Salmon-, On Specialization of Parasitism in the Erysiphaceae. 271 



conidia witliout the fungus deriving any assistance by means of 

 haustoria from the plant on which the spores are sown. It may 

 he mentioned in support of the latter view that we often tind on the 

 inoculated leaves in the cases under consideration scattered groups 

 of a few conidiophores without any trace of a mycelium ohservable 

 under a simple lens. Is it possible that these conidiophores are 

 produced direct from germinating conidia? We may recall the follow- 

 ing Observation made by Dangetird (7) on the germination of 

 the conidia of Sphaerotheca Casfugnei: — ,Cependant, quelquefois, 

 la conidie produit, en plus de quelques fllaments vegetatifs, un conidio- 

 phore qui se forme directement ä ses depens.' On the other band, 

 if the conidia are capable of germinating and living for a short time 

 independently of the plant on which they are sown, we ought to 

 find such cases occurring quite commonlv in inoculation experiments, 

 wiiereas this is certainly not the case. I am inclined to think, there- 

 fore, that in the present case of the Oidium sown on B. brizaeformis 

 and on B. velutinus — and in all similar cases — a faint infection 

 of the host-plant really occurred. 1 propose to call such cases ones 

 of ,subinfection.' 



Jn connection with the present subject some striking cases ob- 

 served by Neger (4) in his experiments may be mentioned here. 



In two instances when Neger sowed the conidia of an Oidium 

 (believed to be that of Erysiphe Polygoni) occurring on Rcmmiculus 

 repens on plants of Galium sihaticuin Single conidiophores appeared 

 after two days on the inoculat(»d plants. By the next day, however, 

 the conidiophores had completely disappeared. The experiment was 

 subseqiiently repeated numerous times, but always gave negative 

 results, and Neger expressly states that in his opinion the two 

 cases in which a few conidiophores appeared on the Galium are 

 not to be taken as proving that the Oidium on Ranunculus repens 

 can really infect this plant. This case is of special interest because 

 strong a priori grounds existed for supposing that this Oidium^ 

 which in all probability was that of E. Polygoni^ would be incapable 

 of infecting Galium^ since in nature we never find E. Polygoni on 

 species of Galium^ but another species, viz. E. Cichoracearum (see 

 foot note, p. 2.) 



In another case, in a Single experiment, the conidia of an Oidium 

 growing on Hieracium murorum — a plant on which E. Cichora- 

 cearum has been recorded — were sown on Leonfodon Taraoracum, 

 and after 3 days a few scattered conidiophores appeared. These 

 had completely disappeared by the next day. Now, on Leontodon 

 Taraxacum the only member of the Erysiphaceae which occiu^s in 

 nature is, I believe — notwithstanding many Statements to the 

 contrary (see 1, p. 199, 201) Sphaerotheca Humuli var. fuliginea. If 

 this be so, and if the Oidium used by Neger was really that of 

 E. Cichoracearum^ the present case like the preceding one would 

 supply evidence that conidia can germinate and produce conidio- 

 phores on hosts which they can never really infect. The fact must 

 not be lost sight of, however, that >S. Humuli var. fuliginea has 

 been recorded on Hieracium sabaudum, and I have personally ob- 

 served it on an undetermined species of Hieracium. For this reason, 

 and also because it was not ascertained to what species the Oidium 



