1877.] 



Dr. E. J. Mills on Electrostriction. 



511 



with copper at least, at the melting-point of ice ; and it is not improbable 

 that copper deposited at or near 100° would not show any such effect at 

 any temperature. One circumstance may be mentioned as bearing on 

 this point, viz. that the electrostrictive effect is greater, and more rapidly 

 produced, at the winter temperature than at that of summer. It may 

 be added that soft metallic deposits are, as a rule, obtained from heated 

 electrolytes. 



(5) Uniform external pressure on a cylinder takes effect chiefly on the 

 diameter, and less on the length ; and these results are practically in- 

 dependent of each other *. If we conceive such pressure to be produced 

 by the electric deposition of successive layers of metal (as was actually 

 the case with thermometer 502), it seems reasonable to suppose that 

 each of the layers will have nearly the same constrictive effect. But the 

 first layer constricts the bulb alone ; the second constricts the first also ; 

 the third its predecessors, and so on. Hence the observed effect upon 

 the bulb should diminish at compound interest. If y be the total 

 obtainable effect after a time oc, D the portion of it due to diametral 

 constriction, L the portion of it due to longitudinal constriction, and 

 d and I the respective geometric factors, we have, in the case of the 

 cylindrical thermometer, 



2/=Def+Lf, 



D being always greater than L. On the other hand, the spherical 

 thermometer necessarily receives more metal upon its equatorial region 

 than can be deposited above or below, that region being nearer to the 

 anode ; and the longitudinal constrictive action must tend to bulge the 

 equatorial part, opposing the constriction at work there. 

 Hence we have, for the sphere, 



y = T)d X -Lf, 



D being the greater as before. Nevertheless it appears probable that in 

 both cases, at the early stage of deposition, the difference between 

 diametral and longitudinal compression is but slight, not having yet been 

 much multiplied by the increased stress. Hence, during that period, an 

 equation 



y = ~Dd 



may doubtless more accurately represent the experimental results. 

 Subject to this and some other minor considerations (the close discussion 

 of which would be out of place in mere indicative work) there is a fair 

 agreement between theory and experiment. The probable error, in fact, 

 of a single observation does not on the whole exceed *053 of a scale- 

 unit, being about *4 per cent, on the total quantity determined ; and this 

 reduces to less than -3 per cent., if we exclude the first comparisons 

 made at a lower temperature, in Table I. 



* Fairbairn, Phil. Trans. 1858, p. 404, 



