o f li ght
N ewt on s observations le d h im to infer th at light
c onsis t e d o f par t i c les o f matt er s h o t o ut from th e
s un
Huy gh ens ins is t e d th at li ght consi s t e d in th e
propagat ion of some kind of dis t urbance in th e man
ner of a w ave mot ion N ewt on s th eory being t aken
as es t ablis h e d, it w oul d nece s s it a t e th at l ight o n
ent erin g a d enser bo dy o f w at er bein g refract e d
more nearly in a direct ion perpen dicular t o th e
surface s houl d accor din gly, m ove fas t er in th e
d enser bo dy th an in th e rarer one out si d e
On
th e o th er h an d, accordin g to Huy gh ens th eory th e
opposit e effect S h ou l d t ake place
li ght bein g re
frac t e d t ow ar d s th e vertical at th e h orizont al sur
face o f a d ens e bo dy such as wa t er it s velocity in
th e d ense bo dy s h oul d be les s th an it s velocity in
.

VERIFICAT ION

th e rare bo dy

D TO

159

AND PRE IC I N

T h e experiment s separat ely m ad e


by F izeau an d F oucaul t both g ave th e resul t th at in
w at er li ght moves slo w er th an in air, an d th erefore
th e th eory o f Huy gh ens wh ich w as in accor d w ith
suc h a fact, w as veri e d an d th e th eory o f N e wt on,
wh ic h w as radic ally o ut of h armony w ith such a
l
fac t w as d iscre dit e d
We canno t th eorize concernin g nature t o any co n
sid e rable ext ent w i th ou t resort in g to nature a g ain
to correc t aberrat ions of reason , an d th e false
fan cies o f t h e ima g ination T h eory, if correc tly
formul at e d, will al ways lea d to a represent ation
o f fac t s as th ey are ; jus t as fac t s as th ey are, if
ri ghtly int erpret e d, w ill al ways lead t o correc t
th eory
T h e follo win g are illu s t rat ions of th e value o f
th is meth o d in pre dic t in g resul t s before unkno wn
Halley h a d th e g lory o f h avin g rs t d et ect e d a
peri o dic comet in th e cas e o f th at wh ich h as s ince
borne h is name I n 17 05 Halley explaine d h o w
th e parabolic orbit o f a plane t may be det erm ine d
fro m th ree observat ions ; and j oinin g ex ample to
precep t, h imself calculat e d th e posit ions an d orbit s
He foun d, as th e rewar d
o f tw enty four comet s
o f h is in d us t ry , th at th e c ome t s of 1607 an d 153 1
A n d h ere th e
h ad th e same orbi t as th at o f 1682
int ervals are nearly th e same, namely, about s even ty
A re th ese three comet s th en i d ent ical
v e years
I n lookin g bac k into th e h is t ory of such appear
ano es h e foun d come t s rec ord e d in 1456, in 1380,
an d 1305 ; th e intervals are s t ill th e same, sev
.

Tait ,

Recent Ad vances

in P h ysical S cience,

pp

65 , 66

INDUC IVE LO GIC

160

se v enty S ix years It w as impossibl e


no w t o d ou bt th at th ey w ere th e perio d s of a rev o l v
in g bo dy it s orbit a lon g ellipse no t a parabola If
th is w ere so th e comet mus t reappear in 17 58 o r
Halley be gan his laborious calculations an d
17 5 9
pre d ict e d th at th e comet w oul d reac h it s perih elion
A pril 13 17 5 9 but claime d th e license o f a month
for th e inevitable inaccuracies of a calculation in
wh ich in a ddition to all o th er sources of error w as
mad e in h as t e th at it m i ght appear as a pre dic tion
T h e comet ju sti e d h is c alculat ion s an d h is caut ion
t o g eth er ; for it arrive d at it s perih el ion o n Marc h
enty-v e o r

13 , 17

Ano th er illus t ration Of a like nat ure is th e pre


dic tion o f F arad ay, base d upon Wh eat s t one s ex
perim e ntal proof th at th e con d uc t ion o f elec t ric it y
requ ire d time ; namely th at if th e conduc t in g
w ire s w ere connec t e d w ith th e coatin g s of a lar g e
L ey d en jar th e rapidity O f con duc t ion w oul d be
necessarily lessene d T his pre d iction w as ma de in
1838 an d w as not veri e d unt il S ixt een years lat er,
a s ubmarine cable w as lai d beneath th e Engl is h
Ch annel A con s id erable re t ar dation of th e elect ric
S park was th en d e t e c t e d by S iemens an d L at imer
F arad ay at once point e d out th at th e wire
Clark
surroun d e d by w at er resemble s a L ey d en j ar o n a
larg e scale : so th at eac h me ss a g e sent th rou gh th e
2
cable veri e d h is remark of
In P as t eur s exper iment s with S ilk worms already
referre d t o, h e ma d e a pred ict ion in 1866 wh en,

W h ew ell His tory of I nd uctiv e S c ie nce


2 Jevo ns P rin c i les o S c ience p 543
p
f
1

3d

ed

V o l II
.

182

VER IF ICAT ION

R D TO

16 1

AND P E IC I N

h avin g inspec t e d fourt een parcels of e gg s int end e d


for incubation an d h av in g examin e d th e mo th s
wh i c h pro duc e d th ese e gg s h e w rot e out th e pre
d ic t ion o f wh at w oul d occur in 18 67 , an d place d
th e proph ecy as a seale d lett er in th e h an d s of th e
mayor of St Hippoly t e I n 1867 th e cul tivat ors
communicat e d t o th e mayor th eir resul t s T h e
le tt er of P as t eur w as then opene d an d read an d it
w as foun d th at in tw elve o u t of fourt een cases
th ere w as absolut e conformity betw een h is pre dic
t ion an d th e observe d fact s Many o f th e group s
h ad peris h e d t ot ally ; th e o th ers h a d peris h e d
almost t o t ally ; an d such w as P as t eur s pre dict ion
I n tw o o u t o f th e four t een case s , in
s t ea d of th e
proph esie d d es tr uct ion, h alf an averag e crop w as
obt aine d l
A noth er int eres t in g illus t rat ion concerns D ar
w in s S peculation s re gard in g th e formation o f c oral
Before D ar win w rot e on th e
reefs an d atoll s
subj ec t i t was g enerally believe d th at th e coral
at olls w ere forme d by th e coral polypes g ro win g
upon submer g e d volcanic crat ers D arwin in sis t e d
th at as th e polypes canno t live belo w a d epth o f
100 fee t an d are kille d by exposure t o suns h ine an d
air an d coul d not th erefore h ave g ro w n upw ar d
from th e vas t d epth s to wh ich th e coral masse s
ext en d each at oll mus t h ave be g un as a frin g ing
reef about an island almos t t o uch in g th e sh ore w ith
only a narro w an d sh allo w ch annel o f w at er be
tw een ; and th en became a barrier reef, th at is o ne
w ith a wi d er and d eeper ch annel of w at er separat
,

Tynd all , F rag men ts


M

f S cien ce,

pp

291, 292

162

IN

DUCTIVE

OGIC

in g from th e sh ore , o w in g t o th e slow but pro g res


th e islan d roun d wh ic h th e
s ive s ubsi d ence O f
polypes rs t beg an t o buil d Th en w ith th e furth er
an d complet e s ubsid ence of th e island beneath th e
w at er th ere remaine d a rin g o f coral w ith a cent ral
lag oon formin g th e s o calle d at oll D arw in says in
h is A ut ob io g raph y th at th e main feat ure s of h is
th eory w ere c onc ei v e d wh ile on th e voya g e, an d
th at even previou s t o seein g a t rue coral reef l He
says : N o o th er w ork of mine w as be g un in so
de d uc t ive a sp irit as this fo r th e wh ole th eory w as
thou ght out o n th e w es t coas t of S ou th A merica,
before I h a d seen a t rue c oral reef I h a d only t o
verify an d ext en d my Views by a careful examina
t ion o f l ivin g reefs Bu t it sh oul d be obs erve d
th at I h a d durin g th e two previou s years been in
c e s s an tly att en din g t o th e effec t s on th e s h ores of
S outh America o f th e int ermitt ent elevat ion o f th e
land, t o g eth er w ith d enudat ion an d depo s ition o f s e d i
ment T his necessarily le d me t o re ect much o n th e
effects o f subsi dence, an d it w as easy to replace in
imag inat ion th e c on tinue d d eposit ion o f se dimen t
by th e up w ar d gro wth of corals To d o th is was
to form my th eory of th e format ion o f barrier reefs

an d at oll s
It w ill th us be s een th at D arw in s d e duc t ion w as
ba s e d u pon previous in duction s in oth er S ph eres ,
th e result of h is o wn observation ; h e also t ells us
in th e same connection, th at h e h ad in th e prep a
rat ion o f his w ork on Coral Reefs, spent tw enty
month s of h ar d labor, readin g every w ork on th e
.

Lif e

and

Letters

f Ch arles Darwin ,

1887 , Vo l I
.

58

VERIFICATION

R DICT ION

16 3

AND P E

islan d s o f th e P aci c an d c onsul t in g many c h art s


He thus ma d e th e w i dely ext en de d O bservat ions o f
o th er men t ribut ary t o h is inferences concernin g
coral reef format ions D r Williams says of D ar
w in s insi ght in th is particular He saw more
clearly th an h is precursors h a d d one th e valid ity
o f th e d ic tum of Jo h annes M uller in th is
an d
in d ee d all h is w orks, th at th e mos t import ant
t ruth s in nat ural science are to be discovere d
neith er by th e mere analy s is o f ph ilosoph ical id eas
nor by simple experience but by reectiv e exp erience,
wh ic h d is t in gui sh e s th e e s s ential from th e accid en
t al in th e p h enomena observe d an d thus n ds
princ iples from wh ic h many experiences can be
d erive d 1 T h is is a very sat isfact ory an d s t rik
in g account o f wh at may be s tyle d th e combine d
in d ucto
d e duc t ive t em per o f min d and especially
as embo die d in so em inent a s t ud ent of nature as
D ar win
Bacon in sist s th at ant icipat ions o f natu re are
sources o f innumerable errors an d th at th e t ruly
scient ic me th o d consis t s in an int erpret at ion o f
nat ure as it is reveale d t o th e percep tion th rough
d ire c t observation an d experiment I t is , h o w ever,
lar g ely th rou gh th ese anticipat ions th at pro g re ss
in s c ience is att aine d T h ere may be anticipat ions
wh ich are cons i d ere d nal an d all att empt s at veri
c atio n re g ar d e d as unnece s sary an d even as im
pert inen t Resul t s d e duc t ively att aine d are th en
assert e d with d o g mat ic insis t ence as th ough po s
.

,
,

Darwin , Coral Ree


fs

ix

Prefato ry n o te by Dr

W illiams

U CTIVE

164

IND

O GIC

sessin g th e convincing po wer o f fac t s th emselves ;


and all appeal t o controver tin g or excep tional cases
are set asi d e w ithout even so much as a respectful
h earin g S uch antic ipat ions o f nature ri ght fully
fal l und er th e scornful repreh ension of a Bacon
Bu t th ere are oth er anticipat ions wh ich serve as
a S pur to a more penet rat in g observat ion an d more
painst akin g experiment, in ord er t o square th eory to
fac t s S uch antic ipat ions are th e glory o f s cience "
S uppose suc h ant ic ipations are disprove d by
subs equent experimen t or observat ion ; th ey h ave
serve d a high purpose in su gges tin g invest igat ion
alon g l in es which oth erw i s e w oul d h ave remaine d
unth ou ght o f Antic ipat ions alone are barren ; an
ticipatio n s lead in g to veri cat ion are pro duct ive O f
valuable resul t s To this th e h is tory o f scient i c
th ought bears abund ant t es timony P rofessor Clif
ford h as mad e th e po w er o f pre dict ion one of th e
e s sent ial ch arac t eris tic s O f s cienti c th ou ght He
says in his essay on th e A ims an d Instru ments of
S cientic Tho ught th at th e d ifference be tw een sci
en tic an d merely t ech nical th ou ght is j us t th is :
Bo th of th em make use of experience t o d irect hu
man ac tion ; but while t echnic al th ought or skill
enable s a man t o deal with th e s ame circums t ances
th at h e h as me t w ith before, scient i c th ou ght en
able s him t o d eal with d ifferent c irc ums t ances th at
h e h as never met with before 1 He c it es two illus
tration s wh i c h are ad mirable example s o f s c ient i c
pre di ct ion Th e rs t relat es t o th e sugg es t ion o f
Fl eemin g J enkin re gar d in g s t r u ctural bracin g
It
,

Cliffo rd

Lec ture s

a nd

E s s ays , V o l I
.

128

VERIF ICATION

AND PRED IC ION

16 5

h a d been kno w n before th at in an arch every part


is compresse d or push e d by o th er part s ; an d every
part of a ch ain is in a s tat e o f t ension th at is,
pulle d by th e oth er part s I n many cases th ese
form s are unit e d in th e common g ir d er wh ich co n
sis t s o f two main pieces o f wh ich th e upper act s
as an arch an d is compre s se d, wh ile th e lo w er o n e

c
act s as a h ain an d is pulle d
N o w, say s Pro fes
so r Cli ffor d
suppose th at any g oo d, practical en g i
neer makes a bri dg e or a roof upon some approve d
patt ern wh ic h h as been ma d e before He d esi gn s
th e s ize an d s h ape o f it t o suit th e openin g wh ich
h as t o be spann e d ; select s his mat erial accor din g
t o th e locality assi g ns th e s t ren gth wh ic h mu s t be
g iven t o th e several part s of th e s t ructure accor d
in g t o th e loa d wh i c h i t w ill h ave t o bear T h ere
is a great d eal of th ou ght in th e making o f th is
d esi gn wh ose success is pre dict e d by th e appli c a
tion of previous experience ; it requires t ech nical
skill o f a ve ry h igh or d er, but it is no t s cienti c
th ou ght O n th e o th er h an d Mr Fleemin g J enkin
d es igns a roof c onsis tin g of two arch es brace d t o
g eth er ins t ea d of an arch an d a ch ain brace d
t o g e th er ; an d al th ou gh th is form is quit e d ifferent
from any kno wn s t ructure , yet before i t is built h e
assi g ns with accuracy th e amoun t o f mat erial th at
must be put int o every part o f th e s tr uct ure in
O rd er t o bear th e require d load, an d this pre dict ion
may be t rus t e d w ith perfect security Wh at is th e
nat ural comment on th is
Wh y th at Mr Fleeming
1
J enkin is a scientific en gineer
1 Cliffo rd Lec tures and E ssa s V o l I pp 127 128
y
,

16 6

IN

DUCTIVE

O GIC

T h e secon d illus t rat ion wh ic h P rofe s sor Clifford


g ives is as follo ws : Y ou kno w th at if you make a
dot o n a piece o f paper, an d th en h ol d a piece O f
I celan d spar over it yo u will see no t one d ot, but
two A mineralo g i s t by measurin g th e angle s o f a
cry st al , can t ell you wh eth er or no it possesse s th is
property w ith out look in g th rough it He requires
no scient i c th ou ght t o d o th at But S ir W illiam
Ro w an Hamil t on th e lat e as t ronomer royal o f I re
lan d kno win g th ese fac t s , and also th e explanat ion
o f th em wh ic h F resnel h a d g iven th ou ght about th e
subj ec t an d h e pre dic t ed th at by looking th rough
cert ain crys tals in a particular direction w e s houl d
see n o t two dot s but a continuous circle Mr L loy d
ma d e th e experiment an d saw th e circle a resul t
wh ich h ad never been even sus pect e d Th i s h as
al way s been consi d ere d o n e of th e mos t s ignal
inst ance s of s c ient ic th ought in th e d omain o f
phy sic s It is mos t dis tinc tly an application of
experience g aine d un d er cert ain circums tanc es to
ent irely d ifferent c irc ums t ances 1
T h ere is al s o an in d irec t meth o d of pre dict ion
varyin g somewh at from th e one alrea dy d e s cribe d
an d yet s imilar t o it It is calle d pre dic tion by
inversion o f cause an d effec t
T h ere are m anv
ca s es in wh ic h c aus e an d e ect are relat e d in a
reciprocal manner so th at not only w ill th e cau s e
pro duce th e effect but th e e ffect operat in g as a
cause w ill bring about th e orig inal cause as an
effect, it may be in a m o die d form but clearly reco g
n izabl e as such
P rofessor Tyn d all sai d of F ara day
,

Cliffo rd

Lec tu res

and

E ss ays , Vo l I
.

pp

128 , 129

V ERIFICATION

th at

R D ICT ION

16 7

AND P E

th e s t ron g t en d ency of h is min d t o look upon


th e reciprocal act ion s o f natural forces gave birth
1
to h is g reat es t d iscoveries
F or in s t ance, O ers t e d
h a d prove d th at an elec tric current w ill pro d uce
mag netism an d F ara day t akin g th is as a sugg es tion,
inferre d th at mag net ism mi ght pro duce an elec t ric
current ; in th e year 1831 h e d evise d a suitable
experimen t o f int ro ducin g a bar magnet int o a coil
o f insula t e d copper w ire , an d th en w ithd raw in g th e
ma gne t wh ils t th e tw o end s o f th e w ire w ere con
n ecte d w ith a d is t an t g alvanome t er, wh ic h in dicat e d
th e presence of the elect ric current T hus , h is in
ference receive d subs tant ial v e ric ation 2
It h as , moreover, been foun d th at wh en a g iven
cause pro duces a cert ain c e ct th en if th e effect be
pro d uce d in some oth er manner th e proces s w ill t en d
to pro d uce th e original cau s e but invert e d as re g ar d s
its direc t ion o r nat ure
Fo r ins t ance i t is know n
th at h eat w ill expan d g ases ; no w if a g as be re
liev ed of th e pres s ure of th e vessel enclosin g it, it w ill
expan d by V irtue o f it s o wn elas tic po w er pro d uc
in g, h o w ever, col d in th e surroun d in g atmosph ere
S o also h eat w ill cause a bar o f iron t o expan d
D r Joule prove d th at if iron w ere expan d e d by
mech anical force it w oul d be accompanie d by col d
Inasmuc h as ind ia ru bber is relat e d t o h eat in an
oppo sit e manner t o th at of iron, being contrac t e d
by h eat inst ea d o f expan d ed, w e w oul d, accor din g
t o th e law above expresse d, nat urally expect th at
a mech an ical expansion of ind ia rubber w oul d g ive
.

f S cience , p 338
The Art of S cientic Discov ery , p 594

Ty nd all , F rag ments

Go re

16 8

OGIC

IN UC IV E L

h eat an d a cont raction pro duce col d A n e xperi


ment may be t rie d by sudd enly s t ret chin g a rubber
band wh ile th e mi ddle part is in th e mouth ; wh en
s tret ch ed, i t g ro w s w arm ; wh en relaxe d, it seems
col d 1
Ag ain as h eat w ill mel t many subs t ances, if w e
can re duce th e same subs tanc e fro m th e soli d to
th e l iqui d s tat e w e w oul d expect as a resul t th e
ne g ative of h eat namely col d Th is occurs in all
free zin g mixt ures , as th e afnity of sal t for w at er
causes it t o melt ic e th us pro ducin g c ol d in th e
surroun din g at mo sph ere s ufcient t o freeze crea m
or oth er S im ilar subs t ance inasmuc h as, pass in g
from soli d t o liqui d w at er absorbs h eat from all
s ubs t ance s near it ; this absorp tion pro ducin g art i
c ial c ol d surroun din g it
T h e re c iprocal action
of h eat an d col d is illus t rat e d in an int eres tin g ex
?
rim en t d escribe d by Tai t
He t ook a bar of ic e
e
p
s uppor t e d h or izont ally at e ith er en d an d over th e
m iddle of th e bar h e put a ne wire and put equal
w eight s t o th e tw o en d s of th e w ire T h e w ire
g ra d ually by th e ac t ion of th e w eight s cut th rou gh
th e bar of ice an d yet i t w as observe d th at th e path
o f th e w ire w as ins t ant ly replace d by th e freezin g
a g ain o f th e mel t e d portion pro duce d by th e press
ure an d wh en th e w ire h a d wh olly t raverse d th e
entire thickne ss o f th e bar th e bar it self w as int act
an d e v en s t ron g er alon g th e line of th e cuttin g th an
before Th e explanat ion of th is experiment is th at
inasmuc h as h eat mel t s ic e th en wh en ice is mel t e d
1 Jevo ns P rin ci les o S c ien ce p 545
p
f
.

Tait , Recent A d v ances in P h ysic a l S c ience ,

pp

9 9 , 100

VERIFICAT ION

R D T O

169

AND P E IC I N

by pressure as in th is case of th e w ei ght e d w ire


col d, th e ne g ative of h eat is in duce d ; th us as th e
w ire w as force d by th e w eight s int o th e ice th e press
ure upon th e ice mel t e d i t makin g i t col d er so th at
th e wat er pro duce d pas sin g aroun d th e c h ille d w ire
an d bein g th us relieve d o f pressure froze ag ain
F ara d ay pre d ict e d cert ain mag ne t ic ph enomen a
by thi s meth o d wh ic h are specially int eres tin g
as illus t rat ions of th is kin d o f pre d ict ion
It
seems th at A ra g o h ad observe d in 1824 th at th e
number O f oscillat ions wh ich a mag ne t ize d nee dle
makes in a g iven t ime un d er th e in uence of
th e earth s mag net is m , is very muc h lessene d by
th e proximity o f cert ain met allic masses , an d es
E mployin g th e latt er subs t ance
pec ially o f copper
in an experiment upon a ma g net ize d n ee dle h e suc
c e ed e d in re d u c in g th e number o f i t s V ibrat ions in
a g iven t ime from th ree h un d re d t o four Takin g
th e experiment as a basis for his inference F ara
d ay pre d ic t e d th at since th e presen ce Of a metal at
re s t s t ops th e os c illat ions of a mag net ic nee dl e th e
nei ghborh oo d o f a ma g ne t at res t ought t o s t op
th e mo tion o f a ro t at in g mass of metal He th ere
fore procee d e d to put h is inference t o th e t es t o f
act ual experimen t by suspen d in g a cube of copper
to a tw is t e d th rea d wh i c h w as place d be tw een th e
W h en th e
poles o f a po w erful elect roma gnet
th rea d w as left t o it self i t be g an t o spin roun d
w ith great velocity but s t oppe d th e momen t a
po w erful current passe d th rou gh th e electrom ag
net 1 Ag ain, as h eat applie d t o th e junction of
,

Gano t

P h ysics ,

pp

7 97 7 9 8

17 0

IN

DUCTIVE

OGIC

two met allic bars as ant imony and bismuth pro


d u c ed an ele c t ri c current, i t w as inferre d th at if an
elec tric c urrent w as mad e to pass th rough such a
j unc tion it w oul d pro duce col d, an d such prove d t o
1
be th e case
I n th e general proces s of veri cation it oft en
h appe n s th at s eem ing exc eptions occur wh ic h are
direct contra dict ions of th e la w w e are att emp tin g
t o prove A n d i t is in d ealing w ith suc h c ases
th at one s po w er of discriminat ion is mo st fully
t axe d It is nece s sary t o make a mos t c areful
distinc tion betw een seemin g an d real exc ept ions
Profe s sor J evon s h as g iven a very exh aus t ive cla s
th e different kin d s o f ex c ep t ional
s ic atio n of
ph enomena which it is w ell t o h ave in min d in
or der t o kno w in any invest i g ation th e various po s
s ible co m pl icat ion s th at may ri s e ? Th e e xc ep
tio n al ph enomena as g iven by J evon s, are
1 I maginary o r false excep t ion s ; th at is fact s ,
obj ect s o r event s wh ic h are no t really wh at th ey
are su ppo s e d to be
2 A pparen t bu t con gruent excep t ions wh ic h
th ough apparently in c on ic t with a law o f nature,
are really in ag reement with it
3 S ing ular ex c ep t ions , wh ic h really a gree w ith
a law of nat ure but exh ibit remarkable an d u nique
re s ul t s of it
4 Diver gent excep t ions wh i c h really procee d
from th e or dinary action of kno wn proce ss es o f
,

Jevo ns P rin cip les of S cience p 547


S ee Jevo n s Ch ap te r XXIX in his P ri ncip les of
,

Exc e tio nal Ph eno me na

S cien ce,

on

VERIF ICATION

R D TO

17 1

A N D P E IC I N

nat ure but wh ich are excessive in amount o r


m ons t rous in ch aract er
5 A cci d ent al excep tion s arisin g from th e int er
ference of some ent irely dist inct but know n law o f
nature
6 N ovel an d u nexpl aine d excep t ions wh ic h lea d
t o th e d iscovery o f a ne w serie s o f law s an d ph e
n o m en a mo d ifyin g o r dis g uis in g th e e ffec t s of pre
v io u sly kno w n law s W i th out
bein g incon s is t ent
w i th th em
7 L imi t in g excep t ions , s h o w in g th e falsi t y of
a suppo s e d la w t o s ome c ase s t o which it h a d been
ex t en d e d, but not aec ting it s t ru th in o th er cases
8 Cont ra d ict ory or real, ex c ep t ions , wh ic h lea d
us to th e conclus ion th at a suppose d hypoth esis o r
th eory is in oppos it ion t o th e p h enomena of nature
an d m us t th erefore be aban d one d
I t will be seen th at amon g so m any po s sibilities
o f in t erpre t at ion an ex c ep t ion d oes no t nece s s arily
prove th e rule as th e o l d a d a g e w oul d h ave it ;
nor d oes th e exception o n th e oth er h an d n e ce s
sarily d isprove th e rule or la w It mus t be in eac h
case s t ric tly an d ad equat ely int erpret e d, whi ch re
quire s a penet rat in g sag ac ity an d a th orou gh kno wl
e dg e Of th e ph enomena un d er inves ti g at ion
In th e process o f ver i cat ion, th e ques tion n at
u ral ly su gg es t s it self : Ho w many verifyin g in
s t anc es are s ufc ient t o d et ermine th e un iversal
val i d ity o f a given law
Th is ques tion will be
reco g nize d as an Ol d d ifcul ty n o w present e d in
anoth er form ; but in real ity it is th e perplexin g
problem o f d e t erm inin g th e lo gical g roun d o f in
,

17 2

I NDU CTIVE

O GIC

d uct ion Wh at is o u r w arrant fo r proceedin g from


kno wn an d veri e d ins tanc es t o unkno wn ph eno m
ena of th e s a me kin d it is t rue but as ye t beyon d
th e pale Of o u r experience
Th e w arrant for o u r
g eneralization d oes not lie wh olly in th e number o f
verifyin g ins t ance s In addition t o th e cect wh ich
mere number pro duces in c onrmin g o u r belief
th ere is th e cond enc e whi ch w e feel in th e co n
an d wh ich w e are
stan c y of th e or d er of nat ure
1
c on s t raine d to assume as a fun d amen t al po s t ulat e
T h erefore w e say th at th e verifyin g fac t s mus t be
o f su c h a number
and of such a nat ure as w ell
th at th ey give evid enc e of a uniform ity which
t ranscend s all s uppo s ition o f mere coinc i d ence an d
compel s us t o att ribut e it t o th e uniform ity o f
nat ure i t self in whic h w e n d a w arrant for our
g enerali zation A s Profes s or Cliffor d h as remarke d :

T h e aim of s cient ic th ou ght is t o apply pas t ex


Th e in st rument
perience s t o ne w c irc ums t ances
is an observe d un iformity in th e cour s e of event s
By th e u s e o f this ins t rum ent it give s us inform a
t ion t ranscen din g our experienc e it enable s us t o
infer th in g s th at w e h ave not seen from th i n g s
th at w e h ave s een ; an d th e e vid ence fo r th e t ruth
of th at informat ion d epend s on o u r s upposin g th at
th e uniformi ty h ol d s goo d beyon d o u r experience 1
I n ex t en din g kno wle dg e and pre dict in g resu lt s
beyon d th e sph ere of experience mo d ern scienti c
inve stig ation is lar g ely ind ebt e d t o th e principles
an d meth o d s of m ath ematics Math ematic al law s ,
.

1
2

348
S ee Sig w art , Log ic , V o l II
Cli ord , Lec tures and E ssays , Vo l I
.

pp

131, 132

VERIFICATIO N

D TO

17 3

AND PRE IC I N

applie d to th e d at a g iven in sense percep tion g ive


indi cations of th e necessary relations th at mus t
exis t in th e observe d ph enomena, and all th at th ey
involve
T hus, th at wh ic h is given d irectly in con
s cio u sn es s is supplement e d by th at wh ic h is g iven
in direc tly as math emat ically necessit ate d
Th e
m ath ematic o experiment al me th o d in p h ysics h as
le d to very ric h and import ant result s wh ic h h ave
prove d prac t ically its efciency as a scient i c
m e th od
-

CH AP T ER " III

HY P O T HE S I S

induct ive process cannot procee d t o any


g reat ext ent or att ain satisfac t ory re s ult s W ith out
th e aid o f some hypoth e sis A n hypoth e s i s is a
supposition re g ar din g th e cause o f a p h enomenon,
which w e make ei th er as prel im inary t o an experi
ment whic h w ill prove or di s prove th e supposition,
o r in lieu O f an experimen t o r sys t emat ic observa
t ion wh en s uch are impossible o win g t o th e peculiar
con dit ions of th e p h enomenon it self We see th ere
fore th at th e fram in g o f h ypoth eses h as a d ouble
fun ction F irs t c onsi d ered as preliminary t o ex
periment We foun d th at in cases wh ere tw o th ree
o r more elemen t s ent er int o a c omplex ant ece d ent i t
is impossible oft en an d al w ays impract icable to t es t
th e various po ssible combinat ions separat ely in or d er
T h e combinations in
to no t e th eir d ifferen t resul t s
c omplex p h enomena are in d e nit ely g reat, an d th e
is olat ion of cert ain element s in or d er to es t imat e
th e exac t resul t of th e c o mbine d force o f th e o th er
comb ination s is ext remely difcul t an d Oft en im
possible Th erefore th e min d d iscard s s ome com
bin atio ns as irrelevan t o th ers as impo s s ible , an d
selec t s one perh ap s as th e mos t likely cause of th e
TH E

"

17 4

HY PO THE SIS

17 5

g iven cect T h is selec t ive func tion of th e min d


th erefore, in d icat es th e line o f experiment in a d e
t erminat e m anner an d d oes no t leave th e p h enom
ena to in d et erminat e an d h ap h azar d inves t i g ation
Consi d er for ins t ance S O em inent an experiment er
as Ch arles D ar w in so fert ile in all kin d s of e xperi
ment al Pesou rc es ; yet it is sai d of h im th at every
experim ent w as th e resul t of a t ent at ive th eory
th ou ght out in a d vance o f all ac t ual t es t by a s ag a
cions insi ght int o th e necessary con d itions of th e
int errelat e d ph enomena before h im His son F ran
cis D arwin say s of h im in h is Reminiscences : He
oft en said th at no one coul d be a g oo d Ob s erver
un les s h e w as an ac t ive th eori zer I t was as th ou gh
h e w ere ch ar g e d w ith th eorizin g pow er rea dy t o
o w into any c h annel o n th e sli ght es t d is turbance,
ever s m all coul d avo id releasin g
s o th a t no fac t h o w
a s t ream of th eory an d th us th e fac t became mag
In th is w ay it nat urally
n ie d in t o impor t ance
h appene d th at many unt enable th eories o c c urre d t o
h im ; b ut fortun at ely h is richness of ima g ination
w as equalle d by h is p o Wer o f ju dg in g an d con d emn
He was
in g th e th ou ght s that occ urre d t o h im
j us t t o h is th eories and d i d no t cond emn th em um
h ear d ; and so i t h appene d th at h e w as willin g to
t es t wh at w oul d seem t o mos t people no t at all
w orth t es tin g T h ese rath er w il d t rial s h e calle d
fool s experime nt s an d enj oye d ext remely A s an
example I may ment ion th at, n din g th e co tyle d ons
o f Bio ph ytu m t o be h igh ly sensi t ive t o v ibration s
o f th e t able h e fancie d th at th ey mi ght perce ive
th e V ibrat ions of soun d, an d th erefore mad e me play
.

17 6

IN

DUCT IV E

O GIC

my bassoon close t o a plant Th e love of experiment


w as very s trong in him an d I can remember th e w ay
h e w oul d say , I s h an t be easy t ill I h ave t rie d it,
as if an out si de force w ere d rivin g h im 1
Hypo th esis an d experiment were in th e h an d o f
D arwin like a two e dg e d S wor d, wh ic h h e employe d
with rare skill an d effe c t An h ypoth esis is t o be
re g ard e d no t only as th e precursor o f experiment,
but it also funct ions as a meth o d of explanat ion
wh en act ual veri cation is impossible We see
th is cons t ant ly in our every d ay l ife W e are com
pelle d ag ain an d a g ain t o account for sit uat ion s
wh ich occur th at are impo s sible for us t o repro du ce
in th e form of an experiment th at w e are able t o
observe but once S ome explanation is require d to
satis fy ment al demand s wh ich are imperative in
th is re g ar d Th e explanat ion wh ich s eems mo st in
keepin g w ith th e sum o f fact s in our posses sion is
th e hypo th esis whic h w e frame ; so also in explain
ing th e cond uc t o f o th ers by conj ecture as t o th e
mo st reason able mot ives th at w ill s atisfact orily
a c count for th e s ame ; s uch hypo th eses w e are con
We are not al w ays
stan tly compelle d t o assu m e
able t o perceive th e relat ions exis tin g betw een
fact s as th ey come int o th e sph ere O f o ur e x peri
en c e an d yet w e are cons t raine d t o think of th em
as relat e d ; bu t in or d er t o sys t emat ize th em w e
must s up ply ment ally th e lacun ae whic h appear in
th e ph enomena as perceive d T h is supposi tion th at
is nece ss ary t o cons t ruc t fac t s int o sys t em is an
hypoth esis
.

Life

an d

Letters

f Ch arles Darwin

Vo l I
.

126

HYPO THE S I S

17 7

A n illus t rat ion o f an hypoth esis su gg est in g sy st e


mat ic observat ion an d experiment is foun d in th e
h is tory of th e dis covery o f vaccination by J enner
It seem s th at wh ile a mere youth , pursuin g h is
s t u dies at S o dbury, h e ch ance d t o h ear a casual
remark ma de by a count ry g irl wh o came t o h is
mas t er s sh op fo r a dvi c e T h e small pox w as men
t io n e d w h en th e g irl sai d
I c annot t ake th at d is
ease for I h ave h ad co w po x 1 Th is O bservation ,
expres s in g th e com m on supers t ition of th e simple
co unt ry folk appe ale d to J en ne r s min d as an in
ch oat e hypoth e s is S ei zin g upon it as a su gg es tion
o f pos s ible value h e procee d e d t o make d ili g e nt
inquiries an d careful O bservat ions wh i c h nally
le d h im t o th e d iscovery o f vaccinat ion
A n illus t rat ion o f hypo th esis as explanat ion of
p h enomena beyon d th e rang e o f experiment is foun d
in th e h ypo th esis as t o th e sourc e o f th e sun s
energ y A n enumeration o f th e different hypo th e
ce d upon th is subj ec t is g iven by Tait in
s e s a d van
h is Recent A d v ances in Physical S cien ce ? Th e
o ld no t ion th at th e sun is a hu g e re , or some th in g
of th at kin d, is one wh i c h w il l only occur t o one
th inkin g o f th e matt er for th e rs t t ime ; but w ith
o u r mo d ern c hemi c al kno w le dg e, w e are enable d t o
say th at, massive as th e sun is if its mat erials h a d
consis t e d o f th e very bes t m at erial s fo r g ivin g out
h eat, th at enormous m ass of some
miles
in radius coul d h ave s u pplie d u s w i th only about
A mass of
5000 years o f th e presen t ra diat ion
.

1
2

G o re

pp

T he A rt

151 ff

f S cien tic Discov er?

495

17 8

IN

DUCT IVE

O GIC

coal of th at size w oul d h ave pro duce d very much


less th an th at amount o f h eat N or w oul d th e
mos t ener g et ic ch emicals kno wn to u s, combine d
in propor tion for givin g th e great es t amount of
h eat by act ual ch em ical combinat ion supply th e
sun s pre s ent w as t e for even 5 000 years
T h ere
fore as w e all kno w th at g eolo g ical fact s, if th ere
w ere no o thers point t o at leas t as h igh a rad iat ion
from th e sun as th e present fo r at all event s a few
h undre d s o f th ousan d s o f years back an d per
h aps al s o in dicat e even a h i gh er rat e o f ra diat ion
from th e sun in Old time th an a t present it is
quit e obvious th at th e h eat o f th e s un canno t
possibly be s uppl ie d by any c h emical process of
wh ich w e h a v e th e sli ght e st conc ep tion

N o w, if w e can n d o n th e oth er h an d, any


physical explanat ion of th is c onsis t ent w ith any
pre s ent kno wledg e w e are boun d to t ake it an d use
it as far as w e c an rath er th an say : T his ques t ion
is t ot ally unans w erable unle s s th ere be ch emical
a g encies at work in th e sun o f a far more po w er
fu l or d er th an any th in g w e meet with on th e
earth s surface I f w e can n d a th orough ly in
telligibl e source o f h eat, wh ic h th ou gh d epen din g
upon a different physical c ause from th e usual
o n e, com
bus tion is amply s ufcient t o h ave sup
plie d th e s un w ith such an amount of h eat as t o
enable it to rad iat e for perh aps th e las t h und re d
million s of years at th e s ame rat e as i t is no w
ra diat in g th en I say w e are bound t o t ry th at
hypoth esis rs t an d ar gue upon it un t il w e n d i t
incon sis t ent w ith some th in g kno w n
An d if w e
.

17 9

HYPO THES I S

do

not n d it inconsis t ent with any th in g th at is


kno w n wh ile w e nd i t complet ely capable o f
explaining o ur d ifcul ty, th en it is not only p h ilo
soph ic t o say th at it is mos t probably th e ori g in
o f th e s un s ener gy , bu t w e feel ourselves co n
s t raine d t o ad mit it
N e wton lon g ago t ol d us
th is Obli g at ion in h is Rules of Philosophizing
N o w it is kno wn th at if w e w ere t o t ake a mass
of th e mos t perfect combust ibles which w e kno w
an d let i t fall upon th e sun merely from th e
earth s dis t an c e th en th e w ork done upon it by
th e sun s att ract ion durin g it s fall w oul d g ive it s o
lar g e an amount o f kinet ic ener gy wh en it reach e d
th e suns surface as to pro duce an impact wh ic h
w oul d represent six th ous an d t imes th e amount of
ener gy wh ic h coul d be produce d by it s mere burn
in g
I t appears th en th at o u r nat ural an d only
,
t rus tw orthy mo d e Of explainin g th e sun s h eat at
present in t ime pas t, an d fo r time t o come must be
some th in g cl Osely analo g ous t o, but not i dent ical
w ith wh at w as c alle d th e nebul ar hypoth esi s o f
L aplace, th e h ypoth esis o f the fall ing to g eth er
ut ion in S p ace
from
r
d
ely
sca
tt
r
d
i
t
rib
e
e
d
s
u
) of
(
th e matt er wh ich n ow forms t h e various suns and
planet s We nd by calculation in wh ic h th ere is
no pos sibilit y of lar g e error, th at th is hypoth es is
is th oroughl y co m
petent t o explain o n e hun dre d

illions
yea
f
r
s
sOl ar rad iation at th e pre s ent
o
m
ble of sh ow in g
n d it is cap a
rate, perh aps more ; a
us h o w it is th at th e sun, for th ousand s o f years
toge ther, can par t wi th ener gy at th e enormous
,

'

'

'

'

180

TV

OGIC

IN UC I E L

rat e at wh ich it does s t ill par t with it an d yet


not apparently cool by perh aps any measurable
quantity
I n con rmation of th is not only is th e hy
po th esis it s elf capable of explainin g th e amount s
o f en er gy wh i c h are in que s t ion bu t also recent in
v e stig atio n s ai d e d by th e s pect roscope h ave s h o w n
us th at th ere are gi gant ic nebular sys t ems at g reat
d is t ance s from our solar sys t em in th e proces s of
physic al d e gra d at ion in th at very w ay, by th e
fall in g t o g eth er of s cat tere d mas ses, an d wi th
numerous consequent d evelopment s of h eat by
impac t s
Wh at are calle d t emporary s t ars form
anoth er splendi d an d s till more s t rikin g ins t ance
of it, as wh ere a st ar s udd enly appears, of th e rs t
ma gnit ud e, o r even bri ght er th an th e rs t ou t
s h ini ng all th e plane t s fo r a month or tw o at a
t ime an d th en aft er a l ittle time become s invisi
ble in th e mos t po w erful t elescope T h in gs of
th at kin d are c onst antly occ urring on a larg er o r
s maller s c ale an d th ey can all be ea sily explaine d
on th is supposition of th e impact o f g rav itat ing

ma ss es
S uc h a h ypoth esis it w ill be s een embraces all
th e fact s observe d in o n e self c onsis t ent sy s t em
T h e o th er hypo th eses are ina dequat e t o ac c ount
satisfac torily for th e ph enomena
T h e vali dity
o f th is hypo th esi s l ies in its be in g bo th a d equat e
an d con g ruent as w ell ; experim ent or corroborat ive
observat ion bein g o ut o f th e ques t ion, w e are, as

Tait says con st raine d to adm it it


Mr D ar win gives
Gnumeratio n an d c rit i c ism
,

HYPOTHESIS
of

18 1

th e

d ifferent hypoth eses wh ich h ave been su g


g es t e d to explain th e ext inc tion of th e gi gantic
an imals kno w n to h ave exis t e d upon th e earth
His accoun t w ill g ive an in d icat ion o f th e nat ural
propensity o f th e min d to fr ame hypo th eses c o n
cernin g ph enomena wh ich lie out si d e th e sph ere
both o f observat ion an d experiment Mr D arwin
says : I t is impossible t o re ect o n th e ch an g e d
s t at e o f th e A m erican Cont inent with out th e d eep
es t as t onis h ment F ormerly it mus t h ave s warme d
with g reat mons t ers ; no w w e n d mere pi gmies
compare d w ith th e ant ece d ent allie d races T h e
g reat er number if no t all, o f th e s e extinct qua d
ru ped s, live d a t a la t e perio d an d w ere th e con
te mporaries of mos t o f th e exis t in g sea s h ells
W h at th en h as ext erminat e d so many spec ies an d
wh ole g enera
Th e min d at rs t is irresis t ibly
h urrie d int o th e belief of s ome g reat cat as t roph e ;
bu t th us t o d est roy animals bo th lar g e an d s mall ,
in S outh ern P at a g onia, in Brazil on th e Cor dillera
o f P eru , in N or th A merica up t o Beh rin g s St rait s
w e mu st s h ake th e en tire frame work o f th e globe

A n exam inat ion, moreover of th e g eolo gy o f


L a Plat a an d P at a g onia lead s t o th e belief th at
all th e features of th e l an d resul t from S lo w an d
g ra dual c h an g es It appears from th e ch aract er
A sia, Aus t ral ia, an d in
o f th e fossils in Europe
N orth an d S outh A merica, th at th ose con d i tions
wh ic h favor th e life o f th e larger quad rupe d s
w ere lat ely coext ensive w ith th e w orl d
Wh at
th ose con dit ions w ere, no o n e h as yet even c on
j ectured It coul d h ar d ly h ave been a ch an g e o f

182

INDU CTIV E

OGIC

t emperat ure, wh ic h at about th e same t ime d e


s t roye d th e inh abit ants of t ropical t emperat e, an d
arct ic latit ud es o n bo th si d es o f th e globe I n
N orth A merica w e positively kno w from Mr Lyell
th at th e lar g e qua d rupe d s live d subsequently t o
th a t perio d wh en boul d ers w ere brou ght int o lati
tu d es at wh ich iceber g s no w never arrive ; from
conclusive but in d irect reasons w e may feel sure
th at in th e south ern h em isph ere th e Macrauch enia
al s o l ive d lon g subsequently t o th e ice t ransport in g
boul d er perio d D i d man, aft er hi s rs t inroa d
int o S outh America, dest roy , as h as been su gg es t e d,
th e unwiel dy Me g ath erium an d th e o th er Ed en
tat a ? We mus t look at leas t t o some oth er cause
for th e d es t ruction of th e l ittle t ucut uco at Bah ia
Bl anca and of th e many fossil mice an d o th er
s mall quadrupe d s in Brazil N o one will imag ine
th at a drou ght even far severer th an th ose whic h
c ause s uc h losses
in th e provinces of L a P lata,
coul d d es t roy every in di vi dual o f every S pecies

frOm S outh ern P at a g onia t o Beh rin g s St rai t s


W h at sh all we Say O f th e ext inct ion o f th e h orse
D i d th o s e plain s fail o f pas ture Wh ich h ave s ince
been overrun by th ousand s an d hund re d s of th ou
sand s o f th e d escen dant s o f th e s t ock intro duce d
by th e S paniar d s
Have th e subsequently intro
d uc ed sp ecie s c ons ume d th e foo d o f th e g rea t
ant eced ent rac es
Can w e believe th at th e Gapy
bara has t aken th e foo d of th e Toxo don th e
G uana c o o f th e Mac rau ch enia, th e exis t in g small
E d ent at a o f th eir numerous gigan tic pro to types 9
Cer t ainly n o fact in th e lon g h is t ory of th e w orl d
,

"

'

'

'

'

'

"

HY POTHES I S

183

is so s t artlin g as th e w i de an d repeat e d ext er


m in atio n s o f its inh abi t ant s 1 Mr D ar win s o wn
h ypo th esis concernin g th is ph enomenon is rath er
in d e nite, bu t neverth eless as d e ni t e as th e ex
t reme complexity o f th e fac t s w ill allo w
He
says th at th ere are cert ain causes operatin g in
nature th eir exact ch aract er remainin g unkno w n,
s uc h t h at th e to o rapi d increase of every species ,
even th e m os t favore d, is s t ea dily ch ecke d pro
d ucin g in some cases rari t y an d in o th er s ex
t in ction if th ese causes O perat e w ith unusual
ef c acy : His h ypo th e sis marks a t en d ency wh ose
nature neverth eless, remains conceale d
I n all th ese w i d ely d ierin g hypo th eses w e see a
cert ain m ental lc o n straint t o o ffer some explanation,
even thou gh it be but a d is g uise d c onfession o f

i gn orance, as In Mr D ar win s hypoth esis


A n illus t ration of an hypo th e sis t o expl ain o h
serve d ph enomen a th at c anno t be furth er t est e d is
th at g iven in th e follo w in g ins t ance c ite d by Pro
fesso r Tyn d all : At E rith, in 18 6 4, th ere occurre d
Th e
a t remen d ous explosion o f a po wd er ma gazine
villa g e o f Erith w as some miles dis t ant from th e
ma g azine, but in nearly all cases th e w in do w s w ere
s h att ere d ; an d it w as not iceable th at th e w in do w s
t urne d aw ay from th e ori g in o f th e explosion suf
fered almos t as muc h as th ose wh ich face d it L ea d
sas h es w ere employe d in Erith ch urc h ; an d th ese
b ein g in som e d e g ree exible enable d th e w in do w s
to yiel d to pre s s ure w ith ou t muc h fract ure of gla s s
Every w in d o w in th e c h urch , front an d back , was

'

Darwin , Vo yag e of

N aturalist, V o l I
.

