GENERA OF FAVOSITID^E. 51 



to three lines (PI. I., figs. 4 and 5). The Devonian examples 

 which I should place in this species, possess an equal variability 

 in this respect; and it is therefore clear that the diameter of 

 the corallites even when we look only to the average tubes 

 of a given specimen cannot, in the case of F. Gothlandica, 

 be regarded as a character of specific value. It is further clear 

 that forms said to be specifically distinct from F. Gothlandica 

 must be proved to possess some character of greater constancy 

 and persistence than the mere size of the average corallites, 

 before their claims for separate recognition can be allowed. 



The walls of the corallites of F. Gothlandica usually have 

 the outer surface smooth, or marked with delicate transverse 

 lines, which indicate the position of the tabulae within. In 

 specimens, however, which cannot be otherwise separated from 

 F. Gothlandica (as previously shown by Mr Billings, loc. cit.), 

 it is not unusual to find the flat outer faces of the corallites 

 exhibiting one or two longitudinal lines or striae separating 

 the rows of mural pores. This is the chief or only character of 

 importance which Milne-Edwards and Haime adduce in their 

 definition of F. Troostii, as separating this species from allied 

 forms ; and its occurrence in well-marked Devonian examples 

 of F. Gothlandica as here understood would seem to greatly 

 diminish or altogether destroy its value as a test of specific 

 distinctness. 



The imiral pores of all the typical examples of F. Gothlan- 

 dica, both from the Upper Silurian and Devonian, are arranged 

 in a double row upon each of the prismatic faces of the coral- 

 lites, those of the two rows alternating with one another, and 

 each being separated from its neighbour by a space of about 

 half a line, measured vertically, and surrounded by a slightly 

 raised rim or border (fig. 14, D). These are certainly the 

 general characters of the mural pores in F. Gothlandica, and 

 they would serve as admirable guides to a specific determina- 

 tion if they could be at all constantly determined, or were in 

 themselves constant in their occurrence. In a great many 

 specimens, however, even though apparently excellently pre- 



