302 PROCEEDINGS OF THE AMERICAN ACADEMY. 



takes place, therefore, only by elongation of the individual somites 

 which compose it. This method of growth is probably what has called 

 forth the phenomenon of ring multiplication. In support of this view 

 may be cited facts such as that stated by Blanchard ('94) in regard to 

 Ozobranchus, that one of the three rings of a typical somite is doubled 

 in the case of large individucds. 



Summary. 



1. The number of somites in the body of the leech has been deter- 

 mined correctly by Whitman ('92) for the Rhyuchobdellid;c, by Bristol 

 ('99) for the Gnathobdellidae. In both cases the number is thirty-four. 



2. The lindts of the leech somite have been placed incorrectly by all 

 students of leech metamerism, with the possible exception of Vaillant 

 ('70)» in the case of a single genus, Pontobdella, from the time of 

 Gratiolet ('62) to the present. 



3. The natural and true limits of the somite coincide with the limits 

 of the neuromere ; that is, a somite includes those annuli which typically 

 are innervated from the same nerve ganglion. 



4. The foregoing statement is confirmed by an examination of meta- 

 merically repeated structures other than ganglia ; namely, septa, testes, 

 and crop diverticula. 



5. Neuromeric groups of rings, that is, somites as defined under 3, 

 behave as structural units (a) in somite abbreviation (reduction in the 

 number of rings in a somite), (b) in somite elongation (increase in 

 the number of rings in a somite). 



6. Both reduction and increase in the number of rings take place 

 chiefly at the ends of the somite. The sensory ring occupies the middle 

 of the somite and is least often and least extensively affected in the two 

 processes just named. It represents the stable component of the somite. 



7. The five-ringed type of somite found in the Gnathobdellidae has 

 been derived from the three-ringed type found in the Rhynchobdellida?, 

 as suggested by Whitman and demonstrated by Bristol. This has been 

 brought about by division of the non-sensory ring at either end of 

 the somite. 



8. The wide prevalence of ring multiplication among the Hirudinea 

 suggests the derivation of the three-ringed type of somite from a still 

 simpler type consisting, as in Chastopoda, of a single ring. 



9. A phylogenetically intermediate stage between the one-ringed and 

 three-ringed types of somite is probably represented in a typical body 

 somite of Branchiobdella. The same type of somite structure appears 



