REGENERATION OF FINS OF FUNDULUS HETEROCLITUS. 389 



two series do not differ as to that character. If, however, the 

 mean difference is I -f- times the probable difference then it is pos- 

 sible that the two series do differ as to the value of that charac- 

 ter. Also if the mean difference is 2 + times the probable dif- 

 ference then it is probable that the two series differ. Finally if 

 the mean difference is 3 -f times the probable difference then it is 

 certain that the two series differ as to the value of that character. 

 To apply the law here it is necessary to find the mean of each 

 series, the probable error, and from these compute the mean dif- 

 ference and the probable error of difference. 



We can arrange the results as to the caudal fin in Table I. in 

 the form of a table. 



Comparing A and B we see at once that the mean difference 

 is actually less than the probable difference, and hence this pre- 

 cludes the possibility of any rational conclusion that there is any 

 difference in regeneration in cases A and B. But we have seen 

 that injury in case of A was less than in B. Hence we cannot 

 conclude that the rate of regeneration is greater in the case of the 

 less injured nor in the case of the more injured. The regenera- 

 tion is the same. 



We find the same result when we compare B and C, and also 

 when we compare A and C. This experiment tends therefore to 

 negative the results of both Zeleny and Emmel. 



We have thus far tested the question by comparing results in 

 these three lots as to regeneration of the caudal fin. But we can 

 also apply the test as to regeneration of the pectorals. We can 

 average the results of the two pectorals in Lot C. 



Mean. Prob. Error. Mean Diff. Prob. Diff. 



Series B = .0694 .002353 



Series C= .0660 .002752 .0034 .0036 



In this case also the mean difference is less than the probable 



