You are on page 1of 19

The Gravitational

Two-Body Problem
Note 5.7. Put the origin in a Cartesian coordinate system at the center of an ellipse
drawn by the technique of a loop of string stretched against two tacks, aligning the
x-axis and the y-axis, respectively along the major and minor axes. Using the
convention (x, y) to denote the coordinates of a point, we note that the two foci are

y ( x, y )

s− s+


− ea

+ ea x

located at (-ea, 0) and (+ea, 0), respectively. If (x, y) now denotes a point on the
ellipse then the application of Pythagoras’s theorem gives the length of string s- from
the first focus as

s − = ( x + ea) 2 + y 2 .
2
(N5.1)

Similarly, the length of string s+ from the second focus is given by

s + = ( x − ea) 2 + y 2 ;
2
(N5.2)

whereas the condition for the taut string path from one focus to the other is

s − + s + = 2a. (N5.3)

If we apply the binomial theorem to expand out the right-hand sides of equations

1
(N5.1) and (N5.2) and if we subtract the latter from the former, we get

4eax = s − − s + = ( s − + s + )( s − − s + ) = 2a( s − − s + ),
2 2
(N5.4)

where we have used equation (N5.3) in the last step. Equation (N5.4) implies

s − − s + = 2ex, (N5.5)

which can be combined with equation (N5.3) to solve for the length of each string:

s − = a + ex, s + = a − ex. (N5.6)

Substitution of equation (N5.6) into equations (N5.1) or (N5.2) produces the same
result, after some algebraic cancellation and collection of terms,

(1 − e 2 ) x 2 + y 2 = (1 − e 2 )a 2 . (N5.7)

If we recall that the square of the semiminor axis b is related to the square of the
semimajor axis a by the equation b 2 = (1 − e 2 )a 2 , we obtain upon division of
equation (N5.6) by (1-e2)a2:

x2 y2
+ = 1, (N5.8)
a2 b2

which is the elegant Cartesian equation for a centered ellipse. It is the familiar
equation for a circle of radius a, x2+y2 = a2, except that the y-axis has been
foreshortened by a factor b/a because of viewing the circle at a tilt. (QED)

Note 6.2. Motion in a central force field will occur in single plane defined by the
intersection of radius vector and velocity vector at any instant in time. Because there
are no forces out of this plane by definition, there can be no velocities generated that
are perpendicular to it. Define now a Cartesian coordinate system (x, y) with origin
at the center of force. The radial distance r of the orbiting body satisfies
Pythagoras’s theorem:

r 2 = x2 + y2 , (N6.1)

while the gravitational field g has the x- and y-components:

2
x y
g x = −g , g y = −g , (N6.2)
r r

where the magnitude g is given by the inverse-square law

GM
g= . (N6.3)
r2

Using Newton’s fluxion notation (see Note 5.6), we may write the gravitational
equation of motion a = g as

x y
&x& = − g , &y& = − g . (N6.4)
r r

The symmetry of equation (N6.4) encourages to multiply the first equation by y and
the second equation by x and to subtract the result, yielding

x&y& − y&x& = 0. (N6.5)

The left-hand side of equation (N6.5) is the time derivative of ( xy& − yx& ) , as may be
verified by explicit differentiation. The derivative of something being zero implies
that something is a constant (a conserved quantity):

xy& − yx& = constant ≡ j. (N6.6)

We refer to j as the specific angular momentum of the orbit (the so-called


z-component of the angular momentum per unit mass of the orbiting body). We note
that we derived the constancy of j without needing to assume any particular spatial
dependence for g, e.g., equation (N6.3). Thus, as shown first by Newton, the
conservation of orbital angular momentum holds as long as the force field is centrally
directed.
We now also prove the conservation of total specific energy. First, we
differentiate equation (N6.1) and cancel a common factor of 2 to obtain

rr& = xx& + yy& . (N6.7)

If we now multiply the first of equation (N6.4) by x& and the second by y& , we get

