48 
BULLETIN 1400, U. S. DEPARTMENT OF AGRICULTURE 
in this first analysis, and then shows the distribution of the same 
items for the 368 farms remaining after excluding the farms growing 
mushrooms from the tabulation, illustrates the difficulty. 
Table 40. — Frequency distributions of farms according to the values per acre of 
several items, both including and excluding mushroom farms 
Value in dollars per acre 
All farms classified by value per 
acre of— 
All farms except mushroom farms 
classified by value per acre of— 
Land 
I 
Dwelling 
A 
"Other" 
buildings 
C 
Land Dwelling 
I A 
"Other" 
buildings 
C 
to 50 
195 
173 
33 
11 
3 
1 
2 
1 
1 
391 
12 
5 
183 
151 
25 
7 
1 
367 
50 to 100 
118 
190 
62 
27 
6 
5 
1 
3 
1 
5 
113 
171 
55 
21 
5 
3 
1 
100 to 150 
150 to 200 
200 to 250 
3 
3 
1 
3 
1 
250 to 300 
300 to 350 - 
1 
350 to 400 
400 to 450 
450 to 500 
500 to 750 
1 
750 to 1,000 
1,000 to 1,500. 
1 
1 
1 
1 
1 
1 
1 
1 
1,500 to 2,000 
2,000 to 2,500 - - - 
2,500 to 3,000--- 
Table 40 shows how much the inclusion of the mushroom farms 
distorted the picture. Although the highest value per acre reported on 
any of the other farms was under $350 per acre, some of the mushroom 
farms ran over $2,000 per acre. And even more striking, although 
only one of the other farms showed values for buildings other than 
dwelling or dairy running above $50 per acre, on several of the 
mushroom farms values up to $2,500 per acre were reported. For 
this reason it was decided to omit the farms reporting mushrooms 
from the balance of the tabulation and analyze the factors responsi- 
ble for land values separately for the other 368 farms. 
Estimating value of the farm per acre (I) for these 368 farms on 
the basis of the net regressions shown on page 47, the correlation 
between the estimated and observed values was found to be but 
i? = 0.64. This showed conclusively that the relation of these eight 
factors to farm value, as determined by the first correlation, was a 
very poor statement of the actual relations when the farms with 
mushroom houses were left out of consideration. 
The analysis for these 368 farms was carried through to show not 
merely the average relation of the value of the dwelling to the value 
of the farm but also the contribution of the dwelling value to the 
farm value for dwellings of various values on farms of various sizes, 
as it was found that the more expensive the dwelling the smaller 
the proportion of its value which was added to the farm value, 'and 
that a small dwelling did not add as much to the total value of a 
large farm as of a small one. This analysis involved the use of 
multiple curvilinear correlation and "solid" multiple curvilinear 
correlation, which is too complicated to present here in detail (5, 8). 
Only the final conclusions reached will be presented. 
A multiple correlation of P = 0.772 for the relation between the 
eight factors and farm values per acre was obtained. This means 
that only about 60 per cent of the causes of variation in land value 
