PERSONAL EQUATION IN BREEDING EXPERIMENTS. 359 



It was not feasible to ask all of the original fifteen observers to 

 go to the labor of recounting these ears. Second counts made 

 after a relatively long lapse of time are, however, available from 

 three observers (namely, VI., VIII. and IX.) for all four ears. 

 \\hile this gives only comparatively meager data, -till some 

 point-- of interest appear. The-e d.iui are given in Tables X., 

 XI. and XII. It should be -aid that the recounting was done 

 in the same way as the original count. In each case the observer 

 had no access to the original data while the second count was 

 in progress. No one of the three had any remembrance of what 

 hi- r her) original counts were. The writer has not been able 

 to di-cover any factor which would make these recount- any- 

 thing other than what they were intended to be. namely, really 

 independent determinations of the same mail-rial by the same 

 observers after a long lapse of time. 



It will be remembered (cf. p. ,}_).<) snf>ni\ that one kernel I'n.m 

 ear No. 10 was lost in the course of th' nal counting, h is 



theieion- obvious that all the recounts of thi- ear mu-t of m . 

 HI y be one kernel smaller than the tir-t count-. 



TAUI.I-: X. 

 INAL AND SECOND COUNTS - 8 TO u i .\. VI. 



I Is. 



The data in the-e table- indicate that, -o far at lea-t a- tlie-e 

 thrc-e observers are con<vrne.d. the judgment of the indi\idiial 

 i- reasonably constant. This i-, plain it' the total deviation of 