3
GM GM
x&&x& + y&&y& = − 3
( xx& + yy& ) = − 2 r&, (N6.8)
r r

where we have made use of equation (N6.7) in the last step. It is easy to show that
the left-hand side is the time-derivative of ( x& 2 + y& 2 ) / 2 while the right-hand side is
the time derivative of GM/r. Thus, we have

d ⎡1 2 GM ⎤
⎢ ( x& + y& 2 ) − = 0. (N6.9)
dt ⎣ 2 r ⎥⎦

The quantity inside the square bracket must be a constant, the conserved specific
orbital energy:

1 2 GM
( x& + y& 2 ) − = constant ≡ ε . (N6.10)
2 r

Note 6.3. To interpret as well as to integrate equations (N6.6) and (N6.10) further,
we introduce polar coordinates:

x ≡ r cos θ , y = r sin θ . (N6.11)

Differentiation then demonstrates

x& = r& cos θ − rθ& sin θ , y& = r& sin θ + rθ& cos θ . (N6.12)

Algebra then obtains for equations (N6.6) and (N6.10) the simplifications:

1 2 GM
r 2θ& = j , (r& + r 2θ& 2 ) − = ε. (N6.13)
2 r

Notice that the product of radius times angular velocity, rθ& , is the component of

velocity in the θ direction, i.e., the circular velocity vθ = rθ&. . Thus, the specific

orbital angular momentum is the combination of lever arm r times the circular

velocity v θ , i.e., j = rvθ = r 2θ&.

Write out equations (N6.13) using Leibniz’s notation:

4
1/ 2
dθ dr ⎡ ⎛ GM ⎞ j 2 ⎤
r 2
= j, = ⎢2⎜ ε + ⎟− ⎥ (N6.14)
dt dt ⎣ ⎝ r ⎠ r2 ⎦

where we have substituted in the first of equation (N6.13) to eliminate θ& in the
second equation. If we divide the first of equation (N6.14) into the second and use
the chain rule of differential calculus, we get an differential equation for r in terms of

θ (the so-called “orbit in space”):

1/ 2
1 dr ⎡ ⎛ ε GM ⎞ 1⎤
= ⎢2⎜ + 2 ⎟⎟ − 2 ⎥ . (N6.15)
r 2 dθ ⎣ ⎜⎝ j 2 j r ⎠ r ⎦

In equation (N6.15), all the r’s appear in the denominator. This suggests that
we should transform variables,

1 du 1 dr
u≡ ⇒ =− 2 . (N6.16)
r dθ r dθ

Equation (B16) now becomes

1/ 2
du ⎡ ⎛ ε GM ⎞ ⎤
= − ⎢2⎜⎜ 2 + 2 u ⎟⎟ − u 2 ⎥ (N6.17)
dθ ⎣ ⎝j j ⎠ ⎦

To “complete the square” inside the bracket, we transform variables one last time

GM
u ≡ w+ , (N6.18)
j2

and equation (B16.19) takes the form

dw

(
= − w0 −w 2
2
)
1/ 2
, where w0 ≡
2 1
j2

⎜⎜ 2ε +
G2M 2
j2

⎟⎟ . (N6.19)
⎝ ⎠

Up to an additive constant, which we may absorb by a proper orientation of the

coordinate axis, i.e., where we define the zero of θ, the solution to equation (N6.19)

5
reads

w = w0 cos θ , (N6.20)

as may be verified by substituting equation (N6.20) into equation (N6.19) and


carrying out the differentiation. Collecting all the variable transformations, we may
now write

1/ 2
⎡ 2 j 2ε ⎤
= 2 (1 + e cos θ )
1 GM
where e ≡ ⎢1 + 2 2 ⎥ . (N6.22)
r j ⎣ G M ⎦

Notice that r does not reach infinity if 0 ≤ e < 1 . In such cases, the orbit in

space is closed, reaching minimum r for θ= 0, and maximum r for θ= 180∘. For e =

1, the solution for r extends to infinity as θ→ 180∘from above and below (i.e. the body

enters and leaves the solar system on parallel paths below and above the x-axis. For
e > 1 , the body enters and leaves the solar system at symmetric angles with respect
to the x-axis.
For the first case, we claim that equation (N6.22) is the equation of an ellipse of
eccentricity e and semimajor axis

j2 ⎛ 1 ⎞
a≡ ⎜ ⎟. (N6.23)
GM ⎝ 1 − e 2 ⎠

To see this, multiply equation (N6.22) by r, obtaining after substitution of equations


