(>ki mann: Families and Genera of Najades. 227 



which term there for this purpose. This "unnatural" discharge is 

 known only in the Lampsilinee, and is unknown in the Anodontina and 

 Unioninee. Finally the familj of the Unionida differs from the 

 Margaritanidce and is more highly advanced in the formation of rudi- 

 mentary siphons. But in this respect this family is not very progres- 

 r-i\r. It has the anal and branchial openings separated only by the 

 (complete) gill-diaphragm, and in addition, it has the anal closed above, 

 thus giving it an incomplete tubular shape. Beyond this, there is no 

 progress in this family. The presence of a supra-anal opening is, in 

 my opinion, only incidental to the closing of the anal. 



The members of the third family, the .1/ ittclidcc, have gone in another 

 direction in t heir development. If the expression may be permitted, 

 they lay chief stress upon the better development of the siphons, while 

 in the differentiation of the gill-functions they have started out from 

 the beginning with another idea, which, however, has not attained a 

 very high degree of perfection. With regard to this it may be said 

 that they have restricted the marsupial function to the inner gills, 

 and very likely the anterior connection of these gills with the palpi is 

 incidental to this function. Not much advance is to be observed in 

 the gill structure, and only two types are met: incomplete septa and 

 intercommunicating water-tubes (a rather primitive condition) in 

 one group (Hyriince) ; and complete septa and water-tubes in the other 

 group (Mutclificc). Very likely the latter structures are not homolo- 

 gous to the septa and water-tubes of the Unionidcc, but have been ac- 

 quired independently, since their finer structure is different. With 

 regard to the siphons, which attain within this family their highest 

 perfection among the Najades, we have first of all a complete separa- 

 tion of anal and branchial openings by a firm mantle-connection, 

 which forms the posterior continuation of the gill-diaphragm, and in 

 addition we have a tendency to close both the anal above, and the 

 branchial below, by mantle connections. It is true that this tendency 

 is not yet perfect in main- Mutelidce, but it is developed within this 

 family, so that in the most highly specialized genera we have two real, 

 tubular siphons, formed by complete coalescence of the edges of the 

 mantle. 



I think the above account of the phylogenetic tendencies within 

 the various divisions of the Najades will make it clear that the mor- 

 phological characters upon which our new system is founded are 



