S al m o n , On Specialization of Parasitism in the Erysiphaceae. 275 



experiment (no. 18) the Oidium behaved similarly to that on B. 

 interruptus iii infecting fiilly B. mollis, and in passing over B. race- 

 mosus, B. arvensis, B. sterilis B. erectus and B. maximus ; but again 

 showed itself to be different by infecting- B. commutatus. In the present 

 experiment, however, the plants in one of the two pots of B. commu- 

 tatus used were fully infected, whilst for some reason those in the 

 other remained free. 



Another point of interest in the present experiment was the 

 ,siibinfection' of B. secalinus. Of the two pots of this species used 

 3 (of tlie 5 inoculated) leaves in one pot bore, on the 5 th day after 

 inoculation, a few patches of mycelium — in one case a large and 

 well-developed patch — but no conidiophores ; in the second pot no 

 infection was then visible. On tlie next day thei-e were a few 

 scattered conidiophores on the 4 infected leaves in the lirst pot, 

 but the fungus was scarcely increasing; the second pot appeared 

 still uninfected, On the 8 th day 5 leaves in the flrst pot (i. e. all 

 that had been inoculated) bore a few patches of mycelium with a 

 few scattered conidiophores; the fung-us, however, was evidently not 

 increasing, and in fact, seemed only just able to maintain its hold; 

 in the second pot 1 leaf (of the 4 inoculated) now showed a minute 

 fleck of mycelium. On the 10 th day the fungus began to disappear 

 from the plants in the first pot. On the 12 th day the second pot 

 showed further sig-ns of faint infection by producing on a second 

 leaf a few scattered conidiophores. On the 16 th day the fungus had 

 nearly died away on the plants in both pots. As the fungus never 

 produced powdery masses of spores, it seems advisable to consider 

 the present case as one of ,subinfection' only, similar to that pro- 

 duced when the Oidium on B. interruptus was sown on B. brizae- 

 formis and B. celutinus. 



In the next experiment (no. 22) the Oidium on B. hordeaceus 

 was again fomid to infect fully B. moUis. 



The Oidium was then sown on B. commutatus again and also 

 on three fiesh species, B. hrizaeformis, B. macrostachys and B. 

 interruptus. In this experiment (no. 23) füll infection ag"ain resulted 

 on B. commutatus] otherwise an agreement with the Oidium on B. 

 interruptus was shown by the füll infection of B. interruptus, the 

 ,subiiifection' of B. brizaeformis, and the non-infection of B. macro- 

 stachys. With regard to B. brizaeformis the fungus behaved as 

 follows. Oü the 9 th day after inoculation, all the inoculated leaves 

 in one pot, and 3 (of the 5 inoculated) in the other, bore minute 

 scattered tufts of conidiophores — only just visible under a lens — 

 at the exact places only where the spores were sown. On the 

 11 th day the fungus began to die away, and on the 15 day — by 

 which time the Oidium produced on B. interruptus and B. commutatus 

 was g-rowing vigorously and producing powdery masses of spores 

 — all traces of the fungus had disappeared. The non-permanence 

 of the fungus produced on B. brizaeformis, and its slight develop- 

 ment on this host, seem to point to the present case as being one 

 of ,subinfection' only. 



In the next experiment (no. 26) the Oidium on B. hordeaceus 

 again fully infected B. commutatus, and like the Oidium on B. inter- 

 ruptus, failed to touch B. madritensis. 