(N6.1), (N6.11), and (N6.23):

1=
1
a(1 − e 2 )
[ ]
( x 2 + y 2 )1 / 2 + ex . (N6.24)

Getting rid of the square root, we obtain

2ex e2 x2 x2 y2
1− + = + , (N6.25)
a(1 − e 2 ) a 2 (1 − e 2 ) 2 a 2 (1 − e 2 ) 2 b 2 (1 − e 2 )

where we have used the relation in Note (5.6) between the semimajor axis a and the

6
semiminor axis b. Moving all terms to the right-hand side except for the 1, we get

x 2 + 2eax y2
1= + . (N6.26)
a 2 (1 − e 2 ) b 2 (1 − e 2 )

Completing the square in the numerator of the first term on the right-hand side,
bringing the extra term to the left-hand side, and multiplying through by (1 − e 2 ), we
obtain

( x + ea) 2 y 2
1= + 2, (N6.27)
a2 b

which is the equation of an ellipse, with a focus rather than the center placed at the
origin of a Cartesian coordinate system. (QED)
The case e = 1 corresponds to a parabola; the case e > 1, to a hyperbola. We
speak of these cases as not bound (because r can reach infinity for such orbits).
From equation (N6.22), notice that e ≥ 1 corresponds to an orbital energy (per unit
mass) ≥ 0. This is the more general statement: bound orbits are those that have
negative energies; unbound ones are those that have positive energies. The critical
case is ε = 0 , which gives a parabola in the case of a 1/r2 force-law.
In the case of an elliptical orbit, Figure 6.8 shows j to equal twice the rate of
sweeping out area by the radius vector, i.e.,

dA 1
= j. (N6.28)
dt 2

The total area A of the ellipse is obtained by integrating the above over one orbital
period P:

1
A= jP. (N6.29)
2

Kepler showed that the area of an ellipse is given by equation (4.5)

A = πab. (N6.30)

Solving for P in equations (N6.29) and (N6.30), we obtain, after using equation
(N6.23) to eliminate j,

7
2πab 2πa 2 (1 − e 2 )1 / 2 2π
P= = = a3/ 2 , (N6.31)
j [GMa(1 − e )] 2 1/ 2
(GM ) 1/ 2

which is Kepler’s third law, now written for elliptical orbits instead of a circular one.

Note 6.4. For a circular orbit, v = vθ = constant ≡ v 0 . The specific angular

momentum is r0 v 0 = j , while the centripetal acceleration v 0 / r0 must balance the


2

gravitational attraction GM/r02. The latter requires

j2 GM 1 2 GM G 2 M 2 G 2 M 2 G2M 2
r0 = , v0 = , ε0 = v0 − = − = − ,
GM j 2 r0 2 j2 j2 2 j2

with the first and third relations recovering equation (6.4). (QED)
For a circular orbit, the kinetic energy is − 1 / 2 of the gravitational potential
energy, a special case of a more general result known as the virial theorem. The
virial theorem has very important applications in astronomy, applying to gravitational
systems that are statistically in equilibrium, i.e., which looks statistically the same
from one instant to another. The virial theorem states that twice the total kinetic
energy of such a system plus its total gravitational potential energy is equal to zero.
A particle in a circular orbit about a gravitational center of force satisfies the virial
theorem because it is statistically the same from one instant to another.

