54 I. IJIMA : HEXACTINELLIDA. I. 



especially iu intimate association with the hilt-rays of the dermalia, 

 with their outer ends at, or sticking out of, the external bound- 

 ing surface (PI. V, fig. 36). In such a situation and arrange- 

 ment, the rhaphides would serve in their own way as a power- 

 ful defense against attack from without. The excessive abundance 

 of broken off oxyhexaster-terminals in perfectly undisturbed 

 specimens of certain Rosselid species has given me the impression 

 that the phenomenon is not peculiar to the graphiocome alone. 



The oxyhexasters (PL II, fig. 15; &c.) have been found in 

 all Eu])lectella species, except in E. simplex and E. nodosa. 

 Unlike the other rosettes already referred to, these occur iu both 

 the external and the internal trabecular layer. In these positions 

 they should serve as a defense against pernicious intruders. 

 While in E. oweni and curvisiellaia the oxyhexasters were quite 

 plentiful, I have found them rather sparingly in E. marshalli 

 and imperialis and particularly so in the outer trabecular layer, 

 — a circumstance which may be correlated with the especial 

 abundance of graphiocomes in the two last mentioned species. 



The entire size of oxyhexasters and the different develop- 

 ment of their parts are of value, though not universally, in the 

 distinction of species. A very noteworthy modification of the oxy- 

 hexaster is the clasp-like or sigma-like fibula of E. jovls (also 

 found in another Eupectellid, Holascus fibulatus), the true nature 

 of which has been perceived by F. E. Schulze ('87, pp. 78, 88). 

 AVe have here to do with a diactinose oxyhexaster, probably 

 derived from such a form as I have called the hexactin-shaped 

 or hexactinose oxyhexaster (Ijima '97, p. 45 ; ' Derivate-Oxy- 

 hexaster of Schulze, '99) in which each principal has only a 

 single terminal, in a manner analogous to that in which a rod- 



