26 SELECTION IN CLADOCERA ON THE BASIS OF 



were discarded as apparently unprofitable material for a selection 

 experiment on the basis of reactiveness to light. The lines of S. ex- 

 spinosus which were retained did not contain fewer non-reactive 

 individuals during their early history, but the experiments were 

 continued, and the results justify the conclusion that there was 

 after all a sufficient basis for an experiment in selection. 



RELATION BETWEEN ENVIRONMENTAL CONDITIONS AND 

 REPRODUCTIVE AGE. 



Reproductive age is very closely related to environmental con- 

 ditions, principally to temperature. It is a matter of common obser- 

 vation in the rearing of this material that the reproductive age is 

 greatly affected by temperature and food conditions. During a local 

 period of high temperature and warm nights in summer, a generation 

 may be obtained in 6 to 7 days in S. exspinosus and in 5 to 6 or 7 

 days in Daphnia and even in 1 or 2 days in Moina, while during a 

 period of continued cool days and nights in fall or spring, when the 

 laboratory is not provided with heat, the reproductive age stretches 

 out to 11 to 14 days for Simocephalus and Daphnia and 8 to 12 days 

 for Moina. 



Curves were made for some of the lines, showing the mean re- 

 productive ages for different two-month periods. At times they show 

 remarkably close agreement in mean reproductive ages, not only 

 between the two strains of the same line but also between the different 

 lines of the same species and to a considerable degree between D. 

 pulex and S. exspinosus. Figure IB (for Line 695) and Figure 17s 

 (for Line 757) show two of these curves. 



STATISTICAL TREATMENT OF DATA. 



Irregularities in phototropic response are common. They are 

 noted in almost every paper dealing with the light-reactions of ani- 

 mals. Dependable results are generally obtained only by securing 

 sufficient data to obtain averages from relatively large numbers. 



Like most other material, the Cladocera discussed in this paper 

 showed considerable irregularities in light-reactions, and compari- 

 sons of the two selected strains of the same line are made by con- 

 sidering the data by longer periods, so as to avoid the extreme effect 

 of local conditions and fluctuations in the behavior of individual 

 broods. 



In the study of a character so complicated as a behavior charac- 

 teristic it is essential to deal with relatively large numbers, and the 

 safest way to deal with large series of data is by statistical treat- 

 ment. Much time has been given to the statistical analysis of this 

 data, although the treatment has been carried only so far as seemed 



