3 a 1 m n , On Specialization of Parasitism in the Erysiphaceae. 277 



ulation, it is possible that tlie lütimate amoimt of infection would 

 have been greater. 



The characteristics shown by this Oidium on B. hordeaceus and 

 its var. glahrescens, as sliown by the above 12 experiments, are as 

 follows. The Oidium is capable of causing füll infection on B. mollis, 

 B. interruptus and B. tectorum; füll infection nearly always on 

 B. commutatus; ,subinfection' on B. brizaeformis (7 times), on B. 

 velutinus (3 times out of 4), and on B. secalinus (3 times out of 5). 

 It fails to infect B. arve?isis, B. rucemosus, B. maximus, B. sterilis, 

 B. asper, B. erectus, B. macrosfachys, B. madritensis, B. gigan- 

 teus, B. inermis, B. patulus, B. criniius, and B. arduennensis var, 

 villosus. 



A summary of the experiments in which the Oidium on B. 

 hordeaceus and its var. glabrescens was used is given in Tables 

 2 and 3. 



A series of experiments was now made with a third Oidium, 

 viz. one occurring on B. commutatus. It become at once apparent 

 from the first experiraent (Exper no. 15) that we were now dealing 

 with a fungus with very different powers of infection from the 

 Oidium both on B. interruptus and on B. hordeaceus. The Oidium 

 was first sown on B. commutatus, B. mollis^ B. racemosus, B. secali- 

 nus, B. sterilis, B. arcensis, B. maximus and B. erectus. Füll in- 

 fection resulted on B. commutatus (3 pots) and füll or nearly füll 

 infection on B. secalinus (2 pots); while no infection occuiTed on 

 potted plants of B. racemosus (1 pot) nor on seedlings of B. mollis, 

 B. sterilis, B. arvensis, B. maximus and B. erectus (2 pots each). 

 For the first time, it must be noted, we were now dealing with an 

 Oidium which failed to infect B. mollis. 



In the next experiment (Exper. no. 24) the Oidium was sown 

 on B. commutatus and B. velutinus, and fully infected both species. 



The Oidium was then tried (Exper. no. 36) on B. commutatus, 

 B. tectorum and B. macrostachys. Here B. commutatus alone was 

 fully infected. On B. tectorum ,subinfection' in a very weak form 

 occurred on the inociüated plants in one pot, while the result was 

 doubtful in the other. No infection occurred on B. macrostachys. 



In the next experiment (Exper. no. 41) the Oidium was found 

 to be incapable of infecting B. interruptus, as it was in the case of 

 B. mollis in the first experiment (Exper. no. 15); it also failed to 

 touch B. asper. 



In Exper. no. 60 it was found that this Oidium — which from 

 the foregoing experiments is seen to differ in several particulars 

 from the Oidium both on B. interruptus and B. hordeaceus — yet 

 agreed with both in causing ,subinfection' on B. brizaeformis. In 

 this experiment the fungus proved to be imable to touch B. madri- 

 tensis. 



In the last experiment (Exper. no. 66) with this Oidium the 

 fungus was sown on seedlings of B. commutatus, B, giganteus, B. 

 inermis and B. patulus, and on a potted plant (about 10 weeks 

 old) of B. racemosus. Of these plants only B. commutatus was 

 infected. 



Beihefte Bot. Centralbl. Bd. XIV. 1903. 19 



