324 



Aaron Franklin Shiill 



no very clear evidence of any influence of food on the proportion 

 of male-producers; what evidence there is in these parts of the 

 experiment would have gone to show that starvation reduced 

 the percentage of male-producers. But in those parts of the 

 experiment from July 28 to August 25, and from October 13 to 

 October 25 the evidence is very decisive on the other side; and the 

 period from September 6 to September 20, while not so marked, 

 also points strongly to the conclusion that starvation is accom- 

 panied by an increase in the proportion of male-producers. 



TABLE IV 



A summary of the data given in Table III, dividing it into periods according to the percentage of male-pro- 

 ducers in the well-fed series. 



These apparently conflicting results do not show that the food 

 culture nas no influence on the proportion of male-producers. It 

 is to be noted that the proportion of male-producers in the starved 

 line is much more constant than in the well-fed line. The change 

 in the proportion of male-producers from one group to the next 

 is always of the same sign in the starved series as in the well-fed, 

 but is less in amount. This greater constancy of the starved fami- 

 lies is well shown in the figure (see opposite page), in which the 

 percentage of male-producers in each series is represented by a 

 curve. To eliminate the minor fluctuations from one generation 

 to the next, the experiment has been blocked oft in five-day peri- 

 ods in July and August, and six-dav periods after September i. 

 The last period includes seven days. The aim was to include 



