ANTHOMEDUSAE 13 



occurs in Haeckel's specimen from the West Atlantic, although Haeckel himself failed to observe it. 

 In each of the four corners of the stomach of the medusa, adjacent to the bases of the radial canals, 

 a typical, well-developed hydranth, mounted on a short pedicel, carries medusa buds as lateral out- 

 growths with fairly long pedicels (PI. Ill, figs. 2-5). The hydranths are fairly broad, each provided with 

 a circle of 10-12 tentacles and a circular mouth opening. They are also able to catch food, as seen in 

 PI. Ill, fig. 5, which shows a hydranth grotesquely expanded by a copepod which it has swallowed. 



(2) In one of the specimens from St. 1581, off Zanzibar (PI. Ill, fig. 6), no hydranths can be 

 observed, but from the corners of the stomach filiform structures have grown outwards, adnate to 

 the epithelium of the subumbrella, exactly like the stolon of a hydroid attached to a solid substratum. 

 They are more or less branched and give rise to pedicels with medusa buds. In the specimen figured, 

 the stolon carries two well-developed medusa buds ; the stolon is adnate up to a point a little beyond 

 the origin of the medusa buds, its terminal end being slightly elevated, the internal structure showing 

 that it is in the act of developing into a medusa. The small outgrowth on the right is in a juvenile, 

 unmodified condition ; it has no terminal cluster of nematocysts and therefore cannot be regarded as 

 a blastostyle. 



(3) Medusa buds situated on the walls of the stomach can be seen in the six specimens from 

 Brazilian waters, three collected in April (St. 680) and three in October (St. 709). Some of the 

 medusa buds are solitary, issuing directly from the stomach wall as in B. niobe, but most of them are 

 collected in clusters on branched pedicels. Hydranths are not developed in these clusters, but the 

 very fact that the pedicels of the medusa buds may be branched seems to indicate that they may be 

 regarded as highly reduced stolonial or polypoid structures. This is the only method of budding 

 found in the Brazilian medusae as well as in Haeckel's specimen (described under (i)) in which 

 hydranths are also developed. It also occurs in the medusa from St. 1581, off Zanzibar (described 

 under (2)), in which some of the medusa buds are also borne on stolonial outgrowths from the 

 corners of the stomach. 



Although examples of propagation by budding have previously been reported in numerous genera 

 of medusae, the only instances at all comparable with Bougainvillia platygaster occur in the limno- 

 medusa, Proboscidactyla ornata, and in the leptomedusa, Phialidium mccradyi. 



Proboscidactyla ornata is a species common to all tropical coasts and in it medusa buds are found on 

 the radial canals. The species has been divided into a number of varieties, or even species and sub- 

 species, based on the position of these buds, which may be either adjacent to the corners of the 

 stomach, or else arise on the first, second or third forking of the radial canals. In every case the 

 medusa buds do not arise directly from the tissues of the radial canals but from a blastostyle, which is 

 a polypoid structure. These buds were first seen by T. Huxley (1877, p. 132, fig. 17) in an unidentified 

 ' Willia' {Proboscidactyla). He described them as issuing from small 'stolons', each terminating in a 

 'knobbed extremity containing many nematocysts'. Huxley's observations were quoted by Browne 

 (1904, p. 727). In a later paper Browne (1916, p. 184) called Huxley's 'stolons' 'blastostyles', a 

 decidedly better term. The occurrence of these structures has been reported subsequently by several 

 other authors, and in 1951, while in the Philippines with the Danish ' Galathea ' Expedition, I found 

 numerous specimens of Proboscidactyla ornata with medusiferous blastostyles in every possible posi- 

 tion on the canals. PI. Ill, fig. 7 shows one of these structures which I have observed. It is certainly 

 a blastostyle, a polypoid individual corresponding to the blastostyles on which the gonophores are 

 developed in many hydroid colonies. 



In Phialidium mccradyi also, medusiferous blastostyles have been found to occur (Brooks, see Mayer, 

 1910, p. 271, PI. 34, figs. 2, 3, PI. 35, figs. 1-3), but in this instance they arise from the immature 

 gonads and are enclosed in gonothecae. 



