intralamellar spongy growth becomes, the more do the original gill-filaments lose the character of blood-holding tubes, and tend to become dense elastic rods for the simple purpose of supporting the spongy growth. This is seen both in the section of Dreissensia gill (fig. 12) and in those of Anodonta (fig. 13, A, B, C). In the drawing of Dreissensia the individual filaments f, f, f are cut across in one lamella at the horizon of an inter-filamentar junction, in the other (lower in the figure) at a point where they are free. The chitinous substance ch is observed to be greatly thickened as compared with what it is in fig. 11, C, tending in fact to obliterate altogether the lumen of the filament. And in Anodonta (fig. 13, C) this obliteration is effected. In Anodonta, besides being thickened, the skeletal substance of the filament develops a specially dense, rod-like body on each side of each filament. Although the structure of the ctenidium is thus highly complicated in Anodonta, it is yet more so in some of the siphonate genera of Lamellibranchs. The filaments take on a secondary grouping, the surface of the lamella being thrown into a series of half-cylindrical ridges, each consisting of ten or twenty filaments; a filament of much greater strength and thickness than the others may be placed between each pair of groups. In Anodonta, as in many other Lamellibranchs, the ova and hatched embryos are carried for a time in the ctenidia or gill apparatus, and in this particular case the space between the two lamellae of the outer gill-plate is that which serves to receive the ova (fig. 13, A). The young are nourished by a substance formed by the cells which cover the spongy interlamellar outgrowths.
Other points in the modification of the typical ctenidium must be noted in order to understand the ctenidium of Anodonta. The axis of each ctenidium, right and left, starts from a point well forward near the labial tentacles, but it is at first only a ridge, and does not project as a free cylindrical axis until the back part of the foot is reached. This is difficult to see in Anodonta, but if the mantle-skirt be entirely cleared away, and if the dependent lamellae which spring from the ctenidial axis be carefully cropped so as to leave the axis itself intact, we obtain the form shown in fig. 15, where g and h are respectively the left and the right ctenidial axes projecting freely beyond the body. In Arca this can be seen with far less trouble, for the filaments are more easily removed than are the consolidated lamellae formed by the filaments of Anodonta, and in Arca the free axes of the ctenidia are large and firm in texture (fig. 9, c, d).
If we were to make a vertical section across the long axis of a Lamellibranch which had the axis of its ctenidium free from its origin onwards, we should find such relations as are shown in the diagram fig. 16, A. The gill axis d is seen lying in the sub-pallial chamber between the foot b and the mantle c. From it depend the gill-filaments or lamellae—formed by united filaments—drawn as black lines f. On the left side these lamellae are represented as having only a small reflected growth, on the right side the reflected ramus or lamella is complete (fr and er). The actual condition in Anodonta at the region where the gills begin anteriorly is shown in fig. 16, B. The axis of the ctenidium is seen to be adherent to, or fused by concrescence with, the body-wall, and moreover on each side the outer lamella of the outer gill-plate is fused to the mantle, whilst the inner lamella of the inner gill-plate is fused to the foot. If we take another section nearer the hinder margin of the foot, we get the arrangement