6.5. Two-Body Problem

The problem solved in Newton’s Motu and in Notes 6.2 and 6.3 is not quite the
problem of the real Sun and one of its planets, because the Sun is not infinitely
massive relative to any planet, and therefore, by Newton’s third law, it should move in
reaction to the planet’s revolution around it. Indeed, as we indicated in Chapter 5,
Newton did not know the value of the AU and therefore he did not know the mass of
the Sun, so the situation could have been worse than it actually turned out to be
(where none of the planets have a mass that exceeds 0.1% the mass of the Sun).
Nevertheless, by the time of the Principia, Newton had dispensed with this difficulty
too (he is a perfectionist, after all). In modern language, he reasons essentially as
follows (see also Note 6.6).
The Sun plus a planet, approximated as a closed system (i.e., ignoring for the
time being, the influence of other planets on a given planet in comparison with the

8
much larger influence of the Sun), requires six coordinate positions as functions of
time to describe the two bodies of the combined system. Three of these can be the
coordinates of the center of mass of the Sun-planet system. The internal forces
between the Sun and the planet cancel and do not affect the motion of the center of
mass, which must therefore travel in a straight-line at constant speed through space.
Constant speed in a straight line does not change the dynamical equations (according
to the Galilean principle of invariance), and therefore we may place ourselves in such
a frame of reference to do the calculations of the relative motion of Sun and planet.
Strictly speaking, we should then use a coordinate system with origin on the center of
mass, follow just the motion of the planet (which requires three coordinate positions)
and calculate the coordinates of the Sun as a scaled negative version of the motion of
the planet. This reduction of the two-body problem to an effective one-body
problem results in the identification of M in all the previous formulae of this chapter

and those in Notes 6.2 and 6.3, not as the mass of the Sun MS, but as the sum of the

mass of the Sun plus the mass of the planet mp:

M = MS+mp . (6.6)

Similarly, the mass m is not the mass of the planet, but the reduced mass of the
two-body problem transformed to an equivalent one-body problem:

mp M S
m= . (6.7)
M S + mp

As a practical consequence of the replacement (6.6), Newton’s derivation of


Kepler’s third law, equation (N6.31) in Note 6.3, becomes

4π 2
P2 = a3. (6.8)
G ( M S + mp )

The proportionality “constant” between the square of the period and the cube of the
semimajor axis is now no longer independent of the planet; it depends on the planet
mass in the term (MS+mp). Since the mass of each of the planets in the solar system
is much less than that of the Sun, mp << MS, the correction for Kepler’s third law is
small, even for a giant planet like Jupiter. However, the reflex motion of the central

9
body is measurable in modern observations of other stars, and it is a major way by
which planets outside of our own solar system are found (Chapter 26).
In the center-of-mass description, where we have a reduced mass m moving in
the central force field of the total mass M, the effective gravitational force is equal to

GMm GM S m p
= , (6.9)
r2 r2

which is the actual gravitational force between the Sun and the planet, with r being
the separation distance between the two, not the distance of one or the other from the
center of mass. This is a source of confusion for beginners. What is the physical
(rather than mathematical) explanation of the replacement (6.6) in a calculation where
the motion of the planet is measured (as Kepler did) relative not to the center of mass
of the system, but relative to the Sun?
Well, relative to the position of the Sun, the planet feels a direct gravitational
acceleration of magnitude,

GM S
(6.10)
r2

pointed toward the Sun. But a coordinate system centered on the Sun is not an
inertial frame (one which moves, at best, at constant vector velocity with respect to
the fixed stars). The Sun is being accelerated by the gravitational attraction of the
planet with magnitude equal to

Gmp
(6.11)
r2

pointed toward the planet. Einstein would later formalize a notion taken from
Galileo’s (thought) experiment from the leaning tower of Pisa known as the Principle
of Equivalence:

There is no way physically to distinguish locally between being in a real field of


gravity and being in an accelerated frame of reference. (6.12)

We will discuss this principle at greater length when we describe general


relativity theory Chapter 13. For the present, let us apply this principle to our
current problem where we have done the calculation (in Notes 6.2 and 6.3) in an

10
accelerated frame of reference. The acceleration is given by equation (6.11) directed
toward the planet. Principle (6.12) says we may replace this “mistake’ by pretending
that it is due to an extra gravitational field in the opposite direction, namely directed
toward the Sun. Thus, the correction amounts to saying that the planet feels two
gravitational accelerations, the direct term (6.10) aimed toward the Sun, and the
indirect term (6.11) also aimed toward the Sun. When we add the two we get an
effective gravitational attraction toward the Sun equal to

G(M S + m p )
(6.13)
r2

which explains why we need to identify the effective mass of attraction at the Sun’s
location for the planet as M = MS+mp.

