According to Slipher the spectroscopic binary γ Geminorum has the comparatively long period (for a spectroscopic binary) of about 3½ years. This period is comparable with that of the telescopic binary system, δ Equulei (period about 5·7 years). The orbit is quite eccentric. I have shown elsewhere[317] that γ Geminorum has probably increased in brightness since the time of Al-Sufi (tenth century). Possibly its spectroscopic duplicity may have something to do with the variation in its light. With reference to the spectra of double stars, Mr. Maunder suggests that the fact of the companion of a binary star showing a Sirian spectrum while the brighter star has a solar spectrum may be explained by supposing that, on the theory of fission, "the smaller body when thrown off consisted of the lighter elements, the heavier remaining in the principal star. In other words, in these cases spectral type depends upon original chemical constitution, and not upon the stage of stellar development attained."[318] A curious paradox with reference to binary stars has recently come to light. For many years it was almost taken for granted that the brighter star of a pair had a larger mass than the fainter component. This was a natural conclusion, as both stars are practically at the same distance from the earth. But it has been recently found that in some binary stars the fainter component has actually the larger mass! Thus, in the binary star ε Hydræ, the "magnitude" of the component stars are 3 and 6, indicating that the brighter star is about 16 times brighter than the fainter component. Yet calculations by Lewis show that the fainter star has 6 times the mass of the brighter, that is, contains 6 times the quantity of matter! In the well-known binary 70 Ophiuchi, Prey finds that the fainter star has about 4 times the mass of the brighter! In 85 Pegasi, the brighter star is about 40 times brighter than its companion, while Furner finds that the mass of the fainter star is about 4 times that of the brighter! And there are other similar cases. In fact, in these remarkable combinations of suns the fainter star is really the "primary," and is, so far as mass is concerned, "the predominant partner." This is a curious anomaly, and cannot be well explained in the present state of our knowledge of stellar systems. In the case of α Centauri the masses of the components are about equal, while the primary star is about 3 times brighter than the other. But here the discrepancy is satisfactorily explained by the difference in character of the spectra, the brighter component having a spectrum of the solar type, while the fainter seems further advanced on the downward road of evolution, that is, more consolidated and having, perhaps, less intrinsic brightness of surface.