little more than the germs from which the chemical elements themselves are subsequently evolved by polymerisations during the process of condensation into stars. Another hypothesis, which has been strongly advocated by Sir Norman Lockyer, is that the nebulae consist of swarms of meteoric stones, and that the spectrum presented to us is simply that of the gases driven into the interspaces by collisions, and rendered luminous by heat or electricity. On this supposition, the metallic elements remain unvolatilised and consequently fail to signify their presence in the spectroscopic record.
Independently of these hypotheses, however, the phenomena of New Stars furnish valuable evidence as to the presence in nebulae of materials other than those which are directly indicated by the spectrum. It is well known that although the spectrum of a new star in its earlier stages is a complex one, great changes accompany the decline of the star, so that finally the spectrum becomes like that of a nebula. Now, it has been shown that the earlier spectrum consists chiefly of the spark lines of various known metals, and unless we are prepared to adopt the improbable hypothesis that the metallic elements have been dispersed into the surrounding spaces, it must be admitted that they are still present when the star has faded and only the nebular spectrum is visible. Hence, it is not unreasonable to believe that other nebulae with similar spectra also contain metallic elements, existing under conditions which prohibit their visible display.
A difficulty formerly presented by such bodies as the Great Nebula in Andromeda, which do not give a spectrum of bright lines at all, has been removed by recent spectrum photographs taken at Lick and Mt. Wilson observatories. These photographs have confirmed previous suspicions that the continuous spectra of these objects are crossed by dark lines, like the spectra of most of the stars, and the unavoidable conclusion is that such bodies belong to the order of star clusters. Their composition must consequently be the same as that of the stars in general.
In opposition to the view here advocated as to uniformity in the chemical composition of the various cosmical bodies, it has sometimes been urged that the apparent occurrence of stars of different spectral types in special localities in space is more easily explained by assuming actual diffe-
