528 
G. R. BISBY 
So, too, the rusts occurring upon these hosts have been found to be 
variable, uncertain, and confusing. Arthur^ has dealt with similar 
conditions in the case of the rose rusts, in which the variability of the 
host appears to be reflected in a similar degree by the variation in the 
rust. In his discussion of this condition Arthur states (p. 28) that 
"each species of Phragmidium has attained a degree of orthogenetic 
development and a diversity of characters corresponding to the hosts 
on which it occurs, always, however, with a certain lag due to the 
inhibiting nature of parasitism." It would seem that parallel con- 
ditions as regards variability, both with the hosts and their rusts, 
obtain in both the Rosaceae and Onagraceae. 
Considerable work has been done upon the rusts of the Onagraceae; 
the results, however, are rather scattered and somewhat conflicting, 
and the species described have seemed often to be of doubtful validity. 
The Sydows, as previously suggested, have done much toward sys- 
tematizing the knowledge of these rusts; Holway has published valu- 
able descriptions and notes on the species of Puccinia occurring upon 
this group. Descriptions of some of these rusts have already appeared 
in the North American Flora. Many other writers have added their 
ideas; yet it was apparent that further work upon a considerable 
number of collections of some of these species of rust should result 
in an attainment of greater order. The facilities of the Arthur her- 
barium have afforded to the writer an unequaled opportunity to study 
a large number of collections; preparation of manuscript upon the 
rusts for the North American Flora rendered it imperative that ad- 
ditional critical studies be made of Onagraceous rusts. 
The ideas regarding relationship, correlation, and classification, 
advocated in this paper, should be considered but the elaboration of 
the opinions held by Dr. J. C. Arthur, and by the workers in his 
laboratory. It has been possible to bring together the data presented, 
through the courtesy of Dr. Arthur in placing the facilities of the 
laboratory at the writer's disposal. Furthermore, thanks are due to 
Messrs. F. D. Kern, C. R. Orton, C. A. Ludwig, and especially to 
Prof. H. S. Jackson, for much assistance. y , . f r 
The term "correlation," as used in this paper and by other writers 
from this laboratory, is intended to express an idea of relationship of 
rusts based on similarities of morphological characters. This simi- 
^ Arthur, J. C, North American Rose Rusts. Torreya 9: 21-28. 1909. 
