310 STAR many telescopic stars reaches us now only af- ter the lapse of many thousands of years. To apply to a Oentauri the method for inferring a -tar's volume, indicated above, we proceed as follows: The star's distance exceeds the sun's 230,000 times, so that the sun removed to that star's distance would shine with only TF,TT5Triinr.Tm> P art of nis observed lustre. But it has been found by Zollner that a Oentauri shines with about TT ,TyTr.tVffTy.innr part of the sun's brightness. Hence the star emits three times as much light as the sun, or (if our as- sumption as to equal intrinsic surface bright- ness be correct) a Centauri has a surface three times, a diameter ^/3 times, and a volume 3>/3 times (L e., more than five times) greater than the sun's. If w& dealt with Sirius in like man- ner, we should deduce a volume exceeding the sun's about 2,700 times (taking the mean of the values above given for his annual paral- lax). But there is reason to believe that the real volume of Sirius, though far exceeding the sun's, is much less than that we have thus deduced; whence it is to be inferred that the larger stars shine with a greater intrinsic lus- tre than our sun, or in other words that a square mile of the surface of a large star like Sirius gives out much more light than a square mile of the sun's surface. It is not improba- ble that we may find hereafter in such consid- erations the means of distinguishing between the various orders of real star magnitudes, since stars of different intrinsic brightness might be expected to give different results under spec- troscopio analysis. We have shown under SPECTRUM ANALYSIS that such differences un- questionably exist in stellar spectra; but as yet it has not been found possible to associate them satisfactorily with differences in the sizes of stars. In fact, when we observe that Ca- pella, though a star not only of the leading order of apparent magnitude, but also, judging from its minute annual parallax, one of the largest in real volume, yet belongs to the sec- ond spectral class, that is, the class of stars re- sembling our sun, we can scarcely place much reliance on this method of discriminating large from small stars. Closely connected with the question of the various orders of stars is the circumstance that many stars are colored. Of stars visible to the naked eye, only the bright- est show recognizable color, at least as so viewed. Antares, Betelgeuse, and Aldebaran are ruddy ; Arcturus, Pollux, and Procyon yel- low; Vega and Altair bluish; Capella, Sirius, Canopus, and many others, brilliantly white. But among telescopic stars more marked in- stances of color occur, some stars being blood- red, garnet-colored, rich orange, golden yel- low, and so on. It is noteworthy that few single stars show such colors as blue, green, violet, or indigo ; but among double and mul- tiple star systems not only are these colors recognized, but such colors as lilac, olive, gray, and so on. A beautiful feature in many star- ivmaiiis to be noticed: it is often found that the components exhibit complemen- tary colors. This is oftener seen among un- equal doubles ; and then the larger component shows a color from the red end of the spec- trum, as red, orange, or yellow, while the smaller shows the corresponding color from the blue end, as green, blue, or purple. The colors are real, not merely the effect of con- trast, for when the larger star is concealed the color of the smaller remains (in most cases) unchanged. Spectrum analysis shows that the colors of many double stars are due to absorp- tive vapors cutting off certain portions of the light. The existence of double and multiple star systems is itself remarkable, and the the- ory of a real physical connection between the members of such systems was long opposed because of the strangeness of a conception which in our own day has become familiar to us. Of course, many stars apparently double are in reality far apart, and merely brought into accidental association because both lie nearly on the same visual line. But not only is the number of such pairs far greater than it should be to be thus explained, but also many pairs have been watched during long periods, and it has been found that the components are circling around each other, or rather around their common centre of gravity. Among the most remarkable instances of this kind are the double star TO Ophiuchi, which completes a revolution in about 80 years ; the stars of the pair f UrscD Majoris, which complete the cir- cuit around their common centre of gravity in about 60 years ; Castor, y Virginis, f Bootis, C Cancri, and other doubles, which exhibit equally noteworthy motions. Many catalogues of double stars have been formed by astrono- mers since Sir W. Herschel first paid special attention to the work. He observed 2,400; W. Struve of Dorpat observed 3,063; Dem- bowski, Secchi, Webb, and others in Europe have observed many double stars, carefully measuring the distance between the compo- nents, the angle of position, color, and so on, thus forming a fund of materials from which future astronomers can determine what changes are taking place in these interesting systems. Among such catalogues, those recently formed by Mr. Burnham of Chicago will hold a dis- tinguished place because of the "difficulty" of the double stars he has observed, arising chief- ly from the nearness of the components, or from the smallness of one or both. It is re- markable that though every region of the heav- ens contains double stars, they are more abun- dant by far in some regions than in others; while again some regions of the heavens con- tain double stars of particular orders only or chiefly. This leads us to notice the circum- stance that aggregations of stars of greater and greater extent are recognized ;is v rxtrml our survey of the heavens. Of all such aggrega- tions the most complex is the galaxy or milky way (see GALAXY), in which millions of stars shine with lustre so blended and softened by
