1. (1) There are three sections of the body: (a) the preoral lobe, (b) the postoral section (collar) which carries tentacles, and (c) the posterior or anal section (trunk); (2) the preoral lobe probably represents the umbrella of the trochophore; (3) the cilia at its margin probably correspond to the preoral ciliated band (Trochus), while (4) the postoral ciliated zone (collar) which carries the tentacles probably corresponds to the postoral band (Cingulum) of the trochophore; (5) this postoral ciliated zone (Cingulum) runs obliquely around the body, being further posterior on the ventral than on the dorsal side; (6) tentacles appear near the ventral midline and fresh pairs are added dorsally.

II. (7) There is an epithelial nervous system, (8) an apical plate, ganglion and (in some species) eye spots, and (9) an oesophageal commissure.

III. (10) The cœlom is composed of an anterior unpaired cavity and two pairs of cavities posterior to this (Masterman);1 (11) the anterior cœlom sac arises as an enterocœl, the posterior paired ones as schizocols; (12) there is one pair of protonephridia, which end blindly internally in connection with excretory cells.

IV. (13) There is a ventral invagination posterior to the zone of tentacles and a peculiar metamorphosis by the evagination of this invagination; (14) during metamorphosis the tentacles turn forward, and (15) the anus comes to lie on the dorsal side of the mouth, the intestine thus forming a loop.

Comparing now the larvae of brachiopods with the Actinotrocha we find that, in addition to the general resemblances to the trochophore which both show, there are the following special resemblances between the two: (1) In both brachiopods and Actinotrocha the postoral ciliated zone (Cingulum) is greatly enlarged and runs obliquely around the body, being farther posterior on the ventral than on the dorsal side. (2) In both cases this forms the mantle or lophophore, though the tentacles or cirri which are borne upon it appear much earlier in Actinotrocha than in the brachiopod larva. (3) Brooks has shown that in Lingula the ventral pair of

See postscript, p. 70.

tentacles appears first and that successive pairs of tentacles are added dorsally, exactly as in Actinotrocha. (4) In the metamorphosis the mantle (lophophore) is turned forward over the preoral lobe in exactly the same way in both cases. These are extremely important resemblances, and in themselves lend support to the view that Phoronis and the Brachiopoda are closely related.' On the other hand, according to Masterman's ('97 and 1900) recent work on Actinotrocha, there are certain important respects in which Actinotrocha differs decidedly from the brachiopod larva : (1) The cœlom consists of an anterior unpaired cavity and of two pairs of cavities, one of which lies in the lophophore and the other in the trunk region. The anterior unpaired cavity somewhat resembles in position and method of origin the anterior portion of the enterocol of Terebratulina, but the lophophoral and trunk cavities of Actinotrocha differ from the mantle and peduncular cœlom of Terebratulina in that the latter are a part of the enterocœl and are never completely separated from one another, whereas in Actinotrocha they arise as schizocols and are always separate. (2) Actinotrocha also has rudiments, at least of a second pair of nephridia. (3) It also has two endodermal outgrowths from the anterior portion of the enteron, which are composed of large vacuolated cells, and are homologized by Masterman with the notochord of the Hemichorda.'

I have had no opportunity of studying the later stages in the development of the brachiopod, in which alone the two last-mentioned structures might be looked for, and cannot therefore determine whether there are real differences between the brachiopod and Phoronis in these respects. With regard to the differences shown by the cœlom, one must bear in mind the fact that in the brachiopod larva the cœlom almost entirely disappears, except in the mantle, and a segmentation of the coelom in later stages could not therefore be observed, even if it had at one time existed in the ancestors of the brachiopods. There can be no doubt however that in Terebratulina the entire cœlom arises from a single enterocœl, in which respect there is a decided difference between the brachiopod and Phoronis. The resemblances mentioned above

1 The presence of "plasmic corpuscles" (Ideka, 1901) in the blastocol of both forms is another interesting resemblance (see p. 45).

2 See postscript, p. 70.

PROC. AMER. PHILOS. Soc. XLI. 168. E. PRINTED MAY 6, 1902.

however are so important and extend to such details that I am inclined to accept the view that Phoronis and the Brachiopoda are related, and to look to future work on the development of both of these groups to harmonize the apparent differences between them. (c) Comparison with Larval Polyzoa.

Brooks in particular has emphasized the resemblance between the larvæ of Polyzoa and Brachiopoda, basing this comparison, however, rather upon the external characters in which both resemble the trochophore than upon a detailed comparison of internal

structure.

Ectoprocta. It is extremely difficult to compare larval brachiopods with larval ectoprocts, owing to the great variety of forms presented by the latter, their many secondary characters, and the conflicting accounts of their structures and homologies which have been given by various authors. There is some reason for believing however that the ectoproct larva belongs to the trochophore type, and that the following parts of the two may be homologous: (1) The retractile disk may correspond (at least in part) to the apical plate, (2) the corona in part to the trochus, (3) the sucker to the trunk of the trochophore, or to the ventral evagination of Actinotrocha. Furthermore one may trace a certain resemblance between the invaginated sucker of Bugula and Lepralia and the peduncle and mantle of Terebratulina. In both cases attachment takes place by the peduncle, while the covering folds (mantle in the case of brachiopods) are turned forward as the peduncle is protruded. However the degeneration and modification of structures, both in the larval stages and in the metamorphosis, are so extreme that any attempt at the present time to trace homologies between larval Ectoprocta and other forms must be accompanied by a lively imagination and a ready facility in guessing.

