of injured apples, — excluding for the moment a single exceptional 
tree in the seventh row, which gave but i6 per cent, of injury. That 
is, with this exception, the ratios ran from row to row with virtual 
uniformity as far as the third row from the dividing line. In the 
other half of the orchard, that left unsprayed as a check, the injury 
on lo trees distributed through six rows — from the fourteenth to 
the twenty-second-T-ranged from 87 to 99 per cent., and in this 
series also there is no uniform progression as one goes from the 
dividing line outward. 
We may make this influence of a proximity of check and ex- 
perimental plots more definitely apparent by saying that if the two 
adjacent rows only had been brought into comparison the difference 
in injured apples would have been only 38 per cent., while the real 
efifect of the treatment was a difference of 63 per cent. Or, to put 
the facts still more definitely, a comparison of the product of the 
adjacent rows would indicate a saving, by spraying, of 47 per cent, 
of the apples which would otherwise have been injured, while the 
real saving was one of 70 per cent. The apparent benefit would be 
but two thirds the actual. 
We find, in short, two to four rows running through the center 
of this orchard, one or two in each of the two plots, the product of 
which has been rendered unfit for use in an exact comparison of 
the condition of these two plots with reference to experimental re- 
sults. Such a strip must evidently always exist under similar cir- 
cumstances ; a kind of neutral zone belonging properly to neither 
plot, and unfit for comparison or for use in a discussion of the out- 
come of an experiment. The importance of this conclusion is evi- 
dent when one scans the ])ublished statements of similar experiments 
and finds that in many cases the adjoining plots are so narrow that 
nearly or quite all of both must lie in this neutral zone. Sometimes, 
indeed, one can not learn from the description of an experiment the 
form or dimensions of the plots, and consequently can not judge how 
far some portions of each or either may have been removed from 
this central belt. Any experiment of the kind on diffusible insects 
or fungi which does not take these facts into account, and from the 
discussion of which a siifficient central strip is not omitted, must 
be regarded as unsatisfactory, if not rejected outright as inexact. 
Both the amount and the extent of the mutual influence of check 
and experimental plots will vary greatly with different insects, and 
with the same insect species under different conditions of abundance 
and different kinds of treatment. Where infestation is unusually 
heavy this interference with effects of treatment would, other things 
being equal, be more pronounced. To determine, in any case, how 
much of a central strip must be omitted, it would be necessary to 
