362
The Galactic System of Stars [ch. xiv
to a single plane, namely the plane of the orbits of the outer planets, while in
the neighbourhood of any one point, the motion of the asteroids is nearly
confined to one direction in that plane, namely the direction at right angles
to the direction of the sun.
The question arises as to whether the motion of the stars is in any sense
an ordered motion like that of the asteroids, or whether it is purely random
like the motion of the molecules of a gas. Until 1904 most astronomers would
probably have conjectured that the motion was of the latter type, but solely
on the grounds that the motion had not so far been found to be anything but
characterless. In 1904 the situation was abruptly changed by Kapteyn’s
discovery of the phenomenon of “ star-streaming.”
Kapteyn found that the motion of the stars in the neighbourhood of the
sun shewed a distinct preference for one direction in space. The preference
was not entirely like the preferential motion of the asteroids, since some stars
moved forwards and some backwards along the same direction, but this dis
tinction is one which can be removed by altering the frame of reference to
which the motions are referred. The stars do not all move along this particular
direction; indeed, the preference for this direction is not very highly marked,
many stars moving absolutely athwart it.
The direction of the preferential motion is found by Kapteyn and many
others to be exactly in the galactic plane. Kapteyn attempted a dynamical
explanation on the general lines that the direction of preferential motion was
at right angles to the direction pointing towards the centre of the galactic
system, the motion of the stars about the centre of the galactic system thus
being supposed similar to that of the asteroids about the sun, except that
Kapteyn imagined two swarms of “ asteroids ” intermingled with one another
and moving in opposite directions.
An alternative mathematical expression of the observed fact was suggested
by Schwarzschild. The molecules of a gas obey Maxwells law of distribution,
according to which the number whose components of velocity u, v, w lie within
a small range dudvdw is'of the form
J q e -A(M 2 +t) s +w 2 ) dudvdw.
Schwarzschild proposed that the motions of the stars conformed to an
“ ellipsoidal ” law of distribution of the type
A-6~h ('“H-« 2 )-*«’ 8 dudvdw.
If h is made equal to k in this formula, the motion reduces to the random
motion specified by Maxwell’s law. If h is infinite in comparison with k, u and
v must be zero for every star, so that the motion reduces to a pure to-and-fro
motion along the axis of £. If h is larger than k , the motion shews a
preference for the two directions along the axis of z, the amount of this pre
ference being determined by the value of the ratio h/k. The observed degree
of preferential motion is represented by assigning to h/k a value of about 2.