35 8 DISCOVERY REPORTS 



order in which they occur ; nor did he state how he deduced his sequence. In the legend to his text-fig. 7, 

 he stated that in the youngest stage of development of the male gonophore, the germ-cells lie inside the 

 branchlets and he indicated schematically certain opaque bodies filling the lumen of the branchlets. 

 The pedicels of the nectophores shown in Richter's fig. 7 have little mesogloea, and on this criterion 

 alone (see page 352), apart from my observations on living material just described, I believe that this 

 figure represents an advanced but not final stage in the development of the male gonophore. Richter 

 described a complicated sequence of developmental events. In the youngest stage of the male 

 gonophore he showed the conventional medusoid bud, an outgrowth of ectoderm, mesogloea and 

 endoderm with a small ' Glockenkern ' (entocodon). He described how the growth of the ' Glocken- 

 kern ' downwards pushes the endoderm, ectoderm and ' Glockenkern ' into the lumen of the branch, 

 and subsequently, how the lower ectodermal layer of the ' Glockenkern ' pushes up again into the half- 

 emptied gonophore. This he appears to have thought resulted in the condition shown in his fig. 25, 

 Taf. XXVIII, which he regarded as the penultimate stage of development of male germ-cells, but 

 which I regard as an early stage. 



Examination of several of my specimens that bore at least ten large gonodendra showed that all the 

 gonodendra of one specimen were either all male or all female. Ripe eggs have never been seen. 

 Haeckel said that Physalia specimens which he observed in the Bay of Algeciras on his way home from 

 the Canaries in March 1867 bore ripe sperm, but he did not state whether the sperms were motile. 

 Because we find ripe eggs neither on the bud-colony nor shed in surface waters, it is probable as 

 Steche (1907) said that they undergo development in the depths, as in Velella. 



ORIGIN OF PHYSALIA AND THE SIPHONOPHORA: THE 

 PAEDOPHORE HYPOTHESIS 



The evolution of Physalia and the other siphonophores seems to be linked with that of certain 

 corymorphine, myriotheline and margelopsine hydroids; with Pelagohydra and with the so-called 

 Disconantha {Velella, Porpema, Porpita). 



Hadzi* (1918), like Chun (1897 a) before him, suggested the derivation of physonect siphonophores 

 from the floating hydroid Margelopsis gibbesi (fig. 8 and legend) ; also of the Chondrophora (Disco- 

 nantha) from corymorphine hydroids (fig. 9 and legend) and of both physonect and calycaphore 

 siphonophores from simple hydroid corms of the Myriothela-type (fig. 10). Garstang (1946), who I 

 think cannot have known of Hadzi's paper, also drew attention to the similarity between disconanths 

 and corymorphines, and between physonects and myriothelines. A review of work on the phylogeny 

 of siphonophores and disconanths may be found in a paper by Leloup (1954). 



I suggest that all these comparable forms, corymorphine, myriotheline and margelopsinef hydroids 

 as well as Pelagohydra and the disconanths and siphonophores, are themselves derivatives of com- 

 paratively lately evolved and new types of animal organization — tentaculate actinula larvae, and that 

 it is through the adaptive radiation of such larvae that these neotenic groups have arisen. The new 

 tentaculate actinula larva must not be confused with the conception of the actinula of Brooks's (1886) 

 hypothesis. Brooks's hypothesis satisfactorily accounts for the origin of the Hydroida, from which 

 stock the precursors of the paedophores must have arisen in the following way : Certain early hydroids 

 tended to retain their larvae which hitherto had been released early as eggs or free-swimming ciliated 

 planulae. We can see this sort of trend in some bougainvillid hydroids today. These retained larvae 

 must have been comparable on release with the actinulae of present-day myriotheline hydroids. 



* So far I have only a translation of some of the figure subscripts in this paper, which is in Serbo-Croat. 

 ■ These subfamily names have been used by evolutionists in a general way and are so used without precise restriction here. 



