THE DEVELOPMENT OF FASCIOLARIA. 157 



which from 12 to 30 eggs develop into embryos, whereas each of 

 the remaining 500 or 600 divides without regularity into from 

 14 to 20 fragments. 



Selenka ('72), who confirmed Carpenter's results, did not con- 

 sider the division of the " sterile " eggs equivalent to the regular 

 segmentation of the fertile ones, not only because of the irregu- 

 larity of the former process, both as to form and occurrence, but 

 also because he found no nucleus. Neritina fluviatiiis, according 

 to Blochmann ('Si), has capsules in which all the eggs are pro- 

 vided with nuclei that take the usual part of the formation of 

 polar bodies and female pro-nuclei. After these processes no 

 regularity can be detected in the divisions of the sterile ova, the 

 later behavior of which led Blochmann to agree with Biitschli 

 ('77) that these eggs are unfertilized. 



Brooks ('77) observed that of the 6 to 20 eggs in the capsules 

 of Urosalpinx cincrca, all undergo development normally, though 

 exceptionally some may break down and serve as food for the 

 survivors. This case of exceptional cannibalism furnished McMur- 

 rich (loc. cit., p. 408) with "a clue to the manner in which the 

 phenomena seen in Fasciolaria, Pitrpura lapillits, etc., have been 

 brought about. An occasional egg in a capsule has from some 

 cause or other broken down, and has been drawn into the diges- 

 tive cavity of the developing embryos. This process having 

 proved useful is continued, and an arrangement such as I have 

 described above for Purpnra floridana obtained. From this it is 

 but a step to what occurs in Bnccimnn, Pitrpitra lapilhis and Ner- 

 itina. In Fasciolaria the process is, as far as we know at present, 

 at its culmination." 



This series was not conceived of as genetic, for this would have 

 been justified only if it had been previously shown that each of 

 the forms mentioned was evolved from one next lower in the scale 

 of cannibalism. That such has been the history of cannibalistic 

 prosobranchs is not supported by any evidence known to me, 

 and the series is therefore probably a collection of graded par- 

 allelisms. 



The real clue seems to me to lie in another factor which, as 

 McMurrich himself pointed out, is invariably present where can- 

 nibalism has advanced to any great extent. " This is (loc. cit., 



