D:DTFL581LB.DVI

Eur. J. Phys.

(2000) 1–18. Printed in the UK

PII: S0143-0807(00)14581-7

Regular Keplerian motions in classical
many-body systems

Eugene I Butikov

St Petersburg State University, St Petersburg, Russia

E-mail:

butikov@spb.runnet.ru

Received 30 May 2000

Abstract.

A clear and simple physical approach to the explanation of exact particular solutions

of the classical many-body problem is suggested. When the motion of individual bodies coupled
by mutual gravitational forces in a many-body system occurs along conic sections, each body can
be treated as moving not under the pull of the other moving bodies, but rather under a stationary
central inverse-square gravitational field. These solutions describing possible amazingly simple
(Keplerian) many-body motions are illustrated by computer simulations. Some pedagogical and
philosophical aspects of the problem are discussed.

1. Introduction: simple and complicated motions of celestial bodies

Exact analytic solutions to the differential equations of motion are remarkable for the simplicity
of motions described by these solutions. In particular, the classical Kepler problem of a
body in the Newtonian inverse-square central gravitational field (and of two interacting bodies
coupled by mutual gravitation) allows such analytic solutions predicting simple motions along
conic sections. Unfortunately, exact solutions are seldom encountered in physics. When
there are perturbing interactions (gravitational forces produced by other bodies, deviations of
mass distributions from exact spherical symmetry, additional non-gravitational forces, etc),
the equations of motion become non-integrable. The mathematical analysis of such perturbed
motions is immensely complicated. The marvel of closed orbits that are found in Keplerian
motion, as well as their wonderful simplicity, vanishes.

For example, the distortion of the Earth’s gravitational field from spherical symmetry

causes the actual orbit of a satellite to differ from an ellipse. The real trajectory is a complex
curve, generally not closed and not lying in a plane. After a revolution, the satellite does
not return to the same spatial point. The trajectory of an equatorial satellite looks like a
multi-petalled flower whose leaves gradually fill out the annular region enclosed between the
two concentric circles. For a hypothetical oblate planet with a somewhat exaggerated axial
distortion, a possible trajectory of a satellite is shown in figure 1.

When the perturbing forces are small compared to the main gravitational force, one can

use approximate analytic methods. Keplerian motion, whose distinctive feature is the striking
mathematical simplicity of possible trajectories, can be assumed in this case as the zeroth
approximation to the actual motion. We can consider that the small perturbations cause
relatively slow variations of the parameters characterizing the corresponding Keplerian orbit,
and try to obtain analytic expressions for these slow variations. Lagrange has called such an
ellipse with varying parameters, grazing the actual trajectory, the

osculating orbit

. In figure 1

osculating ellipses are shown (by thin lines) for three points (

and

) of the trajectory.

0143-0807/00/000001+18$30.00

Regular Keplerian motions in classical many-body systems

Figure 2.

Trajectory of a satellite orbiting in turn a planet and star in the heliocentric (left) and in

the geocentric (right) frames of reference. Identical numbers refer to the same instants of time.

Chaotic behaviour of a nonlinear system (governed by simple deterministic laws) which

we observe in this example is related to the extreme sensitivity of the differential equations
describing the system to the initial conditions: a very small initial difference may result in an
enormous change in the future state and long-term behaviour of the system. Celestial dynamics
gives one of the numerous examples of chaos in physics. We may suppose that in this case
the absence of an analytical solution reflects probably the complexity of the possible motions
of the system rather than the weakness of the analytic capability of the mathematics. Many
examples of such complex motions can be found in the simulation programs of the package

Planets and satellites

[1] developed recently by the author. These simulations allow us to

observe and study many fascinating phenomena that can occur in a system of three or more
bodies attracted to one another by gravitational forces. Their motion delights the eye and
challenges our intuition.

For the planets of our solar system, Kepler’s laws give a good zeroth-order approximation

because the masses of planets are small compared to the mass of the Sun, and the planets are
separated from one another by large distances. That is, with good precision we can neglect
the forces of gravitation between the planets and consider their motion to be governed only
by their attraction to the Sun. Hence, because of the structure of our solar system, the motion
of each of the planets is rather simple. But in a double-star system the motion of a planet can
be very complicated. Many stars in our Galaxy are multiple systems—double and triple stars,
unlike the Sun, which is a single star. In a multiple-star system, stable planetary orbits are also
possible and it is conceivable that a community of animated, thinking creatures could arise on
such a planet. Because trajectories of planets in a double-star system are very complicated, it
would be an immensely difficult problem for astronomers among those creatures to establish
the kinematical laws of planetary motion in the double-star system (Kepler), and even a much
more difficult problem would be to discover that these complex kinematical laws are generated
by the simple inverse-square law of gravitational attraction to each of the stars (Newton). Our
civilization has advantage of a planet orbiting a single star. Mankind has been lucky to travel
so fast along the thorny road of knowledge.

