10 BULLETIN 883, U. S. DEPARTMENT OE AGRICULTURE. 
tional foot of space in the alleys was left to permit room for the binder 
bull wheel at harvest. 
The varieties were seeded in the same order in each replication. 
In 1916 all the varietal plats were in one continuous series, but in 
1914 and 1915 it was necessary to utilize parts of two or more of the 
series into which the station farm is divided. To obtain as repre- 
sentative conditions as possible the varieties were sown in reverse 
order when it was necessary to carry a replication over from one 
series to the adjoining one. For example, two of the replications and 
part of a third might be sown on one series, the varieties appearing in 
the same order in each replication, running from north to south. 
The remainder of the third and ail of the fourth and fifth, replications 
would then be sown on the adjoining series, beginning at the south 
end and running from south to north. In this arrangement, plats of 
the same variety approach each other at one point, as shown in the 
arrangement of the numbers below: 
1 2 3 4:5 
14 13 ' 12 11 10 
6 7 8 9 10 
9 8 7 6 5 
11 12 13 14 - 1 
4 3 2 I 14 
2 3 4 5 6 
13 ; 12 11 10 9 
7 
8 
Different experiments have required nursery rows of various 
lengths. In 1914 and 1915 the agronomic nursery was seeded with 
the ordinary grain drill in single rows 8 rods long, spaced 1 foot apart. 
This was found to be unsatisfactory, and in 1916 the nursery was 
seeded with a hand garden drill in rows 34 feet 4 inches long, spaced 
1 foot apart, and replicated thrice. This permitted a better distri- 
bution of each variety over the plat and a determination of the 
probable error of the experiment. More uniform seeding was ob- 
tained, and the harvesting and thrashing operations were facilitated. 
The number of varieties can always be determined and the series 
arranged so that all replications are evenly distributed. 
In 1914 head selections were grown in 5-foot rows, but in 1915 and 
1916 these and new introductions were grown in rows 17 feet 2 inches 
long. The yield in bushels per acre from a row of this length may be 
calculated readily by weighing the thrashed seed in grams and divid- 
ing by 10. The rows in the replicated nursery are twice this length, 
so that yields in bushels per acre may be figured by dividing the yield 
in grams per row by 5 or multiplying it by 0.2. 
Eows in the breeding nursery where hybrids are studied have either 
been 5 feet in length or have varied with the quantity of seed available. 
PREPARATION OF THE LAND. 
The experiments here reported were all conducted on the same 
general field, being a part of the table-land formed by the postglacial 
terrace already discussed. Part of this land was broken in the sum- 
mer of each of the successive years 1913, 1914, and 1915. This per- 
