134 LIBBIE H. HYMAN 



L. limosus, on which MorguHs ('07) has experimented. Liimbric- 

 ulus inconstans, however, differs from both of these forms in a 

 very important way as will appear shortly. 



If Lumbri cuius inconstans is cut up along the axis into pieces 

 of moderate length, — that is, pieces including more than ten 

 segments — then each of these regenerates completely. A com- 

 plete anterior end is the head of seven segments, of which six 

 are setigerous, a large prostomium, brain, pharynx, etc.; but the 

 transverse green pigment stripes do not appear on the regener- 

 ated head for some time. In long pieces, the regeneration of 

 seven head segments is almost invariable, demonstrating that a 

 definite number of segments is differentiated into a 'head' in 

 Lumbriculus; but in short pieces, the number of head segments 

 regenerated is very variable, and this variability has lead to 

 much controversy over the question of cephalization in the 

 oligochaetes. The regeneration of the posterior end consists 

 simply of the formation of an anal segment, and of an indefinite 

 number of new segments in front of this. As is general in 

 oligochaetes, neither the extreme anterior or posterior ends are 

 capable of regeneration. The head piece must include at least 

 four trunk segments, or eleven in all, before it will regenerate 

 posteriorly; isolated heads shorter than this remain tailless. 

 The posterior end must be twenty or thirty segments long 

 before it will regenerate anteriorly, otherwise it dies without 

 regeneration. 



With shorter pieces, less than ten segments long, an axial 

 difference becomes apparent. Disregarding now the extreme 

 anterior and posterior piece, it is found that the capacity for both 

 anterior and posterior regeneration decreases along the axis. 

 Posterior regeneration may be dismissed briefly. Pieces from 

 the anterior regions, although they may remain tailless in a 

 small percent of cases, in which the pieces are too short to con- 

 tain enough of the axial gradient for formation of both dominant 

 and subordinate parts (see above, and Child, '13 b), in general 

 produce very long posterior ends. As one passes back along the 

 axis, the number of posterior segments regenerated gradually 

 decreases, and in pieces from posterior regions, frequently only 