223

INDU CTIVE LO GIC

184

bent inward s I n fac t, as th e soun d wave reach e d


th e ch urch it separat e d ri ght an d left, an d fo r a
momen t, th e e di ce w as claspe d by a girdl e of in
t ensely compresse d air wh ic h force d all it s win
d o ws inwar d s Aft er compression, th e air in th e
ch urc h no doubt dilat ed, and t en d e d t o rest ore th e
w in do w s t o th eir rs t con dition T h e ben ding in
o f th e w in d o w s , h o w ever, pro d uce d but a small
c on d ensat ion o f th e wh ole mass O f air w i th in th e
ch urc h ; th e force of th e recoil w as, th erefore,
feeble in comparison w ith th e force o f impac t,
an d in s ufcient t o und o wh at th e latt er h a d ac
1
c omplis h e d
Here also is a set of con d it ions
th at mus t be satis e d by a correc t hypo th esis
T h e ph enomenon w as not capable o f repet it ion by
any experiment P rofessor Tyn dall , th erefore pict
ures t o h is min d wh at must h ave h appene d beyon d
th at which w as Ob s erve d, in or d er t o accoun t for
th e result wh ic h act ually h appene d He ll s up
th e unseen from wh at h e kn o ws o f th e nat ure o f
s oun d w aves an d th us cons t ruc t s o n e self consi s t ent
syst em wh ic h inclu d es both th e seen an d th e nu
seen th e kno wn an d th e unkno wn, th e observe d
an d th e inferre d
It will be notice d in th is an d o th er illus t rat ions
of hypoth es is h o w larg e a par t is pl aye d by th e
imag inat ion I t is th e imagin at ion wh ich lls o ut
th e vacant spaces in th e pi c ture of percep t ion
W ith some, th e funct ion o f imag inat ion is asso
It mus t in
c iated w i th fancy rath er th an fac t
this connect ion, h o w ever, be clearly emph asize d
-

Tynd all , ou S ound ,

23

185

YPOTHESiS

th at th e imag inat ion wh ic h cons t ruc t s h ypo th eses


mus t be th rou gh out in t ouc h w ith fac t I t mu s t
represent to th e min d n o t wh at fancy su gg es t s but
wh at th e kno wn fact s necessi t at e Th e unseen is
cons t ruc t e d o u t o f th e d e t erminin g cond itions o f
th e seen
It is th is d e duc t ive fun c tion of th e
ima g ination th at g ives to i t a s t rictly lo gical si g
nic an ce
F or ins t ance P rofes s or Tyn d all s reason
in g concernin g th e Erith ch urc h w as some wh at as
follo w s : T h e win d o w s are all bent in ward th ere
fore th e pressure mus t h ave O perat e d o n al l si d es
from w ith out in w ar d ; such pressure coul d only
occur upon th e supposit ion th at th e soun d w aves ,
separat in g ri ght an d left, wh olly enc ompasse d th e
c hurch, et c In each cas e, th at wh ich h e picture d
to his m in d as h appenin g, w as re g ar d e d by h im as
act ually nec es s it at e d by th e fac t s as ob s erve d
P rofe ss or Ty n d all h as mos t ad mirably d iscusse d
th e S cient i c U se o f th e Im ag inat ion ; an d his
lec t ure un d er th a t tit le every s tu d ent, bo th o f lo g ic
I quot e o ne passag e
o r o f science
S h oul d rea d
from it wh ic h h as S peci al bearin g u pon wh at h as
jus t been said : We are g ift e d with th e po w er o f
I ma ginat ion combinin g wh at th e G ermans call
an
d
by
A
e an d
W
th is po w er w e can li ght en th e d arkness which sur
round s th e w orl d o f th e senses T h ere are t ories
in scienc e wh o re g ar d ima ginat ion as a facul ty t o
be feare d and avoi d e d rath er th an employe d T h ey
h a d observe d its ac tion i n w eak ve s s el s and w ere
un dul y impres s e d by it s disas t ers But th ey mi ght
with equal jus t ice point t o explo d e d boilers as an
.

U TV

186

IND C I E

LO GIC

ar gumen t a g ain st th e use o f st eam Boun de d an d


cond itione d by c o Operant Reason, i m a ginat ion be
comes th e might ies t ins t rum ent o f th e s c ient i c di s
coverer N ewt on s passag e from a falling apple
t o a fallin g moon w as, at th e out set a leap o f th e
imag ination Wh en William T h om s on t ries t o
place th e ul timat e par t icles o f matt er betw een h is
co mpass point s , an d to apply t o th em a s cale of
millimet res, h e is po w erfully ai d e d by th is facul ty
An d in muc h th at h as been recently sai d about
pro toplasm an d life, w e h ave th e outgoin g s o f th e
imag inat ion g uid e d an d cont rolle d by th e kno wn
analo gies o f scienc e In fac t without this po w er
o u r kno w le dg e of nat ure w oul d be a mere t abul a
tion o f coexis t enc es an d sequences We s h oul d
s t ill believe in th e s ucce ssion o f day an d ni ght of
s ummer an d w int er ; but th e soul of F or c e w oul d
be dislo dg e d from o u r universe ; caus al relation s
w oul d d isappear, an d with th em th at scienc e
wh ich IS now bin ding th e part s o f nature t o an
org anic whole 1
I n all th e illus trat ions w h ic h h ave been g iven,
and , in fac t, in all exampl es of th e framin g o f
hypo th eses, it w ill be seen th at th e ment al functions
S pecially in operat ion are th o s e o f analysis an d
a separation p f th e element s as far as
s y nthesi s
pos s ible into th eir S imples t fo rms o f expression,
and the buil ding th em t o g eth er int o some one sy s
t em wh ose unity lie s in t he assum e d hypoth es is
Mr Ven n h as especially emph asized th is aspect of
.

'

Tyn d a11, Use

a nd

Limit

the I mag in a tio n i n S c ienc e,

187

HY PO THE S I S

hypo th esis an d h is ch ap t er o n th is subj ect w ill


w ell repay a careful readin g 1
E very s upposit ion , h o w ever is n o t necessarily an
hypo th esis in th e lo g ical or scienti c si gni cance
of th at t erm I t w ill be necessary th erefore, t o
ment ion s ome o f th e requirement s wh ic h a lo g ical
h ypo th e s is s h oul d sat isfy
1 A n h ypo th e s i s sh oul d be plausible ; th at is i t
sh oul d be no fanciful , o r merely conj ectural e x pla
nation of th e p h enomena in ques t ion
T h e sup
posit ions o f th e int erference o f S pir it s, o r in a
myt h olo g ic al a g e o f th e g o d s , to ac coun t fo r per
ple xin g si tuat ions o r obscure h appenin g s h ave n o
rank as hypo th ese s s o, al s o F at e is oft en referre d
t o as a convenient confession o f ignorance in lieu of
a satisfact ory explanation S pinoza h as remarke d
upon th is as fo ll o Ws z T h ey wh o h ave d es ire d t o
nd ; scope fo r t h e d isplay o f th eir in g enuity in
as s i gnin g c auses, h ave h a d reCo u rse tO a ne w s tyle
o f argui ne nt to h elp th em in th e ir conclusion s ,
namely, by re duct ion n o t t o th e impo s sible or ab
s urd , bu t t o i gnorance or th e unkno w n a proce dure
whi ch s h o w s very plain ly th at th ere w as no o th er

co urse O pen t o th em
T h e difference betw een a scient ic hypbth esis
an d a popul ar explanat ion concerning th e s ame
p h enomena may be foun d In D ar win s account o f
a s t ran g e belief whi c h i s g en eral amon gs t th e
inh abit ant s o f th e Maldiv a at olls n amely th at
corals h ave root s an d th erefore th at if merely
broken do wn to th e surface, th ey g ro w up again ;
,

'

'

Venn

E mp iric al Log ic ,

Chapter " VI

188

iNDUCTIv E Lo cic
'

'

but if roo te d out, th ey are permanently des t roye d

By th is mean s th e inh abit ant s keep th eir h arbors


clear ; an d th us th e F renc h g overnor o f St Mary s
in Ma dag ascar cleare d out an d mad e a beaut iful
little port at th at place 1 T h eir explanat ion,
h o w ever, is purely fanc iful h avin g no bas i s in fac t
I n c ont rast, D arw in s hypoth esis t o explain th e
fact s in th e cas e is o f a lo g ically scienti c nature
an d is as follo w s : I nasmuch as loose se diment is
injurious t o th e l ivin g po lypifers, and as it is prob
able th at s and w oul d accumulat e in th e hollo w s
forme d by t earin g out th e corals but not on th e
broken an d proj ect ing s t um ps th erefore in th e
form er case th e fres h g ro wth o f c oral mi ght be
th us prevent e d by th e deposit e d se diment
2 T h e sec on d requiremen t is th at th e h ypo th esis
mus t be capable o f proof or disproof T his d oes
not d eman d a t es t by experiment necessarily ; for
th at as w e h ave seen, may be impossible I t does ,
h o w ever, require th at some fact s S h oul d be forth
comin g th at will eith er con rm th e hypo th e sis o r
dis prove it T h ere are cases, h o w ever, as L otze
su gg es t s wh ose very nature preclu d es th e possi
bility o f provin g or d isprovin g th e h ypoth e s is
frame d t o account for th em F or ins t ance, th e
very common an d S imple hypoth e s is of re g ar din g
th e st ars, wh ic h are apparently but small point s o f
li ght, as bo d ies of vas t size, only very remot e from
us, is in it self inca pable of bein g eith er refut e d o r
con rme d by su bsequent d isc overy L ot ze s ay s

We mus t abi d e cont ent if o u r hypo th eses are

Darwin , Coral Reefs ,

89

HYPOTHESIS

th ink able

and

189

useful if th ey are capable o f ex


plainin g all int erconnec te d appearances , even such
as w ere still Imk no wn wh en w e cons t ruct e d th em ,
if, th at is to say th ey are indirec tly con rme d by
th e agreement o f all th at can be d e duce d from th em
in th ou gh t w i th th e ac t ual pro gress o f expe rience
But if w e w oul d be so for tunat e as to n d an
h ypo th esis wh ich will not lac k th is subsequent
conrmation w e mus t n o t simply assume anyth in g
th at can be barely c onceive d as real ; w e mus t onl y
as sum e th at which besi d es bein g th inkable, con
form s, so to speak, t o th e un iversal cus toms of real
ity, or to th e special loc al cus t oms wh ic h prevail in
th at d epartment of ph enomena t o wh ic h th e obj ect
w e are inves t i g at in g belon g s 1
It is to be specially observe d th at wh ile th e re
quirement of proof o f an hypo th es is may be waive d
in th e sph ere of ph enomena wh ere proof is mani
festly impossible, s t ill, wh ere proof is available, an
h ypo th esis mus t never be so fram ed as to rend er
th e re quire d te s t ei th er impos sible o r imprac ticable
It mus t
3 T h e hypo th esis mu s t be a d equat e
cover all th e fac ts in th e cas e A n out st an d in g
fac t wh ic h i t cann ot explain is sufcien t to con tro
vert such an hypo th esis A kno w le dge o f th e dis
tinc tion be tw een postulat e an d hyp oth esis, and o f
th e relation wh ic h nomin al ly exis t s betw een th e two,
w ill h elp us to appreciat e more clearly th e forc e
A s d ened by
o f thi s requiremen t o f a d equ ac y
L o tz e, a pos t ulate expresses th e con dit ions wh ic h
mus t be se t up, o r th e groun d o f explanation wh ic h
,

Lo tze , Log ic ,

35 3

190

INDUC IVE LO GIC

mus t be g iven by some real ity, force, o r event, before


w e can th ink th e ph enomenon in th e form in wh ic h
it is present e d t o u s ; it th us requires or pos tulat e s
th e presence of someth in g th at c an accoun t for th e
An hypoth es is is a Conj ec ture whic h
g iven e ffect
seeks t o ll up th e pos t ulat e th us abs t ractly s t at e d
by specifyin g th e concre t e causes forces , or pro
ces s e s o ut O f wh ich th e ph enomenon really arose
in th is partic ular case, while in o th er cases maybe
th e s ame postul at e is t o be sati s e d by utt erly d if
ferent th ou gh equivalent combinat ions of forces o r
1
A ccordin g t o th is di s tinct ion
act i v e element s
as applie d t o th e problem o f th e source o f th e
sun s ener gy t h e po st ulat e w oul d be th e s um o f
cond itions which require explan at ion ; namely th e
tremend ous rad iation o f h eat ex t en din g th rou gh
th ousand s and thousan ds o f years Th e post ulat e
th erefor e requires a force a dequat e t o supply for so
long a perio d so g reat an amoun t o f energy We
foun d th at or dinary combust ion of th e mos t h ighly
combustible mat erial s w oul d not as an hypoth es is ,
sat isfy th e c on di t ions wh ic h Obt ain in th e pos t ulat e ;
nor woul d th e l iberation o f ch emical energy s t an d
as an hypoth es is ad equate to sat isfy th e pos tulat e ;
th e hypoth esis o f impact of masses upon th e sun s
surface from immens e d is t ances present s a force
sufcient t o mee t th e requirement s o f th e pos tul at e
Moreover w e see in this illus t rat ion h o w th e hy
poth esis is a part icular an d c oncret e expres s ion of
th e cond it ion s expresse d in g eneral an d in abs t rac t
t erms in th e postul at e T h e e s s ent ial ch aract eris t ic
.

Lo tze , Log ic ,

pp

349 , 350

19 1

HYPOTHE S I S

th erefore o f th e h ypoth esis is th at it sh all perfec tly


sat isfy all th e c ond it ion s expresse d in th e pos tulat e
T he hypo th esis th at nature abh orre d a vacuum in
or d er t o account fo r th e rise o f w at er in a t ube o r
pump w as s een to break d o wn u tt erly wh en it w as
foun d th at th e w at er d i d no t rise beyon d some
th irty th ree feet Th e d em an d o f th e pos tulat e in
th e case w as a force o f precisely such ma g nit u d e
th at it woul d balance a col u mn of w at er th irty th ree
feet in h ei ght T h is force, precisely satisfyin g th e
con di t ions of th e pos t ulat e, is foun d in th e hypoth
esis that th e at mosp h eric pre ss ure is s uch a ma gni
t u d e as to exert a pressure equival ent t o a column
Th e
o f w at er some th ir ty th ree fee t in h ei ght
s t rength o f th e h ypo th esis lies in i t s exac t a nd
appropriat e tt in g int o th e fact s o f th e problem
Ano th er illus t ration of th e tt in g of hypoth esis
t o pos tulat e an d o n e wh ere th e con d it ion s o f th e
pos tulat e are ex t remely comple x I h ave c h osen
from Mr W allace s w ork , On N atural S election :

T h ere is a Mad agas c ar orch is


th e A ngroecum
n immensely lon g an d d eep
u ip ed ale w i th a
s es q
nect ary
Ho w di d such an ex t raor dinary org an

come to be d evelope d ? Mr D ar win s explanat ion


is th is Th e pollen o f th is o wer can only be
rem
ove d by th e bas e o f th e probosci s of s ome V ery
lar g e moth s ,wh en t rying t o get at th e nect a at th e
bo tt om o f th e v e Ss el Th e m o th s with the long est
probosces woul d do th i s mos t effect ually ; th ey
w oul d be rewar d e d fo r th eir lon g t on g ues by g et
t in g th e mo s t n ec t ar ; whils t, o nth e oth er h an d, t h e
o wers with th e deep es t nect aries woul d be th e bes t
.

"

19 2

IN

DUCT IVE

O GIC

fertilize d by th e lar g es t moth s preferrin g th em


Consequen tly th e d eepes t nect arie d orc h id s an d th e
long es t t on g ue d mo th s woul d eac h confer o n th e
o th er an ad van tag e in th e battle o f life Th is
woul d t end t o th eir re s pec t ive perpet uation, an d
t o th e const ant len gth e nin g o f nect aries an d pro
In th e A ng rcecu m ses qu ip ed ale it is meces
bo sces
s ary th at th e proboscis s h oul d be force d int o a
particular part o f th e o w er, an d th is w oul d onl y
be d one by a larg e mo th buryin g i t s proboscis to
th e very base an d st rainin g to d rain th e nec tar
from th e bott om o f th e lon g t ube, in whic h it ocon
pies a depth o f o ne o r two inch es only
N o w l et
u s st ar t from th e t ime wh en th e nec t ary w as o nl y
h alf its present length o r about six inc h e s, an d was
chiey fertilize d by a spec ies o f m oth wh ich ap
pe are d at th e t ime O f th e plan t s o w erin g an d
wh o s e proboscis was o f th e same length A mong
th e millions of o w ers of th e A n graicu m pro duce d
every year some w oul d al w ay s be s h ort er th an th e
averag e some lon g er
T h e former, o win g t o the
s t ruc t ure o f th e o w er
w oul d no t g et fertilize d,
because th e moth s coul d g e t all th e nec t ar with out
forcin g th eir trunks d own t o th e v ery base By
th is proces s alone th e avera g e len gth o f th e nec
tary w oul d annually increase because th e S h ort
nect arie d o w er s bein g s t erile an d th e lon g ones
h avin g abundant offs prin g exactly th e same effect
w oul d be pro duce d as if a gar d ener d es t roye d th e
s h or t ones , an d so w e d th e s ee d o f th e lon g one s
only ; an d th is w e kno w by experience w oul d pro
d uce a re g ular inc rease of length, s ince it is th is
.

193

HYPO THES I S

very process wh ich h as increase d th e si ze an d


c h an g e d th e form o f ou r c ul tivat e d fruit s an d
o w ers But th is w oul d lea d in t ime to such an
increase d len gth o f th e nect ary th at many of th e
mo th s coul d only jus t reach t o th e surface o f th e
nect ar an d only th e fe w w ith excep tionally lon g
t runks be able to suck up a consi d erable port ion
T his w oul d cau s e many mo th s t o ne g lec t th e s e o w
ers, because th ey coul d no t g et a s atis fyin g supply
o f nec t ar, an d if th ese w ere th e only mo th s in th e
count ry th e o w ers w oul d un d oubt e d ly s uffer and
th e furth er g ro wth o f th e nec t ary be c h ecke d by
exactly th e same process wh ich h a d le d to it s
increase

Bu t th ere are an immense varie ty of moth s o f


various len gth s of proboscis an d as th e nect ary
became lon g er o th er an d lar g er spec ies w oul d be
c ome th e fer t ili zers an d w oul d carry o n th e proce s s
t ill th e larg es t moth s became th e sole a g ent s N o w,
if n o t before th e mo th w oul d also be affec t e d ; fo r
th ose w ith th e lon g est probosces w oul d g et th e mo st
foo d, w oul d be th e s t ron g es t an d mos t vi g orous
w oul d visi t an d fert ilize th e great es t number of
o w ers an d woul d leave th e larg es t number of
d escen d ant s T h e o w ers mos t complet ely fertil
ize d by th ese moth s bein g th ose wh ic h h a d th e
lon g es t nec t aries , th ere w oul d in each g eneration
be on th e avera g e, an inc rease in th e len gth o f th e
nect aries , an d also an avera g e inc rease in th e len gth
o f th e probo s ces of th e m o th s ; an d th is w ou l d be a
n ecess ary result from th e fac t th at nat ure ever u ct
u ates about a mean, o r th at in every g enera t ion
.

19 4

IN

DU CT IVE

O GIC

th ere w oul d be o w ers with lon ger an d sh ort er


nectaries, an d moth s with lon g er and s h ort er pro
I may h ere ment ion th at
bo sc e s th an th e avera g e
some of th e lar g e S ph inx moth s of th e t ropic s h ave
probo s ce s nearly as long as th e nec t ary o f A ngrce
cu m sesquip ed ale
I h ave carefully measure d th e
proboscis o f a specim en of M acrosila clu entiu s from
S outh A meric a in th e collec t ion o f th e Britis h
Museum an d nd it t o be nine inche s an d a quart er
lon g
O ne from t ropical Africa (Macro s ila mor
A species h avin g
gan ii) is seven inch es an d a h alf
a probosc is two or th ree inc h es lon g er coul d reach
th e nec tar in th e l on g es t o w ers o f A ngroecum ses
qu ip ed ale wh o s e ne c t aries vary in len gth from t en
t o fourt een in c h es T h at such a moth exist s in
Mad ag ascar may be s afely pre d ict e d ; 1 an d nat ural
ists wh o V i s it th at islan d sh oul d s earc h fo r it w ith
as muc h cond ence as as t ronomers s earch e d for
th e plane t N eptune an d I ven ture t o pre dic t
th ey will be equally s ucce s sful 2
I h ave given this quo t ation at len gth in or der to
in d icate no t only th e tt in g of h ypo th esis t o th e
fact s Observed but also th e lar ge an d important
part performe d by the ima gination in repro duc in g
along parallel lines th e natural hi s tory of th e orch i d
an d m oth T h e hypothesis reach es back over an
in d enit ely lon g pas t, by virt ue of th e necessit ies
.

'

'

hat

It is interesting to note t

b ve

s inc e

Mr W allace w ro te th e
.

K ir y , in his E urop e an Mo th s and B u tterflies , ma


mentio n o f o n e o f th e S h ing id ae wit a ro oscis twelv e inc
a o

l o ng "
2

Wallac e

o u N atural S elec tio n ,

h p b

pp

2 7 127 5

es

h es

HYPOTHES I S

195

observe d in th e pres ent, an d in accor d ance w ith


w ell es t ablis h e d analo g ies an d approve d in duct ion s
T h e funct ion of th e ima g inat ion especially promi
nent is th at of it s d e duct ive insi ght , wh ic h is able
t o pict ure t o th e min d th e inevit able result s o f th is
an d th at con d it ion as furnish e d by th e pos tulat e,
an d th en t o t such n ecessit at e d result s int o one self
consis t ent sys t em , with noth in g left un explaine d,
incon g ruous o r cont radi ct ory
Anoth er illus t rat ion o f an hypoth esis coverin g a
larg e number of complex fac t s is th at o f th e fert i
lizatio n o f cert ain o w ers by means of th e w in d
A s g iven by S ir J o hn L ubbock w e h ave th e follow
in g fac t s an d th e correspond in g explanat ion o f
th em : W ind fert ilize d o w ers as a rule, h ave no
color emit no scent an d pro duce no h oney , an d are
re gular in form Color scent an d h oney are th e
th ree ch aract eris t ics by wh ich insect s are att ract e d
t o o w ers Again, as a rule w in d fert ilize d o w ers
pro d uce much more pollen th an th ose wh ic h are
fert ilize d by insect s
T h is is neces s ary because
it is obvious th at th e ch ances a g ainst any g iven
pollen g rain reac h in g th e s t i g ma are muc h g reat er
in th e on e case th an in th e o th er Every o ne h as
observe d th e s h o w ers o f yello w pollen pro d uce d by
th e S cot ch r Again i t is an a dvant ag e to wi nd
fert ilize d plant s t o o w er early in th e sprin g before
th e l eaves are o ut because th e latt er w oul d cat c h
muc h o f th e pollen an d th us int erfere with it s
access to th e st i g ma Again, in th ese plant s th e
pollen is less adh erent, so th at i t can be easily
blo wn aw ay by th e w in d, wh ic h w oul d be a d isad
-

19 6

IN DU CTIVE

O G IC

vant ag e in mos t plant s wh ic h are fert ilize d by


insec t s
Ag ain suc h o w ers ge nerally h ave th e
s t igma more or less branc h e d, o r h airy, which evi
d en tly mus t t en d t o increase th eir ch ances o f
1
cat ch ing th e pollen
Th ere is h ere a s t ruc t ural
a dap t ation of th es e plant s t o th e c ircumst ances
d esi g ne d t o explain th em , so th at th e consequent
self cons is t ent syst em thus forme d carries w ith it
th e w eight o f convict ion
Th ere are some explanat ions wh ich d o no t per
fe c tly corre spon d t o reality, an d ye t wh en th eir
nat ure is known th ey may be pro t ably use d, not
t o represent reality but t o assis t th e mind by an
app roximate represent ation t o bett er appreciat e th e
fac t s as th ey really are relat e d o n e t o anoth er
T h e s e s o calle d ct ions are useful especially in
math emat ics We suppose, for ins t ance, ins cribe d
an d circumscribe d poly g ons o f a circle w ith ever
increasin g number o f si d es grad ually approach in g
an d becomin g coinci d ent nally wi th th e curve
it self T his latt er w e kno w t o be impo s sible an d
yet we may t reat th at wh ic h h appens only appro xi
mat ely as th ou gh really h appening merely as an
ai d t o th e ima ginat ion ; an d a ction, if al w ays so
un ders t oo d may th us prove h elpful in th e repre
sentatio n o f reali ty more clearly t o o u r min d s
4 T h e h ypo th esis , moreover s h oul d involve no
cont rad ict ion T h is is clearly a requirement th at
is d e duct ive rath er th an in ductive depen din g upon
th e fun d a men tal principle o f cont radic tion lyin g at
th e basis of th e d e duct ive sys t em o f lo gic
,

Lu

bb ock

S cien tific Lec tu res ,

pp

9 , 10

19 7

HYPO THES I S

Th e hypo th esis sh oul d be as simple as possible


N o involve d explanat ion th at my sties rath er th an
clears th e dif cul t ies present e d can rank as a true
hypo th esis S imp lex v eri s igillum T h is require
ment, o f course canno t in all cases be st rictly co m
plie d w ith ; for th e ph enomena t o be explaine d may
present such a de g ree o f complexity th at a s imple
hypoth esis w oul d be alt o g eth er o ut o f th e ques t ion
Fo r ins t ance, th e hypoth esis of a s ubs t ance llin g
th e un iverse, an d pervadin g all part icle s o f matt er
h o w ever soli d and closely kn it t o g eth er, a subs tance
it self more soli d th an s t eel and more elas t ic as w ell ,
such a s uppo s it ion seem s not only to o involve d, but
also even to belie th e or dinary j u dgment s o f common
sense An d yet th is un dulat ory hypo th esis is more
an d more conrme d by every a d vance of science in
th e kno wle dg e o f th e p h enomena o f li gh t an d h eat
It sometimes h appens th at th e very failure o f an
hypoth esis forms a subs t ant ial c ontribut ion t o th e
pro g ress o f th ou ght, lea din g t o th e rea djus t ment o f
a receive d th eory o r s t imul atin g researc h in ord er
t o d iscover th e t rue in plac e of th e false hypo th es is
A s Mr Tait says : We all kno w th at if th ere
h a d no t been a pursuit aft er th e p hilosoph er s s t one,
ch emis try coul d not yet h ave been any thin g like th e
I n th e same w ay we can
g i g ant ic science i t no w is
say th at mo d ern ph ysics c oul d not y et h ave covere d
th e g roun d i t no w occupies h a d it n ot been for th is
experiment al seekin g fo r th e so calle d perpetual
m ot ion an d th e consequen t es t ablis h ment o f a d e
1
nit e an d scienti cally useful ne g at ive
T h e cir
.

'

Tait ,

Recent A d v ances

in P h ysic al

Science

69

19 8

INDU CTIVE

OGI C

th eory o f th e orbit s of th e planet s, while


incorrec t, yet m ad e th e t ransition easier from th e
hypo th es is of circular mot ion t o th at o f mo tion in
an ellip tical orbit wh ich is th e t rue th eory I t
oft en h appens th at an h ypo th esis may no t be
wh olly wron g but may nee d c orrec t ion and this is
oft en provi de d for, no t by a t ot al rej ec tion o f th e
hypoth esis in quest ion but by supplementin g it by
s o calle d subsi d iary hypo th eses
A s t o th e t es ts of a correc t hypo th esis in addit ion
to th e ful lment o f th e requiremen t s alrea dy men
tion e d D r Wh ew ell h as e s pec ially emph asize d th e
impor t ance o f wh at h e h as s tyle d a Consil ience o f

I n ductions
A n hypo th esis receives a conrmatory
s t rength ening o f its vali dit y wh en it enable s u s t o
explain an d d et ermine ca s e s n o t only o f th e s ame
kin d as th e p h enomena out o f wh ich th e hypo th e
sis it self h as d evelope d, bu t case s which arise in
a sp h ere entirely different from th at wh ich g ave
mat erial ori g inally for th e formation o f th e hy
An hypoth esis th at can thus be carrie d
po th e sis
int o ne w t errit ory as an effec t ive ins t rum ent o f
researc h is th ereby d oubly accre dit e d A s D r

Wh e w ell remarks
A ccor din gly th e cases in
which in duct ions from classe s of part s alt o g eth er
different h ave th us j ump ed together belon g only t o
th e bes t e st abl ish e d th eories whic h th e h ist ory o f
A n d as I sh all h ave occas ion t o
s c ience con t ains
refer t o th is peculiar feat ure in th eir evi d ence I
will t ake th e l iberty of describin g it by a part icular
ph rase ; an d will t erm it th e Co nsilience of Ind uc
It is exemplie d princ ipally in some of th e
tions

c ul ar

199

HYPOTHESI S

g reat es t d iscoveries T hus it w as foun d by N e wt on


th a t th e d oct rine o f th e att ract ion o f th e sun vary
in g accor d in g to th e inverse square of th e d is tance,

wh ich explaine d Kepler s Third L aw, o f th e pro


po rtionality o f th e cubes o f th e d is t ances t o th e
squares o f th e perio d ic t imes o f th e planet s ex
plaine d also h is Firs t an d S econd L aws o f th e
ellip tical motion o f each planet ; al th ou gh no
conn ect ion of th ese la w s h a d been visible before
Ag ain, i t appeare d th at th e force o f universal g rav i
tation wh ic h h a d been inferre d fro m th e p ertu rba
tion s o f th e moon an d plan et s by th e sun an d by
each o th er,also account e d for th e fac t, apparently
al t o g eth er d issimilar and remo t e of th e p recessio n
of the equ in o xes
Here was a mos t s t rik ing an d
surprisin g coinci dence which g ave to th e th eory a
s t amp o f t ruth beyon d th e po w er o f in g enuity to
count erfeit 1
W h en tw o rival h ypo th eses can be submitt e d t o
th e t es t o f an experiment wh ich ne g at ives one an d
conrms th e o th er suc h a t es t in g is c alle d an eco
T h e name was rst given by
p erimen tu m cru cis
Bacon, an d h as me t w i th universal accep t ance in
scien t i c p h raseolo gy A crucial t es t as d e c isive
be tw een th e emission an d th e un dulatory th eory
o f li ght, is g iven in an experim
ent rs t t rie d by
F ath er G rimal d i, a Bolo g nese monk , in 166 5 If
a s h utt er be pierce d w ith a very small h ole, an d
th e luminous cone wh ich passes th rou gh th e orice
be examine d, th e cone w ill be foun d t o be much
.

Wh ewell

Art 110
.

N o vum Org anon Renov atw n, Bk

11 Ch
.

200

INDUCTIVE LO GIC

less acut e th an w oul d be expe ct e d consi d erin g only


th e rec tilinear tran s miss ion of th e rays as accor din g
t o th e em ission th eory I f th ere be int erpose d in
th e path o f th e luminous ray a secon d sh utt er,
pierce d w ith a h ole also it will be not ice d th at th e
ray s of th e s econd cone are even more d iver g ent
th an th ose o f th e rs t If th e imag e o f th e ori ce
be rece ive d upon a screen, a wh it e circle is seen
s urroun d e d by a d ark rin g next a wh it e rin g, even
more brilliant th an th e cent ral portion th en a secon d
dark rin g an d nally anoth er very faint wh it e rin g
I f in th e s hutt er wi th wh ic h th e experiment is
ma d e tw o very small h ole s are pierce d at a d is
t ance from eac h oth er o f one or tw o millime t res
an d th e two ima ge s receive d upon a screen in s uc h
a manner th at th ey o v erlap eac h o th er it is foun d
th at in th e cuticular se g ment forme d by th e over
lappin g o f th e imag es th e c irc les are more obscure
th an in th e p art wh ere th ey are s eparat e d Th us
l
by adding light to li ght d arkness is pro duce d
T h ese ph enomena are no w kno wn t o be consist ent
only w ith th e un dul at ory th eory an d di rect ly in
contradiction t o th e emission hypoth esis
M Romane s performe d several experiment s upon
bees which h a d th e fo rce o f crucial t es t s of two
oppo s e d hypoth eses one th at bees possess a g eneral
sense of direct ion irrespect ive o f any special kn o wl
e dg e o f th eir particular surroun din g s ; th e o th er
th at th ey are g ui d e d in th eir i ght by a kno wle dg e
of th e localit ies Wh ich th ey h ave been w ont to fre
quent M Romanes t ook a score of bees in a box out
,

'

Saigey, The Unity of N atural P henomena,

66

201

HYPOTHESIS

t o sea, wh ere th ere coul d be no land marks t o g ui d e


th e insects h ome N one of th em returne d h ome
T h en h e liberat e d a secon d lot of bees on th e sea
sh ore, an d none o f th ese returnin g h e liberat e d
anoth er lo t o n th e law n between th e sh ore an d th e
h ouse N one o f th ese ret urne d al th ou gh th e d is
t ance from th e law n t o th e h ive w as not more th an
two h un dre d yar d s
L as tly h e liberat e d bee s in
d i fferent part s o f th e ow er gar d en on eith er s id e
o f th e h ouse , an d th ese at once re t urne d t o th e
h ive ; an d with repet ition o f th e experiment, a
similar resul t even arrivin g at th e h ive before
h e h imself h a d t ime t o run from th e place wh ere
h e h a d liberat e d th em t o th e h ive A s th e g ar d en
w as a lar g e one, many o f th em h a d to y a g reat er
d is t ance in or d er t o reac h th e h ive th an th ose
liberat e d on th e front lawn T h e ir uniform su c
cess th erefore, in ndin g th eir w ay h ome so im
me diat ely w as no d oubt due t o th eir spec ial
know le dg e of th e o wer g ar den, an d n o t to any
g eneral sense of d irect ion l
T h e hypo th esis th at lead s t o veri cat ion by ex
periment represent s t rue scien t i c proce dure , an d
th at wh ic h h as ac t ual ly been th e m o s t e ffec tive
ins t rum ent o f researc h in all th e various sph eres of
h uman inves t i g at ion Th e ol d cont roversy be tw een
Mill an d W h e w ell a d mit s of a ready a djust ment in
thi s re g ar d W h ew ell emph asize d d is c overy as th e
h eart o f th e syst e m o f in duction lead in g to th e
framin g of hypo th eses wh ose ch ief t est w as not
.

Lubb o c k , 0 n

A nimals ,

pp

th e

269 , 270

S e nses , I ns tincts ,

and

I ntellig ence

202

INDUCTIVE LOGIC

experiment al so muc h as th e capabili ty of ac count


in g fo r th e g iven p h enomena
Mill on th e oth er
h an d insis t e d th at lo gic was essent ially proof, an d
not dis covery He, accor din gly empha size d th e
experime nt al t es tin g by mean s o f h is s everal me th
as bein g th e all important part of th e ih
ods
He h a d l ittle c onc ern fo r th e
d uc tiv e me th o d
ori gin o f th e sugg estion s as to th e mo s t likely
causal element s in th e mi d s t of a complex p h e
Th e primary func tion of lo g ic accor d in g
n o me no n
T h e i d eas
to h im is merely t o prove o r d isprove
of Wh e well an d M ill are not ne c essarily c o ntrad ic
t ory ; th ey can be re g ard e d as mutually s upple
mentary wh ich give s u s a t rue account o f th e i d eal
lo gic al metho d wh ere hypoth es is s u gg e s t s th e l ine
an d experiment in turn con rms
o f experiment
hypo th e sis I n suc h a meth o d, as can be s een in
th e ill u st ratio n g iven th ere is a bl endin g of d e d u c
t ive an d ind uc tive rea s onin g, wh ich is th e g eneral
charact eris tic of all ac t ual proces s es o f th ought
A s S i g wart h as so admirably put it : Wi th out
quickness of combination by which w e can call up
a number o f possible analo g ies an d apply th em t o
th e unexplaine d case ; w ith out a h appy po w er o f
divination which is guide d by un analy zable a ss oci a
t ions to discover th at analo gy wh ic h embrac es mos t
aspect s o f th e even t ; nally w ith out imaginat ion
t o const ruc t c onnec tion s for which th e only groun d
may be a h i dden sim ilarity, o u r th ought s if co m
pelled t o procee d s t ric t ly acc or din g to me th o d w oul d
frequently be cond emne d by th e impossibility of
d iscoverin g in th is way a sufcien t ly groun d e d con
,

203

HY POTHES I S

t o comple t e s t a gnation But th e fact is in


no w ay cont rary t o th e nature of in duct ion ; it is a
necessary consequence of it We canno t even be g in
th e process o f inference w ith out making g eneral
assump tions ; an d th e g eneral proposit ion wh ic h w e
g et by summin g up a number o f ins t ances is really
a h ypoth esis, to wh ich, it is t rue w e are le d c learly
an d c ert ainl y in th is case But betw een th ese mo st
g eneral presupposit ions upon which all ind uct ion
is g roun d e d, an d th e s imple s t case s to wh ich th ey
can be applie d, th ere is a w i d e re gion w ith in wh ich
th e h ypoth ese s wh ic h are al w ays necessary for in
duc t ion can only be forme d t ent atively , in or d er t o
g ive some d enit e d irec t ion t o inves ti g ation t o serve
in our analysis o f p h enomena int o th eir elem ent s as
a mean s o f breakin g up complet e ph enomena o n c er
t ain lines , an d t o invent th e experiment s wh ich will
1
make it possible t o conrm or refut e an opinion

nec tio n,

S ig wart , L og ic , V o l II
.