Moreover, we can even suppose without irony that Newton’s law of gravitation would

have been more difficult to discover if Kepler’s model of planetary motion had been quickly
invalidated by more accurate astronomical observations than those of Tycho Brahe. We can
regard this page in the history of science as a surprising piece of evidence that a greater precision
of the experimental data obtained too early can impede the scientific progress.

E I Butikov

The three-body problem is frequently mentioned in various intermediate mechanics texts

as an example of extreme complexity of possible motions generated by simple and precise
physical laws. However, since Lagrange it is well known that in generally unsolvable many-
body problems there exist several particular solutions describing simple Keplerian motions
of the bodies. It may seem a real wonder that such an unexpectedly simple finite subset of
motions falls out of the continuous set of tremendously complex general three-body motions.
And these simple solutions certainly should allow an equally simple physical explanation.

There is a serious lack of relevant information regarding these simple solutions in standard

texts on general physics and even on celestial mechanics. Several papers on the subject
published during the last decades (see, for example, [2]) deal only with the triangular libration
points in the restricted circular three-body problem. In serious advanced courses of celestial
mechanics the simple result concerning the Lagrangian points is presented in an enormous heap
of extremely complex formulas and therefore leaves little aesthetic satisfaction: we expect that
simple and beautiful results certainly deserve simple and clear ways of their derivation.

In this paper we suggest a simple physical explanation for exact particular solutions to

the many-body problem. We claim that in all cases in which the motion of individual bodies
(coupled by mutual gravitational forces in a many-body system) occurs along conic sections,
each body can be treated as moving not under the pull of the other moving bodies, but rather in
a stationary central gravitational field whose strength is inversely proportional to the square of
the distance of the body from the centre of mass of the system. Under certain conditions this
effective gravitational field can be stationary in spite of the fact that it is created by moving
bodies. We illustrate this idea first by the simplest example of the two-body problem.

2. The two-body problem

In most textbooks on mechanics the two-body system generally is not referred to as a many-
body system because the problem of relative motion of two interacting bodies (irrespective of
the physical nature of the interaction) mathematically is equivalent to the problem of motion
of a virtual single body with the reduced mass

/(m

)

under a stationary

central force equal to the force of interaction between the actual bodies. The solution to this
problem actually describes the motion of one body relative to the other. Under the inverse-
square gravitational force of interaction, this relative motion of the bodies occurs along a conic
section and obeys Kepler’s laws. Knowing that if one body moves, say, in an ellipse about the
other (a binary star), we can show that they both move synchronously in homothetic ellipses
about the centre of mass of the whole system. As the bodies move, they are always at the ends
of a rotating straight line that passes through the common focus of their orbits located at the
stationary centre of mass (figure 3). The linear dimensions of these similar elliptical orbits are
inversely proportional to the masses of the bodies.

Therefore, the concept of the reduced mass enables us to regard the two-body problem as

one-body. This traditional approach, being quite correct and mathematically simple, may seem
especially amazing to some students and cause confusion since it allows us to treat the non-
inertial frame of either of the bodies as inertial. The explanation of this apparent inconsistency
is all too subtle for most students who study physics at an introductory level. Moreover, the
business of transforming from one frame to another in this case can be also rather confusing.
(After all, both Copernicus and Galileo had difficulty getting the world to accept such ideas.)

However, dealing with the two-body problem, it is possible to use a somewhat different

approach that is free of the difficulties mentioned. We consider the motion of each body in the
inertial centre-of-mass frame of reference. Since the force of gravity between the bodies lies
at each instant along the line joining the bodies, the force vectors are always directed through
the centre of mass. In order to explain why the motions of the bodies relative to the centre of
mass obey Kepler’s laws and occur along conic sections, it is sufficient to show that each of the
bodies coupled by mutual gravitation can be treated as moving not under the pull of the other
moving body, but rather in a stationary central gravitational field whose strength diminishes

Regular Keplerian motions in classical many-body systems

Figure 3.

Trajectories traced by components of a double star in the centre-of-mass reference frame.

