[BARNES] ANCHOR-ICE FORMATION 79 



is sometimes a great increase, to which a new frost succeeds, so that 

 there is in the river, but especially near the banks, two layers of ice 

 superimposed at a distance; the one at the height of the first level of 

 the water, the other at the height which this level has attained on the 

 rise of the water. This theory, which refers to a peculiar case, does 

 not explain, in any point of view, the observations just made, and in 

 which natural philosophers have actually seen ice formed on the surface 

 of pebbles placed at the bottom of tbe water in the beds of certain 

 rivers. 



"We now come to jMr. ]\tcKeever, who, confining himself closely to 

 the most subtle principles of the theory of heat, has not, on this account, 

 been more fortunate than his predecessors. 



According to this author, the rocks, stones, and gravel which gen- 

 erally cover the bottom of rivers, have powers of radiation superior to 

 those of mud, perhaps on account of their peculiar nature, but chiefly 

 because they have rough surfaces. Thus rocks, in large or small masses, 

 will become much cooler in consequence of radiation: when the atmos- 

 pherical temperature is very low, they will, of course, freeze the water 

 which touches them. 



It is unnecessary to examine here, whether heat radiates through 

 a, thick layer of water, as Mr. McKeever siipposes, as the most simple 

 observation is sufficient to overthrow it. 



"Where is the person who has not observed, that the strong radiation 

 which the Irish philosopher admits, would be more plainly manifested, 

 or as Completel}', in still water than in running water; but no one 

 has seen a piece of still water frozen at the bottom? 



Let us throw aside all these absurd explanations, and, for want 

 of better, analyze perspicuously the physical condition of the question. 



If liquids of different densities are thrown into a vessel, the heavy 

 will sink to the bottom, the light keep at the top. 



This principle of hydrostatics is general. It applies as well to 

 liquids possessing different chemical properties, as to portion of one and 

 the same liquid whose densities are dissimilar, in consequence of inequal- 

 ities in the temperature. 



Liquids, like all other bodies, solid or gaseous, increase in density 

 as their temperature diminishes. 



Water alone, in a certain small extent of the thermometric scale, 

 presents a singular exception to this rule. Suppose water is taken at 

 +10° centigr. and gradually cooled, at 9*^ we shall find it denser 

 than at 10", at 8° more than 9°, at 7° more than 8°, and so on 

 till 4° ; at this point condensation [?] will cease. In going from 4° to 

 3°, for example, there is a manifest diminution of density. This dim- 