Note 6.6. In this note we give the standard treatment for the two-body problem,
which does not make explicit use of the principle of equivalence and therefore
transfers easily to problems involving non-gravitational forces. In an inertial
reference frame, let x and X be the vector positions of masses m and M, respectively.
The equations of motion for the two under mutual gravitational attraction read

d 2x GMm d 2x GMm
m = ( X − x), M = (x − X). (N6.32)
x−X x−X
2 3 2 3
dt dt

The right-hand sides are equal and opposite in sign to each other (Newton’s third law
of motion), suggesting that it would be profitable to add the two equations:
d2
(mx + MX) = 0.
dt 2
which we rewrite in the form,
mx + MX d 2R
R≡ , (m + M ) 2 = 0,
m+M dt
with the following physical interpretation. In the absence of any external forces, the
position R of the center of mass moves as if it were a free particle of total mass m+M.
In fact, motion at a constant speed in a straight-line describes how the center of mass
of any number of particles moves if the system is free of external forces.
Of the six independent components of the two vector variables x and X, three
combinations in the components of R behave very simply.. What about the other
three? As the appropriate independent variable for the other three, we choose

r ≡ x − X,

11
which describes the position of m relative to M. Since the origin of r is at X, a
coordinate r centered on M does not constitute an inertial reference frame;
nevertheless, we can make good use of it providing we derive the equation governing
it from valid relationships that are referenced to an inertial frame. Thus, if we
multiply the second relation in equation (N6.32) by m and subtract the result from the
first relation multiplied by M, we get
d 2r GMm( M + m)
Mm 2 = − r, (N6.33)
dt r3
which we may interpret, after cancellation of Mm and multiplication by μ , as the
motion of a hypothetical point particle of mass μ moving in the central force-field
of the combined mass M+m:
d 2r G ( M + m) μ
μ 2
=− r. (N6.34)
dt r3
In the above derivation, we can choose μ ≠ 0 to be anything, since it was arbitrarily
introduced on the left- and right-hand sides of equation (N6.33). This is peculiar to
gravity because of the principle of equivalence. In general, the term − GMm on the
right-hand side of equation (N6.33) would not be available for cancellation of Mm on
the left-hand side. For example, in the corresponding problem involving two electric
charges q and Q of masses m and M, respectively (see Chapter 8), the term − GMm
would be replaced by the product Qq. In this case, as in all other two-body problems,
the only correct choice for μ is the reduced mass that comes from dividing Mm on
the left-hand side of equation (N6.33) by M+m on the right-hand side:
mM
μ≡ . (N6.35)
M +m
In the gravitational problem, μ cancels out from equation (N6.34), which
allows motion to take place in a single plane defined by the joint occupation of the
radius vector r and the instantaneous velocity v = dr/dt. To see this, note that
equation (N6.34) can be written
dv G ( M + m)
=− r;
dt r3
thus, the right-hand side lies in the plane defined by r and the initial v, so there is no
force to change v to come out of this plane. This remains true throughout the entire
orbit, so the entire orbit lies in a single plane. Let us adopt (x,y) as the Cartesian
coordinates in this plane, with origin, not at the center of mass, but at the zero of r
(see Note 6.3):

x = r cos θ , y = r sin θ . (N6.36)

