22 BULLETIN 1224, U. S. DEPARTMENT OF AGRICULTURE. 
deeds for each county to the secretary of state. A careful exami- 
nation of the data makes it evident that these reports are often 
erroneous, so that the land-value data for this State are not highly 
reliable. In Minnesota and Wisconsin, where returns were received 
on the questionnaire which was sent to the cash-renting landlords, 
from only a part of each State, the average value of land is based on 
those counties from which returns were received. 
The ratios of rent to value in Table 7 should not be looked upon as 
showing the true relationship between land income and land value 
because, as pointed out, the sample of farms on which cash-rent data 
were available from 1910 to 1920 may not be typical of the average 
grade of farms for the area they represent. They may be poorer or 
better than the average. But by comparing the yearly average 
rents based on different samples, it appears that any sample shows 
about the same increase in average rents from year to year. Any 
sample may be taken, therefore, as showing the extent to which cash 
rents and land values move together, or as showing the extent to 
which the ratio of rent to value varies from year to year. Since the 
samples on which the rents are based are small, they will not bring 
out this relationship with perfect accuracy. It will be noted, for 
instance, that the ratios of rent to value are more erratic for the areas 
in which the average rents are based on small samples than in the 
areas where the averages are based on large ones. 
Two points of importance are brought out in this table, showing 
the relative movements of cash rents and land values: (1) Land 
values have increased at a greater rate than land income in the last 
10 years; (2) the ratios of rent to value are remarkably consistent 
for any 3 or 4 consecutive years, considering the probable inaccuracies 
of the data. 
The fact that land values have been increasing faster than rents 
is a matter of considerable importance, and will come up for discussion 
later. The present problem is to find out how typical the 1920 ratios 
of rent to value are of other years. In other words, if the census had 
been taken 2 or 3 years earlier or later would the same result have been 
obtained ? 
The greatest fluctuations in the ratios of rent to value are found in 
Minnesota and Ohio. In Minnesota this ratio increased 1.3 points 
from 1910 to 1911. From 1911 to 1912-13 it fell 1.8 points. The 
high ratio in 191 1 was due to a drop in land values. Apart from 1911, 
the greatest difference in the ratios for any 3 or 4 consecutive years 
was 0.6 per cent. In Ohio the greatest difference between any 2 
years was 0.3 per cent, and for any 4 years 0.6 per cent. The varia- 
tions in the ratios in these two areas are probably not real, but are 
due on the one hand to the inadequate samples on which the average 
rents are based, and on the other to inaccuracies in the land value data. 
The actual variation in .the ratios was probably less than this. In 
Wisconsin the greatest variation for the whole period was 0.7 per cent. 
The variation from year to year in all but five cases was only 0.1 per 
cent. For any 3 or 4 consecutive years the greatest difference was 
0.3 per cent. In Iowa the greatest variation between any 2 years 
was 0.5 per cent, and this was from 1919 to 1920. From 1912 to 
1913 the ratio decreased 0.3 per cent, and from 1915 to 1916 it 
decreased again by 0.3 per cent. In all other pairs of years where 
there was any change it was only 0.1 per cent. 
