PHOTIC REACTIONS OF HONEY-BEE 381 



For example, in the first pair of determinations on bee no. 21, 



957 mc. 24 mc. 



+29.417cm. - ( + 15.847cm.) = + 13.577cm. 



If the great majority of d values are of the first category, viz., 

 negative, we may conclude that the animal experiences a greater 

 impulse to turn toward the functional eye in an illumination of 

 24 mc. than it does in one of 957 mc. If equal numbers of nega- 

 tive and positive values occur, there is no relation between the 

 intensities of photic stimulation employed and the amount of 

 turning. If, however, d is generally positive, we may conclude 

 that the tendency to turn toward the functional eye increases 

 if the intensity of photic stimulation is sufficiently increased. 



Experimentation soon demonstrated that the only satisfactory 

 solution of the problem was to be had through a statistical treat- 

 ment of large numbers of data. Even the more constant animals 

 often varied widely from one pair of determinations to another 

 without any apparent external cause. Therefore, a large num- 

 ber of bees were experimented upon and each individual was sub- 

 jected to many tests, the averages of which were relied upon to 

 indicate the general trend of behavior. In table 2 (appendix) are 

 presented the results obtained from a careful measurement and 

 computation of over two thousand records taken on fifty-two 

 bees. On some individuals as few as sixteen records were taken; 

 on others as many as seventy-four. This difference was due in 

 part to varying longevity of individuals and in part to the fact 

 that more favorable animals were frequently experimented with 

 longer than less favorable ones. The determinations of approxi- 

 mately the first half of the animals were made on the plan of the- 

 first type of experiment, while the remainder were carried out 

 according to the scheme used in the second type. 



From the figures presented in columns F and G of table 2, 

 it is evident at once that there is a marked preponderance of 

 the positive d values over tl^e negative. The ratio of the two is 

 strikingly shown in the frequency polygon (fig. 11) Since the 

 number of d values varied considerably with the individual, due 