The simultaneous positions of the bodies are marked by the same numbers.

as the square of the distance from the centre of mass. The source of the field (located at the
stationary centre of mass) is characterized by some effective mass

eff

Indeed, let

and

be the radius vectors denoting momentary positions relative to the

centre of mass of the bodies with masses

and

respectively. Then

and

[1 +

]

. Therefore, in the formula for the gravitational force

exerted

on the first body, we can express the distance

)

between the bodies in terms of the sole

distance

of the first body from the centre of mass:

= −

)

= −

)

eff

)

eff

3
2

)

(1)

Thus, in the inertial centre-of-mass frame, the motion of the first body in the two-body

system should be exactly the same as in a central inverse-square gravitational field created by
a stationary source of mass

)

eff

(in the absence of the second body). As we know, such a

motion obeys Kepler’s laws.

Similar considerations can be applied to the other body of the two-body system: the

gravitational influence of the first body can be replaced by the stationary source of an effective
mass

)

eff

3
1

/(m

)

. It remains only to prove that these Keplerian motions of both

bodies occur synchronously along homothetic (closed or open) complanar orbits whose linear
dimensions are inversely proportional to the masses of the bodies. This follows immediately
from the equation

0, which holds for an arbitrary moment during the motion.

In another inertial reference frame (that is, as seen from aside) the bodies of a two-body

system move non-uniformly along complicated wavy or looped trajectories. This apparent
complexity is generated by the superposition of their rather simple periodic Keplerian elliptical
motions around the centre of mass and quite simple uniform rectilinear motion alongside the
centre of mass. Figure 4 shows an example of such trajectories.

3. A collinear symmetric three-body system

We start our analysis of exact particular solutions to the many-body problem with the simplest
possible example. When a pair of heavy bodies have equal masses, the restricted three-body
problem has an evident exact solution, provided the third body (of negligible mass) is placed
exactly halfway between the members of the pair (at the centre of mass of the system), and
provided its velocity relative to the centre of mass is exactly zero. Then, since the gravitational

E I Butikov

Figure 4.

Trajectories of the components of a binary star as seen from aside (in an arbitrary inertial

reference frame). The simultaneous positions of the bodies are marked by the same numbers.

forces exerted on the central body due to each member of the pair are equal and opposite, the
central body remains at the centre of mass. This simplest particular solution is a special case
of the collinear interior Lagrangian equilibrium point

(see below).

This solution holds for arbitrary (not only circular) motions of the massive bodies,

including cases in which they trace synchronously congruent elliptical orbits (and open
parabolic or hyperbolic trajectories) with the centre of mass as their common focus. Moreover,
the existence of this exact particular solution is almost evident also in the case when the central
body has a finite mass. That is, the solution holds for the unrestricted three-body problem, say,
for the special case of an imaginary system of two planets of equal masses that orbit a single
star.

Indeed, let two massive planets (of equal masses) initially lie on the same straight line

with the star at the mid-point. If the planets have equal and opposite initial velocities (in the
frame of reference of the star), this configuration of the three bodies is preserved during their
further motion. The only difference from the case mentioned above is that the third body of
an arbitrary mass, being placed halfway between the bodies of equal masses, influences their
motion because of an additional gravitational pull: the net force on either one of the pair is
the sum of two gravitational forces (pointing in the same direction), one from the central body
and the other from the other member of the pair. However, the net force on either body of the
pair, until the symmetric configuration is violated, is in this case also inversely proportional to
the square of its distance from the central body (from the centre of mass of the system).

Therefore we can consider any of the planets to move in a

stationary

Newtonian inverse-

square gravitational field whose source is located at the centre of the star. The effective mass
of the stationary source

eff

is somewhat greater than that of the star by virtue of the additional

gravitational pull of the other planet. In this effective gravitational field the planet traces a
closed Keplerian ellipse. The second planet moves in an equivalent effective gravitational field
and traces synchronously a congruent ellipse. Although this exact solution is of no practical
importance, its existence is interesting in principle and deserves discussion.

Figure 5 shows the simulation of possible simple motions in a system of two planets of

equal (arbitrarily large) masses. Initially the planets

and

are on the same straight line

with the star

, at equal distances on opposite sides of the star. The planets have equal and

opposite initial velocities (in the heliocentric frame of reference, shown in the left-hand side
of the figure). We see that in this symmetric configuration the motion of the system is regular
and very simple. The star is stationary, while the massive planets trace closed orbits that are
congruent ellipses with the common focus at the centre of the star. At any moment the planets
are at the opposite ends of the straight line passing through the centre of the star, and their
velocities are equal and opposite.

Regular Keplerian motions in classical many-body systems

Figure 5.

