S al m on , On specialization of Parsaitism in the Erysiphaceae. 269 



hj means of spores from infected plaiits had taken place as 

 described above. 



In Exper. no. 2 (Table 2), on the 15 th day of the experiment, a 

 powdery mass of spores was seen to faU from an infected leaf in 

 pot no. 2 a on to the leaf of a control plant; on the 21 st day of 

 the experiment nearly all the controls had become in this way 

 infected. 



In the open, where the wind coiüd blow the spores, this secon- 

 dary infection was very fi'equent. 



The results of the main series of experiments may now be 

 given. It must be pointed out however that it is impossible from 

 want of Space to describe here in detail the 55 inoculation experi- 

 ments carried out. I therefore summarize the important facts in 

 the following Tables and notes. 



The first material used for infection pui"poses was the Oidium 

 of E. Gramiuis on Bromus interruptus, — a species belonging to 

 the section Serrafalcus of the genus, and a plant closely allied to 

 B. mollis, to which species it is sometimes referred as a variety. 

 In the &st experiment (Exper. no. 1) the conidia of this Oidium 

 were sown on the following species of Bromus ; — B. secalinus, B. 

 mollis, B. commutatus, B. aroemis, belonging to the section Serra- 

 falcus; B. maximus and B. sterilis belonging to the section Steno- 

 hromus; and B, erectus and B. asper belonging to the section Festa- 

 coides. As in nearly all these experiments a number of plants in 

 two pots of each species of Bromus were inoculated. Five days 

 atter the conidia had been sown vigorous patches of mycelimn bearing 

 young conidiophores appeared on the inoculated leaves of B. mollis. 

 On the 7 th day all the inoculated leaves in each pot of B. mollis 

 were richly covered with patches of Oidium, composed of vigorously 

 growing mycelium and powdery masses of spores. No signs of in- 

 fection occurred, during the period of three weeks in which the 

 plants were kept under Observation on any of the other species of 

 Bromus, nor on the control plants in either of the pots of B. mollis. 



There seemed here, therefore, good reason to infer that the 

 Oidium on B. interruptus was a specialized form, — a ,biologic 

 form' — capable of infecting B. mollis, the plant most closely allied 

 to B. interruptus, but incapable of infecting the tlu-ee species named 

 above belonging to the same section as B. hiterruptus, as well as 

 foui' other species belonging to other sections of Bromus. 



The Oidium on B. interruptus was then sown on B. unioloides 

 (Sect. Ceratochloa) and on B. ciliatus (Sect. Festucoides). None of 

 the plants in the six pots of these two species Avhich were inocu- 

 lated showed any signs of infection (Exper. no. 6). 



In the next experiment (Exper. no. 8) the Oidium was sown 

 on B. mollis and on B. sterilis. The same result as in the first 

 experiment was obtained ; that is, füll infection occuiTed on B. mollis 

 while B. sterilis was untouched. 



The next experiment (Exper. no. 16) was a repetition of the 

 first, with the addition of B. racemosus to the list of plants inoculat- 

 ed, and gave precisely the same results as before; that is, füll 

 infection resulted on B. mollis, wliile all the other species were 

 passed over. 



