CASTLE. METAMERISM OF THE HIRUDINEA. 291 



midway between the sensory rings, exactly the position indicated for the 

 somite limits by the study of the nervous system (Fig. 1, right half). 



In his work on "The Embryology of Clepsine," Whitman (78) states 

 that the testes appear in the septa between the mesodermal somites. 

 Now the position of the testis in the adult is regularly in the two non- 

 sensory rings, that is, exactly midway between successive sensory rings. 

 This is just the position in which the boundary between somites should 

 be on the grounds already examined. 



Crop diverticula are repeated at metameric intervals throughout the 

 middle part of the body in most species of Glossiphonia. They inva- 

 riably arise, so far as my observations go, in the seusory ring of the 

 somite, a fact readily explained by the occurrence in that position of the 

 interseptal portions of the coelom, represented by the transverse lymph 

 vessels. 



There is one other set of organs which is metameric in its arrangement, 

 — the nephridia. A single uephridium, according to the account of 

 Oka ('94), lies principally in the ring which contains the nephropore 

 (seusory ring, Fig. 1, nph'po.) and in the two rings which precede it. 

 Accordingly the nephridium is not confined to a single somite, either as 

 somites are commonly limited or as I have suggested placing their limits. 

 This need not surprise us, since in the Cha^topoda also the uephridium 

 lies partly in the somite preceding that in which the greater part of it is 

 found. 



III. Limits of the Somite in the Gxathobdellid.e. 



Let us next inquire whether the suggested criteria of somite limits are 

 applicable also in the case of the Gnathobdellidae. 



1. Relation of Five-ringed to Three-ringed Type of Somite. 



Whitman suggested and Bristol ('99) has demonstrated that the five- 

 ringed type of somite, found in the Gnathobdellidae, may be derived from 

 the three-ringed type, found in the Rhynchobdellidae, by supposing that 

 each of the two non-sensory rings of the latter type has divided, while 

 the sensory ring has remained unchanged. 



A comparison of the distribution of the nerves arising from a single 

 ganglion in Glossiphonia (Fig. 1) and in Nephelis (Fig. 4) shows the 

 correctness of the homology suggested. The anterior two of the three 

 metameric nerves of Glossiphonia are united in the case of Nephelis 

 (Fig. 4, a). They are distributed, as we should expect, to the sensory 

 ring and the two rings which precede it. The posterior nerve of Glossi- 



