263 DISCOVERY REPORTS 



to the dorsal side of the primary polyp. A specimen from St. 1644 (total length 15-5 cm., text-fig. 3, 6) 

 shows that new buds may also arise on the dorso-lateral sides of the primary polyp (i.e. also on the 

 axis), ventrally between, or a little below, the lowermost pair of secondary autozooids. 



This development leads to the formation of a cluster like that figured in text-fig. 3, 7 from a 

 specimen with total length 86 cm. (St. 366, no. 5 of the table). Here the lower part of the rachis has 

 a bowl-like shape, and the secondary autozooids are situated along its brim. The axis enters the centre 

 of the bowl at the bottom, but from here it follows the dorsal side of the rachis in a deep and narrow 

 furrow to the brim between two dorsal secondary autozooids, bends inwards and ends in the basi- 

 dorsal wall of the primary polyp, which has been displaced into the central field by the growth of the 

 whorl of secondary autozooids. 



oOo o*o oOq 



000 1?, o 9 n 



o o 



Fig. 3. Unihellula lindahli. Diagrams showing the arrangements of autozooids in clusters with comparatively few polyps. 

 Where distinguishable the primary polyp is hatched, i. Type specimen E of Umbellula carpenteii KoUiker, 1880. 2. Kolliker's 

 specimen D of the same species. 3. Type specimen of Umbellula magniflora Koiliker, 1880. 4. Two small specimens from 

 Discovery 'At. 181 (nos. i and 2 of the table, p. 264). ^. Discovery 'Stt. 181, no. 3 (p. 265). 6. The specimen irom Discovery At. 1644 

 (p. 267). 7. Discovery St. 366, specimen no. 5 (p. 265). 



In other specimens this dorsal furrow has disappeared because of the growth of the rachis-tissues 

 and can only be made out by dissection. Here we might speak of typical magniflora specimens, 

 when only some few further autozooids have developed in the central field. The type specimen of 

 Kolliker's (1880) magniflora (text-fig. 3, 3) evidently shows the commencement of the development 

 of such clusters. 



On the other hand, we cannot look on this development of the cluster as absolutely characteristic of 

 the species. In several cases the primary polyp and the axis retain their peripheral dorsal position, and 

 the entire cluster in the main keeps its predominantly ventral position on the rachis, with generally 



