THE UREDINIUM IN PUCCINIASTRUM AGRIMONEAE 
59 
that the urediniospores in certain species of Melampsorella and 
Pucciniastrum are catenulate, although sometimes appearing much 
as if pedicellate ; and it does not seem to be unreasonable to anticipate 
that the same condition is to be found in Melampsoridium. In addi- 
tion, it is now also definitely established by Bartholomew's work that 
the true pedicellate habit is to be found in Hyalopsora; and it is at 
least highly probable that the same condition is typical of the other 
fern rusts. Now, if these assumptions prove correct, there is provided 
a set of invariable characters upon which the larger divisions of the 
group may be separated. The points of distinction formerly used, 
such as wall color or number of cells in the teliospore, are often variable 
to a large degree within the same sorus. 
This method of dividing up the group would give two subgroups, 
in one of which the urediniospores are borne singly on pedicels and 
in the other of which, .when present, they are borne in chains with 
each chain maturing but one spore at a time. To the first of these 
two subgroups would belong the three fern rust genera Uredinopsis, 
Milesia, and Hyalopsora; while to the latter would belong the genera 
Pucciniastrum (Thekopsora) , Melampsorella, Melampsoridium, and 
Calyptospora. Calyptospora, a genus without uredinia, is neverthe- 
less included because such characters as it possesses show close affinity 
to Pucciniastrum. This genus, therefore, gives no real trouble in 
this connection, even though it could not be taken care of in a working 
key if uredinial characters only were used in the division. 
The division of the Pucciniastratae along the line suggested has 
the additional fact in its favor that it groups the fern rusts together 
and separates them from those rusts of the group which have the 
sporophytic stage on spermatophytes. Because of the evident simi- 
larity of the fern rusts, such a change has always been desirable, but 
until now lack of sufficient morphological information has prevented it. 
It is possible, of course, that further study within the group will 
bring to light facts which will make necessary a further realignment 
of the genera. However, from the data now at hand it would seem 
that the grouping here suggested is more nearly along natural lines 
than any heretofore proposed. 
The writers wish to acknowledge here the aid received from Dr. 
J. C. Arthur, Prof. H. S. Jackson, and Dr. F. D. Fromme, to whom 
they are indebted for a number of helpful suggestions. 
Purdue University, 
Agricultural Experiment Station, LaFayette, Indiana 