42 3

CH AP T ER " IV

A N A L O GY
oft en h appens th at th e cause of a ph enomenon
is dis clo s e d by th e fac t th at th e c ause of a s imilar
p h enomenon is kno wn an d th e inference th en fol
lo ws th at th e sim ilar ph enomena h ave s imilar
c auses
S uc h a pro c ess o f inference is d et ermina
tion by analo gy Analo gy cons id ere d in its rel a
t ion t o th e in duc t ive pro c e s se s occupies a tw ofol d
posit ion In th e rst place, wh en a complex p h e
n o m en o n is g iven as preliminary t o th e format ion
o f any hypo th e sis , as t o th e probable cause wh ic h
will in t urn lea d to experiment al d et ermination
by one o f th e inductive meth o ds th e min d instino
tiv ely examine s, w ith s w eepin g glance every d et ail
o f th e p h enomenon for th e purpo s e o f d is c over in g
s ome famil iar feat ures th at may prove su gg es t ive
o f kno wn relat ions an d funct ion s occurring in oth er
sph ere s Analo gical su gg es tion th erefore, initiat e s
every in duct ive in quiry
I n th e secon d place, in every in duc tive g eneral
izatio n th ere is an ext ension o f th e kno wn int o
unknow n re g ion s, by virt ue o f th e principle o f
analo gy expre s se d in wh at w e may s tyle its limit
ing case
Fo r ins t ance wh en w e h ave examine d
Ir
'

204

205

AN ALO G Y
a r
h

an d n d th em al w ays ch arac te r
izo d by th e mark B an d th en by g eneralization
rise t o th e proposition All A s are B, w e d o so by
reason of pos tulatin g an anal o gy betw een all th e

i
in d vid ual A s o f so st rictly an accurat e nat ure, that
it amount s to essent ial i d entity I h ave th erefore
cal le d th is th e limi tin g case o f analo
gy ; an d th is
resemblance o f par t iculars is th e groun d of all uni
versal s wh ereby th ey man ifes t an id entity in th e
mid s t of d ifferences We are th erefore j us ti e d in
af rming th at all in ductive g eneralizations present
an aspec t o f analo g ical inference
A nalo gy, consi d ere d as a mental process, is
g roun d e d in th e law of similarity T h is t en d ency
o f no tin g resemblance makes possible th e ex t ension
o f k no w le dg e
T h e format ion of o ur concept s is ,
in th e m ain, an analo g ical proce dure ; j ust as th e
g eneral izat ion of an un ivers al d epen d s upon our
discriminat ion o f th e element s wh ic h are similar
from th ose wh ich are d ierent W hi le analo gy
th us functions in all th e lo gic al processes of
th ough t, i t i s us ed in a more res tricte d sense t o
in d icat e th at mo d e o f inference especially whic h
procee d s from a numbe r o f observe d ch arac teris t ics
th at are similar, to oth ers whi ch are th ereby judg e d
T h is meth o d is very pot ent as
to be similar al so
I n 1845, F ara day dis
an ins t rum en t o f discovery
covere d the m a gnet ic ro t ary pol arizat ion o f light ;
by an alog ical reasonin g W aitman n in th e follo win g
y ear inferre d th at a s imilar resul t w oul d be att aine d
w ith a beam o f h eat, wh ic h w as aft er ward s experi

o
a
l
en
t
ll
ve
i
e
d
T
h
e
calle
d
na
r
kin
d

s
t
u
r
s
m a y
mber

of

206

INDUCTIVE

LOGIC

furni s h manifol d illus t rat ions o f conclusive anal o


g ies T h ey posse s s numerous properties, some Of
th em kno wn an d oth ers unk no wn T h ro u gh larg e
groups o f th e m are foun d similar c h aract eris tics
si d e by sid e with manifes t differences , an d yet th e
similarities are s o s t rikin g th at Oft en, wh en ne w
properties are discovere d in certain members O f th e
g roup th ere seems t o be g roun d for inferrin g th eir
exi st ence in o th er members of th e g roup also
Cert ain propert ies kno wn t o exis t in po t as s ium an d
s o d ium w ere inferre d t o be present in rubi d iu m an d
caesium ; th e carbonat es o f so dium an d po t a ss ium
are no t d e c omposed by a re d h eat an d it w as in
ferre d th at th e same w oul d prove t rue o f th e car
bonate s of rubi d ium an d c ae siu m ; an d such prove d
t o be th e cas e S ome o f th e st at ement s wh ich are
t rue of ch lorine are foun d t o be t rue o f bromine
an d io dine Mr G ore, h aving foun d th e m ol ecu
lar ch an g e in ant imony elec t ro d eposi t e d from it s
h e inferre d an d d is c overe d th e s ame in
c h lori d e
th at d eposit e d from its bromide and io di de S ir
Humph ry D avy h avin g discovere d th at pot assium
mi ght be isolat e d by means of elect rolysis imme
d iately inferre d an d procee d e d t o prove by e x peri
ment th at it woul d be possible also t o isolat e so d ium
an d oth er subs t ances of analo g ous properties 1
T h e principle of analo gy lies at th e basis o f all
clas s ication, th e separat in g and g roupin g t og eth er
in appropriat e d ivision s ind iv id uals whic h possess
certain salient att ribut es in c ommon
P rofessor J evons d enit ion o f classi cat ion em
.

Gore, The Art of S cientic Discovery,

522

20 7

ANALO GY

bo d ies at th e s am e t ime a full s tat ement of i t s


exact lo g ical si gnicance as an ins t rument of re
search , an d th erefore I g ive it in full : By th e
classi cat ion Of any series of obj ect s, is meant
th e ac tual o r i deal arran g ement t o g eth er o f th ose
wh ich are alike an d th e separation O f th ose wh ich
are unlike, th e purpose of th is arran g ement bein g
primarily, t o di sclose th e correlat ions o r law s O f
union of propert ies and circums t ances an d secon
d arily to facilit at e th e O perat ion s O f th e min d in
clearly conceivin g an d ret ainin g in th e memory th e
c h arac t ers O f th e Obj ec t in quest ion 1 I n d es crib
in g th e purpose of classi cat ion, th e latt er clause
is more a psych olo g ical d esi derat um th an lo g ical ;
th e former specicat ion cont ains i t s lo gical pur
pose ; namely t o d isclose th e correlations o r law s
O f union o f propert ies an d circum s t ances
T his
may be ill us t rated in th e g roupin g t o g eth er o f
pot assium , so dium , c aesium , rubi dium an d lith ium
T h is was
and callin g th em th e alkaline met als
d one by v irtue o f th e common c h arac teris t ics in
th e mi d s t o f th eir in d ivi dual peculiarit ies ; namely
th ey all combine very ener g etically w ith oxy gen
to d ecompose w at er at all t emperat ures , an d form
s t ron gly basic oxi des, wh ic h are high ly soluble in
w at er, yiel d in g po w erful caus t ic an d alkaline hy
d rate s from wh ich w at er canno t be expelle d by
b eat ; th eir carbonat es are also soluble in w at er,
Th e
an d each met al forms o nl y one ch lori d e
manifes t ad vant ag e O f classifyin g th ese m et als
t o g eth er lies in it s s u gg es tive capacity, as w e h ave
,

Jevons

P rincip les

f S c ience , p 677

208

INDU CTIVE

LO

GIC

already no t e d in illus t rations above g iven


So
many Observe d s im ilari ties su gg es t inferences by
analo gy ; wh en for ins t ance a ne w proper ty is
discovere d in any o ne or tw o o f th e met als of th is
cla s s , th e i dea imme d iat ely sugg es t s it self th at th e
same property may possibly ex t en d over all th e
me t als of th e same class N ot only is such an i d ea
s u gg e s t e d bu t alon g w i th it th ere exis t s an ant e
ce dent probabil ity respec tin g it s s olut ion in accord
anc e with th e sugg es t ion wh ich analo gy s tart s
A n excellent illus t ration O f th e prac tical resul t s
att aine d th rou gh a scienti c use O f classication is
foun d in Mr Lo ck yer s research es o n th e sun 1 A s
a g uid e as t o wh at element s t o look for in th e sun s
ph ot osp h ere, h e prepare d a classi c at ion o f element s
accor din g as th ey h ad or h a d no t been t rac e d in
th e sun, t o g eth er w ith a d et aile d s t at ement o f th e
ch emical nat ure Of eac h element
He w as th en
able to observe th at th e element s foun d in th e sun
w ere for th e mo st part, th o s e formin g s t able co m
pounds wi th oxy g en He th en inferre d th at th e
oth er elements wh ich w ere know n t o form s table
c ompoun d s w ith oxy g en w oul d , in all probabili ty ,
be foun d present in th e sun Startin g upon th is
s u gg es t e d t rack , h e succee d e d in d iscover in g v e
such met als
A nalo g ical inference carries special w ei ght wh en
i t is base d u po n th e principle of t eleolo gy ; th at
is wh en any observe d ph enomena s eem t o po s sess
s t ruc t ural cont rivan ces ad ap t e d t o en d s in s ome
de gree at leas t, similar to h uman c ont rivances
.

Q uo ted by Jevo ns in P ri ncip les of S c ie nce,

676

209

O GY

A NA L

d esi gne d to pro duce certain propos ed en d s W h en


th is similari ty is apparent it sugg es t s th e possi
bility th at an O bserve d con t rivance in nature may
subserve en d s beyon d th e possibility Of O bserva
t ion an d wh ich th erefore, may be inferre d really
t o exis t We h ave seen th at th e g roun d o f all
inference lies in th e represent at ion o f any g iven
p h enomena o f conscio u sne s s as co h erin g in o n e
sy s t em wh ic h compreh end s th e several part s in
a common unity Of such a nat ure th at kno w in g
some Of th e part s an d th eir relation s, w e infer th e
ch arac t er an d funct ion o f o th er part s no t kno wn
an d ye t wh ich th at already kno w n necessit at es
A n d amon g th e many kin d s of relation th at may
O bt ain bet ween par t an d part, o r part an d wh ole
th e t eleolo gical is a very common o n e an d more
over by i t s nat ure nec essit at es cert ain c onsequenc es
th at lie beyon d th e sph ere Of O bservat ion an d ye t
neverth eless , may very properly be supplie d by in
ference I n o th er w or d s th e causal connections in
a s ys t em are no t m erely th ose o f an efcient o r
a formal cause ; th ey m ay w ith a like force an d
su gg es t iveness , be consi d ere d in th e li ght o f a nal
cause ; th at is, th e presence o f mean s a d ap t e d t o
cert ain en d s , o r of or g an s a dap t e d t o cer t ain n ece s
sary funct ions o r O f cont rivances o f a mec h anical
nat ure as th ough d esi g ne d fo r a speci c purpose
J ane t h as specially emph asize d th e import anc e
an d prevalence o f th is kind of inference, an d, as an
illus tration O f th e co g ency O f inference base d upon
nalit y h e ur g es th at th e cert it u d e wh ich th e belief
in th e int elli g en c e of our fello w men g ives us is
.

,
.

I NDU CTIVE

2 10

LO

GIC

ba s e d upon analo gical reasonin g o f th is type ; an d


th at, moreover this belief, res t in g upon such a bas i s
is o n e of th e s tron g es t beliefs wh ic h w e po ss es s
He say s : N o w if w e ask our s elve s why w e
suppose th at o th er men th ink , w e s h all see th at it
is in virt ue o f th e principle o f nal causes In
effec t wh at is it th at experience s h ow s in th e
ac t ions of oth er men, but a cert ain number O f
ph enomena c o o rd in ated in a cert ain manner, an d
boun d no t only to g eth er, but also t o a future ph e
n o m en o n more o r less remo t e
T hu s wh en w e s ee
a man prepare his foo d by means of re, w e kno w
th at this assemblag e O f ph enomena is connect e d
with th e act O f t ak in g foo d ; wh en w e s ee a paint er
d rawin g lines o n a c anvas, w e kno w th at th e s e
apparently arbit rary ac t s are connec t e d w ith th e
execut ion O f a pic ture ; wh en w e s ee a d eaf mut e
makin g s igns which w e do n ot un ders t and w e be
l ie v e th at th ese g est ures are connect e d w ith a nal
effect, which is t o be un d ers t oo d by h im t o wh om
h e makes th em ; in ne wh en men s peak, w e s e e
th at th e articulation s O f wh ic h a p hrase is c o m
pose d are c o or d inat e d t o eac h o th er so as t o pro
duce a certain nal effect which is t o aw aken in us
a cert ain th ought an d s entiment N o w w e cannot
s ee suc h co or dinations , wh eth er ac tual o r fut ure
w ith out supposing a cert ain cause for th em ; an d
as w e kno w by int ernal e xperience th at with o u r
selves such co or dinat ions only t ake pl ace un der
th e conditio n th at th e nal effec t is prev ious ly
repre s ent e d in our con s cious ne ss w e s uppo s e th e
same th in g in th e case of oth er men ; in a w or d,
,

'

O GY

211

A N AL

w e suppose fo r th em th e consciousne ss of an end


a consciousnes s re ec t in g more o r le ss acc or din g
as th e c ircum s t ances more or le ss re s emble th ose
th at accompany in our selves th e re ec tin g con
s c io usn ess
T hu s wh en w e af rm th e int ell ig ence
o f o th er men, w e af rm a t ru th of i nd ispu t able cer
tit ud e ; an d ye t w e only af rm it o n th e g roun d of
anal o gy an d O f analo gy g ui d e d by th e princ iple of
nal cau s es 1
In th is illus t rat ion of J anet s w e h ave th e i d ea
o f a s ys t em of c O or dinat e d par t s espe c ially promi
nen t ; an d for a sati s fac t ory acc o rm t of th e rela
tions O bt ainin g in s uc h a s y st em , it w ill be s een
h o w in di s pensable it is to pos t ulat e th e th eory O f
nal caus e This m o d e o f in ference n d s a st r ikin g
illus t rat ion in th e famous dis covery of Harvey, con
cernin g th e circul at ion of th e bloo d I n th e early
part o f th e sevent een th c entu ry wh ile Harvey w as
h is pupil th e celebrat e d anat omi st F abric ius A qua
pen d ent e of P a d ua obs erve d th at many veins co n
t ain valves wh ic h lie open as lon g as th e bloo d is
owi n g to war d s th e h eart Harvey, learnin g of
thi s f ac t saw in it th e su gg es t ion of an ad apt at ion
of means t o an en d ; nam ely a cont rivance s o fas h
io n e d by nat ure as t o permit th e bloo d t o o w
al w ays in one direct ion only , an d to preve nt its
o w in an O pposit e di rect ion
O b s ervat ion of o th er
port ions of th e circulatory mech anis m le d t o a c o n
rm atio n O f th e id ea, an d t o th e d iscovery of th e
2
circulat ion o f th e bloo d
,

1
2

Janet

Fin al Caus es ,

pp

113 , 114

Go re , A rt of S c ien tific Dis cov ery ,

57 1

INDUCTIVE

212

LO

GI C

Ag ain many int subs t ances h ave been d isc o v


ere d, as th ou gh c uriou sly w rou ght w ith sh arp
e dg es an d a place as th ough d esigne d fo r a h andle,
w ith wh ic h t o w iel d th e s t one as a w eapon o r a
t ool ; it h as been inferre d from th ese g eneral ch ar
ac teristic s th at th e s t ones w ere s o cons t ruc t e d by
h uman effort an d use d by h uman bein g s fo r th e
purposes for which th ey evi d ently seem t o be
a dapt e d T h i s inference is base d upon an analogy
betw een th e peculiar s h apes o f s uch s t ones, an d
know n sh apes d esi gne d an d us e d by man
T hi s form o f analo gy h as prove d special ly sug
g e stiv e in research e s re gar d in g plant an d an imal life
S ir J oh n L ubbock g ives th e follo w in g d escrip t ion O f
th e common wh it e d ead nettle w ith th e explanat ion
O f its functions th at is evi d ently a t eleolo gical in
ference : Th e o wer consi st s o f a narro w t ube
some wh at expan de d at th e upper en d wh ere th e
lo w er lobe Of th e c orolla forms a platform o n each
si de Of wh ich is a s mall proj ect in g lobe T h e upper
portion of th e corolla is an arch e d h oo d un der
which lie four anth ers in pairs , wh ile betw een th em
an d proj ectin g s ome wh at do w n w ar d s is th e poin t e d
pist il At the lo wer en d, th e t ube cont ains h oney,
and above th e h oney is a ro w of h airs al mos t clos
in g the tube N o w why h as th e o w er th is pecul
iarform
Wh at re gulat es th e len gth o f th e tube ?
W hat is the u se of this arch ? Wh at lesson s d o
th ese lobes t each us ? Wh at a dvant a g e is th e
h oney t o th e o w er 9 Of what use is th e frin g e Of
h airs
Why d oes th e sti g ma proj ect beyon d th e
Wh y is th e corolla whit e, W hil e the re st
anth ers
,

'

213

O GY

ANAL
of

th e plant is green ?

S imilar ques tions may o f


course be a ske d w ith referenc e t o oth er o w ers
At th e close Of th e las t c ent ury Conra d S pren g el
publish e d a valuable w ork in w h ic h h e point e d o ut
th at th e forms an d colors th e scen t, h oney an d
g eneral s t ruc ture o f o w ers, h ave reference to th e
visi t s O f insec t s wh ic h are o f import ance in t rans
ferrin g th e pollen from th e st amens to th e pis til
Mr D ar win d evelope d th is th eory an d prove d ex
perim entally th at th e s pecial service wh ich insect s
perform t o o w ers consis t s not only in t ransferrin g
th e pollen from th e s t amens t o th e pis t il but in
t ran sferrin g it from th e s t am ens of o n e o w er t o
1
Th e l ine of subsequent Oh
th e pis t il o f anoth er
servation an d experiment w as th us ori g inally sug
g es t e d by th e st ruc t ural appearance of th e s e o w ers
wh ich seeme d forme d fo r some spe c ic en d T h e
que s tions once s t art e d,
TO wh at en d 9
To wh at
purpose ? Fo r wh at u s e 9 le d to th e th eory o f
S pren g el an d th e corroborative experiment s of
D ar w in
T h is is furth er illus t rat e d in some v ery int er
estin g o w er st ruc ture s, also d escribe d by S ir
Joh n L ubbock, wh ich in dicat e pecul iar cont rivances
for th e des truc t ion O f in sect s T h e peculiarity o f
format ion rs t s u gg es t e d some such en d as th is ,
wh ic h h as since been prove d by careful Observation
t o be th e c as e T h e rst observation o n insec t
eat in g o w ers w as mad e about th e y ear 186 8 by
He Observe d th at in D ion aea, a N orth
E llis
A merican plant th e leaves h ave a joint in th e
.

Lu

bb oc k

S cien tic Lectu res ,

pp

1, 2

INDU CTIVE

214

LO

GI C

m iddle, an d th us close over, kill, and ac tually di


g e st any insect wh ich may alight o n th em A n
o th er cas e is th at O f Ut ricularia an aquatic species
which bears a number O f ut ricles o r sacs which
Branch es,
h ave been suppose d t o ac t as oat s
h o w ever, whic h bear no bla dd ers oat jus t as w ell
as th e o th er s an d th ere seems n o d oub t th at th eir
real use is t o cap ture small aquat ic animals , which
th ey d o in consi d erable numbers T h e bla dd ers in
fact are on th e princ iple of an eel t rap h aving an
ent ranc e close d w ith a ap, wh ich permit s an easy
entrance but effect ually prevent s th e unfortunat e
victim from g ett in g o ut a g ain I n th e g enus , S ar
rac enia some o f th e leaves are in th e form o f a
pit ch er Th ey secret e a ui d an d are line d int er
nally w ith h airs pointin g do wnwar d s U p th e o ut
si de o f th e pit ch er th ere is a line o f h oney glan d s
wh ic h lure th e insect s t o th eir d es t ruction F lies
an d oth er insect s wh ic h fall int o this pit c h er cann ot
g et o ut ag ain an d are act ually dig es t e d by th e plant 1
In th e example wh ere th e i dea O f an eelt rap sug
g e st e d th e possible funct ion Of th e similar s t ruct
ure in th e plant U tric ularia w e n d o ne of th e
mos t st rikin g illus trat ion s o f this mo d e o f ana
It w as an easy an d nat ural
lo gical inference
t ransition from s imilarity o f s t ruc ture t o si milarity
To g ive an i dea O f th e great number
o f func t ion
o f t eleolo g ical p h enomena in th e ve g e t able an d
animal w orl d, an d th e w eal th Of possible su gges
t ion s t ore d a way in th ese various st ructure s an d
disclose d by a sag acious analysis, I quo t e a remark
.

bb oc k

Lu

S cientic Lec tures ,

pp

4, 5

2 15

ANALO GY

John Lubbock s in commentin g upon th e


variation O f color an d marking s of cat erpillars : I
s h oul d pro duce an impression very different from
th at wh ic h I w is h t o convey, w ere I t o lea d you t o
suppose th at al l th ese varieties h ave been explaine d
o r are un d ers t oo d
F ar from i t ; th ey s t ill O er a
lar g e el d for s t udy ; neverth eless, I vent ure t o
th ink th e evi d ence n o w brought for war d, h o w ever
imperfec tly is at leas t sufc ient to jus tify th e con
elusion th at th ere is not a h air o r a line not a spo t
or a c olor fo r wh ic h th ere is not a reason wh ic h
h as no t a purpose o r a meanin g in th e e c onomy o f
nat ure 1
A n illus t rat ion g iven by D ar win s h ow s th is mo d e
of inference applie d t o th e s p h ere of animal l ife
also He says : Th e g reat size of th e bone s O f
th e me gath erioi d animals w as a complet e puzzle t o
n at ural is t s unt il P rofessor O w en lat ely s olve d th e
problem w ith remarkable in g enuity
T h e t eeth
in d icat e by th eir s imple s t ruct ure th at th ese me g a
th erio id animals li v e d on ve g e t able foo d, an d prob
ably o n th e leaves an d small twi g s of t ree s ; th eir
pon d erous forms and g reat s t ron g, curve d claw s
s eem s o little a d ap ted for locomo t ion th at some em i
nent nat uralis t s h ave ac t ually believe d th at like
th e slo th s, to wh ich th ey are int imat ely relat e d
th ey subsis t e d by climbin g back d ownw ard s o n
trees an d fee d in g on th e leaves It w as a bol d
not to say prepos t erous , i d ea to conceive even ant e
dil uvian t rees w ith branc h e s s t ron g enough t o bear
ani mals as larg e as eleph ant s P rofessor O wen,
of

Sir

Lu

bb ock

S cientic Lec tu res ,

pp

66 , 6 7

216

IN DUC T IVE

O G IC

w ith far more probabili ty believes th at inst ead o f


climbin g o n th e t rees th ey pulle d th e branche s
d o wn t o th em , and t ore up th e smaller ones by
th e root s , and so fe d o n th e leaves T h e colossal
breadth an d w ei ght of th eir h in der quart ers, wh ich
can h ardly be ima gi ned with out h avin g been seen
become on this vie w O f obvious serv ice ins t ead o f
bein g an encumbrance : th eir apparent clum s iness
disappears With th eir g reat t ail s an d th eir hug e
h eel s rm ly xe d l ike a t ripo d on th e g round th ey
coul d freely exert th e full force Of th eir mos t
pow erful arm s an d g reat cla ws St ron gly root e d
ind ee d mus t h ave been th at tree wh ic h coul d h ave
resis t e d s uc h force " T h e Mylo d on moreover was
furnish e d with a lon g ext ensile t ong ue like th at of th e
giraffe wh ich by one O f those beaut iful provis ions
of nat ure thus reach es w ith th e aid of its lon g neck,
1
i t s leafy foo d
T hrou gh out w e Observe analo g ical
inference ba s e d upon th ese t eleolo gical marks, an d
furnishin g a basis for a sat isfac t ory hypoth esis
We see wh at a w i d e el d thu s O pens in th e
re gion O f biolo gy alone for th e di s covery Of resem
blances leading t o th e apprec iation of th e ful ler
t eleolo g ic al si gnic ance of plant an d animal life
In th e illus t rations given, bo th of th e teleolo gi
cal an d o th er form s O f analo gy w e no t ice th at it s
ch ief lo g ical funct ion is th at Of su gg es tion Of s ome
h ypoth esis wh ich may or may n o t be aft erw ar d s
conrme d by subsequent experiment S ome of th e
mo st import ant dis coveries of science h ave arisen
from analo gical s ugg es tions S ir J oh n Hersch el
,

Darwin , Voy ag e of

N aturalis t,

pp

106 , 107

217

O GY

AN AL

w as le d by observe d analo g ies t o pre dic t cert ain


ph enomena aft erw ar d s veri e d experiment ally by
F arad ay Hersch el h a d not ice d th at a screw like
form kno w n as h elicoidal dissymmet ry was o b
serve d in th ree cases namely in elec t rical h elices,
pla g ih e d ral quartz cryst al s (th at is, crys t als h avin g
an O blique spiral arran g ement of planes) an d th e
ro t at ion of th e plane of polarizat ion o f li ght A s
Hersch el h imself s ai d
I reasone d thus : Here
are th ree ph enomena a g reein g in a v ery strang e
P robably this peculiarity is a connec t
p ecu liarity
ing link, p h ysically speakin g amon g th em
N ow
in th e case o f th e crys t al s an d th e li ght th is prob
abilit y h as been t urne d int o cert ainty by my o wn
experiment s T h erefore, in duction le d me to con
an d mus t
c l ud e th at a similar connec t ion exis t s
t urn up, someh o w o r o th er, be tw een th e elect ric
current an d polarize d l ight, an d th at th e plane Of
polarization woul d be d e ect e d by mag net o elec

Hersch el th us ant icipat e d F ara day s ex


tric ity
perime ntal d iscovery of th e in uence of m a gne t ic
s t rain upon polarize d li ght 1
A noth er import ant discovery th e g erm th eory
o f epi d emic d isease w as rs t su gg es t e d by an
analo gy I n th e th eory , as expre s se d by K irch er,
an d favore d by L inn aeus, an d aft erward s support e d
by S ir Henry Holland i t s special s trength accor d
in g t o P rofessor Tyn d all , consis t e d in th e perfect
parallelism O f th e ph enomena O f cont ag ious d isease
w ith th ose o f life A s a plant e d acorn g ives birth
to an o ak compet en t to pro du ce a wh ole crop o f
-

Jevo ns

P ri n cip les

f S cien ce

630

218

IND U C T IVE

LO

GIC

ac orns each g ift e d with th e pow er of repro ducin g th e


parent t ree, and as thus from a s in gle s ee dlin g a
wh ole fores t may sprin g, so it is con t en de d th e s e
epi d emic diseases lit erally plant th eir see d s, g ro w
an d sh ake abroad ne w g erms wh ich, mee tin g in th e
h uman bo dy th eir proper foo d an d t emperat ure,
1
nally t ake possession O f wh ole populat ions
T h e th eory of evolution w as rst su gg es t e d to
Mr D arwin by th e analo g ous ph enomena Observe d
in articial selec tion an d bree din g T h e t rans ition
t o natural select ion w as eas ily ma de especially as
o n rea din g Mal th us , 0 n P op u latio n h e conceive d
th e id ea o f a s t ruggle fo r exis t enc e as th e inevit able
re s ul t O f th e rapid increas e o f org anic bein gs T h is
id ea ne c essi t at e d th e nat ural sele c t ion wh ic h h e
nee d e d t o account for re s ul t s s imilar to th e arti
c ial selec t ion an d th us h is th eory g re w o u t of an
analo gy as its be g innin g Moreover in th e d evel
o pm ent o f th e th eory in i t s m anifol d d e t ail s o th er
analo g ies prove d al s o sugg est ive
F or ins t ance
th ere is th e suppose d analo gy be tw een th e g ro wth
O f a specie s an d th e g ro wth of an ind ividual
It supposes, for example as P rofessor Clifford has
put it th at th e rac e of c rabs h as g one th rou gh
much th e s ame s ort of c h an g es as every crab g oe s
th rou gh no w in th e course Of its formation in th e
e gg, ch ang e s represent e d by it s pris t ine s h ape
utt erly unlike wh at it aft er w ar d s att ains, an d by
its g ra d ual met amorp h os is an d format ion o f s h ell
and claw s 2
,

1
2

p 287
p 86

Tyn d all , Frag ments of S cience,


liffo rd , Le ctures an d E ss ays ,

219

ANALO GY

T h e g erm th eory O f pu t refact ion rs t su gg es t e d


by S chw ann receive d conrmat ion th rough cert ain
resemblances no t e d by P rofessor L is t er be tw een
ferment at ion an d put refaction I n h is Intro d u c
t ory L ec t ure before th e University o f E dinbur gh,
P rofessor L is t er calle d att ent ion to th e fact th at
ferment ation an d put refact ion present a very
s t rikin g parallel I n eac h a s t able compoun d
sug ar in o n e case albiun en in th e oth er und er
g oes ext raor dinary c h emical ch ang e s un der th e
in uence of an excessi v ely minut e quant ity O f a
subs t ance wh ic h, re g ar d e d ch emic ally w oul d be
consi d ere d inert It w as point e d out also by P ro
fesso r L is t er in th is connec t ion th at as w as w ell
kno wn, one of th e c h ief peculiari t ies of livin g o r
g an ism s is th at th ey possess ext raor d inary po w ers
O f eecting c h emical c h an g es in mat erial s in th eir
vicinity out of all proportio n to th eir energy as
mere c h emical compoun d s S uch bein g th e fac t s
in th e c as e an d, moreover th e fer ment at ion O f
su g ar being g enerally allo we d to be occa sione d by

th e pre s ence of livin g or g anisms , P rofessor L is t er s


inference w as th at put refac t ion w as d ue t o an
l
analo g ous a g ency
A d iscovery in quit e a d ifferent sph ere, th at of
math emat ics, l ead in g t o th e branc h o f analy tical
g eomet ry , w as rs t sugg es t e d t o D escart es th rou gh
Observin g th e resemblanc es exis tin g be tween g eom
I n a sim ilar manner Boole w as
etry an d al g ebra
le d by th e resemblances not e d be tw een al g ebra
an d lo g ic t o g ive express ion t o th e same in a sys
-

T nd all , Frag men ts

S cience,

pp

300302

220

IN DU CTIVE

LO

G IC

t em wh ich h e calle d th e la ws Of th ought an d wh ic h


h as become th e basi s of a g eneral or symbolic lo gic
Wh ile th ere are th us unques tionable evid ences o f
th e value o f analo gy as a form o f inferen ce, th ere
are also ca s es O f false analo gy unfort unat ely so nu
me ro u s as t o d iscre dit th e process wh olly in some
quarters It will be w ell th erefore t o in dicat e
some o f th e requirement s o f t rue analo gy :
1 I n th e rs t place th e resemblance mus t be a
prepon deratin g one ; th at is th e ph enom ena c o m
pare d must sh o w a more s t rik in g ag reement th an
difference S o m e w rit ers h ave balan c e d ag reement
ag ain s t difference upon a purely numerical bas is of
comparis on formin g wh at may be calle d an analo gi
cal ratio, w ith point s o f s imilarity formin g th e numer
at or an d th e po int s bo th O f similarity an d difference,
plus th e unkno wn th at is th e t o t al number form
in g th e d enom inat or S uch a represent at ion Of th e
force o f an analo gy is given by Mill Bain an d
oth ers I think h o w ever th a t this represent ation
is ap t t o be mislea din g in pro ducin g th e impress ion
th at th e mere number of point s of a greement irre
spec tiv e O f th eir s i g ni cance is th e c h ief feat ure o f
analo gy Wh ereas it is th e w ei ght of th e agreein g
att ribut es an d not th e number, th at count s A s h as
been before sai d in analo gy w e w ei gh inst ances ,
an d d o not count th em T h e analo gical rat io ex
presse d numerically as above is really equivalent
to th e rat io of probability wh ic h w ill be d escribe d in
th e follo w in g ch apt er I h ave th erefore c h an g e d
th e usual w ordin g Of th i s requirement so th a t it
read s, th e resemblances m us t be more st rikin g th an
,

'

221

ANALO G Y

th e d ierenc e s Th is prov i d es for cases wh en per


h aps a fe w point s of resemblance will be of such a
nat ure as t o outw ei gh many point s o f d ifference in
th e t ot al es timat e
T h is requirement also exclu d es all fanciful anal
o g ies an d all resemblanc es res t in g upon a gurat ive
rath er th an a real basis F or inst ance th e ad v o
cat es O f annual P arliament s in th e t ime o f th e
Common w ealth ur g e d th eir c ase o n th e analo g ical
g roun d th at a bo dy pol it ic is similar t o a l ivin g bo dy
an d th at serpen t s annually cas t th eir sk in wh ich
bein g no d oubt for a bene cial purpose , m ight w ell
be imit at e d
2 I n no t in g th e point s O f resemblance betw een
two ph enomena, all c ircums t ances wh ic h are kno w n
to be eects of o n e cau s e mu st th erefore be re
g ard e d no t as m any but as on e Fo r ins t ance, two
c h emical oxid e s may be compare d ; th e effec t s com
m o n to each may be d ue t o th e presence O f th e
oxy g en wh ic h eac h con t ains an d th erefore m ust no t
be re g ar d e d in th e li ght o f in d epen d ent marks Of
similarity
3 I f w e infer by analo g y th at a s ubs t ance po s
sesses a cert ain property wh ich w e kno w is inc o m
pat ible w ith some o ne or o th er kno wn properties
of th e s ubs t ance th e analo gy is at once d iscre d it e d
We may infer th at th e moon is inh abit e d by virtue
O f th e many poin t s of re s emblance betw een th e
moon an d th e earth Ho w ever th e fac t th at th e
moon h as no atmosph ere necessary t o sus t ain life
at once m akes such an ar g ument base d upon ah
alo g y wh olly out o f th e ques t ion
.

222

INDU CTIVE

LO

GIC

Th ere are cert ain spec ial requirement s refer


ring to th at part icular form O f analo gy wh ich is
base d upon t eleolo g ical consi d erat ions T h ey are
as follo w s
a T h is principle mus t never be use d as an ar g u
ment agains t an observe d fac t or an es t ablis h e d
law of nat ure While th is precaution is not n ec e s
s ary at th e present t ime in scient i c circles at
leas t, s till th ere w as a tim e wh en its counsel was
sorely n ee d e d Wh en in ast ronomy it w as prove d
th at th ere w ere suns g ravitat ing aroun d oth er suns
w ith o ut o ur s olar sys t em th is was obj ect e d to upon
th e follo win g g roun d, as given by o n e N ic h olas
F u ss a celebrat e d ast ronomer at th e en d o f th e
ei ght eenth c ent ury : Wh at is th e g oo d of some
luminous bo d ies revolvin g roun d o th ers 9 Th e s un
is th e only s ource wh ence th e planets d erive li ght
an d h eat Were th eir entire sys t em s of s uns co n
t rolle d by oth er suns , th eir nei ghborh oo d and th eir
mo tions w oul d be O bj ectless th eir ray s useless T h e
s un s h ave no nee d t o borro w from s t rang e bo dies
wh at th ey th emselves h ave rec eive d as th eir o wn
I f th e s econ d ary st ars are luminou s bo d ie s, what is
the e nd O f th e ir mo t ive s
T h ere is moreover anoth er abuse of th e princ iple
Of nal causes wh ich h as al s o h is t oric int eres t
rath er th an any pre s ent pertinence ; n amely oppos
ing cert ain false t eleolo gical i deas t o es t ablis h e d
discoverie s o r invent ion s w ith a mist aken zeal in
d efence Of a Divine Providence Fo r in st ance at
th e t ime of J enner s g reat disco v ery an Eng lis h
physician, D r Ro wley, sai d of small pox : I t is a
4

223

AN ALO GY

mal a dy impose d by th e d ecree O f h eaven, an d

v ac

c in atio n

is an au d acious an d sacrile g ious violat ion


O f our h oly reli g ion Th e d esi gn s O f th ese vaccina
t ors appear t o d efy h eaven it self, and th e very w ill

o f Go d
T h e int ro duc t ion of w inno w in g mach ines
int o S co tland met w ith bitt er O pposi t ion on th e
g roun d th at th e w in ds w ere th e w ork O f G o d, an d
th at th e win d th us articial ly raise d w as a verit able

d evil s wind as th ey w ere won t to c all it S ir


Wal t er S c ott in Old M o rtality, h as th e ol d Manse
say t o h er mi s t res s : Your ladysh ip an d th e st e w
ar d are wis h in g Cuddie to use a ne w mach ine t o
w inno w th e corn T h is mach ine Oppo s es th e de
sign s O f P rovi d ence by furnis h in g w in d fo r your
spec ial use an d by human means in place of askin g
it by pray er an d waitin g with pat ienc e t ill P rov

i d ence i t self sen d s it


b F inal c auses s h o ul d never be employe d to
explain ph enomena which d o no t exis t A s M
F lorens h as s aid : We mus t procee d n o t from n al
causes t o fac t s, but from fac t s t o n al cau s es ; th at
is, w e sh oul d no t s uperim pose nal cause s upon
ph enomena
We mus t see th em in ph enomena
th emselves , an d w e mus t not arbit rarily proj ec t a
t eleolo g ical i d ea, purely subj ect ive upon an Obj e c
t ive g roun d T hus in anc ient t imes Hippocrat e s
is sai d to h ave a dmire d th e skill w ith wh ich th e
auricles O f th e h eart h ave been mad e to blo w the
.

in to t he heart

air

W e mus t dis tin g uis h acci d ent al from essen tial


marks O f nality an d not be le d int o fanciful o r
V ol t aire h as expresse d such
far fe t ch e d analo gies
c

22 4

INDU CTIVE

LO

GI C

d efect wh en in satire h e mad e that famous re


mark N oses are mad e in ord er t o bear spec t acles
Bernar din d e S aint P ierre says : D o g s are usu
ally O f tw o O ppo s it e colors th e one li ght th e o th er
d ark in or d er th at wh enever th ey may be in th e
h ous e th ey may be d is ting uis h e d from th e fu rn i
ture wi th th e color of wh ic h th ey m ight be con
foun d e d
Wh erever eas are th ey j ump o n
wh it e colors T h is ins tinct h as been given th em ,

that we may th e more easily cat c h th em


An d
a gain th e same w rit er say s : T h e melon h as been
divid e d int o sec tions by nature, fo r family eatin g 1
All s uch gro t esque inferenc es will g ive an id ea o f
h ow read ily th e ima ginat ion w ill run riot if allo w e d
to remain un c urbe d by th e reason
5 Analo gy sh oul d never be re g ar d e d as h avin g
more w e ight th an th at of ext remely h i gh probabil
i ty, even in cas e s seemin gly mos t conclusive It s
true funct ion is sugg es tive leadin g t o hypo th esis
an d experiment an d it nee d s th i s supplement ary
proof It was an inference ba s e d on analo gy for
ins tance wh ich sugg es t e d th e probability th at th e
Bin omial L aw h avin g prove d t o be vali d as re g ar d s
th e sec on d th ir d an d fourth pow ers , mi ght also be
ext end e d t o th e fth, and s o on t o th e oth er po w ers
in d e ni t ely T h is su gg est ion Offere d n o real basis ,
h o w ever, upon wh ich th e Binomial T h eorem coul d
re st ; i t nee d e d math ematic al d emons trat ion t o con
rm an d g eneralize its expres sion in th e spec ial
case s already experiment ally te s t e d s o as t o c over
a

pon t h e ab use f nal ause I h ave tak en


Chapter V III o f A p end ix
from Janet s ad mi abl e chapte
p
1

Th e illu st ratio ns u

r,

225

O GY

A NAL

all possible exponent s , both positive an d ne g at ive


fract ional an d int e gral
S O also th e d iscovery O f th e circulat ion o f th e
bloo d was rs t su gg es t e d to Harvey, as h as been
sai d, by analo g ical consi d erat ions upon observe d
t eleolo g ical p h enomena Harvey h o w ever was not
cont ent w ith this sugg es t ion merely He w as le d
to experimen t upon th e veins an d art eries
h e tie d
an art ery an d vein an d carefully O bserve d th e me
c h anic al e ffect s upon th e tw o si d es o f th e t ie d par t s
E xperiment s O f th is nat ure w i th c lo s e O bservation
an d st udy, w ere kept up m ost d ili g en tly an d w ith
rare perseverance fo r nineteen years before h e h ad
t race d th e ent ire c ourse of th e bloo d th rough all
part s O f th e human bo dy an d, in a manner wh olly
sat isfac t ory to h imself, veri e d th e rst s t at ement
O f thi s th eory
,

CH A PT ER " V
P R O BA BI L I T Y

T HE RE are cert ain ph enomena o f suc h a nat ure


th at th eir ant ec e d ent s bein g ex t remely complex,
cannot be a dequat ely c ompreh end e d by observat ion
h o w ever s earc h in g it may be ; nor can th ey be s ub
j e cted t o any analy s is th at w ill d is c lose th e c aus al
element s t o which th e effect in que st ion is due
Moreover, with s eem ingly th e same ant ece d ent s
th e event sometimes h appens an d sometime s does
not ; an d even with ant ec ed ents a s so c iat e d w ith an
event as cause an d effec t re s pec tively, neverth eless
th e event d oe s no t occur as w e sh oul d naturally
expect wh ile with ant ec e dents assoc iat e d with th e
contra dic t ion O f th e event as caus e an d effec t re
spe ctiv ely w e n d th e occur ren c e of th e even t qui t e
cont rary t o wh at w e sh oul d naturally expec t T h e
evidenc e of a con st ant c onnec tion betw een antece
d ent an d con s equen t th at w e h ave foun d in so many
cas e s which w e h ave exam ine d is h ere wh olly lack
in g Re gularity h as been replac e d by irre gul arit y
respe ctin g s uch ph enomena Fo r inst ance I thro w
dic e repeat e dly ; th e ant ece d ent s h ak ing o f th e
box, an d t os sing th e dic e upon th e t able, is about
th e s ame eac h time, at lea st th e dierenc e c an
,

2 26

I I Y

PR BAB L T

d et ermine d, an d ye t th e resul t s vary w ith


eac h succes s ive t h ro w
T h e causal d et ermination
in each cas e is so complex as to be beyon d com
put at ion ; th e init ial posi t ion O f th e d ice, th e force
o f th eir ej ec t ion from th e bo x th e h ei ght O f th e
hex above th e t able wh en th ey leave it, th e ine
qualit ie s Of th e t able it self a variation be tween
th e ph y s ical an d g eomet rical cen t res Of g ravi ty
o f th e d ice
et c , all th ese m ake th e antece d ent
so c omplex th at a sli ght vari at ion in any o n e o f
th ese con d it ions w ill aect th e result We n d,
th erefore, d ouble sixes at o n e t ime, a th ree and
four at ano th er an d so o n in d e nit ely
O r again, i t s ometimes h appens th at w ith perfect
s anit ary con dit ions a cont ag ious disea s e will appear
th at h as al w ays been re g ar d e d an d th at correc tly as
d ue t o imperfec t sanit ation o r an en tire dis re g ar d
o f sanit ary r equirement s an d Of all th e law s of
h eal th may ye t g ive ri s e t o no disea se O f special
moment
Cert ain c on d i tions O f t emperat ure at
m o s ph eric pressure veloci ty an d dire c t ion o f th e
w ind, may o ne d ay bring s t orm an d rain, an d as
far as O bservat ion c an d e t ect, similar con d it ions
may a g ain bring fair w eath er S o , also, th e rise
an d fall in st ock an d money market s is ext remely
sus cep t ible t o th e varyin g con dit ions o f in d e
po w ers
n itely complex force s wh olly beyon d all
of d e t erminat ion Or of pre d i c t ion
S uch ph e
n o m e na present a problem wh i c h th e meth o d s o f
in duc t ive inquiry cannot d eal w ith O bservat ion
is no t far reach in g enou gh t o provi d e th e d at a for
th e solut ion o f th e problem, an d, even if
not

be

'

228

IND U C T IVE

LO

G IC

meth o d s of c omput at ion an d d etermin ation


are no t suf c ien t ly a d equat e t o solve problems o f
so many t erms an d O f s o complex a nature
T h e experimen t al meth o ds are d esi gne d t o tes t
causes su gg es t e d by analo gy or a ment al anal ysis ;
bu t in such ph enomena as th ese, th e problem is
not simply t o nd a caus al connect ion Th e causal
connection m ay be e s t ablish e d beyond all reas ons
ble doubt, an d yet th e c ause Ob tains in th e mid s t
o f so complex a s e ttin g th at th e problem is really
this, t o d e t ermine wh eth er a cause, wh ose exac t
nat ure may be know n o r unkno wn, w ill prove
O perat ive or inoperat ive T h e cause may be al
w ays presen t an d even it s exac t nature may be
kno wn, an d yet th e complex circum s t ances at
t en din g it may be of s uch a ch arac t er th at one
alone, o r tw o or more c ombinin g may neut ralize
th e operation o f th e cause an d o n th e o th er h and
a slight variat ion O f th e c ombine d circums t ances
may promo t e an d even ac c elerat e th e operat ion O f
th e cause in que s tion T h e problem th en is to
d et ermine h o w oft en th e event h appens, an d h o w
O ft en it fail s of h appenin g th e complex an d in
d et erminat e ant ec e d ent bein g pre s ent in al l th e
ins t ances examin e d
Wh en w e be gin t o coun t ins t ances, w e are re
mind e d th at w e mus t be in th e near nei gh borh oo d
of th e sph ere o f enumerat ive in duc t ion Enumera
t ive in duction it will be remembere d, t reat s ih
s t ances by not in g th e num ber of Observe d coinci d ent
h appenin g s o f th e ant ec e d ent an d c ons equent un d er
invest igat ion, n o att empt being mad e t o analyze
our

'

I I

229

PR BAB L TY

th eir respective cont ent s

t o d et ermine a c ausal
connec tion more d eni t ely by means of any one o r
more o f th e in d uc t ive meth o d s O f re s earc h and veri
c atio n
Th e resul t O f suc h an inves t i g at ion may
be formulat e d in a proposit ion o f th e form E very
A is B
T h is , s t rictly int erpr et e d h as th e force of,
Every A th at h as been O b s erve d is B
T h e enu
m eratio n of th e kin d of in s t ance s wh ich w e are d is
cu ss in g ih this c h apt er, h o w ever, differ s from this
in th at th e observat ion l ead s t o a tw ofol d resul t,
a s et Of ins tances in wh ic h i t is observe d th at
th e A s are B s al s o, anoth er set, h o w ever in
wh ich th e A s are no t B s
T h ese ins tances are
of s uch a natur e th at th e Observe d A is an ant e
ce d ent so ext remely complex th at th e element
th in it, wh ic h is a c ause capable Of pro ducin g
B, may eith er be absent w i th out pro d ucin g an
appre c iable c h a ng e in th e g eneral nat ure o f A or
bein g presen t may be n eut ralize d by some oth er
element O f A its elf T h e res ult g ives a basis for a
probable inference only; an d th e nature Of th at in
ference will d epen d upon th e prepon d erance of th e
obs erve d h appenin g, o r o f th e failure of th e event
un der inves t i g at ion
T h e probability att ach e d t o suc h an inference,
h o w ever is different from th e probability which
c h aract eri zes th e nat ure of enumerat ive in d uction
I n th e latt er wh en th e observat ion h as been w id ely
ext en d e d an d no except ions n ot e d it is u sual t o s ay
th e resul t expre ss e d in th e proposit ion Every A is
B h as th e force o f a h i gh d e g ree O f probabili ty
I n th e ins tances, h o w ever, whose invest i gat ion
,

or

INDU CTIVE

23 0

LO

GI C

s h o w s th e resul t th at some A s are Bs an d some


not an d yet wh ere th e former for ins t ance far
outnum ber th e latt er cases , th en it may be inferre d
th at th e A s wh ich in future w e may meet w ith
w ill probably be B s ; an d th e d e g ree Of probability
expresse d in such a proposit ion is c ommensurat e
with th e prepond erance of th e num ber o f O bserve d
afrmative inst ances over th e ne g ative Here th e
probability refer s t o th e vali d ity o f an inference
concernin g cert ain p articu lar ins t ances, be th ey
many o r be th ey fe w wh ich lie beyond th e sph ere
o f our presen t kno w le dg e ; in enumerat ive in d u c
t ion th e probabil ity is at tach e d t o th e u n iv ersal ity
o f th e proposi t ion af rme d as a res ul t O f observ a
t ion th at h as not so far d et ect e d an except ion In
th e former case th e ques t ion Of the universality
of th e resul t is conclusively ans w ere d an d th at in
th e ne gative ; th ere can be no univers al proposit ion
possible as some ins t ances give A an d B t o g eth er,
o th er s g ive A with th e abse nce Of B ; an d th e
ques t ion O f probability th at h ere arises , th erefore,
refers to in divi dual cases not ye t exam ine d as t o
wh eth er th ey severally will more likely correspon d
t o th e set of af rm at ive, or to th at o f th e ne g ative
ins tances already no t e d
Th e comparison of th e number O f h appening s
w ith th at of th e failures O f an event affor d s a
basis for th ree k in d s o f inferenc e, all O f th em in
th e sph ere Of probability
1 We n d in su c h a c ompari s on a basis for th e
c alculat ion of th e probabili t y of a par t icular even t
h appening, in case th ere is a repet ition O f th e cir

231

PROBABILITY
cumstan ces

wh ich , in form er cases, h ave sometimes


pro duce d th e event, an d s ometimes h ave faile d t o
pro duce i t I f, accor di n g t o former O bservation,
th e event h as h appene d , le t us say , seven t imes ,
an d faile d thr ee, th e probability, expresse d nu
merically, of it s h appenin g a g ain is 316
T h e rul e
is, to express th e probabil ity Of an event, take as
numerat or th e number o f times wh ich th e event
h as been observe d to occ ur, an d as d enominat or
th e to t al nu mber observe d, both o f h appenin g an d
failure ; th e fr ac t ion th us expresse d w ill represent
th e probability of th e event h appenin g T h e coun
t er probability may be represent e d by th e number
o f observe d failures of th e event di vi d e d by th e
t o tal number of cases Observe d T h e count er pro b
ability, plus th e probability, evid ently is equal t o
unity I f, th erefore, th e probability is unity, th e
count er probabili ty w ill equal z ero ; th at is, th e
probability in th at case h as mer g e d int o certainty
Zero, th erefore , represen t s absolut e impossibili ty
A ll frac tions be tw een th e l imit s zero an d o ne rep
resent vary in g d e grees Of probability from impo s
sibility at o ne extreme to cer t aint y at th e o th er
N o t only may th ere be th is in d uct ive basis for
th e c al cul at ion o f probabil i ty, aris in g from ac tually
obs erve d in s t ances ; th ere may be also a d e d uc tive
cal culation o f probabil ity base d upon th e kno wn
s t ruc ture o r natur e o f th e p h enomena th emselves
in ad vance O f any O bservat i o n as t o th eir ac tu al
beh avior F or ins t ance, w e s ay th e probability
Kno w in g th e
o f a penny t urnin g up h ea d s is 3
form of th e penny an d th at th ere are but two
.

