But, besides all this, we must remember, in the second place, that all the fossiliferous deposits in the world, even if they could be thoroughly explored, would still prove highly imperfect, considered as a history of extinct forms of life. In order that many of these forms should have been preserved as fossils, it is necessary that they should have died upon a surface neither too hard nor too soft to admit of their leaving an impression; that this surface should afterwards have hardened sufficiently to retain the impression; that it should then have been protected from the erosion of water, as well as from the disintegrating influence of the air; and yet that it should not have sunk far enough beneath the surface to have come within the no less disintegrating influence of subterranean heat. Remembering thus, as a general rule, how many conditions require to have met before a fossil can have been both formed and preserved, we must conclude that the geological record is probably as imperfect in itself as are our opportunities of reading even the little that has been recorded. If we speak of it as a history of the succession of life upon the planet, we must allow, on the one hand, that it is a history which merits the name of a "chapter of accidents"; and, on the other hand, that during the whole course of its compilation pages were being destroyed as fast as others were being formed, while even of those that remain it is only a word, a line, or at most a short paragraph here and there, that we are permitted to see. With so fragmentary a record as this to study, I do not think it is too much to say that no conclusions can be fairly based upon it, merely from the absence of testimony. Only if the testimony were positively opposed to the theory of descent, could any argument be fairly raised against that theory on the grounds of this testimony. In other words, if any of the fossils hitherto discovered prove the order of succession to have been incompatible with the theory of genetic descent, then the record may fairly be adduced in argument, because we should then be in possession of definite information of a positive kind, instead of a mere absence of information of any kind. But if the adverse argument reaches only to the extent of maintaining that the geological record does not furnish us with so complete a series of "connecting links" as we might have expected, then, I think, the argument is futile. Even in the case of human histories, written with the intentional purpose of conveying information, it is an unsafe thing to infer the non-occurrence of an event from a mere silence of the historian--and this especially in matters of comparatively small detail, such as would correspond (in the present analogy) to the occurrence of _species_ and _genera_ as connecting links. And, of course, if the history had only come down to us in fragments, no one would attach any importance at all to what might have been only the _apparent_ silence of