12
The equation of motion (N6.34) written out in Cartesian coordinates now reads
d 2x G ( M + m) d2y G ( M + m)
=− 2 x, =− 2 y, (N6.37)
dt 2
(x + y 2 )3/ 2 dt 2
(x + y 2 )3/ 2
which is the same as equation (N6.2) if we replace equation (N6.4) by
g = G ( M + m) / r 2 . (QED)
If we use the polar coordinates (N6.40), the solution in space now reads (cf. eq.
N6.22):
1 G ( M + m)
= (1 + e cos θ ), (N6.38)
r j2
where the eccentricity e can be regarded as an integration constant, and where the
specific angular momentum j is another integration constant from the equation

r2 = j.
dt
With the equation (N6.38) giving r (θ ) , we may now integrate for the orbit in time:
G 2 ( M + m) 2 θ dθ
t=∫ ,
j 3 π (1 + e cos θ ) 2

where we have arbitrarily set t = 0 when θ = π (apoapse) for reasons that will
become clear shortly. In any case, j can be expressed as (cf. eq. N6.23):
1/ 3
⎛ P ⎞
j = G ( M + m)a(1 − e ) = [G ( M + m)] ⎜ ⎟ 1 − e2 ,
2 2/3

⎝ 2π ⎠

where we have used a = [( P / 2π ) 2 G ( M + m)]1 / 3 ; consequently, we have


2π t dθ
P
(
= 1 − e2 )
3/ 2 θ
∫0 (1 + e cos θ ) 2
≡ Fe (θ ). (N6.39)

The integral function Fe (θ ) must satisfy Fe (2π ) = 2π since θ = 2π when t


= P. Because this has to hold for all e, we suspect that it must be possible to perform
the integration analytically. The trick to doing so is to define the transformation,
1 1 + e cos χ
= . (N6.40)
1 + e cos θ 1 − e2
which makes θ = 0 correspond to χ = π , and θ = π to χ = 0. Thus, θ and
χ both have cycles of 2π . This allows the ability to express sin θ relatively
easily in terms of sin χ . The reason that this is helpful becomes apparent when we
take the differential of equation (N6.40) and get
sin θdθ sin χdχ
=− .
(1 + e cos θ ) 2
1− e2
Now, we are motivated to solve equation (N6.40) for cos θ :

13
1 ⎛ 1 − e2 ⎞ e + cos χ
cos θ = ⎜⎜ − 1⎟⎟ = − . (N6.41a)
e ⎝ 1 + e cos χ ⎠ 1 + e cos χ

If we square the above equation, we may write


e 2 + 2e cos χ + cos 2 χ
1 − sin 2 θ = .
1 + 2e cos χ + e 2 cos 2 χ
Solving for sin 2 θ , we get
(1 − e 2 )(1 − cos 2 χ ) sin 2 χ sin χ
sin 2 θ = = (1 − e 2
) ⇒ sin θ = − 1 − e 2 .
(1 + e cos χ ) 2
(1 + e cos χ ) 2
1 + e cos χ
Collecting expressions, we now have
dθ (1 + e cos χ )dχ
= .
(1 + e cos θ ) 2
(1 − e 2 ) 3 / 2
This relation allows us to perform the integration in equation (N6.39) analytically:
2π t χ
= ∫ (1 + e cos χ ) = χ + e sin χ . (N6.41b)
P 0

Equation (N6.41) demonstrates explicitly that t advances by P when χ (and θ )


advances by 2π . Equations (N6.41a) and (N6.41b) yield a parametric solution
through χ for t as a function of θ (or vice-versa). The parametric variable χ is
given the picturesque name of the (orbital) anomaly in celestial mechanics. The
name arises because when e is nonzero, equation (N6.41b) shows that χ does not
advance at a uniform rate with respect to t, in violation of Plato’s preconceptions
about the nature of Heavenly motion.

Figure 16.2. (Top) The wobbly proper motion of Sirius. (Bottom) Schematic
depiction of the orbital motion of Sirius and its companion after removal of the
straight-line motion of the center of mass.

