132 BULLETIN OF THE BUREAU OK FISHERIES. 



which are similar to those occurring on the interlamellar junctions farther in. The 

 respiratory canals must be capable of expansion and contraction to a considerable 

 degree, as a rich supply of smooth muscle fibers, passing in both a vertical and a hori- 

 zontal direction, may be seen underlying the epithelium of the canals everywhere except 

 in the septum (fig. 58, pi. xiv). Large blood sinuses (b. s.) are found in the lamellae 

 just outside of the canals, as seen in this figure, which shows how close the blood must 

 come to the water within the canals (r. c). There can be no doubt that the water passing 

 through the canals is a respiratory current. 



Although the respiratory canals open dorsally into the suprabranchial chambers, 

 the marsupial division of the water tubes is completely closed off from the latter, as 

 Ortmann has stated, by a roof which is developed in connection with the septa forming 

 the respiratory canals. The dorsal free border of each interlamellar junction at the 

 level of the suprabranchial chamber expands both anteriorly and posteriorly, but only 

 over the marsupial division of the tube. The anterior and posterior edges of these 

 umbrella-like expansions fuse with each other in exactly the same way as do the septa 

 already described, and, since they also become continuous laterally with the vertical 

 septa which separate the respiratory canals from the marsupial spaces, the latter 

 thereby come to be completely roofed over and do not open at all into the suprabranchial 

 chambers, unless the covering is broken. Of course, the formation of the roofing mem- 

 brane does not take place until after the marsupium is fully charged with eggs. Owing 

 to the gorged condition of the marsupium in these genera, the egg masses cause the 

 roof to bulge up into the suprabranchial chamber over the marsupial division of each 

 water tube, and on exposing the chambers the upper ends of the egg masses, covered, 

 however, by the delicate transparent roofing membrane, are seen protruding beyond 

 the dorsal boundary of the gill. In the drawing of Syni phynota complanata (fig. 3, pi. 

 vi), in which a portion of the suprabranchial chamber is exposed, the condition just 

 described is distinctly shown. 



As Ortmann has described, the secondary division of the water tubes entirely 

 disappears after the discharge of the glochidia. The dorsal expansions of the inter- 

 lamellar junctions, which united to form the roof, give way along the original sutures, 

 and the glochidia are enabled to pass out; the septa separate in a similar manner, and 

 are gradually retracted, and when the marsupium returns to the resting condition no 

 trace of these structures is to be seen. 



We have confirmed Ortmann's discovery of the respiratory canals in Alasmidonta, 

 Anodonta, Sirophiius, and Symphynota. Figures representing the marsupial structure in 

 Anodonta cataracta (fig. 51, pi. xiii; 57, 58, pi. xiv) have already been referred to. Figure 

 56, plate XIV, is a section taken from near the ventral end of a water tube in the gravid 

 marsupium of Alasmidonta truncata; the young embryos with which the marsupium 

 is filled are not shown. The respiratory canal (r. c.) at this level is quite small and 

 less sHt-like than in Anodonta, but it widens out toward its dorsal end. The nuclei of 

 the interlocking cells where the edges of the opposite septa have fused are quite distinct 

 in the section. Figure 52, plate xin, shows a horizontal section from the gravid mar- 



