850 PROCEEDINGS OF THE NATIONAL MUSEUM. VOL.XX. 



efferent orifice differentiated. The ligament may be wholly external, 

 as in Mallet-ia, Tindaria, etc., or it may be rudimentary and replaced 

 by an internal cartilage or "resilium," or both may coexist in varying 

 degrees of development and degeneration. The hinge teeth may be very 

 numerous and regularly V-shaped in each series, or they may be com- 

 paratively few and irregular, sometimes becoming oblique and lamelli- 

 form (Silicula). The beaks generally turn backward (Yoldia, Leda, 

 Nucula], but in Malletia, Tindaria. and some other genera they turn 

 forward. On this account, when there is neither pallial sinus nor 

 external ligament, it is often difficult, if not impossible, to tell which is 

 the anterior end of the shell without the soft parts. Hence many fos- 

 sil and some recent species have probably been reversed in the descrip- 

 tions. Thus many of the Palaeozoic species referred to Nucida are 

 described as having the beaks turned forward, the longer end of the 

 shell being considered posterior, but in modern Rucuhc the beaks turn 

 backward and the shorter end is posterior. Many of the deep-sea 

 species with small, thin shells show no distinct muscular nor pallial 

 scars, which increases this difficulty. When a differentiated external 

 ligament is present, we have assumed that it is posterior to the beaks 

 (opisthodetic), though a narrow extension usually runs under and for- 

 ward of the beaks in a groove. When the shell of a dimyarian bivalve 

 gapes posteriorly, the existence of a siphon may generally be assumed; 

 for otherwise the internal soft parts would be exposed to enemies. The 

 existence of a posterior rostrum or a protrusion of the posterior margin 

 defined by an inferior emarginatiou indicates the existence of a siphon, 

 or at least an anal tube, but these organs may exist without such mod- 

 ifications of the shell. If these rules be applied to Palaeozoic forms we 

 must conclude that the rostrate and subrostrate forms of Palcconeilo, 

 etc., had some sort of a siphon, and therefore were not true Nuculidie. 



Numerous Palaeozoic species referred to the genus Paheoneilo proba- 

 bly belong to or near the Tindarinae. Some of the species ] from the 

 American Devonian rocks can hardly be distinguished from Tindaria 

 by any important structural characters, unless it be the form of the 

 teeth. It is probable that Nuculites and several related genera belong 

 near this division, for they have an external ligament and no resiliuni. 

 In these genera the plain, transverse teeth are very numerous and 

 more simple than in the modern genera, seldom showing any trace of 

 the acute, V-shaped form characteristic of most modern genera, though 

 in some species the teeth are slightly angulated in the middle. 



Mr. Dall has proposed the family Ctenodontid;e 2 to include numerous 

 Palaeozoic species belonging to Ctenodonla, and allied genera, some of 

 which Zittel and others refer to Arcidie on account of their thickened 

 pectuuculoid shells. They seem to be allied rather to Tindariuoe. 



1 For example see P. constricta Hall, P. plana Hall in Paleontology of New York, 

 V, Pt. i, pp. 333, 334, pi. XLVIII, figs. 1-28, 1885. 

 - Trans. Wagner Free Inst., Ill, p. 515, 1895 