232

INDU CTIVE

LO

GI C

possibili ties , h ea d s or t ails an d th ere bein g no


reason why one s h oul d more likely t urn up th an
th e oth er w e say th ere is o ne ch ance favorable t o
h ead s as over a g ains t th e tw o ch ances wh ich rep
resent th e t ot al number o f possibilitie s un d er th e
existin g c ircums t ances With a d ie, in th e form
of a perfect cube, w e say th ere is one c h ance of
it s turnin g up th e face m arke d 1 as over a g ains t
th e six ch ances represent e d by th e six faces , th e
tot al number ; h ere th e probability is
T h us th e
basi s fo r th e calculat ion O f probability may be a
th eoret ic al as w ell as an empirical o ne
In th e e stimat e O f th e probability of an e v ent in
th e ac t ual con duct O f affairs w e sel d omexpre s s
th at probability num erically I woul d say th at w e
express a d e gree of probability a dverbially rath er
th an numeri c ally ; th at is w e say an event is quite
probable or it is v ery probable o r i t is extremely
probable T h e fact is th at, as re g ar d s mos t p h enom
ena, w e d o not keep an exact o r even approximat e
memoran dum O f th e number Of h appenin g s compare d
with th at O f th e failures We rath er clas sify our
observat ions in t erms o f more or less F or ins t ance,
cert ain c ircums t ance s w e Obs erve pro duce about as
many failures as h appenin g s O f an event ; o th er
circums t ance s pro duce far more h appenin g s th an
failures ; oth ers far les s , an d so on Consequent ly
w e rec ei ve cert ain psych olo g ical impressions of
varying d e g rees O f int ensity accor d in g t o th e pre
pond eranc e of h appenin g over failure or v ice v ers a
th is impression be c omes th e ba sis for es t imatin g
th e probabili ty in ques t ion, an d th e d eg ree o f th at
,

'

233

OBABILITY

PR

is commensu rat e w ith th e int ensity O f


th e ori g inal psych olo g ical impression ari s in g from
concep t s Of more or o f less I n such a sph ere, h o w
ever as th at d evo t e d t o th e int erest s of bettin g,
g amblin g pool sellin g, book makin g et c , probabili
t ies are est imat e d accor din g t o Observat ions an d
th eore t ical consi d erat ions , wh ose condi tions are ex
pres s e d numerically ; an d th e amount riske d in
each case is s t ric tly est imat e d ac c ord in g t o th e
exact ratio O f probability t o count er probability
un d er th e exis t in g circums t ances
T h e es t imat ion o f probability in t erms of a
g reat er o r less d e g ree is , h o w ever, m ore usual , and
appl icable to th e con duc t O f h uman l ife g enerally
I t h as special force an d ut ilit y as a mo d e of infer
ence, wh en th e O bserve d inst ances so far outnumber
t h e excep t ions as t o creat e an im pre s sion of suc h
a h i gh d eg ree o f probability as t o approximat e
prac t ical if no t th eoretical cer tainty Fo r ins t ance,
it h as been no t e d over a w i de el d of O bservat ion
th at a secon d att ack o f s c arle t fever is extremely
rare Excep t ions h ave occurre d and th erefore, by
enumerat ive in duc t ion it is impossible t o g eneral
ize th e universal propos it ion th at a secon d att ack
will never occur It is, h o w ever possible t o a s s ert
w ith some wh at positive assurance th at i t is h ighly
probable th at a person w ill be exemp t from a secon d
att ack
O r you h ear th at a person, wh ose name is nu
kno wn t o you h as me t w i th an acci d en t in th e
cit y Of N e w York resul t in g fat ally You are no t
alarme d, and perh aps th e possibility d oes not even

probabili t y

'

IND UCTIVE

2 34

LO

GIC

su gg est it self to you th at th e unkno wn person may


pro v e t o be a member O f your o wn family , or a
frien d wh o at th e time is kno w n t o be in N e w
York T h e probabil ity a g ains t su ch a su gg es t ion
is so lar g e as t o preclu d e even th e th ou ght Of it
S uppo s e, h o w ever, th e acci d en t occurre d at one of
th e suburban s t at ion s Your kno wle dg e th at your
frien d rid es on one Of th e suburban t ra in s each d ay
to an d from t o wn, may be th e groun d of some
anx iety bec ause in th i s ca s e th e ran g e O f possibil i
t ie s is mat erially narro we d S uppose, moreover,
t hat th e s t ation wh ere th e acci d en t o c curre d is at
th e Yillag e wh ere your frien d res id es your anxiety
r e c e ive s an a dd itional increment ; an d
a g ain, sup
po s e it is a t th e h our at wh ic h your frien d ordi
narily reach e s th is st ation th ere is th en increase d
appreh ension on his ac count Thu s, as furth er
kno wle dg e l im it s th e num ber O f t ot al pos s ible case s
th e d enominat or of th e probabil ity fraction is c ont in
u all
y d e c rea sin g an d th erefore th e probabili t y it s el f
c on tinually increas es un t il it h as d evelope d fro m a
frac tion Of insi gnic ant proport ion s t o one wh ich is
s u gg es t ive o f g reat anxie ty an d s uspense
2 T h e c om par is o n of failure an d h appenin g of
event s ba s e d upon Observation or th eoret ical co n
s id erat ions of s t ruct ure an d nat ure
lea d s also to
inferences c oncernin g lar g e numbers Of ins t ances
consid ere d t o g eth er I f a memoran dum is kep t of
th e number of t imes an even t h as h appene d an d
th e number of t imes it h as faile d an d th e t o t al
number of ins t anc e s examin e d be s ufciently g reat,
th en th e resul t in g rat io O f favorable ins tance s t o
,

2 35

PROBABILITY

total number w ill

th e

be

foun d a pproximat ely


repeat e d, if a secon d set O f an equal number of
ins tances be likew ise examin e d T h ere is a law o f
t en d ency w h ereby nature seems to repeat h erself
even wh en th e att en d ant circums t ances O f an e vent
are mos t complex , an d beyon d al l po w ers Of accu
rat e d e t erminat ion A s th e resul t o f O bservations
ext en d in g over th ous ands an d thous an d s o f in
s ta
nces , it is afrme d th at about o n e fourth o f th e
chil dren born in th e w orl d die before th e a g e of six
years, an d about one h al f before th e ag e of sixt een
T ak e a group o f t en ch il d ren, th e rat io s w oul d per
h aps be d evi ate d from very mat erially ; in a group
o f a h un dr e d th e d e v iation is ap t t o b e less ; in a
group o f a th ousan d, s t il l less ; an d in a g roup o f
o n e hun d re d th ous an d th e ratios as above g ive n
T h e appro xima
w oul d be subs t ant i al ly re al ize d
t ion w oul d be so near th at th e error w oul d be ins ig
n icant as compare d with t o t al number of cases
T h e follo win g law, th erefore, expresses th is t en
d eney , th at wh ile in a small number o f ins t ance s
th ere is irre gul arity in th e observe d ratio betw een
th e num ber o f tim es a given event h as h appene d
an d i t s failure s , s till in a lar g e number of ins tances
th is rat io t end s t o w ard s a const ant limit Th is is
clearly seen in th e pit c hin g o f a penn y ; 10 th ro ws
mi ght very possibly res u l t in 7 h ead s an d 3 t ails ;
in 100 th ro w s, h o w ever th e rat io expressin g th e
resul t as t o h ead s an d t ails Observe d w ill be muc h
nearer th an in th e former case while if 1000 o r
th ro w s be Observe d, th e resul t w ill appro xi
mate th e ratio
The com parison of obs erve d cas es
.

'

INDUCTIVE LO G I C

23 6

with th e number g iven by th e calculation O f th e


probabilit ies in ques t ion h as been ma d e by Quet ele t
al s o by J evons T h eir resul t s are mos t si gnican t
an d int erest in g
Que tele t mad e 409 6 d raw in g s
from an urn cont ainin g 20 black balls an d 20 whit e
T h eoretically h e sh oul d h ave draw n as many wh it e
as black balls, 2048 eac h ; th e act ual drawin g s re
J evons
su lted in 2066 wh it e balls an d 203 0 black
mad e
th ro w s o f a penny ; th e th eoret ic al
re s ult sh oul d h ave been
h ead s ; th e ac tual
h ea d s
resul t w as
T h e t en d ency to ward s a const ant ratio in aggre
g at es cont aining a consi d erable number of ins t ances
is s t rikin gly illust rat e d in th e recor d O f baptis m s
t aken from an Ol d paris h re gist er in En glan d Th e
number O f male bapt isms re gi st ere d t o every 1000
female ran as follow s fo r th e re s pective years from
182 1 t o 1830 : 104 8 ,
104 1 104 9 104 6
104 7 1043 1043, 103 4
We see with wh at surpris in g
accuracy th e con st ant ratio w as repeate d substan
tially year aft er year T his t en d ency t o approxima t e
a const ant ratio is s een even in suc h in det erminat e
event s as railroad acc id ent s Here th e causes pro
d u c ing th e acci d ent s are so numerou s so d iverse , so
complex an d ex t en din g over so lar g e an area as
for example th e wh ole O f th e Unit e d St at es th at
w e s h oul d th ink th at th e resul ts w oul d ex h ibit so
many variations from any d enit e ratio as abso
l u tely t o elu d e all att empt s at accurat e d e t ermi na
tion Th e follo win g gures h o w ever g iven by th e
I nt ers t at e Commerce Com mission in dicat e re sult s
w on derfully correspon d in g for year aft er year
,

23 7

'

PROBABILITY
d

w m
w o
w m

3 8
m

o
n

m m
o
o

m c o
m m w

w
a

m
Q

H
H

E
D

m
o

a
r

m
m
s

3
3

5
9

i
t

m 2
w g 3
?

r 1

2
:

s
e

3 w

w
m

aE

E
E

3
3

o
a
o

B
o

9
4

m
m

<

O
O

m
a

o
m

we

8
9

3 5
0
o

w
m
m

m
.

abm

o
.

6
~

3
3

d
s

w
a
s

w
e

e
e

N
e

w m w

m
e

a
s

m3
a

3
n
:
5

INDUCTIVE LO G IC

238

m
m

9
6
a

o 6
n m

3
$

mm
a
8
m2

5
3

9
5

u
s

w
w
w
w

w
e

"
a

I:

'o
v

5
8

w
e

8
8

m
m

m
d

m
a

n
a

m
d
w

3
3
.

w
w
w
w

d
a
m

5
m

2
3
.

M
a

m
i
e

3
8

a
s

3 m
o

a
:

u
o

9
3
5
8
5

mm @
n m8 8 5 5 2 8
m8
m
a
8
8 2 g 8g 8 8
ng 2
o
8 o 5 p 5
o m5 0 0 4 4 0
c

la
S

J:
l

88
83

'

a e

l 0$

: 2
.

22

532
: 2
B

8e
2:

2
:

l2
$

l5w

l
"

w
w
h

S
l
u

m
n
a

a8

8n

w
u

8
6

m
a
n

8
n 2

@ am
5
9

E
m

m
6

m l
e

5
2

3
8
d

s
e

a
.

B
Z
u

e
o
o

<

m
.

o
A

S
S

e
e

5
8

8
a
e

E
a
c

h
.

23 9

I I Y

PR BAB L T

A n examination o f th es egures w ill d isclo s e th e


fac t th at th ere is a s t rikin g approximat ion t o an
accurat ely proportionat e di s t ribution O f th e number
O f accid ent s , of th e kille d an d o f th e injure d,
th rou gh out th ese s everal years It w ill be not ic e d,
al s o th at th e dis t ribut ion am on g employs , pa s sen
g ers, o th er per s ons et c t end s t ow ard s a re gularity
t h a t is remarkable wh en w e c ons id er th e ex t reme
complexity O f th e c ircum s t ances th at mu st c ombine
t o pro d uce th ese re s ult s A like re g ularity seem s
t o pervad e every d epart ment O f life Th e t ot al
number O f c rimes is a
pproximat ely th e same year
aft er year ; th e annual d eath rat e, th e appor tion
m ent of d eath s , moreover, t o th e several d isea s e s as
th eir evi d en t cau s es th e number o f mi s s ent l ett ers
th at reach th e D ead L ett er O f ce at Wa sh in gton
eac h year th e annual number O f s uic i d es of d i
v o rc e s all th ese diverse event s in d i c at e a re g ul ari t ,
y
in th e lon g run as re g ard s th eir numeric al est imat e
T h e resul t s wh ich are th us att ai n e d re gar din g ag
If a
g re g ates c anno t be st at e d as probable re s ul ts
sufcien tly lar g e num ber of in st ance s are t aken, th e
res ul t wil l be cert ain w ithin a very s mall an d in
many ca s e s an insi gni cant marg in In e st imat in g
th e probability o f a sin gle event, th e que s tion is
wh e th er i t w il l h appen o r no t h appen, an d th e ele
ment of unc ert ainty is th erefore prominen t In
d ealing w ith a gg re g at es , h o w e v er, no suc h elemen t
O f uncert ainty ent ers ; th e ques t ion is not wh eth er
th e quest ion
o r no t th ere w ill be cer t ain re s ul t s
c onc erns rath er th e d e g ree O f exac tness w ith wh i c h
th e resul t s will approximat e a definit e rat io; An d

"

"

'

240

IND U CTIVE

OG IC

th e la w O f t en den cy is th at th e lar g er th e numbe r


O f ins t ances th e g reat er w ill be th e approximat ion
t o an ac curat e an d d e nit e resul t
Th is is especially illus t rat e d in th e numerous in
surance companies , wh o s e business is con duct e d upon
th e basi s Of an approximat ely cons t ant d eath rat e
F or ins t anc e th e g eneral proce dure is some wh at as
follo w s : S uppose
pers on s ins ure th e ir lives at
$1000 per in divi dual, an d th e annual d eath rat e Oh
serve d over a rud e ext ent O f t errit ory, an d inclu din g
a very lar g e number O f ins t ances amo unt s to 200
person s out Of
T h e losses th en t o th e in
suranc e company will amo unt annually t o
on s uc h a ba sis T h e s e lo s s es dis tribut e d amon g
th e
insurin g in th e company w oul d amount
T h e company th erefore, h as a
to $20 apiece
numerical basis for calculatin g th e amount whic h
eac h per s on mu st pay in or d er to cover th e annual
losses an d provid e an assure d revenue for the co m
pany
I h ave, o f course s t at e d th e problem in roun d
numbers merely t o illus t rat e in g eneral th e pri nc i
ple involve d ; th e act ual calculat ion is more compli
cate d because in eac h part icular c ase th e age O f th e
in divi dual an d th e varying death rat es for d ifferent
year s mus t be t aken int o ac coun t T h e s ubs t ant ial
s t an din g Of th e innumerable insurance c ompanie s
in o ur co unt ry bears wit nes s t o th e fact th at th e s e
ent erprises are bas e d upon a prac t i c al cert ainty re
g ar ding d eathrat es wh en appl ie d t o lar g e ag gre
g at es Ch ance is thus eli minat e d almos t entirely ;
th at wh ic h w oul d be a serious risk as re g ar d s an ih
,

241

PROBABI LITY

d iv i d ual is subst antially voi d o f all ris k wh en lar g e


numbers are concerne d
Moreover, s t atis tics coverin g different classes are
O ften mos t val uable in in d icatin g t en d encies opera
t ive in th e classes wh en co mp ar e d one w ith an o th er
A ccor din g to M Loua ( E co no miste Frangais 1882,
V ol I p
th e follo w ing are th e g ures Of th e
annual mort ality in P aris :
.

Th e

h and well-to-d o
inh abitants
ric

c las se s ,

156

v ry

o ut o f e e

Th e

po o r

h b

v ry

out o f e e

285

in a itants

so in En glan d, th e avera g e d uration O f life


am on g th e w eal th y classes is from 55 t o 5 6 years ;
for th e w orkin g classes i t falls to 28 years o r even
lo w er 1 S uch comparison s are si g nicant in i nd i
catin g un derlyin g force s in society th at o th erw ise
mi ght be overlooked, o r, at leas t, not a d equat ely
appreciate d, an d wh ich a limit e d O bser v at ion c oul d
Mr D ar w in, aft er O bs ervin g
n o t accurat ely reveal
an d experimentin g upon a very larg e number o f
plan t s , foun d th e follo w in g gures respec t in g th e
relat ive pro duct ivit y of c ross and spon taneously
selffer til ize d o w ers : A s re gard s th e number o f
see d s per po d yiel de d by cross and self fert ilize d
o w ers, th e ratio was 100 to 41 respec tively ; th e
crosse d see d s compare d w ith an equal number o f
th e spontaneously self fertilize d see ds w ere h eavier,
in th e rat io of 100 to
Th e ratios th us disclosed
in ex aminin g a l arg e nu mber o f inst ances coul d
n o t h ave be en g aine d by any experimental me th o d
SO

al

1
9

Gide

405
P olitical E co no my ,
Darwin, Cross and S elf Fertilization, p 165
,

IND U CTIVE

LO

GIC

a d apte d for dealin g w ith in divi dual ins t ances Al


th ough th e cause is n ot quant it atively d et ermine d ;
a t enden c yOf a con st ant nature t o w ar d s a de nit e
end is clearly in d icat e d
ce c h aract eris t ics are O ft en disclose d by co m
Ra
parativ e st at istic s an d th e presence or absence o f
moral causes especially are thu s reveale d wh ich
o th erw ise coul d no t be d et ermine d w ith any con
deg ree O f d enitene ss T h e follo win g
s id erable
t ables w ill in dicat e th is :
.

'

In

c it ie s th e number of
suici d e s per
inh abi t an t s is as follo w s :
P aris 42 ; Lyon s 2 9 ; St P et ers bur g 7 ; Mosco w,
11 ; Berlin 3 6 ; Vienna 28 ; L on d on 23 ; Rome
8 ; Milan 6 ; Ma d rid 3 ; G enoa 3 1 ; Brussels 15 ;
Am st er dam 14 ; Lisbon, 2 ; Ch ris tiania, 25 ; Stock
h olm 2 7 ; Const ant inople 12 ; G eneva 11 ; D res
den, 51
Mad ri d an d L isbon sh o w th e lo w es t,
D res den th e high est gure
T h e averag e annual suici d e rat e in count ries o f
th e w orl d per
person s livin g is given by
D enm ark
Barker as follo w s : S axony
S chle s wig Hols t ein
A ust ria
S wi t zer
lan d
Franc e,
G erman Empire,
Hanover
P russia
Q ueens lan d
Vict or ia
N e w S outh Wale s,
Bavaria
N e w Zealan d
S outh A us tralia
Sw e d en,
N orw ay
Bel gium
En glan d an d
Wales
Ta smania
Hun gary
S co t
lan d
It aly
N eth erlan d s
U nit e d
St at es
Irelan d,
S pain,
Ru s sia,
S u ic id es

European

PRO BABILITY

243

T h e causes o f suici d e in European coun t ries are


reporte d as follo w s : O f 100 suici d es : Ma dness , d e
liriu m 18 per cent ; alcoh olism, 11 ; vice, crime, 19 ;
different d iseases, 2 ; moral sufferin g s, 6 ; family
matt ers , 4 ; poverty, w an t, 4 ; loss of int ellec t, 14 ;
consequence o f crimes, 3 ; unk no w n reasons, 19
Homicid es It a1y t akes th e lea d O f European
nations, w ith an avera g e annual crop of mur d ers o f
247 0 a rat io per
d eath s of
S pain fol
lo w s w ith a rat io O f
an d 1200 murders ; A us
t ria, ratio o f
an d 600 murders ; F ranc e ratio of
an d 662 mur ders ; En glan d, rat io of
an d
3 77 mur d ers
T h e gures , h o wever, represent
ac tual mur d ers , no t h om icides from all causes , as
d o th ose in th e U nit e d St at es t able
I llegitimacy Of eac h 1000 birth s , th e number
ille g itimat e, accor d in g t o s t atis tics publis h e d in
Lon d on, 189 2, w ere : Russia 27 I relan d, 2 8 ; Hol
lan d 33 , E n glan d and Wales , 46 ; Sw itz erl an d, 47 ;
I t al y, 7 3 ; N orw ay, 7 4 ; S co tlan d, 7 9 ; P russia, 80 ;
F rance, 84 ; Hung ary, 85 ; Bel g ium, 88 ; D enmark,
93 ; Sw e d en, 101 ; S axony , 125 ; Bavaria, 14 1 ; Aus
t ria, 147 N o accurat e st atist ics for th e Unit e d
S tate s exis t T h e lo w es t rate in Europe is th at
o f Conn au ght, in W es t ern I relan d, 7 per 100 0
,

'

Dr Al bert Leingwell, S ummit, N J

Wh en ph enomena indicat e a marke d d eparture


from th e rat io of frequency as d et ermine d by prior
or by th eoretical consi derati ons, th en
o bservat ion
i t is ordinarily inferre d th at a ne w cause h as be
come op erat ive , not before exis t ent, or, if present,
3

244

INDU CTIVE

LOG IC

it s effect neutralized F or in st ance, w e w oul d nat u


rally expect a die t o s h o w th e face th ree o n an
averag e about once in six th ro ws But if it re
peate dly t urns up th ree in succession an d no o th er
number appears , o r appears but rarely, w e are
w arrant e d in inferrin g th at th e die is load e d
T h e num ber of h omici d es in th e U nit e d St at e s in
189 4 far excee d e d th e annual number O bserve d for
th e several years pre c edin g Th is discrepancy is
easily account e d for by th e fact th at th e nat ural
nu mber w as s wollen by th e dea th s cause d by th e
st rikers an d riot ers in th e month Of J uly of th at
year S o al so a marke d d eparture from th e annual
d eath rat e O f s uch a c ity as N e w York is at once
an urg ent su gg es tion t o th e Boar d of Heal th t o
s t art inve s ti g at ion s th at will unearth th e h i dden
caus e th at one is c ons t raine d t o believe must be
present S uch caus es as d efect ive d rains prev a
lence of epid emic s etc are ag ain an d a g ain foun d
t o accompany an increas e o f th e avera g e d eath rat e
U n d er such c ircum st ance s th e me th o d Of invest i
gatio n wh en prac t ic able whic h s h oul d be pur s ue d
is t o en deavor t o break up th e t ot al int o smaller
g roups of a spec ic nat ure T hus if th e d eath rat e
for th e year is appre c iably increased, examine th e
d eath rat e per month S ee if any month sh o ws a
marke d d eparture from th e avera g e I f so th i s
w ill su gg est a careful inves ti g at ion O f th e c ircum
st ances an d ch aract eris tic s O f th e mon th in ques t ion
O r it may be possible to make a g eo graphical dis
tributio n O f th e t o t al over different sec tions of th e
city und er inv estigation S ome s pecial locali ty may
.

245

I Y

PR BABIL T

in d icat e an unusually lar g e d eath rat e I nves ti ga


t ion, th erefore, at th at point may reveal a lurkin g
cause of d isease oth erw ise unno t ice d
By similar consi d erat ions also, it is O ft en possible
to d is t in g uis h betw een a c h ance coinc id ence an d a
d e t erminat e cause wh ich h as pro du ce d th e event in
ques tion Fo r, if th e possibility of some o ne d e
nit e cause is consi dere d ou t o f th e ques t ion an d
th e ori g in o f th e event is foun d amon g complex
ph enomena O f such a number an d variety th at th ey
m ay form an in d e nit e number O f com binat ions ,
only o ne o f wh ic h can po s sibly pro duce th e event
in ques t ion th en th e probability th at th e event h as
ac tual ly been pro duce d by such a ch ance combina
tion is ext remely small We are th en th rown back
upon th e o th er hypoth esis, th at, ins t ea d Of one o ut
O f many possible combinat ions th ere is some one
d et ermin at e cause O perat ive in th e cas e It s nat ure
may no t be d enit ely ind icat e d, but at leas t th e
possibility of its presence is sug g es t e d
T his line o f reasonin g is illus t rat e d in th e fo l
lowin g account O f th e di scovery of th e exis t ence o f
iron in th e sun, in th e research es o f Bunsen an d
O n co mparin g th e spec t ra of sunl ight
K irchh o
an d o f th e li ght procee d ing from th e inc an d escent
vapor of iron it becam e apparen t th at at leas t six ty
bri ght lines in th e spec t rum Of iron coinci d e d w ith
dark lines in th e s un s spec t rum S uch c o inci
d enc e s coul d never be O bserve d w ith cert ainty ,
because, even if th e lines only closely approach e d,
th e ins t rument al imperfect ions o f th e spect roscope
w oul d make th em apparently coinci d ent, an d if one
-

24 6

INDUCTIVE

LO

G IC

line came w ith in h alf a millime tre of anoth er, on


th e m ap o f th e spect ra, th ey coul d no t be pro
noun ce d di s t inc t N o w th e averag e d is tance o f th e
solar lines o n Kirchh off s map is two millimetres,
and if w e th ro w d o wn a line, as i t w ere by pure
ch ance, on such a map, th e probability is about 4
th at th e ne w line w ill fall wi th in one h alf milli
met re on o ne si d e or th e oth er of some one o f th e
solar lines To put it in anoth er w ay , w e may
suppose th at eac h solar l ine, eith er o n account o f
it s real brea dth, or th e d efect s o f th e in s t rum ent,
pos s es s es a brea dth of one hal f millimet re, an d th at
each line in th e iron spect rum has a like breadth
T h e probability, th en, is just
th at th e centre o f
each iron line will come by ch ance wi th in on e milli
met re of th e cent re o f a s olar line, so as t o appear
t o coinci d e w ith it T h e probability of casual coin
c id en ce o f each iron line w i th a solar line is in l ike
manner
Coinci d ence in th e case o f each o f th e
sixty iron lines is a very unl ikely event if it arises
ca sually, for it w oul d h ave a probability of only
eo
(g) or less th an one in a t rillion T h e o dd s , in
sh ort, are more th an a million million millions t o
unity a gains t such a casual coinci d ence Bu t on
th e o th er hypoth esis, th at iron exis t s in th e sun, it
is hi gh ly probable th at s uc h coinci d ences w oul d be
observe d ; it is immensely more probable th at s ix ty
coinci dences w oul d be observe d if iron exis t e d in
th e sun, th an th at th ey sh oul d arise from ch an ce
Hence by our principle, it is immensely probable
th at iron does exist in th e su n 1
.

'

Jevons

P rincip les of S cience,


.

pp

244 , 245

247

I I Y

PR BAB L T

T h is principle is also illus t rat e d in inst ances Of


circums t an ti al evi dence I n suc h cases th e observe d
combinat ion o f so many d iverse circums t ances , e v en
as re g ar d s an in d e nit e num ber o f minor d et ails ,
preclu d es th e hypoth esis of ca s ual coinci d ence, an d
su gg es t s some one d e nit e cause th at w ill prove a
uni fyin g principle of explanat ion o f all th e a tt en d
ant circums t ances A s Mr J us t ice Bullen says :

A presump t ion wh ich necessarily aris es from cir


c u m stan c e s is very O ft en
more convinc in g an d
more sat isfact ory th an any o th er kin d o f evid ence
I t is n o t within th e reac h an d compas s of h uman
abil ities t o invent a t rain of circum s t ances wh ich
s h all be so connect e d t o g eth er as t o amount t o a
proof o f guil t w ith out affor ding opportunities to
cont ra d ict a g reat part if n o t all , Of th ese circum

st ances
T h e follo w in g account, t aken from The N ew York
Law J o u rnal illus t rat e s th e probat ive force o f cir
c um stantial evi d ence :
.

h ( Marc h 1895 21 S E R
364) th e S upreme Co urt o f A pp eals o f V irginia sus taine d a
co nvictio n o f mu rder th e c rimin al agenc y b eing es tablis h e d
by circumstantial evidence Th e follo wing extract fro m th e
o pinio n pres ents t h e m ain fac ts wh ic h implic ate d th e d e
fend ant
O n th e 8 th d ay of Dece mb er 189 2 Ph ilip Norm an
Nich o las th e pl aintiff in erro r o ne Jam es Mills and h is
w ife A n na A Mills and th eir th ree small c hil dre n were
living in th e u p p er p art o f Henrico Cou nty o n a farm
kno wn as the W ic kh am Plac e ab o ut o ne m ile fro m James
River Nic h o las w as th e re nter o f this farm an d c ultivate d
it o n s h ares
He w as h ims elf ho wever ch ie y engaged as
In Nic

h olas v

C o mmonwe al t

.
'

248

INDU CTIVE

LO

GI C

r pper h aving a numb er o f traps s et alo ng b oth s ides o f


He em p lo y e d J ames Mills with wh o m h e lived
th e river
and o n e W illiam J u ds on W ilkerso n as s ubten an ts to d o
th e farm wo rk for a p o rt ion o f h is sh are o f th e c ro ps
W ilkerso n live d with an aged mother in a small house very
ne ar t o Mills h o us e near enough to s ee into th e wind ows
Philip N Nich olas th e pris
o f o ne ho us e fro m th e oth er
o ner was an unmarrie d man and lived in a ro o m o f th e
a t a

h o use occupied b y James M ills and his family The evi


denc e sh ows th at on th e night b efore th e dro wnin g th e
prisoner James Mills and William J W ilkers on were
to geth er at th e h o use o f Mrs W ilkerso n th e latter s
mo ther and th ere arranged and determin ed u p o n a trip
Th is
acro s s th e river th e n ex t mo rning to take a b ee tree
exp ed itio n was s u gg es ted p lann e d and c arrie d o ut b y th e
p riso n er W ilkerso n was very u nwilling to go and nally
c o nsente d at th e s ugges tio n o f h is mo th er wh o s aid th at as
Mr Nic h o las s eem ed s o anxio u s for him to go he h ad better
M ills was unwilling to go u nless W ilkerso n went
d o so
W il kerso n said h e wo ul d rath er plo ugh th an go Th e priso ner
replied If yo u will g o yo u sh all no t lo s e anything In the
c o urse o f co nvers at io n wh ich resulte d in th is exp ed ition
b eing agreed u po n b o th M ills and W ilkers on stated in th e
presence o f Nich o las that th ey c o uld no t swim and were
very m uc h afraid o f water ; that th ey d id no t like water
m o re th an kne e deep
Th e fact th at th e y c o ul d n ot swim
w as gen erally kn o wn to th eir frien ds
It is f urth er sh own
th at it was th e h ab it o f Nic h o las to g o every m orn in g early
to th e rive r to examin e h is traps
A nd it ap p ears fro m th e
evid ence th at o n th e mo rn ing o f th e d ay th e dro wn ing
o ccurre d h e wen t to th e river ab o ut daylight and retu rn ed
ab o u t b re akf ast time an d wh en qu estio n e d ab o ut it s aid :
I d id n o t g o to my trap s th is mo rning
I was sic k
He
af terwards to l d M rs W ilk erso n h e d id no t c atc h any th ing
Every th ing b e ing in read in es s to c arry o ut th e plan for th e
d ay th ese th ree men s tart e d fro m h o me ab o ut nine O c lo c k
in th e mo rn ing equ ipp e d with every th ing necessary fo r
.

'

249

PROBABI LITY

k g th e bee tree h aving with th em two buckets h olding


t wo and o ne h al f to three gallo ns e ac h fo r th e h o ne y two
ax e
o n e h atc h e t and a p iec e o f ne tt ing t o pro tec t th e per
so n f o m th e b ees
Th e b o at used b elo nge d to o ne Joseph
Bru in and o n th eir way to th e iver an u ncle o f the owner
w as as ke d if th ey migh t us e th e b o at and w as to l d th e y
c o uld ge t th e k ey wh ic h u nlo ck e d th e b o at fro m its fas tening
to th e b ank f m Bruin th e o wn er
Th e pri o n r re p l ie d
th at h e h ad a k ey o f h is o wn and h ad o f ten use d it b e f o e
with o u t p e rmiss io n
It app e ars th at th ey lan de d o n th e
C h e terel d s id e o f th e river at a p o int o ne mile and a h alf
fro m wh ere any o ne l ived and proceed ed to the bee tree
wh ich was o ne mile fro m the p o int o f lan din g I nves tiga
tio n s h o we d th at th ere w ere n o tracks ab o u t th e p o int o f
lan din g b u t th ose o f th e th ree me n go in g fro m and eturnin g
to th e b o at
It fu the app ears fro m th e s tat e me nt o f th e
priso ner th at after reach ing th e tree th ey conclu d d no t to
c u t it b ec au se it w as a l arge tree n ear th e main ro ad and
migh t g e t them in to tro u b le an d f th e fu rth e easo n th at
th e h o le w as sm all and it m igh t no t h ave an y h ney in it
anyh ow
Th e tree w as af terwards cu t by o rde o f th e M ag is
t ate and fo un d to b e full o f h o ney
It furth e r app e ars th at
th e b o at was a s mall o n e ab o ut te n feet lo n g and ab o ut t wo
and o n e h al f feet w id e an d th at b o th in go in g o ver and
re tu rning th e p iso ner s at in th e extreme rear of th e b o at
with h is fac e to th e fro nt and that Wilkerso n and Mills s at
in fro nt o f him with th ei fac es to th e fro nt and th eir b acks
This po sitio n o f th e p a ties th e p riso ner
to th e acc used
When
ad mitte d ve ry relu c tan tly w h en qu stio ne d ab o u t it
returning and ab o u t fty yards fro m th e Henrico s h o re
th e b o at s udde nly lle d with water and M ills and W ilker
Th e
so n were d ro wn e d an d th e priso ner sw am to s h o re
next d ay th e M agis trate o f th e district w as n o tie d o f the
Th e b o at
o cc urre n c e an d an in ves tigatio n w as se t o n f o o t
w as go tte n o u t o f th e water and it w as fo und th at im me d i
th e se at wh e re Nich ol as s at th ere we e th ree
ately u n d e
h oles freshly b o red with an inch and a h alf auger The
ta in

s,

ro

or

r r

250

INDUCTIVE

LO GIC

vidence o f the o w ner o f th e b o at sh ows th at o n Tues day


evening th e 6 th o f Decem b e r h e u s e d his b o at and it was
sound
It was tak en by Nic h o las for th is fatal tr ip Th u rs
Furth er investi gation
d ay mo rning th e 8 th o f De c em b er
disc overe d fresh pine sh aving s corresp o n ding to s ize o f th e
h o les and to th e wo o d th e b o at was mad e o f wh ich h ad
b een th rown into th e water but h ad drifted upon th e sh ore
near th e p o int where th e b oat had sto o d f astened to th e
He nrico side Th ere were also found c o rn c o b s wh ic h h ad
been cut to exac tly t th e h o les in th e b oat which h ad also
drifted to th e same p oint It was s h own th at th e prisoner
h ad in h is p o ss ession an auger jus t th e s ize o f th e h o les
This th e p riso ner at rst d enied b ut afterwards s aid th at it
m u st b e ab o u t th e pl ac e so m ewh ere
Diligent search was
m ad e for th e auger b u t it w as never f ound
Taken to geth er th e c ase is an interestin g illustration o f
th e conc lus ive pro b ative f o rce o f c irc u ms tantial evidenc e
provided th ere is enou gh of it Th e o ld s aying th at mu r
d er will o ut is almo st u nex c ep tio nally true as to mu rd ers
o f elab o rate s teal th an d c o m p lex ity o f detail
O nc e l e t a
clue b e o btaine d to th e c h ain o f c aus ation and mo tive an d
th e mys tery unravels almost o f itse lf
It is quite n atu ral
th at m os t o f th e el ab orately pl ann ed murd ers o f rec ent
A nd th e Harris B u ch
times sh o u l d h ave b een b y p oison
anan and Mey er c ases in New Y o rk d is clo s e h o w c o mp ara
tiv ely eas y d etec tio n and c onvic tion are in c rim es o f suc h
c lass
It is s ig nicant th at two o f th e greates t enigmas in
A meric an c riminal annal s during th e last quarte r o f a c en
tu ry h ave b een th e N ath an murder and th e B o rd en murd er
In b oth c ases th e killing was d one n o t by meth o d s c alcu
lated to c o nc eal th e agenc y o f a mu rderer bu t in th e m ost
prim itive and bru tal mann er No trac eable ph ysical clue
to any p articul ar p erson was lef t an d we are inc lin e d to
b elieve that in b o th c as es th e c onnec tio n o f th e murderer
1
W ith th e crime was merely c asual o r accid ental
e

Th e N ew Yo rk L aw J o u rnal , Th urs ay ,

May

2 , 1895

251

PROBAB I LITY

In th e various illus t rations which h ave been given


w e n d th at th e th eory of probabili ty provi d es a
me th o d o f d ealin g wi th p h enomena wh ic h c anno t
be subj ect e d t o th e ordinary in duc tive meth o d s
T h e ph enomena are so complex th at a speci c cause
canno t be d et ermined, for th e real cause in ques tion
is a correlat ion of many diverse forces , an d if only
a fe w in s t ances are ex amine d no causal connec t ion
will be d isclosed ; i t is necessary, th erefore, t o d eal
w i th lar g e num bers, s t atis tical averag es , et c in
or d er t o d et ec t an emer gin g relation o f a causal
ch arac te r expresse d by a cons t ant ratio T his ratio
once d et ermine d i t becomes a furth er t es t, as w e
h ave alrea dy seen wh en th e resul t s w i d ely d epart
from i t t o su gg es t th e presence of a ne w force o u t
si d e of th e comb inat ions t o wh ich th e effec t w oul d
be natu rally referre d ac cor din g t o th e in dicat ions of
th e probabili ty ratio Th is latt er mo d e of in ference
i s akin to th e meth o d o f resi d ues , for th e inf erence
in ques tion is bas e d upon th e fac t th at th e pro babil
ity ratio will accoun t for only a cert ain frequency
of occurrence of th e event un d er invest i g ation ; a
m arke d excess mus t be account e d for by posi tin g a
d enit ely operative cause A n d if an ant ec e d ent
of suc h a nature is know n to be present, th e su gg es
tion at once rises in our th ou ght th at this in all
probabili ty is th e cause pro ducin g this excess in th e
resul t s
.

