26o 
H. S. REED 
law. This tendency was especially pronounced in the case of plants which 
ended in quartiles I and IV, in other words, in the extreme classes. This 
tendency naturally became more pronounced as growth progressed, so 
there were no changes in the quartile positions of the plants in the last part 
of the season. This seems to indicate that the plants, as a rule, had a 
uniform rate of growth, and that the plants which were small at the end of 
the season were in that class because they had an inherent tendency to 
smallness all through the season. Conversely, the plants in the highest 
group were not there because they were mediocre plants which grew longer 
than the rest, but because they were superior all through the season. 
Of the 150 observations made on the plants ending in the first quartile* 
67 percent were in that quartile (table 8) and 58 percent of the 140 obser- 
vations made on the plants ending in the fourth quartile were in that quar- 
tile (table 11). 
As in the former comparisons, we see that plants which fell into a given 
class at the end of the season showed a tendency to group themselves in or 
near that class from the outset. Table 12, which assembles the data of the 
four preceding tables, affords a convenient means of making comparisons. 
As in table 7, the actual standard deviation shows that the height of the 
plants varied more widely than would be the case if their heights were de- 
termined by pure chance. These relations make it quite evident that the 
height of any given group of these plants is not determined by such casual 
factors as variations in soil, water, light, and other external factors, since 
they would produce a more nearly random quartile distribution. It seems, 
on the contrary, that the quartile distributions of the measurements of a 
plant are determined by internal, inherited factors which operate so strongly 
and characteristically that they outweigh the influence of casual factors 
and assert their dominant influence in the behavior of the sunflower. In 
short, there appears to be some agency at work on the plants which pro- 
duces deviations from their theoretical means greater than those which are 
to be expected upon the basis of pure chance, and this influence appears to 
be most effective on the plants in the extreme classes. The attempt will 
be made, in a subsequent paragraph, to elucidate somewhat the nature of 
this influence which manifests itself in the quartile positions of these groups 
of the population. 
It may be of interest, especially for those not familiar with the use of 
the above described methods, to examine the actual figures. It might be 
thought that the classes are too broad to reveal actual differences, or that 
rapid growth in the first part of the season would sec a plant so far ahead of 
others that it would stand in an upper quartile, though making smaller 
growth subsequently. I have thought it worth while, therefore, to present 
in table 13 the successive increases in mean height of the plants ending in 
the .various quartiles. Inspection of the table shows that after the 14th 
day the relative rates of growth are consistently larger in the higher quar- 
