4 BULLETIN 881, IT. S. DEPARTMENT OF AGRICULTURE. 
rotations 40 and 44, the alfalfa period covers two years and in rota- 
tions 60 and 61 it covers three years. In the above-enumerated 
rotations the 1913 potato crops followed only one year of alfalfa. 
In 1914 they followed two years of alfalfa, which is the normal course 
for the first two rotations, 40 and 44, while for rotations 60 and 61 
the normal course of three years of alfalfa was first reached for the 
crop of 1915. 
Table I shows the yields of potatoes for the four pairs of rotations 
at each of the three stations. These yields, given in bushels per acre, 
are for the total crop, including small and diseased potatoes. The 
annual difference in yield between the crops following alfalfa and 
the check plats are also shown in the table, together with the mean 
annual yield of the crop for each rotation and the mean of the 
annual differences in yield. With each of the means the probable 
error is given. 1 
The results given in Table I show that in nine cases out of twelve 
the mean of the annual differences in yield was in favor of the crops 
following alfalfa. But in only six of the nine cases is this mean 
difference significant; that is, more than three times the probable 
error. In the other six cases the mean is less than three times the 
probable error and is not regarded as significant. 
It has been customary in connection with these experiments to 
sort the potatoes in the field at the time of digging. This sorting 
has been done by means of a wire screen with 2-inch meshes. The 
smaller potatoes that pass through this screen are classed as un- 
marketable, while those passing over the screen are marketable. 
It has been observed that at Scottsbluff the percentage of marketable 
potatoes has been significantly larger from the plats following alfalfa 
than from the check plats. This has not been true, however, at 
Belle Fourche or at Huntley. 
It is clear from the facts set forth in Table- 1 that the beneficial 
effect of alfalfa on subsequent yields of potatoes has been much more 
marked and consistent at Scottsbluff than at the other two stations. 
This may be due to the fact that the soil at Scottsbluff is a light sandy 
loam, while at both the other stations the soil is a heavy clay loam. 
The essential facts concerning these yields for each of the three sta- 
tions are summarized in the following paragraphs. 
At Scottsbluff the yield of potatoes, large and small, for the 48 
plat years has averaged 236 bushels per acre, with a mean annual 
difference resulting from alfalfa of 100 ± 7. The yield of marketable 
1 The probable error of the mean as used in these tables is obtained by Merriman's formula 36, which is 
stated as follows: r o p=. ' In other words, the probable error of the mean is obtained by multiply- 
n-yjn— 1 
ing the sum of the departure fr:>m the mean by the quotient of n^n—l into 0.4853, where n equals the 
number of yields involved. (Merriman, Mansfield. Method of Least Squ'ares, ed. 8, p. 223. New York, 
London, 1913.) 
