of the strobila are sexually matured or maturing individuals which, as medusæ, are fitted for swimming about, multiplying, and reaching other habitats; while each detached proglottis of the cestoid is, by the nature of its medium, limited to creeping about. Clearly this fissiparous process in such Annelids as the _Syllidæ_, which has similarly been compared to the strobilization of the _Scyphomedusæ_, differs simply in the respect that single segments are not adapted for locomotion, and it therefore profits the species to separate in groups. All these facts and analogies point to the conclusion that the remote ancestor of the Annelids was an unsegmented creature homologous with each of the segments of an existing Annelid. This conclusion is supported by other kinds of evidence here to be added. The larvæ of Annelids are very various; but amid their differences there is a recognizable type. “The Trochophore is the typical larval form of the Annelid stem”: a trochophore being a curious spheroidal ciliated structure suggestive of cœlenterate affinities. And this unsegmented larva, representing the remote ancestor from which the many Annelid types diverged, is similar to the larvæ of the _Rotifera_ and the _Mollusca_: a trochophore is common to all these great classes. Moreover since, among the _Rhizota_ (a sub-class of the _Rotiferæ_), there is a species, _Trochosphæra_, solitary and free-swimming, resembling in form and structure a trochophore, though it is not a larva but an adult, we get further evidence that there was a primitive creature of this general character, of which the trochophores of _Mollusca_, _Rotifera_, and _Annelida_ are divergent modifications, and which was unsegmented: the implication being that the segmentation of the _Annelida_ was superinduced. That this segmentation resulted from gemmation is implied by what are called polytrochal larvæ. These “sometimes appear as a stage succeeding other larval types. Thus those of _Arenicola marina_ arise from larvæ which at first were monotrochal, later became telotrochal, and finally, by the appearance of new ciliated rings between those already present, assumed the stage of polytrochal larvæ.... This condition warrants the assumption that the segmented forms are to be looked upon as the younger, the unsegmented, on the other hand, as the phylogenetically older.” (Korschelt and Heider, i, 278.) And that the above-described rings of cilia mark off segments is shown by the case of _Ophryotrocha puerilis_, which “remains, as it were, in a larval condition, since the segments retain their ciliation throughout life.” (_Ib._, 277.) Yet one more significant fact must be named. In early stages of development each segment of an archiannelidan has cœlomic spaces separate from those of neighbouring segments, but in the adult the septa “generally break down either partially or completely, so that the peri-visceral cavity becomes a continuous space from end to end of the animal.” (Sedgwick,