122 Mr. H. M. Bernard on the 



have travelled. Hence, although one would never have 

 selected a tissue like the ccenenchyma, which all experience 

 shows to be dangerously variable, as a basis for classification, 

 there is in the present case really no choice, i. e. if our mor- 

 phological diagnosis is correct. Hitherto the variations of 

 the surface ccenenchyma — very superficially handled — have 

 been accorded only a secondary place. Dana, deducing Monti- 

 pora [Manopora) from Madrepora by the degeneration of the 

 calicles, classified its species accordingly into those in which the 

 protuberant calicles persisted and those in which they had 

 quite disappeared, the latter group being further subdivided 

 according to the form of the corallum and the character of 

 the surface. Milne-Edwards and Haime divided the Monti- 

 porce primarily according to the form of the corallum. We may 

 at once dismiss this latter classification as purely artificial. 

 Returning, however, to Dana, it must be noted that there 

 is no evidence whatever to make us believe that Monti- 

 j)ora is deducible from Madreptora by gradual degeneration of 

 protuberant calicles. The only Montiporan forms which 

 Dana adduced as transitional hardly support his contention : 

 one — M. gemmidata — has been removed by Verrill to the 

 Turbinarians, while the protuberant calicles in the other — 

 M. caliculata — are not true calicles in Dana's sense, but a 

 peculiar specialization of the interstitial ccenenchyma which 

 will be referred to again. 



We have, then, no choice but to accept the variations in 

 that tissue, the specializations of which are essentially the 

 peculiarity of the genus, as the basis of classification. Be- 

 neath all its baffling superficial variations the laws of its 

 growth can be made out and the main lines along which 

 it has diverged can be traced. This serves to divide the 

 genus into groups which have some claim to be natural. 

 Uncertainty, however, comes in when, in further subdividing 

 these groups, we come within range of the superficial varia- 

 bility due to accidents of position and nutrition. It must 

 therefore be at once confessed that many of the assumed 

 specific variations are not to be relied upon. The " species " 

 established are in many cases only descriptions of individual 

 specimens the surface characters of which give no clue as to 

 their affinities with other specimens. Of course in many 

 cases there are other characters sufficiently striking to justify 

 us in confidently claiming new and distinct types. 



The following analysis of the development of the coenen> 

 chyma was only very gradually arrived at after studying 

 series of sections revealed by fractured specimens. It will be 

 best understood if we reverse the process of its discovery, 



