156 



Pearl and Surface. 



An examination of tables 13 to 15 shows that these standard 

 deviations range from zero to 1'54. Only two plants (table 13) show a 

 standard deviation of 1-5 or over, and only two other standard deviations 

 are as high as 1*4 (table 15). This shows at once that the measure- 

 ments of most of these plants are not as widely scattered about their 

 means as would be the case if the observations were equally distri- 

 buted among all five quintiles. 



In order to get at the general trend of these constants, frequency 

 distributions of the standard deviations for each series may be made. 

 Such distributions together with their means and standard deviations 

 are given in table 17. 



Table 17. 



Frequency Distributions of the Standard Deviations of the 



Mean Quintile Positions of Individual Plants. 



From this table it is noted that the distributions are fairly smooth 

 considering the small number of observations and also that they are 

 reasonably symmetrical. The means of these distributions range from 

 0-77 to 0-87. Now it was shown above that if the observations were 

 distributed among the quintiles at random the average standard deviation 

 would be 1'4142. The difference between this and the largest observed 

 mean standard deviation is more than 25 times the probable error. 



This small mean standard deviation and the uniformity with which 

 it occurs in all three series shows conclusively that there is a strong 

 tendency for the measurement of any individual plant to be closely 

 grouped about its mean. That is, this shows again the point which has 

 been previously brought out from other evidence, viz., that there is a 

 strong tendency for plants relatively small at one stage to remain rela- 

 tively small throughout their 'growth and similarly for medium sized 

 plants and for large plants. 



