2 BULLETIN" 529, U, S. DEPARTMENT OF AGRICULTURE. 
shows this relation for a group of 378 farms in southeastern Penn- 
sylvania. The results indicate that under the conditions prevailing 
in this locality, and with the methods practiced by local farmers, 
the point of diminishing returns is reached when the yield on a given 
farm reaches about 40 per cent above the general average of the com- 
munity. Yields higher than this appear to be obtained at an ex- 
pense greater than the increase in income due to the increased yields. 
The figures would naturally differ for different regions. 
Table I. — Relation of crop yield to labor income. 
Groups of farms based on yield per acre. 
Average yields ex 
munity average. 
in percentage of the corn- 
Average labor income expressed in percentage of the 
community average 
84 and 
less. 
140 and 
over. 
HOW FARM RECORDS ARE OBTAINED. 
Knowledge of the details of farm practice and of the results arising 
from this practice may be obtained in two ways. First, careful 
records may be kept of the details of the farm work and the business 
transactions of the farmer. Second, such details may be obtained 
by interviewing farmers who 'give them as accurately as may be 
from memory, or from such desultory records as may have been made 
of the farm operations. The first of these methods involves years of 
labor and enormous expense; the second gives an enormous amount 
of data in a short while and at a nominal expense. The question is 
as to the relative accuracy of these two methods. 
When farm management investigations first began it was supposed 
that the only way to get at the facts of farm practice with a degree 
of accuracy sufficient for investigational purposes was b} T means of 
carefully made records. Accordingly, cost-accounting records were 
begun on a large number of farms. It was soon perceived, however, 
that the cost of such records and the time required for their accumu- 
lation were serious obstacles. Furthermore, practice differs so 
widely in different regions, on different farms in the same locality, 
and even on the same farm from year to year, that it would be an 
interminable task to collect sufficient data in this manner to solve 
the numerous problems which the study of farm practice had re- 
vealed. Because of the amount of time involved the results would 
frequently be out of date before the work could be finished. Finally 
it was decided to give the second method a trial. At first many 
students of farm management had misgivings as to the validity of 
data obtained from farmers who keep few or no records. Accord- 
ingly, in order to test this point a number of investigations were un- 
