]32 Pearl aud Surface. 



The values given in the percentage columns of these tables are 

 shown graphicallj- in Fig. 7. Tliis figure corresponds to Fig. 6 in all 

 particulars. 



Fig. 7 shows in general the same kind of facts noted in Fig. 6. 

 In most instances a larger proportion of the observations fall in the 

 quintile in which the plants end than in any other. On the other hand 

 the deviations are in no case so great as in the corresponding diagram 

 in Fig. 6. In the majoritj' of cases the observations do not deviate 

 from the theoretical mean more than three times the standard deviation. 

 The greatest deviations are again to be found in the case of very large 

 and very small plants. 



These differences can best be comprehended in the form of root- 

 mean-square deviations. These have been calculated in the same way 

 as for tables 7—9. These constants are tabled in the next to the last 

 column of tables 10 — 12. They are shown graphically in Fig. 9. 

 Figs. 8 and 9 have been placed next to each other to facilitate com- 

 parison between them. 



Fig. 9 shows in a more exact manner the differences which we 

 have already noted from Fig. 7. In every case these root-mean-square 

 deviations are significantly greater than the standard deviations of simple 

 sampling. On the other hand, the deviations in quintiles I and V are 

 not nearly so great as the corresponding ones in Fig. 8. In Series B 

 and C the greatest deviations occur in quintile I. Series C does not 

 show any greater deviation in quintile V than in any of the three 

 preceding quintiles. 



On the whole the deviations shown in Fig. 9 are considerably 

 smaller than those shown in Fig. 8. From this it follows that plants 

 relatively large at the end of the season are less likely to have been 

 relatively large during the entire season than are the plants which are 

 large at the beginning of the season. The same is true of small plants. 

 Thus the relative position of the plants at the beginning of the season 

 is a much better criterion of their average relative size for the whole 

 season than is their relative position at the end of the season. Or 

 again, on the hypothesis suggested earlier, the relative position of the 

 plants at the beginning of the season is a better criterion of their 

 probable genotypic constitution, so far as height is concerned, than is 

 their relative position at the end of the season. The reason for this 

 difference will be discussed at another place in this paper. 



