VARIABLE STARS There are three main classes of variable stars: (I) eclipsing variables, (2) Cepheid variables, (3) long period variables. In a sense the sun is a variable star, since the changes of its surface condition which follow the sunspot cycle of 112 years must involve some small variation in the rate of emission of light and heat (see SUN) ; but this is not analogous to any of the recognized forms of stellar variability. Novae or "new stars" might also be counted technically as variables; it is, however, no longer be lieved that they have anything in common with the ordinary types of variables, and we shall not include them here. (See NovA.) Eclipsing Variables.—In this type of variation there is no physical change in the star. It emits light steadily all the time, but it is occulted at times by another body passing in front of it. The present explanation dates from 1782, when John Good ricke, having discovered the regular period of fluctuation (2d 2oh 49m) of the variable star Algol, suggested that the periodic dimming was caused by a dark body revolving round it and eclipsing it once a revolution. Knowing, as we now do, that there are multitudes of close binary systems, it is only to be expected that we should sometimes observe an eclipse of one component by the other. Each system will provide regularly recurring eclipses for observers situated sufficiently near the plane of the orbit; the eclipsing variables are therefore ordinary binaries whose orbits we happen to be viewing nearly edge on. For Algol and many other variables of this class it has been verified that the object is a binary system, and that the decline of light occurs when the principal star is in the part of its orbit most remote from us, which is the time when the second component would intervene.
It is not necessary to assume that the intervening object is a dark body. Even if the system consisted of two equal stars a total eclipse of one by the other would halve the light received. In most systems that have been studied, one component is con siderably brighter than the other, but the light of the second component is quite recognizable. Half a period after the main
eclipse, when the faint star occults the brighter, a secondary eclipse is observed owing to the bright star occulting the fainter. There is, of course, a considerable period when no eclipse is occurring and the system shines steadily with its maximum light. (By this feature eclipsing variables are easily distinguished from all other types.) The steadiness in these intervals is, however, only approximate, and is complicated by two phenomena of singular interest: (I) Reflection Effect.—Systems in which the two stars are very close together have the best chance of showing eclipses, since the orbit does not then need to be so accurately orientated with respect to our terrestrial post of observation. The closeness of some of the best-known eclipsing variables is remarkable, the components being separated by no more than their own diameter. Each is therefore strongly illuminated by the other, and the hemisphere turned towards the companion becomes heated and shines more brightly than the other hemisphere. For the fainter star in particular, the disparity in brightness of the two hemi spheres is considerable, and as it moves round its orbit it exhibits variations of brightness analogous to the phases of the moon. Owing to this reflection effect, the end of the principal eclipse is followed by a gradual brightening, which continues until the dip representing the secondary eclipse begins.
(2) Ellipsoidal Effect.—Owing to their close proximity the two stars are distorted, being elongated towards one another. Thus, in addition to the variation due to eclipses and reflection, there will be a variation depending on the aspect of these elongated objects. Their apparent sections will be smallest when we see them end-on near the time of the eclipses ; half-way between eclipses we see the prolate spheroids broadside-on and correspond ingly brighter.