452 SEMI-CENTENNIAL OF TORREY BOTANICAL CLUB 
2. By comparing the progeny of different plants which vary 
in the average of their total product, with the estimate of individual 
variabilities which takes into account the partial variabilities that 
are present. 
. Either method is good if properly carried out. The first 
method, however, is less satisfactory in that it does not consider 
partial variability. The extremes of partial variations may include, 
as they do in chicory, the very lowest and the highest values of 
range. It is, however, the first-named method that Johannsen 
chose to use. Furthermore, the far-reaching conclusions published 
by Johannsen ('o3) were based on the results of оле crop for 
which the offspring of small and large seeds produced by the same 
plant were grown and compared as to size and weight of seeds 
which they produced as classes. As far as any line was concerned, 
the selection constituted a test for the heredity of partial vari- 
ations only, as Belling (12) has already noted. The negative 
results obtained indicate that the weight of individual seeds 
produced in the same pod or in different pods on the same plant, 
as was the case in each of the 19 lines, is largely due to differences 
in nutrition (phenotypic), resulting from position in the pod or 
to relative position of the pod. It seems clear that adequate tests 
for the heredity of variations in the character of weight of seeds 
in beans should be based on the performance of plants as wholes 
rather than on the weight of the individual seeds. 
Johannsen (713) reports progenies beyond his first generation from 
a known parent plant (his 1902 crop) from only two of the original 
nineteen plants. These two were plants of near the average 
rather than the extreme performance. In these two lines of pro- 
genies, however, the performance of individual plants was not 
considered and as selection was on the basis of the weight of indi- 
vidual seeds, those selected for high and low weights may have 
repeatedly had the same parentage. 
Even with this method, however, there is evidence that heredi- 
tary variations of fluctuating nature arose within a line. In 
line I, the weight range of beans produced by the parent plant 
(1901) extended from 550 to 750 milligrams. The next crop of 
seeds (produced by at least several sister plants) ranged from 350 
to 900. For several years later the average weights of the heavier , 