Simple periodic motions described by an exact particular solution to the three-body

problem for a symmetric configuration of two identical planets.

The unperturbed ‘heliocentric’ elliptical orbits that each of the planets would trace in the

absence of the other planet under the gravitational pull of the star are shown by thin lines in
the left-hand side of figure 5. These osculating orbits that graze the actual elliptical orbits
of the planets (thick lines) are shown for perihelia

and

(only portions of the ellipses)

and for points

and

that are closer to aphelia (whole ellipses). The right-hand side of

figure 5 shows the trajectories of the star

and planet

in the reference frame of planet

(in

a somewhat smaller scale).

The net force exerted on each of the planets is formed by addition of the forces of

gravitational attraction to the star and to the other planet. This resulting force is always
directed toward the centre of the star, and its magnitude is inversely proportional to the square
of the distance from the star:

(

m(M

)

(2)

Here

is the gravitational constant,

is the mass of the star,

is the mass of either

of the planets, and

is the distance from the star to either of the planets. It follows from

equation (2) that in the symmetrical configuration the motion of each of the planets occurs
along a Keplerian ellipse, as if this motion were governed solely by an effective stationary
central Newtonian gravitational field whose source is characterized by an effective mass of

With the help of Newton’s second law, we can easily calculate the velocity of the planets

for the special case of circular orbits. Equating the force given by equation (2) to the product
of the planet’s mass

and the centripetal acceleration

, we obtain for the velocity

the planet in the circular orbit of a radius

(3)

The period of revolution of the planets along such circular orbits is found by dividing the

length of the orbit 2

πr

by the circular velocity

πr

G(M

)

(4)

This expression is a generalization of Kepler’s third law for the special case of the planetary

motion under consideration. Equation (4) is equally valid for elliptic motions of the planets
provided we replace the radius

with the semimajor axis

of the elliptical orbit.

E I Butikov

The symmetric configuration of the system is preserved during the motion provided the

initial velocities of the planets relative to the star are exactly equal and opposite. If the velocities
differ slightly in magnitude or direction, or the distances from the star to the planets are not
exactly equal, or the three bodies do not lie exactly on the same straight line, the paths of
the planets sooner or later deviate from Keplerian ellipses, and these deviations progressively
increase. Hence the periodic motion described by this exact particular solution of the three-
body problem is unstable. Eventually the motion of the system becomes irregular and very
complicated.

4. A ‘round dance’ of identical planets

Similar periodic exact solutions in which the bodies trace closed Keplerian orbits exist for
systems of several bodies of equal masses surrounding a central body. Let

bodies (‘planets’)

of arbitrarily large but equal masses be located at all

vertices of a regular (equilateral) polygon,

and one more body (a ‘star’ whose mass can differ from the masses of the other bodies) be
located at the centre of the polygon. In this symmetric configuration the central body is in
equilibrium under the joint gravitational pull of all other bodies. The resulting gravitational
force exerted on any of the other bodies (on a ‘planet’) by the central body and by the other
planets is directed toward the centre, and its magnitude is inversely proportional to the square
of the distance from the centre (or, which is the same, to the square of a linear dimension of
the polygon, e.g., of the length of its side).

Therefore identical ‘planets’ in an equilateral configuration can trace congruent Keplerian

ellipses (or even open parabolic or hyperbolic trajectories) with the common focus at the ‘star,’
provided the initial velocities of the planets are equal in magnitude and make equal angles
with the corresponding radius vectors of the planets. The symmetric polygonal configuration
of the system is preserved during the motion (figure 6).

In particular, the ‘planets’ can move uniformly at equal distances from one another along

the same circular orbit (circumscribed about the polygon). In this case the polygon, with
the planets at its vertices, rotates uniformly about its centre. For elliptical trajectories of the
planets, the angular velocity of the polygon is greatest when the planets pass simultaneously
through the perihelia of their orbits. In this non-uniform rotation of the polygon, the lengths
of its sides vary periodically.

The upper part of figure 6 shows examples of these exact solutions for systems of three (left)

and four ‘planets’ (right). Moving along elliptical trajectories, at any moment the bodies are
at the vertices of a regular triangle and a square respectively. Thin lines show the unperturbed
orbits that the ‘planets’ would trace in the absence of the other planets under the gravitational
pull of the ‘star’ (about the centre of mass of the two-body system consisting of the star and
the single planet). These osculating orbits are shown for the perihelia of the actual orbits and
for the moments at which the ‘planets’ pass through the points marked by small circles.

