ARBACIA. 67 



Ambulacral plates at ambitus with 3 (exceptionally 4) pairs 



of pores ; secondary spines and tubercles wholly wanting Arbacia. 

 Ambulacral plates at ambitus with 4 or 5 pairs of pores ; sec- 

 ondaries present in interambulacra at and above ambitus Tetrapygus. 

 Abactinal surface with numerous short, non-articulated spines ; 

 primary spines flattened, with more or less serrate edges, 

 confined to ambitus or actinal surface of test. 



Abactinal system small or of moderate size, not .60 h. d. ; 

 anal plates, 4. 



Abactinal system about .35 h. d. ; non-articulated 

 spines not arranged in horizontal series on each 

 abactinal coronal plate ; tridentate pedicellariae small, 



with valves only about .30 mm. long Podocidaris. 



Abactinal system about .50 h. d. ; non-articulated spines 

 arranged in horizontal series of 4-7 on each abac- 

 tinal coronal plate ; tridentate pedicellariae very 

 large, with compressed valves, 2-2.5 mm. long . . DiaUthocidaris. 

 Abactinal system very large, about .66 h. d. ; anal plates, 5 . Pygmoeocidaris. 

 Test thin and delicate; anal plates, 5; primary spines decidedly 

 triangular in cross-section ; valves of tridentate pedicellariae re- 

 markably wide and flat; valves of ophicephalous pedicellariae 

 extraordinarily constricted, with notably expanded and hollowed 



apophysis Hahrocidaris. 



Primary spines very long and usually tapering, greatly exceeding twice 



diameter of test ; sphaeridial pits 6-12 in each ambulacrum actinally . Ccelopleurus. 



Arbacia. 



Gray, 1835. Proc. Zool. Soc. London, p. 58. 



Type-species, C idans jmstuhsa heske, 1778. Add. Klein, p. 85. 



Tertiary (?) and Recent Arbaciadie. 



There seem to be only five valid species of this genus, although Lov^n, 

 in his revision of it in 1887, recognized double that number. Loven, how- 

 ever, considered as a good subgeneric character, one which we find cannot 

 be relied on even to distinguish a given species, namely, the extent to which 

 the abactinal interradial areas are free from spines, and he recognized indi- 

 vidual differences such as the form of the test as valid specific characters. 

 We are entirely unable to find any character which will distinguish speci- 

 mens from Brazil from those taken in the Mediterranean, while Loven con- 

 sidered them as two quite distinct species. It is possible that further material 

 from the west coast of Africa will make the recognition of the species africana 

 Troschel desirable, and there is also a possibility that altenians Troschel may 

 ultimately be separable from Dufresnii Bl., but in the light of such material 

 as is now available either in the M. C. Z. or elsewhere, these three names 



