\2i Pearl and Surface. 



has been calculated for each case aud found to vary from 3"2 to 3 "5 

 percent, with an average of about 3" 3-4 percent. Since this deviation 

 is comparativelj' slight it has not been thought necessary to encumber 

 the tables with these figures. All the necessary data are given for 

 the calculation of these constants if anyone desires them. 



The data given in Tables 7 to' 9 are shown graphically in Fig. 6. 



From these figures it is at once noted that the plants starting 

 in quintiles I and V show very marked deviations from the theoretical 

 means. The greatest positive deviations are, in each of these cases, in 

 the quintile in which the plants start. The greatest negative deviations 

 occur in the quintile farthest removed from that in which the plants 

 start. This of course indicates a strong tendency for plants which are 

 relatively small at the beginning of the season to remain relatively small 

 throughout their growth. Likewise relatively large plants tend to remain 

 relatively large. 



An idea as to the significance of these deviations can be obtained 

 by comparing them with the limits of 3 ö shown by the dotted lines in 

 the figure. If the quintile distributions had been governed by the con- 

 ditions of simple sampling none, or very few, of the observations would 

 deviate so far from the mean value as 3 a. 



Turning now to plants starting in quintiles II, III and IV, it is 

 evident that, in no instances, are there such wide deviations from the 

 theoretical mean. In some cases, for example plants starting in 

 quintile IV for series A and C, the deviations are little if any greater 

 than might be expected on the theory of probability alone. In some 

 instances, especially the plants starting in quintile II, there is a tendency 

 for a large percentage of the observations to fall in this quintile. In 

 other cases there does not appear to be any greater tendency for the 

 measurements to fall in the quintile in which the plants started than 

 in a neighboring quintile. 



In general then there is not such a marked tendency for medium 

 sized plants to remain medium sized throughout the season as has been 

 noted for the extremes. Instead some of the medium sized plants 

 become relatively large and some become relatively small, so that for 

 the whole season there is a much more even distribution of their 

 measurements than of the very large or the very small plants. The 

 significance of this fact will be considered after we have discussed a 

 method of measuring these deviations. 



