A PHYSIOLOGICAL CHARACTER. 131 



struck the surface of the water. These differences in reactiveness 

 were not due to temperature differences, since they occurred in cases 

 where there were no temperature differences in the water. It is 

 assumed that this difference in reactiveness was due to the accumu- 

 lation of carbon dioxide from the observer's breath in the water 

 contained in the tank, though exhaled volatile organic substances 

 may have been wholly or partially responsible. 



3. RELATIVELY TEMPORARY CHEMICAL DIFFERENCES IN WATER USED IN 

 EXPERIMENTAL TANK. 



Probably the environmental factor of greatest importance con- 

 sisted in differences in the chemical content of the water used in the 

 experimental tank, although this water was always obtained from the 

 same pond and handled in the same fashion. Such chemical differ- 

 ences in the water would account for the pecularities in reaction-time 

 which were so apparent in the reactions of all the broods tested on 

 certain days. Not infrequently, between successive days or be- 

 tween periods of a few days' duration, considerable difference was 

 noted in the general reactiveness of the stock as a whole. On one 

 day the individuals tested would react fairly promptly, while on the 

 following day they might react conspicuously less promptly. This 

 clearly points to environmental influences operative temporarily but 

 pronouncedly. These temporary differences in reactiveness, pre- 

 sumably due to differences in the water, are well illustrated in table 

 2 (page 18), showing in abbreviated form the data for the broods of 

 Line 695, tested on August 29, 30, and 31, 1913. 1 



The data given in table 2 present an extreme case selected to 

 show the environmental influence upon reaction-time (and upon the 

 occurrence of negatively reacting individuals) at its maximum. The 

 averages for the data for the two strains of Line 695 for the 3 days 

 shown in table 2 are 401, 636, and 398 seconds, based upon 87, 190, 

 and 50 individual reaction-times. The mean for the second of these 

 days is 235 seconds or 58.6 per cent greater than that for the first 

 day and 238 seconds or 59.8 per cent greater than that for the third 

 day. The standard deviations and probable errors for these means 

 were not determined, but obviously (from other experience in deter- 

 mining probable errors with such data) these differences are of large 

 statistical value. As already noted, this is an extreme case (and it is 

 doubtful if many as extreme could be found in any portion of the 

 later data), but the factor operative here presumably was also 

 operative to some extent in many of the tests conducted in making 



1 These data were obtained with due regard to the necessity for changing the water in the 

 experimental tank at frequent intervals and they fit the requirements of the present purpose very 

 well. They are part of the data of the first test series conducted. As a test series this was not 

 satisfactory, as the numbers of individuals in the corresponding plus and minus broods were not 

 as nearly exactly the same as in later test series. Also, the broods were not always paired here, 

 i. e., in the usual course a plus brood was always tested just before or immediately after a minus 

 brood of the same age and containing the same number (within one or two) of individuals. 



