-105- 
ncrmal distribution susceptible of the usual statistical treatment. Such an 
experiment would be time-consuming and of little, if any, practical value. 
Although the 1-square-foot samples varied less among themselves in this 
experiment and hence were possibly more reliable indicators of -nooulation 
than the 1/5-square-foot samples, there is probably not enough difference in 
variation to justify the increase in work involved in their examination. 
Furthermore, reducing the size of the sample makes it possible to handle a 
larger number of them per given area in the time available, probably afford- 
ing better chances of finding bugs if present. 
The l/5-square-foot samples also indicated a much higher infestation 
than did the 1-square-foot samples. Probably this would be generally true 
because the smaller samples would naturally be taken from the best parts of 
the grass clumps, while the larger samples would necessarily include a vary- 
ing amount of the sparse vegetation less likely to harbor bugs. This very 
limited experiment would indicate that the number of bugs per 1/5 square foot 
should be multiplied by two instead of five, to give the infestation to be 
expected in 1-square-foot samples composed entirely of grass. This relation- 
ship would undoubtedly be variable, depending on kind of grass, method of 
sampling, and other factors. 
Replicated Survey of Tippecanoe County 
Between November 22 and December lU , 193^? 180 l/5-square-foot samples 
were taken in Tippecanoe County, 20 being taken from each of the 9 areas 
sampled during the previous fall survey in standing corn, the entire county 
being roughly divided into three rows of 3 areas each. The sampling was re- 
stricted to well-drained south and west exposures and to the bunch or tuft- 
forming grasses, exclusive of timothy. The most favored grasses were not 
present throughout all the areas, hence it was necessary to take the samoles 
from the best ones available. Twelve species were involved, the number of 
samoles being as follows: Androoogon f urcatus , SI; Andropogon scoparius, 2; 
Slymus canadensis, 19; Elymus virginicus, 15; Elymus striatus, 2U; Sporobolus 
asper , 15; Tridens f lavus , U; Pan i cum virgatum, 13; Atheropogon curtipendulus, 
2; Hystrix hystrix, 1; Sorghastrum nutans, J>] Spartina michauxiana, 1. 
The number of chinch bugs in each sample was determined and the data 
were made up into 20 sets, each set including 1 sample from each of the 9 
areas so as to be representative of the entire county, no sample being used 
in more than 1 set. Thus each representative county sample was a completely 
independent set of 9 l/5-square-f oot samples. The results are summarized as 
follows: ITumber of l/5-square-f oot samples in county samole , 9"> number of 
county samoles, 20; average number of bugs per 1/5 square foot in county 
samples, minimum, 57; maximum, 35^; mean, lUo. 
It is seen at once that the average number of bugs per county sample 
varied greatly, notwithstanding the attempts to reduce variation by restric- 
tion of the sampling to the most favorable host plants and exposures. Sta- 
tistical analysis of these county samples by the usual methods was found to 
