SYSTEMATIC CLASSIFICATION OF ECHINI. 215 



on account of differences in spicules of the tube-feet which are very large and peculiar, and on 

 differences of the pedicellariae. It differs from the Echinometridae also in that the oculars 

 enter the periproct in the sequence I, V, IV (p. 113), not V, I, IV, as in that family. The most 

 important character, however, is that the teeth are keeled and epiphyses narrow, not meeting 

 in suture over the foramen magnum (Plate 4, figs. 8-10). Other important peculiarities exist 

 as described (p. 186). The character of the lantern places this genus and family as a member 

 of the Stirodonta. Its affinities are nearest to the Phymosomatidae as indicated by spurs 

 from the dorsal tips of the pyramids, which support the teeth (Plate 4, figs. 8-10). The ambu- 

 lacral plates are composed of three elements each, and where fully developed, as at the mid- 

 zone, four or five plates are combined and grown over by a large primary tubercle, a very strik- 

 ing character. The periproct has numerous small plates, as in Phymosoma. The ocular 

 plates of the bivium are typically insert in Stomopneustes variolaris, but I only or I, V, IV may 

 be insert as described (p. 113). Also, as an aberrant variation, three specimens in a total of 

 64 have oculars I, V, II insert. This is an essential character of the Echinidae and Strongylo- 

 centrotidae, and suggests a connection between the families. The small suranal plate of Glypto- 

 cidaris crenulare, as figured by Mr. Agassiz (1873, Plate 7a, fig. 8), suggests the same affinity, 

 and it seems that in one of these two families may lie the ancestral stock of the Echinidae and 

 allies. 



Of the Arbaciidae the most striking feature is that there are only four, or five, nearly equal 

 plates in the periproct, a feature otherwise known only in Parasalenia. Another important 

 feature not so readily ascertained is the retention of the primordial interambulacral plates in 

 the basicoronal row, as pointed out by Mr. Agassiz (1904, p. 54). A third unusual character 

 is that ocular plates, when they enter the periproct, do so in the sequence V, I, IV (p. 158). 

 This is known elsewhere only in the Echinometridae, as here shown. The auricles exist as widely 

 separate styloid processes (text-fig. 227, p. 193) or slightly arched and joined by suture (text-fig. 

 228). Tube-feet are dorsally modified as ambulacral gills in Arbacia (A. Agassiz, 1872, p. 264). 

 In the genera Habrocidaris and Podocidaris, as shown by A. Agassiz and Clark (1908, p. 77), 

 and Pygmaeocidaris, as shown by Doderlein (1906), the primordial interambulacral plate is 

 retained as in Arbacia. 



The suborder Camarodonta may be considered the most specialized of modern regular 

 Echini on the basis of the lantern, and also in various genera by the sculptured test, the degree 

 of specialization of the ambulacrum, peristome, perignathic girdle, or the elliptical form through 

 a sidewise axis. The essential feature is the fact, that while the teeth are keeled as in the last 



' Doderlein figures (1906, p. 184, fig. 36e) a third median plate in tiie third coronal row of Pygmaeocidaris. As this 

 genus so closely resembles Habrocidaris, in which there is a median tubercle but no median third plate, it is possible that 

 Doderlein was misled in introducing the sutures of this plate, a point difficult to ascertain in such a small specimen (4.8 mm.) 

 as he had. If Doderlein is correct, it is a unique case in the Centrechinoida. 



