S a 1 m o n , On Specialization of Parasitism in the Erysiphaceae. 263 



2. E. Graminis f. spec. Hordei, sur Hordeum hexastichon, vul- 

 gare^ trifurcatum, nudum, jubatum et murinum, non sur H.. 

 maritimum, secalinum, et bulbosum. 



3. E. Graminis f. spec. Seealis, sui" Seeale cereale et anatolicum. 



4. E. Graminis f. spec. Avetiae, sur Avenn satioa, orienialis, fatua 

 et sur Arrhenatherum elatius. 



5. E. Graminis f. spec. Poae, sur divers Poa, notamment P. 

 annua, tricialis, prate?isis, caesia, mutaletisis, nemoralis, et 

 seroiina. 



6. £. Graminis f. spec. Agropijri, sur les Agropyrum. 



7. ii. Graminis f. spec. Brotni, sur divers Bromics, notamment 

 sur -B. moUis et sterilis. 



We may note tliat of tlie seven ,formes' liere enumerated, five 

 are ranked as thougli being- able to infect all of the species of tlieir 

 respective genera. We shall see later tliat tliere is reason to doubt 

 if such infection-powers are common; whilst we can prove in one 

 at least of the above cases, that not one, but several jbiolog'ic 

 forms' exist within the genus of host-plants. In the case of the 

 tvvo exceptions given by Marchai, the , forme spec. Hordei'' is 

 stated to be able to infect Hordeum vulgare and five other species 

 of the genus, but to be unable to infect H. maritimum, H. secali- 

 num, and H. bulbosum. Now, E. Graminis occurs on at least one 

 of these last three species, viz. H. secalinum (1, p. 211) so that 

 we may assume that it is probable that there exist at least two 

 ,formes specialisees' within the genus Hordeum. 



In the second instance, it was found that the ,forme spec. 

 Tritici'' would not infect T. durum, T. monococcum, or T. dicoccuni; 

 as however, E. Gramitiis has not up to the present been recorded 

 on any of these three species, there is no evidence here, as yet, of 

 specialization occnrring within the genus. 



In a paper (3) dealing with the American Gooseberry-mildew 

 I liave lately pointed out that the conidia of Sphaerotheca mors-uvae 

 on Rihes Grossularia will infect R. Cynosbati. 



In August last I was able, through the kindness of Prof. 

 Mar shall Ward, to carry out, in the Cambridge Uni versity Labo- 

 ratory, a series of experiments with the Oidia of certain species of 

 Erysiphe. I wish to take this opportunity of expressing my thanks 

 for the excellent laboratory facilities wliich were afforded me, and 

 especially to thank Prof. Marshall Ward for the kind and valu- 

 able assistance given to me during the work. 



Before giving the lesults of these experiments a few remarks 

 may be made on the morphological characteristics of the Oidia 

 used, and on the methods employed. 



The Oidium used in the main series of experiments was that 

 of E. Graminis on species of Bromus, Trificum, Avena, Festuca 

 and Lolium. The general morphological characteristics of this 

 Oidium can be Seen on referring to Figs 1 — 3. As in all the 

 conidial forms of the Erysipheae, we find a creeping mycelium 

 consisting of delicate branched septate hyphae, which send at inter- 

 vals haustoria into the epidermal cells. On this mycelium a great 



18* 



