S a 1 m n , On Specialization of Parasitism in the Erysiphaceae. 279 



was sown on 1 pot eacli of mature plants of Festuca elatior var. 

 pratensis and Poa annua\ and the Oidium on B. hordeaceus var. 

 glahrescens on 1 pot each of seedling plants of Arrlienaiherum ela- 

 tius, Lolimn perenne^ Poa annua, Festuca elatior var. arundinacea^ 

 Dacfylis glomerata and Alopecurus pratensis. No sig^ns of the least 

 infection resulted. 



Referring- now to the main series of experiments, which are 

 summarized in Table 8, we find that this gives evidence that four, 

 if not five, of the Oidia used are ,biologic forms,' each with different 

 physiolog-ical characteristics as shown by their distinctive powers of 

 infection. At Table 9 five Oidia which were used in the experiments 

 are so arranged as to show these biolog'ical characteristics. In this 

 Table the number of times that each species of host was tried is 

 given in brackets. We see that four or more probably five, of 

 these Oidia must rank as ,biolog'ic forms,' since they show re- 

 spectively the following- distinctive peculiarities. 



The Oidium on B. interruptus diifers from that on B. hordea- 

 ceus in never infecting' B. commufafus or B. secalinus; from that 

 on B. commutatus in causing füll infection on B. mollis, B. inter- 

 ruptus and B. tectorum, and in not infecting B. commutatus or B. 

 secalinus, and apparently in never fully infecting-, and sometimes not 

 infecting" at all, B. velutinus; and from the Oidia on B. tectorum 

 and B. arvensis in not infecting* B. sterilis or B. arvensis. 



The Oidium on B. hordeaceus difi'ers from that on B. inter- 

 ruptus in nearly always (9 out of 10 cases) infecting B. commutatus, 

 and apparently in sometimes ,subinfecting' B. secalinus\ from that 

 on B. commutatus in infecting fully B. mollis, B. interruptus, and 

 B. tectorum, and in not fiüly infecting or quite failing to touch 

 B. secalinus and B. velutitius; and from the Oidia on B. tectorum 

 and B. arvensis in not infecting B. sterilis or B. arvensis. 



The Oidium on B. commutatus differs from that on B. inter- 

 runtus in not infecting B. mollis, B. interruptus, and apparently 

 only slightly B. tectorum, and in infecting fully B. commutatus, B. 

 secalirius and B. velutinus; from that on B. hordeaceus in not in- 

 fecting B. mollis, B. interruptus, and apparently only slightly B. 

 tectorum, and in infecting full}^ B. secalinus and B. velutinus; and 

 from the Oidia on B. tectorum and B. arvensis in not infecting B. 

 sterilis or B. arvensis. 



The Oidium on B. tectorum may be the same as that on B. arven- 

 sis. although it is very improbable that this is thecase; the Oidium 

 on B. tectorum and B. arvensis differs from those on B. interruptus, 

 B. hordeaceus and B. commutatus in infecting respectively B. sterilis 

 and B. arvensis. 



It may be noted here that the only mention by Marchai of 

 the Oidium on Bromus is found in the following Statement (6, p. 212): 

 ,E. Graminis, forme specialisee Bromi, sur divers Bromus, notamment 

 sur B. mollis et sterilis.'- We see, however, from the foregoing ex- 

 periments that not merely one, — but several, specialized or ,bio- 

 logic' forms exist on the species of the genus Bromus, and that as 

 a matter of fact the Oidium on B. hordeaceus (which is probably 

 synonymous with B. mollis) is unable to infect B. sterilis. 



19* 



