16 BULLETIN" 1334, U. S. DEPAETMEN'T OE AGEICULTURE 
when few varieties were grown. On the other hand, the weighted 
mean does not properly care for the differences of season. The 
weighted mean, however, is a correct mathematical statement, and it 
probably is as good a basis of comparison as any which can be used. 
At stations where the number of varieties tested has been large, 
they can be arranged in groups. At many places, for example, 
several varieties of the Hanna group and several of the Manchuria 
group have been grown. Where the results are not evident in the 
first compilation, a digest has been made by placing all varieties of 
the various types in groups and a comparison made of the resulting 
groups. This is not a very satisfactory method, in that the groups 
contain varying numbers of varieties, and these seldom can be com- 
pared for the same number of years. However, the adaptation of the 
groups is so different that the errors due to the difficulties of compila- 
tion usually are less than the variation in yield of the groups. The 
performance of barleys of the different types is much more easily 
determined in a group digest than in the tabulation of individual 
yields. The details oi this digest are not presented in tabular form, 
as the relationship can be clearly and concisely stated in the text 
itself. 
Most agronomists are interested in a few prominent well-known 
varieties. A second digest, therefore, has been made and the results 
reported in tabular form. In this second digest a well-known variety 
of each group has been taken and compared with well-known varieties 
from the other groups. In many cases the choice has been difficult. 
It is necessary for reasonable comparison that each pair of varieties 
shall have been in the experiment during several coincident years. 
This limits the choice of varieties. Those used have not always been 
the best of the group to which they belong. They are representative 
in most instances, and they have been chosen with a view both to 
their local importance and to their inclusion in experiments else- 
where. Usually varieties that have produced good yields over sev- 
eral years at any station have been sent to other stations. Occasion- 
ally a high-yielding variety has been grown at one or only a few sta- 
tions, but this is the exception. For the most part the digest is 
intended to cover the period of experimentation as well as can be 
done. The treatment necessarily varies as the data from the sta- 
tions vary. 
Later in this bulletin varieties which have shown promise in recent 
jesiTS are described. In this discussion and the table accompanying 
it the varieties most worthy of distribution in the light of recent tests 
are given. This is the only place in the bulletin where it is felt that 
definite recommendations as to specific varieties should be made. 
The earlier work forms an invaluable background for the later. 
The earlier results in a way are a check upon the behavior of types 
over a long period of time. If a selection of Hanna has yielded well 
for two or three years in an area where the Hanna group has been 
promising in only 6 of the last 20 years, these recent results are not so 
conclusive as if the Hanna group had been uniformly good. 
USE OF CEREAL INVESTIGATIONS NUMBERS 
^ In preparing the material for publication it was necessary to use a 
single nurnlxT to designate the same variety at the various stations. 
It is probable that State officials would prefer that the identity of 
