Selective illumination is beyond doubt largely concerned in modifying the information we are able to obtain as to the composition of remote masses, and its modes of action seem capricious because they are very imperfectly understood. Hence, spectral modifications may take place merely through the substitution of some elements for others in carrying (let us suppose) an electric discharge, though all were from the first simultaneously present in unvaried proportions. Moreover, chemical immutability can no longer be taken for granted. We have learned of late that even elementary individuality breaks down under the battering-ram of time. Sooner or later the stamp, however apparently inviolable, will be defaced, transformations of species will ensue, and novel combinations of material will subtly accommodate themselves to the needs of a growing world. These things, it is true, are involved in much obscurity, but we have caught glimpses of instability clear enough to convey an emphatic warning against dogmatic interpretations of spectral characters. Physical science may then license M. du Ligondès' theory of comets with a provisional _Nihil obstat_. The zodiacal light suggests a different set of considerations. Comets being of pre-planetary origin, the ecliptical glow must be supposed post-planetary. It belongs to a later epoch, being composed, according to an accepted opinion,[110] of superfluous materials left over from the construction of the train of globes to which our own belongs. It might be compared to whey from which the curd has been separated. All the good has been got out of it; we might be tempted to throw it aside upon the rising rubbish-heap of the skies, with the importunate asteroidal throng, a few dozen undistinguished comets, and some hundreds of ill-defined meteoric systems. But celestial refuse is always worth sifting, above all, for tokens of genealogical descent, and we should be unwise to neglect the chance of finding them in the peculiar relations of the zodiacal light.