CASTLE. — METAMERISM OF THE HIRUDINEA. 299 



takes place (1) most often by division of one or both of the non-sensory 

 rings ; (2) occasionally by division of the sensory ring also. In other 

 words, ring multiplication is most active at the ends of the somite. 



2. GnathobdelUdcB. 



The typical five-ringed somite of Nephelis, Hirudo, and other genera 

 of Gnathobdellidae (Fig. 7, H) has undoubtedly been derived from a 

 three-ringed somite through division of each of its non-sensory rings. 

 This idea, suggested by Whitman ('92), has been abundantly confirmed 

 by the careful work of Bristol ('99). 



In the two genera Dina and Trocheta, as described by Blanchard ('94), 

 we find a further multiplication of annuli taking place. In the former 

 genus (Fig. 7, /), the last of the five rings which, according to my view, 

 constitute a somite (ring 5, Fig. 7, /) is broader than any of the others 

 and is partially divided by a transverse furrow (not shown in the figure). 

 In Tiocheta (Fig. 7, J) the fifth ring is completely but unequally di- 

 vided, the posterior of the two rings thus formed being narrower than its 

 mate. Sometimes no further evidences of ring multiplication are found 

 in the somite of Trocheta, but usually other divisions occur forming a 

 somite of seven (Fig. 7, K) or more frequently of eight rings (Fig. 7, L). 

 Next in order after the division of ring 5 comes that of ring 1 (Fig. 7, A'), 

 then that of ring 2 (Fig. 7, Z). Blanchard finds that individuals obtained 

 from the Crimea often show evidence in certain somites of still further 

 divisions affecting the sensory ring and the two broad rings which follow it. 

 This brings the number of rings in the somite up to eleven (Fig. 7, H). 



It is interesting to note how these eleven rings are related to the three 

 primary rings of a typical somite of Glossiphonia. The primitive sensory 

 ring of Glossiphonia (2, Fig. 6, G) is represented in Trocheta by two 

 narrow rings (stippled in Fig. 7, M). It may, therefore, be regarded as 

 having divided once, the division being among the latest to occur in any 

 of the rings. In place of ring 1 of Glossiphonia (Fig. 6, C), there are 

 four rings in Trocheta (Fig. 7, M). Primitive ring 1 may accordingly 

 be said to have divided twice (compare Figs. 6, C; 7, H\ and 7, M). 

 In place of ring 3 of Glossiphonia (Fig. 6, C), we find in Trocheta (Fig. 

 7, il/) five rings ; in other words, the primitive posterior ring of tlie somite 

 has divided twice, and the most posterior of the four rings thus formed 

 has divided a thvdtime (compare Figs. 6, C ; 7, H\ 7, /; and 7, M). 



To recapitulate: — In ring multiplication among the Gnathobdellidas 

 (1) all five rings occasionally divide ; (2) more often only the non-sensory 

 rings divide, the sensory ring remaining unaffected ; (3) division takes 



