If, then, there is no break in the law of evolution here, the race which was predominant at the time when the vertebrate first appeared must have been that from which the first fishes arose, and these fishes must have resembled, not the crustacean proper, or the arachnid proper, but a member of the palæostracan group. Moreover, just as the Labyrinthodonts show special affinities to the fishes which were then living, so we should expect that the forms of the earliest fish would resemble the arthropodan type dominant at the time more closely than the fish of a later era. At first sight it seems too great an absurdity even to imagine the possibility of any genetic connection between a fish and an arthropod, for to the mind's eye there arises immediately the picture of a salmon or a shark and a lobster or a spider. So different in appearance are the two groups of animals, so different their methods of locomotion, that it is apparently only an inmate of a lunatic asylum who could possibly suggest such a connection. Much more likely is it that a fish-like form should have been developed out of a smooth, wriggling, worm-like animal, and it is therefore to the annelids that the upholders of the theory of the reversal of surfaces look for the ancestor of the vertebrate. [Illustration: FIG. 9.--_Apus_ (from the Royal Natural History). Dorsal view.] [Illustration: FIG. 10.--_Branchipus stagnalis._ (From CLAUS.)] {29}We must endeavour to dismiss from our imagination such forms as the salmon and shark as representatives of the fish-tribe, and the lobster and spider of the arthropods, and try to picture the kind of animals living in the seas in the early Devonian and Upper Silurian times, and then we find, to our surprise, that instead of the contrast between fishes and arthropods being so striking as to make any comparison between the two seem an absurdity, the difficulty in the last century, and even now, is to decide in many cases whether a fossil is an arthropod or a fish. I have shown what kind of animal the palæostracan was like. What information is there of the nature of the earliest vertebrate? The most ancient fishes hitherto discovered have been classified by Lankester and Smith Woodward into the three orders, Heterostraci, Osteostraci, and Antiarcha. Of these the Heterostraci contain the genera Pteraspis and Cyathaspis, and are the very earliest vertebrates yet discovered, being found in the Lower Silurian. The Osteostraci are divided into the Cephalaspidæ, Tremataspidæ, etc., and are found in the Upper Silurian and Devonian beds. The Antiarcha, comprising Pterichthys and Bothriolepis, belong to the Devonian and are not found in Silurian deposits. This, then, is the order of their appearance--Pteraspis, Cephalaspis, and Pterichthys.