42 
AMERICAN JOURNAL OF BOTANY 
[Vol. 9 
In the second place, since we have Httle knowledge of the relations 
within the plant, of the interrelations of the plant's activities, even in an 
environment whose composition is measured, but which in kind and quantity 
of components is chosen at random, there is little chance of success because 
the main current or trend of the plant's activity may, and surely does, 
change if the environment is sufficiently altered, or altered in particular 
respects. We know enough on this point to be sure that such an altered 
activity is common. In the third place, since we select our plants on the 
basis of the magnitude of an external character such as height of seedling 
at a certain age, weight of seed, color of seed, etc., and not on the basis of 
a relation between external characters, we are again dealing, not with the 
plant's internal constitution but with the product of the interactivity of 
environment and plant, even though that interactivity occurred before the 
materials came to hand, that is, during seed formation. This should be 
taken to indicate, not that work of this kind is without value, but that it 
would progress more rapidly and more surely if the ever-present variable of 
plant constitution could be evaluated. 
This brings us to another objection. Suppose for the sake of argument, 
it may be said, that internal activity and external character are related 
quantitatively, does it follow that the relationship is either direct or easy of 
discernment? Of course not; if the relationship were direct and easy it 
probably would have been noticed long since. Hope lies in the general 
success of such endeavors in physics and in physical chemistry, but this 
point of view is not proposed as a panacea for all experimental difficulties; 
it is a proposal for a campaign rather than for a coup. A little later I 
shall attempt to consider an avenue of approach, general considerations 
upon which methods may be based, and this will probably prove a more 
satisfactory form of answer. 
It has been objected that, even supposing we have found a mathematical 
expression connecting the activities of a small number of individual plants 
with their measurable characters, such a relation must involve one or more 
constants peculiar to the individual, and since these constants will vary 
in magnitude from individual to individual over an enormous range, the 
information obtained from the expression could not be applied to a large 
assemblage of individuals, such as a field of wheat. This does not seem to 
offer a real difficulty at all. We have information now at hand to furnish 
the probable answer — which is that the information could be applied. 
Physiology is proceeding on a general conviction that contradicts this 
objection. The success of the statistical method, the success attending 
the use of the average in physiological work, both evidence that physi- 
ologists firmly believe that, if a sufficient number of individuals be con- 
sidered, the changes in the results produced by adding more individuals 
will be negligible. In other words, characteristics are not variable through 
an infinite range, nor irregularly through a limited range. On the contrary, 
