Salmon, On Specialization of Parasitism in the Erysiphaceae. 283 



relations of certain Oidia with certain of their liost-plants brought 

 to light by these experiments shows the existence of problems lesemb- 

 liüg- the theoretical ones discussed above. The inter- relations of 

 the Oidium on B. interruptus and that on B. hordeaceus in connec- 

 tion with certain host-plants may be represented diagrammati- 

 cally thus (Fig. 5): — 



We find on B. hordeaceus, «, an Oidium which will infect B, 

 commutatus and B. interruptus; we will call this Oidium Olo. On 

 B. interruptus^ c, we find an Oidium which, while infecting B. 

 hordeaceus, cannot infect B. commutatus. On account of this ditfe- 

 rence, the Oidium on B. interruptus must, adopting cnrrent termi- 

 nology, be considered a distinct ,biologic form'; we will call this 

 Oidium 02 c. Further, on B. commutatus, b, we find a third Oidium, 

 which diifers biologically from the two Oidia, Ol and 02, in not 

 being able to grow on B. interruptus. (The behaviour of this 

 Oidium towards B. hordeaceus is not definitely known, although in 

 all probability the Oidium is incapable of infecting this species also, 

 since it was proved to be unable to infect B. mol/is, and, as menti- 

 oned above, B. hordeaceus is probably synonymous mth that species). 

 We will call this third Oidium, O'db. The problem to be solved 

 may be stated thus: does there exist on c two ,biologic forms' of 



03b% - --¥.Q2c 



Oidium, 02. c and another 1 r, derived from the spores of 1 a ; 

 or does 1 a on infecting c become at once physiologically identical 

 with 02 c, i. e. does it at once lose its power of being able to infect B. 

 commutatus'^ Similarly, there must exist on a either only one ,bio- 

 logic form' Ol«, which would mean that the spores of 02c on 

 infecting a give rise to an Oidium which at once acquires the power 

 of infectüig B. commutatus; or there must exist here two ,biologic 

 forms,' Ol« and another form 02« derived from the spores of 02c. 

 Again, have we on 5 one ,biologic form' only, Odb, which would 

 mean that the spores of Ol« on infecting b at once give rise to 

 an Oidium which differs from Ol« in being unable to iirfect c ; or 

 havewe on b two ,biologic forms', 03 i and another 1 <?•, derived 

 from the spores of Ol«? 



A more complicated case presents itself when we consider the 

 inter-relations shown by the three Oidia which grow on B. tectorum. 

 These relations may be expressed diagrammatically thus (the sign 

 — >? = ,subinfection') (Fig. 6): — 



On B. tectorum, a, we find an Oidium diifering from all the 

 others here represented in being able to infect B. sterilis, e\ we 

 will call this Oidium Ol«. On B. hordeaceus, b, we find another 

 Oidium, and on B. interruptus, c, a third Oidium. Both of these 

 are able to infect B. tectorum, but are Seen to differ biologically 

 from each other in that the former can infect B. commutatus, while 



