GRASSHOPPER EGG-POD DISTRIBUTION 15 
In many instances from 15 to 30 fields are needed to obtain a satisfac¬ 
tory standard error of the area mean. For this purpose not many more 
stops will be required for a survey of several similar counties than would 
be needed for a single county. Distribution of stops over an area should 
provide for representation of principal subareas or types, should consider 
economy in travel, and should exclude personal choice. Randomness is 
desirable but may be hard to attain. It may be attained in part by sam¬ 
pling more fields per county or district. 
Several suggestions for improvement in efficiency of sampling have been 
considered. The greatest contribution to accuracy would be made by 
increasing the number of fields sampled, but this procedure would be 
rather expensive. Restriction of within-field sampling to permit increase 
in the number of fields is not promising, because so much work would be 
needed to sample only a few more fields. Modification of within-field 
sampling is also unpromising; increase in amount and distribution of 
more than five 1-square-foot units does not give much gain in informa¬ 
tion, and more “bunching” results in considerable loss. Surveying grain- 
fields alone does not seem to offer enough advantage to compensate for 
the reduction in information. Reduction in work in the margins seems 
well justified. Stratification, and also modification of number of fields in 
accordance with intensity of infestation, have been previously discussed 
and seem to have promise. Prediction of regional needs and transportation 
must be attacked before local distribution. 
The idea of consolidating similar and contiguous counties into a 
single district seems very promising. If there could be some lessening of 
dependence on county lines, adequate estimates could probably be made 
from fewer fields, travel could be planned more efficiently, and more 
representative sampling of important crop types could be carried out. 
Estimates of bait and transportation for regional needs could readily be 
governed by such a survey, although the utilization of county machinery 
might still be required to ascertain local distribution. The development of 
district sampling would make it easier to stratify and modify the intensity 
of the survey. Under this plan the total number of fields would be larger 
than for any single county, and greater freedom should be possible in 
planning for their selection. 
CONCLUSIONS 
It is suggested that five 1-square-foot units be examined in each field 
and two similar units in its margin. For convenience, the field units may 
be located on an arc cutting well into the field, and the sampler may 
return along the margin. Units should be taken by a method that would 
ensure freedom from personal choice. Stops should be distributed with 
respect to crops and localities so that the samples will be as representative 
as possible within the limitations of time and accessibility. Randomness 
is desirable but not fully attainable. A restricted random plan is worthy 
of study. Probably not less than 10 field stops should be made in a county 
or a group of similar counties, and 15 to 30 stops would be better. 
