THE SPONGES. 49 



some being surrounded by otbers, Fig. G, Plate 5. The terminals are 

 roughened and very slender, and taper toward the apex, where they 

 bear small end-plates of a watchglass shape. In the larger spicules the 

 end-plate is obviously divided into marginal teeth, and the ray in its distal 

 half is not merely roughened but bears small recurving spines. The size 

 of the spicule varies considerably in the same specimen. The terminals as 

 a rule considerably exceed the principals in length, being from 1.5 to 2.5 

 the length of the latter. The principal measures 36-50 /a x 6 /a, the 

 terminal 60-100 [x, x 2 fi. 



Occasionally this discohexaster exhibits an abnormality of some in- 

 terest. One or several of the principal rays, in addition to bearing terminal 

 umbels, bear one or in some cases two lateral branches, one above the 

 other, Fig. 2, Plate 5. Such lateral branches resemble the terminal rays. 

 Moreover, examination of the larger discohexasters shows that the base of 

 the umbel is frequently asymmetrical. An extreme case of this kind is 

 shown in Fig. 8, Plate 5. These appearances receive an explanation on 

 the hypothesis that the umbel of terminals represents an aggregation of 

 lateral branches, and that during the growth of the spicule some of the 

 lateral branches may become separated from the main cluster. 



A good many small discohexasters occur, having a similar distribution 

 to the large form just described. Some of these are doubtless stages in the 

 development of the latter type, although the principal ray is often about 

 equal in length to the terminals, as in Fig. 3, Plate 5. The principal may 

 bear one or two lateral rays. In a selected spicule of this character the 

 principal rays and the terminals are both 40 [i long ; in another such 

 spicule the principal is 26 fi, the terminals 28 /x long. Together with these 

 spicules occur discohexasters of a different type, one of which is shown in 

 Fig. 6, Plate 4. The principals and terminals in this spicule are com- 

 monly subequal in length, 16 to 24 [x. long, but not infrequently the prin- 

 cipal is perceptibly longer than the terminals. The brush-like clusters are 

 relatively wide and include numerous, from 20 to 30, terminals. The 

 principals are smooth, or bear one or two comparatively large tubercles, or 

 sometimes a lateral ray. It is possible that this spicule is of foreign origin. 

 But against this supposition speaks its distribution, as does also the fact 

 that other small discohexasters occur, which are intermediate in structure 

 between the types shown in Fig. 3, Plate 5, and Fig. 6, Plate 4. As an 

 example of such intermediate forms I select a spicule in which the principal 



7 



