On the 12th of December, 1878, Sir Norman Lockyer formally expounded before the Royal Society his hypothesis of the compound nature of the "chemical elements."[651] An hypothesis, it is true, over and over again propounded from the simply terrestrial point of view. What was novel was the supra-terrestrial evidence adduced in its support; and even this had been, in a general and speculative way, anticipated by Professor F. W. Clarke of Washington.[652] Lockyer had been led to his conclusion along several converging lines of research. In a letter to M. Dumas, dated December 3, 1873, he had sketched out the successive stages of "celestial dissociation" which he conceived to be represented in the sun and stars. The absence from the solar spectrum of metalloidal absorption he explained by the separation, in the fierce solar furnace, of such substances as oxygen, nitrogen, sulphur, and chlorine, into simpler constituents possessing unknown spectra; while metals were at that time still admitted to be capable of existing there in a state of integrity. Three years later he shifted his position onward. He announced, as the result of a comparative study of the Fraunhofer and electric-arc spectra of calcium, that the "molecular grouping" of that metal, which at low temperatures gives a spectrum with its chief line in the blue, is nearly broken up in the sun into another or others with lines in the violet.[653] This came to be regarded by him as "a truly typical case."[654] During four years (1875-78 inclusive) this diligent observer was engaged in mapping a section of the more refrangible part of the solar spectrum (wave-lengths 3,800-4,000) on a scale of magnitude such that, if completed down to the infra-red, its length would have been about _half a furlong_. The attendant laborious investigation, by the aid of photography, of metallic spectra, seemed to indicate the existence of what he called "basic lines." These held their ground persistently in the spectra of two or more metals after all possible "impurities" had been eliminated, and were therefore held to attest the presence of a common substratum of matter in a simpler state of aggregation than any with which we are ordinarily acquainted. Later inquiries have shown, however, that between the spectral lines of different substances there are probably no _absolute_ coincidences. "Basic" lines are really formed of doublets or triplets merged together by insufficient dispersion. Of Thalèn's original list of seventy rays common to several spectra,[655] very few resisted Thollon's and Young's powerful spectroscopes; and the process of resolution was completed by Rowland. Thus the argument from community of lines to community of substance has virtually collapsed. It was replaced by one founded on certain periodical changes on the spectra of sun-spots. They emerged from a series of observations begun at South Kensington under Sir Norman Lockyer's direction in 1879, and continued for fifteen years.[656]