14
Note 16.3. Note 6.6 for the gravitational two-body problem gives the center of mass
position R and relative dispacement r in terms of the vector positions x and X of
masses m and M in an inertial frame of reference:
mx + MX
R= , r = x − X.
m+M
If we solve these relations for x and X, we get
M m
x=R+ r, X = R − r,
M +m M +m
where R = V (t − t 0 ) describes a straight-line path through space at constant velocity
V and the tip of relative displacement vector r traces out an elliptical orbit in time.
Therefore, x and X trace out smaller ellipses about the straight-line motion of the
center of mass, as drawn in Figure 16.2. The center of mass is a focus for the two
latter ellipses, which are smaller than the relative ellipse by the complementary mass
fractions.
Bessel could only see the position X of Sirius A as a function of time in the
plane of the sky. Denote the unit vector in the direction line of sight to the system by

k̂ and the dstance to Sirius by d; then what Bessel could measure was kˆ × X(t ) / d ,

the angular motion of X(t) across that line of sight. This angular motion is

composed of two parts; the straight-line contribution from kˆ × R (t ) /d, which he

could remove (thereby determining the rate of proper motion of the system kˆ × V / d ) ,

and the wobble contribution from kˆ × rM (t ) / d , where r M ≡ − m /( M + m)r (t ) is the

15
position of Sirius A relative to the center of mass. This position rM (t ) generally
traces out in time an ellipse that does not lie in the plane of the sky. Suppose the
fixed normal n̂ to that plane makes an angle i relative to our line of sight, i.e.,

kˆ ⋅ nˆ = cos i. The orbital ellipse rM (t ) projects into the plane of the sky also as an

ellipse, but the projected ellipse does not have the center of mass as a focus. Thus,
the observer needs to deproject the observed ellipse until the deprojected ellipse does
have the center of mass as a focus. (The center of mass remains fixed in any
deprojection.) This gives the inclination i and the true eccentricity e of the binary
orbit. The semi-major axis a M = ma /( M + m) traced by Sirius A in its elliptical
motion about the center of mass can then be measured since Bessel knew the distance
d (by trigonometric parallax) so that angular displacements can be converted to linear
displacements. From the requirement that the known orbital period P satisfies

P 2 = 4π 2 [( M + m)a M / m]3 / G ( M + m) = [4π 2 a M / G ][(M + m) 2 / m 3 ],


3

we can now deduce the mass m of Sirius B, but only if we know the mass M of Sirius
A. The latter was not known in Bessel’s time. However, once Sirius B was also
found, then the ratio m/M could be determined from the relative displacements of A
and B from the center of mass. This information added to the above determination
for the combination m 3 /( M + m) 2 gives M and m separately without making any
assumptions about the physical or astronomical nature of the two stars. From this
discussion we see that good observational data on just one, or better two, of the
components of a visual binary can provide invaluable information about the masses of
stars, which can then be correlated with their other physical properties. For this
reason, visual binaries constitute a fundamental database for stellar astronomy.

Note 16.16. From Note 6.3, we may derive the velocity of the star M in a binary
system by differentiation as
& =R
X & + r& ,
M

where r&M = − m /( M + m)r& is the velocity of M relative to the center of mass. By


spectroscopic technique, the observer is able to measure only the line of sight
& , which has a steady component kˆ ⋅ R
component kˆ ⋅ X & due to the center-of-mass

motion at constant velocity and a time-variable component kˆ ⋅ r&M (t ) due to the

orbital motion of M relative to the center of mass. From Note 6.3, we may obtain for
the two components of the velocity r& of m relative to M:

16
j G ( M + m)
v θ = rθ& = = (1 + e cos θ ),
r j
G ( M + m)
v r = r& = e sin θ ,
j
where r and θ are the distance and angle of the primary with respect to the
secondary, with θ = 0 defining the direction of least separation (periapse). In the
above, we have made the Newtonian substitution M → M + m , and j is the angular
momentum per unit mass of the fictitious reduced mass. To get rid of j, we use the
formula (N16.23),
1/ 3
⎛ P ⎞
j= G ( M + m ) a (1 − e ) = ⎜ 2
⎟ [G ( M + m]
2/3
1 − e2 ,
⎝ 2π ⎠
where we have used equation (N16.31) for the revolution period P. Thus, the two
velocity equations become

v r = eV0 sin θ , v θ = V0 (1 + e cos θ ),

where V0 is the velocity scale:

⎡ 2π
1/ 3
⎤ 1
V0 ≡ ⎢ G ( M + m ) ⎥ .
⎣P ⎦ 1 − e2

To convert the two velocity components relative to the center of mass, let us
first resolve them into Cartesian components:
v x = v r cos θ − v θ sin θ = −V0 sin θ ,

v y = v r sin θ + vθ cos θ = V0 (e + cos θ ) .

