316 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



responding to the radius opposite the interradius containing the water pore. The 

 infrabasals, like the other plates, seem at first to avoid the ventral side, and in 

 the rare cases where five infrabasals are developed, they appear to be arranged 

 in the form of a horseshoe, quite as widely open ventrally as that of the basals 

 and orals. At the time of the fixation of the larva the inner border of each infra- 

 basal becomes smooth and concave, and they then arrange themselves in a circle 

 around the chambered organ just above the topmost columnal. The arrangement 

 of these plates is still the same as in the earlier stage, the smallest plate being in 

 radius A. At a slightly later stage these three plates fuse with one another and 

 with the topmost columnal so as to form one large plate. Though the sutures of 

 the infrabasals still persist, the plates themselves have grown out into five angles; 

 these angles are radial in position, fitting in between the edges of the basals and, 

 while the infrabasal in radius A produces only one angle, each of the other two 

 grows out into two angles; at a slightly later stage the sutures disappear, though 

 the groove separating the infrabasals from the topmost columnal persists for some 

 time. The whole large plate formed by the coalition of the circlet of infrabasals 

 with the topmost columnal is therefore in reality a double structure, the lower half 

 only being the true centrodorsal. 



In Antedon adriatica Seeliger found that the infrabasals are developed at a 

 little over four days; they are usually four or five, rarely three, in number. The 

 two lateral infrabasals on either side lie moderately near together, and may be 

 the morphological equivalent of Bury's large lateral infrabasals observed in A. 

 mediterranea. 



In Promachocrinus Tcerguelensis the infrabasals, which are five in number, are 

 much larger than in the two species of Antedon in which they have been found, and 

 remain distinct from the centrodorsal until a considerably later period. They are 

 all of approximately equal size, forming a circlet of rounded plates about the top of 

 the column. 



It is indeed strange that such painstaking and accurate observers as Thomson, 

 Perrier, and the two Carpenters should have overlooked such prominent structures 

 in Antedon lifida if they really occur in that species. Antedon adriatica is the 

 least specialized of all the species of the genus, and A. mediterranea is only slightly 

 more advanced; the former has four or five underbasals, the latter three. Antedon 

 bifida, A. moroccana, A. petasus, A. liupjen and A. dubenii represent phylogeneti- 

 cally a great step in advance over the two Mediterranean forms, and it is quite 

 within the bounds of possibility that, as a result of acceleration of development, 

 all traces of infrabasals have been lost in the ontogeny of these five Atlantic species. 



Basals, and structures formed from and associated with them. 



The basals, primarily five in number, in the later crinoids typically form a circlet 

 about the apical portion of the body between the circlet of infrabasals and the 

 circlet of radials, with both of which they alternate in position, being midsomatic 

 or interradial (figs. 565, 566, pi. 7, 576, pi. 9, 579, pi. 11, and 583, pi. 12); they cor- 

 respond to the genitals of the echinoids. 



In nearly all of the recent crinoids the basals are abnormal in their develop- 

 ment; they may be reduced to three, as in Hyocrinus, Thalassoerinus (fig. 145, p. 209), 



