VALIDITY OF THE SURVEY METHOD OF RESEARCH. 
13 
nearly the same as on the corresponding full pages. Thus, in the 
case of the drill press the third page contains only 44 replies, but the 
percentage of owners among them is nearly the same as on the pre- 
ceding full page. Even here the number of replies is sufficient to 
permit the law of chance to make itself evident. 
Table IX. — Illustrating operation of law of chance. 
Implement. 
Carborundum or emery wheel. 
Set of stocks and dies 
Gasoline blowtorch . 
Wrecking tool 
Pinch bar 
Hack saw 
Small hoist 
Drill press, or breast drill. 
Combination vise, drill, and anvil . 
Combination pliers 
Expansion bit 
Combination bevel square 
Heary sbears, or tinner's snips 
No. 
Number 
of farms 
having. 
488 
369 
367 
5 
205 
105 
537 
532 
133 
612 
598 
87 
478 
450 
38 
520 
519 
148 
454 
420 
26 
370 
345 
638 
634 
190 
349 
296 
375 
274 
448 
425 
53 
Percent- 
age of 
owners 
on two 
pages. 
50.00 
50.00 
50.14 
49.86 
50.3 
49.7 
50.5 
49.5 
51.5 
48.5 
50.1 
49.9 
52.0 
48.0 
50.15 
49.85 
51.32 
48. 68 
Number 
of farms 
recom- 
mending. 
700 
700 
700 
700 
6 
700 
426 
700 
700 
180 
700 
700 
104 
700 
700 
53 
700 
700 
220 
700 
700 
44 
700 
654 
700 
700 
216 
700 
56S 
700 
494 
700 
700 
86 
Per cent 
of those 
recom- 
mending 
having. 
69.71 
69.71 
52.71 
52.43 
24.64 
76.71 
76.00 
73.88 
87.43 
85.43 
83. 65 
68.28 
64.28 
71.69 
74.28 
74.14 
67.27 
64.85 
60.00 
59.09 
52.85 
52.75 
91.14 
90.57 
87.96 
49.86 
52.11 
53.57 
55.46 
64.00 
60.71 
61.62 
The fact that in each case it is page 1 that has the highest propor- 
tion of owners is of no significance, since the full pages were deliber- 
ately arranged in this order after the tabulation was completed, the 
original chance arrangement being thus lost. 
MISTAKEN NOTIONS OF ACCURACY. 
In the endeavor to find the average value of a variable quantity, 
such as annual rainfall, the yield per acre of a crop under given con- 
ditions, etc., there is such a thing as gross inaccuracy in the final re- 
sult even where the individual measures are made with a high degree 
of precision. Suppose, for instance, it is desired to ascertain the 
average yield of winter wheat after summer fallow as compared with 
the yield after a preceding crop of small grain, under the soil and 
climatic conditions prevailing on a particular tract of uniform soil. 
