50 HAROLD C. BINGHAM 



intermediate stage. On the other hand, one might assume 

 with some justice that, in more than 2000 tests distributed 

 among eight individuals, all irregular fluctuations would tend 

 towards an equal distribution among the different size relations. 

 In this event, the time records represented by the final averages 

 of table 4 might be said to substantiate the preceding conclu- 

 sion, based upon the number of correct choices, that the maxi- 

 mum which the chick can discriminate under the conditions of 

 this experiment lies somewhat above o 15-f-. 



Statistically considered, however, the averages of this table 

 have very little significance. The mean variation, for example, 

 from the 11, 16, and 20 seconds averages are respectively 8, 13, 

 and 12 seconds. These mean variations are nearly as great as 

 the averages themselves, and they are, in every case, as great 

 as or greater than the differences between any two of the aver- 

 ages. 



These averages of table 4 are even more questionable when 

 one considers the individual records on which the table is 

 based. A notion of these individual records may be derived 

 from Test Sheet 1 and Test Sheet 2 which are presented to 

 illustrate how the behavior respectively differed under easy 

 and difficult discrimination. Taking these two series as ex- 

 amples, we find that the mean variation represented by Test 

 Sheet 1 is 1.88 seconds which is nearly 59 per cent of the average 

 for the ten tests. In the records presented by Test Sheet 2, 

 the mean variation is a little lower being slightly above 51 per 

 cent. This wide variability suggests that the averages for the 

 different individuals are themselves questionable. Combining 

 with this the variability noted within the table itself, these 

 averages appear as doubtful criteria. 



A number of things might be mentioned in explanation of the 

 variability between the tests of a single series. Wholly secon- 

 dary influences, such as a severe shock, the slamming of a door, 

 or a snow slide on the roof, are often responsible for a sudden 

 fluctuation in the reaction time. A more natural difficulty 

 with the chick which is responsible for even greater fluctuations 

 in reaction time is a tendency to go to "roost" in the dark room. 

 It can usually be aroused by blowing lightly against the feathers, 

 but such behavior plays havoc with the reliability of the indi- 

 vidual time records. 



