Reed—Infection Experiments with 21ildeiv. 543 
all our previous results, we should expect a specialized form 
of mildew for each genus of the cucurbits. Nothing of th© 
sort occurs, however. There is no evidence of any specializa¬ 
tion in the mildew of this family. 
Some experiments, however, indicate that the mildew on the 
cucurbits is limited to this one family. Conidia from squashes 
have been sown on species of the golden rod ( Solidago ) and 
Aster , which are reported as host plants of Erysiphe cichora - 
cearum DC., without causing infection, and I am carrying on 
still further experiments in this line. 
Why results so different from those reported for the grass 
mildew, for example, should he obtained with the mildew on 
the cucurbits is by no means apparent. It is true, of course, 
that all the cucurbits studied are quite similar in their mor¬ 
phological characters and form a group of plants quite similar 
also in their ecological relations. They all thrive well under 
approximately the same conditions of soil, moisture, light, etc. 
Comparisons between two such widely separated families as 
the Gramineae and the Cucurbitaceae are, of course, very diffi¬ 
cult, and it is a question whether we can find from this stand¬ 
point any light on the question as to why the grass mildew is 
split up into physiological species while the cucurbit mildew 
is so entirely adaptable and cosmopolitan. Many species of 
Poa, for example, differ more among themselves ecologically 
than do Cucurbita maxima and C. pepo, and this may be a rea¬ 
son for greater specialization among the mildews of the Poos. 
On the other hand, there are no greater ecological differences 
between wheat, barley and rye than between squashes, pump¬ 
kins and bottle gourds. Whether in morphological characters 
Triticum and Secale are more different than Cucurbita and La - 
genaria is a difficult question. 
Salmon (29) claims to have found four or five physiological 
species of mildews on species of Brome grasses, and here we 
certainly do not find differences in structure and habit any 
more marked than are present in the cucurbits. Salmon as¬ 
serts, for example, that conidia of Bromus commutatus will not 
infect B. racemosus, and vice versa. These two species, how¬ 
ever, are essentially the same with reference to their ecological 
