UMBELLULIDAE 257 



also been obscured by the larger clusters, the tightly packed autozooids being grouped on an extremely 

 short terminal part of the rachis. 



In these two best preserved specimens, siphonozooids were observed everywhere on the rachis, 

 both between the autozooids and on the rapidly tapering part below the cluster. It was impossible 

 to find any siphonozooids on the extremely thin upper part of the stem as well as on the holdfast. 



The axis is extremely thin. In its upper part, immediately below the cluster, its diameter is 

 scarcely 0-5 mm., nor does it increase until the middle of the stem is reached; from here downwards 

 the axis slowly grows stouter and attains its greatest diameter — about i mm. — in the middle of the 

 holdfast, tapering again towards the basal end. This entirely agrees with Marshall's observations, and 

 with his remark that ' for the greater part of its length the stalk is extremely flexible, so much so that it 

 can readily be coiled in circles of 5 mm. diameter without the slightest danger of breaking '. Although 

 in live specimens the flexibility of the axis in most other deep-living, slender pennatularians is also 

 great, such an extreme flexibility like that of U. huxleyi is very seldom encountered. It must be cor- 

 related with the extraordinarily scanty calcareous incrustation. On the other hand, the species has a 

 very efficient holdfast, as shown by the photographs (PI. i, fig. i). These two characters explain why 

 the species has been reported from remarkably few localities, although its habitat, as collections show, 

 is very extended and shallower than those of most other species of the genus. 



Hickson's (1937) U. huxleyi with its rigid and quadrangular axis belongs to U.pelhicida (see p. 261). 



U. huxleyi is characterized by its obviously concentrated, pompon-like cluster, whereas the second 

 spicule-free species spicata with a round axis is distinguished by the long, spaced tassel of very slender, 

 long autozooids. 



(Lectotype: British Museum (Nat. Hist.). Reg. No. 1881. 2. 11. 25. Specimen D of Kolliker, 



1880.) 



Umbellula spicata Kiikenthal 1902 (PI. i, fig. 2) 



U. spicata + U. valdkiae Kiikenthal 1902, p. 594. 



W. elongata+U. radiata+U. pendula Thomson & Henderson 1906, pp. 96, 98, 99, Pi. 7, fig. 6. 



U. spicata+U. valdiviae Kiikenthal & Broch 191 1, pp. 294, 295, PL 16, figs. 17-19, PI. 20, fig. 510, b; PI. 21, 



figs. 52, 53. 

 U. spicata+ U. valdkiae+ U. elongata (?)+ t/. radiata (?)+ t/'. pendula (?) Kukenthal 1915, pp. 53, 55, 57, fig- 60. 



According to Kukenthal and Broch (191 1), U. spicata and U. valdiviae differ as follows: U. val- 

 diviae has a weak, more slender stem (this is not perceivable in the excellent figures). The rachis 

 is not so contracted ('viel dicker') in U. valdiviae as in U. spicata. In U. valdiviae, the tentacle 

 pinnulae alternate in size, whereas they increase successively in length towards the end of the tentacle 

 in U. spicata. Neither of these features, which obviously are results of contraction, nor the dif- 

 ferences in the arrangement of the siphonozooids mentioned in the detailed description, furnish a 

 sufficient fimdamentiim divisionis. It must also be emphasized here that none of the specimens from 

 the ' Valdivia ' Expedition were adult. 



In the collections at the British Museum (Nat. Hist.), an unrecorded specimen is labelled ' 1914. 

 I. 27. I. Between Aden & Bombay, 13° N, 76° E, 600-700 fthms. J. E. Purton.' The specimen is 

 broken, the lower part of the stem with the holdfast is in three pieces, whereas the upper part of the 

 stem with the rachis and the large autozooid-tassel is comparatively very well preserved (PI. i, fig. 2). 

 If these four fragments represent the entire specimen, its total length was about 55 cm., including 

 the 9-5 cm. long rachis with its cluster of autozooids. In the lower part of the rachis the autozooids 

 are placed in a couple of very indistinct whorls round the stem, but in the upper two-thirds or 

 more, no regularity in the arrangement of the autozooids can be traced. These are separated by 