C HAPTER

" VI

EM P I RI CA L LA W S

T HE RE is a class of law s wh ich are int erme diat e


betw een a universal , in d uctively g roun d e d by scie n
tic d e t erminat ion an d a law of t en d enc y wh i c h is
th e expression of th e probability o f th e h appenin g
o f an even t in spi t e o f reco gnize d ex c ept ions
T h ese
are law s whic h h ave been observe d t o obtain un d er
given condi t ions of time plac e an d c ircums t ance
an d yet th e causal relation no t suffi cient ly d et er
mine d to w arrant a nec essary ext ension of th e
same to a sph ere beyond th at wh erein it h as been
observe d t o be operat ive S uch law s are kno wn as
Empirical Law s
We h ave th erefore, th ree c lasses
Th e
o f law s of varyin g d e g rees o f probabili t y
rs t is wh ere th ere has been a scienti c ally d et er
mine d causal connec tion betw een ant ec e dent an d
consequent ; an d no t only h ave no exceptions been
no t e d but th e possibili ty o f th ere bein g an ex c ep
t ion h as been eli minat e d by s t ri ct experiment al
meth o d s T h e secon d is wh ere th e re gul arity of
sequence h as been broken by ac tual exceptions , an d
th e resul t o f th e observations of in s t ances gives an
in dication only of th e relative frequency of occur
renc e an d failure wh ich w ill probably c h aract eri z e
,

2 52

253

LAWS

EM IRI C AL

o th er event s of th at n atu re T h e thir d class an d,


as h as been sai d , an int erme d iat e class comprises
all expressions o f uniform sequence or coexis t ence
wh ere no exception wh at soever h as been no t e d , and
yet th ere is no g roun d fo r necessit atin g a uni versal
expression of th e observe d uniformity T h ere is
h ere al ways a possibili ty of an excep tion appe arin g,
o r o f an excep tion h avin g been overlooke d
Th is
pro duces an element o f un cert ainty wh ich perva d es
all ph enomena of th is sort
T h ere are several kin d s o f empiric al law s , as
follo w s :
1 Wh ere th e causal relat ion is in process of
scienti c d e t erminat ion ; a uniform connect ion
be tw een p h enomena h as be en observe d, an d as
yet h as no t been prove d All law s , n ally d et er
mine d as expressions o f causal connec tion pass
th rou gh th is empiric al s t ag e
S ome expressions
o f uniform relat ions never p ass beyon d th is s t a g e
because as w e h ave seen, th e nature of th e ph enom
ena may be suc h as t o preclu d e all experiment or
even in direct veri cation
E mpirical l aw s may beco m e ul timat e law s o r
d erivat ive law s , as th e case may be U l timat e
law s are th ose wh erein th e caus al relation be tw een
a simple ant ece d ent an d it s correspon din g couse
quent h as been scienti cally d et ermine d in t erms
of th eir exact quantit at ive variation an d expresse d
in th e simples t form possible
Th e derivative
la w s , h o w ever as th e name in dicat es, are m ore
concret e expressions o f th e ul timat e an d simpler
l aw s t o wh ich th ey are re ferre d as special cases
.

254

IND U CTI VE

LO

GIC

A n empirical law may be prove d direc tly an ul ti


mat e law or be prove d a derivative law di rec t ly
trac eabl e t o an ultimat e law, as its basis o r lo g ical
g roun d We may observe th at a glass o f ice w at er
al w ays sh o w s d rops o f mois ture on i ts out er su r
face T his uni formi ty as thus expresse d h as th e
forc e only o f an empirical la w N o att empt h av
ing been made as yet, t o explain th e presence
of th e mois ture, it s empirical nature is evi d ent
Bu t as soon as th e mois ture on th e glass is t rac e d
t o th e cond ensation o f th e mois ture in th e at mos
ph ere o w in g t o th e di fference o f t emperat ure be
tween th e at mosph ere and th e col d surface of th e
glass w e h ave th e empirical law becomin g a d eriva
tive law ; th at is th e expression o f a uniform
sequence direc tly t raceable t o th e more ul timat e
law of th e saturation an d condensation of vapors
T h e pro g ress of scienti c an d lo gically acc urat e
th ou ght is al ways marke d th erefore by th e resolu
tion of empirical g eneralit ies int o d erivative an d
ul timat e law s
2 T h e ch aract er o f an empirical law is attac h e d
to th e relat ion exis t in g be tw een ant ece d en t an d
consequent wh e n th at relation is a c omplex o n e in
which a simple causal relation is so involve d wi th
oth er element s ent erin g int o combination w ith it,
that its real na ture is thus hi dd en an d c annot read
ily be disclose d T h is class inclu d es all causal rela
t ions due t o collocations of various kind s th at are
necessary t o pro duce th e require d effect A s Mill
h as point ed o ut : It is th e nature o f an empirical
law th at w e do not kno w wh e th er i t result s from
,

P I

EM IR CAL LAW

255

th e different effect s o f o ne cause or th e effec t s of


d ifferent causes
We cannot t ell wh e th er i t d epen d s
wh olly upon la w s, or partly upon law s an d p artly
upon a collocation If i t d epen d s upon a colloca
t ion , i t w ill be t rue in all th e cases in which th at
part icular colloc ation exist s
But since w e are
entire ly i g nor ant in c ase of it s d epen ding upon a
collocat ion wh at th e collocation is w e are no t safe
in ex t en din g th e law beyon d th e l imit s of time an d
place in wh ic h w e h ave ac tual experience of its
t ruth Kno win g of no rul e o r principle to wh ic h
th e collocations th emselves conform w e canno t
conclu d e th at because a collocation is prove d to
exis t wi th i n cer t ain limi t s o f place o r time i t w ill
1
exis t beyon d th ose limit s
T h ere are many illus t rations of such observe d
g enerali ties wh ere th e eect is due larg ely, if no t
t o coll ocations Th e effect o f certain
al t o g eth er
m e d icines upon th e human sys t em, th e openin g an d
s huttin g of some o w ers at cert ain h ours of th e
d ay th e local ac tion of ti d es at various places on
th e earth s surface, th e a d apt ation of cert ai n plant s
t o a peculi ar kin d of soil, th e reappearance o f some
c h ronic di seases, as h ay fever at th e same season
e ach year even to th e very d ay of th e month, all
suc h g eneralit ies h ave merely an empirical w ei ght,
an d th e eects men t ione d are larg ely due t o collo
cat ions th at cannot be d e ni t ely det ermine d S o
also cer t ain la w s o r cus t oms may h ave prove d
bene cial in th e coun t ries in which th ey h ave been
t rie d, an d yet, in count ries W h ere con di tion and
.

Mill , Log ic, B o o k III

Chapter "VI

256

INDU CTI VE

OG I C

circums t ance are radically different th ey may fail


wh olly of bene cial resul t s T h ere may be also
cert ain in dus t rial c ircums t ances wh ich in one c oun
t ry mi ght be con ducive t o prosperity an d in an
oth e r country to a dversi ty
C ertain a gricul tural
metho d s which in o n e section of th e count ry t en d
t o an increase of pro duc tive po w er in ano th er
mi ght prove a complet e failure A g overnment al
policy may in o ne count ry lea d to unparallele d s u c
c ess ; in ano th er h o w ever a like policy mi ght lea d
t o disas t rous resul t s
T h e famous formula of M al thus, th at population
t en d s t o increase in a g eomet rical pro g ression whils t
th e means of subsist ence c an only increase in an
ari th metical pro g ression can h ave only an empirical
force It s ext ension in t o an in d en i t e future is
unw arrant able A s is kno wn, pro d uction h as in
crease d enormously an d at a ratio vas tly g reat er
th an any cont empl at e d by Mal th us as at all in th e
ran g e of possibili ty Many causes, on th e o th er
h an d may combine t o ch eck th e rapi d increase of
populat ion Th e collocations h ere are so complex
as t o defy any d enit e pre dic tion T his is t r u e of
all t en d encies wh ich are due t o present social con
ditio n s ; th e con dit ions th emselves may so vary in
t ime to come as t o ch an g e t ot ally th e accept e d
g eneralizations of t od ay T h eir empirical ch arac t er
is th erefore mos t evi d ent
3 A th ir d class of empirical law s comprise s all
those g eneralizat ions wh ich represent an a gg re g at e
o f proper t ies in th e same in d ivi dual
In all suc h
cases no c ausal relation h as been speci c ally d e
,

EMPIRI

C AL

LA W

25 7

termin ed

be tw een th e properties th emselves, o r


betw een th e properties and th e wh ole in which th ey
coinh ere O ut si d e o f our experience th e proper
ties obser v e d mi ght be mat erially ch an g e d, an d ye t
n o t affec t th e in t e g ri ty of th e conc ep t in g eneral
A proposi tion su c h as all s w ans are wh it e can h ave
only e m pirical force ; fo r beyon d our experience,
th e discovery of black s w ans w oul d forbi d th e prop
o sitio n bein g re g ar d e d in th e li ght of a universal
Many properties o f subst ances are thus referre d t o
th e nature of th e subst ance i t self as th eir g roun d
an d ye t because th e exac t c ausal relat ion i s n ot d e
term in e d th e connec t ion can be consi d ere d only as
an empirical one In o th er w or d s , referenc e t o some
g roun d as explanat ion o f a ph enomenon wi th out
explainin g why o r h o w su c h reference is mad e h as
al w ay s th e force of an empirical la w only T h e fol
o w in g are empiric al g eneralizat ions o f this nature
C opper is duct ile ; s t eel is elas t ic ; glass is brittle an d
t ransparent ; th e compoun d silicat es o f alkalies an d
alkaline met als are t ransparen t an d oth er inst ances
o f like nature th at can be m ul t iplie d in d e nit ely
In th e sp h ere o f biolo gy, Mr S penc er h as drawn

durin g th e era in which


att ent ion t o th e fac t th at
un iformi ty of many qui t e simple inor g anic rel a
tions w as s till unreco gni z e d, cert ain or ganic rel a
t ions, int rinsically very com plex an d special, w ere
g enerali z e d T h e cons t ant coexis t ence o f feath ers
an d a be ak o f four le g s wi th an int ernal bony
frame w ork are fact s wh i c h w ere and are, familiar
Did a sava g e n d a bir d w i th
to every sava g e
t eeth, or a mammal cloth e d wi th feath ers h e w oul d
.

2 58

be

INDU CTI VE

LO

GIC

muc h surprise d as an ins truc t e d naturalis t


N o w th ese uniformi t ies o f or ganic s t ructure th us
early perceive d, are of exactly th e same kin d as
th ose more numerous ones lat er est ablis h e d by
biolo gy T h e cons tant c oexist ence of mammary
gland s wi th tw o o c cipit al con dyles t o th e skull
of vert ebrae w ith t eeth lo dg e d in socket s , of front al
h orns with th e h abit of ruminat ion, are g en eralize
tions as purely empirical as th ose kn own t o th e
O ri ginal hu nt er T h e bot anis t cannot in th e leas t
un ders t an d th e complex relat ion betw een papilio
naceo u s o w ers an d see d s borne in a tt ene d po d s ;
h e kno w s th ese an d like connections simply in th e
same w ay th at th e barbarian kno w s th e connections
betw een par ti cular leaves an d particular kind s o f
1
S uc h know le dg e as Mr S pencer h ere d e
w oo d
scribes is a kno wle dg e o f th e coexist ence o f tw o
ph enomena in th eir t ot ality wh ic h resis t all at
t emp t s t o analyze int o th eir component part s
Moreover, law s wh ich are bu t g eneral d escript ions
of c orrelat e d event s h ave th e same force as th e
d escriptions of coin h erin g attribut es of subs t anc es
T h ey too rank as empiri c al g eneralizations T h e
successive s t ag es in th e g ro wth of a plan t from
seed t o o w er an d frui t th e embryoni c as w ell as
th e post-nat al d evelopment s in animal life, th e
h abit s an d ins tinct s of animals ,
all th ese are d e
s c riptive g enerali z at ions wi th ou t any att emp t at
c ausal d e t ermination
4 All g eneralizat ions expresse d in t erms of prob
ability only bec ause Of kno wn ex c ep tions rank as
as

S p enc er, Classic atio n

th e S c ien ces ,

53

LAws

PIRI CAL

EM

259

em p iric al law s Here, even in th e t ime pl ace, and


circums t ance of observa tion, th e law has not been
foun d al w ays vali d T h e si g ni cance of an empiri
cal l aw, if w e allo w th is la tt er cl ass to b e inclu d e d
un d er th em , is evi d ent ly th at of th e cont r ad ict ory
o f a law wh ic h is th e resul t o f a causal d etermi
nation Every g eneral i z at ion not caus al ly d e ter
mine d is th en t o be re g ar de d as an empirical law
T h ere is h o wever, a narro w er us ag e of th e t erm
wh ic h d oes no t inclu d e th is l att er class ; n amely
a res t ric t ion o f th e t erm empiric al la w t o si g nify
th e expression of a relation wh ich h as been foun d
constant th rou gh out th e sph ere of observ ation, an d
yet wh ere th ere exis t s no kno wn c ausal g roun d by
re ason of whic h w e w oul d be warrant e d in infer
rin g th e continu ation o f th is relation in a sph ere
beyon d th at already observe d We mi ght add th at
wi th th is th ere is al so th e expec tation gre at er or
less , accor din g t o th e circu ms t ances att en din g th e
ph enomena, that th e g enerali z e d experience will
be furth er con rme d by subsequent observ ation in
a wi d er sph ere This res t ric t e d meanin g of an em
pirical law is th e one g enerally un d ers too d, unless
i t i s impl ie d t o th e contrary
A n empiric al uniformi t y g ener ally resul t s from
th e meth o d of agreement
Observe d ins t ances ,
even so sel ec t e d as t o v ary th e antece d ent s as
muc h as possible, cann o t al one es tablis h a law
o f uniformi ty th at s h all h a v e universal v ali d i t y
The me th o d of ag reement w e h ave seen, nee d e d t o
be supplement e d by th e meth o d of difference if
possible, or by an hypo th esis cap able of subse
.

I NDU C TIVE

2 60

LO

G IC

quent vericat ion A n empirical law is th erefore,


due ei th er t o some d e ciency in meth o d, or t o th e
nat ural limi tations O f our kno wle dg e
T h e element o f uncert ainty att ach e d to all infer
en ces d epen din g upon th e extension of an empirical
l aw in t o unkno w n t erri t ory i t h as been insis t e d
upon in several quart ers , m ay apply equally as w ell
t o all inferences d epen din g upon th e resul t s o f th e
in du ctive meth o d s even wh en mos t scienti cally
d et ermine d I t is cont en d e d th at even a causal
relation h o w ever rmly groun d e d and h o w ever
simple may be it s nature neverth eless present s an
empirical c h arac t er I t may g ive assurances of a
hi gh d e g ree o f probability but c an never pro duce
absolut e certit u de in ou r mind s Mr Venn, for in
s tance, h as style d his w ork on induct ion E mp irical
L og ic th at by th e t i tle h e mi ght in dicat e h is point
He says in th e preface t o
o f vie w in this re g ar d
his work By th e int ro duc tion of th e t erm em pir
ical into th e title I w is h to em ph asi z e my belief
th at no ul timat e obj ective cert aint y such as Mill,
for ins t ance seem e d to att ribut e to th e resu lt s o f
induction, is attainable by any exercise of th e hu

man reason
Re g ar d e d in th is li ght all law s are
empirical
Th e distinc tion h o w ever betw een empirical la ws
in th e sense g enerally un d ers t oo d an d law s ex
pressin g causal relations scienti cally d et ermine d
is a real dis tinc tion, an d a si gni cant one as w ell
A n d this mus t no t be overlooke d ; an d it cann o t
be obli t erat ed by any shift ing of th e point o f
v ie w
F or t o doub t th e vali dity o f an empirical
.

,
.

P IRI CAL LA WS

2 61

EM

la w wh en ex t en d e d t o a sph ere beyon d th at whic h


h as been observe d cas t s a re ec tion merely u p

on one s abili ty a d equat ely t o d et ermine th e con


n ectio n s exis t in g be tw een
th e various element s
involve d in th e par tic ular ph enomena un d er in
v estig atio n
T h is is , h o wever n o confession of
inabili ty t o discover th e causal connections o f
p h enomena in g eneral in such a manner as t o
d et ermin e law s o f universal validi ty To s ay that
all la w s h ave onl y empirical si gni cance is t o re
ect upon th e basal pos t ulat es of kno w le dg e
Ou r
w orl d is th e w orl d as w e kno w it th e w orl d of our
consciousness To discre dit th e uniformi ties an d
re gulari ties th erein exis tin g an d wh ic h n d ex
pression in universal la w s , is to discre dit th at
which w e feel obli g e d t o think in or d er th at our
w orl d O f kno wle dg e re gar d e d as a sys t em m ay
remain consis t ent w ith it self th at is part t o part,
an d part t o wh ole
We must, therefore re gar d
an empirical law no t as th e nal form of kno wl
e dg e o r th e g oal o f in duc tive research but rath er
as m arkin g a t ransit ion s t ag e t o w ar d s complet e
c aus al d e t erminat ion An d even wh en, o win g t o
th e nature of cert ain p h enomena w e are no t able
to pass beyon d th is t ransi t ion s t a g e o f empirical
d et ermination nevert h eless suc h ins t ances by con
tr as t bear unimpeac h able t es t imony t o th e fac t
th at th ere are o th er p h enomena o f suc h a nature
th at it is possible t o subj ec t th em t o an analysis
wh ic h will disclose causal connec tions of such a
ch arac t er as to form a basis for th e formul ation o f
un iversal la w s
,

C H APT ER

" VII

FA LLA C I ES

A CO N S I DE R AT I O N o f th e various kind s o f in duc


t ive fall ac ies , an d th eir ch aract eris tic features may
be re gar de d as th e obverse represent ation of th e
g eneral th eory of in duct ion F rom th e o ne poin t
of vie w w e consi der th e positive con di tions of t rue
in ductive inference ; from th e obverse point of vie w
w e re gar d th e various breach es of th ese inductive
con ditions Th e discussion of fallacies th erefore
ind icat es no pro gress in th e eluci d ation of th e sub
j ect un d er consi d eration ; it rath er serves t o em ph a
size dis tinct ions an d requirement s alrea dy indicat e d
by presentin g th em in a new li ght an d from a dif
Th e subj ec t of fallacies is g enerally
feren t an gle
t reat e d by exhibitin g th rou gh various illus t rations
th e cases in which th e positive con di tions of induc
tive inference h ave faile d o f satisfac tory ful lment
S uc h illust rat ions o f th e infrin g ement of th e require
ment s of vali d in duc tion I h ave end eavore d t o
incorporat e in th e bo dy o f th e t ext in connect ion
w ith th e exposition o f th e various con di tions an d
requirement s th emselves In thi s ch apt er I sh all
att empt t o in dicat e th ose fallacies especially wh ich
are due t o th e psyc h olo gical dis turb ance o f o u r
,

26 2

26 3

FA LLA C IES

normal lo gical processes A n enumeration of th ese


t end encies , partly psy c h olo gi c al an d partly lo g ical
may serve t o i m press u pon us th e d an g er of fallin g
in t o e as y errors , t o wh i ch th e hum an min d g enerally
is liable T h ese errors emerg e in th e various mental
processes T h ey are as follo w s
.

I
II
III

IV

E rrors

of P ercep tion
E rrors of J u dg ment
E rrors Of th e Imagination
E rrors of th e C oncep tual P rocesses
.

I E rrors of P ercep tio n


O bservation is th e in
s t rument o f research preeminently in all in ductive
E xperiment is but a meth o d fo r increase d
I nquiry
facil ity and accuracy o f obser v ation We may say
th er
efore, th at all th e d at a of in ductive inference
are furnis h e d by this one process Observation Any
d eran g ement o f o ur po w ers of observation wil l affect
th e nature o f th e d at a an d th erefore th e nature of
th e resul t s of in duc tion It becomes th erefore all
import ant th at w e s h oul d be appraise d at leas t of
th e various circums t ances wh ose t en d ency is t o
o perat e in th e mi d s t of th e perceptive processes as
dis turbin g forces We h ave th e follo win g possibili
t ies of e rror in th e sph ere of pePCeption :
1 Errors due t o a failure t o t ake in th e wh ole
el d o f vi sion T h ere may be port ions omi tt e d
wh ich possess a de t erminin g si g ni cance as re g ar d s
th e obj ec t of invest i g ation T hus exceptions m ay
be overl ooke d th at m i ght h ave an import ant bearin g
upon some receive d hypoth es i s ; or a fac t mi ght be
passe d by wh ich, if known, w o ul d prove h i gh ly su g
.

'

26 4

I NDU CTIVE

O G IC

Various

d evices h ave been employe d t o


enlar g e th e sph ere o f observat ion beyond th e n atu
ral limi t s of th e senses A s , for ins t ance, soun d s
wh ich are inau d ible t o u s may be d et ect e d by means
o f a sensi tive ame ; th e t elescope, th e microscope ,
serve t o rend er th e dist ant near, an d th e small
lar g e I t h a d been no t e d th at th ere w as a su dd en
elon g at ion o f an iron w ire at a particular t empera
ture whils t und er longitu dinal s t rain d urin g th e ac t
o f coolin g from a re d h eat ; an a dd i tional circum
s t anc e was not e d by Professor Barrett wh en per
formin g th e experiment in a d arkene d room namely,
th at at th e moment of elong ation th e wire su dd enl y
evol v e d h eat an d exhibi t e d a visible an d c o nspic u
re dness l T his c ircum
o u s moment ary g lo w o f
s t anc e it w oul d be impossible to not e unless in a
d arkene d room Thus , a prominent charact eris tic
o f scient i c observation is th e en d eavor t o ex t en d
Here also
c ont inually th e sp h ere o f observat ion
muc h d epend s upon th e ment al h abit T h ere are
some who naturally see wi d er an d farth er th an
o th ers A n d it is absol ut ely necessary th at th e
t rue observer shoul d cul tivat e wi th all assi duity
suc h a h abit wh en i t is not a nat ural possession
Th ere is a slov enliness in observa tion wh ich gives
t o th e inferences base d upon i t s resul t s a color of
ind eniten e ss an d inaccurac y , an d wh ic h proves a
fertile source of error
I t also oft en h appens, th at o wing t o th e min d
bein g prepossesse d by a c ert ain i dea or theory re
search will be th ereby res t ric t e d t o a limit e d re gion,
gestiv e

Gore , The A rt of S cientic Dis covery ,

32 1

26 5

FALLAC IES

an d nei ghborin g re gions be wh olly overlooke d


A n ope n ey e d vision in s pit e o f all preconceptions
or preju dic es is th e prim e requisi t e for securin g
from all quart ers th e g reat es t possible array of
fac t s that may in an y w ay t en d t o th e formation
of a clearer an d more a dequat e judgmen t
2 A secon d error of observat ion arises from an
opposi t e ment al h abit a failure properly t o coneen
t rat e th e att ention upon th e relevant fact s an d s o
t o discriminat e as t o exclu d e from consciousness fo r
th e t ime bein g a t leas t all irrelevant det ail s T h e
lack of su c h a dis c riminatin g facul ty lea d s ei th er
t o error, or to th e d earth of all si g ni c ant result s
I t is necess ary t o avoi d ei th er ex t reme, so that th ere
may be a s w eepin g survey of all th e possible fac t s
relevant t o th e subj ec t un der investi g ation, c o m
bin ed a t th e same t ime w i th a concent ration of
att ention th at is th e prerequisit e o f a deep insi ght
int o th e i nner connections an d int errelations of
th ese fac t s T h ere mus t be a d epth as w ell as a
w i d eness o f vision
T h ere are also errors arisin g from a failure t o
not e si g ni cant d ifferences in p h enomena th at
present s trikin g surface resemblances Here th e
closest scrutiny is necessary Th e ol d er ch emis t s
coul d n o t d is tin gui s h pot as h from so d a ; bary t a an d
s t ront ia w ere formerly confound e d t o g ethe r, so also
pot ash an d c aesia T h ro u gh ou t th e wh ole realm of
scient i c research, it s h oul d be ever kept promi
n en tly in min d th at surface d ierence S may h i d e
essential resembl ances, an d th at surface resem
blances m ay h i d e essential differences
.

'

26 6

I NDUC TI VE L O GIC

may arise from appercept ive proj e ction


Here th e O bj ec t ive element s o f percep tion combine
with th e subj ective so th at th e complet e perception
may cont ain element s which d o no t correspond wi th
reali ty T h e min d th us proj ect s upon th e el d of
vision it s o wn colorin g We s ee oft en th at which
w e wis h t o see an d fail t o see th at wh ic h w e d o no t
Wh en palla dium w as ori ginally mad e
Wis h t o see
kno wn t o th e publi c Ch en ev ix procee d e d t o examine
it, prepossesse d with th e i d ea th at i t w as an alloy of
some two kno wn me t als T h is i d ea was so proj ec t e d
upon h is experiment s th at h e at las t came t o th e
conclusion th at i t was a compoun d o f platinum and
merc ury Ch enev ix was le d into an error of o bser
vation as w as aft erwar d s prove d by Dr Wollas t on
wh o himself h a d obt aine d palla dium fromth e solu
"
tion o f cru d e platina in aqua re g ia
Th is error of
observation was due to th e fac t th at h e approac h e d
th e experiment s wi th a xe d i d ea in h is min d as t o
wh at th ey s houl d prove ; an d bein g d et ermine d t o
see evi denc es of this in th e ph enomena, h e u n co n
s cio u sly rea d int o th e m th at wh ic h w as no t really
th ere
II E rro rs of Jud gment
T h ese errors occur in
th e int erpret ation o f th e d at a o f perception F or
th at which is observe d mus t be referre d t o it s
appropriat e pl ace in th e bo dy o f kno wle dg e re
g ar d e d as a syst em, in wh i ch, part mus t t t o part
Inac curat e reference resul t s in
an d par t t o wh ole
manifes t imperfections an d inc on gruities in th at
p art of th e syst em of kno wle dg e t o which th e ref
3

Errors

Go re , The Art of S cientic Dis cov ery

26 7

FA LLACI ES

erence h as been m a d e An d th e inferences b ase d


th ereupon are nat urally affect e d by th is fun da
men t al error o f j u dgm ent Th ese errors are as
follo w s :
1 E rrors d ue t o false associat ions
Here, wh ere
ar t i cial or super ci al associa tions are int erpre t e d
as th ou gh th ey w ere real caus al connections, th e
mis t ake may prove mos t serious Th e mos t fertile
Source o f suc h fallacies is th e w ron g int erpret ation
Of space an d t ime associat ions , re g ar din g mere con
ti g uity in space an d t ime as evi d ence o f causal
connec tion U n d er thi s h ea d may be classe d th e
fallacies , n o n cau sa p ro causa, an d p o st hoc ergo
P rosperi ty , fo r ins t ance, follo w in g th e
prop ter hoc
enact ment o f cert ain in dust rial o r t ariff measures ,
is oft en att ribut e d as th e eect of th e same merely
because th ey appear in s t rikin g sequence Ho w
ever, it m
ay be th at th e prosperi ty h as follo w e d in
Spi t e o f the law s an d not o n accoun t o f th em
2 E rrors o f ju dg ment due t o emot ional pertur
b atio n s
Wh en th e int ellec t is d e ect e d from i t s
t rue point in g by passion, o r prejud ice, o r super
or any s t ron g em o tion th e consequent
stitio n
ju d gmen t is th e resul t ant o f tw o forces rath er
th an th e expression of on e A s Bacon say s : T h e
h um an un d ers t an din g resembles no t a dry li ght, but
a d mi t s a tinc ture O f th e w ill , an d passions which
generat e th eir o wn sys t ems accor d in gly ; for man
al w ays believes more rea dily th at wh ich h e prefers ;
h is feelin g s imbue an d corrupt h is un d ers tan din g i n
innumerable an d sometimes impercep tible ways 1
.

'

'

1 Baco n

Novum Org anon, Bo ok I Apho rism " LI"


.

26 8

INDU C TI VE

OGI C

Th e necessi ty o f ju dgin g in a d ry li ght as far


as possible , is especially emph asize d in th e e th ical
positions of Ad am S mi th an d lat er o f Mr S i dgw ick
Ad am S mi th cont en d s th at one s duty mus t be
es timat e d from th e s t an dpoint o f an impartial
spec tat or an d critic
T h at is man mus t as i t
w ere s t ep ou t o f h imself leavin g feelin g beh ind,
and ju dg e of himself and of his duty from a
purely objective point of vie w S o also Mr S id g
w ick says th at one o f th e ch ie f difc ulties in th e
u tilit arian posi tion namely th e discrepancy be
tween th e e gois tic an d alt ruis t ic claims upon
o u r ac t ivi ties c anno t be h armonize d s at isfac t orily ,
wh en s t at ed as a problem of mere feelin g Here
a gain man mus t eliminat e feelin g an d ju dg e of
himself merely as o n e amon g many, wh ere each
c ou n ts for o n e an d n o o n e fo r more th an one
In
th e li ght of pure reason h e may be able t o see
th at th e goo d o f all is h is hi gh es t g oo d
But
wh en th at dry li ght is colore d by feelin g, suc h
ju dgment is impossible
F ara d ay, in his O bs erv ation s on M en tal E d ucatio n,
h as borne t es t imony d irec t ly to th e necessi t y of
eliminatin g feelin g from o u r ju dgment s He says :

T h e t end ency t o d eceive ourselves re g ardin g all


w e wis h for sh oul d be kept in min d, an d th e n eces
s ity also of resis t ance t o th ese d esires
T h e force
o f th e t emp t a tion w h ic h ur g es us t o seek for suc h
evi d ence an d appearances as are in favor o f our
d esires , an d t o disreg ar d th ose wh ich oppose th em,
is w
on d erfully g reat In th is respect w e are all
more or less ac tive promot ers of error
I will
,

26 9

FA LLA C IES

simply exp ress my s t ro n g beli ef th at that point of


self e duc at ion wh ic h consis t s in t e achin g th e min d
t o resis t its d esires and inclinations un til th ey are
prove d t o be ri ght, is th e mos t important of all,
no t onl y in thin g s of natural ph ilosoph y, bu t in
1
every d epart ment of d aily life
3 Errors of ju dgment due to the common frail
t ies of h uman nature S uch errors B ac on h as s tyle d
I d ol s His enumeration is no t only comple t e,
but is cl as sic in it s w ay, an d th erefore I quote i t at
thi s pl ac e : Four sp ecies o f i dols beset th e h um an
min d, t o wh ich, for dis tinction s s ak e, w e h ave as
si g ne d n ames , callin g th e rs t I d ols of th e Tribe,
th e secon d I dols of th e Den, th e th ir d I dols of th e
M ark et, th e fourth I d ols of th e Th eat re

T h e form ation o f no tions an d axioms on th e


foun dat ion of t rue in duc t ion is th e only ttin g
reme dy by wh ic h w e can w ard o ff an d expel th ese
i d ols I t is, h o w ever o f g reat service t o point th em
o ut ; for th e d oc t rine of i d ols be ars th e s ame rel a
tion to th e in terpre tation of n ature as that of the
confu tation of sop h isms d oes t o common lo gic Th e
i d ols of th e tribe are inh erent in hum an nature, and

th e very t ribe or race of m an ; for man s sense is


falsely as sert e d to be th e s t an dard of th in g s ; on
th e con t rary , all th e perceptions , bo th of th e senses
and th e min d, be ar reference to man and n o t t o th e
universe, and th e hu man min d is lik e th ose uneven
mirrors which imp art th eir o wn properties to dif
ferent o bj ec t s from wh i c h rays are emi tt e d, and
d is to rt and d isg ure th em
.

Gladstone , Mich ael Fara

day p
,

128

INDUC TIVE LO G I C

27 0

Th e i d ols of th e d en are th ose o f each ind ivi dual ;


for everybo dy (in a ddition t o th e errors common t o
th e race of man) has h is o wn in divi dual d en or
cavern wh i c h int ercept s an d corrupt s th e li ght of
nat ure, ei ther from h is ow n peculiar an d sin gul ar
disposition, or from h is e duc ation an d int ercourse
with o th ers, or fro m his rea di n g, an d th e au th ori ty
acquire d by th ose whom h e reverences an d ad mires ,
or from th e different impressions pro duce d on th e
min d as it h appens t o be preoccupie d and pre dis
pose d o r equable and t ranquil , an d th e like ; so
th at th e Spirit of man ( accor din g t o its se v eral d is
positions) is variable confuse d, an d as it w ere ac tu
at e d by Ch ance ; an d Heraclitus sai d w ell th at men
search fo r kno wle dg e in lesser w orl d s, an d not in
th e g reat er or common worl d

T h ere are also id ols forme d by th e reciproc al


int ercourse an d So c ie ty of man wi th man wh ich w e
call i dols of th e market from th e commerce an d
association of men with each o th er ; for men con
vers eby means o f lan g ua g e, but w or d s a re forme d
at th e will Of th e g enerality an d th ere arises from
a ba d an d unap t formation of w or d s a w on derful
obst ruction t o th e mind N o r can th e deni tions
an d explanations with wh ich learne d men are w ont
to guar d an d pro t e c t th emselves in some ins t ances ,
affor d a comple t e reme dy,
w or d s s t ill manifes tly
force th e un d ers t an din g thro w eve ry th in g int o con
fusion an d lea d mankin d int o vain an d innu mer
able cont roversies an d fallacies

Las t ly, th ere are i dols wh ic h h ave crep t int o


m en s min d s from th e various do gm as o f peculiar
'

F ALLA C IE S

27 1

sys t ems o f p hilosoph y, an d also from th e pervert e d


rules o f dem ons t ration, an d th ese w e d enominat e
i dols o f th e th eat re ; for w e re g ar d all th e sys t ems
o f phil o so ph y h i th ert o rec eive d or im ag ine d as s o
m any plays brou ght ou t an d performe d, creatin g
c tions an d th eat rical worl d s N o r d o w e allu d e
merely t o g eneral sys t ems but also t o many ele
ment s an d axioms o f sciences whi c h h ave become
inve t erat e by t rad ition, implici t cred ence, an d ne g

l ec t

A ll such t en d encies, as thus present e d by Bacon


clo g an d h amper th e normal func tionin g o f th e
j u dg ment T h e min d mus t be alert an d on guar d
t o eliminat e suc h fat al see d s of error
III E rr ors d u e to the Imaginatio n Here th e
imag ination supplies inner c onnections an d rela
tions , lyin g beyon d th e sph ere of observation in
or d er t o explain th e nat ure of th e observe d p h e
Th e d an g er h ere is th at th e
n o m en a th emselves
ele ment s supplie d in or d er t o make th e self con
sis t ent wh ole d o no t correspon d t o realit y
Th e
syst em re g ar d e d as a ment al cons t ruc tion may be
complet e in all of its c o or dinat e d part s an d never
U n d er th is
th el ess possess n o obj ec tive reali t y
h ea d fall all loosely cons t ruct e d hypoth eses In th e
framing o f an hypo th esis in g eneral, th e ima gina
tion func t ions very lar g ely I t is th e inner vision
th at represent s t o th e min d th e thin g s n ot seen
Moreover, th e ima gination is peculi arly liable t o
error an d t o s win g clear o f th e t rammels o f fac t,
an d in th e region o f pure fancy cons t ruc t Sys t ems
,

Bac o n, N ov u m Org ano n B oo k


,

I Apho isms " "" I "


.

etc

27 2

INDU CTIV E LO GIC

th a t res t upon no soli d basi s of reality


T h ese
d an g ers in d et ail h ave been point e d out in th e

ch apt er on Hypoth esis


T h e mos t fertile source of error, h o wever, arises
from th at natural elation of min d upon th e di scov
ery even of sli ght con rming evi d ence o f th e t ruth
of th e assume d hypo th esis This enthusiasm is apt
t o ma g nify unduly an inad equat e veri cation, an d
t o res t satis e d in an hypoth esis th at is g round e d
upon an insufcient basis T hus S ince th e year
17 7 0 more th an forty d iscoveries of ne w element ary
subs t ances h ave been announce d t o th e w orl d by
enthusiastic experiment ers an d, in all cases, th eir
dis c overies h ave been prove d t o be absolut ely w orth
less F or ins t ance it w as cond ently announce d
th at To rbern Berg marm in 17 7 7 h ad ext ract e d fro m
diamond s wh at h e consi d ere d t o be a new earth,
and c alle d i t t erra nobilis
We dgwoo d in 17 90,
discovere d aus t ralia in san d obt aine d from th e
continent of th at name ; but Hat ch ett prove d i t
t o be merely a mixture of silica alumina, oxi d e o f
iron an d plumbag o In 1805 Ri cht er discovere d

n icc o l aniu m
; i t w as aft er w ar d s prove d to be a
mix ture of iron cobal t, nickel , an d arsenic T h ese
inst anc es are bu t a few o f th e many wh ich ch arac
terize th e h is t ory o f c h emi cal research , an d s tan d
as c onspicuous w i tnesses o f th e d an g er of divorcin g
fancy from fac t
T h e imagination, h o w ever, properly cons t raine d is
most po t ent in su gg es tin g possible causal relations
in cons t ructin g hypo th eses, in d evisin g me th o d s o f
experiment in ord er t o verify th em , and in formin g
.

FA LLA

C I ES

273

universal concept s in which all th e particulars o f


observation must coinh ere
Davy an d F ar ad ay
w ere both conspicuous in th is men tal ch arac t eristic
A n d t o th is source th eir eminent discoveries may be
t race d Dr Wh ew ell says , for ins t ance o f F ara

d ay
In d iscoverin g th e nature of vol t ai c act ion,
th e essent ial int ellec tual requisit e w as t o h ave a d is
tinc t concept ion o f th at wh ich F ara day expresse d
by th e remarkable ph rase, An axis o f p ower hav
in g equ al an d opp o site fo rces
A n d th e d is tinc tness
o f th is i d ea in F ara d ay s min d s h ines for th in every
part of his w ri tin g s He appears t o possess th e
i d ea of th is kin d of force w ith th e s ame eminent
dist inc t ness wi th wh ic h A rchime d es in th e ancient
an d S t ev inus in th e mo d ern his t ory o f scienc e po s
sesse d th e i d ea o f pressure, an d w ere thus able t o
foun d th e i d ea of mec h anics A n d wh en F arad ay
canno t obt ain th ese di s tinct mo d es o f con c ept ion, h e
is d issat is e d an d conscious o f d efec t 1
I t is i n d ee d a t ouch o f g enius th at enables o ne t o
g rasp an d formulat e a cent ral i d ea th at will unify
an d also universali z e a lar g e bo dy o f seemin gly dis
connect e d an d incong ruous fact s But such an i d ea
mus t be th e expression o f th e relations actually o b
t ainin g, an d no subj ective fancy proj ect e d upon th e
ph enomena th emselves h ow ever clever o r in g en
ious such an ima ginative creation may be If o ne
w ere aske d wh at is th e most efcient ins t rument
of scienti c research , th e ans w er mus t be Th e
Im ag ination
A nd if one w ere aske d wh at is th e
.

W h ewell His tory of

147

In

u c tiv e

S cien ces , V o l

III 3d
.

ed

27 4

INDU CTIVE LO GIC

most fertile source of error th e ans wer lik e wise

mus t be, Th e Ima g ination "


I t mus t also be
remembere d th at i t is no t sufcient merely th at an
hypoth esis Sh oul d be in h armony wi th th e fac t s in
th e case it mus t be prove d also th at th e fact s are
connec t e d wi th th e hypoth esis th rou gh necess ary
links
A n d it is w ell also t o bear this in min d wh en
ar guin g ag ains t a rival hypo th esis th at may h ave
been a dvanced by an opponent wh o has claime d for
it only th e possibility o f its vali di ty and wh o h as
not afrme d it s necessity I t is manifes tly unfair,
as w ell as fallacious t o d eny th e possibili t y o f th e
hypoth esis merely by in dicat in g cert ain uncertain
ties connect e d wi th es t ablish in g it To con t ra dic t
possibili ty one mus t prove th e hypoth esis im
possible Re gardin g such a conflic t betw een rival
hypo th eses U eberweg su gg es tively comment s as
follow s : In cases of th is kin d i t is one of th e
h ar des t of scienti c an d ethical p roblems t o give
fair play t o one s opponent
O u r o wn preju di c es
are sure t o in uence us Yet th e e ffec t of th e inu
ence of anoth er s st an dpoint wh en i t is reach e d is
of immense value in scienti c kno wle dg e P olemic
easily lead s t o exasperation ; i t is easy bo th t o
abuse it an d t o l et it alone because of dislike t o
th e con ict s which i t pro duces ; but i t is di fcul t
t o reco gnize i t, an d u s e it in th e ri ght sense as th e
ne c essary form which th e labor of inves ti g ation
al w ays t akes
Man never att ains to a scienti c
kno wle dg e of th e t ruth wi th out a ri ghtly conduc t e d
battle of scienti cally just i able hypoth eses, th e one
,

FALLACIE S

27 5

a g ainst th e o th er : th e scien ti c gui d ance of th is


b attle is th e true d ialectic metho d 1
I V E rrors of the Concep tu al Pro cesses
T his
cl as s o f errors arises in th e form ation of g eneral
concep t s an d th eir expression in univers al law s
The n at ural t en d ency of th e min d t o g enerali z e
oft en lead s t o ill consi d ere d resul t s Th e universal
uni t es m any dierenc es int o an i d entity, an d th e
min d will re ad ily mini mize th e d ifferences in or d er
to form a d esire d univers al ; thus d isp ar at e attri
bute s may be incorrec tly c o or dinat e d in on e an d
th e same sy stem
Hersch el h as remark e d th at
h as ty g enerali z ation is th e bane of science A n d
Bacon h as s ai d our in t ellec t s w ant no t win g s, bu t
rath er w ei ght s of lea d t o mo derat e th eir course
T h e me th o d o f ag reement, wh en relie d upon t o
th e exclusion o f furth er experiment al d e t ermina
ti on is a fert il e source o f error in this respe ct T h e
possibility of a plurality of causes s h oul d be ever
kept prominen tly in mind O ne re ad ily assi gns an
effec t to a causal element whi c h is onl y part ial ly
it s cause ; th e consequent g enerali z ation is, th ere
F or ins t ance, it oft en h appens th at
fOre, incorrec t
ac tiv i ties of youn g animals are d escribe d as in stin c
tive an d con g enit al ; an d univers al proposi tions are
foun d e d th ereupon A n d yet i t m ay be th at th e
ac ti vi ties referre d solely to ins tinc t are due par
In or d er t o avoi d this error
tially to imit at ion
and eliminat e th e facto r of imi tation, inves ti g ato rs
in th is line are accus t ome d t o s tu d y th e ac t i v i ties
of anim als h at ch e d in incub at ors an d purposely
.

wg

r
Uebe

A S y stem of Log ic , etc

509

27 6

I NDU CTIVE

LO

G IC

kept from all o f th eir kin d T h is illus t ration will


serve t o s ho w th e prec autions th at mus t be taken
in ord er t o eliminat e all possible error from th e
data wh ich th e process of g enerali z ation cons t ruct s
into universal forms S o also inaccuracies in any
o f th e o th er in duc t ive me th o d s lea d t o g ross errors
in th e consequent g enerali z at ions base d upon th em
U n d er this h ead also are th e fallacies resul tin g
from th e ext ension of empirical law s t o sph eres
beyon d th e experience wh ich th ey embo dy an d
express T h is sourc e of error is espec ially ill us
trated in law s expressin g some quanti t at ive relat io n
between ant ec ed ent and c onsequent ; i t is a nat ural
supposi tion in such cases and yet a very mislea d
in g one oft entimes th at a simple proportional rel a
tion w ill exis t betw een ph enomena of th e same
but wi th g reat er or lesser magnitu d e as th e case
may be Bacon g ives th e followin g illus t rations of
this fallacy : S uppose a lea d en ball of a pound
w ei ght let fall from a s t eeple reach es th e earth in
t en second s will a ball of two pound s wh ere th e
po w er of natural mo tion, as th ey call i t s h oul d be
double reach it in ve ? N o th ey will fall almos t
in equal tim es and n o t be a cc elerat e d ac c ordin g t o
quantity
S uppose a d rac h m of sul ph ur woul d
liquefy h alf a poun d o f s t eel w ill th erefore an
ounce o f sulphur liquefy four poun d s of st eel ? I t
does no t follo w ; for th e s tubbornness of th e matt er
in th e patient is more increase d by quanti ty th an
th e ac tivity o f th e ag ent Besi d es t oo much as
w ell as to o little may frus t rat e th e e e ct, thus,
in s mel t in g and re nin g of met als i t is a c ommon
.

