UMBELLULIDAE 269 



also a more or less distinct bilateral structure. In these specimens the rachis below the autozooids 

 generally retains a more slender and obconical shape in preserved specimens, although we also find 

 stages, where the rachis is more fleshy, or less strongly contracted. This development leads to the 

 formation of polyp-clusters like those of the largest specimens from the Swedish Deep-Sea Expedition 

 (Broch 1957, text-fig. 4, VII). In that paper the possibility was indicated, that such clusters with 

 two whorls of autozooids may give the final and normal arrangement of the species, as found in 

 specimens with total heights of 105 and 119 cm. This, however, is not the case in antarctic waters. 

 In the magniflora type especially, there is a tendency for several autozooids to develop in the central 

 field within the first whorl of secondary autozooids. Great crowding results and arrangement in 

 whorls is soon lost. At the same time the inner part of the ' bowl ' develops into a spigot carrying 

 autozooids on all sides. In typical antarctica forms, like those from St. 1948 (see also PI. 2, fig. 6), 

 these features are distinct, and in the largest of them (total height 38-5 cm.) new autozooids are 

 budded everywhere between the old ones. Evidently the number of autozooids in the cluster in some 

 way or other depends on local conditions rather than on hereditary fixed characters. On the other 

 hand, collections like those from Stns. 366 and 371 indicate that specimens may develop divergently 

 in one locality. Great individual variations can occur under apparently uniform conditions. 



No correlation exists between total length and number of autozooids in the specimens, when they 

 have passed their first stages of development. The specimen described by Kiikenthal and Broch (191 1) 

 as the type of U. antarctica was rather short (total length 465 mm., a cotype 460 mm.); on the other 

 hand, its 'stalk' (both holdfast and stem) is remarkably fleshy. Although this is the case in many 

 specimens of typical antarctica forms, comparatively slender specimens with very thin coenosarc on 

 stem and holdfast, but with typical antarctica clusters, also occur fairly frequently. In most cases 

 such slender specimens have fewer autozooids in their clusters, which vary from the bilateral car- 

 penteri to the radial magniflora type. The collections corroborate the suggestion of Hickson (1907) and 

 of Jungersen (1907) that all antarctic Umbellulae with square axes and without spicules (apart from 

 the calcareous concretions in the inner tissues of the terminal part of the holdfast) belong to one very 

 variable species. 



A study of the descriptions of other non-spiculate species of UmbeUula with square axes shows that 

 it is exceedingly difficult to trace differences, which might serve as reliable fundamenta divisionis. 

 Many so-called characters are due to contraction ; others to a certain degree are apparently dependent 

 on local conditions. Polyp clusters have a tendency to develop more autozooids in subantarctic- 

 antarctic as well as in subarctic-arctic seas, i.e. in colder areas, than in temperate and tropical waters. 

 One might suggest that this is not only due to lower temperature, but also to an admixture of water 

 from melting ice, i.e. to variations in salinity. In these colder seas, UmbeUula is also found at much 

 shallower depths than elsewhere. It is accordingly necessary to be most cautious in using the numbers 

 of autozooids in the clusters for specific determination, because these numbers show no correlation 

 with the length of the specimens (Broch 1957). 



It has been mentioned earlier that Kiikenthal was at first inclined to look on U. antarctica as a 

 variety of U. encrinus. Later on he (191 5) reduced U. ambigua Fischer (also retained by Marion (1906) 

 as a separate species) to a variety of U. encrinus. This course was already indicated by Jane Stephens 

 (1907), who had no less than 17 specimens at her disposal, the stems of which (from the clusters down- 

 wards) ranged from 1 12 to 420 mm. (Marion's specimen measured in the same way was about 350 mm.). 

 The numbers of autozooids in the clusters varied from 6 to 43, and the polyp bodies from 8 to 30 mm. 

 Marion's drawings (1906, PI. 15, fig. 22, and 22 A) might well have been made from a typical specimen 

 of U. magniflora Kolliker, and in his text Marion (1906, p. 146) writes: 'V UmbeUula ambigua se 

 rapproche intimement de V UmbeUula grandiflora Kolliker prise dans I'Extreme-Sud, aux iles 



