444 DISCOVERY REPORTS 



prominent, and this calcareous thickening is sometimes continued down the sides of the 

 ovicell. 



In var. elatior the ovicell is flanked by the pointed corners of the zooecium as in var. 

 compacta, but is wider in proportion to the width of the zooecium and there is no 

 marked thickening of the shoulders (Fig. 44 D). 



In var. magna the ovicell occupies nearly the whole width of the zooecium, the distal 

 corners being small and inconspicuous, the lateral walls of the ovicell converge at an 

 obtuse angle and there is no thickening of the shoulder (Figs. 44 E, 45 A, B). 



In var. quadriavicularis the ovicell is as wide as the distal end of the zooecium 

 (Fig. 44 A, B). 



In all four varieties the rosette-plates connecting the ovicell with the distal zooecium 

 are very variable, and the ovicell may be symmetrical or asymmetrical according to the 

 position in which the distal zooecium arises from the fertile zooecium. 



The type material of C. bicornis agrees with Kluge's varieties in the absence of the 

 thickening at the base of the avicularian beak. This thickening is present in some of the 

 long-headed avicularia of C. bicornis var. quadriavicularis. 



The correlated differences in the general build of the colony (cf. pi. X, figs. 1-3), the 

 form of the zooecia (pi. XI, figs. 1-3 and Kluge's figures), the avicularia and the ovicells 

 are such that these forms might well be given specific rank, but, until more is known of 

 typical C. bicornis, I have preferred to leave untouched Kluge's classification of them as 

 varieties of that species. 



Both Kluge (1914, p. 621) and Waters (1904, p. 21) refer to cross-connexions between 

 the branches in C. bicornis. In my material these connexions are found in the older 

 parts of the colonies, and may be numerous, but, except in one specimen of var. quadri- 

 avicularis, they are very irregular in arrangement and the tips of the branches are free. 



Livingstone's material (1928, p. 27) was not in good enough condition for him to 

 determine its relation to Kluge's varieties, and has therefore had to be omitted from 

 my statements of distribution. 



Calvet (1909, p. 8) evidently included at least two species in C. bicornis, as noticed 

 by Kluge (1914, p. 621). Both had secondary branches and the three types of avicularia 

 found in C. bicornis and several other members of the bicornis group. The specimen from 

 Booth- Wandel Island had biserial branches which became tri- or quadriserial at the 

 bifurcations. As the only known members of the bicornis group showing this character 

 are C. bicornis var. quadriavicularis, which has four kinds of avicularia, and C. giganteus, 

 Calvet 's specimen may have belonged to C. giganteus. The specimens from Schollaert 

 Bay had three to ten series of zooecia. This clearly distinguishes them from any of the 

 varieties of C. bicornis and points to C. lewaldi as the only known form with which they 

 agree. A process of elimination thus suggests that Calvet had C. giganteus and C. 

 lewaldi, but there is not positive evidence to prove either suggestion. 



Ancestrulae believed to belong to two of the varieties of C. bicornis are described 

 above (pp. 438, 439), as species 2 and species 7. 



