clusions drawn from analytical results. Proof of homogeneity is an essential datum in the establishment of a mineral species. Many a supposed species has been overthrown by the microscope.
But a mineral may be apparently homogeneous, and yet indefinite chemically; for, apart from mere impurities which are unrecognizable by physical means, there are modes of admixture even more difficult to determine. Two distinct compounds may crystallize together in varying proportions, so as to yield definite forms which are, to all physical tests, perfectly homogeneous. Such mixtures of "isomorphous" substances are almost infinite in number; they are among the commonest occurrences in Nature; and they complicate the mineralogic problem enormously. Theoretically, a species is easily defined; practically, the definition is most troublesome. Oftentimes all the members of an isomorphous group are regarded as one species, in which certain analogous elements are said to "replace" each other. Iron and alumina are thus mutually replaceable; so are the oxides of the magnesia group; so also are sodium and potassium. But this usage, though common and sanctioned by weighty authorities, is not rigidly scientific. It is allowable conventionally, but only so long as we do not lose sight of what it really means. The so-called replacement is in reality a phenomenon of mixture between isomorphous salts of allied metals or acids, which salts are the true, definite species. For example, garnet varies in composition in just this peculiar way, and six or more compounds, all different but similar, are represented in it. Sometimes we find one of these compounds nearly pure; but oftener two or more exist in a given crystal. Garnet, therefore, is not one species, but a group, and should be so treated. A mixture is a mixture, whether visibly so or not, and has no title to specific naming. On a systematic basis the current policy needs modification; for it varies too widely and can not be universally applied. Seeking to evade one set of difficulties, it creates new ones.
In consequence of the tendency toward mixture among species, and of the wide-spread fashion of regarding the crystal as the mineralogical unit, there has grown up a general belief that minerals are in most cases very complex chemically. Some species, undoubtedly, are quite simple, like quartz, fluor-spar, or calcite; but others, especially among the silicates, appear to be most complicated, and even variable, in composition. This complexity, which is in great part due to the influences already mentioned, is perhaps apparent rather than real. Mixtures, whether crystalline or mechanical, can hardly be given either simple or definite chemical formulæ. The true individual units are probably not very complex, for their modes of origin favor simplicity. A complex molecule is likely to be unstable—the more complex, the more unstable; while minerals seem to be generated under conditions adverse to instability. Some have been deposited from solutions in which many reactions were possible; others originate under conditions