GAMMARIDEAN AMPHIPODA 137 



opod 4 but have the broader spread of posterior setae on gnathopod 1, 

 sharper cusps on the coxae, a slightly shorter first antenna combined 

 with no spines on the outer uropodal rami, were assignable to H. 

 grandicornis. 



Other problems also were noted by Stephensen. He was of the 

 opinion (as evidenced by his synonymy) that Iwasa's H. novizea- 

 landiae might not belong with H. grandicornis, in view of the absence 

 of "fur" on pereopod 4, of spines on the outer rami of uropods 1-2 

 and of the strongly tapering hand of male gnathopod 2. Stephensen 

 had New Zealand specimens which he also rejected from H. grandi- 

 cornis on the basis of a longer first antenna, a longer dactyl of gnathopod 

 1 (this is unclear to the writer), the lack of fur on the posterior edge 

 of pereopod 4, and, indeed, the absence of spines on the outer rami of 

 the uropods. But Stephensen already had included in H. grandicornis 

 his Tristan specimens lacking pereopodal fur. 



One point not considered by earlier workers is the very sharp 

 posterodistal corner of article 2 on pereopod 5 which is characteristic 

 of the type specimens figured by Stephensen. He writes that the 

 Tristan specimens have that article more strongly rounded below 

 than in the types; Hurley's figure of f. novizealandiae has it rounded 

 and the present specimens as well as those of Iwasa have it perfectly 

 rounded. Hurley does not mention its condition in f. thomsoni. 



The specimens at hand are very like those figured by Iwasa, with 

 remarkable similarity in the third uropods, the pleonal epimera, 

 maxillae 1-2, the pereopods, and in some characters of the gnathopods. 

 However, article 6 of the male second gnathopod in Iwasa's figure 

 tapers considerably, the male first gnathopod has the posterior setae 

 of article 6 located more distally and nearly at the palmar corner 

 (the female gnathopod is more like the representation herein), and 

 the coxal cusps are slightly sharper. 



The various forms assignable to H. grandicornis seem to be divisible 

 into at least two groups: (1) the novizealandiae form of New Zealand, 

 Japan, and California, lacking fur on pereopod 4, with a strongly 

 rounded ventral edge of article 2 on pereopod 5, spines occurring on 

 the outer rami of uropods 1-2, the presence of a subdistal spine on the 

 ramus of uropod 3, strongly sinuous second and third pleonal epimera; 

 and (2) the grandicornis form of Chile, combining fur on pereopod 4 

 with no marginal spines on the outer rami of the uropods, straight or 

 scarcely sinuous second and third pleonal epimera, and the absence of 

 a subdistal spine on the ramus of uropod 3. A grandicornis-\ike form 

 also occurs in New Zealand, and one might speculate whether it exists 

 there as a sibling species. South Africa apparently has a novizealandiae 

 form (K. H. Barnard, 1916) which may have gene flow with the 

 grandicornis form because of the frequent mixture of spines and setae 



