1891 
THE RURAL NEW-YORKER'. 
479 
that as between a very wet season and a very dry one, I 
prefer the latter even though my land is so well drained. 
In a wet season the land is kept cool, and even if drained, 
it is more or less both cool and damp, and the temperature 
I have as yet no means of controlling, except by the plan 
above described, and the results of this are not sufficiently 
definite to be fairly estimated. My general practice, in a 
very few words, is to prepare the land in the best possible 
manner for the seeds, plowing and cultivating deep before 
planting, but, as an almost invariable rule, my cultivation 
is afterwards shallow, but so frequent and thorough that 
the land need not be touched when it is in a wet and sticky 
condition. 
Brown County, Wis. 
THE SIZE OF EXPERIMENT PLOTS. 
LARGE VS. SMALL PLOTS FOR ACCURACY. 
American Experimenters Talk. 
On page 388 The R. N.-Y. di-cussed Prof. C. S. Plumb’s 
remarks on the value of small plot experiments. Prof. 
Plumb stated that “ among station investigators of repu¬ 
tation the plot system is considered an enigma that yields 
more dissatisfaction than profit.” The question has been 
submitted to the directors of our experiment stations with 
the followiDg results: 
From The California Station. 
crop of the various plots. The water content of the crop 
from each plot should be determined by chemical control, 
wherever it seems possible. The writer has observed dif¬ 
ferences of seven to nine per cent in the moisture of hay 
from different plots of a grass experiment with fertilizers, 
even when the crops from all of the plots were harvested 
and weighed in one afternoon. Similar differences of 15 to 
20 per cent have been seen in the case of stover. If the 
field weighings alone are depended upon very erroneous 
conclusions are often the result. 
Another obstacle that lies in the way of field experi¬ 
ments doing the greatest good, i3 the fact that farmers 
and too often experiment station workers are over-anxious 
for immediate results. The slow, painstaking, but indus¬ 
trious plan of our German friends might be a good pattern 
for many of our younger investigators of agricultural 
problems, to reflect upon. The succes of the field work at 
Rothamsted is another illustration of what may be accom¬ 
plished by painstaking and loog-continued work. Sir 
John B Lawes was once quoted as having said: ‘‘We 
should experiment and continue to experiment under 
similar conditions, until we can predict the results.” 
Storr’s Agr’l. Exp. Station, Conn. C s. PHELPS. 
Large versus Small Plots. 
In the class of questions to which answers are sought in 
plot experiments, two important sources of error must be 
eliminated if the results are to be considered trust wor- 
not because their work was not conscientiously done, but 
because their plots were too small to eliminate the varia¬ 
tions of individual plants. It is often very delusive, for 
example, to report the comparative earliness of several 
varieties of tomatoes from five or six plants of each vari¬ 
ety, or the yields of numerous varieties of potatoes from 
five to ten hills of each. Mistakes of this kind may be par¬ 
donable in beginners, but the time is at hand when they 
should no longer be regarded with lenity. E. s. GOFF. 
University of Wisconsin. 
Small Plot Experiments Oft Repeated. 
The whole matter in a nutshell is this, that In nine- 
tenths of the plot experimentations the person in charge 
endeavors to prove too much by a single experiment, and 
as The Rural says, ‘‘ What is needed is a greater number 
of repetitions of the same trials (experiments).” The ele¬ 
ment which causes the chief trouble is that of soil varia¬ 
tion, and this cannot be eliminated. By certain careful 
arrangement of plots only can the errors arising be re¬ 
duced to a minimum. If an acre be divided into two 
equal parts, and potatoes or corn be treated with a fertil¬ 
izer on one half, and the other half be left free, we have 
one test, with the factor of soil variation undisturbed. 
When the year Is closed and the result is to be Interpreted, 
suppose the fertilizer has given a magnificent return appar¬ 
ently, and that is announced and interpreted as “ increase 
caused by the fertilizer.” This increase may be due to the 
I entirely agree with The R N.-Y. that a given area thy, viz., the variability of individual plants and the natural fertility of the soil on that plot, and the poverty 
to be devoted to a culture experiment is better utllizfd by variability of the soil upon which the plots are located. of its neighbor; or, suppose the reverse be true, then a 
really meritorious fertilizer gets a blow which it does not 
deserve ; hence I believe in small-plot experiments, repeat¬ 
ing in the same manner, on the same field, always under 
the same conditions, and not alone for one year, but for 
two or three, so as to eliminate any peculiarity of condi¬ 
tions arising from extremes 
of climatic conditions, e. g % , 
a case where an excessively 
dry season would nullify 
all or nearly all of the effects 
of a good fertilizer, or a very 
wet season produce an oppo¬ 
site effect. The average re¬ 
sults of three years ought to 
give good indications at least, 
if not positive results. I have 
often been astonished at the 
interpretation which many 
persons give to certain ex- 
' periments; results that could 
be obtained only by the most 
ImoPIIkS Ifimll fallacious and specious 
Hull reasoning. Put me down for 
accurate small-plot experi- 
HjUll ments often repeated, as a 
result of my personal observa- 
tion and study of this ques- 
llliP ’wif The Mississippi Station’s 
■ ... .. n - I fully agree with The R. 