27 7

FALLA IES

error t o increase th e h eat of th e furnace or th e


quanti ty o f th e ux ; bu t if th ese excee d a due
proportion th ey preju dice th e O perat ion bec ause
by th eir force an d corrosiveness th ey turn much of
th e pure met al int o fumes an d carry it o ff wh ence
th ere ensues no t only a loss in th e metal but th e
remaining mass becomes m ore slu ggis h an d intract
able Men s h oul d, th erefore, remember h o w ZEso p s
h ouse wife w as d eceive d wh o expec t e d th at by
d oublin g h er fee d h er h en sh oul d lay two e gg s a
d ay ; but th e h en g re w fat an d lai d none I t is
absolut ely un safe t o rely upon any natural e xperi
ment before proof be ma d e of it, bo th in a less an d
1
a lar g er quant ity
Ano th er fallacy o f th e same ord er oft en occurs in
th e inference concernin g th e int erpolat e d element s
o f a series w h ose su cc essive v alues h ave n o t all been
observe d T h e inferen c e ex t en d s th e nature of th e
kno w n to th e un kno wn part s , an d presum es th at
th e in t erme d iat e links betw een actually observe d
part s of th e series are in acc or d ance with th e
g eneral nature of th e latt er S uch inferences very
oft en give correc t res ul t s as in th e plottin g of a
curve some salient point s may be d e t ermine d ac
cordi n g t o observe d quanti t ativ e variations an d th e
remainin g portions supplie d as upon th e above
supposi tion T h is ext ension t o cover int erme diat e
an d unobserve d ins t ances may h o w ever be some
t imes very fallacious F or a force may be assume d
t o be such th at i t s eec ts increase s t ea dily an d i t
,

B aco n , A dv ancement of Learning , B o o k

190

V Ch apter
.

II

27 8

INDU CTIVE L OG I C

may be th at th ey operat e perio di cally ; int erpol a


tion upon one assume d basis wh en th e o th er is th e
t rue o ne, w oul d of course in t ro duce g r ave errors
To eliminat e such errors , d evices h ave in many
cases been resort ed t o by which a self reg is t erin g
apparatus will record all successive values of th e
ph enomena under inves ti g ation
U n d er th e fallacies o f h as ty g enerali z at ion nat
urally fall all provincialisms wh ic h aris e from a
narro w nature an d habi t of min d Th e loc al t ra di
tions an d supers t itions th e prevailin g win d s, th e
social cust oms and m anners, are taken as types of a
univers al experience T h e inferent ial wi d enin g of
th e circle of a li mit e d experience is al ways prov o
cative of false inf erence an d misleadi n g res ul t s
We h ave also false analo g ies wh ich consis t in th e
ext ension of o ur experienc e of cert ain ph enomena
th at w e h ave observe d t o be al ik e in some respec t s
t o inclu de o th er att ributes not observe d, concern
in g which w e assume a c orrespon din g simil ari ty ;
th e abuse of nal c auses m ay be re gard e d as a
special c ase o f false analo gy Moreover a t end ency
t o consi d er causat ion in th e li ght exc lusively of nal
causes has oft en ret ar d e d th e advance of science, in
w ithdrawing th e att ent ion an d energies of th e ih
v e s tigato r fro m a se arc h aft er physic al causes , as, for
inst ance amon g th e ancient s i t w as d ecl are d th at
th e leaves of th e t rees are to d efen d th e fruit from
th e sun and win d Res t in g s at is e d in such an
explanat ion, th e precise function of th e leaves in
th e economy o f th e plant s gro wth w as not furth er
invest igat e d, an d thus pro g ress was impossible
.

FALLA

CIES

27 9

Again, incorrect cl assi c at ion is a source of error


In groupin g t o g eth er d isparat e p h enomena, w e h ave
a basis for for min g a g eneric concep t th at w ill in
o lu d e incompat ible spe c ies
in oth er w or d s , a
, or
Im iv ersal th at w ill h ave evi d ent except ions
More
over, if th e classi cation is partial , th e resul tin g
law s base d upon i t will h ave o nl y empirical force
I h ave en d eavore d in th is c h ap t er t o in dicat e
errors th at are mainly psych olo gical in th eir ori gin,
for tw o reasons In th e rs t place such errors
operat e as di s turbin g forces in th e mi d s t of purely
lo gical processes Th e d at a of inference are psy
c h o lo gic al as re g ar d s th eir source an d errors th us
ori ginatin g aec t th e inference base d upon th em
appeari ng in th e nal resul t as lo g ical fallacies
A n error of observation becomes an error in th e
ju dg ment th at is base d upon th e ori g inal percep
tion an d per dures in th e hypo th esis classi c ation
et c , fou nd ed on th at ju dgm ent an d nally emer g es
in th e con clusions base d upon th ese processes In
th e secon d place th e fallacies th at are purely
formal an d in th e s t ric t sense lo g ic al are no t as
apt t o d eceive an d mislea d th e mi n d
In th e
mat erial d at a especially lurk th e g erms of fallacy
O n th e th eory th at it is wiser an d also more lo g ical
to s t amp ou t an error in i t s incipiency I h ave
place d special emph asis upon th e various psych o
lo g i c al processes as initi al sources Of error More
over it is more rational to d eal w ith errors of pro
A m ach ine th at
ce s s rath er th an aw s o f pro d uc t
t urns out i mperfec t articles coul d h ave i t s imperfec
tions rec ti e d by repairin g eac h arti c le thus pro
.

,
.

,
i

280

INDUCTI VE

LO

G IC

or th e machine i t self coul d be re adjus t e d


so as t o produce th e ar ti c les w i th out aw I t is
nee d less t o say whi ch meth o d is th e more lo gical,
an d mos t sat isfact ory as w ell
Th e d esi deratum is accurat e and compreh ensive
observat ion ; a discriminat in g ju dgment forme d in

th e dry li ght of reason ; an imag ination th at h as


d eep insi ght into th e h eart of surface appearances ;
an d po w ers o f g eneralization wh ich t ranscen d o h
serve d ph enomena by a d equately int erpretin g them
d uced ;

C H APTER

" VI II

IN D U C T I VE ME TH O D S A s A PPL IED
V A R I O U S S CIE N CE S

THE

THE

TO

nat ure O f e ach sep arat e science will d et er


mine cert ain peculiarities o f m eth o d fo r th at s ci
ence ; an d its peculiar meth o d will be lar g ely a
matt er o f g ro wth, as experience accre dit s or dis
cre di t s th e various resul t s which it s operation may
att ain I t will thus be correc t e d or supple ment e d
accor din g as i t s t and s th e t es t of ach ieve d resul t s
T h ere are h o w ever some g eneral features an d e s
pecially some nat ural limi t ations of th e in duc tive
meth o d s th at may be properly in dicat e d
I In th e rs t place, th e nature of th e meth o d
use d, an d it s efciency, as measure d by i t s resul t s
wil l be foun d t o vary as th e nature o f th e ph e
S ome ph enomena a dmit of
n o m en a th emselves
analysis an d furth er d e termination by experiment
Ins t ea d of att emp ting to d e t ermine th e relation of
a c omplex ant ece d ent to a complex consequent th e
ant ece d ent is rs t separat e d int o its componen t
part s , an d one element is t es t e d alone in ord er t o
d et ermine it s precise effect T h e relation can th eir
be expresse d betw een th e simple ant ece d ent an d
simple consequen t, as a causal connec tion ; an d it
THE

28 1

INDU C TIVE

28 2

LO

G IC

a dmit s moreover o f a quant it ative d et erminat ion as


w ell S uc h a meth o d of proce dure by analy tical ex
perim en t enables us t o rise to la w s h avin g universal
vali di ty This meth o d is ch arac t eris tic especially
of th e ph ysical sciences because th e ph enomena
rea dily a dmit of resolut ion int o component part s,
and th e isolation of one simple forc e so as t o d e
t ermine i t s t ot al effec t Th e physical forces are
mos t readily a dap t able t o experiment T h ey th ere
fore affor d th e wi d es t el d for th e application
O f th e experiment al meth o d of in ductive inquiry
Moreover w e may readily pre di ct th e resul t s of a
combination of simple forc es wh en w e kno w th e
law s governin g th eir component elemen t s Th e
in d u ctO d e du c t ive meth o d th erefore becomes e s
pecially efc ient in ext en din g th e d omain of th e
physical scienc es Here also , math ematical analy
sis an d calcul ation is mos t valuable as an ai d to
experiment al inves ti g ation an d in d e t erminin g
quanti t ative relations as necessi t at e d by th e math
e m atic al la w s t o wh ic h th e d at a g ath ere d in duc
tiv ely mus t conform
T h ere are h o wever sciences wh ich present ph e
n o m ena of such a hi gh d e g ree of complexi ty th a t
an analysis of a complex wh ole int o i t s separat e
part s or element s of force is impossible More
over th e ph enomena c annot be analyze d in this
respec t eith er th at a cert ain part o f th e c omplex
whole can be indicat e d as th e wh ole ant ece d ent ,
an d th e remainin g portion as th e ent ire couse
quent
Th e di fcul ty th erefore is tw ofol d ; it
is i mpossible t o separat e th e complex wh ole int o
,

DIDUCTIVE METHODS A ND THE S CE NCES

283

two

o th er complex wh oles , ant ece dent and couse


quent, an d st ill furth er impossible t o separat e such
antece d ent an d consequent even if th ey coul d be
d et ermined, int o th eir simples t componen t part s
T h e p h enomena present e d are h ere n o t in th e
form of a sequence so often as in th at of c oexis t
ence, as in th e s c iences of bo t any , z o olo gy an d th e
like Here th e meth o d s of analo gy an d clas s ic a
tion mus t be resort e d t o, and w e ob tain d escriptive
law s as th e resul t
T h e forces manifes t e d in th e processes o f vit al
g ro wth are especially d ifcul t to d et ermine by ex
perim ent ; for th ey not only op erat e t o pro duce
cert ain e ects , but per dur e in th e eec ts t o pro
duce cert ain o th er effec t s in a process of c o n
tin u ou s cons t ruc t ion
S eparat ion by mech ani c al
analysis m eans ins t ant c essat ion o f th e pro c ess
i t self Dissect ion means d eath Here th en, is a
natural lim i t ation Moreover, th e la w s of d evel
o pm en t are furth er mo d i e d by ex t ernal c h an g es ;
T h e resul t o f th e inner force an d th e out er in
u en c es w orkin g t o g e th er c ompli c at es th e prob
l ems t o such an ext ent th at th e pure in duc tive
m eth o d s are w ell ni gh impossible of application
Resort is th en h a d t o d et ermination by st ati s t ical
meth o d s Larg e g roups of plant s an d animals are
examine d fo r th e purpose of noting t endencies
di sclose d in th e a gg re g at e bu t h i dd en as re gar d s
th eir manifes t ation in th e in divi d ual Here o f
course classi cation is an ai d in disclosin g simi
laritie s an d differences th at may su gg es t h ypo th
eses t o explain cert ain dominant ch arac t erist ics
,

INDUCTI VE LO GI C

2 84

We m ay, moreover h ave merely perm anent ef


fee t s present e d in perception th e cause h avin g
cease d t o ac t lon g since T hus in g eolo gy w e h ave
fact s th at h ave bee n c ause d it is t rue bu t th e
causes can be discerne d only as manifest e d in
th e effect s an d th erefore, th ey can be d e t ermine d
only by th e me th o d of hypo th esis, wh ich may
lead t o veri cation or not as th e case may be
Ag ain cert ain sciences may su gg es t problems
wh ich concern th e explanation or si gni c ance
not of particular p h enomena within th e sph ere
of th at s c ie nc e but rath er th e int erpret ation of th e
whole bo dy of ph enomena whic h th e science in
ques tioncompreh en d s T h e problem is no t solve d,
therefore by any att emp t in th e way o f analysis
by experiment but rath er in th e way o f synth esis
th rou gh h ypoth esis ; th at is th e i d eal cons t ruc t ion
o f a wh ole wh ic h w ill unify an d account for all
fac t s or in o th er w ord s t o discern th e sys t em
wh ic h underlies an d c o or dinat es th e various par
tic u lar manifes t ations
T his is seen espec ially
illus t rat e d in th e problems wh ich g eolo gy and bi
olo gy present concernin g th e int erpret ation of th eir
respect ive ph enomena re gar d e d in th e li ght of
th eir to t ality A s t ronomy also present s a mass
of seemin gly c h ao tic ph enomena an d ye t th e aim
of th is science is t o reduce th e m all t o some one
self c onsis t ent sys t em
F or ins tance Mr S pencer remarks concerni n g th e
g eolo gi s t : He d oes no t t ake for h is problem only
th ose irre gularities of th e earth s crus t th at are
w orke d by d enu dation ; or only th ose wh ic h i gneous
,

INDU C TIVE METH ODS

A ND THE

285

S CIE NCES

ac tion causes He d oes not seek simply t o un d er


s t an d h o w se d imentary st rat a w ere forme d ; or h ow
faul t s w ere pro duce d ; o r h o w moraines ori g inat e d ;
or h o w th e be d s of Alpine lakes w ere scoope d o u t
But takin g int o account all a g encies c o operatin g
in en dless an d ever varyin g combinations h e aims
t o int erpret th e entire s t ructure o f th e earth s c rus t
If h e s t udies separately th e action of rain rivers
gl aciers , iceber g s t i d es , w aves volcanoes earth
quakes etc h e does so th at h e may be bett er
able t o c ompreh en d th eir j oint ac tions as fac t ors
in g eolo gical ph enomena ; th e obj ect of his science
bein g t o g eneralize th ese ph enomena in all th eir

involve d connections as part s of one wh ol e


Mr
S pencer also d escribes th e nature of biolo gy in
muc h th e same w ay In like mann er biolo gy
is th e elaborat ion of a complet e th eory of life
in each an d al l of its involve d manifest at ions If
di fferen t aspec t s of i t s p h enomena are inves ti
g at e d apart, if o ne observer busies h imself in
classin g or g anis m s ano th er in dissec tin g th em
anoth er in ascert ain in g th eir ch emical composi
t ions, anoth er in stu dyin g func tions ano th er in
tracin g law s of m o dic atio n, th ey are all c o n
unconsciously h elpin g t o w ork out
sc io u sly or
a solut ion of vit al p h eno m ena in th eir en tirety
bo th as d isplaye d by in divi dual org anisms an d
1
by or g ani sms at lar g e
Mr S pencer makes th e di s t inc tion be tween in
v estig atio n of par ti cular causal relat ions on th e
one h and, and , on th e o th er th e int erpret ation of
.

S p enc er, Classic atio n of the S c iences ,

pp

19 , 20

286

INDU C TIVE

LO

GI C

th e t o t al ph enomena of a science as th e basis


of classication of th e sciences He divi d es th e
sciences int o th ose wh ich tr
eat of th e forms in
which ph enomena are kno wn su ch as th e A bs t rac t
S ciences of math ematics an d lo gic ( t e formal
lo gic ) ; an d th ose which t reat of th e ph enomena
th emselves o r th e C oncrete S ciences Th e latter
re divi d e int o th e Abs t rac t C oncret e S ciences so
called, because th e speci c element s of th e c on
cret e ph enomena are abs t rac t e d fro m th e p h e
n o m en a consi d ere d as a wh ole an d so d et ermine d
relat ive t o parti cular referenc es
as causal law s
w ithin th e wh ole bo dy o f ph enomena wh ich th e
science compreh en d s ; th ese are such sciences as
mech ani c s physics ch emis t ry et c : th e secon d is
th at of th e C onc ret e S c ienc es simply, wh ich re
g ar d ph enomena in th eir t ot ali ties as above d e
scribe d A difference O f meth o d is h ere in dicat e d
Th e Abs tract C oncret e S eries are t o be inves ti gat e d
analy tically th at is by experiment principally
with a vie w o f exhibi tin g c omplex ph enomena
in th eir simples t t erms T h e simple C oncret e
S ciences are t o be t reat e d synth e tically, th at is
by th e framin g o f an hypo th esis th at w ill com
preh en d all particular ph enomena in one c o or di
nat e d wh ole
Th e division of Herber t S pencer can only be
accep t e d in a g eneral w ay as in dicatin g pre
d ominant ch aract eris tics o f th e tw o kin d s of sci
It w ill not d o t o lay d ow n h ar d an d fast
e nce s
li nes h ere for ev ery scien c e present s two kin d s
d et ermine par ticular
o f problems ; th e rs t, to
.

INDU CT IVE

METHODS AND

THE

S C IENCES

287

causal relation s ; th e secon d, t o co or dinat e all suc h


rel ations int o a self consis t ent sys t em whic h will
un ify all separ at e an d in d ivi d ual ins t ances
F or
ins tan ce , t ake th e ph enomena o f li ght in p h ysics
P articular problems as re gar d s in t ensi t y v eloci t y
composi tion of li ght etc present th emselves ;
th en an un d erlying problem Ho w explain all
th e ph enomena of li ght upon some one single
basis re g ar din g th e essential nature o f li ght
Hence arose th e emission and un dulatory th eories
of li ght, an d al l ph enomena bearin g upon th e
th eory w ere mars h alle d in support of one an d
of th e o th er, unt il th e con ic t w as conclusively
d eci d e d
An d a g ain th e th eory o f li ght th e
th eory of elect rici ty , th e th eory of h eat et c
su gg es t s till ano th er problem Ho w unify all th e
s eparat e th eories in one all compre h ensive th eory
to wh ic h th e separat e p h enomena may be alike
referre d T hus every science present s particular
problems an d a g eneral problem as w ell An d
h erein lies a su gg estion th at all inves ti g at ors in
any branch of science woul d d o w ell t o bear in
min d S peciali z ation in any o n e line of particular
problems sh oul d al ways lea d to a consi d eration
of th e relations o f th ese particular problems t o th e
g eneral sys t em o f wh ic h th ey are part s S pecial
iz atio n th at d oes n o t th us supply its o wn corrective
by th e natu ral insist ence o f th e min d t o int erpret
th e particular in th e li ght of more g eneral law s
t en d s t o narro wn ess o f min d and barren resul t s
II In reference to meth o d in th e s c iences it
mus t be observe d al so th at in certain ph enomena
-

INDUCT IVE

288

LO

G IC

th e simple th eory wh ich re g ard s th e c ausal con


nec tion as a t ransfer of ener gy accor din g t o th e
doctrine of th e conservat ion o f ener gy , is f urth er
complicat e d by certain variations and mo di cations
of th e ener gy in th e process o f t ransference Wh en,
for ins t ance, a billiard ball s t rikes ano th er an d th e
secon d ball by virt ue of th e impact, rec eives th e
ener gy o f th e ini tial movin g ball t ransferre d t o i t,
th e problem is simpli e d by th e fac t th at th e mo t ion
of th e rs t is easily t raceable in th e second bein g
a t ransfer of energy whic h manifes t s it self in th e
same manner in th e tw o cases Ho w ever th e prob
lem is c omplicat e d at onc e wh en in ch emis t ry for
ins t anc e th e tw o combinin g element s form a th ir d
in whi ch th e c h arac t eristi c features o f th e former
are wh olly los t in th e new form Here is likewi se
a transfer o f ener gy wh ich may h ave mech anical
equivalent s i t is t rue an d yet s o ra dical a ch an g e o f
form ac c o m panies th e t ransfer th at it complicat es
th e problems which arise in this science We h ave
seen h o w th e combine d ind u c to de duct i v e meth o d
oft en pre dict s event s an d the nature of ph enomena
no t yet observe d An d yet thi s becomes mos t dif
cu lt wh enever t ransfer of ener gy is ac c om panie d
by a c h an g e in th e peculiari ties o f its manifes t ation
as well Kno wle dg e of th e nature of two element s
an d all th eir separat e ch arac t eris t i c s will n o t be
s ufc ient d at a for any pre di c t ion as to th e nature
of th e compoun d T hus c h emist ry confront s a
natural d if cul ty as re g ard s meth o d, which d oes no t
affect ph ysical scienc e g enerally
Another d ifc ul ty appears in psyc h ology , for h ere
,

'

INDU CTIVE ME T HO DS

AND THE

2 89

S C IENCES

s timuli from th e out er w orl d, expresse d in t erms of


ph ysic al ener gy an d quan ti t atively d et ermine d, pro
d uc e psych ical reac t ions, th at cannot be expresse d
in p hysic al t erms A n d, on th e oth er h an d, proc
esses o f i d eat ion pro duce muscular activi ties th a t
may be es t imat e d in physic al t erms I t h as been
ur g e d th at h ere th e th eory of conser v ation of energy
breaks d o wn, th at th e t ransferre d energy is wh olly
account e d for by th e nerve an d brain mo di cations ,
an d th at th e psyc h ical accompaniment s are wh olly
unaccount e d for upon th is basis Th ey st an d out
as unexplore d remain d ers
T his obj ec tion is met in two w ays
O n e is
th at th e physical an d psych ical are as i t were,
tw o close d circles an d while simultaneous in th eir
f unct ionin g d o not mutually int eract
T h is is
th e th eory o f th e so c alle d psych ophysical paral
I t necessi tat es me t aphysic al explanations
l elis m
an d pos tulat es th at seem t o complicat e rath er th an
Simplify th e difcul ties T h e secon d is th e more
reasonable th at psych ical ac t ivity may be radically
different fro m p hysical and yet th e two capable of
reac tin g upon eac h o th er, so as t o liberat e th e
po t enti al ac tivi ties in eith er sph ere, and thus ini
tiate a series of causally connec t e d p h eno mena
Suc h a th eory is butt resse d by subs t antial analo gies
in th e p hysical sph ere i t self ; namely th at in m any
ph enomena th e imping in g forc e is s o mo di e d by
th e reac t ion due t o th e nature of th e subs t ance ac t e d
upon, as t o lose to all observation at leas t i t s ori g
inal ch arac t eris tic feat ures Fo r ins t ance fric tion
p asses over int o elect ricity because of th e nature of
.

INDU CT IVE LO GIC

290

subs t anc e th at is rubbe d ; thun d er sours cream ,


an d th us soun d vibrat ions cause effec t s wh olly
in c on g ruous t o th em
T h ese illus trat ions mi ght be mul tiplie d th rou gh
all th e realm of physical s c ienc e
Th ey are so
many as to prepare u s for reali z in g th e possibility
at leas t
th at physical excit ations may produce
psychical ph enomena in th e sense th at th e ou ter
s ti mulus c alls int o ac tivi ty psychical ener g ies th at
thus s t irre d m an ifes t th emsel v es a c cor d in g t o th e
forms of th eir o wn nature rath er th an th e forms
of th e physical ph enomen a exci ting th em U pon
such a th eory we may procee d by observat ion an d
experiment to measure d uration int ensity et c , of
mental reactions respon din g t o ext ernal s timuli
A s re gard s th e meth o d h ere employe d th e series is
consi d ere d as one an d c omplete so th at physical
ex c i tations are trace d, as it w ere th rou gh an u n
broken causal ch ain t o th eir psy c h ical effec t s an d
v ice v ers a
O n th e th eory o f two close d c irc les i t
is di fcul t t o in dicat e a lo gical meth o d of e xpe ri
ment al inquiry unless it be furth er postulat e d that
act ivities in th e o ne accor din g t o its kin d may
in du c e mo dications of th e oth er accordin g t o its
kin d T his reservat ion is g enerally insis t e d upon
III I t sometimes h appens th at th e ph enomena
of o ne s c ienc e are t o be int erpret ed in th e li ght of
th e resul t s of anoth er scien c e T hus th e law s of
one science become g ui din g principles in inves ti
gatin g th e causal rel ations exis tin g in anoth er
S ph ere T his can only be done wh en there is some
similarity betw een th e ph enomena of th e two sci
th e

INDU CTIV E M THO DS


E

29 1

SCIENCES

AND THE

T his meth o d is especially illus t rat e d in h is


to ric al explanation
T h e problem present s a mass
of event s th at mus t be co
or dinat e d in a sys t em
wh erein th eir several causal relations w ill be e x
hibited
A n d no t merely must d etach e d epoch s be
prove d caus ally int errelat e d as re g ar d s th e event s
o c cur rin g in th em but h ere, also th e special prob
lems g ive rise t o a g eneral problem , to discover in
th e wh ole th e p h ilosophy of his t ory an d to d et er
mine th e several his t orical t endenc ies in one syst em
wh ose c h arac t eris tic features w ill reveal th e fac t th at

th rou gh th e ag es o n e in creasin g purpose run s


To solve th e special an d th e g eneral problems of
hi s t ory, recourse is h a d t o an analysis o f event s on
th e basis of w ell es t ablish e d psych olo gic al resul t s
T h e p h enomena of his t ory are subs t antially th e
activi ties of man, bo th in his in divi d ual an d col
lec tiv e capaci t ies
Event s bein g given, an hypo th e
sis concernin g th e motives , an d en ds wh ic h ac tuat e d
th em is frame d upon th e supposition th at men ordi
n aril y are impelle d by similar mo t ives u n d er similar
in or d er t o ac hieve similar en d s
c ircums t an c es
Here th e analo gies drawn betw een men of th e
present an d men of th e pas t or between men mov
in g in th e ordinary rout ine of e v ery day life and
men wh ose ac t s may be epoch makin g furnis h a
bas is for h is t orical int erpret ation We say th at
a series of event s perh aps o f a ve ry complicat e d
nature can be explaine d only by an h ypoth esis th at
a w ell d en ed pu rpose an d a s t ron g d et ermine d will
w ere fas hion in g th em an d movin g th rou gh th em t o
an en d th at was in th e c hief act or s min d from th e
enc e s

INDUCT IVE LO GIC

29 2

be ginnin g An d so th e rise of social h abit s, cus


toms , t raditions , law s, th e reli gion, th e g overnment,
an d national ins titutions of a peo ple h ave an ori gin in
psy c h ical an d no t ph ysic al element s a d eeper un d er
s t an din g o f wh ose nature an d all th at it necessi t at es
t en d s t o a c learer eluci d at ion of th e probl ems th erein
present e d T h e kno wle dg e of man, th e m icrocosm
is a g ui din g th rea d ami d th e be wil d ering mazes of
th e macrocosm I t is possible o f course, for th e
imag ination of th e h is torian to lea d him to w and er
far a el d an d invent fanc iful mo tives , purposes ,
publi c policies, etc t o explain g iven event s Ho w
ever h ere as in th e physical an d o th er sciences, th e
hypo th eses frame d mus t meet th e g eneral require
ment s an d con ditions of a vali d hypo th esis
IV T h ere h as been a g ro win g t en d ency in s c iences
re gar de d as solely or lar g ely de duct ive to c orrec t
and supplement th e t ra di tional meth o d and resul t s
by a more search in g in duc tive inquiry T his is
espe cially t rue of political economy T h e d e du c tive
meth o d procee d e d to buil d up a bo dy of doct rine
compose d of inferences nec essitat e d by a fe w fun
T h e premises w ere suc h as
d am ental premises
th e follo win g : T h e princ ipal mo t ive of act ion
is self int eres t ; th e earth as man s g reat s upply
h ouse is limit e d in ext ent an d pro ductivi ty ; th e
physi c al an d psyc h olo gi c al t end encies o f man lea d
him to multiply his own spec ies w ith a rapi d
ity whi c h if not c ount erac t e d by obs tac les woul d
brin g about an unlimit e d increase o f population
All ec onomi c law s w ere thus d educe d from some
The t e
such fun d ament al p ro p osi t ions as th e se
.

INDUCT I VE
sults

METHO DS AND THE

S CIENCES

of

29 3

th is d e ductive meth o d, h o w ever h av e


been brou ght t o th e bar o f anoth er meth o d for
search in g examinat ion an d ju dicial sent ence In
1848 Hil d ebran d an d K nies in 1853 w i th Ro sch er
in 1854 se t forth th e principles o f th e his t orical
sch ool of politi c al ec onomy
T h ey h el d th at an
inductive inquiry mus t be s tart e d in ord er t o
es timat e th e physical, eth nical an d h is t orical c o n
di tio n s of a nat ion an d its s t a g e of c ivilizat ion
T h ese forces correc tly assesse d, will give th e eco
nomic condi tions of a par ticular perio d of his tory
o r of a p ar t icular na tion
T his is not th e place t o
crit icise th e t ene t s of th is sch ool , but merely t o
point o ut th e fac t th at its in uenc e h as been pot ent
in correc tin g an d supplementin g th e result s o b
tain ed in a purely d e d uc t ive manner
De duct ion may g ive th e j oint effec t of universal
psyc h olo gic al impulses , operative un d er cert ain nat
ural con d itions of environm ent, e tc , provi d e d no
d is turbin g force is present But th e ques t ion h ere
is no t wh eth er a cert ain cause, if ac tin g alone w ill
pro duc e a cert ain effec t ; but wh eth er count erac ting
c auses w ill be presen t, o r mo d ifyin g c auses as th e
c ase may be To es timat e th e resul t s of c ollocations
an d no t S im ple causes becomes, as w e h ave seen a
complex problem F or it s solut ion recourse mus t
be h a d lar g ely t o s t at ist ic al meth o d s wh ereby larg e
a gg re g at es re v eal t en d encies th at are ac tual an d not
th eoretical merely
In a similar way th e hist orical s ch ool of juris
pru d ence asso c iat e d with S avi gny h as inuence d
th e soc alle d p hilosop hical sch ool in d emons t ratin g
,

29 4

INDU CTIVE

LO

GI C

th at resul t s th eoretic ally d et ermine d by d e duction


are const ant ly mo di e d by th e real c on ditions an d
limit ations o f each particular nation s life T h e
in uence o f th is sch ool is in dicat e d by a Si gni cant
fact th at wh en He g el w ro t e h is th eory o f law
(Rechtslehre) h e pai d more re g ar d t o th e his t oric al
formation o f s t at es th an di d th e earlier th eoris t s
of nat u ral law "
Ag ain ano th er illus t ration of th e growin g prev
al en c y of in d uc t ive me th o d is foun d in th e mo d
ern psyc h olo gical meth o d T h e sole meth o d was
c onsi d ere d from t ime immemorial to be th at of
int rospect ion I t s resul t s h ow ever, were mea gre ;
th e meth o d it self w as in d eni t e an d l ac ke d c er
t ainty an d uniformity In duc tive inquiry, th ere
fore, pro c ee d ed by it s o wn metho d s t o sec ure an d
int erpret mat erial in o th er an d various el d s A S
P rofessor La dd says : T h e meth o d of ps ych o lo g
i c al science is peculiarly int rospec tive an d analytic
of th e envisag e d ph enom ena calle d st at es of co n
Bu t i t is far broad er an d more effec t ive
s c iou sn es s
th an it coul d be if it w ere merely int rospect ive
It pus h es its analysis of th e g enesis of th e ph e
momena as far bac k as possible by th e use of
experiment al meth o d s an d meth o d s of ext ernal
observation applie d t o th e wh ole pro c ess o f ment al
evolution ( s tu dy of infant s of primit ive man an d
of th e lo wer animals
evolutio nary an d c ompar a
tive psych olo gy) It int erpret s th e psych olo g ical
life of th e in divi dual min d in th e li ght o f knowl
e dg e gath ere d concerning th e psych ical d evelopment

Blu ntschli, Th e Th eo ry of

th e S tate,

69

INDUC T IVE

METH ODS A ND THE

S C IENCES

29 5

of th e race (th e psych olo gic al s tu dy o f lit erature,


socie ty, art, reli gion
I t lays peculiar e mph a
S is upon abnormal an d path olo gical p h enomena of
th e ner v ous an d ment al life (psych iat ry hypno tism,
p h enomena of insani ty an d of th e c riminal classes ,
I t t akes acco unt o f th e rise an d fall of par
ticul ar forms of psyc h olo g ical th eory (th e h is t ory
of psyc h olo gy ) I t s t rives t o t ranscen d experien c e
by hypoth et ical principl es of explanation But in
th e employment of al l th ese meth o d s th is science
differs in no i m port an t respect from th e sciences
whic h d e al wh olly with ph ysical ph enomena I t
is only th e use of introspec tion for th e possession,
an d , t o some ex t en t at le as t for th e analysis, o f
its obj ec t s wh ic h makes psyc h olo gy as respect s
1
its me th o d d ifferen t from th e o th er sciences
In th e above w e see th at in duc tive inquiry lays all
possible el d s of researc h un d er t ribut e t o th e one
en d of explainin g an d correla tin g psyc hical ph enom
ena T h e sys t ems o f ethics also wh ic h are foun d e d
upon an a p rio ri basis, are becomin g more in d ebt e d
to empirical inves t i g at ions wh i c h h ave g iven a ric h er
cont ent t o th e s t rictly formal ethic Advance d
psyc h olo gic al researc h, th e stu dy of race ch aract er
istic s, t ribal cus t oms , h abi t s la w reli gion, et c , th e
in dications of moral pro g ress all g ive mat erial
whic h if int erpret e d by ri ght h ypo th eses, will
th ro w li ght upon th e th eory of eth ical principles
re g ard e d merely from a S peculative point of vie w
We may conclu d e th erefore th at th e in d uctive
meth o d an d th e d e duc tive are n o t mutually exclu
,

dd

La

In tro

du

c tio n

to P hi lo sop h y,

116

INDU CTIVE

29 6

S ive pro c esses

O GI C

T h ey m ay be so combine d as m u
Wh at Blunt schli
tu ally t o s t ren gth en o n e anoth er
says of jurispru d enc e may be applie d equally as
well to all sciences th at claim some a p rio ri basis :
Th e ol d s t rife be tw een th e philosop h ical and
his t ori c al sc h ools in G ermany h as alt o g e th er cease d
P eace w as ma d e as early as 1840
S ince th en it is
reco gnize d on all si d es th at th e experiences an d
ph enomena of his t ory mus t be illumine d wi th th e
li ght of i deas an d th at spec ulation is ch il dis h if it
d oes not consi d er th e real con ditions of th e nation s
life 1
It w ill be seen h o w impor t an t a fac t or h is t orical
d at a becomes in all th e scienc es that d eal wi th
human voli tion an d acti v ities Wh at ever hypo th
esis may be frame d, i t must correspon d to th ese d ata
because th ey represent ac tual c on d itions th at mus t
be co or dinat e d in a self consist ent sys t em an d th eir
nature an d relations satisfact orily int erpre t e d
.

B l untsc hli, Th e Theory of th e S tate, p 7 0


.

C H APTER
HI ST O R I CA L S K ET CH

" I"
OF

I N D U CT I O N

W e n d th e be gi nnin g s
of in ductive inquiry in th e S ocratic or maieutic
meth o d th at art of ment al mi dwifery by wh ich
concep tions w ere t o be delivere d from th e mass o f
in divi dual experiences an d opinions in wh ich th ey
lie conceale d T h e S ocratic proce dure in th e form a
tion of concept ions is t o ques tion every particu lar
vie w an d es timat e it by brin gin g t o g eth er analo g ous
cases an d discoverin g th eir nat ural connection s so
as to expli c at e th e g eneral no tion whic h i t cont ains
an d thus procee d from comparison of par ticulars t o
th e framin g o f g en eral propositions S ocrat es gen
e ralizatio n s w ere many of th em h as t y an d in h is
d esire t o formul at e a g eneral c onception h e over
l ooke d excep t ions an d minimize d d i fcul t ies bu t in
his me th o d th ere w ere th e g erms o f t ruly scien ti c
proce d ure Th e sp h ere o f his meth o d w as h o w ever
limi t e d, as h e applie d it only t o th e illumination o f
e thic al cont roversies
Plat o enrich e d th e S ocrat ic
Pla to (42 7 3 4 7
meth o d of in duct ion by removin g its limit ation t o
ethical inquiry
Plat o w as especially concer n
ed

w ith inves ti g atin g th e relations of h is i d eas t o


eac h o th er, and this le d t o th e appreh ension of th e
S o crates (47 0399

29 7

298

INDU C TIVE

O GIC

lo gical relations betw een conceptions, especially as


re g ar d s th eir subord ination an d co or dination This
forms a basis fo r c lassi cation, Plat o s di v ision o f
class concep t s or lo gical g enera i nt o th eir species is a
prominent feature o f his meth o d He also su ggest s
th e hypoth etical meth o d of t reatin g th e relations of
concept s ; namely t o examine a t entatively propose d
c onc ep tion by d evelopin g all th e possib l e co n s e
qu en ces th at w oul d follo w from its union w i th kno wn
concep tions T his is in keepin g w ith th e ind u cto
d e duc tive meth o d of Mill an d th e mo d ern lo gici ans
A ri sto tle (3 8 4 3 22
A ris t otle s name is es
pecially an d i t may be sai d almos t exclusively,
and syllo g is tic
associat e d w i th d e duc tive lo gic
reasonin g
Al th ou gh h e di d no t d evelop fully
th e in ductive lo g ic, h e neverth eless d i d not i gnore
it, in some of it s essential feat ures at leas t He
ackno wle dge d th e necessity o f inves ti gatin g th e
s t arting poin t of d e duct ion, n amely, th e ul timat e
groun d s of proof an d of th e principles of explana
t ion
T his process h e calle d di alectic I t is a
d ouble process th at procee d s from th e particulars
given in per c ep t ion, an d from th e i d eas current in
cus t omary opinion to discover th e g eneral , an d
th en from th e g eneral t o d e duce th e particular
which is th ereby veri e d in th e pro c ess
Th e
former proce dure is th e reverse of th e de ductive
an d is epago gi c or induc tive In duct ion, accordin g
t o h im is a syllo gism in wh ic h th e inference th at
th e maj or belon g s t o th e mi ddle is me diat ed
th ro ugh th e minor directly ; an d not in direc tly
th rou gh th e mi ddle Thus, to u s e A ris t otle s illus
-

HIS T ORIC AL SK E T CH
tratio n

T IO N

29 9

o r I DUC

th e inves ti g ation of th e conn ec tion be tw een


th e absence o f g all in animals an d lon g evi ty in a
n umber o f ins t ances , as in man h orse m ule , e tc ,
m ay disclose th eir coexis t ence 1 T h ey are th en
unit e d direc tly wi th out me di ation o f a mi ddle t er m
If w e h ad g iven th e universal propo s i tion to s t art
w i th , Wh at ever animal h as no g all is long li v e d,
an d th e minor premise th at man h orse mule et c
are animals h aving no g all th en th e con clusion
woul d follow th erefore th ey are lon g live d T his
is th e d e duc t ive syllo g ism
T h e in duc t ive me tho d,
on th e oth er h an d s t art s from particular observa
t ion th at th e h orse wh ich has no g all is lo n g lived,
so also th e mule, so also man et c ; th erefore
w i th ou t any m i ddle t erm , a coexis t ence is t aken
as equivalen t t o a causal relation be tw een th ese
att ribut es , an d th e inference is d ra w n th at all ani
S uc h an infer
m als h avin g no g all are lon g live d
ence is vali d syllo gis tically , h o wever only on th e
assu m p tion th at th e ins t ances examine d comprise
th e W h ole c lass h avin g th e att ribut es un d er inves t i
T his in ductive syllo gism , th erefore, ex
g atio n
presses inferences only o f complet e enumeration
T h e form o f suc h a syllo gism is as follo w s :

minor t erm
F
maj or t erm
mi dd le t erm
M
T h is, th at an d th e o th er S is P
T his th at an d th e o th er S is all M
A ll M is P
Let

Aris to tl e , P rior A nalytics , II xxiii


.

