238 ANNALS OF THE CARNEGIE MUSEUM. 
there is or has been a general tendency to restrict the marsupium to 
the outer gills, and that this forward step in a quite natural direction 
has been made independently in various groups. That is to say, the 
Unio-type of marsupium has repeatedly developed from the Quadrula- 
type by parallel evolution. 
Of the other features of the soft parts only three furnish some help 
for the distinction of genera. The first and most important is the 
character of the placenta, revealing differences which are of prime 
value, but affect only a few forms, as will be seen below. The second 
is the separation of the anal and supra-anal openings. Although 
characteristic of certain forms (in one case these openings are not at 
all separated), it is somewhat variable in others, even individually. 
Thus we can use this character only to a limited degree. The third 
is the connection of the inner lamina of the inner gills with the ab- 
dominal sac. Here there seems to be a difference between certain 
forms of the Old and the New World. But, unfortunately, too few of 
the former are known for me to express a final judgment. 
Thus the soft parts alone would furnish only few criteria for the 
distinction of genera, and we should direct our attention to the shell. 
Here we have indeed great variety, and the shapes of the shell have 
been largely used heretofore for the definition of genera. The most 
important feature, in my opinion, is the beak-sculpture, which, how- 
ever, has been largely misunderstood by Simpson. In fact in this 
primitive subfamily we have, side by side, all the different types 
of beak-sculpture, and, as we shall see, they may be used to great 
advantage. 
Since various types of shell-structure are frequently combined with 
various types of soft parts, it would not do to make only a few large 
generic divisions. For if we recognize, for instance, only two main 
genera according to the character of the marsupium, the same types 
of shell would turn up in either of them, which surely would give an 
incomplete or wrong impression of affinities. Thus, in my opinion, 
it is advisable to admit a larger number of genera founded upon both 
the structure of the soft parts as well as of the shells. Such a scheme 
is introduced here, at first, tentatively, but I hope it finally will prove 
to be the most convenient. 
Finally I should mention the glochidia of these forms. The latter 
are known in a number of North American.species, where they always 
are of a primitive shape (see Plate XIX, fig. 1). They are also 
