530 
Journal of Agricultural Research 
Vol. XXIII, No. 7 
between the characters here considered must be based upon a large 
amount of data secured from analyses of pure lines and made by the same 
laboratory, rather than any analyses of commercial varieties. In the 
light of this proposition the present paper is to be regarded as a pre¬ 
liminary study, as the amount of data from pure lines here included is 
very limited. 
Notwithstanding these limitations the present study is of some interest. 
It represents the first attempt to analyze statistically a large amount 
of raw data by computing from them correlation coefficients for a number 
of characters. It should throw some light upon the relationships of the 
chemical characters where formerly they were based merely upon 
observation or inspection of data. It should also be instructive to com¬ 
pare, in the future, the results obtained in the main from commercial 
varieties with those to be secured from pure lines of wheat. Finally, it 
may furnish suggestions of service in appraising the bread-making value 
of wheats. 
MATERIAL AND METHODS 
The material used in the present study is the data from chemical, 
milling, and baking tests published by the agricultural experiment 
stations of Colorado (7), Idaho (<£), Kansas (id), Maine (18), Minnesota 
(. 2 , j, 4 , 5), Montana (13), North Dakota (d, 9, jo, J5), Ohio (j, iy ) , 
Wisconsin (n), and the Central Experimental Farms of Canada at 
Ontario (12). While the choice of the material was naturally limited by 
the availability or lack of data for the different grain-growing legions, 
an inspection of the localities stated above will show that they include 
the important districts of the hard red spring and semihard red spring 
wheats, the hard red winter, and semihard red winter wheats. No data 
are included on the white soft wheats of the Pacific coast, the soft winter 
wheats of the Mississippi district, and the durum wheats. 
The classification of the material is first based upon the geographic 
distribution of the wheats and flours, the data of each experiment station 
being treated separately. The data of each experiment station are 
further subdivided according to the kind of wheat, the spring wheats 
and winter wheats being grouped separately. A further subdivision of 
the material was effected by grouping the spring wheats and winter 
wheats according to whether they are commercial varieties or pure strains. 
The coefficients of correlation computed in this study relate to 10 
important values secured from chemical analysis and milling and baking 
tests. Protein content in wheat is correlated with protein content in 
the flour, dry gluten, gliadin, and yield of flour. Loaf volume is corre¬ 
lated with protein content in wheat, protein content in the flour, dry 
gluten content, wet gluten content, quality of gluten, gliadin water 
absorption, and yield of flour. Dry gluten is correlated with gliadin 
and water absorption. As already noted, the whole material was ana¬ 
lyzed into four groups: Spring wheats, winter wheats, commercial 
varieties, and pure strains. Of the total number of 100 computed coeffi¬ 
cients of correlation, 30 relate to commercial varieties of spring wheat, 
27 to pure strains of spring wheat, 32 to commercial varieties of winter 
wheat, and 11 to pure strains of winter wheat. 
Since the data used in this paper have all been published by the 
experiment stations enumerated above, it is not necessary to publish 
the large number of correlation tables. However, for each kind of corre¬ 
lation, with a few exceptions, a representative correlation table is here 
