9G THE MEDUSAE. 



from the parent larva while still a mere sac (PI. 25, fig. 6), or it may remain 

 attached and develop tentacles and a mouth which breaks through at the 

 pole opposite that of attachment (PL 24, fig. 6) ; the tentacles are developed 

 in the sequence already described. A second bud may be formed from the 

 aboral plate of the parent while the first is still attached (PI. 24, fig. 6; PI. 

 25, figs. 7-9), but I have never seen more than two ; nor have I seen any 

 trace of the development of tentacles in the second while the first was still 

 attached to it. Bud larvae being indistinguishable after detachment from 

 larvae developed directly from the egg, it is impossible to tell how many 

 generations of polyps may be formed by budding, or whether all develop into 

 medusae. 



The metamorphosis of the polyp larvae into medusae consists chiefly of a 

 flattening caused by disproportionate growth on the part of the aboral disc 

 between the bases of the tentacles. Successive stages in this process, which 

 has been described by Metschnikoff ('86 a ), are shown in PI. 26, figs. 1 to 5. 

 As noted by that author, the collar region first appears as a ridge sur- 

 rounding the larva just oral to the tentacles. The collar, then, does not 

 pass through a lobed stage, but is continuous from the beginning. We 

 must, however, regard the peroniae as representing obliterated incisions 

 between ancestral lobes. Thus it appears that, although morphologically the 

 tentacles may be regarded as standing on the bell margin, in this species 

 they never do so in actual development, since they are already some distance 

 within the margin when the latter is first distinguishable as such. In this 

 respect P. smaragdina differs from Cunoctantha octonaria, in which, according 

 to both Brooks ('8G) and H. V. Wilson ('87), the marginal lappets result from 

 an outgrowth of the aboral surface between the tentacles, phylogenetically a 

 more primitive condition. 



The flattening continues with growth until the largest specimens, 12 mm. 

 in diameter, arc disc-like (PI. 26, fig. 5), and at the same time the number of 

 tentacles increases, the greatest number observed being thirteen. Octocysts 

 are first distinguishable in the stage represented in PI. 26, fig. £, where 

 they are merely papilliform processes of the margin, comprising both germ 

 layers exactly like rudimentary tentacles (PI. 26, fig. 6'). Soon from one 

 to five otoliths are formed, appearing first at the distal end (PI. 24. fig. 8), 

 and as the sense organ attains its mature form the connection of its endo- 

 dermic core with the vascular endudermie layer is lost (PI. 24, fig. 7). There 

 are no otoporpae. 



