The time required for the eye to change the direction of the line of regard to an object appearing suddenly somewhere in the field of vision usually ranges from 150 to 250 σ, depending somewhat upon the amount of practice that the subject has had in the test. This measurement was used in the normal series of 1917; the apparatus and procedure were identical with that employed in the low-diet research. From 10 to 35 reactions were obtained on each subject. A larger number than 10 was taken on every subject, but occasionally, for one cause or another, some of the records would not be legible. Records from the 63 men in the series of 1917 show a general average of 244 σ, with a standard deviation of 46 σ; the individual averages range from 176 to 415 σ. Besides the individual with the average of 415 σ, which may not be a typical eye reaction, there were 3 other men in the normal series of 1917 who had reaction times of 340 σ or longer, that is: 342, 347, and 346 σ. Not one of the 63 subjects had previously served in this measurement. It was deemed of interest to take account of the shortest reactions which the individual subjects made,1 as these minimum reaction times would appear to be less complicated by fluctuations in attention than is the general average of a series of reactions. The five shortest reactions made by each of the 63 men were therefore averaged. The general average of the 63 minimum averages shows 193 σ, with a standard deviation of 44 σ.

With but few exceptions, all of the eye reactions for the men of the normal series of 1917 were included in the averages. An inspection of the records of prolonged series of reactions taken with other individuals and also in the present low-diet research reveals the fact that frequently the first reaction made by a subject is unusually long and apparently quite out of uniformity with the subsequent reactions. On the other hand, these long initial reactions were not invariably found. While on general grounds it is not permissible to omit certain data on the supposition that it fails to fit in with one's standard of normality, yet it is possible to adopt some arbitrary method of dealing with such difficulties. The following rules were employed in averaging the eye-reaction data for this research:

(1) If, in the first set of eye reactions taken with a subject, that is, the first time this measurement was made on him, the first two reactions were abnormally long (350 σ or longer), they were both omitted from the average and considered as preliminary practice.

(2) Two plates were taken on the subject at each sitting or experimental evening. At each subsequent evening after the first, if the first reaction on either plate was abnormally long, it was also omitted from the average and considered as preliminary practice for the evening.

(3) Abnormally long reactions which occurred at any other places in the records than those specified were invariably included in the averages.

The individual eye-reaction time averages with standard deviations and coefficients of variability are given in tables 167 and 168 for Squads A and B, respectively. Squad A shows 231 and Squad B 237 σ as averages for the first reactions taken from the two groups of men with this technique. The average standard deviation in each case amounts to 20 per cent of the reaction time, the two standard deviations being 47.9 and 50.6 σ respectively. These values correspond very nearly with the 244 σ, and coefficient of variability of 19 per cent for the normal series of 1917. The averages for Squads A and B are about 10 σ less, due, no doubt, to our method of dealing with the preliminary long reactions just mentioned. The average reaction times for Squads A and B for other dates than October 28 and November 4, with the exception of the fragmentary data for November 11 (Squad A), always show smaller values than on the original first date. Such a practice effect is normal for untrained subjects in this measurement. It was found by Dodge and Benedict in the eye reaction data in their alcohol investigation.1 It is clearly evident in an unpublished series of eye reactions secured with trained subjects in this Laboratory; it also shows definitely in this low-diet research.

1Dicfendorf and Dodge, Brain, 1908, 31, p. 472.

Table 167. - Squad A - The Eye-Reaction Time And Its Variability

[M. in σ, S. D. in σ, and C. in per cent].

1Very few data secured, but all are averaged in.

2Action of shutter defective.