INDUCT IVE LO G I C

3 00

Here i t w ill be observe d th at th e p articul ar in


s t ances comprisin g th e minor t erm S wh en summe d
T h ere is no inference
u p, equal th e mi ddle t erm
in this if w e h ave re g ar d t o th e s t rict sense in
wh ic h th e wor d i s use d A ris totle, ind ee d, co nsid
ere d th e only scienti c in duc tion t o be th e so calle d
perfect ind uc tion an d says th at t o g eneralize many
experiences of th e same kind is ad missible only
wh en th ere is no contrary case T h e th ou ght th at
causal connect ion enables us t o g enerali z e is s t at e d
by A ris t o tle but as Ue berweg says , it d oes not
att ain t o a fun dament al Si gni cance in his lo g ical
th eory 1
The P recu rs ors of Baco n
T h e revol t a g ains t
th e sc h olas t icism of th e Mi ddle Ag es an d th e
fe tt ers o f th e A rist o telian lo g ic was many voice d,
culminat in g, h ow ever as re g ard s th e emp h asis
plac e d upon ind uc tion as a s cienti c meth o d, in th e
w orks of F ran c is Bacon
F oremos t amon g th e early ch ampions of in duc
t ive inquiry w e n d Ro g er Bacon born in 1214
a F rancis c an monk, yet d evot e d h eart an d mind
t o th e cause of science His Op u s Maj u s was
pu blis h e d rs t in 173 3 by Dr S J ebb principally
from a manuscript in th e library o f Trinity C olle g e,
Dublin
Th is w ork is ch aracterize d by a spirit of
prot es t ag ains t auth ority in g eneral an d th at of
A ris t otle an d his lo gic especially
He rec o m
men d s math ematics an d experimen t as th e two
g reat ins t rument s o f scienti c inves ti g ation
In
this p articular i t is int eres ting t o not e his antici
,

Ueb erweg , Log ic ,

479

SK E T C H

HI S TO RI C AL

or

INDU CT ION

301

patiou of th e mo d ern math ematicophysical mo des


of scient i c inquiry T h e follo win g quot at ion wil l
g ive an in dicat ion of his spirit and aims :

E xperimen tal science, th e sole mis t ress o f spec


ul ativ e scienc es , h as th ree g reat prero g atives amon g
o th er part s of kno wle dg e : Firs t S h e t es t s by ex
periment th e nobles t conclusions of all o th er s ci
e nc es ; nex t, S h e d iscovers , respec t in g th e no tions
wh ic h o th er sciences d eal w i th ma gni cent t ruth s
t o wh ic h th ese sciences of th emselves c an by no
means att ain ; h er th ir d di g ni ty is th at sh e by h er
ow n po w er an d w i th ou t respec t o f o th er sciences,
1
inves t i g at es th e sec ret s of nature
Leonar d o c o m
L eona rd o d a Vinci ( 14 52
bin e d in one personalit y many brilliant t al ent s , be
ing eminen t as sculp t or, paint er arc h i t e c t en g ineer,
ast ronomer, an d natural ph ilosoph er His w orks
un publis h e d exis t in manuscrip t s in th e library of
th e Ins ti tut e at P aris He expresses h imself very
clearly an d emp h atic ally c onc ernin g th e relation of
experience t o S peculation : Th eory is th e g eneral ;
experim ent s are th e sol diers We mus t consul t ex
perien ce an d v ary th e circums t anc es t ill w e h ave
d rawn from th em g eneral rules ; for it is sh e wh o
fu rnis h es t rue rules Bu t of wh at u se you ask are
th ese rules
I reply th at th ey d irect us in th e
research es o f nature an d th e operations o f art
Th ey prevent our imposin g upon ourselves an d
o th ers, by promisin g ourselves resul t s wh ich w e
can no t obt ain Bu t see th e absur dity of men "
.

Wh ewell ,

333

P hilos op hy

t he I n u ctiv e S ciences , Vol

II

3 02

INDU C TIVE

O GIC

Th ey turn up th eir noses at a man wh o prefers t o


le arn from nature h erself rath er th an from au th ors
wh o are only h er clerks 1 T his latt er remark is
similar in its reference to th e epith et o f Galileo
applie d t o men wh ose know le dg e comes wh olly
from books an d no t from O bservation ; namely

paper philosoph ers


His w ork , en
Bern ardina s Telesius (1508
tit le d D e Rerum N atu ra anticipat e d in some d e gree
a t leas t, th e N ovu m Organo n of Bacon
Bacon
h imself says of him : We th ink w ell concernin g
Telesiu s an d ackno w le dg e h im as a lover of t ru th ,
a useful cont ribut or t o science an amen d er o f
some t enets , th e rst o f recent men Tele siu s
but in
s et for h imself a h i gh aim an d purpose
th e application of h is me th o d h e w as no t so fortu
nat e bein g d ominat e d in h is researc h es by specu
lat io n rath er th an th e resul t s o f experiment al
inquiry A s t o his professe d meth o d h e announc es
in th e t it le of h is De N atura th at th e cons t ruct ion
of th e w orl d th e ma gnit ud e an d nature of th e
bo dies cont aine d in it are n o t to be inves ti gat e d by
reasonin g wh ic h w as done by th e ancient s , but are
t o be appreh ende d by th e sen ses an d collec t e d from
th e th in g s th emselves
An d in th e P roem of th e
same w ork h e says in th e same st rain that th ey
wh o before u s h ave inquire d concerning th e c on
s t ruct ion o f th is worl d, an d of th e th in g s wh ich it
con tains seem in d ee d to h ave prosecut e d th eir ex
amin at ion w i th pro t rac t e d vi g ils an d g reat labor
.

Wh ewell

369

P h ilosop hy

the I n uc tive S ciences , Vol

II

H IS T ORICAL S K E T C H

INDU CT IO N

or

3 03

but never t o h ave looke d at it F or, as i t w ere,


att emp t in g to riv al Go d in w is d om and ven turin g
t o seek for th e principles an d causes o f th e w orl d
by th e li ght of th eir o w n reason an d thinkin g th ey
h ad foun d wh a t th ey h ad only inven te d, th ey m ad e
an arbi t rary w orl d of th eir o wn
We th en, no t
relyin g on ours elves an d of a d uller int ell ec t th an
th ey, propose t o ou rselves to turn our re gard s t o

th e w orl d i t self an d its p art s


F oll o w in g Tel es iu s , an d of hi s sch ool, w as T h om as
C amp anell a ( 1568
He w as a cont emporary
of B acon, and, un d er th e in uence o f Tele sius early
conceive d th e i d ea o f an in duc tive me th o d o f re
search At th e a g e o f tw enty
tw o h e publis h e d a
w ork wh ose c harac t er may be ju dg e d by its title,
T h om as C ampanella s Philosoph y d emons t rat e d
t o th e senses, ag ains t th ose wh o h ave ph ilosophi z e d
in an arbi t r ary and d o g matic al manner, no t t ak in g
nature for th eir gui de ; in whic h th e errors o f Aris
to tle and his foll o w ers are refut e d from th eir o wn
as sert ions an d th e la w s o f na tur e ; an d al l th e
imaginat ions feig ne d in th e place of nature by th e
P eripat etics are al t og e th er rej ec t e d ; wi th a t rue
d efence of Bern ar di n Tel esiu s of C osen za, th e
g reat es t of p hil osop h ers ; conrme d by th e opin
ions o f th e ancient s, h ere eluci d at e d an d d efen d e d,

especi al ly th ose of th e P lat onis t s


T h e i d e as of Bac o n wi th th eir impetu s t o th e
in ductive meth o d of research w ere not only antici
p ate d by w ri ters of books ; but actual di scoveries
by z ealous inves ti g at ors w ere turnin g th e att ention
of th e think in g w orl d t o nature and h er secre t s
.

INDU CTIVE

3 04

LO

G IC

Th ere was an il lus t rious line of pioneers in th is


undiscovere d count ry T h ere w as A n d rew Cze sal
pinus (1520
th e found er of th e s c ience of
ad
bot any ; an d earlier C operni cus (14 7 3
v an c ing h is h eliocent ri c th eory ; an d Gilber t ( 15 40
th e court physician o f Eli z abeth an d J ames,
con ductin g wi th untirin g perseveranc e his inves ti
g atio n s of th e nat ure of magne tism an d elec t rici ty
Kepler, born t en years aft er Bacon 157 1 an d
Galileo born in 15 64, an d th eir cont emporary,
Ty ch o Brah e born in 1546 forme d a t riumvirat e
of scienti c po w er an d brillianc y ma d e resplendent
by th e g lory of th e h eavens i t self It must be
remembere d, t oo th at at this time a ne w w orl d h ad
been disc overe d ac ross th e seas ; th e recent in v en
tions of gunpo wd er of th e mariner s compass, an d
of th e art of print in g all t en d e d to s timulat e th e
th ou ght of th e worl d an d ush er in a ne w epoch in
th e hist ory o f civilization
Bac on s in duc tive
Fran cis Baco n (15 6 1
syst em is given, for th e mos t part in th e N ov um
T h e title o f th is w ork was in it self a
Org ano n
pro t es t a g ains t A ris t o tle an d his lo gi c , implyin g
th at A ris to tle s Organo n w as no w ou t of d at e an d
w as t o be superse d e d by th e ne w Bac on insis t s th at
all kno wle dg e of nature h as for i t s en d th e d is cl o s
in g of th e causes o f th ing s A ccor din g to th e A ris
tote lian sc h eme causes are formal mat erial efc ient
or nal Bacon is only concerne d wi th th e formal
c auses
F or h e says all event s h ave th eir groun d
in th e forms o f th ing s
By th e form of a
thin g, h e meant its essent ial nature Wh ere h e
.

HI S TO RICAL S K E TC H

INDUCTION

or

3 05

uses th e form w e may w ell supply th e w or d


law To discover th e forms o f ph enomena, i t is
necessary, accordin g t o Bacon t o collec t as many in
s tances as possible in wh ic h th e ph enomenon un der
inves t i g ation appears ; t o g eth er th ey form a ta bul a
In like m anner, th e inst ances in which
p rces entiae
th e p h enomenon is l ac kin g are g roupe d in a tabula
a bs entiw ;
an d a third group must be formed a
tabu la g ra d u a m in wh ic h th e variat io n s of int ensi ty
in th e p h enomena are compare d wi th th e varyin g
inte nsi ty o f o th er p h en omena T h e problem is
th en t o be solve d by a process of exclusion (exc lu
s o ; th a t is th e rej ec t ion or exclusion o f th e several
)
qual i t ies wh i c h are no t foun d in some inst ance
wh ere th e given qu al ity is present or are foun d in
some in st anc e wh ere th e given quali ty is absent
o r are foun d t o increase in some ins t an c e w h ere th e
given quali t y d ecreases, or t o d ecre as e wh en th e
By thi s process an in dica
g iven quali ty increases
tion will be given by wh ich an h ypoth esis may be
frame d an d nal ly verie d by subsequent observa
In th e ske t ch o f this meth o d
tion an d experimen t
i t wi ll be seen th at h is th ree t ables o f inst ances
closely resemble th e me th o d s of a g reement of
di fference an d of concomi t ant variations T h ey
h o wever lack th e precision of th e lat er formula
tion of th ese meth o d s Th ere is no h int at a syst e
matic selec tion an d variation of th e inst ances ; an d
no requirement as in th e me th o d of d ie rence,
th at tw o ins t ances s h all be so experiment ally d e
term in ed th at th ey w ill a gree in every poin t save
th e g iven ph enomenon, wh ic h is p resent in the on e
.

INDU C TI VE

30 6

LO

GIC

an d absent from th e o th er Bacon h o w ever ma de


a subs tantial c ontribution t o th e meth o d o f in du c
tion in g eneral in insis ting upon th e examination
of ins t ances th emselves an d in ascen di n g from
th em qui t e gradually th e scale of th e more g eneral
up to th e mos t g eneral an d in th is h e ent ere d a
vi g orous prot es t a g ains t h as ty g eneralizat ion
A s t o th e manner of certifyin g th e hypo th esis
forme d aft er th e pro c ess of c ollectin g an d sifting
ins tan c es Bacon h as no recou rse t o d e duc tion
base d upon th e hypo th esis an d consequent v erica
tion He seems to d espise math ematic al metho d
as an ally of in duc tive inquiry ; an d th erefore has
no place in his sch eme for th e pre dict ion o f n e w
ph enomena by means of c al c ulation O f his nine
divisions o f ai d s to in duction h e c omplet e d only
th e rst P rero g ative Ins t ances O f th e ins t ances
whi c h h e enumerat es tw en ty seven in all only a
fe w h ave any bearin g direc tly upon th e in duc tive
meth o d proper T wo set s of th ese inst ances may
be c onsi d ere d as a cru d e s t at ement o f th e meth o d s
of a greement an d difference ; th e S oli tary Inst ances ,
wh ic h eith er exhibit a ph enomenon wi th out any of
i t s usual ac c ompaniment s o r which a gree in every
thin g except some par ti c ular ph enomenon an d
Mi g rat ory Ins t ances , wh ere quali ties are pro du c e d
in bo dies by evi dent causes , as fo r ins t ance th e
pro ducin g of whi t eness by poun ding g lass also by
s t irrin g w at er int o fro th Th ese ins t ances h o w
ever as exhibi t e d by Bacon lack precision an d th e
possibili ties o f accurat e d et ermination of causal con
n ec tio n s
Th e only o th er g roup of ins t ances h av
,

H IS T O RI C AL S K ET C H

OF

INDUCT IO N

307

in g speci al in duct ive si gni cance is th at of th e In


stantia Crucis ; as before m en tione d suc h ins t ances
,
are v aluable in d eci din g be tw een rival h ypo th eses
Wi th all th e d e ciencies o f Bacon s metho d, h o w
ever, his service t o th e thinkin g w orl d is ind ispu
t able, in emph asizin g th e nee d of inves ti gatin g
ph enomena th emselves as a correct ive o f fanciful
spec ul ation s an d in h is vi gorou s warnin g s ag ains t
preju dice, a g ains t int ellec tual in dolence, ag ains t
subj ection of th e min d t o th e t rammels o f auth or
it y an d a g ains t over h as ty g eneralizations
Lo c ke applie d th e meth o d
L ocke ( 163 2
o f Bacon t o th e obj ec t s of inner experience
He
d ecl are d th at th e d at a of all know le dg e come from
s ens ation or sense percep t ion, an d from re ection,

an d th at th ere are no inn at e i d eas an d th ere


fore no s t artin g point for a p rio ri speculations
T h e meth o d th at h a d been foun d useful in actual
di scove ri e s such as th ose of N e wt on, Kepler, an d
oth ers, Lock e insis t e d w oul d prove pro ductive also
of ri ch res ul t s in th e int ellec tual S p h ere
S cientic me th o d
Is aac N ewto n (16 42 17 27
w as gra dually formulatin g i t self in th e ac tual pur
suit s o f s c ienti c inves ti g ation, no t th ou ght ou t
a s muc h as w orke d ou t, an d i t s e fciency t es t e d an d
N e wt on gives form t o th at
c on rme d by resul t s
w h ich w as a res ul t o f many experiment s an d o f a
m ass of various expe riences , in h is Rules o f Phi
l o so ph izing (Regu lae Philosop handi) pre xe d t o th e
.

P rin cip ia

T h ese rules are as follo w s :


1 Th e rs t rule is tw ofol d
.

INDUCT IVE LO GI C

308

real c auses are t o be admitt e d in ex

ph enomena
b
N o more c auses are t o be ad mi tt e d th an suc h

as sufce t o expl ain th e ph enomena


In as far as possible th e s ame c auses are to
2
be as si gne d for th e same kin d of natural effect s

3
Qu ali ties th at can neith er be increase d nor
diminis he d in int ensity , and th at obtain in all
bo dies accessible to experiment, mus t be consi d ere d
qual i ties o f al l bo dies wh at soever
4 In p h ilosophi c al e x
periment, propositions
collect e d from ph enomena by induc tion are t o
be h el d, notwith s tan din g cont rary hypoth eses, as
ei th er exac tly or approximat ely true, unt il oth er
ph enomen a occur wh ereby th ey are ei th er rend ere d

more ex ac t or are prove d liable t o exceptions


Hypoth eses non
N e wt on s celebrat e d sayin g,
n go was ori gin ally a prot es t ag ains t th e sup
position o f th e exi s t ence of occul t or imag inary
cau ses t o expl ain ph enomena, no tably th e C ar
te s ian explanat ion of th e celes t i al movemen t s by
vortices Hypoth eses of a different nature, an d
ration ally g roun d e d of course, did no t fall un d er
N ewtOns repreh ension
S ir John Herschel (17 92
Hersch el s D is
co urs e o h the S tu d y of N atu ra l Philosop hy w as pub
l ish ed in 18 32 J oh n S tuart Mill revie w e d th is book
in th e E xaminer and w as evi dently impresse d an d

in uence d by i t
Hersch el s d esi g n was t o mak e
th e meth o d s of science more explicit T h ese are
cont aine d in nine proposi ti ons readily applicable
to particul ar c ases, or rul es of philOS Ophizin g

Only
plan ation of
a

H IS T ORI CAL S K E TCH


Of

0 F IND

U C T IO N

309

th ese proposit ions th e second, seventh, ei ghth,


and ninth present subs t ant i al ly th e experiment al
me th o d s as aft er ward s more precisely formulat e d
by Mill T h ese meth o d s h o w ever h e re g ar d s sim
ply as mean s t o discovery, an d no t meth o d s o f
O f th e remainin g proposi tions th e rs t
proof
is a more precise s t at emen t o f Bacon s principle
o f exclusion, an d is th e foun d at ion of th e j oint
m eth o d of a greemen t an d d ifference
T h e th ir d
proposit ion is th at we are not t o deny th e exis t
ence of a caus e in favor of whi ch w e h ave a
un animous a greement of st rong analo gies, th ou gh
i t may not be app arent h o w suc h a cause can pro
duce th e c ect, or even thou gh i t m ay be di fcul t
t o conceive i t s exist ence un d er th e circums t ances
of th e case T h e four th is th at con t rary o r
opposin g fact s are equally inst ructive for th e dis

c ov e ry of causes w i th favorable ones


T he fth
recomm en d s th e t abul ation of fact s in th e or d er
o f int ensity in wh ic h some peculiar quali t y sub
sis t s T h e sixth rul e in sist s upon th e investi
g at or keepin g prominently in min d th e possibil i ty
th at count eractin g o r mo difyin g causes may sub

sis t unperceive d, and th at th is fac t may be th e


means of expl ainin g m any apparent except ions
Hersch el also emph asi z es th e necessity of com
binin g in duct ion an d d e duction in complic at e d in
qu irie s ; an d, fur th er h e explains th e nature o f
e mpirical la w s , as also th e nature an d t es t s o f
hypo th eses We can no w se e th at th e bo dy of in
d u ctiv e princ iples be gin s at len gth to assume n al
form and proportion
,

INDU CTI VE

3 10

O GI C

Dr

Whewell (17 9 5

Wh e w ell publis h e d
his P hilosop hy of the I nd u ctiv e S cien ces in 1840,
cont ainin g his sys t em of induction His me th o d in
volves tw o principal processes , th e colli g ation of
fact s and th e explicat ion of conceptions T h e inv e s
tigato r is to g ath er all th e fact s at his disposal , an d
upon th em h e is t o superin duce a concep tion which
will unify th em o r collig at e th em He says th ese
conceptions are supplie d by th e mind, while fact s are
supplie d by th e sense Th is h o w ever, is a dis tino
tion th at separat es so w i dely th e sph eres of th e
par ti c ular fact s , an d th e g eneral concep tions th at
upon suc h a basis a union of th e tw o as explainin g
one by th e oth er w oul d be arti cial an d with no
correspondin g bon d of reali ty
T h e colli gatin g
c oncep t ion d oes no t exis t in th e min d before or
apart from it s exi s t ence in fact T h e att empt t o
t fact s t o rea dy ma d e concep tions is of th e nature

Kepler s ninet een guesses reg ar ding


o f g uess w ork
planetary orbit s is an ins tance o f att emp tin g t o
superin duce c oncept ions upon a mass o f fact s It
is not a t ruly sci enti c o r lo gical proce dure an d
th e great d ang er of applyin g it lies in th e fac t
th at th e min d all to o readily t en d s t o mou l d fact s
int o th e forms o f prior conceptions

Th e Meth o d s employe d in th e F ormation o f

S cience, th e t itle o f his conclu din g ch apt er are


th ree, as follo w s : Meth o d s of O bser v ation Meth
o d s of O bt aining C lear I deas an d Meth o d s o f
In duction
T h e last principally conc erns our
present purposes T h e meth o d s o f in duction are
meth o d s of discovery rath er th an proof, save th e
.

H IST O R I CA L S K ET C H

o r INDUC

T ION

311

las t wh ich is one o f the experiment al m eth o d s


T h ey are th e me th o d of c urves t o express g raph i
cally th e g ra duat e d res u l t s of several observat ions ;
th e me th o d o f means , an d th e meth o d o f leas t
squares both d esi gne d t o eliminat e acci d ent al ao
co m pan ime nts of cons t an t causes by s t rikin g an
avera g e of several observations ; an d th e meth o d of
resi dues W h e w ell s meth o d may be c h arac t eri z e d
in brief as a meth o d of d is c overy rath er th an proof
Mill s L o ic
J o hn S tuart M ill (180 6
g
publish e d in 184 3 , w as essenti ally a me th o d o f
proof rath er th an a meth o d of discovery His aim
in form
ulat in g th e meth o d s in vo gue in experim en
t al science was t o discover th e precise mo d es of
th eir Operat ion in or d er t o apply th e same in in v e s
tigatin g th e various ment al , moral soci al , an d poli t
ic al p h enomena Bacon in th e N ov um Organo n h a d
as ser t e d th a t thi s in duc t ive m e th o d w as applicable
t o th e int ellectual an d moral sciences Th is was no
d oubt su gg es tive to Mill, as i t h a d been t o Locke
Wh at ely s Logic, publis h e d in 18 2 7 , in uence d Mil l,
an d w as th e means of turnin g h is att ention t o lo g i
cal s tu d ies Wh at ely s book w as review e d by Mill,
wh en onl y tw enty one, in th e Westminster Rev iew
T h e revival in lo g ical int eres t at th is t ime an d
th e d epartu re from sc h olas t ic trad i tions h ave been
t race d t o th e in uence of E dw ar d C oples t on, tut or
at O xfor d, an d aft er w ard s Bis h op of Llan d aff
Wh at ely s w ork represent e d th e rst frui t s , an d
Mil l s th e rich er an d riper pro duc t of this revival
I t is a matter of more th an passin g
o f lo gic
int eres t to not e th at o ne of W h at ely s mos t active
,

IND UCTIVE LO GIC

312

collaborat ors in th e w ork w as J oh n Henry N e wm an


so th at, as P rofessor Mint o says , th e common room
bf O riel , wh ic h Mr Frou d e d escribes as th e cent re
from whic h emanat e d th e Hi gh C hurc h Movement,
may also be sai d t o h ave been th e cent re from which
emanat e d th e movement that culminat e d in th e rev

o lu tio n of lo g ic
Mill s special office as re g ar d s in duction consist s
in his cryst allizing th e princ iples an d practices of th e
scienti c investi g at ors wh o h ad cau ght and re ect e d
th e S pirit o f mo dern research Th e form ul at e d meth
o d s of inductive lo gic , subst antially as given by Mil l,
h ave become th e reco gni z e d meth o d s o f all invest i
gatio n th at is ac t uat e d by a sc h olarly spiri t an d a
sc ienti c h abit Mill s cont ributions t o th e in duc
t ive lo g ic h ave been so lar g ely d rawn from and so
frequent ly referre d t o in th e composi t ion of this
book, as t o nee d n o furth er c omment h ere Th e
w orks o f th e more recent w ri t ers as Lotz e, S i g wart,
Bosanque t J evons , Venn etc h ave also been notice d
in th e bo dy of th e t ext Th eir w ork is larg ely crit
ical, an d no dis tinct in duct ive s yst em is especially
associat e d with any o f th eir n ames
,

C H A P T ER

""

LO G I CA L E " E R C I S E S

th e fol lo win g examples indicat e th e n ature


of th e inferences th e meth o d s employe d, an d th e
c h arac t er of th e resul t s
obtai ne d, wh e th er vali d
or invali d an d th e reasons for th e same
1 In all un h eal th y count ries th e g re at es t risk of
fever is run by sleepin g on s h ore I s th is o win g to
th e s t at e o f th e bo d y durin g sleep o r t o a g reat er
abun dance o f miasm a a t suc h times ? I t appears
cert ain th at th ose wh o s tay on boar d a vessel
th ou gh anc h ore d at only a sh ort dis t ance from
th e coast g enerally su ffer less th an th ose ac tually
on Sh ore DA RW I N in Vo yage of N aturalist
2 T h at th e perio d of th e ti d e s h oul d be acc id e n
t ally th e same as th at o f th e culmination o f th e
moon th at th e perio d o f th e hi gh es t tid e s h oul d he
ac ci d ental ly th e same as th at o f th e sy zy g ies is
possible in abs tracto ; but i t is in th e hi gh es t d e g ree
improbable ; th e far more probable assumption is ,
ei th er th at sun and moon pro duc e th e ti d e, o r th at
th eir mo tion is d ue t o th e same g round s as th e
mot ion of th e t i d e
3 In measurin g th e veloci t y o f soun d by e xperi
ment s con duc t e d at ni ght wi th cannon th e resul t s
IN

3 13

314

INDUCTIVE

GIC

LO

at one s tat ion w ere never found to ag ree exactly


with th ose at th e oth er Moreover, i t w as no tice d
th at on th e ni ght s wh en th e d iscord anc e w as g reat
e s t, a s t ron g w in d was blo w in g nearly from o n e
s t ation to th e o th er
M Mello n i, observin g th at th e maximum
4
point of h eat is t ransferre d farth er an d farth er
t o war d s the red en d of th e S pect rum , according
as th e subs t ance o f th e prism is more an d more
permeable t o h eat, inferre d that a prism o f rock
sal t wh ic h possesses a g reat er po w er of t ransmit
t in g th e calori c rays th an any kno wn bo dy ou ght
t o th ro w th e poin t o f g reate s t h eat t o a consi der
abl e dist ance beyon d th e visible p ar t o f th e spec
t rum ; an d his pre diction w as veri ed by subsequent
experiment
5 Durin g th e mi ddle of th e ei ght een th c ent ury
Bonnet an d S pallan z ani discovere d th at th e h orns,
t ails, le g s , eyes , or even h ea d o f some creat ures, if
o ut o ff, w oul d g ro w a g ain
T h e t ail an d le g s o f a
salaman d er w ere remove d and repro duce d th em
selves ei ght times in succession By means o f a
number of experiment s it h as been foun d th at th e
more simple th e s t ruct ure of an animal is, th e
more d o its several part s possess a po wer of in de
pend en t exist ence an d th at in th e more com plex
animals, th e d eran g ement o f o n e part much more
affect s th e act ion o f th e entire or g anism
6 P rofessor J evons h as observe d th at economic
crises h ave occurre d at re gul ar int ervals of a bout t en
years T his t en year perio dicity moreover seems t o
correspond t o a Similar perio di c i ty of ba d h arvest s ;
.

LO G I CAL

IS ES

315

E" ERC

an d th e cause o f thi s seems to be a d ecennial peri


o di c it
y in th e spot s on th e s u n
7 W h at is th e si g ni cance o f th e remark of
Ch evreul , th e F rench scientis t : Every fac t is an

abs t rac ti on
8 A l so of th e follow in g remark of M E spinas :
If h um an ac tivi ty was incompatible wi th th e ord er
o f thi n g s , th e ac t of boilin g an e gg w o ul d h ave t o be

re gar d e d as a miracle
9 I t h ad lon g been known th at gras s h oppers
an d cricke t s h ave on th eir an t erior le gs t wo pecu l
iar g l assy, g eneral ly more or less ov al , d r um l ike
s tr u ctures ; bu t th ese w ere suppose d by th e ol d er
ent omolog is ts t o se rve as res onat ors an d t o rein
force o r int ensify th e w ell known chi rpin g sound s
whic h th ey pro duce J oh ann es Mull er w as th e
rs t wh o su gg es te d th at th ese d rums or tympana
ac t like th e tympanum of o ur o wn c ars an d th at
th ey are real ly th e ex t ern al part s of a t rue au di
t ory app aratus T h at any ani mal s h oul d h ave its
ears in its le g s soun d s no d oubt, a p rio ri very
unlikely and h ence probably th e t rue fun c tion of
this Or g an w as s o lon g un suspec te d S I R J O HN
.

LU BBO CK

In S imple frac tu re of th e ribs, if th e lung


be puncture d by a fra gmen t th e bloo d effuse d into
th e pl eural cavity , th ou gh freely mixe d w ith air
un d er g oe s n o d ecomposi t ion
Wh y air int ro d uce d
int o th e pl eural cavi ty th rou gh a w oun d e d lun g
s h oul d h ave such wh olly dieren t effect s from th at
en t erin g dir e c tly th rou gh a w oun d in th e ch es t was
t o me a complet e my s t ery until I h eard of th e germ
10

I NDU CT IVE

3 16

LO

G IC

th eory of put refaction wh en i t at once occurre d t o


me th at i t w as only natural th at air s h oul d be l
tered of g erms by th e air passag es o ne o f wh ose
of ces is t o arres t inh ale d part icles of dus t an d
PR o
prevent th em from ent eri ng th e air cells
,

RE

SS O E

LI STE R

If th e lu n g s be emptie d as perfectly as po s
sible an d a h an dful of c o tt onwool be place d a g ains t
th e mouth an d nos t rils and you inh ale th rou gh it
i t will be foun d o n expirin g this air th rou gh a glass
tube th at it s free do m from oatin g matt er is mani
fes t T h e application o f this is obvious ; if a phy
s ician w is h es t o h ol d back from th e lun g s of h is
pat ient or from his o wn th e g erms , or virus by which
cont agious disease is propag at e d h e will employ a
cott on w ool respirat or PR O FE SS O R T Y N D A LL
12 In th e d esert of N orth A frica wh ere nei th er
t rees brus hw oo d, nor even un dulat ion of th e surface
afford th e sli ght es t prot ection t o i t s foes a m o di
cation of color in ani mals wh ic h sh all be assimilat ed
t o th at o f th e surroun din g c ount ry is absolut ely
necessary Hence wi th out exception th e upper
plumag e of every bir d, wh eth er lark ch at, sy lvian,
or san d g rouse, an d also th e fur o f all th e smaller
mamm al s an d th e skin of all snakes an d li z ar d s is
o f o n e uniform is abelline , o r san d color
W A LL A CE
13 Dar wi n in inves t i g at in g th e d i fference in
w ei ght between cros s an d self fer tilize d plant s,
foun d th at th e six nes t crosse d plant s avera g e d
ounces , whils t th e six nes t self fer tilize d
plant s avera g ed only
ounces or as 100 t o 22
14 Bees i ncessant ly visi t th e o w ers of th e com
11

'

LO GI C AL

E" ER CIS ES

3 17

mon broom and th ese are ad a p te d by a c urious m ec h


anis m for cross fertil izatio n
Wh en a bee li ghts
on th e W in g petal s o f a youn g o wer, i t i s S li ghtly
o pe ne d, and th e s h o rt s tamens sp rin g o ut, wh i c h
rub th eir po llen ag ain s t th e ab d omen of th e bee
If a rath er ol der o w er is vi sit e d for th e rs t time
( or if th e bee exert s great force on a youn g er
owe r) , the k eel O pens alon g its wh ole length an d
th e lon ge r as w ell as th e sh orte r s tamens, to g eth er
with th e much elon gate d curved pis til, sprin g forth
wi th v i ol ence Th e atte ne d spoon l ik e ex t remity
of the pistil res ts for a time on th e bac k of th e
bee , an d l eav es on i t th e lo ad of po ll en wi th wh ich
i t is char g ed As soon as the bee ies away, th e
pistil i ns tan tly curls roun d so th at th e s ti gmatic
surfac e is no w upturne d an d occupies a posi tion, in
w hic h i t w oul d be rubbe d again s t th e ab d omen of
T hus wh en
ano th er bee v isitin g th e s ame o w er
th e pis til r s t escapes from th e k eel, th e s ti g ma is
rubbe d again s t th e bac k of th e bee, dus te d wi th
poll en fro m th e lon ger s tamens , eith e r of th e same
or anoth er o w er ; and afterw ard s agains t th e lo w er
su rfac e of th e bee dus t ed with poll en from th e
s ho rt er s tam ens , which is of te n sh e d a d ay or two
befo re that h om th e lon g er s t amens If th e v isits
o f bees are prevent e d, an d if th e o w ers are not
d as h e d by th e win d ag ains t any obj ec t, th e k eel
never o pe ns , so that th e s tamens and pis til rem ain
enc l o sed Plants th us pro t ec te d yiel d very fe w
p od s in c omparis on wi th th ose pro duce d by nei gh
bo rin g un co vere d bus h es, and sometimes none at
.

all

DA RW IN

INDU C TI VE

3 18

O GIC

Baron Zach r eceive d a lett er from P ons a


successful n d er o f com et s, c omplainin g that for a
cert ain perio d h e h a d foun d n o c o m ets, th ou gh h e
h a d sou ght d ili g ently Zac h a man Of muc h S ly
humor, t ol d him th at n o Spo t s h a d been seen o n
th e sun fo r abou t th e same time
whic h w as t r ue
an d assure d him th at wh en th e spo t s came back
,
th e comet s w oul d come with th em S ome time
aft er th at h e g ot a lett e r from P ons wh o informe d
h im w ith g reat sat isfac tion th at h e w as qui t e
ri ght th at very lar g e spo t s h ad appeare d on th e
sun, an d th at h e h ad fo un d a ne com et sh ortly
DE M O R GAN S Bud get of P arad oxes
aft er
15

'

16

If Tellus win g e d be,


T h e earth a mo tion roun d ;
T h en much d eceive d are th ey
Wh o nere before i t foun d
S olomon w as th e w isest,
His wi t nere this att aine d ;
C ease, th en, C opernicus
T hy hypo th esis vain "
.

S Y LV AN U S

M O R GA N

16 52

Weath er F orecast er Dunn h as prepare d a


ch art s h ow in g th e number of d eath s from grip in
N ew Y ork C i ty durin g th e perio d from Marc h 22
to May 16, 189 1, es t ablis h in g th e relat ion betw een
th e d eath rat es an d w e ath er condi t ions durin g th e
g rip epi demic of th at year Mr Dunn h as ma de a
careful study O f recor d s of th e disease an d selec t e d
th e epi d emic o f 189 1 as bein g th e t ime wh en th e
g rip w as mos t pronounce d
17

'

L O G I C AL

RC ISES

3 19

E" E

He h as apparently d emons t rat e d th at th e w eath er


is an importan t fact or in th e mo rt ali ty o f g rip cases
He says that h umi di ty or mois ture in th e air seems
to be th e mos t import ant element in c ausin g th e
disease t o S pread Th ere i s a correspondin g
creas e of d eath s wi th in creas in g humi d ity
T h e fat al ity is mos t marke d wh en th e h um i dity
is at i t s maximum and th ere is a su dd en fall of th e
t emperature This is s h o wn by th e record of A pril
2 1, wh en th e d eath rat e from g rip was th e h i gh es t
e ver k now n
Durin g th e tw enty four h ours of th at
d ay 2 5 0 d e ath s w ere report e d
On A pril 1 an d
A pril 3 0 th e d eath rate w as also hi gh T h ese w ere
d ays follo w in g a su dd en fall in t emperat ure
Al l th rou gh th e epi d emi c th e c h art s Sh o w an
increasin g d eath rat e w ith h i gh or incre asin g h u
mid ity T h e h i gh er th e humi di ty and th e more
su dd en th e fall in t emperatu re, th e g reat er w as th e
nu mber of d e ath s Wh en th e temperature an d th e
h umi dity droppe d at th e same time, th ere w as a d e
creas e in th e d eath rat e, as Mr Dunn poin t s out by
several ex ampl es He says that th e lesson to be
learne d from his chart is th at th ose sufferin g from
an inci pient att ack of th e grip s h oul d be mos t
cautious Of th e col d, h umi d days th at imm ed iat ely
follo w th e w arm, damp ones
18 If in a reservoir immerse d in w ate r, th e air
be compres se d t o th e exte nt of t en atmosph eres,
an d suppos in g th at n o w wh en th e compresse d air
h as acqui re d th e t emperature of th e w at er, i t be
allo w e d t o ac t upon a pis t on loa d e d by a w ei ght,
th e wei ght is raise d At th e s ame t ime th e w at er
.

INDU C TIVE

3 20

LO

GIC

becomes cooler S h o win g th at a certain quan ti ty of


h eat h a d disappeare d in pro ducin g th e mech anical
effort of raisin g th e w ei ght
19 T h at th e feelin g of e ffort is lar g ely if not
entirely o f periph eral rath er th an c en t ral ori gin,
appears from such experiment s as th e follo win g
Hol d th e ng er as if t o pull th e t ri gg er of a pis
T hink vi g orously o f ben din g th e n g er, but
to l
d o not bend i t A n Im mistak able feelin g of effort
resul t s Repeat th e experiment, an d not ice th at
th e bre ath is involunt arily h el d, an d th at th ere are
t ensions in th e oth er muscles Repeat th e experi
ment a g ain t aking care to keep th e breath in g re gu
Li tt le or no
lar an d th e oth er muscles passive
feelin g of e ffor t will no w ac company th e im ag inary
ben din g of th e n g er FE RRIE R
2 0 A S t o th e natu re of pet ri e d s h ells Qu irin i
conceive d th a t as earthy particles uni t e d in th e sea
s o as t o form th e s h ells of Mollusca th e same crys
tallizin g process mi ght be eected on th e lan d ; an d
th at in th e latt er cas e th e g erms of th e anim al s
mi ght h ave been disseminat ed th rou gh th e sub
s t ance of th e rocks, an d aft erw ar d s develope d by
virtue of humi dity
2 1 Vol taire su gg es t e d th at th e marin e s h ells
foun d on th e t ops of moun t ains are East ern S pecies
droppe d from th e h at s o f pil g rims as th ey returne d
from th e Holy Lan d
22 T h e epicyclical th eory of th e h eavens was
con rme d by it s pre d ictin g eclipses of th e su n an d
moon, congurat ions of th e planet s , an d o th er celes
tial p h enomen a
,

L O G IC AL
23

R IS ES

3 21

E" E C

A rfv e d so n

discovere d lithia, by perceivin g


an exc ess of w ei ght in th e sulph at e pro duce d from
a small port ion of wh at h e consi d ere d as m a g nesia
present in a mineral h e h a d analy z e d
24 We see amon g th e nebul ae (wh ich are d if
fuse d al on g th e Milky Way) ins t ances o f all d e grees
o f con d ensation
from th e mos t loosely diffus e d
ui d , t o th at separation an d soli di cation of part s
by wh ich s uns an d s at elli t es an d plane t s are
forme d ; and thus w e h ave before u s ins tances of
sys t ems in all th eir s t ag es ; as in a fores t we see
t rees in every perio d of g ro wth LA PLA CE
25 I t h a d been d e duc t ively inferre d from th e
C opernican th eory th at th e planet s Venus an d Mer
cury ou ght t o pass th rou gh p h ases like th e moon,
an d th e t elescope reveale d this to be th e c ase
2 6 Werner says S ir C h arles Lyell , h a d n o t
t ravelle d t o dis tant count ries ; h e h ad merely ex
pl o red a small por t ion o f Germany, an d c onceive d
an d persua d e d o th ers t o believe, th at th e wh ol e su r
face o f our pl anet, an d all th e mount ain ch ains in
th e w orl d, w ere m ad e aft er th e mo d el of h is o wn
province
2 7 S ch einer w as a monk ; an d o n communic at in g
t o th e superior of his or d er th e account o f th e S pot s
on th e sun , receive d th e reply : I h ave search e d
th rou gh A rist ot le an d can n d no th in g of th e kin d
mentione d : be assure d th erefore th at it is a d ecep

t ion of your senses , or of your glasses


2 8 Wh en w e are t ol d th at a man h as become
d erang e d from anxiety or g rief, w e h ave learne d
very little if w e res t cont ent w ith th at How does
.

I NDU CT IVE

3 22

LO

G IC

it h appen th at ano th er m an subj ec t e d t o an ex


ad
ac tly similar c ause of g rief, d oes n o t g o m
M A U D S LEY
2 9 I t was a g eneral belief at S t Kil d a th at th e
arrival of a Sh ip gave all th e inh abit ant s col d s Dr
J o h n C ampbell t ook pains to as c ert ai n th e fac t an d
t o explain i t as th e effect o f efuv ia arising from
human bo dies ; it w as discovere d, h o w ever, th at th e
situation Of S t Kil d a ren d ers a north east w in d in
dispensably necessary before a s hip can make th e
landin g
gh tin g w as
3 0 C h rysippus main t aine d th at cock
th e nal cause of cocks th ese bird s bein g ma de by
P rovi d ence in or d er t o inspire u s by th e example of
th eir courag e
3 1 Touc h in succession various obj ect s on th e
t able A paper w ei ght if met alli c is usually col d
t o th e t ouc h books paper, an d especially a woollen
t abl ec over, comparatively warm Tes t th em by
m eans o f a th ermomet er an d th ere will be li ttle o r
no difference in th eir t emperatures Why th en do
some feel col d, o th ers w arm t o th e touc h ? T h e
sense o f t ou c h d oes n ot inform us d irec tly o f t em
peratu re but o f th e rat e at wh ich our ng er g ains o r
loses h eat A s a rul e bo dies in a room are c ol d er
th an th e h an d an d h eat al ways t end s t o pass from
a warmer t o a c ol d er bo dy Of a number of bo dies,
all equally col d er th an th e h an d th at o ne will seem
col d es t to th e tou c h as th e met allic wh ic h is able
mos t rapi dly t o c onvey aw ay h eat from th e h an d
,

'

TA IT

32

O ne

of

J oule s

experime nt s conc ernin g th e

LO

G I C AL

R ISES

3 23

E" E C

mech anical value of li ght is as follo w s : He com


pare d th e h eat evolve d in th e wire conductin g a g al
v an io current wh en th e w ire w as i gnit e d by th e
pas sa g e of th e current, w ith th at evolve d wh en w i th
an equal c urrent it w as kep t cool by imm ersion in
w at er T h ese experiment s s h o w e d a small but u n
mistak able d im inution of th e h eat wh en l i ght also
w as g iven o ut T A I T
33 I t is an illusion in psyc h olo gy an d a corrup
tion o f lo gic t o t ake th e con di tions whi c h occasion
th e lo g ical operations of th ou ght fo r th e operat ions
th emselves T h ere is onl y on e d elusion more d es
perat e still, t o imagine th at a complet e p hysical
th eory of th e nervous sys t em w ill explain th at
whi c h is i t self th e con di t ion o f any th eory bei ng
possible at al l L O TZE
34 Durin g th e re t re at of th e Ten T h ousan d a
cuttin g north win d ble w in th e faces of th e sol diers ;
sac ri ces w ere o ffered t o Boreas, an d th e severi ty of
th e w i nd imme di at ely cease d, whi c h seeme d a proof

of th e g o d s c ausation
3 5 I t h as been Sh o w n by observat ion th at over
driven catt le, if kille d before recovery from th eir
fat i gue, become ri gi d an d pu t refy in a surprisin g ly
Sh ort time A sim ilar fac t h as been observe d in th e
case o f anim al s h unt e d t o d eath, cocks kille d durin g
T h e cont rary
a ght an d sol d iers slain in ba tt le
is remarke d wh en th e musc ul ar exercise h as no t
been g reat o r excessive
s su s
3 6 A correc t an al ysis of lapis lazul i w a
peCted to be erroneous becau se th ere S eeme d t o be
no thi n g in th e element s as si gne d t o i t wh ich w ere
.

I NDU CT I VE

324

O GI C

Silica alumina so d a sulphur an d a t race of iron t o


,

accoun t fer th e brillian t blue c olor of th e s t one


3 7 A ccor din g t o th e th eory th at th e earth h as
b ut a th in crus t, i t is s till subs t ant ially a liqui d
globe, and th erefore und er th e att ractive inuence
o f th e sun and moon, i t ou ght t o be h ave like a yiel d
in g liqui d A ccor din g t o Hopkins, T h omson and
o th ers , th e earth in all its as t ronomical relations be
h aves like a ri gi d soli d
a soli d more ri gi d th an a
soli d globe of glass, an d th e difference betw een th e
beh avior o f a liqui d g lobe and a soli d globe coul d
easily be d et ect e d by as tronomical ph enomena
.

