ME. P. H. CAEPENTER ON THE GENUS ACTINOMETEA. 95 



nouucecl with the subsequent increase of its size (PI. IV. fig. 13, ^;). The rosette is thus 

 essentially formed at the expense of the secondary or ventral layer of the original basals, 

 the ends of the curved spout-like processes being the sole residue of their primary or 

 dorsal layer ; and since, by the removal of the median portion of that layer in each plate, 

 the primary basal cords are left bare vipon their dorsal aspect, they now pass from the 

 angles of the quinquelocular organ into the central canals of the first radials, on the 

 inferior or dorsal side of the calcareous skeleton which occupies the base of the calyx ; 

 instead of reaching them by passing, as they did in the first instance, along its superior 

 or ventral face or, as at a later period, through the middle of its substance. 



Each of these primary basal cords, X, Y, &c., which are thus interradial in position, 

 divides into two branches, X,, X,, l"i, 3^, &c., towards the periphery of the rosette, on 

 the dorsal surface of which it rests. These branches lie in the shallow channels which 

 mark the union of the base of each interradial triangular process (fig. 13 b, o') with the 

 two curved lateral processes above mentioned, each of Avhicli unites with a corresponding 

 process from the adjacent basal to form one of the five spout-like rays (j)) of the rosette. 

 The apex or peripheral end of each triangular process is directed to the suture between 

 two contiguous radials (figs. 3, 7, 16, o'), to which it is attached just between the two 

 adjacent apertures (x, x') of their central canals. Into these canals pass the secondary 

 basal cords Xi, X, one into each of the two contiguous radials, so that one lies on each 

 side of the interradial process of the rosette. 



(§ 65) As a general rule, this process, both in Jnt. rosacea and in Ant. celtica, is short, 

 triangular, and slightly curved towards the ventral side. It is not always so, however, 

 for I have frequently met with specimens of Ant. rosacea in which one or more of tlie 

 interradial processes of the rosette, after bending for a short distance towards the ventral 

 side, turns suddenly downwards, and extends towards the dorsal surface of the radial 

 pentagon. At the same time the parallel margins of each of these abnormally developed 

 processes are so inflected towards the dorsal surface as to form a narrow interradial 

 spout-like process. This is so applied to the projecting and similarly inflected outer 

 edges of the adjacent openings of the central canals {x, x') in two contiguous radials 

 as to convert the interradial furrow lying between them into a complete axial inter- 

 radial canal, precisely similar in character to the radial axial canals already described 

 (§§52,57). 



In one case which I have met with, four out of the five interradial processes of the 

 rosette were of this character. In the rosette represented in PI. IV. fig. 13, only two of 

 the interradial processes (o) are long and spout-like, the other three (o') are short and 

 triangular, like those of a normal rosette. 



This abnormal condition of the interradial process of the I'osette of Ant. rosacea is of 

 considerable interest, as it is the normal one in Actinometra and in many si^ecies of 

 Antedon. 



Not only the interradial, but also the radial processes of the rosette of Ant. rosacea 

 may exhibit departures from their usual shape; for the removal of the primary or 

 dorsal layer at the salient angle of one or more of the five embryonic basals may be 

 incomplete, so that the ends of the curved rays of the rosette exhibit lateral processes, 



