ary
systems accompanies unlikeness in colour, and that stars more equally
matched in one respect are pretty sure to be so in the other. Besides,
blue and green stars of a decided tinge are never solitary; they
invariably form part of systems. So that association has undoubtedly a
predominant influence upon colour.

Nevertheless, the crude notion thrown out by Zoellner in 1865,[1374] that
yellow and red stars are simply white stars in various stages of
cooling, obtained for a time undeserved currency. D'Arrest, indeed,
protested against it, and Angstrom, in 1868,[1375] substituted
atmospheric quality for mere colour[1376] as a criterion of age and
temperature. His lead was followed by Lockyer in 1873,[1377] and by
Vogel in 1874.[1378] The scheme of classification due to the Potsdam
astro-physicist differed from Father Secchi's only in presenting his
third and fourth types as subdivisions of the same order, and in
inserting three subordinate categories; but their variety was
"rationalised" by the addition of the seductive idea of progressive
development. Thus, the white Sirian stars were represented as the
_youngest_ because the hottest of the sidereal family; those of the
solar pattern as having already wasted much of their store by radiation,
and being well advanced in middle life; while the red stars with banded
spectra figured as effete suns, hastening rapidly down the road to final
extinction.

Vogel's scheme is, however, incomplete. It traces the downward curve of
decay, but gives no account of the slow ascent to maturity. The present
splendour of Vega, for instance, was prepared, according to all creative
analogy, by almost endless processes of gradual change. What was its
antecedent condition? The question has been variously answered. Dr.
Johnstone Stoney advocated, in 1867, the comparative youth of red
stars;[1379] A. Ritter, of Aix-la-Chapelle, divided them, in 1883,[1380]
into two squadrons, posted, the one on the ascending, the other on the
descending branch of the temperature-curve, and corresponding,
presumably, with Secchi's third and fourth orders of stars with banded
spectra. Whether, in the interim, they should display spectra of the
Sirian or of the solar type was made to depend on their greater or less
massiveness.[1381] But the relation actually existing perhaps inverts
that contemplated by Ritter. Certainly, the evidence collected by Mr.
Maunder in 1891 strongly supports the opinion that the averag