seemed to be excluded. But this supposed variability of the sun is
disproved by Guthnieck and Prager's (4) photoelectric measurements
of the brightness of the planet Saturn. The planet, being illumi-
nated by sunlight, would reflect any changes in intensity of the
sun's radiation ; the delicate measures possible with photoelectric cells
showed that the light is very steady, variations of the amount deter-
mined by Abbot being quite excluded.
Solar System (see 25.357).- A ninth satellite of Jupiter was discovered by S. B. Nicholson at the Lick Observatory in 1914. Like the eighth satellite it revolves round the planet in the opposite direction to the other seven. The periods of satellites VIII. and IX. are about 739 and 745 days respectively, and the two bodies are revolving in almost equal interlocked orbits in planes inclined at about 10. Satellites VI. and VII. form a somewhat similar interlocked pair, their periods being 251 and 260 days respectively; but their motions are in the " direct " sense.
Much interest has been taken in the " Trojan Group " of minor planets. These illustrate a special case of the problem of three bodies discussed by Lagrange, viz. that in which the three bodies are situated at the vertices of an equilateral triangle. The Trojan planets have almost the same mean distance and revolution period as Jupiter, and the equilateral condition is roughly fulfilled. The problem of the small librations of such a planet about the triangular point of equilibrium has been discussed by E. W. Brown (5); the condition of stability is that the mass of Jupiter must be less than -0385 times that of the sun a condition which is easily satisfied and the period of the libration is about 140 years. Actually the Trojan planets are at some considerable distance from the triangular points, and the problem of determining the finite librations (as opposed to in- finitely small librations) has provided much exercise for mathe- maticians. Six members of the group are now known, Nos. 588 Achilles, 617 Patroclus, 624 Hector, 659 Nestor, 884 Priam, and 911 (unnamed); of these Patroclus and Priam are near the tri- angular point 60 behind Jupiter, and the others 60 ahead of Jupiter.
A very curious minor planet was discovered by W. Baade on Oct. 31 1920, temporarily designated 1920 HZ. Its orbit is extremely elliptical (eccentricity 0-65); and its perihelion lies near the orbit of Mars, whilst its apheh'on reaches to near the orbit of Saturn. It is generally thought that a body with this eccentricity must necessarily be, or become, a comet, the extreme alternations of heat provoking the disruption characteristic of comets; but HZ shows no signs of a cometary envelope, and is provisionally classed as a planet.
The period of rotation of Uranus round its axis has been de- termined by V. M. Slipher from measures of the line of sight velocity of the advancing and receding limbs. The result is io h 5o m and the direction of rotation agrees with that of revolu- tion of the satellites. Leon Campbell subsequently found that the light of the planet is variable with the same period, pre- sumably owing to unequal brightness of different parts of the surface. The rotation period of Venus still remains a mystery; and there are advocates of the long period of 224 days as well as various estimates of short period (one to three days).
Latitude Variation (see 16.267). The study of the small periodic motion of the earth's axis of rotation (relatively to the earth) which gives rise to " variation of latitude " has been continued at the six international stations (reduced in number during the later stages of the World War). The effect is made up of (a) The free precession of a spheroid rotating about an axis which does not coincide with its axis of figure ; the period of this precession determined from the observations is 43 2 J days; (b) an annual term, which is a forced oscillation due to meteoro- logical and seasonal causes. Owing to interference of these two terms, there is an effect analogous to " beats " in sound waves, the amplitude of the motion alternately rising to a maximum of about o"-3 (30 ft.) and dying out in about six years' period. The annual term appears to be nearly circular (6) and of ampli- tude o"-o85; the possible causes contributory to this, such as seasonal circulation of the atmosphere and ocean, snowfall, and vegetation have been investigated by H. Jeffreys (7), who
finds a fair agreement between predicted and observed values. A mysterious Kimura or Z term, which appears in these interna- tional results, would, if interpreted literally, indicate an annual approach to the pole and recession by all stations on the same latitude simultaneously or a shifting of the earth's centre of gravity to and fro along its axis. It is, however, now believed that the term arises from a small systematic error in the observations; independent observations made at Greenwich and Pulkovo (not belonging to the international chain) show either a reduced or zero Kimura term.
The Stars (see 25.784). Progress in our knowledge of the stellar universe must depend largely on the patient accumulation of accurate statistics as to the parallaxes, motions, spectra, magnitudes, etc., of large numbers of stars; it may therefore be well to review the great advance in these data in recent years.
The first photographic determinations of stellar parallaxes reaching a modern standard of accuracy were made by H. N. Russell and A. R. Hinks at Cambridge, and F. Schlesinger at Yerkes, in 1903-7; earlier results are now superseded except for a few of the best heliometer measures made chiefly by Gill. Extensive programmes have since been carried out with large telescopes at the Allegheny, Greenwich, Leander McCormick, Mount Wilson and Sproul observatories, and by 1921 parallaxes of about i, 600 stars had been measured with probable errors generaUy not greater than o"-oi. The use of a rotating sector to reduce the bright ness of the star under observation to that of the comparison stars has made a considerable improvement in the accuracy. Unfortunately it does not follow that we know the distances of 1,600 stars, for many of these parallaxes turn out to be inappreciable. The results emphasize the fact that very few of the stars are sufficiently near for the method to give any close measure of the distance; and a large proportion of the measures are of little use individually though they may throw light on questions of statistical distribution when taken in conjunction with other evidence. We cannot resist the impression that inves- tigation of stellar parallaxes by the trigonometrical method is reaching its limit with present instruments; and perhaps for that reason special interest is attached to a new method of determining the distances of stars described below under " Spectroscopic Parallaxes."
Lewis Boss's Preliminary General Catalogue of 6,188 Stars published in 1910 has been an invaluable aid to research with regard to proper motions. It comprises all the brighter stars, and the proper motions constitute a great improvement as re- gards both accidental and systematic error on anything previous- ly available. Of other catalogues the most notable is the Green- wich 1910 catalogue containing the proper motions of 12,368 stars in the zone Decl. + 24 to 32; the accuracy, of course, does not equal that of Boss's catalogue, but it carries our knowledge of the motions of stars in this region as far as the ninth magni- tude. We have still very little systematic knowledge of the motions of still fainter stars, which can be measured photo- graphically; attention has chiefly been directed to the detection of exceptionally large motions by the " blink " microscope or by other methods.
The first reaUy extensive lists of radial velocities were pub- lished by the Lick Observatory in 1911. At present (1921) about 2,070 have been determined; these have been collected in a catalogue by J. Voute. Progress would have been more rapid but for the large proportion of spectroscopic binaries, which makes it necessary to repeat the measures several times at suitable intervals in order to discriminate between orbital motion and the true secular motion which is looked for. Orbits of 172 spectroscopic binaries are known; and in addition there are about 450 spectroscopic binaries with orbits as yet undeter- mined. It appears therefore that approximately one-quarter of the stars examined have proved to be spectroscopic binaries. Al- lowing for systems of wider separation (not detected by varying radial velocity) the actual proportion of binaries must be still higher.
The apparent magnitudes of stars range from i m> 5 for Sirius, to 2o m and upwards for stars obtained by long exposures with the
