THE GONOPHOEES OF THE HYDEOCOEALLIN^. 401 



no difficulty in believing that the ancestral form was a colonial 

 trophosome, and that medusae of different kinds may have 

 originated quite independently of one another from the 

 Hydroid stocks. 



The original position of the gonads was the centre of the 

 concavity of the umbrella. As they became larger and larger 

 in phylogeny a conical growth of the endoderm^ with re- 

 spiratory and nutritive functions, penetrated them, and 

 became the manubrium. All of these stages may be seen 

 repeated in the ontogeny of the medusa of Millepora. When 

 a mouth was formed at the end of the manubrium the gonads 

 were in some forms (anthomedusse) restricted to the sides of 

 that organ ; but in other forms (leptomedusse) they were 

 shifted to a more convenient place in the radial canals. Ac- 

 cording to my view, then, the manubrium of the male 

 gonophore of AUopora does not prove that it is a degenerate 

 medusa, but, rather, that it is one stage further than 

 Distichopora on the road that all medusae have travelled in 

 the early history of their phylogeny ; that is to say, a stage 

 with a larger spermarium, and a special process of endoderm 

 for its more perfect nourishment and respiration. 



Another question arises in connection with the gonophores 

 of the Hydrocorallinse that at one time would have been 

 considered one of vital importance. 



In the description given above of the development of the 

 medusa of Millepora, I have shown that it is formed by a 

 metamorphosis of a dactylozooid. This would support the 

 view, then, that the medusa is a modified trophosorae. 



In the description of the development of the gonophores of 

 Allopora and Distichopora I do not mention the zooids at all. 

 The gonophores are not developed in these genera (figs. 12, 

 19) in connection with either the gastrozooids or dactylo- 

 zooids, they arise quite independently from the coenosarcal 

 canals. They have no particular relation to the systems in 

 which the zooids are arranged, and there is every reason to 

 suppose that they are quite independent of them. Further, 

 these gonophores are not, according to my view, degenerate 



