35 2 RAYMOND PEARL. 



yellow at the other end, with each intermediate step practicallx 

 as small as one cared to make it. An attempt was made to 

 obtain photographs of such series of kernels which would demon- 

 strate the fact of this gradation pictorially, but the photographic 

 resources at command were not equal to the task and it had to 

 be abandoned. 



4. The data presented fully demonstrate, I think, the interest- 

 ing fact that if each of these fifteen competent, and with one 

 exception (No. X.), specially trained observers had independently 

 undertaken an investigation of Mendelian inheritance in maize, 

 and all used the same seed, of at least the two strains here em- 

 ployed, grown their crops in the same place, and even studied 

 identically the same progeny ears, no two would have fully agreed 

 in the numerical values of the F 2 ratios. 



II. 



Let us now consider the question as to whether these deviations 

 due to personal equation are of sufficient magnitude to be prac- 

 tically significant. The whole of the remainder of this paper 

 will be devoted to a discussion, from different standpoints, of 

 the quantitative aspects of the recorded classifications of the 

 several observers. All these data will bear upon this general 

 point. To answer the question specifically raised in this section 

 it will only be necessary to show the range of the variation 

 exhibited in the counts made. Table VI. gives for the four ears 

 and the four classes of kernels on each ear (a) the mean numbers 

 of kernels found by averaging the counts of all observers, (6) 

 the minimum and the maximum recorded number of kernels, (c) 

 the total range of variation shown in the records, and (d) the 

 percentage which this range is of the mean of the same class. 



It is evident from this table that the personal element is one 

 of real significance. When two careful observers can diflVr in 

 their count of the same set of objects by as much as one and a 

 half times the actual number of the objects counted the factor 

 which leads to this difference is certainly not to be neglected- 



An examination of the standard deviations and coefficients 

 of variation of the counts leads to the same result. These 

 constants are shown in Table \ II. It should be said in this 



