508 Mr. H. M. Bernard on the 



borderland between convex and concave, may, according as it 

 is the one or the other, develop either hemispherical masses or 

 the ty])ical cup modifications. In the former case the disk 

 expands on all sides till it covers an enormous area, the centre 

 constantly thickening till it approaches the true glomerate 

 forms, without, however, being really the same ; while in the 

 latter its form-changes may follow, somewhat stiffly, the more 

 immerous and luxuriant growths of the true cups. 



The instability of the cup-shape is hardly to be wondered 

 at. Even though the first ring of buds is horizontal and 

 uniform, it must obviously become increasingly improbable, 

 as the edge of the cup expands, that the radial polyps round 

 the edge should bud regularly enough and uniformly enough 

 to keep the cup symmetrical. Large regular cups a foot in 

 diameter must thus excite our admiration. There is only one 

 really large cup in the National Collection ; it is 16 inches 

 across. 



A point which remains to be established by further research 

 is whether this persistence of the cup-shape is accidental or a 

 normal specific character. It is at present impossible finally to 

 decide this question. For the practical purposes of classifica- 

 tion we are, however, provisionally compelled to assume that 

 it is a reliable character. 



In view, then, of the great improbability that the budding 

 round the edge should be so regular as to keep the cup 

 symmetrical, it is not to be wondered at that in the vast 

 majority of cases the young cup is sooner or later completely 

 obscured by the subsequent growth. The edge begins to fold 

 or frill in various ways ; the folding or frilling becomes more 

 and more complicated as it continues ; the coral-substance 

 continually streams downwards until the early cup is buried 

 up in the ever-thickening base of the enlarging corallum. 



Apparent Periodicity in the Groicth. — Before describing 

 the subsequent forms assumed by the cup which it is so far 

 possible to distinguish, an apparent periodicity in the growth 

 requires to be mentioned. In many corals, as is well known, 

 the living colony, secreting the coral-substance, is progres- 

 sively withdrawn from the older parts of the corallum. The 

 process seems to be uniform and continues as long as the 

 stock lives. The Turbinarians apjyear to differ from this. An 

 old Turbinarian stock is found to consist of many apparently 

 distinct growths. The whole corallum appears to die down 

 periodically, starting into life again along its edges, where 

 growth had temporarily ceased. These new points of growth 

 are not tresh Turbinarians ; they form no stalked cups, but 

 they continue the growth of the old and dead stock. This 



