354 RAYMOND PEARL. 



This table brings out several points which need discussion. 

 These are: 



1. The amount of variation, both absolute and relative, in 

 the counts is shown by the measures here used to be very large 

 for some ears and classes of kernels. For no ear, taken as a 

 whole, can the variation fairly be considered negligible. Thus 

 the conclusion previously reached by another method is con- 

 firmed. 



2. The amount of variation in the sorting and counting is 

 distinctly different for the different ears. From the values of 

 the constants it would appear that ear No. n presented the 

 greatest difficulty in respect to the classification of starchy 

 kernels. In respect to sweet kernels ears No. 8 takes rank as 

 offering the greatest difficulties. The starchy kernels of ear No. 

 10 were the easiest to classify of all starchy kernels. In the case 

 of sweet kernels ear No. 9 had fewer intermediates (i. e., was 

 easier to classify) than any other ear. 



3. Relatively there was closest agreement among the observers 

 in respect to yellow starchy kernels, and least agreement in 

 respect to white sweet kernels. This table illustrates the fact 

 which was evident to the observers themselves, that there were 

 marked differences in the ease with which the kernels of different 

 ears and different classes could be sorted. 



Now while it has been shown that the fifteen observers do 

 not agree in their classification and counts, and that the dif- 

 ferences are too large to be neglected, it may fairly be asked if 

 the same result would appear if the group of observers participat- 

 ing were not merely scientifically trained and familiar with maize, 

 but in addition had had a considerable amount of actual expe- 

 rience in the detailed study of variation and inheritance in plants. 

 In other words, is not that special familiarity with the object 

 which comes with the active prosecution of research in a particu- 

 lar field worth something in reducing the magnitude of one's per- 

 sonal error or "equation"? To get some light on this point Table 

 VIII. has been prepared. This is made up in exactly the same 

 way as Table VI., except that only observers VI., VI I ., VI 1 1 ., IX., 

 XI., XII., XIII. and XIV. are included. These eight observers, 

 comprising the staffs of Professor Johannsen's and the \\riirr'- 



