GRASSHOPPER EGG-POD DISTRIBUTION 
9 
Number and Location of Field-Sample Units 
Data presented above show that populations vary considerably more 
between fields than between units within fields, that there is a tendency 
to greater variation between distant units in a field than between adja¬ 
cent units, but that in light populations neither type of variation is 
pronounced. Two adjacent units of square foot each can be viewed as 
one 1-square-foot unit, thus affording a basis for comparing numbers and 
sizes of units. If two adjacent units were exactly similar, one would be 
as good as two, but if they differ greatly no special gain would come from 
spreading the sampling. 
The South Dakota data (table 5) has been used in calculating expected 
standard error of different combinations, using established methods. 7 
These calcu lations are given in table 7. 
Table 7. —The reliability of a grasshopper egg-pod survey based on different 
combinations of fields, field-sample locations, and units per field location 1 
Fields 
sampled 
Sample locations 
per field 
Units per sample 
location 
Standard error of 
mean egg pod 
population per unit 1 
Number 
Number 
Number 
20 
5 
1 
0.052 
20 
5 
2 
.042 
10 
10 
1 
.057 
10 
5 
2 
.059 
10 
5 
1 
.074 
10 
1 
10 
.077 
5 
5 
2 
.084 
10 
1 
5 
.089 
^ach field unit contained square foot. To place the standard error on a 1-square-foot basis it must 
be doubled. 
It can be seen that there is an advantage, but a very limited one, in 
spreading sampling within fields. Taking }^-square-foot units instead 
of 1-square-foot units, but doubling the number, will give only a slightly 
lower standard error. Taking the same number of units, but reducing 
their size by one-half, will increase the standard error considerably. To 
take all units in one place is going too far in the direction of “bunching” 
and increases the standard error markedly. In light infestations, such as 
the example just given, five 1-square-foot units are practically as good 
as twice the number of units half that size; but in denser infestations 
there would probably be more advantage in spreading within-field sam¬ 
pling. The five 1-square-foot units seem to offer a workable and sound 
combination. An increase in number of fields is more potent in reducing 
standard error than are increases in within-field sampling. 
Distribution of Units Within Fields 
The data in tables 2 and 3 show that location within the field has little 
systematic effect on the accuracy of the results. This fact indicates that 
sampling within the field need not follow an exact pattern that gives 
7 See p. 3, footnote 4 (sec. 17.8). 