The lower part of figure 6 shows similar systems of six and eight ‘planets’ of equal

masses orbiting the ‘star’ in symmetric equilateral configurations. The regular polygon (at
whose vertices the ‘planets’ are found) rotates non-uniformly, and the lengths of its sides vary
periodically during the rotation. The osculating ellipses shown by thin lines correspond here
to the unperturbed orbits of individual ‘planets’ in the frame of the star (rather than in the
centre-of-mass frame).

5. Keplerian motions in the triangular and square equilateral configurations

We note that in the exact solutions to the many-body problem considered above, the mass of
the central body can be zero. That is, a system of

bodies of equal masses located at the

vertices of a regular

-sided polygon, under their mutual gravitational attraction, can perform

a beautiful ‘round dance’ even in the absence of a central body. In particular, three bodies

Regular Keplerian motions in classical many-body systems

Figure 9.

Regular motions of three bodies of unequal masses in the equilateral configuration.

Figure 10.

Transition to an irregular motion in the system whose initial motion is close to that

described by the exact solution.

An example of such a simple periodic motion is shown in figure 9 (

). In the inertial centre-of-mass frame (left) the bodies trace homothetic elliptical

orbits of different sizes and orientations. In the ‘heliocentric’ frame associated with the body

of greatest mass (right-hand side of the figure), the bodies

and

trace the congruent ellipses

shown by thick lines. The major axes of these ellipses form an angle of 60

◦

. The thin lines

show the (non-congruent) heliocentric osculating orbits that each of the bodies

and

would

have traced around

in the absence of the other body (for the moment at which

and

pass

through the aphelia of their orbits).

This regular periodic motion of the three bodies is unstable with respect to (small)

variations in the initial conditions that disturb the symmetry of the system. This instability of
motion in the initially equilateral configuration of the bodies is illustrated by figure 10.

E I Butikov

6. Lagrangian points in the circular restricted three-body problem

In the special case of zero mass of one of the three bodies moving in the equilateral
configuration, we arrive at the triangular libration points of the restricted three-body problem,
often mentioned in advanced mechanics texts and discussed in several publications [2]. It is
widely known that if two massive bodies orbit each other in circles, there exist five positions
at which an infinitesimal test body may be placed so that it orbits circularly about the centre of
mass of the system in the same plane and at the same angular speed as do the massive bodies.
That is, the whole system rotates rigidly, as if the three bodies were the points of a solid rotating
uniformly about the centre of mass of the system. In other words, in the rotating reference
frame associated with the line joining the primaries, the test body of a negligible mass is in
equilibrium at any of these positions. These five positions are called circular libration points
(or Lagrangian points) of the restricted circular three-body problem. We note that Lagrangian
equilibrium points are formed by the combined gravitational forces of both massive bodies
and centrifugal force of inertia.

Three of the libration points

and

are located on the line passing through the

massive bodies (one point

between the bodies). They are called collinear libration points.

Each of the other two points

and

(triangular libration points) is located at the apex of an

equilateral triangle whose base is formed by the segment joining the primaries.

The stability in the motion in the vicinity of the Lagrangian triangular libration points in

the restricted planar circular three-body problem remained a subject of intense investigation in
celestial mechanics for more than two centuries. It was found (see, for example, [3]) that the
triangular libration points are stable for the mass ratio

/(m

)

(where

< m

)

satisfying the following condition:

µ(

−

µ) <

with the exception of three particular values (0

0243

. . . ,

0135

. . .

, 0

0109

. . .

). That is,

the triangular libration points are stable if the mass of one of the massive bodies is much
smaller than that of the other (if the ratio

does not exceed approximately 0.04). In the

Earth–Moon system

0123,

01215, and so its triangular libration points are

stable.

In the solar system, stable triangular libration points are also formed by the combined

gravitational forces of the most massive planet, Jupiter, and the Sun. There are two groups of
asteroids (named Greeks and Trojans) that are trapped at Jupiter’s leading and trailing triangular
Lagrangian points and move around the Sun synchronously with the planet. The discussed
exact solutions are of some practical interest in space dynamics because of the possibility (even
if only in principle) of launching a stationary satellite located at one of the Lagrangian points
in the Earth–Moon system.

The motion at collinear libration points is always unstable. One of these points lies

between the massive bodies. If their masses are equal, this point is just halfway between
the bodies. This case is discussed above (see figure 5). For a system with

(figure 11), this interior libration point is displaced from the centre of mass towards the lighter
body

by 0.237 of the distance

between the bodies. Its distance

from the lighter body

is approximately 0.43

, while the distance

from the heavier body

is 0.57

. In

this position the resulting force of gravitational attraction by the bodies

and

is directed

towards the centre of mass, and its magnitude is just sufficient to provide the satellite

with the

centripetal acceleration necessary for circular motion about the centre of mass with the same
angular velocity as that of the uniform rotation of the line

joining the massive bodies. Thus,

the rectilinear configuration of the system is preserved during the motion. The simultaneous
positions of all the bodies are marked in figure 11 by equal numbers.

