EXPERIMENTAL WORK WITH RAW ROCK PHOSPHATE, oe 
creased where it is applied in conjunction with commercial forms 
of nitrogen and potash. 
NUMBER AND DISTRIBUTION OF PLOTS, 
9 It has been pointed out by various investigators that a very much 
clearer and more accurate knowledge of the value of fertilizer treat- 
ments can be gained by employing on a limited area many small 
plots, repeating each fertilizer treatment as well as the checks on sev- 
eral well distributed plots, than on the same area where larger plots 
but fewer duplicates are employed. 
Piper and Stevenson? state that a long period of experimentation 
as well as the rephcation of plots tends to reduce probable error. 
Mercer and Hall? in summing up the results of an investigation on 
“Errors of Field Experiments,” state that in field trials the error 
diminishes with increasing size of the plot, but the reduction in error 
is small when the plot is above one-fortieth acre. These authors 
recommend that in any field experiment each unit of comparison 
(variety, method of fertilizing, etc., according to the object of the 
experiment) should be given five plots of one-fortieth acre each, sys- 
tematically distributed within the experimental area. Lyon® con- 
cludes that an area of one twenty-fifth acre of land, in four widely 
separated plots, devoted to any one test, secures a much greater de- 
gree of accuracy than the same area of land in one body. Olmstead * 
states that the replication of plots is a great satisfaction both to 
the experimenter and the readers of the literature, since it enables 
both to determine whether the results show any valid conclusions 
by giving them a means to estimate the precision of the work. 
Whitney ® states, “It is obvious that if the range of the yields 
of the different fertilizer plots among themselves is no greater than 
the range of the yields of the check plots among themselves, the 
relative effect of the different fertilizers would have no practical sig- 
nificance. Very rarely are there duplicates of any fertilizer treat- 
ment, and this is a weakness of the system. There are admittedly 
wide variations in the yield of check plots, and to determine the real 
yield under fertilization, it is Just as important to have eight or ten 
duplicate fertilizer plots as it is to have that number of duplicate 
check plots.” 
1 Standardization of Field Experimental Methods in Agronomy. Proc. Am. Soc. Agron., 
25-72 (4910). 
2 Experimental Error of Field Trials. Jour. Agr. Sci., 4, pt. 2, 107-127 (1911). 
* Experiments to Estimate Errors. Proc. Am. Soc. Agron., 3, 114 (1911). 
4Some Applications of the Method of Least Squares to Agricultural Experiments. Jour. 
Am. Soc. Agron., 6, 202 (1914). 
5 Unpublished work. 
