122 W. J. CROZIER. 



agency of temperature or in some other way, a nice adjustment 

 of the two dissimilar methods of multiplication practiced by this 

 species. Asexual increase through self-division of the body is at 

 a minimum just before and during gametic reproduction. It is 

 not difficult to suppose that an undivided individual is better 

 suited for gamete formation and for such coordinated move- 

 ments as probably are concerned in bringing the sexes near to- 

 gether for fecundation of the eggs. The general energy of the 

 animal is then directed toward the elaboration of gametes, rather 

 than diverted to the formation of new tissue. To what extent 

 the elements of this correlative mechanism are causally related, 

 however, one must refrain from guessing. Judging from the 

 statements of Clark ('130, '13^) and Mortensen ('17) regarding 

 the asexual multiplication of Linckia gulldingii, and the period of 

 its egg-reproduction, I suspect that the supplementary relation 

 here pointed out for Coscinasterias may not obtain with Linckia, 

 where, however, the method of self-division is different, consist- 

 ing not in the division of the body across the disc but in the ab- 

 striction of a portion of a ray. Nor in Holothuria surinamensis 

 is there good evidence of an alternation of the sexual and asexual 

 methods of multiplication (Crozier, '17). 



II. 



The number of madrepores in Coscinasterias is quite variable, 

 from i to 5 being found on a disc. As earlier suggested (Cro- 

 zier, '15), the number of madrepores is correlated rather defi- 

 nitely with the total number of rays (Tables III., IV.). It was 

 observed that during the addition of new rays, subsequent to 

 self-division, one or more new madrepores are also added. In 

 almost every case, throughout the several hundred carefully ex- 

 amined, each moiety of a divided body bore one or several " old " 

 madrepores. Moreover, some of the very largest specimens col- 

 lected, with ray lengths of 12-13 cms., had but 5, 6, or 7 rays, of 

 equal length, with but one madrepore. These facts suggested 

 that the development of several madrepores is perhaps a more 

 or less necessary preliminary to normal self-division. The madre- 

 pores are usually formed at widely separated points on the periph- 

 ery of the disc ; hence the probability is that each part of an 



