GAMMARIDEAN AMPHIPODA 39 



approach genera outside this group (e.g., the Lysianassinae of Hur- 

 ley) because of the near simplicity of the palm; (g) degree of lobation 

 of the upper lip often has generic value, but just as often does not; 

 for instance, all species of Lysianassa JMilne-Edwards have a lobate 

 lab rum (from lateral view) but occasionally the epistome is also lobate 

 and masks the labrum; lobation of the labrum in Anonyx, (sensu 

 Gurjanova, 1962) is highly variable; (f) the size of the head may 

 have generic value but it is difficult to measure exactly and it has 

 been described so infrequently that its use is reserved until its char- 

 acter value is better demonstrated; (e) the length and setation of 

 mandibular palp article 3 appear to be highly variable in the species 

 of several genera even though they have been used for the definition 

 of Elimedon^ closely allied to Hiyfomedon; (d) the depth of the tel- 

 sonic cleft is variable in some lysianassid genera {Onisbnus) and con- 

 stant in others {Lysianassa) ; (c) incision of the inner ramus of uro- 

 pod 2 is a variable character in Anonyx, as demonstrated by Gurjanova 

 (1962) but is believed to be a stable generic character m Lysianassinae 

 (sensu Hurley, 1963). The remaining two character alternatives (a, 

 b, table 1) may have more value as generic criteria than the others. 

 The length, shape and degree of tapering of coxa 1 are relatively 

 stable except in the genus Tmetonyx. The mandibular molars of the 

 hippomedon section not only differ from those of the tryphosa section 

 but they differ within the hippomedon group in the degree of ridging. 



A tendency to degeneration is seen in the order of species arranged 

 in table 1. Typical species of IJippomedon have a long, distally ex- 

 panded coxa 1, but in Hippomedon strages J. L. Barnard and H. 

 kerguelenl (Miers) the anterior and posterior margins of coxa 1 are 

 parallel ; this is also true of the type-species of EUmedon and Para- 

 centromedon. In species of SchistureUa^ Pseudoneslmus, and the re- 

 maining members incorrectly assigned to Lakota^ the first coxa is re- 

 duced in length, tapers, is often covered partially by coxa 2, and may 

 be very small and hemioval. 



It is difficult to set generic limits in a series of progressive charac- 

 ter modifications especially in view of the fact that occasional species 

 have a unique combination of characters (e.g., the type-species of 

 Parace7itro7nedon) . If no generic boundaries were drawn in the 

 scheme of table 1 lysianassid systematics would, at the present state 

 of development, disintegrate. A major discontinuity, however, seems 

 to occur between Paracentromedon carahlcus and Sclilsturella p>ulclira 

 (see table 1). Above that discontinuity all species could be assigned 

 provisionally to Hippomedon^ taking into account the various levels 

 of simplification in the loss of molar ridges and in the reduced ex- 

 pansion of coxa 1 m all deep-sea species. In this group the simplicity 



