292 W. C. ALLEE 



but frequently recurring periods, equally weakened responses, 

 and hence is of equal importance in the ecology of the species. 

 The minor variations in positiveness are probably of little ecologi- 

 cal importance. The isopods that sometimes showed a high 

 percentage of negativeness gave on the average (table 7) as many 

 positive as negative reactions. This means that for these isopods 

 the sign of the rheotactic reaction is of practically no importance 

 in maintaining the position in the stream. 



Following the experiments upon the rheotactic reactions of 

 individual isopods the question arises as to the relation of aver- 

 aged .results obtained by this method with those obtained by 

 group tests as performed in preceding experiments ('12). The 

 two types of results may be seen plotted in chart 8 (p. 283). 

 The group responses are taken from table 2, page 280 (Allee 

 '12). .The chart shows that the results given by the two meth- 

 ods are essentially the same. 



Four of the twenty isopods tested showed a tendency to give 

 a large number of negative reactions under normal conditions. 

 This shows why five stream isopods taken at random give a 

 positive response of from 70 to 80 per cent. The chances are 

 that four of the five will give a positive response of from 80 to 

 100 per cent averaging some place in the upper eighties. The 

 other will give a response of about 50 per cent positive which 

 will cause the average reaction to be between 70 and 80 per 

 cent positive. 



By substituting the plan of individual for group tests the 

 amount of individual variation becomes emphasized. In the 

 highly positive isopods this consists of 20 per cent variation 

 both in the degree of positiveness and in the efficiency of the 

 response. In other and more exceptional cases the range is still 

 greater. It also shows that in some stream isopods a high nega- 

 tive response may accompany a high rate of efficiency although 

 this condition is too infrequent to greatly affect average results. 



In the preceding paper ('12, pp. 338-39) it was shown that 

 the degree of positiveness depends upon the rate of metabolism 

 and can be controlled by certain factors known to affect animal 

 metabolism. In the present series of experiments it has been 



