[485] Shapley. [486] Nature, 115, 419, 1925. [487] M. N. R. A. S., 84, 665, 1924. [488] Clarke and Washington, Proc. N. Ac. Sci., 8, 108, 1922. [489] H. C. 256, 1924. [490] Fowler, Report on Series in Line Spectra, 170, 1922. [491] Hopfield, Nature, 112, 437, 1923. [492] H. C. 256, 1924. [493] Clarke, U. S. Geol. Surv. Prof. Pap. 132, 1924. [494] G. P. Merrill, quoted by Clarke, U. S. Geol. Surv. Bul. 491. [495] Professor Russell believes that iron is much more abundant, at least in the sun, than calculated above. He writes: “More than half of all the strong winged solar lines are iron lines, and the strength and evident saturation of even the faint satellites in the iron multiplets is remarkable.... There are a great many multiplets of nearly equal strength arising from the low triplet level in iron.... Nothing like this happens for the lines, or for and , although it may hold true for the Mg triplets. I should consequently favor multiplying the percentage for iron by a factor of at least 3 and probably 5—which would put it where it obviously belongs.” [496] Chapter V, p. 56. [Pg 190] CHAPTER XIV THE MEANING OF STELLAR CLASSIFICATION IT is not necessary to discuss the possibility or desirability of classifying stellar spectra. Both have been adequately demonstrated by Miss Cannon in the Henry Draper Catalogue,[497] which contains the classification that has been accepted as standard.[498] The catalogue will undoubtedly long remain the authoritative source of spectral data for the major part of the stars bright enough to be accessible to the spectroscopist. The uses of the material that it contains are so numerous and so direct that the basis and meaning of the classes seem to deserve attention. In classifying a number of objects, an attempt should be made to select criteria that will distribute the material into the most natural groups. A classification devised with one point of view will not necessarily appear natural from another, and the best that can generally be done is to select the standpoint that seems to be the most important. From all other standpoints the classification is empirical, and must be treated as such. It seems necessary to emphasize this empiricism with regard to the classifying of stellar spectra, for reference is often made to the Henry Draper Classification as though it had a theoretical, even an evolutionary, basis, whereas it is essentially arbitrary. It is true that a classification based on theoretical principles is very desirable, but at present there is no adequate physical theory on which to found one.