162 R. G. Ccirruthers — A liermon of some Carboniferous Corals. 



thickness, and as they rarely reach the rim of the calyx they may at 

 first sight seem to be absent. This may be termed the cornu-hovis 

 phase (ex. Caninia '■ cornu-lovis^^ Mich.), and with it the develop- 

 ment of the corallum is completed. 



As above noted, the minor septa tisually arise at the conclusion of 

 the dumonti phase, but their appearance may be delayed to a very late 

 stage, and thej- are sometimes hardly visible. They are best 

 developed in broad, widely expanded specimens (e.g. Diagram E, 

 Fig. 1), but even here they remain rudimentary. 



The (cardinal) fossula, whose varying characters are referred to in 

 the foregoing paragraphs, typically lies on the convex side of the 

 corallum, but is often laterally disposed. 



But, save for the appearance of the dissej)imental ring (a feature 

 confined to the mature part of the corallum), none of the phases above 

 described are truly ontogenetic, for, as will now be explained, they 

 are intimately connected with the habits of growth of the corallum. 



Prom an examination of many broken specimens which are devoid 

 of an infilling of foreign matter, and clearly show the internal 

 structures, two factors controlling septal development are found. 



First, it is seen that as the coral becomes cylindrical, the septa die 

 away beneath each tabula and are rebuilt upon its upper surface 

 (see Diagram F, Fig. 1, A); i.e. they become amplexoid in 

 character. But the cylindrical habit may be acquired at any period 

 of growth (e.g. Diagram E, Fig. 3), and consequently amplexoid 

 septa, characteristic of the nystiana and succeeding phases, may occur 

 at any time. It often happens that examples showing the long septa 

 of the dumonti phase are found to have amplexoid septa in their 

 lower portions. 



There is, however, a second factor influencing septal development. 

 It will be noticed from the vertical sections (PL VI, Figs. Ih and \i, 

 and Diagram P, Figs. 2 and 3) that the tabulae are by no means 

 equidistant, and from this fact one may infer that the growth of 

 the coral varied somewhat in rapidity from time to time. Hence, 

 although septal information was slower in the cylindrical part of the 

 coral as opposed to the conical part, the development of the septa seen 

 in the calyx, and the depth of the latter, are also dependent on the 

 rapidity of growth between the formation of the last tabula and the 

 death of the animal (Diagram P, Fig. 1, B). 



The vertical sections (PI. YI, Fig. Ih, and Diag. F, Pigs. 2 

 and 3) also show that the depression of the tabulae in the cardinal 

 fossula, though usually considerable, is by no means constant in 

 amount. On the whole the depression lessens as the cylindrical 

 habit is assumed, but the sections show that the size of the fossula 

 in the calyx must be by no means a constant feature. 



Conversely to the connection of amplexoid septa with a cylindrical 

 habit, it may be expected that specimens showing continuous expansion 

 of the coral should also show the long septa of the dumonti phase, 

 even at a relatively late period of growth ; this is indeed the case, 

 and is well exemplified in the broad form of the coral (Diag. E, 