From the point of view of an observer on the heavier body

(see the right-hand side

of figure 11), the lighter celestial body

, moving around

in a circular orbit, is continually

eclipsed by satellite

, since the visible position of

always coincides with that of

. Similarly,

an observer on

perceives the situation as a uniform revolution of

about himself in a circular

Regular Keplerian motions in classical many-body systems

Figure 11.

Motion of the satellite at the interior collinear libration point.

orbit of radius 0.43

. This revolution visually coincides with the revolution of the celestial

body

around

The ellipses in the right-hand side of figure 11 show the osculating orbits that the satellite

would trace around

were to suddenly vanish. (The first ellipse corresponds to the initial

moment, and the second ellipse to the moment when the satellite is at the point

.) Indeed,

the circular velocity of the unperturbed orbital motion around

is greater than the circular

velocity of the actual motion, when the satellite is also subjected to the gravitational pull of
the other body

. This additional pull reduces the centripetal acceleration of the satellite, and

thus a smaller velocity is required for the circular motion.

Two collinear libration points lie outside the segment

joining the massive bodies. For

the system with equal masses (

) these points are located symmetrically at a distance

of 1.198

from the centre of mass, that is, at a distance of 0.698

beyond either of the

bodies. If

< m

, one of the outer points is located closer to

. For

2 its

distance from the centre of mass is 1.249

, so that this point of libration is separated from

the lighter body

by a distance of 0.582

. The opposite collinear libration point is located

at a distance of 1.136

from the centre of mass, so that its distance from the heavier body

is 0.803

For the Earth–Moon system, the distance of the interior libration point from the Moon is

approximately 58 000 km, or 0.15 of the mean distance

between the Earth and the Moon

(384 400 km). The distance of the exterior point from the Moon is 65 000 km, or 0.17

. The

third collinear libration point lies on the opposite (with respect to the Moon) side of the Earth.
Its distance from the Earth is 380 600 km, or 0.993

The motion of a satellite in any of the collinear libration points (as well as the relative

equilibrium in the rotating frame of reference) is unstable. An interesting example of instability
of the interior libration point in the Earth–Moon system is illustrated in figure 12, which shows
the motion of the system in the frame of reference associated with the Earth

. The initial

position of the satellite

is very close to the libration point. If the Moon

were absent,

the satellite (whose initial velocity is zero in the rotating frame) would have moved in the
gravitational field of the Earth along an ellipse, grazing the actual circular orbit at the initial
point

. This osculating ellipse is shown by a thin line in figure 12. The additional gravitational

pull of the Moon causes the satellite to move in a circle. One more osculating ellipse is shown
for the point

of this circular orbit.

The satellite moves with the whole system in close proximity to the libration point only

during approximately one revolution (for the given initial displacement from the libration

E I Butikov

Figure 12.

Motion of the Earth–Moon satellite with repeating protracted visits to the interior

collinear libration point. The dashed circle around the Moon shows the sphere of gravitational
action of the Moon.

point). At the end of this revolution it leaves the vicinity of the libration point and becomes
an independent satellite of the Earth

. Its orbital motion around the Earth occurs in almost

elliptical orbit

(lying inside the first osculating ellipse) which is slightly perturbed by the

Moon. After several revolutions of the Moon, the satellite and the Moon occur in a close
vicinity, and the gravitational field of the Moon tears the satellite away from its geocentric
orbit. The satellite is again trapped at the same libration point, and for the next half-revolution
of the Moon moves near this point synchronously with the whole system. Then the satellite
once more escapes this unstable libration point, and again orbits the Earth independently of the
Moon along another nearly elliptical orbit

. But after several revolutions the gravitation pull

of the Moon once more returns the satellite to the vicinity of the libration point, thus making
it move again during some time synchronously with the Moon. Such protracted visits of the
satellite to the libration point may repeat many times. At the final moment of the simulation
shown in figure 12, the Moon and the satellite are found at points

and

respectively.

