DYNAMOGENIC INFLUENCE OF LIGHT 363 



ishing the average by 100. The remainder directly expresses 

 M A . 



The distribution of these ratios was found to justify the appli- 

 cation of the conventional methods of statistical presentation. 



The number of measurements (N) being small for each indi- 

 vidual subject, elaborate treatment was considered unnecessary, 

 and the measures of variability and reliability were got by the 

 use of approximate formulas. The distribution in each indi- 

 vidual-case was assumed to be normal; the mean variation was 

 calculated by Dunlap's method; 7 and the probable error of the 

 mean was taken as 



0.8453 X MV 



PEM= VN-I 



The improvement during the first sitting being sometimes so 

 large as to distort comparison between P L and P D , and as other 

 sources of disturbance (such as illness, etc.) occasionally intro- 

 duced questionable values, Chauvenet's criterion of exclusion 

 was applied to the extreme measurements. The excluded meas- 

 urements are shown in the tables but are omitted from the aver- 

 ages. 



By reference to table 2 it will be seen that very important in- 

 dividual differences exist. Six subjects gave differences in rela- 

 tive performance in favor of the light, varying between 1.9 and 

 6.7 per cent. Being fairly large with respect to their probable 

 errors, these differences may be considered statistically reliable. 

 Seven subjects showed differences in favor of the light varying 

 between 2.7 and 0.01 per cent, but these differences are so small 

 with respect to their probable errors that they may reasonably 

 be attributed to chance. That is, on the basis of the laws of 

 chance, one would expect a reversal of the direction of the dif- 



7 Dunlap, Knight: Obtaining the mean variation with the aid of a calculating 

 machine. Psychol. Rev., 1913, xx, 154-157. For correction of typographical 

 errors in his formulas, and for an extension of the method, see Johnson, B. J. : 

 Dunlap's method for the mean variation. This Journal, 1918, i, 325 f. Cf. also 

 Whipple, G. M. : Manual of Mental and Physical Tests, i, 21 f . Baltimore, War- 

 wick & York, 1914. 



