230 THE AMERICAN NATURALIST [Vol. L 



all the individuals for each successive trial were averaged, 

 and the probable error calculated for each point on the 

 curve. In accordance with a plan proposed by Professor 

 Cattell, the limits of the probable error are shown by 

 broken lines. The chances are even that with a greatly in- 

 creased number of cases the time would have remained 

 within these limits, and a nearly smooth curve can be 

 drawn within them. A notable exception is the tenth trial 

 with the yellow mice. At this point there is an unusually 



