264 Alfred Sherwood Romer 



because of my own limited work on the fossil evidence as because of the seemingly 

 conclusive nature of the evidence from kidney studies, I find it difficult to accept the 

 conclusion to be drawn from Gross's figures that fishes were entering fresh waters in 

 major numbers for the first time at such a late period in the evolution as the end of the 

 Silurian. Further, I find it difficult on ecological grounds to conceive of circumstances 

 under which a sudden and almost unanimous surge of fishes from the ocean into the 

 streams at the end of the Silurian would be immediately followed by a Devonian 

 reverse migration of nearly as strong an intensity. An inquiry into the nature of the 

 supporting evidence seems warranted. The reader, however, must be warned that this 

 present inquiry lays no claim to the objectivity in treatment which I believe was true 

 of the work in the Romer-Grove paper. Our conclusions there favoured a fresh water 

 origin, but were, of course, far from definitive ; the evidence for this same conclusion 

 based on kidney structure is, however, of so substantial a nature that I am firmly 

 convinced of its validity and hence find it difficult to believe that the palaeontological 

 evidence truly indicates the opposite conclusion. The present discussion is thus, 

 frankly, an attempt to reconcile Gross's presentation of the Silurian fauna with the — 

 to me — more probable thesis that the ancestral vertebrates (and hence presumably 

 those of this period) were in the main fresh water forms.* 



GENERAL CONSIDERATIONS 

 There may be noted here certain of the problems and difficulties encountered in 

 attempting to evaluate the environment of early fishes. Most of the factors concerned 

 (to my anguished regret) increase the difficulty of gathering evidence in favour of a 

 fresh water origin . 



Paucity of early continental sediments 



If we were concerned with Tertiary deposits, a solution might be much more readily 

 attained. From the late Cretaceous onward the fossil record includes not only 

 abundant marine beds but also numerous formations of a definitely continental 

 nature. But, as even an elementary consideration of geologic history would suggest, 

 the continental record becomes increasingly scanty as we travel backwards in time. 

 Period by period, successive cycles of erosion have tended to do away with inland and 

 upland deposits, so that for the older epochs the sediments preserved include very few 

 truly continental beds. The Mesozoic record of fresh water and terrestrial life is a 

 fragmentary one. In the Upper Palaeozoic, extensive continental beds are rare, 

 although they may be preserved in a few relatively stable areas, such as the South 

 African Karroo. Going back to the Lower Palaeozoic, the elapsed time for the 

 operation of diastrophic and erosional forces has been so great that practically no 

 truly continental beds of that age survive today; the record is almost entirely marine- 

 Our knowledge of the possible continental life of the Silurian and older periods is thus 



* It would be unfair to be unduly sceptical of the reality of the " double shift ", salt-to-fresh-to-salt, 

 which Gross's statistics would indicate, unless the data be broken down into its components. Might 

 it be that this seeming anomaly has no real existence, and that the shift in the statistical totals is 

 caused by changes in abundance of various groups and by the appearance of new groups which alter 

 the total complex of the fauna without themselves changing notably in their environment? If the 

 faunal components in Gross's Table IV be considered separately, it will be seen that of the major 

 groups present in the Silurian, the ostracoderms (Agnatha), show no reversal of trend. However, 

 the two remaining Silurian groups, the acanthodians and arthrodires which together form a large 

 percentage of the statistical material, show, in Gross's table, a very sharp reversal. 



