Madreporarian Siihfamihj Jrontipori'nfe. 123 



and, assuming our conclusions to be correct, start from the 

 hypothetical parent-polyp of the genus. 



This polyji, as we have seen, difFered from that of 

 Madirpora, Turhinaria, and Astrcvopora in that the porous 

 tlu'ca did not rise up into a cone, but was low and thick, 

 filling up, but probably not much overtopping, the rim of the 

 epithecal saucer. Primarily the porous walls consisted of 

 radial laminae joined together by synapticulw *, so that the 

 coenenchyma surrounding this parent polyp may be said to 

 radiate outwards towards the edge of the epitheea. When 

 the first ring of buds ap])eared just within the edge of the 

 epitheea their walls would also radiate outwards, either con- 

 tinuing to be supported by an extension of the primitive 

 epitheea or shooting out freely beyond its edge. From this 

 initial stage in the development of Montipora we should 

 expect to find the coenenchyma consisting of laminate plates 

 standing at right angles to the epitheea and radiating out- 

 wards on all sides. We should expect to find this because 

 the coenenchyma is nothing but the resultant of fusion of the 

 porous walls of the component polyps, and their laminate 

 costje would necessarily be arranged in the manner described. 



Now this initial stage in the growth of the Montiporan 

 corallum is traceable in almost every type. A surface of 

 fracture through any exjdanate Montiporan will almost inva- 

 riably reveal a thin basal layer streaming outwards towards 

 the growing c(\^Q. While this basal "streaming layer" 

 is of fairlv uniform thickness and the direction of its fibres 

 is always outwards — i. e. in the line of growth — its texture 

 may vary. (I) It may be composed of ribbon-like bands 

 running outwards more or less at right angles to the epitheea, 

 but so united as to form a system of flat canals apparently 

 freely communicating with one another. This laminate reti- 

 culum, occurring as it invariably does in the "streaming layer," 

 may, 1 think, safely be regarded as a vestige of the primitive 

 laminate costee which were once the most important element 

 in the calicle walls. (2) The primitive band-reticulum may, 

 owing to the extensive perforation of the laminae, have lost 

 this character and have become a filamentous reticulum. In 

 this case also the direction of the threads is typically very 

 pronounced, streaming outwards towards the growing edge. 

 There are, lastly, a few cases in which the reticulum shows 

 no special streaming ; these would appear to have been 

 secondarily modified. Here it should be remarked that only 



* I use this term generallj* for all outgro'wi;lis from the faces of septa 

 for their mutual support, irrespective of their form and position. 



