686 
AGRICULTURE: LIPMAN AND LINHART Proc. N. A. S. 
4. The kind of fertilizer used seems to have been without significance, 
one being just as good as another, subject to the Hmitations above noted. 
5. The amount of fertihzer used seems Hkewise to have been without 
significance (subject to the same hmitations). 
6. Even when fertihzer experiments are properly planned and the results 
adequately studied by statistical methods, our present knowledge of the 
enormous variability of all soils and plants renders the data from any 
given fertilizer plot of value only on that plot, no matter how near the ex- 
perimental one. This important consideration renders it highly probable 
that no fertilizer experiment as ordinarily conducted is possessed of suffi- 
cient practical value to justify the large expenditure of money, time, and 
energy involved. 
We wish to emphasize that we are not desirous of making a fetish of the 
application of statistical study to fertilizer experiments. Nevertheless, 
we must say that if statistical methods, inadequate as they may be, should 
not prove applicable to an evaluation of the results of fertilizer experiments, 
the latter must be accounted of even less value and significance than we 
have here accorded to them. 
