Metchnikoff, however, discovered among primitive multicellular animals, such as sponges, hydroids, and lower medusæ, a stage of development still more simple than the gastrula; this stage is without a digestive cavity and only assumes the gastrula form in its ulterior evolution. He also made the remarkable discovery that, in the most primitive multicellular animals, the endoderm is formed, not by means of invagination, but by the _migration_ of a number of flagellated cells from one pole of the wall of the blastula into the central cavity. These cells draw in their flagellum, become amoeboid and mobile, multiply by division, fill the cavity of the blastula, and become capable of digesting. They originate the digestive cells of the complete organism and give birth to the mesoderm, which explains how the latter comes to contain a number of devouring cells even though these do not constitute digestive organs properly so called. Metchnikoff gave to that stage the name of _parenchymella_, for the migrating cells constitute the endoderm in the condition of a parenchyma. The invariable presence of this stage in the simplest multicellular animals, the primitive amoeboid state of the endodermic cells, cases of ulterior transformation of the parenchymella into the gastrula form in certain animals, the absence of a differentiated digestive cavity,--all that proved, according to Metchnikoff, that the parenchymella is more primitive than the gastrula, and is therefore entitled to be considered the prototype of multicellular beings. He saw a confirmation of this in the fact that primitive adult animals also have no digestive cavity but merely an intracellular digestion (sponges, turbellaria). He concluded that the common ancestor of multicellular beings was a being constituted by an agglomeration of cells without a digestive cavity, but endowed with intracellular digestion, like that of the "parenchymula" stage of development. He therefore gave to that hypothetical ancestor the name of _parenchymella_. Later, in 1886, he definitely formulated his theory of the genesis of multicellular beings, and having already stated the phagocyte theory, he substituted for the name _parenchymella_ that of _phagocytella_, which indicated at the same time the primitive mode of digestion of that hypothetical ancestor. Reduced to its simplest form, it presented, according to Metchnikoff, a certain analogy with a colony composed of unicellular beings of two kinds: the first, flagellated, forming the external layer, and the others, amoeboid, occupying the centre of the colony and capable of digesting. It may be interesting to mention here that, in this hypothetical description, Metchnikoff foresaw the existence of similar, but real, beings discovered a year later by Saville Kent, namely, the flagellated colonies of _Protospongia_.