128 DISCOVERY REPORTS 



metamorphosing into pentacrinoids. Mortensen has described how short the free- 

 swimming stage is in Isometra vivipara : the embryos travel no farther than the upturned 

 cirri of their mothers before settling down and turning into pentacrinoids. This has not 

 been observed in any of the other species. In Phrixometra lo/igipi/ina and one of the 

 females of the var. antarctica all the embryos in the brood-pouches are at the same stage 

 of development. In the other species and the other female of P. longipinna var. ant- 

 arctica they are, as in Notocrinus mortenseni, at various stages of development. 



In the single known female of Phrixometra nutrix the brood-pouches lie on the oral 

 side of some pinnules, on the aboral side of others. In Kempometra grisea the brood- 

 pouches are on the aboral side of the pinnules, but they lie for the most part beyond, 

 not alongside, the ovaries. In these two species the care of the brood is carried even 

 farther than in Notocrinus virilis. Phrixometra nutrix protects the young throughout their 

 larval existence so that they leave the parent as young comatulids. It seems probable 

 that Kempometra grisea does the same. 



The larvae of Phrixometra nutrix change into pentacrinoids in the brood-pouch. The 

 stalk is attached to the wall of the brood-chamber or to the pinnule segment on its 

 floor ; the head projects through a slit in the wall. In the single known female the brood- 

 pouches hold one or two pentacrinoids each ; where there are two they are at the same 

 or at diff'erent stages of development. No brood-pouch contains developing embryos 

 (Mortensen, 1918). 



In the two females of Kempometra grisea in this collection I found one large embryo 

 in one pouch, two in another, a pentacrinoid larva in a third. The pentacrinoid is still 

 completely enclosed in the pouch, which is not ruptured. No pentacrinoid emerges from 

 any brood-pouch as in Phrixometra nutrix, nor is one attached to any part of the body. 

 The ovary contains few, four to nine, very large eggs; they are up to 0-6 mm. long. 



It is interesting to see that in those species in which the care of the young is carried 

 the farthest — Notocrinus virilis, Phrixometra nutrix and Kempometra grisea — the number 

 of young produced is the smallest. 



DISTRIBUTION AND RELATIONSHIPS 



The table below shows the localities from which the Antarctic comatulids are known. 

 The localities within the Antarctic are divided into "continental coasts and adjacent 

 islands" and "outlying islands". In the former category the term "and adjacent 

 islands" has an application only to the Weddell Sea sector in which the South Shetland 

 Islands, separated from the shores of Graham Land by neither great distance nor deep 

 water, are grouped with it. No crinoid has yet been named from the South Orkney 

 Islands though the ' Scotia ' took feather-stars in Scotia Bay (Wilton, Pirie and Brown, 

 1908, p. 21). The South Sandwich Islands, South Georgia and the Shag Rocks are 

 outlying islands in the Weddell Sea, or South American, sector. Heard Island and 

 Kerguelen are in the Indian Ocean sector. Kerguelen lies on the extreme northern 

 edge of the Antarctic area : the Antarctic convergence, the boundary between the sur- 