7. Non-circular motions at the libration points

It is not widely known that exact solutions to the restricted three-body problem also exist
for non-circular motions. When the two heavy bodies move around their centre of mass in
elliptical orbits, the third body of negligible mass placed at one of the five libration points can
also move in a closed elliptical orbit provided its initial velocity has a certain value required
for this motion. During such regular non-circular motions the distances between the bodies are
subjected to periodic variations while all three bodies trace homothetic ellipses with common
focus at the centre of mass of the system. The program

Planets and Satellites

[1] allows us to

simulate these extraordinary motions.

Figure 13 shows the periodic motion described by such an exact solution with the satellite

at the apex of the equilateral triangle

ABS

whose base

is the line joining the massive

bodies

and

(

2). The left-hand side of the figure corresponds to the inertial

centre-of-mass reference frame in which all three bodies move in homothetic elliptical orbits.

Regular Keplerian motions in classical many-body systems

Figure 13.

The periodic elliptic motions of the bodies described by an exact particular solution of

the restricted three-body problem with the infinitesimal test body at the triangular libration point.

The right-hand side shows the motions of the satellite

and of body

in the frame of the

heavier body

. The simultaneous positions of the bodies are marked by equal numbers.

The equilateral triangular configuration of the bodies is preserved during the motion; that

is, the satellite remains at all times at the corresponding libration point. However, in contrast
to the case of circular motion, here the triangle formed by the bodies rotates non-uniformly
(together with line

joining the bodies), and the lengths of its sides vary periodically during

the motion (just as does the distance

between the heavy bodies). The major axis of the

ellipse traced by the satellite is at an angle with major axes of the ellipses traced by the
heavy bodies. The three bodies pass simultaneously through the corresponding points of their
elliptical orbits (say, through the ends of the major axes). At points marked as

in the figure

the bodies are at their shortest distances from the centre of mass, and their angular velocity
(the same for all bodies) is greatest. At the remotest points

the angular velocity is smallest.

In the frame of body

(the right-hand side of figure 13), body

and the satellite

move

in congruent ellipses around

. The major axis of the ellipse traced by

makes an angle of

◦

with the major axis of the ellipse traced by

. If body

were suddenly to vanish, the

satellite would leave its elliptical orbit and move after this moment along a larger osculating
ellipse. A part of this osculating ellipse grazing the actual trajectory at point

is shown in the

right-hand side of figure 13.

The motion of the satellite at the interior Lagrangian point of the two heavy bodies tracing

elliptical orbits is illustrated in figure 14. In this motion the satellite

remains between the

primaries, on the line joining them. This line rotates non-uniformly while the bodies move
along the ellipses. The position of the satellite divides the line in a constant ratio, and therefore
the satellite traces an ellipse homothetic with those traced by the heavy bodies. The position of
the interior Lagrangian point between the primaries depends on their masses in the same way
as it does in the circular problem. For example, in a system with

2 (figure 14), the

interior libration point is displaced from the centre of mass towards the lighter body

through

0.237 of the distance

between the bodies. Its distance

from the lighter body

is 0.43

, while the distance

from the heavier body

is 0.57

From the point of view of an observer on the heavier body

(see the right-hand side of

figure 14), the lighter celestial body

, moving around

in an elliptical orbit, is continually

eclipsed by satellite

, since the visible position of

always coincides with that of

. The

small ellipses in the right-hand side of figure 14 show the osculating orbits that the satellite
would trace around

were to vanish suddenly. The first ellipse corresponds to the initial

moment, and the second ellipse to the final moment of the simulation.

E I Butikov

Figure 14.

The periodic elliptic motions of the bodies described by an exact particular solution of

the restricted three-body problem with the light body at the interior libration point.

The motion of a satellite at any of the collinear Lagrangian points is always unstable

(whatever the ratio of masses of the heavy bodies may be). That is, sooner or later its simple
elliptical motion inevitably transforms into irregular, chaotic orbital motion around one of the
bodies, and eventually ends with an ejection of the satellite from the system or with the satellite
crashing against one of the heavy bodies.

Appendix. An arbitrary three-body system in the equilateral configuration

Here we give a detailed explanation of the exact particular solutions to the unrestricted three-
body problem for the triangular equilateral configuration of the bodies. We show that simple
Keplerian motions are possible even when the masses of all the bodies are different.

Let the bodies

and

(of masses

, and

respectively) be located at the vertices

of the equilateral triangle with sides of length

(figure A1 shows the system with

and

). We denote by

and

the radius vectors of bodies

and

relative to body

Figure A1.

Regular Keplerian motions of three bodies of different masses in the equilateral

configuration.

