TECHNICAL NOTES AND BIBLIOGRAPHY 235 



ject to rapid erosion, and the mud, sand, and gravel pro- 

 duced by their decay were deposited in the river basins 

 at a rapid rate. Periodic exposure, permitting prolonged 

 aeration, oxidized the iron-bearing sediments to produce 

 the typical red color, which is imparted by iron oxide. 

 Such lakes as might appear at flood time in the flatter 

 basins were shallow and inconstant, and during the dry 

 season shrank to stagnant pools or dry mud flats. 



With Chamberlain, Barrell believed that the early 

 Devonian ostracoderms, sharks, and ganoid fishes were 

 confined to the continental rivers and lakes; in the Mid- 

 dle Devonian, competition and increasing aridity drove 

 the sharks into the sea and gave over the domination of 

 the continental waters to the ganoids. In answer to the 

 question of why the ganoids could survive where the 

 sharks could not, Barrell proposed that it had been 

 among these fresh-water fishes, and in direct conse- 

 quence of the periodic contraction of the continental 

 waters, that aerial respiration and its sequel of terrestrial 

 life had been evolved. 



The fresh-water theory failed to attain acceptance, 

 however, or even serious consideration, partly because 

 only a tentative analysis of the fossil record had been 

 presented by its adherents. In 1923 such an analysis was 

 attempted by the botanist MacFarlane {36}, who came 

 to the same conclusion as Chamberlain and Barrell; but 

 his argument suffered from the fact that his opinion 

 could not carry the weight of one experienced in either 

 the nature of geological formations or the invertebrate 

 faunal associations, and lacked the authority requisite 

 to the final interpretation of a question capable of being 

 warmly debated between paleontologists. Moreover, he 

 approached the problem with the conviction that the 

 vertebrates had been evolved from fresh-water nemer- 

 tean worms, and it would appear that in some instances 

 he was led by this a priori bias to interpret the data in 

 a somewhat arbitrary manner. 



In 1930, the writer {46} suggested that the glomerular 



