260 SECTIONAL ADDRESSES. 
Complexity of the Problem : Methods of Attack. 
The agricultural investigator is thus confronted with three closely 
interlocking agencies—the plant, the climate, and the soil—each of which is 
variable within certain limits, and each playing a large part in the crop 
production which it is his business to study. 
Confronted with a problem of this degree of complexity there are two 
methods of procedure : the empirical method of field observations and ex- 
periments, inwhich there is no pretence of great refinement and noexpecta- 
tion that the same result will ever be obtained twice, it being sufficient 
if over an average of numerous trials a result is obtained more often than 
would be expected from the laws of chance ; and the scientific method, in 
which the factors are carefully analysed and their effects studied quantita- 
tively ; a synthesis is then attempted, and efforts are made to reconstruct 
the whole chain of processes and results. The scientific method is, of 
course, the one to which we are naturally attracted. But common 
truthfulness compels one to admit that up to the present the greatest 
advances in the actual production. of crops have been effected by the 
empirical method, and not infrequently by men who are really artists 
rather than men of science, in that they are guided by some intuitive pro- 
cess which they cannot explain, and that they have the vision of the result 
before they obtain it, which the scientific man commonly has not. 
The best hope for the future lies in the combination of the empirical 
and the scientific methods. This is steadily being accomplished by the 
recent strong infusion of science into the art of field experimentation, 
which has much enhanced the value of the field work and the trust- 
worthiness of its results. Modern methods of replication, such as have been 
worked out at Rothamsted, and in the United States by Harris of the 
Carnegie Trust (Cold Spring Harbor), Kiesselbach in Nebraska, Myers 
and Love of Cornell, and others, constitute a marked improvement in plot 
technique. And the figures themselves, besides being more accurate, can 
be made to yield more information than was formerly the case. 
Great advances have been made in the methods of analysing the results. 
The figures are never the same in any two seasons, since the climatic con- 
ditions profoundly affect the yields. A few men, like J. H. Gilbert, have the 
faculty of extracting a great deal of information from a vast table of figures, 
but in the main even the trained scientific worker can make very little 
of them. The reason is that he has been brought up to deal with cases 
where only one factor is varying, while the growth of plants involves the 
interaction of three variable factors: the plant, the soil, and the climate. 
It is impossible to apply in the field the ordinary methods of the scientific 
investigator where single factors alone are studied; very different 
methods are needed, adapted to the case where several factors vary simul- 
taneously. 
Fortunately for agricultural science, statisticians have in recent years 
worked out methods of this kind, and these are being modified and developed 
by R. A. Fisher and Miss Mackenzie for application to the Rothamsted 
field data. Itso happens that this material is very suitable for the purpose, 
since a large number of the field experiments have been repeated every 
year for seventy or eighty years on the same crop and on the same piece 
of land, using the same methods ; the field workers also remain the same 
for many years, the changes being rare and without break in continuity. 
