In a very large number of types the gonophores do not develop directly on the hydroid stem, but arise on specially modified zooids resembling rudimentary trophosomes which have been named blastostyles by Allman. On the sides of each blastostyle a series of gonophores usually becomes developed. The blastostyles either remain exposed as in all the Gymnoblastic or Tubularian Hydroids, or as in all the Calyptoblastic Hydroids (Sertularidæ and Campanularidæ) they become invested by a special case—known as the gonangium—which is formed of perisarc lined by epiblast. In the forms with blastostyles three generations interpose between the successive stages of sexual reproduction, (1) the trophosome developed directly from the ovum, (2) the blastostyle budded from this, (3) the gonophore budded from the blastostyle.
Such being the main facts, in order to prove that the existing condition of polymorphism amongst the Hydromedusæ is to be explained as hypothetically suggested above, it is still necessary to shew that (1) the free medusiform gonophores are really only modified trophosomes, or rather that the trophosomes and gonophores are both modifications of some common type, and (2) that the fixed so-called adelocodonic gonophores are retrograde derivatives of the free medusiform gonophores. Unless these points can be established it might be maintained that the Medusæ were special zooids, developed de novo and not by a modification of trophosome zooids. To demonstrate these propositions at length would carry me too far into the region of simple Comparative Anatomy, and I content myself with referring the reader to a discussion of the Hertwigs (No. [146], p. 62) where the first point appears to me fully established. With reference to the second point I will only say that the structure and development of the adelocodonic gonophores can only be explained on the assumption that they are retrograde forms of the phanerocodonic gonophores, and that the opposite view, that the phanerocodonic gonophores are derived from the adelocodonic, leads to a series of untenable positions.
The Trachymedusæ, as has been shewn above, develop directly. They are probably derived from gonophores in which the trophosome has disappeared from the developmental cycle.
To sum up, three types of development are found amongst the Hydromedusæ.
(1) No alternations of generations. Permanent form, a sexual trophosome. Ex. Hydra.
(2) Alternations of generations. Trophosome fixed, gonophore free or attached. Ex. Gymnoblastic and Calyptoblastic Hydroids, and Hydrocoralla.
(3) No alternations of generations. Permanent form, a sexual Medusa. Ex. Trachymedusæ.
Siphonophora. In the Siphonophora alternations of generations take place in the same way as in the Hydromedusæ, but the starting point appears to be a Medusa. The gonophores may remain fixed or become detached.
Acraspeda. With the exception of Pelagia and Lucernaria, in which the development involves a simple metamorphosis, all the Acraspeda undergo a form of alternations of generations. The ovum, as already described, develops into a fixed form—the Scyphistoma—which increases asexually by normal budding, and can even form a permanent colony.