The velocity of the star M relative to the center of mass is then given by
⎛ m ⎞ & ⎛ mV0 ⎞ ˆ
r&M = −⎜ ⎟r = ⎜ [
⎟ e x sin θ − eˆ y (e + cos θ ) . ]
⎝M +m⎠ ⎝M +m⎠

The z-axis makes an angle i with respect to the viewing direction k̂ ; therefore, we

suppose that k̂ has the following decomposition in the (x, y, z) frame:

kˆ = eˆ z cos i + sin i (eˆ x cos φ + eˆ y sin φ ),

with φ being a constant angle if the line of apsides (i.e., the x-y axes) does not rotate
with time. With sin θ cos φ − cos θ sin φ = sin(θ − φ ), the line-of-sight Doppler
velocity of star M now reads

kˆ ⋅ r&M = V [sin (θ − φ ) − e sin φ ], (N16.31)

17
where V is the velocity scale,

⎛ m sin i ⎞
1/ 3
mV0 sin i ⎡ 2π ⎤
V ≡ =⎢ 2 ⎥ ⎜ ⎟. (N16.32)
M +m ⎜ ⎟
⎣ G ( M + m) P ⎦
2
⎝ 1− e ⎠
2

The angle θ varies with time t with a period of P; this variation of θ is


known from the parametric solution for the orbit in time, which is derived in Note 6.6:
2π t e + cos χ
= χ + e sin χ , cos θ = − .
P 1 + e cos χ
The important point for us here is that the shape of the variation with time allows one

to determine the eccentricity e from the measured Doppler velocity kˆ ⋅ r&M of the

visible star. The same measurements yield V, from which we can derive a value for
the combination m sin i /( M + m) 2 / 3 knowing the orbital period P.
Note that in comparison with the case of a single visible component in a visual
binary (see Note 16.3), the method requires no knowledge of the distance d to the
source, a considerable advantage. On the other hand, we are worse off here in the
measurement of m /( M + m) 2 / 3 by an indeterminate sin i. For single-lined
spectroscopic binaries that are not also visual binaries (which requires high-accuracy
positional-astrometry measurements for distant systems that may become available
only from interferometry carried out in space), we can determine the smallest possible
value for m by assuming that sin i = 1, provided we can estimate M from other
astrophysical information. For application to Cygnus X-1, this method yields m = 7

M⊙ as the smallest possible value for the companion’s mass when we assume that the

primary star has a mass equal to 30 M⊙.

The method of single-lined spectroscopic binaries has recently been applied by


Michel Mayor and Geoff Marcy and their colleagues with spectacular success to the
reflex motion of stars with planets orbiting them. In the extrasolar planets
application, we can assume that m << M . Therefore, astrophysical knowledge of
the star’s mass M yields m sin i for the extrasolar planet. The eccentricity e of the
orbit is found as before. Moreover, the orbital period P yields the semi-major axis a
of the orbit since we are in the regime where Kepler’s third law in the form
P 2 = 4π 2 a 3 / GM is a good approximation. Thus, familiarity with Newtonian
dynamics and very accurate measurements of a star’s periodic Doppler variations
(with amplitudes of only meters per second) allow astronomers to deduce many
interesting things about planetary companions around stars without ever having seen
the planets directly!

18
In double-lined spectroscopic binaries, both stars are visible. In that case, we
can determine both m sin i /( M + m) 2 / 3 and M sin i /( M + m) 2 / 3 . To get m and M
separately, we still need sin i. The simplest case arises when the binaries also eclipse,
in which case sin i must be close to 1. The shape of the eclipse light-curve then
allows a modeling for the radii of one or both stars. Such binary systems therefore
provide an invaluable database for the empirical properties of stars.

19

You might also like