Proceedings of the Royal Society. 267 



a rotation in one direction ; secondly, on that of a rotation in the 

 opposite ; and thirdly, in that of no rotation at all. From these, 

 by comparing them with a few of the observations, he deduced 

 numerical values for the constants of the formulae, and then 

 employed them to compute the deviations due to rotation in all 

 the rest. He regarded the discrepancy between the calculated 

 and observed results, as in few cases, larger than what he con- 

 sidered may be fairly attributed to error of observation ; and that 

 the theory above stated is at least a general representation of 

 what passes in fact : admitting, however, that it does not give 

 the exact position of the point where the deviation due to rota- 

 tion vanishes, and suggesting partial magnetism in the iron plate 

 used as one mode of accounting for the difference. At all events, 

 by an examination of the case on the ordinary supposition of 

 induced magnetism in the iron, he shewed that a greater coinci- 

 dence between theory and fact would not result from that hy- 

 pothesis than from the one here employed. 



He then proceeded to inquire into the degree of permanence of 

 the polarity thus produced in iron by rotation, from which inquiry 

 it appeared that (at least during 18 hours after the plate was 

 brought to rest) the influence of a single rotation had scarcely 

 suffered any diminution. It appeared also that the effect is so far 

 from depending on the rapidity of the motion, that the plate can 

 scarcely be made to revolve so slowly as that the whole effect shall 

 not be produced. 



Lastly, by a slight change in the formulae, the results of com- 

 putation it is found can be made to agree with observation to a 

 degree of exactness as near as can be wished. This change con- 

 sists in the omission of certain terms introduced by the theory, 

 and the author regards it as very possible so to modify the theory 

 as to get rid of them. 



Tlie author closed this communication with an appendix com- 

 paring the magnetic effects produced by slow and rapid rotation. 

 The result of the comparison was, that the forces exerted on the 

 needle during rapid rotation were always in the same direction as 

 those derived from the slowest rotation, arid which continue to 



