230 "TERRA NOVA" EXPEDITION. 



(3) The first and second antennae in S. glacialis are longer than in S. septem- 

 carinata. In S. glacialis the peduncle of the first antenna is equal to the first four 

 joints of the peduncle of the second antenna, and the whole first antenna is if anything 

 slightly longer than the second. In S. sejytemcarinata the peduncle of the first antenna 

 reaches only about half-way along the fourth joint of the peduncle of the second, and 

 the whole first antenna is considerably shorter than the second antenna. 



(4) In S. glacialis each of the thoracic and first three somites of the abdomen has 

 the median dorsal posterior border produced into a spine-like process, which is absent 

 in S. septemcarinata. 



(5) In S. glacialis the coxal plates of the seventh thoracic segment extend 

 backwards as far as the pleural plates of the third abdominal somite, and not cj^uite so 

 far as those of the second abdominal somite. In S. septemcarinata the coxal plates of 

 the seventh thoracic segment are much shorter than the pleural plates of the third 

 abdominal segment, which in turn are equal to those of the second. 



(6) In *S. glacialis the last somite of the abdomen bears anteriorly a prominent 

 median dorsal spine, which is not present in S. septemcarinata. The latter has seven 

 carinas on the last segment of the abdomen, while S. glacialis has but five. Moreover, 

 the shape of the abdomen differs considerably in the two forms. 



(7) In »S. glacialis the outer branch of the uropods is much shorter than the inner. 

 In S. septemcarinata the two branches are much more equal in size and broader than 

 in S. glacialis. 



S. glacialis, like the majority of species of the genus, has the terga of the first 

 three free thoracic somites separated by a suture from their coxal plates. This character 

 serves to separate it from *S. schytkei, Liitken, S. p)aradoxa, Fabr., and S. polar is, 

 Eich., which have the first four free thoracic coxal plates separated by a suture, and 

 from S. gracilis, Bedd., and S. latifrons, White, in which the number is five and six 

 respectively.* 



S. glacialis is also separated readily from the Australian group of species, 

 S. tuberculata, Grube, S. jmllida, Bedd., S. australiensis, Bedd., S. elongata, Bedd., 

 S. longicaudata, Bedd., S. minuta, Bedd., and S. bakeri, Chilton, in that the tergum of 

 the fourth free thoracic somite is not unduly narrow, and that of the fifth free thoracic 

 somite is not obsolete in the middle dorsal region. From the remaining species of the 

 genus the characters of the last abdominal segment and of the uropods will serve as 

 distinguishing marks.* 



The armature of the inner palmar margin of the propodus of the second thoracic 

 appendages in this genus seems to be a matter that is not quite clear. Beddard, 

 speaking generally, states that there are two kinds of peculiarly formed spines regularly 



* See Caiman (1920) for a suggested re-grouping of the species of this genus based on the structure 

 of the uropods and the segmentation of the thorax. S. glacialis belongs to Caiman's group of S. paradoxa. 

 The Australian species form a distinct group, and the third group comprises S. latifrons and Caiman's new 

 species S. heddardi. 



