26o Journal of Agricultural Research voLxix.no. 6 



In order to learn the percentage of infection and also to be absolutely 

 certain that all seed used in experimental work was infected with rust, 

 the following method of selecting rusted kernels was employed. The 

 samples of wheat to be examined were spread out in a shallow dish where 

 the light was good, and the discolored kernels were taken out one by one 

 by means of small forceps. A common 5-inch reading glass usually was 

 used to facilitate making the selections. These discolored kernels were 

 then placed one by one under a low-power binocular microscope where 

 it could be easily determined at a glance if any rust sori occurred on 

 their germ ends. As will be shown later, some infected kernels may 

 have been missed, for sometimes the sori on the germ ends are broken 

 off with the flowering glumes in thrashing. 



Bolley and Pritchard (5, p. 646) state that — 



in some samples of wheat in the rust-infected crop of 1904 as high as 30 per cent of 

 all grains harvested showed such rust infection. 



Pritchard {22, p. 153) also states that in 1910, a rust-free year, wheat 

 from elevators at Brookings, S. Dak., showed some rusted kernels in 

 every sample and many in some varieties, especially Bluestem. The 

 writer's observations do not agree with this. In all the hundreds of 

 samples examined the largest percentage of kernels found in any one 

 sample showing rust sori was only about i per cent of the total. Many 

 samples were examined in which no infected kernels could be found. 

 In fact, even in 1916, a very bad rust year, the varieties having kernel 

 infection were the exception rather than the rule. Moreover, varieties 

 of the durum wheat were the ones which most often were found to be 

 infected. This was the case in both years and seems to be consist- 

 ently so. One sample of mixed wheat from Reeder, N. Dak., collected 

 in 1 91 6, contained about i per cent of infected kernels. Although the 

 sample contained Marquis, durum, and Bluestem in the mixture, only 

 durum kernels were found infected. This has been found to be the case 

 in many mixed samples examined. Only in a few cases have any number 

 of infected kernels of other varieties been found. This may be due to 

 the fact that the spike of durum wheat is so compact that it dries very 

 slowly after rains or heavy dews and these moist conditions favor infec- 

 tion by rust. It is a well-known fact that durum varieties are very 

 susceptible to Fusarium scab, possibly for the same reason. 



To illustrate this point the following observ^ation is of interest. The 

 writer noted in 1916 at Dickinson, N. Dak., that all the durum wheats 

 were more or less badly rusted on the heads. (See PI. 38.) This was 

 especially true of the Kubanka strain known as selection No. 8, C. I. 

 No. 4063 (PI. 38). A large number of heads of this variety were collected 

 which were Hterally covered with stemrust sori (PI. 39, A). Mr. Ralph 

 Smith, of the Office of Cereal Investigations, stationed at Dickinson, 

 kindly furnished the writer some of the seed of this variety when the 

 plots were thrashed. This seed was all examined carefully, and it was 



