Now the first attempts resulted in artificial classifications, much like our grouping of bats with birds and whales with fish.  All animals, like coral animals and starfishes, whose similar parts were arranged in lines radiating from a centre, were united as radiates, however much they might differ in internal structure and grade of organization.  But this radiate structure proved again to be largely a matter of adaptation.

Practically all animals having a heavy calcareous shell were grouped with the snails and oysters as mollusks.  But the barnacle did not fit well with other mollusks.  Its shell was entirely different.  It had several pairs of legs; and no mollusk has legs.  The barnacle is evidently a sessile crab or better crustacean.  Its molluscan characteristics were only skin-deep, evidently an adaptation to a mode of life like that of mollusks.  The old artificial systems were based too much on merely external characteristics, the results of adaptation.  When the internal anatomy had been thoroughly studied their groups had to be rearranged.

Reptiles and amphibia were at first united in one class because of their resemblance in external form.  Our common salamanders look so much like lizards that they generally pass by this name.  But the young salamander, like all amphibia, breathes by gills, its skeleton differs greatly from, and is far weaker than, that of the lizard, and there are important differences in the circulatory and other systems.  Moreover, practically all amphibia differ from all reptiles in these respects.  Evidently the fact that the alligator and many snakes and turtles (of which neither the young nor the embryos ever breathe by gills) live almost entirely in the water, is no better reason for classifying these with amphibia than to call a whale a fish, and not a mammal, because of its form and aquatic life.

When the comparative anatomy of fish, amphibia, and reptiles had been carefully studied it was evident that the amphibia stood far nearer the fish in general structure, while the higher reptiles closely approached birds.  Then it was noticed that our common fish formed a fairly well-defined group, but that the ganoids, including the sturgeons, gar-pikes, and some others, had at least traces of amphibian characteristics.  Such generalized forms, with the characteristics of the class less sharply marked, were usually by common consent placed at the bottom of the class.  And this suited well their general structure, while in particular characteristics they were often more highly organized than higher groups of the same class.