Regular Keplerian motions in classical many-body systems

(that is, the vectors joining

with

and

respectively), and by

and

the gravitational

forces exerted on

and

respectively. According to the law of gravitation,

We add

and

vectorially to find the total gravitational force

exerted on body

(A1)

This force

is directed to the centre of mass

of the system. Indeed, the radius vector

of the centre of mass relative to body

(the vector joining

with

) is given by

)

(A2)

where

is the total mass of the system.

With the help of equation (A2), we can express the total force

exerted on body

the other two bodies

and

in terms of

and

GMm

(A3)

We conclude from equation (A3) that the acceleration

of the body

produced by the

combined gravitation of bodies

and

is proportional to

. It is clear from symmetry that

similar expressions are valid for the accelerations of the other two bodies

and

of the system:

(A4)

Here

and

are the vectors joining the bodies

and

with the centre of mass

Therefore the accelerations of all three bodies are directed to the centre of mass, and the
magnitudes of these accelerations are proportional to the distances of the bodies from the
centre of mass. This conclusion means, in particular, that the system of three bodies in the
equilateral configuration can rotate as a whole (as a solid) about the centre of mass under the
forces of mutual gravitation. We can find the angular velocity

of this rotation with the help

of Newton’s second law. Equating the product of mass of one of the bodies (say,

) and the

centripetal acceleration of its rotation about

to the total force exerted on this body by the

other two bodies, equation (A3), we obtain

GMm

whence

G(m

)

(A5)

Such a uniform rotation of the entire system in the equilateral configuration can occur only

if the initial velocities of the bodies in the centre-of-mass frame are exactly perpendicular to the
radius vectors of the bodies relative the centre of mass, and magnitudes of the velocities exactly
equal the product of

and the distances of the bodies from the centre of mass. The motion

is unstable. That is, if one of the above conditions is even if slightly violated, the equilateral
configuration soon becomes distorted, and the motion of the bodies becomes irregular.

Uniform rotation is not the only possible regular periodic motion of the system in the

equilateral configuration. We can show that the total gravitational force exerted on any of the
bodies by the other two, being directed toward the centre of mass, is inversely proportional to
the square of the distance to the centre of mass. Therefore under such an effectively stationary
central Newtonian force (although created by the moving bodies) the body can trace a closed
elliptical Keplerian orbit (or an open parabolic or hyperbolic trajectory).

In order to prove the above-mentioned property of the effective gravitational field, let

us express the distance of one of the bodies (say,

) from the centre of mass

in terms of

the distance

between any two bodies (the side of the equilateral triangle) and masses of

the bodies. Calculating the square of

, equation (A2), and taking into account that the

E I Butikov

magnitudes of vectors

and

equal

, and that the angle between them equals 60

◦

, we

find

2
2

2
3

)/M

, whence

2
2

2
3

Substituting

into equation (A3), we obtain:

G(m

2
2

2
3

)

(A6)

Equation (A6) shows that the total gravitational force exerted on body

by the other

two bodies is directed to the centre of mass

of the system and is inversely proportional

to the square of the distance between the body and the centre of mass. Under this force
the body moves in a Keplerian ellipse with one focus at the centre of mass. The same
is true for the other two bodies. And since the accelerations of the bodies, according to
equations (A4), are proportional to their distances from the centre of mass, all three bodies can
move synchronously in homothetic ellipses with a common focus at the centre of mass, thus
preserving the equilateral configuration. To simulate this regular, periodic motion, we should
also choose certain initial velocities of the bodies. In the centre-of-mass reference frame, the
velocities must be proportional to the distances of the bodies from the centre of mass and must
make equal angles with the corresponding radius vectors.

References

[1] Butikov E 1999

Planets and Satellites

(New York: American Institute of Physics). This package of interactive

educational computer programs (with a detailed User’s Manual) can be ordered from Physics Academic
Software Publishing Organization through the web:

http://www.aip.org/pas

[2] Blackburn J A 1977 Satellite motion in the vicinity of the triangular libration points

Am. J. Phys.

1077–81

Greenberg R and Davis D R 1978 Stability at potential maxima: The L4 and L5 points of the restricted three-body

problem

Am. J. Phys.

1068–70

Zapolsky H S 1981 More on the restricted three-body problem

Am. J. Phys.

889–90

Styer D F 1990 Simple derivation of Lagrange’s three-body equilibrium

Am. J. Phys.

917-9

[3] Leontovich A M 1962 On stability of Lagrangian periodic solutions to the restricted three-body problem

Sov.

Math. Dokl.

425

Deprit A and Deprit-Bartholome A 1967 Stability of the triangular Lagrangian points

Astron. J.

173–9

Document Outline