Salmon, On Specialization of Parasitism in the Erysiphaceae. 289 



An Oidium occurring" on A. nuda was then sown on A. nuda^ 

 A sattva, and A. brems. Füll infection resulted on eacli plant. 

 (see Table 12). 



Finally, an Oidium on A. orientalis was sown on A. sativa 

 (Exper. no. 49). Here also füll infection was obtained. 



We see, therefore, tliat in the genus Aoena the rule appears 

 to be that the Oidium occurring on the different species is specia- 

 lized only to the extent that it cannot — so far as it has been 

 tried — infect species belonging to other genera of grasses; but 

 that within the genus Avena the Oidium on any species is capable 

 of infecting all the other species. Through tliis latter character- 

 istic, then, the Oidium on Avena differs markedly from the Oidia 

 found on Bromus. 



The above results, while generally confirming those given by 

 Marchai (6), differ in one point. Marchai separates as ,formes 

 specialisees' the Oidia on Wheat, Barley, ßye, and Oats, and 

 also on Agropyron. The Oidium on Oats is described as follows: 

 ,£". Graminis f. spec. Acenae, sur Avena sativa, orientalis, fatua et 

 sur Arrhenatherum elafius.^ In the Single experiment, however, in 

 which I infected Arrhenatherum elatius with spores from the Oidium 

 on Avena sativa, no infection occurred. Tliis was also the case 

 when the same Oidium was sown on Trisetum pratense. The fact 

 is of interest because these two species were at one time, under the 

 names Avena elaiior and A. Jiavescens , included in the genus 

 Avena. 



It may be noted here that an Oidium was observed in the 

 Cambridge Botanic Gardens on Avena brevis, A. strigosa, and 

 A. sterilis. 



In the next experiment (Exper. no. 31) an Oidium that was 

 found growing" on Festuca elaiior var. pratensis was sown on seed- 

 lings of the same species and on Avena sativa. Infection (althoug-h 

 slight) resulted on the former plant, but not on the latter. 



Next, an Oidium growing on Lolium perenne var. italicum 

 was sown on seedlings of its host-species, of L. perenne, and of 

 Avena sativa (Exper. no. 48). Infection resulted in both cases on 

 the Lolium, but not on the Avena. In the Cambridge Botanic 

 Gardens an Oidimn occurred on Lolium perenne, and vars. multi- 

 ftorum, and italicum, L. temulentum, and L. linic.ola. 



This concluded the series of experiments in which the Oidium 

 of E. Graminis was used. 



The conidial form of E. Polygoni DC. growing on Trifolium 

 pratense and Pisum sativum was then used in a short series of 

 experiments. 



The Oidium on Trifolium pratense was first sown (Exper. no. 

 10) on four other species of the genus, viz. on a fuUy grown plant of 

 T. agrarium, T. repens, T. medium and T. montanum, and on two 

 pots o± seedlings of T. pratense. Infection occurred on T. pratense 

 alone. The same Oidium was then (Exper. no. 13) sown on two 

 pots of seedlings of T. pratense, and on mature examples of three 

 other Leguminous plants, viz. Lotus comiculatus, Melilotus arvensis, 

 and Medicago sativa, as well as on Trifolium repens. Again the 

 Oidium infected only T. pratense, failing completely as before to 