N.-Y. in regard to the advisa- 
bility of using small plots 
with a number of duplicates 
rather than an attempt to 
draw conclusions from single 
If all plants of the same variety, or of the same se¬ 
lection of seed, under similar conditions, grew exactly 
alike, a single plant would be the ideal unit for com¬ 
parison. But since this is not the case, it is necessary to 
take a sufficient number of plants so that their variations 
dividing it into a number of small plots than if occupied 
by only two, or a relatively small number of large ones In 
fact, unless we were justified in considering the land so 
used as absolutely uniform, the preference to be given to 
the former plan over the latter is settled by the elementary 
principles of the doctrine of 
probabilities. An objection f £ f l, ; 
to numerous small plots arises ' ' . if ’.•; ; j £ 
when conclusions as to the p: | 
probable absolute product |;p(| | 
per acre on the large scale are Mtj&j 
intended to be drawn: for # 
then the fact that a very — .. r . 
large proportion of the crop r. ‘ Jap -; 
has the advantage of grow- v '. M M ap 
ing on the ‘‘outside row” a' ial pi 
vitiates the numerical result. ; •.l 
But when it is intended 'L,.■ 
simply to compare the pro- 
duction of a crop treated in Uj:- 
a certain way with that of j 
check plots not so treated, t '‘b’.S’Sv 
the addition of the same ad- 
vantage to each side of the ^ ’Ipdr* 
equation leaves the latter a 
true one and the larger num- .<‘•$111 
ber of tests promise a closer 
approach to the practical ^T 
outcome. e. w. hilgard. 
California Station. „ 
Plot and Soli Tests Com- ’’ *•’*' 
blned. " " ALT*' 
Plot experiments should ,'i- 
by no means be discouraged. * ‘ ^ 
Many experiment station in- 
vestigators of reputation are , * ' 
not yet prepared to believe ( _3 
that the plot system “yields large areas. If we could se¬ 
niors dissatisfaction than MOTIVE POWER ON A VERMONT FARM. Fig. 180. cure a perfectly uniform soil, 
profit.” In all lines of invest!- - . , the single, large plot system 
gation much is indicated, while little is positively proved. 
This is especially true where the indirect methods of investi¬ 
gation are adopted. The so called “soil tests ” conducted 
during the past 10 years, under the direction of Dr. At¬ 
water, of Connecticut, Dr. Neale, of New Jersey, and 
Prof. Scovil, of Kentucky, indicate much regarding- the 
deficiencies and needs of particular soils. It is quite evi¬ 
dent that nothing but “ plot experimentation ” would have 
brought out these results. 
I believe there is a “ happy medium ” between the very 
small and the very large plots. In a test of varieties, 
methods of cutting seed, etc., where the treatment of the 
soil is uniform, very small plots are much less objection¬ 
able. In work of this kind, the best plan may perhaps be 
to plant two or three rows of each variety, and then have 
two or three duplicates of each test distributed across the 
field. Where the value of fertilizers and the needs of soils 
are to be tested, large plots, with unmanured strips be¬ 
tween, are to be preferred. There are two great obstacles 
to be met in undertaking field experiments. The first is 
the difficulty in finding a soil that is fairly uniform both 
in mechanical texture and natural fertility. The second 
is the difficulty of giving to each plot, and the crop there¬ 
on, uniform treatment throughout the season, wherever 
uniformity is desired. It is not so easy to find a field of 
two or three acres that is fairly uniform, as it is one of an 
acre. One plot one-tenth of an acre in size for each test, 
with several unfertilized spots for comparison, is generally 
better than larger plots covering several acres. Soil tests 
can thus be made on about IX acre of land. If the plots 
are made long and narrow, the difficulty resulting from 
lack of uniformity in the soil will, in a great measure, be 
overcome. Where plot experiments are among the farmers, 
the work should be under the control of the stations, as far 
as possible. The laying out of plots, the application of fer¬ 
tilizers, and the weighing and harvesting should be 
closely controlled by experienced workers. Experiments 
of this class are not only a profit to the farmer upon whose 
land they may be carried out, but are also a means of edu¬ 
cation to his neighbors and to the surrounding com¬ 
munity. 
