THE NAUTILUS. 115 



Some corrections of Simpson's system have been advocated by 

 Sterki (Amer. Naturalist 37, '03 p. 103 ff.), and, generally, I find 

 that these are well supported. Yet there are other differentiations 

 in structure, which have been overlooked hitherto, and which are of 

 prime systematic value. 



The most important (and gratifying) result of my investigations is, 

 that the gills or parts of gills, which serve as marsupium during the 

 breeding season, are permanently differentiated in their anatomical 

 structure from those gills (or parts of gills'), which never serve as mar- 

 supium. Thus it is possible to recognize the type of marsupium of 

 any species, if only sterile females are at hand. 



My investigation will be published in full in the "Memoirs of the 

 Carnegie Museum," amply illustrated by figures of the structures 

 discussed. Although I am not quite done yet, and although I hope 

 to secure additional material next summer, I think it advisable to 

 publish my preliminary results as soon as possible, in order to direct 

 the attention of the collectors to those genera, which 1 have not 

 been able to investigate. It would be desirable for anybody, who 

 intends to collect Unionidce next summer, to make up his mind from 

 the beginning, that shells without the soft parts are not the thing 

 that is wanted, but that the soft parts, at least of some specimens, 

 should always be preserved in alcohol. 



The family Unionidce,, in Simpson's sense (1. c, p. 514, 515, 516), 

 is accepted. Simpson divides it into two subfamilies, of which one, 

 Hyriincz (= Hyriamz Swains., Simpson, pp. 515 and 806), is extra- 

 limital, and which cannot be discussed at present. The other is the 

 JJnionincE Swains. All our North American forms belong to the 

 latter. But I think they should be divided into four groups, which 

 I would designate as subfamilies. Thus, disregarding the Hyriina, 

 the Unionidce are divided into four subfamilies, as shown below. 



In the following I shall give a systematic arrangement of the 

 Pennsylvanian species. Of the characters, I have given the most 

 important ones of the soft parts for the subfamilies, and for the 

 genera where necessary. It is not my intention to go into any de- 

 tail, since further investigations may possibly necessitate minor 

 changes in the arrangement of the genera. Some notes are ap- 

 pended at the end, in order to explain the most striking changes 

 introduced here. 



