434 EXPEEIMENT STATION EECOED. 



protein 1.84 and 2.53, nitrogen-free extract 22.09 and 23.11, crude fiber 0.52 

 and 1.17, and asli 0.99 and 1.55 per cent, respectively. 



A rapidly growing- new forage plant, M. Denaiffe and J. Colle {Jour. 

 Agr. Prat., n. ser., 21 (1914), No. 10, pp. 300-302, fig. i).— This describes Loliiim 

 mxiltiflorum and gives results of cultural tests in which this grass yielded 4,880 

 kg. per hectare (4,343 lbs. per acre) in comparison with 3,200 kg. of Italian rye 

 grass and 3,200 kg. of English rye grass. 



Chemical analysis shows it to contain less cellulose and more ash than the 

 other two grasses. 



Oats for New York, H. H. Love {l^ew York Cornell 8ta. Bui. 343 {1914), PP- 

 S63-416, figs. 19). — This bulletin reports work in cooperation with the United 

 States Department of Agriculture with strains of hybrids, selections, and com- 

 mercial varieties of oats grown at Cornell in 1911-1913. Some methods and 

 apparatus used in the field and laboratory are described. Tables give yields in 

 ■bushels per acre, together with gain or loss over the check variety, yields of 

 straw, ratio of straw to grain, weight per bushel, percentage of meat, and 

 weight per 100 kernels. 



In regard to transmission of high yield it is noted " that many of the strains 

 under test are not constantly high-yielding. There are some strains, however, 

 which are always among the best in yield." 



Some correlations of characters of oats as developed in these tests are noted 

 as follows: Correlation coeflicients between a yield of straw and yield of grain 

 for the 3 years, 1911-1913, " are 0.357±0.082, 0.714±.03, and 0.5±0.043, showing 

 that there is a very good correlation between the two characters in question." 

 *' When all the varieties that have been grown in 1912 and 1913 are considered, 

 it is found that weight per bushel is correlated with yield per acre. The co- 

 efficients for the two years are 0.442±0.05, and 0.533±0.041. . . . The length 

 of kernel was correlated with weight per bushel for the two years 1912 and 1913, 

 and the correlation coefficients were found to be — 0.221±0.059 and — 0.5±0.043, 

 respectively. In other words, as the kernels become longer the weight per 

 bushel became less. The varieties having a long kernel, in general, have a 

 smaller weight per bushel than those having a shorter kernel. 



" The breadth of kernel was divided by the length in order to find the ratio 

 between the two characters. The kernels that were broader in comparison 

 to their length would have a higher ratio. This ratio was then correlated 

 with weight per bushel for the two years, and the correlation coefficients were 

 found to be 0.751 ±0.027 and 0.626±0.035, resi>ectively. thus indicating that 

 those varieties possessing kernels with greater breadth in comparison to their 

 length gave a higher weight per bushel." 



" The highest percentages of meat are obtained from those varieties having a 

 comparatively small kernel. Of 19 varieties having a percentage of meat of 

 more than 70, all but 4 have a kernel weight of less than 2.5 gm. per hundred ; 

 while of the 7 varieties whose percentage of meat is 65 or less. 4 have a kernel 

 weight of over 3 gm. This indicates that there is no relation between large 

 kernels and high percentage of meat in the sense that the two are found 

 together. 



"All the varieties studied for 1912 and 1913 were arranged in correlation 

 tables with respect to the two characters, weight per hundred kernels, and 

 percentage of meat, and there was found a negative correlation of —0.237 

 ±0.058 and — 0.188±0.055. These coefficients are not large, and when one con- 

 siders their probable errors they are not conclusive other than for the fact 

 that they are negative both years. This indicates that if any relation exists 

 between these characters it is in the direction that large seeds would tend 

 to have a thick hull and therefore a low percentage of meat. For the condi- 