LE CO N TE

Many years a g o I w as st ruck wi th th e fac t


th at h umblebees, as a g eneral rule perforat e o w ers
only wh en th ese g ro w in lar g e numbers near t o g eth er
In a g ar d en wh ere th ere w ere some very larg e
be d s of S tachys coccinea an d of P ents temon argu tu s,
every sin gle o wer w as perforat e d ; but I found two
plant s of th e former species g ro win g quit e separat e
w ith th eir pet als much scrat ch e d, sh o win g th at th ey
h a d been frequently visi t e d by bees and yet not a
sin gle ow er was perforat e d I found also a sepa
rat e pl ant of the P ent st emon, an d saw bees ent erin g
th e mouth o f th e corolla an d not a sin gle o wer h ad
been perfor at e d In th e follo win g year ( 1842) I
visit e d th e same g ar d en several times : on th e 19 th
of J uly hu mblebees w ere suckin g th e o wers in th e
proper manner an d none o f th e corollas w ere per
fo rate d
On th e 7 th o f Au g us t all th e o w ers w ere
perforat e d even th ose o n some few plant s of th e
s alvia, whic h grew at a litt le dis tance from th e g re at
38

O G I C AL

C I S ES

3 25

E"E R

be d O n th e l t o f Au g ust only a few o w ers on


th e summit s of th e spike s of both s pecies remaine d
fres h an d not one of th ese was no w bore d Ag ain,
in my o wn g ard en every plant in several ro w s of th e
common bean h a d many o w ers perforat ed ; but I
fou n d three plant s in separat e part s of th e g ar den
wh ic h h a d sprun g up acci dent ally, an d th ese h a d not
a sin gle o wer perforat e d G eneral St rac h ey for
merly sa w many perforat e d o w ers in a g ar d en in
th e Himalaya an d h e wrot e to th e o wner t o inquire
W h e th er th is relat ion be tw een th e plant s g ro win g
cro wd e d an d th eir perforat ion by bees th ere h el d
g oo d, an d w as ans w ere d in th e afrmat ive Hence
it follo w s th at th e re d clover an d th e common bean
wh en cul t ivat e d in g reat mass es in el d s , E rica
tetralicc g ro w in g in lar g e numbers o n h eath s ro w s
o f th e scarle t ki dn ey bean in th e ki t c h en g ar d en,
an d masses o f any species in th e o w er g ar d en are
all eminent ly liable t o be perforat ed T h e explana
t ion o f th is is n ot d if cul t Flo w ers g rowin g in
lar g e num bers att ract cro wd s of insec t s T h ey are
thus s t imul at e d t o w ork quickly by rivalry Also
many o w ers h ave th eir nect aries dry wh ich is mos t
quickly discovere d by bit in g h oles in th em
CH A RLES D A RW I N
39 T h e seat o f sensat ion is in th e h eart, as it is in
th e cent re of th e bo dy ; th e brain is c old in or d er
th a t it may coun t eract th e h eat of th e h eart I n o r
d er t o t emper th e col d nes s of th e brain, bloo d is con
v eyed t o th e membrane wh ich envelops i t by means
But, les t th e h eat so conveye d
o f veins o r c h annels
sh o ul d inj ure th e brain, th e veins , instea d of being
.

INDUCTI VE

326

LO

GIC

lar g e an d fe w, are small an d many, an d th e bloo d


conveye d, ins t ea d o f bein g copious an d th ick , is
thin an d pure
ARI S T O T LE
4 0 T h e lun g s o f a fox mus t be a speci c for
asthm a because th at animal is remarkable for its
s t ron g po w ers o f respiration PA RI s Pharmaco
.

logia

41

G alileo discovere d by th e use of h is t elescope


,

th e fo ur small sat ell it e s which circulat e roun d J upi


t er It w as th en inferre d th at wh at h appene d o n
th e smaller sc ale mi ght also be fo un d true o f the
l ar g er planet ary sys t em
4 2 T h e r s t s t ep t o w ar d th e d iscovery o f p h o t o g
ra h
p y w as th e kno wle dg e th at visual light cause d
a c h em ical ch an g e in io di d e o f silver T h e secon d
s t ep w as t o x in permanent posit ion th e port ion
o f th e s ubs t ance c h an g e d by th e li ght, wh ile th e
unch an g e d port ion was remove d
F rom wh at is kno wn of th e ch emical element s
th eir compoun d s , it seems h i gh ly probable
an d
th at nume rous compoun d s may exis t wh ich are
s en s i t ive in th e same w ay to w aves of ent irely
di ere nt len gth s from th ose th at pro d uce vis ion
E ven W ith th e sal t s of s ilver it h as lon g been kno w n
th at th e rang e o f wave len gth s capable of pro ducin g
phot o g raphic effect is muc h great er th an th e visual
rang e ; an d th at th e wave le n gth s wh ich pro duce
th e maximum physiolo g ical effec t (li ght) are no t
th e same as th ose th at pro duce th e maximum ph ot o
graph ic effect
It h as been s h o w n by P rofessor S R L an gley
th at int glass is t ransparent to waves about four
.

OGI C AL

3 27

C IS ES

E"ER

t im es as lon g as th e lon g es t in th e visual ran g e ;


an d th at rock sal t is t ran s paren t t o a ran g e belo w
th e re d en d of th e v isible spec trum tw enty nine
t imes as lon g as th e ent ire visual ran g e G lass is
opaque t o very s h ort w aves it s limit in th at direc
t ion being nearly coinci d ent w ith the visual l imit
Quar t z on th e oth er h an d, is t ransparen t t o a ran g e
o f s h or t w aves ex t en d in g far beyon d th e visual
lim it, but is opaque to very s h ort w aves May not
th ese subs t ances prove valuable in th i s ne w el d
o f ac t ino g rap h y as quar t z t rain s h ave prove d in
p h o t o g raphi n g th e ul t ra violet s pec t rum 9
Sh oul d th e report o f th is discovery (Rontgen s)
be con rme d w e canno t fail t o accord th e hi gh es t
prais e to th is ne w t riump h o f science an d to pre
d ic t a d evelopment o f th e ne w el d of ac tino g rap hy
th at may prove of g reat er import ance th an ph o
-

to graph y

F rom th e analo gy betw een th is form o f ra dian t


ener g y an d d ark h eat it might appropriat ely be
calle d d ark li ght
T he E lectrical Wo rld
4 3 A s to th e th eory o f g eyser erup t ion th e
follo win g prin ciples h ave been es t abli sh e d Th e
boil in gpoint o f w at er rises as th e pre s sure in
creases , bein g 2 93 fo r a pressure o f four at mos
Also, if th e pressure be dim inis h e d wh en
ph ere s
th e w at er is u nd er very s t ron g pressure th e w at er
w ill imm e d iat ely as h int o st eam
Moreover, if
th e circulat ion is impe d e d, as wh en th e w at er is
cont aine d in lon g, narro w, irre gular tubes, an d
h eat e d with g reat rapi di ty th e boilin g point wil l
be reac h e d belo w W h ile it is far from th is poin t in
.

'

328

INDUCTIVE L OG IC

th e upper part of th e tube T herefore at th e mo


ment of erup t ion th e boilin g point fo r th e lo w est
d epth is act ually reach ed T h e w at er th ere bein g
t ransferre d int o s t eam , th e expandin g s t eam w oul d
l ift th e wh ole column o f w at er in th e t ube, causin g
an over o w T h is woul d d im ini s h th e pressure i n
every part o f th e t ube an d consequently a larg e
quant ity of w at er before very near th e boilin g point
w oul d as h int o s t eam an d ins t antly ej ec t th e
wh ole o f th e w at er in th e pipe th e s t eam rush in g
o ut imme d iat ely aft er w ar d s
T h e premonit ory can
n o n ad in g
beneath is evid ently pro duce d by th e
c ollapse of larg e s t eam bubbles risin g th rough th e
c ooler par t of th e w at er o f th e t ube BU N S E N S
T HE O RY
44
Mackenzie s th eory o f g eyser erupt ion is
th at th e g eyser pipe is connect e d by a narro w con
duit with th e low er part of a s ubt erranean cave
wh ose w alls are h eat e d by th e near vicinity of
volcanic re s T h e w at er risin g above th e openin g
o f th e con d ui t, an d c h an g in g int o s t eam an d h avin g
n o w ay of escape, w oul d th rough pressure th us
cau se d be force d up th e pipe an d th e st eam rush in g
aft er it P rofessor L e Cont e say s o f th is th eory : I f
th ere w ere but o ne g eyser this woul d be consi d ere d
a very in g enious an d probable hypo th esis ; for w e
may conceive o f a cave an d a con d uit so cons t ruct e d
as t o account for th e ph enomena But th ere are s o
many g eysers, th at it is inconceivable th at all of
th em sh oul d h ave caves an d con duit s so peculiarly
cons truct e d T h is th eory, th erefore, is ent irely un
t enable
.

O G I C AL

45

C ISES

3 29

E"ER

I t h as been foun d by experiment th at a

curren t movin g at th e rat e of th ree inch es per


secon d wi ll t ake up an d carry al on g ne clay ;
movin g six in ch e s per se c on d, w il l carry ne sand ;
ei ght inch es per secon d coarse san d th e size of lin
see d ; tw elve in ch es , g ravel ; tw enty four inc h es,
pebbles ; three feet an gul ar s t ones of th e size of a
h en s e gg I t wil l be readil y seen th at th e carry
in g po w er in cre as e s much more rapidl y th an th e
velocity F or ins t ance a current of twelve inch es
per secon d carries gravel while a curren t o f thr ee
fee t per secon d onl y th ree t imes great er velocity
carries s t ones many h un dre d t im es as lar g e as
g rains o f gravel
4 6 I f w oo d be soake d in a s t ron g solution o f
s ul p h at e o f iron ( copperas) an d d rie d an d th e s ame
process be repeat e d unt il th e w oo d is h i gh ly ch ar g e d
with th is s al t an d th en burne d th e s t ructure o f th e
w oo d w ill be preserve d in th e peroxid e o f iron left
A lso it is w ell know n th at th e smalles t s s ures and
cav it ies in rocks are spee d ily l le d by inlt rat in g
wat ers w ith mineral matt ers N o w, woo d burie d
in soil soake d w ith some petrifyin g mat erial becomes
h i ghly ch ar g e d with th e s ame an d th e c ells l le d
with in lt rat e d matt er, an d wh en th e woo d d ecays
th e pet rify in g mat erial is left, ret ain ing th e s truct
I n nature also th ere is an a ddi
ure of th e w oo d
t ional proces s , n o t illus trat e d by th e experiment
A s each
o r by th e example o f in l t rat e d l lin g s
pa rtic le o f org anic matt er pas ses aw ay by d ecay a
particle o f mineral matt er t akes i t s place until
nally th e wh ole o f th e org anic matt er is repl ace d
,

3 30

I NDUCTI VE LOG IC

A s t o th e orig in of bit um en th e follo w in g


observations h ave been ma d e : Certain or ganic
matt ers at or dinary t emperat ure in presence of
abun dant mo ist ure an d out of c ont ac t of air will
un d er go a species o f d ecomposit ion or ferment at ion
by wh ic h an oily o r t arry subs tance, S imilar t o bitu
men is form e d I n th e int erior of h eaps o f veg etable
subs t anc e suc h bitu m inous matt er is oft en foun d
F o s sil cavit ie s h ave been foun d in soli d limes t one
c ont ainin g b it umen ev ident ly forme d by d ec o mpo
S o also, S hales h av e
sitio n of th e animal matt er
been foun d in S co tlan d, lle d with s h es wh ich
h ave c h an g e d int o bit umen
4 8 Count Rumfor d in 17 9 8 prove d th at th e com
mon no tion th at h eat was a subst ance was false by
borin g a larg e piece o f brass un d er g reat pressure
o f th e borer wh ils t th e brass w as in a g allon of
w at er ; and at th e end of tw o an d o ne h alf hours
th e w at er act ually bo ile d
4 9 Kenelm D i gby s t reat ment o f w oun d s was t o
apply an oint ment not t o th e w oun d it self, but t o
th e s w ord th at h ad in ic te d it, t o d ress th is care
fully at reg ular int ervals, an d in th e meantime,
h avin g boun d up th e wound, t o leave i t alone fo r
s even d ays
It w as observe d th at many cures fol
lo w e d upon th is treat ment
5 0 W h en P ascal s barome t er w as carrie d t o th e
t op o f P uy d e D ome and th e mercury in it fell it
w as inferre d th at th e fall of th e mercury w as due
t o th e ch an g e in elevat ion Before nally accepting
th is conclusion, th e baromet er was place d in expose d
posit ions an d in s h elt ered, wh e n th e win d blew an d
47

LO

GICAL

33 1

E" ERCI S ES

wh en it w as c alm in rain an d in fo g ; an d th ese


varyin g c ircum s t ances d i d n o t mat erially a ffect th e
resul t
51 A F rench experiment er Po uch et, th ou ght h e
h ad ob t ain e d in d ubit able evi d ence of spont an eous
g enerat ion He took i nfusions o f ve g et able matt er
boile d th em t o a pit ch sufcient to d est roy al l g erms
o f lif e an d h erme tically seale d th e l iqu i d in glass
as ks Af t er an int erval micro or g anisms appeare d
It seems th at at a cert ain st a g e in Po u ch et s process
h e h a d occasion t o d ip th e mouth s o f th e asks in
mer c ury It occurre d t o P as t eur in repeatin g th e
experiment s th at g erms mi ght h ave foun d th eir
w ay in from th e at mosp h eric d us t o n th e s urface of
thi s mercury An d wh en h e carefully cleans e d th e
s u rfac e o f th e mercury , no life appeare d aft er war d s
in h is as ks
5 2 T h e cau s e s t o whi ch th e d ecay o f th e nat ives
o f N e w Z e al an d h as been assig ne d are g iven as fol
lo w s : drink d ise as e E uropean clothin g peace, an d
w eal th
Jo u rna l of the A nthrop o log ical Institute
53 A n eminent ju dg e w as in th e h abit o f j oco s ely
propoun d in g, aft er dinner a th eory th a t th e caus e
o f th e prevalence o f J ac o bin ism w as th e prac tice
He quo t e d o n one s id e
o f bearin g three names
Ch arles J ames F ox, Ri ch ard Brin sley Sh eri d an
J oh n Horne Tooke, John P hilpot Curran S amuel
T aylor Coleri dg e T h eobal d Volfe Tone
On th e
oth er h an d th ere w ere, William Pitt J ohn S co tt
W illiam W indh am S amuel Horsley Henry Dim
Moreover th e prac t ice o f
d as , Ed mu n d Burke
giving ch il dren th ree names h as b een a g ro wing
,

,
,

INDUC T I VE

33 2

LO

G IC

prac tice an d J acobin ism h as also been gro win g


T h e pract ic e o f g ivin g c h il dren th ree names is more
common in A merica th an in En g lan d I n En gland
w e st ill h ave a Kin g an d a House of L or d s ; but th e
A mer icans are Republicans Burke an d T h eobal d
Wolfe Tone are bot h I ris h men ; th erefore th e be
ing an I rish man is no t th e cause o f J acobinism
Horsley an d Hom e Tooke are both clergymen ;
th erefore th e bein g a clergyman is not th e cause of
J acobin ism F ox an d Win dh am w ere both e ducat e d
at O xfor d ; th erefore the bein g e ducat e d at O xford
is not th e caus e of J ac o binism P itt an d Hom e
Tooke w ere both e ducat e d at Cambridg e th erefore
th e bein g e duc at e d at Camb ri dg e is not th e cause o f
J aco binism Th e cause is, th erefore th e h avin g
th ree names MA CA U L A Y
54 T h e exo tic Pelargon ia h ave a peculiar h errin g
bone s t ru ct ure in th e pet al s ; moreover th e h erring
bone st ruct ure is conj oine d in th e Pelargonia with
th e g eneral c h arac t eris t ics of th e Geranieae A lso
th e o w ers with such see d vessels as o ur wil d g era
n ium s h ave th e c h arac t ers of Geranieae
It is, th ere
fore excee din gly probable that o u r w il d g eraniums
s h oul d h ave th e pecul iar h erring bone s t ruct ure
5 5 Colonie s ou ght not to rebel ag ains t th e
moth er country, since th ey are it s ch il dren an d
c h il d ren ou ght no t t o rebel a g ains t th eir parent s
5 6 F in d in g th at th e s ize of t o wns varies con
c o m itantl
y w ith th e si ze of th e ri vers o n wh ich th ey
are buil t an ob s erver might infer th at th e size o f
th e river w as due t o th e S ize o f th e t own
57 An eminent auth or, w rit in g o n th e w ork o f
.

L OG ICAL

33 3

C ISES

E" ER

th e Englis h Ch urch before th e Tract arian move


ment c ont rast s th e n e w er s t at e of th in g s unfavor
ably w ith th e ol d er because th e Churc h in th ose
former d ays t au ght us to use reli g ion as a li ght
by wh ich t o s ee o ur w ay alon g th e roa d o f duty
With out th e sun our eyes w oul d be o f no u s e t o us ;
but if w e look at th e sun , w e are simply d azzle d an d
can see neith er it nor anyth in g else It is precisely
th e same w ith th eolo g ical speculat ions I f th e bea
c o n lamp is s hin in g
a man of h eal th
y min d w ill
no t discuss th e composition o f th e ame
58 S carle t color preva ils amon g balsamina
E uph orbia, P elar g onium , poppy, S alvia, Bouvar d ia
an d V erbena ye t none o f th e scarlet s are o f s w ee t
perfum es S ome of th e li ght Colored bal s ams an d
verbenas are s wee t sce nt e d but none of th e s carlet s
are T h e common sa g e w ith blue blooms is o d o rif
e ro u s bo th in o w er an d folia g e ; bu t th e scarle t
s al vias are d evoi d o f smell
N one of th e s w ee t
scent e dleave d P elar g oniums h ave scarlet blooms,
an d none o f th e scarlet bloomers h ave s w eet scent
o f leaves n o r o f
blooms
S ome of th e wh it e
mar g in e d poppies h ave pleas ant o d ors ; but th e
Briti s h scarlet s are no t s w ee t scent e d T h e Brit ish
wh it e bloomin g h awt h orn is of th e mos t d eli ght ful
frag rance ; th e scarlet o w er h as n o smell S ome
o f th e h oneysuckles are s w eet ly perfum e d, but th e
scarlet t rumpet is s centless
59 T h e pro d uc t ive po w ers o f plant s, ju dg in g
from th e increase d fert ili ty o f th e parent plant s
an d from th e increase d po w ers o f g ro wth in th e
o ffsprin g, are favore d by some d e g ree o f d ifferen
,

'

"

3 34

INDU CTIVE L O GIC

tiatio n in

th e element s wh ic h int erac t an d unit e


so as t o form a ne w bein g Here w e h ave some
analo gy w ith ch e mical afnity o r att raction, wh ich
comes int o play only betw een at oms o r molecules
o f a d i fferen t
nat ure
A s Professor Miller re
mark s : G enerally speakin g, th e g reat er th e dif
ference in th e properties of tw o bo dies, th e more
int ense is th eir t en d ency to mutual ch emical ac tion
But betw een bo dies o f a s imilar ch aract er the t en

i
d eney t o unit e s feeble
6 0 I n afrmin g th at th e gro wth o f th e bo dy is
mech anic al an d th at th ou ght, as exercise d by u s,
h as it s correlat ive in th e ph ysics o f th e brain, I
th ink th e posit ion of th e mat erialis t is s t at e d, as
far as th at position is a t enable one I th ink th e
mat eriali st w ill be able nally t o maintain th is po
sitio n a g ains t all attacks ; bu t I d o no t th ink in
th e present con dit ion of th e hu man mind th at h e
can pas s beyon d th is posit ion I d o not th ink h e is
ent itle d t o say th at h is molec ular g roupin g s and his
molec ular mo tions explain every th in g I n reality
th ey explain nothin g T h e ut most h e can afrm i s
th e associat ion of th e two classes of p h enomena, of
wh ose real bon d o f union h e is in absolut e i gnorance
T h e problem o f th e connec t ion of bo dy an d s oul is as
insoluble in it s mo d ern form as i t w as in th e presei
Ph osph orus is kno wn t o ent er int o th e
entic a g es
composit ion of th e human brain, and a trench ant
G erman w rit er h as exclaime d Ohne Ph o sph or
kein G e danke
T h at may o r may not be th e case ;
but even if w e knew it t o be th e case, th e kno wle dg e
w oul d no t li ght en o ur d arkness
TY N DA LL
.

LO G ICAL
61

33 5

ExERCIS Es

G rant in g th at He g el was more o r less success

ful in cons t ruc tin g, a p riori, th e lead in g result s o f


th e moral sciences , s t ill i t was no proof of th e correc t
ness o f the hypoth esis o f i d ent ity , w ith wh ich h e
s tart e d Th e fac t s o f nat ure w oul d h ave been th e
crucial t es t T h a t in th e moral sciences t races o f
th e act ivity of th e h uman int ellect an d o f th e several
s t a g es o f it s d evelopment s h oul d present th em
selves, w as a matt er o f course ; but surely if nature
reall y re ect e d th e res ul t o f th e th ou ght of a cre
at ive min d th e sys t em ought, w ith out di f cul ty, t o
nd a place fo r h er comparat ively simple p h enom ena
an d processes
HE L M H O L T Z
6 2 Wh en youn g G alileo w as a s t u d ent at P isa
h e not ice d on e d ay durin g th e service at th e g reat
Cath e d ral , th e ch an d elier s win gin g back w ar d s an d
forw ard s an d convin ced h imself, by count in g h is
p ul se, th at the d uration o f th e oscillations w as
in d ep end ent of th e arc th rou gh wh ic h it move d
63 G oeth e enunciat e d th e exis t ence of a resem
bl ance between th e d iffe ren t part s o f o ne an d th e

same or g anic bein g


A ccor d in g t o G oeth e s o wn
account th e i d ea rst occurre d t o h im wh ile lookin g
at a fan palm at P a d ua He w as s t ruck by th e im
mense variety of ch an g es o f form wh ic h th e succes
s iv ely d evelope d s t em leav es exh ibi t, by th e w ay in
w h ic h th e rs t sim ple ro o t lea e t s are repl ace d by
a series o f more an d more divi de d leaves , t ill w e
come to th e most complicat e d He aft erward s suc
c eed e d in d iscoverin g th e t ransform at ion o f s t em
leaves in to sepal s an d pet als an d o f s epals an d
pe t als int o s t amens, nec t aries, an d ovaries an d th us
.

INDU CTIVE LO G IC

3 36

h e was le d t o th e d oct rine o f th e met amorph osis of


pl ant s wh ich h e publish e d in 17 90

6 4 A fort unat e glance at a broken s h eep s


S kull,
wh ic h G o eth e found by acci dent on th e s an d o f th e
L ido at Venic e sugg e st e d to h im th at th e skull it
self consis t e d o f a s erie s o f very muc h alt ere d verte
brae At r st si ght no tw o th in g s can be more un
like th an th e broa d, uniform , cranial cavity of th e
mammalia en close d by smooth plat e s, an d th e nar
ro w c yl in d ric al tube o f th e spinal marro w, compose d
o f s h or t, massy j agg e d bones H E L M H O L T Z
65 T h e exis t ence o f th e s o calle d blin d s pot in
th e eye was rs t d emon st rat e d by th eoretical ar gu
ment s Wh ile th e lon g c ontroversy wh eth er th e
perception o f li ght resi d e d in th e re tina o r th e
c h oro id w as s t ill un d eci d e d, Mario tt e aske d h imself
wh at percep tion th ere was wh ere th e ch oroi d is
d eficient He ma d e experiment s t o discover th is
point an d in th e course o f th em d iscovere d th e
bl ind s po t
6 6 Hairy observe d th at crys t als of h eavy spar
from S icily an d those fro m D erbys h ire (wh ich w ere
c on s i d ere d to be th e same subs t anc e) diere d in th e ir
an gles of cleava g e by th ree an d on e h alf d eg rees an d
rem arke d : I coul d no t suppos e th at th is difference
w as th e effect o f any law of d ecrement ; for it w oul d
h ave been ne c essary t o s uppose so rapi d an d complex
a l aw that such a h ypo th esis mi ght h ave been jus tly
re g ar de d as an abuse o f th e th eory Vauquelin
by ch emic al analysis discovere d th at th e base of
th e crys tals from S icily was st rontia, an d th at o f
th ose fro m D erbysh ire was bary ta T h ese fact s ,
.

LO GIC AL

C I SES

3 37

E"ER

becomin g known to Hauy enable d h im by inference


t o di scover th at th e an gles o f crys t al s m i ght be
employe d as a t es t fo r th e presence of d ifferent
subs t ances which very nearly resemble eac h o th er
in o th er respec t s
6 7 G raebe, a G erman c h emis t, in inves ti gat in g
a clas s of compoun d s, calle d th e quin ones d e t er
mine d inci d en tally th e molecular s t ruc t ur e o f a
bo dy closely resemblin g ali zarin e, wh ich h a d been
d iscovere d s everal years before T his bo dy w as
d erive d from n aphth aline an d lik e m an y sim il ar
d erivat ives w as re duce d back to naphth al ine wh en
h eate d with z inc d us t T h is circum s t ance le d th e
ch emis t t o h eat al so ma dd er aliz arine wi th zinc
dus t wh en t o his surprise h e obt aine d anth r ac ene
O f course th e inference w as at onc e draw n th at
alizarin e mus t h ave th e s ame relat ion to anth rac ene
th at th e al lie d colorin g matt er bore t o nap hth a
line ; an d, more th an th is it was also inferre d
th at th e s ame ch emic al processes whic h pro duce d
th e colorin g matt er from naphth aline wh en applie d
Th e re s ul t
to an thracene w o ul d yiel d alizarine
ful ly ans w ere d th ese expec t ation s an d n o w ali
zarine is manufac ture d on a lar g e s cale from an
th rac en e obt aine d from coal tar
CO O K E The N ew
,

Chemistry

L yell by s tu dyin g th e fact th at


th e river G an g es yearly conveys to th e ocean as
muc h earth as w oul d form S ixty o f th e great pyra
mid s of E gy p t w as enable d to in fer th at th e ord i
n ary s lo w causes n o w in operation upon th e earth
wou l d accoun t fo r th e immense g eolo gical ch an ge s
68

S ir Ch arles

338

I NDU CTIVE

LO

G IC

that h ave occurre d, w ith out h avin g recourse t o th e


less reasonable th eory of su dd en cat as t rop h es

Joule
69
s exper iment s s h o w th at wh en h eat is
pro duc e d by th e consumpt ion o f w ork a d enit e
quant ity of work is require d t o pro duce th at amount
o f h eat wh ic h is kno wn t o th e p h ysicis t s as th e uni t
o f h eat ; th e h eat th at is t o say , wh i c h is necessary
t o rai s e o n e g ramme of wat er th rough one d e g ree
centigra d e T h e quanti ty o f w ork necessary for
th is is accor ding to Joule s bes t experiment s, equal
t o th e w ork wh ic h a gramme woul d perform in fal l
in g th rou gh a h eight o f 425 met res
I n or d er t o s h o w h o w closely concord ant are his
num bers I will adduce th e resul t s o f a fe w series
of experiment s wh ich h e obt aine d aft er intro d u c
in g th e lat es t improvement s in h is metho d s
(1) A series of experiment s in wh ich w at er w as
h eat e d by fric tion in a brass vessel I n th e int erior
o f this vessel a ver t ical axle provi d e d w ith six t een
pa ddle s was rot at ed th e e ddies thu s pro duce d bein g
broken by a serie s o f proj ec t in g barriers , in wh ic h
part s w ere cut out lar g e enou gh for th e pa ddles t o
pas s th rou gh T h e value o f th e equivalent w as
met res
(2) Two S imilar experiment s, in which mercury
in an iron vessel was subs t it ut e d for w at er in a
bra s s o ne gave 425 an d
met res respec t ively
(3 ) T w o series of experiment s, in wh ich a
conical rin g rubbe d ag ainst anoth er both sur
roun d e d by mercury, g ave
an d
met res
respec t ively
Exact ly th e sam e relat ions be tw een h eat an d
.

LO

G ICAL

339

C ISES

E" ER

w ork w ere also foun d in th e reverse process ; th at


is wh en w ork w as pro d uce d by h eat H EL M H O L T Z
7 0 A g as wh ic h is allo w e d t o expan d w ith mo d
c rat e velocit y becomes coole d
J oule w as th e rs t
t o S h o w th e reason of this coolin g Fo r th e g as h as
in exp an di n g, t o overcome th e res is t ance whic h th e
pressure o f th e atmosp h ere an d th e s lo wly yiel din g
si d es o f th e vessel oppose t o it ; or if i t cannot o f
it self overcome th is resis t ance it support s th e arm
G as th us perform s
o f th e observer wh ic h d oes i t
w ork , an d th is w ork is pro d uce d at th e cos t o f its
h eat Hence th e coolin g I f, on th e cont rary th e
g as is su dd enly allo w e d t o issue int o a perfec tly ex
h an s t e d sp ace wh ere it n d s n o res is t ance it d oe s
no t become cool as J oule h as S h ow n
HE L M H O L T Z
7 1 T h e pr incipal feature in th e plan of my at
t empt t o penet rat e in t o th e N orth P olar re gion o r
if possible t o cross it is in brief t o t ry t o make use
o f th e curren t s of th e sea ins t ea d o f ghtin g a g ains t
th em My opin ion is as I h ave alrea dy explaine d
th at th ere mus t some wh ere
o n several occ asions
run c urrent s int o th e P olar re g io n wh ich carry th e
oe ice across th e P olar S ea rs t northw ar d t o w ar d
th e P ole an d th en southw ar d a gain in t o th e At lan
T h at th ese current s really exist all
tic O cean
A rct ic expedi t ions prove as mos t of th em h ave h a d
t o ght a g ain s t th e current s an d a g ains t th e ice
drift in g southw ar d, because th ey h ave t rie d t o g e t
northw ar d from th e w ron g si d e
I th ink a very
simple conclusion mus t be d rawn from th i s fac t th at
current s an d d rif t in g ice are c ons t antly co m in g from
th e unkno w n nor th, v iz Current s an d perh aps al s o
,

INDU CTIV E LOG I C

3 40

ice mus t pass int o th is same re gion, as th e w at er


running out mus t be replac e d by w at er runnin g in
T his conclusion is base d upon th e simples t of all
natural law s ; but th ere seem t o be people wh o will
not even a d mit th e neces s ity of th is
Th at such current s run across th e N orth P olar
region is also pro v e d by many fac t s I may men
tion th e great quantit ies of S iberian driftwoo d wh ich
are annually carrie d t o th e S h ores of S pitzber g en
an d G reenlan d ; i t comes in such abun dance an d
w ith such re gularity th at it is quit e impo s sible i t
s h oul d be carrie d t o th ese S h ores so far from th e
orig inal h ome, by occa sional w inds or current s
Th ere mus t be a re g ular communicat ion betw een
th e c oa sts of S iberia and th ose o f S pit zber g en an d
G reenl an d
By th is s ame communicat ion w ere
several obj ect s from th e unfort unat e Jeannette car
rie d t o th e G reenlan d coas t T h e Jeannette sank
in J une, 188 1, t o th e north o f th e N ew S iberian
I slan d s, an d th ree years aft erwar d, in June, 18 84 a
g reat many obj ect s belon g in g t o h er o r h er crew
o e on th e southw es t coas t o f
w ere foun d on an ice
G reenlan d T his o e can only have been brou ght
th ere by th e same current wh ich carries th e drift
woo d By th i s s ame current an E squimau impl e
ment, a th row in g s ti ck or h arpoonth row er w as also
carrie d th e lon g w ay from Alas ka t o th e w e st coas t
o f G ree nlan d
T h ere c an in my opinion be no
doubt o f th e exis t ence of s uch a communicat ion o r
curren t across th e N orth P olar re gi on from th e
S iberian s i de t o th e Greenland si de
DR N A NS E N
in The S trand M ag azin e
.

I N DE "
Bo tany 283
Bo yl e 104 132
Brad l ey , 22
Brah e Tyc h o 76 304
Brew s ter 40 97 98
Bro wn 59
Bull en , 247
Buns en, 150 , 245
.

Ad am s , 147
A d er ial

v b

p ob ability 232
eg at es P o b ability as
r

Agg r
r
ai
,
fec ting , 234 if
A g reeme nt , Me t o d o f , 84 , 86
104 , 128 , 259 , 27 5
Al g e raic al lo gic , 219

pp

Anal o g y , 35 , 39 , 44 if
204 ff , 28 3 , 29 1
An al ys is , 60 , 6 2 , 186
A
e rc e tio n , 3 , 266
A q u a en d ente , 211

202,

p
Arithmetic al m eth o d

37

A ris to tl e 16 , 46 , 50 , 304
Asso c iatio n o f id eas , 267
Astro no my, 284

C
C

Attentio n , 265

p b b

Cze salpin us , 304


alc ul atio n
of
ro a ility ,
230 if
am anella , 303
au s al analy sis , 35 , 40 , 4 2, 45 ,
4 7 , 60 , 64 ff
ausatio n , 45 if , 47 , 50 ff , 209 ,
227 , 228
h al me rs , 6 9
h anc e , 240 , 24 5 ff
e mic al c o m in atio ns , 71
h em is try , 288
Ch ene v ix , 266
irc u m st ant ial e v id enc e , 247
lassic at io n , 206 ff , 279 , 283
liffo rd , 50 , 16416 6 , 17 2, 218
o existenc e , 6 6 , 9 2 , 25 8
o inc id e nc e , 245 , 246
o l lo c atio n , 68 , 69 , 255 , 256
T eo ry o f in
o m inatio n s ,
c au sal an al ys is , 77 ff , 105
o nc e t , 9 0 , 9 1 , 205
ro c esses ,
o nc e tual
allacies
o f, 26 3 , 27 5

Bac on ,

rancis ,

16 , 46 , 7 2 , 73 ,
94 , 163 , 199 , 267 , 269 , 27 1, 27 5 ,
27 6 , 300 , 303 , 304 , 309 , 3 11
Bac o n , Ro g er, 300
Bain , 16 , 220
.

Barrett , 6 1, 264
Bas is o f ro a ility , 232
Ben e e 3 6
Be ud ant, 141
Bio l o gy, 216 , 284
Bl u ntsc li, 294 296
Bo o l e , 2 19
Bos an qu et , 7 , 26 27 , 29 , 40 , 42
.

p b b

58 , 8 1, 3 12

C
C
Ch
C

C
C
C
C
C
C
C
C
C

p
p

3 41

INDE"

942

Co nc o mitant v ariations Meth o d


,

84 , 85 , 130 if
o n se rv at io n o f ene rgy , 5 2 ff ,
7 1, 7 2 , 28 8 if
o n silie nc e o f ind uc t io ns , 198
o nte nt , ex lic it and im lic it ,
o f,

C
C

Co -o
'

pe

ve c irc ums tanc es

rati

C o pernic us 304
Co r el atio ns 67
Cou nteracting c aus e 7 0
Co u nter pro b ability 231
Cuvier 9

F
a

123 , 126 , 139 ,

6 8 , 7 7 , 113 ,
16 1, 17 5 ,

s,

Darwi n , G H , 111
Davy , S ir Hu m ry , 149 , 206
27 3
Ded uc tio n

ph

ind uc tio n , 16 ff

and

Ded u ctive met o d , 147


DeMorg an , 74
Derivat ive l aw s , 253 , 254
Descarte s , 219
Devel o m ent , 6 8, 218
Differenc e , M eth o d o f, 79 , 84
.

85 , 10 1 ff , 117 , 149
Disco e ry , 202
.

hamel

Du

j min

en a

48 , 49 , 107 ,
275 ff
,

204 , 20 5 , 256 ff ,
eo l o g y , 284 ff
id e , 144 , 241
il e rt , 3 04
l au er, 155
o re , 167 , 17 7 , 206 , 211, 264 ,
266
ra e r, 115
re en , 13 , 45, 5 0, 58
.

G
G
G b
G b
G

G b
G
Guyot

97

28

Eliminatio n , 90
Ellis , 213
Em iric al asis o f ro a il it y ,
232
Em iric al l aws , 3 8 , 2 52 , 276
En um e ratio n , 4 3 , 45 ff , 9 1,
228 ff
E is te mo l o g y , 2 6 , 5 8
Eth ic s , 295
.

p b b

Halley , 152 159


Harvey 211, 225
Hatc he tte 141
He g el 294
Herac litus 270
Hersc h el 5 9 7 3 ,
.

44

Galileo 302 304


Generalizatio n 23

S ee Tel e ol ogy

c au se .

G
.

160,
273

18 1, 187 ,
.

F
Fi l
Fl n 223
F " 114
F anklin B
o re

ru

o re

Ch arles

F t nd t th 19 20
Fall i 262 ff
Fa d ay 6 1 74 112 114
na

Darwin ,

9 7 , 17 6 ,

166 , 169 , 205 , 2 17 , 268 ,


ic tio n s , 19 6

p
p
p

ra

ac es ,

10 , 77 if

282
Ex lanat io n , 14
Ex l anatio n , Histo ric al , 291
Ex lic it c o n tent , 6

ac

69

p eriment

Ex

149 , 152 , 2 17 ,

27 5 , 308 ff
Hi o c rates , 223
Hist o ric al e x l anatio n , 291
Ho ll and , 217
.

pp

Hume , 4
Huyg h ens
,

15 8

3 43

I N DE "
Hypo th es is ,

42, 93 , 17 4 it , 210
6 . La ate r, 18
27 1 ff , 284, 308
Law , 30 , 3 1
H po t etic al uni ersal , 31
Le o nard o d a inci, 32, 301
Le errie r, 147
Linn aeus , 217
I
Lis te r, 219
o l s o f Bac on , 269 ff
Lo ck e , 16 , 307
ma inatio n , 185 , 202 , 263 , 27 1 ff Lo c er 208
y ,
t
if
m lic it con ent, 6
Lo t z e , 22 , 25 , 26 , 29 , 31, 46 , 58 ,
ncrement of pro a ili ty , 233 ,
109 , 188 , 189 , 312
Lo n a, 241
nd uctio n and
e uc tio n , 16 if
Lu o c , 83 , 114 , 120 , 124 ff ,
Ind ucto -d ed uc ti e me t o d , 156
195 , 2 12 , 213
if , 282
Lyel l , 138 , 182
nd uc ti e azard , 24
nference , 1 if
nstances , Num er o f, 42
Malt u s , 219 , 256
nsurance , 240
M ans el , 57
ntermixture o f effec ts , 59
Math em atic al met o , 172, 173
n aria ilit
o f c au s atio n, 51
M ech anic al co m ination, 7 0
54, 58
Med ic ine , 99 , 100
n erse ro lem , 27
M ill , J S 16 , 25, 30, 54 57 , 59 ,
6 8, 7 2 , 7 9 , 84 , 86 , 129 , 156 , 201,
202 , 220 , 254, 260, 298 , 309
ames , 10
Minto , 3 12
Janet, 209 , 210, 211, 224
Mo lar fo rc es, 7 1
en in, 7 2 , 164 , 165
M
o l ec ul ar o rces, 7 1
r
1
2
22
enne , 77 ,
Je o n , 27 29 , 115 , 143 , 17 0,
208, 217 , 236 , 246 , 312
206
o int met o d , 84 , 85 , 117
N at ural ind s 67 205 206
.

Id
I g
I p
I

b b

dd

bb k

v h

I
I
I
I
I
I v

b
.

I v

hd

p b

J k
J

J
h
Jo ul e 167
Judgment

Fall ac ies

o f,

263

Neg atio n, Determination by ,


7 8 ff , 104
Nega tive c o nd itio ns , 7 0
3 12
N ewman , J
N e wto n 39 , 153 158, 15 9 ,
N o n caus a pro causa, 267
.

Juris pruden

c e,

293 t

K ant , 53
K e ler, 7 6 , 199 , 304 , 3 10
K irc er, 217
K irc h off, 245

h
h

Ob servati on
O wen 215
,

72 ff

91, 227

d
p

La d , 294
La lac e, 153 , 179
.

Pas te u r, 9 8 , 99 , 160
Pe rc e tio n , 2 ff

INDE"

3 44

Fallacies

Perc e tio n,

26 3 ff

o f,

Perfec t ind uc tio n , 3 6 ff


Ph ysics , 282
Plate au , 115 , 125
Plato , 50 , 297
Pl urality o f c auses , 59 , 9 3
Po l itic al ec o no my , 292 if
P os t ho c erg o p rop ter h o c, 59
267
Po st ul at e ,
und amental , 8, 26
189 if , 26 1
Po t e ntial c ause , 66 , 108
Po te ntial in in fe re nce, 11
Pred ic tio n , 156 ff
P rese ntatio n , Data o f, 5
Preyer, 56
Priestl ey , 108
Pro a ility , 29 , 38 , 43 , 56 , 226 ff
Pro o f, 20 2
Psyc h ol o gy an d l o g ic , 2 if
14 , 15 , 47 62 , 133 , 232 , 262 ff
.

Sc wann, 219
S c ience , 33
Sc ienc es and th e in uctive
m eth o d , 281 ff
S cie ntic analysis
S ee Caus al
.

anal ysis

S c ientic s irit, 33 , 16 4 , 201


S e q ue nc e , 64 ff , 92
S id g wic , 26 8
S ig wart , 26 , 29 , 57 , 58, 68, 94 ,
.

b b

279 , 289 ff
Psyc o lo g y and
.

h
history 291 E
Ps yc ho l og y Meth o d in 294
,

282
Q u etel et , 236
.

p
p
Sp

reng el ,

2 13

S ufc ien t re aso n , 5 8


S ug g es tio n o f c ausal relatio n ,
99 , 224
Synt esis , 186
S ystem , 7 if , 12 , 17 if , 44 , 45 ,
5 8 , 14 8 , 176 , 209 , 211, 266
.

T
.

Q uantitative d eterminatio n , 7 2 ,
13 1 ff

Sim ilarity , Law o f, 205


S mith , A d am , 268
S oc rate s , 297
286
S enc er, 257 , 284
S ino z a, 187

Tait, 52 , 133 15 1, 153 , 159 , 168 ,


,

17 7 , 180 , 19 7
Tel eo lo gy , 208 ff
.

222 if

278

Tel esius , 3 02
Tennyso n , 12 , 51
Th o m s o n , 13 3 , 15 1, 153
To tality , Law o f, 7
Truth and fac t , 19 , 20
Tynd all , 7 5 , 7 6 , 80 , 81, 102, 13 5 ,
166 , 183 , 185 , 217
.

Rail ro ad
Reality ,

acc id ents ,

Inference

referenc e

to , 13

236
an ind irec t
.

Red uc tio n, 28 , 42
Resid ues , 84 , 85 , 146 if
Ric ard o , 13 2
Ro manes , 200
Rul e , 30

Ueb erweg , 32 , 43 5 8, 274 , 300


Ultim ate l aws , 253 , 254
Unifo rmity o f nat ure , 22, 25,
.

S aig ey , 89 , 135
S aint-Pierre , 224
S c h l eie rmac h er, 16

51, 54 if
ni e rsal , 12 , 18 , 27 5
ni e rsal c ausat io n , 51
.

U v
U v

3 45

IM ) EK
V

Venn

77 , 106 , 111,

Veric atio n

W all ac e

Wb

aw ,
e er s r

156 if

19 1

W h ateiy
W hew el l

30 1 ,

57 , 3 11

3 0 , 198 , 201,
3 10
.

Zo dl ogy, 283

You might also like