KCHINOIDKA. II. 8l 



disconnected like that of /'. laguncula. It is beyond doubt that in a restriction of the genus the name 

 Pourtalesia lias to be retained for the group of species to which /'. miranda belongs. Thus far I agree 

 with Agassiz, whose above cited proposition of a subdivision of the genus is evidently much more 

 in accordance with the natural relations of the species than Duncan's and Pom el's subdivisions. 

 Nevertheless I cannot fully accept Agassiz' subdivisions either. 



On reviewing the characters of the species it seems to me that one feature may reasonably be- 

 taken to be of primary importance for a grouping of the species, viz. whether the bivial ambulacra are 

 interrupted by the postero-lateral interambulaera or not. Also the shape of the test seems rather im- 

 portant, whereas pedicellarice and spines seem to be of secondary importance. The character of the 

 apical system, whether it is disconnected or compact, cannot be used, all the species thus far known 

 having in fact a disconnected apical system '. 



The bivial ambulacra are continuous in carinata (almost certain!), pkiale, paradoxa and pro- 

 bably ceratopyga, disconnected in the other species (P. rosea, hispida and miranda are unknown in this 

 respect, but the two latter may well be supposed to have them disconnected). Further it is to be re- 

 marked that P. carinata differs from all the other species in having two pores and tube-feet in the 

 ambulacral plates I. a. i and Y. b. i. (P. rosea and miranda again are unknown in this respect, though 

 the latter may doubtless be supposed to have the pores single as in laguncula etc.). Finally it may 

 perhaps be a character of some importance whether the dorsal plates of the odd posterior interambu- 

 lacrum are paired or alternating, the latter being, of course the more primitive structure; they are 

 alternating in P. carinata and ceratopyga, paired in Jeffreys:, Wandeli, hispida , laguncula, Tanneri, 

 phiale and paradoxa. Upon the wdiole this character evidently cannot, however, be taken too rigorously, 

 the paired plates generally showing more or less distinct traces of their originally alternating condition. 

 In typical examples the difference between these structures is very conspicuous, as seen e. g. by a 

 comparison of Figs. 51 and 52. PI. VII in Loven's: On Pourtalesia. In accordance with the characters 

 pointed out here as the more important, I think the following grouping of the species will prove to 

 be the natural one: 



1. Bivial ambulacra continuous; two pores in the ambulacral plates La. 1 and 

 Y. b. 1. Test not especially widened or elongate. Dorsal plates of odd 

 interambulacrum alternating P. carinata. 



2. Bivial ambulacra continuous; one pore in the plates I. a. 1 and V. b. 1. Test 



very elongate; dorsal plates of odd interambulacrum paired P. phiale and paradoxa. 



3. Bivial ambulacra (probably) continuous; one pore (sometimes two) in the 

 plates I. a. 1 and V. b. 1. Dorsal plates of odd interambulacrum alternating. 



Test much widened anteriorly P. ceratopyga. 



4. Bivial ambulacra disconnected; one pore in the plates I. a. 1 and V.b. 1. 

 Dorsal plates of odd interambulacrum paired. Test not especially widened 



or elongated P. laguncula. miranda (?), 



Unknown: P. rosea. Tanneri, Jeffreysi, Wandeli and hispida. 



1 Whether the genital plates be separate or not, seems to be a character of small importance, since both cases may 

 occur in the same species. Likewise the presence or absence of the labrum is of small importance, as shown by its great 

 variation in P. Jeffreysi ami Wandeli. 



The [ngolf-Expedition. IV. 2. 1 j 



