58 
net results, obtained while eliminating the effect of other factors, 
verify the conclusions shown by simple tabulation in Table 22. Even 
a large herd would not return a good income if the milk was not 
produced efficiently and economically; with efficient production, the 
larger the herd the larger the earnings. 
Taking into account the difference in the relation of dairy index to 
operators earnings for small and for large farms raised the multiple 
correlation slightly, but the increase was not significant enough to 
indicate that the exact relation could be measured very closely from 
the available data. For that reason the regression surface, from 
which the values quoted in the preceding paragraph were read, was 
not considered of sufficient accuracy to publish. 
The ''coefficients of determination" for the crop and pasture acre- 
age in this last solution confirm the conclusion, previously illustrated 
by simple tabulation, that difference in the size of farms outside of 
the size of the herd had very little effect upon farmer's earnings the 
year of the survey. This checks the conclusion shown by the result 
as to the proportion of the dairy feed purchased and the proportion 
raised: For the year of the survey it made little difference whether 
the feed was raised or purchased; it cost the farmer the same. But 
it should be remembered that this was by no means a typical year, 
and that with more usual conditions the crop enterprises would 
undoubtedly add something to the farmer's earnings. 
The result as to crop index shows, as is invariably the case, that 
the farmers with higher yields tended to receive the larger profits. 
This was more important than any other factor except the size and 
efficiency of the dairy business. To the extent that these higher 
yields are owing to natural superior fertility of the soil not offset by 
a higher charge for the use of land, this indicates not better ability 
on the part of the operator but merely that he benefits from using 
superior resources. 
As has already been shown, farms with a crop index of 110 were 
worth about $10 an acre more than farms with a crop index of 90, 
other things being equal. For a typical farm of 100 acres that would 
mean a difference in value of about $1,000. As shown by the net 
regression for this group of dairy farms, an increase of 20 points in 
crop index would increase earnings about $180, on the average. From 
this it would appear either that (1) a large part of the differences in 
yields were owing to the skill of the individual farmers and were not 
reflected in land values, or (2) part of the observed variation in 
yields was owing to conditions of the individual year and were not 
representative of the individual farms, or (3) the appraisal of indi- 
vidual farms and the competition for farms is not keen enough in 
this area to reflect in the price of the farm all of the differences in 
fertility. This latter point might also be explained in terms of the 
importance of the differences in the ability of men as well as of land 
in determining the rents to be paid (18, pp. 194-198). Each of 
the alternatives is probably true in some measure. At any event, it 
apparently pays to give some attention to procurimr good yields in 
Chester County: but, at least on dairy farms, the room for improve- 
ment in this direction is nothing like so great as the room for improve- 
ment in selecting, breeding, and feeding the dairy herd, variations in 
the efficiency of milk production being over five times as important 
as a determinant of farm profits as variations in the yields of crops. 
