[Illustration: FIG. 84.—Transverse section through the body-wall of _Grantiopsis_. _d.o_, Dermal ostium; _fl.ch_, flagellated chamber; _i.c_, long incurrent canal traversing the thick cortex to reach the chamber layer; _p_, apopyle. (After Dendy.)] [Illustration: FIG. 85.—Transverse section through the body-wall of _Leucilla_. _d.o_, Dermal ostium; _ex.c_, exhalant canal; _fl.ch_, chamber; _i.c_, inhalant canal. (After Dendy.)] Above are described three main types of canal system—that of Homocoela, of _Sycon_, and of _Leucandra_ and _Leucilla_. These are conveniently termed the first, second, and third types respectively, and may be briefly described as related to one another somewhat in the same way as a scape, umbel, and compound umbel among {192}inflorescences. These types formed the basis of Haeckel's famous classification.[221] It has, however, been concluded[222] that the skeleton is a safer guide in taxonomy, at any rate for the smaller subdivisions; and in modern classifications genera with canal systems of the third type will be found distributed among various families; while in the Grantiidae, _Ute_ and _Leucandra_ stand side by side. This treatment implies a belief that the third type of canal system has been independently and repeatedly evolved within the Calcarea—an example of a phenomenon, homoplasy, strikingly displayed throughout the group. It is, remarkably enough, the case that all the canal systems found in the remainder of the Porifera are more or less modified forms of one or other of the second two types of canal system above described. The families Grantiidae, Heteropidae, and Amphoriscidae, all possessing a dermal cortex, are distinguished as follows:—The GRANTIIDAE by the absence of subdermal sagittal triradiate spicules and of conspicuous subgastral quadriradiates; the HETEROPIDAE by the presence of sagittal triradiates; the AMPHORISCIDAE by the presence of conspicuous subgastral quadriradiates. Two families of Calcarea, possibly allied, remain for special mention—the Pharetronidae, a family rich in genera, and containing almost all the fossil forms of the group, and the Astroscleridae. The PHARETRONIDAE are with one, or perhaps two exceptions, fossil forms, having in common the arrangement of the spicules of their main skeletal framework in fibres. The family is divided into two sub-families:— I. DIALYTINAE.—The spicules are not fused to one another; the exact mode of their union into fibres is unknown, but an organic cement may be present.