THE COMMON MUSSELS. 33 



others, several together, higher up near the middle of the lamellse 

 (i I j, Fig. 38). In M. latus, in all I have examined, there were two 

 or at most three interlamellar junctions to one filament, and these 

 were placed one, as in M. edulis, near the ventral side of the lamella, 

 one near the middle of the lamella (i I j, Fig. 36), and the third, 

 which existed only in very long filaments, still higher up near 

 the base of the gills. In no case were there two interlamellar 

 junctions close above one another as we have them represented 

 near the middle of the filaments in M. edulis. The outer and 

 inner lamellae are thus more strongly connected by interlamellar junc- 

 tions in M. edulis than in M. latus ; and this may be necessary, as 

 the ascending lamellse in M . edulis are free at their upper edge, and 

 would thus be more liable to be torn apart from the descending 

 lamellse than in M. latus, whose ascending lamellse are joined at their 

 upper edges to the body or mantle (Figs. 35 and 37). 



The interlamellar junctions about the middle of the filaments 

 are found to be arranged in a very peculiar fashion, when we regard 

 the whole length of the lamella in M. latus. On looking through the 

 two lamellae of a gill, about the middle there are seen a series of 

 short oblique portions more opaque than the rest (i I j, Fig. 39), and 

 occupying the middle line or long axis of the gill. On separating the 

 filaments, it is found that these opaque portions are formed by the 

 middle interlamellar junctions. Each of these opaque bars extends 

 over about fifty filaments, and they are obliquely placed because in 

 each set of fifty filaments the middle interlamellar junction of any 

 one filament is a little higher than that of the gill preceding. In 

 i I j, Fig. 40, I have attempted to show this arrangement diagram - 

 matically. This figure represents a portion of a gill in which a 

 number of succeeding filaments have been separated and turned round 

 to show the interlamellar junctions. If, in this figure, instead of 

 only five filaments there were fifty or more, it would represent 

 roughly a part of a gill crossed by one of the oblique bars of inter- 

 lamellar junctions. 



The interfilamentar junctions in M. latus are not exactly in the 

 central line of each filament, as represented in Peck's figure of M. 

 edulis {ifj, Fig. 38), but are near the edges (ifj, Fig. 36). 



The ascending lamellae in M. magellanicus are free, as in M* 

 edulis. 



" Cavite des flancs " (Sabatier) . 



This opening is of no great importance, except as marking a con- 

 stant difference between M. latus and edulis or magellanicus, as it 

 3 



