S a Im 11 , On Specialization of Parasitism in the Erysiphaceae. 28 1 



ities, as is shown by the fact tliat certain individuals are not ca- 

 pable of infecting all the hosts concerued. If we deflne a ,biologic 

 form' merely as coiisistiug- of individuals whicli possess certain con- 

 stant physiological peculiarities as shown by their powers of infec- 

 tion when on one host, then 0« and Od must be regarded as two 

 jbiologic forms,' since they are not capable of infecting respectively 

 d and a. If however we add to the definition that a certain fixity 

 of the characteristics by heredity is necessary, then we requii^e 

 more evidence than the fact of the non-infection of a by the Oidium 

 on d, and of d by the Oidium on a, gives, before we can call Qa 

 and d two distinct ,biolog-ic forms.' Thus, if it were found that 

 Oa on passing- on to b at once became capable of infecting t?, this 

 would prove that the peculiarity of 0« in not being able to infect 

 d was not hereditarily fixed; if on the other band, it were found 

 that a on passing to b (= 1 6) required to grow for several 

 generations on b before it was rendered capable of infecting d, then 

 the Claim of Oa to be regarded as a distinct ,biologic form' would 

 be a strong one. 



2. The next case we may consider is that in which the 

 Oidium on a (Ola) infects another host than b, namely c, while 

 02 on (/ infects e as w^ell as 6; 02 not beiiig able to infect c, 

 and Ol not being able to infect e. These relations may be ex- 

 pressed thus (Fig. 2). — 



In such a case the question arises, have we (2a) on 6 two 

 ,biologic forms' one infecting a b c, the other d b c^ according to 

 origin f rom a or d respectively ; or (2 ß) does a tliii-d ,biologic form' 

 come to exist on b, capable of infecting all or most of the plants 

 a b c d e'^ 



O/c yfOjb Oje 



0;a'k ^^-^«^ 



fig. 2. 



3. Now let US suppose that we find in natui^e an Oidium on 

 b (03 6), which proves to be a ,biologic species' capable of infecting 

 f and .9, but not a c d e. These relations may be expressed dia- 

 grammatically thus (Fig. 3j. — 



Complicated problems would arise in such a case as this. It 

 might possibly be that we should find on b three ,biologic forms,' 

 01^», 02 b, 03 6, restricted in their powers of infection, according 

 to their origin, to respectively a b c, d b c, f b g. The problems 

 that present themselves in connection with Ol 6 and 02 6 are those 

 considered in case 2 above (2« and '2ß). The fact, however, that 

 we have here assumed the occurrence in nature of an Oidium on 

 6, confined io b f g, suggests the further question: is it possible 

 that there would be on b only this 03 {Obfg); that is, is it possible 

 that the Oidium found on b, w^hether derived from a (Ol 6) or 



