364 DISCOVERY REPORTS 



scuta, which appear to have a rather narrow, unbranched lucida, running straight from 

 the stalk to the opposite border of the blade, i.e. down the middle of the long axis of the 

 scutum and transverse to the long axis of the zooecium. The zooecia are a little shorter 

 and stouter than those of S. frondis and the ovicells are also shorter in proportion to 

 their width and have more numerous pores with raised borders. The frontal avicularia 

 are raised on short columns, those of 5. frondis being sessile, and the cervicorn spine 

 also differs. In Thornely's specimen its main axis is broad and flattened and the tines 

 are more numerous and shorter than in true S. frondis. Finally the vibracular chambers 

 are much larger than those of S. frondis, though similar in shape. I therefore regard 

 Thornely's specimen as representing a distinct species. 



The Tortugas material of S. frondis consists of a number of colonies growing to- 

 gether on weed. In most of them the ancestrula, which is evidently joined to the first 

 zooecium by a very fragile connexion, has broken away, but one was found complete 

 (Fig. 18 C), and another ancestrula which had not yet formed a whole daughter- 

 zooecium was growing on the same weed and evidently also belonged to this species 

 (Fig. 18 D). From Ascension Island (St. 1, 16 November 1925) there are two similar 

 ancestrulae, each with two daughter zooecia and the beginning of a third. The 

 ancestrulae are short, with straight or convex sides and an oblique terminal opesia. 

 They have ten spines surrounding the opesia and no scutum. They are slung by two 

 rootlets, one issuing from each proximal corner of the first zooecium, and the second 

 zooecium has a vibraculum which gives rise to another anchoring rootlet. All the 

 zooecia have scuta and the typical number of spines. The number of tines on the 

 branched spine, which is bifid in the first zooecium, gradually increases, and both the 

 proximal spines gradually become more curved over the aperture. The typical arrange- 

 ment is attained by the end of the second or beginning of the third internode. 



Canda Lamouroux, 18 16 



1. Canda arachnoides Lamouroux. 



Canda arachnoides Lamouroux, 1816, p. 132; 1821, p. 5, pi. lxiv, figs. 19-22; Busk, 1852^, 

 p. 26, pi. xxxiii; 1884, p. 25; Waters, 1887, p. 89; Levinsen, 1909, p. 142; Harmer, 1926, 

 p. 385, text-fig. 17. 



not Canda arachnoides Waters, 1909, p. 165. 



Station distribution. New Zealand: Sts. 934, 935. 



Geographical distribution. Timor (Lamouroux) ; Australia (Busk ; Waters ; Levinsen ; Harmer) ; 

 New Zealand (Waters ; Discovery). 



I have examined Waters 's specimens (kindly lent by the Manchester Museum), and 

 confirmed the Australian and Tasmanian records (1887). The Red Sea specimens 

 (1909) are clearly distinct from Canda arachnoides but not readily to be identified with a 

 described species, though agreeing in many ways with C. caraibica Levinsen. The 

 specimen from Enoshima, near Yokohama, recorded in the same paper, belongs to 

 C. pecten Thornely. 



Waters wrongly quoted Philipps's record of C. retiformis from Brazil as C. arach- 



