120 



W. J. CROZIER. 



viduals which may have divided some little time previous to 

 others, as indicated by the various stages in regeneration classi- 

 fied under this head. A " lag " in this curve- of division with 

 respect to water temperature might therefore be expected. The 

 sudden rise of percentage of " non-regenerating " specimens at 

 December, however, is probably adequate proof that a month or 

 so is sufficient time for the full growth of new rays. The actual 

 " lag " in the curve of autotomy is of the wrong sort for the idea 



c 



30 



28 

 26 

 24 

 2,2 

 20 

 18 



r'. 



\ 



\ 



\ 



pa 



f- 

 O 



o 



o 



- -J >- z: -j o - 



^LJ^O^^^^Z^IUOQW 



"-^u-Zr^SI^^^c. 



FIG. 3. Mean monthly temperatures of seawater at Bermuda. Heavy line, 

 data from Verrill ('02) ; dotted line, averages from records of W. J. C. The 

 dotted line records for winter months are higher than those given by Verrill, 

 probably because my own observations were restricted for the most part to the 

 waters of the partially enclosed Sounds. 



of complete temperature control, because the decrease in fre- 

 quency of self-division perceptibly precedes the seasonal decrease 

 of temperature. This might be explained on the basis that the 

 onset of higher temperature in spring serves indirectly to induce 

 self-division, but that the capacity for this method of propaga- 

 tion is gradually exhausted, or in some fashion checked, before 

 the warmer season has closed. 



It is interesting, however, to notice how the frequency of self- 

 division is enormously decreased at the approach of the period 

 of breeding. There is thus accomplished, whether through the 



