298 ANTHOPHYSA ROSEA. 



numerous palpons lie in groups between the siphons, close below the gono- 

 dendra, and alternating with the latter. New ones in various stages in growth, 

 are to be seen in various regions. 



The gonodendra have already been described by Chun ('97b), I need merely 

 add that they bear palpons as well as gonophores. The siphons, each with its 

 well-marked basigaster, offer no features of special interest. The palpons 

 are of the usual type, each bearing a long filament charged with nematocysts. 

 Fewkes ('82), observed these filaments, but was not certain as to their nature. 



Tentilla. The tentacles bear two kinds of tentilla (Plate 20, fig. 10, Plate 

 23, fig. 4) corresponding to the involucrate tricornuate, and dendritic types de- 

 scribed by Fewkes ('82a) and by Haeckel ('88b). Chun ('97b) and Lens and Van 

 Riemsdijk ( : 08) observed the tricornuate only. But the absence of the dendritic 

 in their material was probably due to iinperfect condition. Bedot ( : 04) describes 

 no less than six other forms of tentilla, in addition to these two types. But 

 comparison of his figures with the various stages in development exhibited by 

 the specimens studied shows, as Lens and Van Riemsdijk suggest, that they 

 all represent dendritic or tricornuate stages in growth. Thus Bedot's types 

 7], I, y, and ^ are successive stages of the tricornuate (cf. his figs. 6, 7, 10, 

 11, Plate 1, with Plate 20, figs. 8, 10), his 0, 8, e, are undoubtedly three 

 stages reducible to the dendritic type. This is the more likely, as the latter 

 varies in the form of its papilliform processes. 



Pneumatophore. A study of serial sections allows me to add certain details 

 as to the septa and the structure of the pneumatosaccus to the accounts of Chun 

 ('97b) and Bedot ( : 04). The septa were described by Bedot as thirteen in num- 

 ber, unequally developed, only the four on the dorsal side reaching the pneuma- 

 tocodon and forming complete partitions. Bedot's conclusion that such a 

 condition is normal is supported by transverse sections of one of the "Albatross " 

 specimens; but in this case there are sixteen septa, the five dorsal being complete. 

 These five (Plate 21, fig. 1) arise near the apex; and continue downward to the 

 level of the lower end of the bracts, as can be seen on longitudinal sections. 

 But they are connected with the pneumatocodon for only about one half this 

 distance. On the ventral side the septa arise at a somewhat lower level. As 

 Bedot observed, the septae contain numerous "giant cells", usually longitudi- 

 nally placed. In favorable sections the ectoderm cells which separate the giant 

 cells from the supporting layer can be distinguished (Plate 21, fig. 5), much as 

 Chun ('97b, taf. 4, fig. 8) has figured them Athorybia. 



Pneumatosaccus. The pneumatosaccus, best studied on longitudinal sec- 



