This bulletin summarizes a comparison made between 1055 farm records and a geographically stratified random sample of 782 farm survey schedules both covering the year 1939. The record group was found to contain many more large farms than the representative sample. Also it contained more farms of the cattle-feeding type, and fewer crop farms and dual-purpose cattle farms than the sample.

Even when the record farms were compared to random sample groups of farms of the same size and the same type there were many pronounced differences. The record farms commonly produced a gross income at least half again as great as the representative farms. Data from the random sample, however, were obtained by the survey method rather than by records; consequently they may be incomplete by some 12 or 14 percent. But this still leaves a wide difference between the two groups in production per 100 acres.

Net income per 100 acres also varied significantly between the two groups of farms, with the record farmers well above the representative sample. Thus an extra acre on the record farms added $9.18 to net income on the average while an extra acre on the random sample added only $5.01.