4 BULLETIN 529, U. S. DEPARTMENT OF AGRICULTURE. 
the seven farmers' estimates was 1,369 pounds, the difference being 
about one-half of 1 per cent of the gin record. 
The same investigator obtained the percentage of lint to seed 
cotton from gin records of 1,192 bales of cotton at Atkins, Ark., and 
from farmers' estimates on 151 bales in the same locality. According 
to the gin records, the average turnout of lint cotton was 32.5 per 
cent; the average of the farmers' estimates was 33.1 per cent. He 
made a similar investigation at Dermott, Ark., the gin results being 
31.75 per cent (on 907 bales) and the farmers' estimates being 31.2 
per cent (on 65 bales). 
Records kept on 15 farms in an Arkansas community on the 
amount of cotton picked per day per man gave an average of 110.4 
pounds. The average of the estimates of 50 farmers in the same 
locality was 140.3 pounds per day. 
On 23 plantations in Coahoma County, Miss., on which were 
9,326 acres of share croppers' cotton and 1,509 acres of share crop- 
pers' corn, the number of days of labor on these crops as shown by 
planters' estimates was 129,347. Each planter also estimated the 
number of days of " outside labor " performed, and this amounted 
to 14,018, or a grand total 143,365 days of labor for the share crop- 
pers and their families. On the same plantations each owner was 
asked to estimate the gross yearly value of the labor of the share 
croppers and their families, and the total for the 23 plantations was 
$144,007. This sum of money is equal to a daily wage of $1,004 
for each of the 143,365 days of labor performed, which is a close 
approximation to the current wage of $1 per day. 
From the above data it appears that in the case of important items 
of the farmer's business he has knowledge which is quite accurate. 
Matters of less importance are usually not kept in mind so ac- 
curately. In the matter of the amount of labor done in producing 
a crop, which involves a knowledge of an average day's work at 
plowing, harrowing, seeding, cultivating, etc., the farmer's knowledge 
is based on experience usually covering many years, and the 
answers he gives to such questions are averages rather than figures 
applicable to any one year. Because of differences in the preceding 
crop, amount of rainfall from year to year, variations in tempera- 
ture, etc., the work done on an acre of corn, for instance, may in any 
one year depart quite widely from the average. It is therefore im- 
possible to test adequately the accuracy of the farmer's estimates of 
items of this character by comparison with actual records for any 
one year. 
In order to show the variations that may occur between estimates 
based on many years' experience and accurate records for a single 
season, the following data relating to a group of 29 farms at Conway, 
Ark., are given. Each of these farmers was asked to estimate the 
