1907.] 



THE SCALES OF FISH. 



75& 



as cosmine ; but tlie resemblance to the typical cosmine of Megcd- 

 ichthys, or even to the cosmine of Palaeoniscids, is not at all close, 

 and the two structures are probably not homologous. Since in 

 the lepidosteoid scale there are neither pulp-cavities, nor vascular 

 networks giving rise to canaliculi, it seems advisable not to apply 

 to them the name cosmine at all. The tubules, with their inner 

 branching ends, may very well merely represent modified bone- 

 cells, which, instead of being buried in the matrix they produce, 

 get carried outwards further and further from their first position 

 as the scale grows older. They do not all start from the same 

 region. Only the oldest tu.bules reach the central parts ; younger 

 ones start at various points among the later foi"med laminfe. 

 Occasionally the tixbules seem to traverse the ganoine in its outer- 

 and thinner region ; but as a rule they either do not run upwards 

 to the exposed surface or they get cut off by the newly deposited 

 layers of ganoine, each of which of course extends a little further 

 than the last (text-fig. 199). 



Text-fig. 199, 



(From Lankester's ' Treatise on Zoology,' by pennission of Messrs. A. & C. Black.) 



Much enlarged view of a piece of the scale of Lepidosteus osseus L. d., suijerficial 

 denticles ; g., ganoine layer ; ?"., inner bony layers, or isopedine ; t., tubules with 

 branching inner ends ; vc, vascular canal. 



We have seen what are the three chief kinds of scales commonly 

 called ganoid : to the first it is proposed to give the name cosmoid, 

 while the second and third are varieties of the true ganoid scale. 

 Other and less important varieties exist and Avill be dealt with 

 later on (p. 765). 



Origin of the Cosmoid and Ganoid Scales. 



According to Williamson's theory, which has already been 

 mentioned, the cosmoid scale arose by the fusion of a large number 

 of denticles, and their combination with a bony plate developed 

 below. In fact the theoiy which we have just found to appl}^ so 

 admirably to the explanation of the plates of the Heterostraci,. 