There is good evidence in the degeneration of the intestine and cœlom of the ectoproct larva, and in the general degeneration which accompanies its metamorphosis, that we are dealing with a highly modified type of development, which is little likely to throw light upon the affinities of the Polyzoa. However the resemblances between the adult Polyzoa and Phoronis and the Brachiopoda is such as to warrant the conclusion that these groups are at least remotely related to one another.

Entoprocta.-Among larval entoprocts there are few, if any, undoubted homologies with either the trochophore, the actinotroch,

or the brachiopod larva. It is possible that the ciliated disk of Pedicellina and Loxosoma is homologous with the retractile disk of the ectoproct larva and with the apical plate of the trochophore, and that the margin of the vestibule (ciliated ring) in the former corresponds to the trochus of the latter, but these possible homologies are too hypothetical to be affirmed with any degree of

5. Conclusions.-Neglecting the older views as to the affinities of the brachiopods with lamellibranchiate Mollusca, which were founded merely upon superficial resemblances, we find that within recent times the brachiopods have been associated, at different times and by different authors, with Chatopoda, Polyzoa, Chatognatha and Phoronis.

Both Morse ('73) and Kowalevsky ('74) independently reached the conclusion that the brachiopods are chatopod annelids. Morse says in summing up his work on the subject ('73', p. 57): "We must regard the brachiopods as ancient cephalized chatopods, while Serpula, Amphitrite, Sabella, Protula and others may be regarded as modern (later) cephalized chatopods"; and Kowalevsky ('74) maintained that the brachiopods ought to be considered simply as an order of the annelids, which present at least as many resemblances to the chatopods as do the leeches.

Morse has enumerated twenty-four characteristics in which brachiopods resemble more or less closely Vermes, sedentary annelids and Gephyreans. Kowalevsky also names a considerable number of points in which brachiopods resemble chatopods. Some of these features are not actually characteristic of the brachiopods, as, for example, the segmentation of the larva; others are of such a general character as to apply to almost all Bilateralia, as Brooks has shown, while still others represent real resemblances between the brachiopod larva and the trochophore. The trochophore larva however is of such wide occurrence among bilateral animals, that the mere classification of the brachiopods among the Trochozoa throws no light upon the nearer affinities of this group.

Huxley, Lankester, Claus and others have regarded the brachiopods as more or less closely related to the Polyzoa, and Brooks in particular has held that the two groups belong to the same phylum and class. "The organization of the Lingula larva," he says, "shows that it is not merely like a Polyzoon, but that it actually is one; as much so as the hydra stage of an Hydro-Medusa is a

Hydra, or the tailed larva of Botryllus is an Appendicularia, and more so than a tadpole is an urodellan Batrachian." This close relationship he bases largely upon the external resemblances between the larvæ of Thecidium and various Polyzoon larvæ. It seems to me that some of these resemblances are real homologies, but on the other hand the differences between these larvæ, as well as between the adults of these two groups, are so great that it would be inadvisable to place them together in the same class; though I believe they should be placed in the same phylum. Moreover it seems to me that Brooks' view, that the Polyzoa are the ancestral form of which the Brachiopoda are a specialization, is just the reverse of the real relationship; larval as well as adult Brachiopoda show less specialization and certainly less degeneration than the Polyzoa.

The resemblances of the brachiopod larva to the Molluscan veliger, upon which Brooks lays emphasis, are in the main the same as the resemblances to the trochophore, the veliger and trochophore belonging to the same type of larva.

The idea that the brachiopods are related to the chatognaths, which was suggested by Bütschli and Hertwig ('80) and maintained by van Bemmelen ('83), has little more in its favor than the supposed resemblance in the method of formation of the cœlom and in certain histological details.

So far as the formation of the coelom is concerned, I have already pointed out the fact that in Terebratulina it forms in a very different manner from what obtains in Sagitta, and as for the histological resemblances they are by no means confined to the two groups in question. On the other hand there are so many important differences between the two groups, both in their embryology and in their adult structure, that one could as well maintain the affinity of the Brachiopoda with Echinodermata, Enteropneusta or Chordata, as with Chaetognatha.

Caldwell ('82) first pointed out in detail the resemblances between Phoronis and the Brachiopoda. In this paper he has urged "an entirely new view of the homologies of the body surfaces in Brachiopoda." He regards the Brachiopoda as fixed by their ventral surface, and both valves of the shell as ventral in position, the peduncle of the brachiopod corresponding to the ventral invagination of Actinotrocha. While there are some facts which may be urged in favor of this view there are many which may be used