282 Salinon,On Specialization of Parasitism in tlie Erysiphaceae. 



from d (02 b) becomes at once from its residence on b capable of 

 infecting/ and ff but not a c d e? This would mean the trans- 

 formation of Oacb and Odbc into Ob/ff on either passing 

 on to b. 



026 



Om)k 



■^02d 



fig. 3. 



4. Lastly, we will suppose that 0« and Od are capable of 

 mutually infecting their liost-plants, but behave differently towards 

 other hosts, 0« infecting c but not e, and Od infecting e but 

 not c; thus (Fig. 4). — 



In such a case we must suppose either (4«) that there exist 

 on host a two jbiologic forms' and two also on host 5, viz. on a 

 0\a infecting d and c, and 02^/, infecting d and e, or even by 

 virtue of its residence on f/, c also ; similarly, mufafis mutandis, with 

 the Oidia on d. Or, (4 ß) that 2 on reaching a from d loses the 

 power of infecting e, and by acquiring the power of infecting r, 

 becomes then identical with Ol«; similarly with Ol on r/. 



0/Ci 



■^'X 



0/a ••-^„„^^^^ ^^^^02d 



O2 a '^'^ ^^^--^ Oid 



fig. 4. 



In connection with the above problems, the question may be 

 asked: how do we know that the Oidium found on «, c?, etc. 

 belongs really, as is here assumed, to one ,biologic form' only. In 

 other worde, may not the host-plants a, d, etc. be the meeting-place 

 of jbiologic forms' of Oidium from various sources, such as the host- 

 plant b is in diagram 3? Whilst admitting that as a rule we have 

 no absolute proof that the Oidium on any given host-plant, which 

 may have been used in experiments, always belongs to one specialized 

 form only, it may be remarked that we have grounds for assuming 

 it in the cases where spores are obtained from an Oidium which 

 has been growing in nature for a long time, i. e. through many 

 generations, on the same plants of the host-species. 



We will return now to the series of experiments with the Oidia 

 on Bromus, tabulated below (Table 8). A consideration of the 



