56 CEREAL RUSTS OF THE UNITED STATES. 



The results above given show that as a rule this rust form passes 

 readily from wheat to barley and from barley to wheat, the exceptions 

 not being more numerous than the faihires from wheat to wheat, and 

 such exceptions were probably generally caused by the presence of mil- 

 dew or some other unusual condition. Indeed, it usually seemed that 

 the transfers from wheat to barley and from barley to wheat, especially 

 the latter, resulted more easily than did transfers from wheat to wheat 

 or from barley to barley. As shown by the writer's investigations, it is 

 the only cereal rust that readily infects more than one of the five cereals. 

 In a few instances there were apparently slight infections of oats, but it 

 was evident, if they were not accidental from Uredo graminis of oats, 

 that the infections took place with difficulty. 



One interesting feature of the experiments is that although the bar- 

 berry rust in one set of experiments readily infected barley, it failed to 

 infect wheat or oats and produced but one spot on rye. If the bar- 

 berry rust was originally produced by the sporidia of teleutospores 

 from wheat, which is not known, the failure to infect oats would be in 

 accord with Eriksson^s experiments (20, pp. 304, 305; 29, pp. 200-202), 

 which showed that rust of barberry would infect only that cereal the 

 rust form of which originally produced the barberry rust. On the 

 other hand, the fact that it also failed to infect wheat, and infected rye 

 but slightly, . can not be thus exj)lained; and this, with a few other 

 peculiar results obtained, has led the writer to suspect that in this 

 country there may be two distinct forms of the black stem rust on 

 barley, one of which also infects rye but not wheat, and the other 

 wheat but not rye. Like barley, Hordenm juhatum seems to also act 

 as a host for both forms. Similarly peculiar results were obtained with 

 certain other wild grasses and need to be further invCvStigated. 



As previously stated, the leaf rust seemed to infect the true bread 

 wheats more readily than it did the durums and poulards, but in these 

 experiments the reverse was true, the stem rust infecting durums and 

 poulards more easily than it did the bread wheats. Moreover, larger 

 and darker sori seemed to be formed on the durums and x)oulards than 

 on the other wheats (PI. IV, fig. IS) — an interesting fact in connection 

 with the subject of rust resistance of varieties. 



For the inoculations from the wild grasses Hordtum juhatum, Agro- 

 pyron teneruin^ and Elymus canadensis glaucifoliuSj material was of 

 course obtained in the field on these hosts. Two rusts are found on A. 

 ienernm^ both of which seem to produce the same sort of sori on wheat, 

 but only one is believed to be P. gyam'mis. The spore forms were 

 quite different. 



Under ordinarj^ greenhouse conditions the i)eriod of incubation for 

 this rust in the uredo ranges from eight to twelve days. Under dates 

 of September 2G, 1896, November 6, 1897, and February 21, 1898, the 

 shorter periods noted include only the time until signs of rust appeared, 

 the epidermis not being broken until one or two days afterwards. 



From experiments and observations so far made, the following may 



