228 Royal Society .— 



always been in the entire want of any accurate definition of that term. 

 When such definition was demanded, it was said that glacier ice must 

 be viscous, because a glacier adapted itself to the inequalities of its 

 valley as a viscous mass would do. This was equivalent to saying 

 that the mass was viscous because it moved in a particular manner, 

 instead of asserting that the mass moved in that particular manner 

 because it was viscous. Now this kind of inversion of the direct 

 enunciation of the proposition is only admissible when there is no 

 other physical cause than the one assigned, to which it is conceivable 

 that the observed phenomena should be ascribed. Thus we may 

 assert with perfect conviction, that gravity exists as a property of 

 matter and acts according to a certain law, because the bodies of the 

 solar system move as if such were the case ; but the conclusiveness 

 of this inductive proof of the proposition — that " gravity is a property 

 of matter" — rests entirely on our conviction that matter has no other 

 property by which we could equally account for the phenomena of 

 the celestial motions. And so with regard to glaciers. If viscosity 

 were the only conceivable property of ice by which we could possibly 

 account for the observed motion of glaciers, then would the observed 

 phenomena of that motion perfectly convince us of the existence of 

 the property in question. But here the two cases entirely differ, 

 inasmuch as there was no general conviction, nor even a decided pro- 

 bability at the time I allude to, that no physical property of ice could 

 exist besides viscosity which might account for the observed pheno- 

 mena of a glacier's motion ; and at the present time it is proved that 

 there is another property of ice by which those phenomena are per- 

 fectly accounted for, and the inductive proof becomes altogether 

 valueless. Moreover, in the case of universal gravitation, the induc- 

 tive proof is the only possible one, whereas in glacier motion we are 

 concerned with a property which, in whatever sense the definition of 

 it may be regarded, must be as capable of being rendered patent by 

 experiment in ice, if it exist, as in any other substance. 



The answer, then, that was given to the question — what is visco- 

 sity 1 — comprised no definition at all of that term. The viscous 

 theory ignored the possibility of the molecular mobility of a glacial 

 mass united with the preservation of its continuity, being attributable 

 to any other property than that which was designated as viscosity, 

 but without giving any exact definition of the term. If it was meant 

 to define by it the property which is here defined by the same terms, 

 the theory had a legitimate claim to be considered a physical theory, 

 because it assigned a determinate physical property as the cause of 

 certain observed phenomena. In this sense, however, the author 

 conceives that it would now be admitted to be entirely disproved by 

 Professor Tyndall's experiments, in which the ice exhibits so clearly 

 the property of solidity, and the absence of all indication of plasticity. 

 It may be presumed that the hypothesis of viscosity could only 

 have been adopted in the first instance from the apparent absence of 

 any other property of ice which might account equally well for the 

 molecular mobility of the glacial mass. 



4. But if the determinate property of viscosity, as here defined, be 



