Trembley (see p. 57) had shown that Hydra, though an animal, multiplies by budding like a plant. He got indications, upon which he did not altogether rely, that it also propagated by eggs, and ten years later (1754) this supposition was confirmed by Roesel, who figured the egg, though he was unable to demonstrate that a young Hydra issues from it; subsequent inquiry has placed the fact beyond doubt. In 1819 Chamisso announced that Salpa, a well-known Tunicate which abounds at the surface of the sea, exhibits a regular alternation of the two modes of increase, the egg-producing form being succeeded by a budding form, the budding form by the egg-producing form, and so on indefinitely. Sars a few years later showed that the common jelly-fish Aurelia also propagates by eggs and buds alternately. Here the familiar swimming disks, which are of two sexes, produce eggs from which locomotive larva issue. The larva at length settles down and takes a Hydra-like form. It pushes upwards an ascending column, which divides transversely and forms a pile of slices, each destined to become a free, sexual Aurelia. The alternation of generations may be regarded as resulting from the introduction of budding into the early stage of a life-history which culminates in sexual reproduction, much as if a caterpillar were to divide repeatedly and form more caterpillars, each of which ultimately became a moth. The case which has been given as an illustration actually occurs in nature. A parasitic caterpillar, that of Encyrtus, divides while still an embryo, so that one egg produces several moths.[35] Many other cases of alternation have since been found among animals, and it seems to be the rule among plants. Alternation of generations may be complicated by association with transformation, by the omission of stages usual in the class, and by budding-out from one part instead of from the whole body. In particular cases the complication becomes so great that biological language breaks down under it. Such terms as _generation_, _individual_, _organ_, _larva_, _adult_, cannot always be used consistently without either being strained or artificially limited. Baer and the Development of Animals.