Computing S

ien e 272
Notes on the Fermat Numbers

November 18, 2003

The Fermat numbers fn gn0 are dened to be the positive integers
n = 22n + 1

for  0
The rst ve Fermat numbers
n

0 = 3 1 = 5 2 = 17 3 = 257 4 = 65537
;

are all primes, and this led Fermat to onje ture that every Fermat number n is a prime.
Almost 100 years later, in 1732 Euler showed that this onje ture was false, and he gave the following
ounterexample:
5 = 4294967297 = 641
6700417
The Fermat numbers that are prime are alled Fermat primes, and to this day, it is not known if there are
innitely many Fermat primes. No fermat primes beyond 4 have been found. However, the largest known
omposite Fermat number is 23471 and was dis overed by W. Keller in 1984.
The mathemati ian G. Bennet gave an elementary proof of the fa t that 5 is omposite without doing any
division! His proof is as follows: First note that
;

641 = 5
27 + 1

Now,
5

5 = 22 + 1 = 232 + 1 = 24
228 + 1 = 16
228 + 1

= (641 625)
228 + 1 = (641 54)
228 + 1


= 641
228 5
27 4 + 1 = 641
228 (641 1)4 + 1


= 641
228 6414 4
6413 + 6
6412 4
641 + 1 + 1


= 641
228 6413 + 4
6412 6
641 + 4


so that 641  5
There is a lot known about the Fermat numbers, but probably the most important is the onne tion, found
by Gauss in 1796 between the Fermat primes and the ruler-and- ompass onstru tion of regular polygons.
His result is as follows:
Theorem. A regular polygon with
sides is onstru tible using only a ruler and ompass if and only if
:

= 2r
1 2



f f

fk ;

where  0 and 1 2
k are distin t Fermat primes.
The early Greeks knew how to onstru t regular polygons with 2k 3
2k 5
2k and 15
2k = 2k
3
5 sides.
They also knew how to onstru t regular polygons with
r

f ; f ; :::; f

3 4 5 6 8 10 12 15 and 16
;

sides, but did not know how to onstru t a regular polygon with 17 sides. This was done by Gauss when he
was 19 years of age, and (so the story goes) he de ided to devote the rest of his life to mathemati s. He also
requested that a 17-sided regular polygon be engraved on his tombstone (it is not there).

Now we give a re urren e relation satised by the Fermat numbers and use this to prove some interesting
results.
Theorem. For every integer
 0 we have
n

n+1 2 = 0
1
2


n
Proof.

Sin e 220 1 = 1 then we have

;

0
1
2
3


n = 1
0
1
2
3


n








= 220 1 220 + 1 221 + 1 222 + 1 223 + 1


22n + 1







= 221 1 221 + 1 222 + 1 223 + 1


22n + 1







= 222 1 222 + 1 223 + 1


22n + 1






= 223 1 223 + 1


22n + 1
..
.




= 22n 1 22n + 1
= 22n+1 1
= 22n+1 + 1 2
= n+1 2
:

Any two distin t Fermat numbers m and n with

are relatively prime.
Proof. Let m and n be distin t Fermat numbers with
and suppose that
0 is a ommon
divisor of m and n then divides 2 = m 0
1


n


m 1 Therefore, = 1 or = 2 but m
and n are odd, so we must have = 1 Therefore, for
the Fermat numbers m and n are relatively
prime.
Theorem.

m > n

m > n;

d >

m > n;

And now we give G. Polya's elementary proof that there are innitely many primes, that is, there is no
largest prime.
Theorem. There are innitely many primes.
Proof. There are innitely many distin t Fermat numbers, ea h of whi h is divisible by an odd prime, and
sin e any two Fermat numbers are relatively prime, these odd primes must all be distin t. Thus, there are
innitely many primes.
Finally, we an show the following:
Theorem. For every integer
 0 the positive integer
n

= 22n 1

is divisible by at least dierent primes.

Proof. Note that for ea h
 0 we have
n

22n 1 = 22n + 1 2 = n 2 = 0
1
2
3


n 1
and all of the k 's are pairwise relatively prime.

Re all that a positive integer is a triangular number if and only if

n

= ( 2+ 1)
for some positive integer As an exer ise, prove the following result:
Theorem. Among the Fermat numbers n there are no squares, no ubes, and no triangular numbers
(ex ept 0 = 3 = 22
3 ).
n

k:

k k