40 R- T. YOUNG. 



fall of the fission rate, due to some unknown factor in cell metab- 

 bolism, from which recovery is autonomous" ('05, p. 604), we 

 must consider the then "unknown factor," which virtually makes 

 "rhythm" synonymous with inter-endomictic period, a sense in 

 which the term "endomixis was not used . . . when first em- 

 ployed by Woodruff and Erdmann" (Woodruff, '176, p. 447). 

 But if we do not consider this factor in defining "rhythm," its 

 definition becomes a wholly arbitrary matter dependent upon the 

 character of the graph employed, which as Woodruff and Erd- 

 mann ('14, p. 447) have shown is at best an artificial means of 

 representing the history of the race in question, and which tends 

 to emphasize or obscure the rhythms as the average period is 

 shortened or lengthened. 



While cessation of endomixis was followed by death in two of 

 Woodruff's experiments (A T and AE T , '176), both my own and 

 Woodruff's experiments show many cases in which the process 

 occurred shortly before the death of the race in question see my 

 Figs. II, VIII, Villa, Xa, etc., and Woodruff's ('176) Figs, i, O, 

 and 8, B s and M s . It is of course entirely possible that death in 

 the former cases resulted from cessation of endomixis, and in the 

 latter from some other cause, but on this question the experi- 

 ments throw no certain light. 



Woodruff's recent work has further disclosed several additional 

 cases of the exceptional occurrence of endomixis at a high, or 

 rising, rather than a low, or falling point in the division rate. I 

 have previously noted two of these exceptions (/. c., p. 47) one in 

 my own experiments and one in those of Woodruff and Erdmann 

 (1914), to which may -be added here the occurrence of the process 

 at periods 68 O, 81 B, Fig. 7; 61 A s) 82 and 86 O 8 , 70 B s , 70 and 82 

 M s , Fig. 8 and 76 M, Fig. 12. 



In most of Woodruff's graphs the variations are too irregular 

 or the period too short to show any regular rise and fall in the 

 division rate over considerable periods, but in one case (A, Fig. 

 9, p. 455, 176) such alternate periods are clearly shown, without 

 however any evident correspondence between endomixis and the 

 periods of low and high division rate. An endomixis is as liable 

 to occur at a high point as at a low point on the curve, neglecting 

 the minor fluctuations of a few tenths of a division which mark 