One of the sources of “liability to wrong conclusions” 
rests in the differences found in the water content of the 
will balance each other, in order that the average of the 
whole number may represent a truly normal plant. This 
point theoretically defines the size of the plot. A smaller 
number of plants is untrustworthy; a larger number is 
not only superfluous, but it renders more difficult of eli¬ 
mination the other source of error—the variability of 
the soil. As a rule, the greater the area occupied by an 
experiment the greater the range of soil variability. 
In like manner, the number of duplicate plots should be 
sufficiently large so that the average of the soil conditions 
of each set of plots under comparison shall exactly equal 
the average of those of every other set. 
But while It is thus very easy to theoretically designate 
the proper size and number of plots for an Ideal experi¬ 
ment, this theory, when applied to practice, involves so 
many unknown quantities that it becomes well nigh use¬ 
less. We cannot know the prospective limits of variation 
that lie hidden within the mystery we call a seed and the 
freaks of the soil of experiment plots certainly surpass our 
powers of anticipation or explanation. What, then, can 
we do ? It is safer to be guided by the best light we have 
than to go hap-hazird. A few principles may be laid 
down that will be of some assistance. 
1. With a given area, a large number of small plots is 
preferable to a small number of large plots, because the 
number of plants grown will not differ much in the two 
cases, and if one plot does not furnish sufficient plants so 
that the average shall represent a truly normal specimen, 
the errors of this kind in a large number of plots will bal¬ 
ance each other; while numerous small plots are much 
better calculated to furnish an average of the soil con¬ 
ditions of a given area than a few large ones. 
2. The greater the fixity of a variety and the greater the 
purity of the seed, the smaller may be the plots. As a rule, 
varieties propagated by division may be tested upon 
smaller plots than those propagated by seed. Plants com¬ 
paratively little changed by culture, as the parsnip and 
salsify, may be tested on smaller plots than those very 
highly developed, like the cabbage and cauliflower. 
Too much stress can hardly be laid upon the danger of 
error in reporting the yield of numerous varieties from a 
single small plot of each. Many of the reports of our ex¬ 
periment stations have been misleading in this respect. 
would be preferred on account of its greater simplicity and 
its less liability to error in measurements and weighings; * 
but such uniformity in soil being out of the question for 
most of us, I can see no other way than to use small plots 
in duplicate. A given field may be a fair type of thousands 
o’ farms in the same State, and still vary widely in its 
different parts, especially on the alluvial soils In the South¬ 
ern States. 
In Mississippi the work of the station and that of the 
agricultural college are entirely separate from each other, 
though they are parallel as much as possible. The work 
of the station Is almost wholly confined to plots contain¬ 
ing one-tenth of an acre each, and each plot Is duplicated 
several times in different parts of the field, as it is only in 
this way that we can balance the differences in the soil 
and obtain a fair average for the field. By taking such 
samples and arranging them we believe that we can ac¬ 
complish more work for the State at large and make It 
fully as accurate as if we were to use 10-acre lots for each 
experiment. When decisive indications have been secured 
in this way, the college uses some one or more of the most 
promising station methods upon fields containing several 
acres, with results which usually agree very closely with 
those obtained by the station. s. w. TRACY. 
Mississippi Station. 
How Small May Plots Be? 
The term “ small plot” is very indefinite. I do not re¬ 
gard plot experiments as useless, but on the other hand, I 
consider them very valuable. The Rural’s idea of small 
plots is, as it seems to me, an extreme one. I think one- 
twentieth of an acre as small as I should want to use, 
though half this amount might be all right. I know it 
has become popular to assail many existing .methods of 
Investigation; but in too many cases this comes from the 
acknowleged fact that field work cannot be conducted 
with the precision that laboratory work attains to, but to 
the practical man the field experiment stands in the same 
relation that the laboratory experiment does to the scien¬ 
tist. Plot experiments will not go out of the scheme of 
agricultural investigation, because there is merit in the 
system and valuable results have been and will be obtained 
from it. G. h. whitcher. 
New Hampshire Experiment Station. 
