S a 1 m o n , On Specializatioii of Parasitism in tlie Erysiphaceae. 285 



B. sterilis. If so, B. tectorum will, from the point of view of the 

 fiingus, serve as a bridge between the species of Serra- 

 falcus and Stenobromus. In other words, the species of. Bromus 

 belongiag to the section Stenobromus appear, as far as the experi- 

 ments have gone, to be safeguarded against the attacks of the 

 ,biologic forms' of Oidium occiirring on species of Bromus of the 

 section Serrafalcus^ except as regards the species B. tectorum in 

 Stenobromus, which lias remained susceptible, and wliere consequently 

 the defence breaks down. 



It is not necessary to point out in detail other probleras of 

 similar nature to those discussed in the hypothetical cases above, 

 as these will readily suggest themselves on glancing at Table 8. 

 Attention may however be drawn to the interesting fact of the infection 

 by Oidia derived from three different sources of both B. secalinus 

 and B. brizaeformis, so that each of these species of Bromus becomes 

 thus a meeting-place for three distinct ,biologic forms' of Oidium. 



A few experiments bearing on the question of the infection- 

 powers of a transferred Oidium were carried out. 



In two cases (Exper. nos. 51 and 70) the Oidium produced on 

 B. Jwrdeaceus by the sowing of the spores of B. interruptus in 

 Exper. no. 28 (see Table 1, 28 b) were sown on B. commutatus. 

 The germination of the conidia was good in both cases. In Exper. 

 no. 51 two leaves were inoculated on the under surface, and after 

 15 days (Sept. 3—18) no infection was visible. In Exper. no. 70 

 1 leaf was inoculated on both surfaces, and after 6 days (Sept. 

 12—18) no infection was visible. 



In the third case (Exper. no. 52) the Oidium produced on B. 

 commutatus by the sowing of the spores of B. hordeaceus in Exper. 

 no. 26 (Table 2, 26 a) were sown on B. mollis. Five leaves were 

 inoculated, some on the under and some on the upper surface. The 

 germination of the conidia was good. After 14 days (Sept. 3—17) 

 no infection was visible. 



In the fourth and fifth cases (Exper. nos. 63 and 69) the 

 Oidium produced on B. interruptus by the sowing of the spores of 

 h. hordeaceus m Exper. no. 40 (see Table 2, 40 a, b, c, d) were 

 sown on B. commutatus. In botli cases two leaves were inoculated 

 on both surfaces. The germination of the conidia was very good. 

 After 9 and 6 days respectively (Sept. 8 — 17 and Sept. 12 — 18) 

 no infection was visible. 



In the first two cases it would seem as though the characte- 

 ristics of the Oidium on B. interruptus persisted when the fungus 

 was transferred to B. hordeaceus^ since no infection of B. commutatus 

 resulted, whereas this species is nearly always infected by the 

 Oidium found growing in nature on B. hordeaceus. 



In the other cases, however, the Oidium on B. hordeaceus trans- 

 ferred to B. commutatus and B. interruptus showed the characte- 

 ristics of the Oidia found growing in nature on these plants, appa- 

 rently losing at once on these hosts the power of being able to 

 infect respectively B. mollis and B. commutatus, which the Oidium 

 found in nature on B. hordeaceus possesses. 



