Chemical Affinity in terms of Electromotive Force. 33 
with zinc-sulphate solution about two thirds saturated gave 
the values 101*07 and 100*92 during the first few weeks 
after construction ; cracks then formed, and efflorescence and 
evaporation took place, so that the zincs became partially 
bared, during which time the electromotive forces gradually 
sank. After some months the paste became covered with crys- 
tals of zinc sulphate, indicating that the residual solution 
moistening the mercurous sulphate was saturated: the elec- 
tromotive forces were then 100*13 and 99*73 respectively, 
which values were subsequently retained almost constant for 
several months longer, notwithstanding that a considerable 
portion of each zinc rod was out of the paste and exposed 
to the air. 
A number of observations made with cells containing zinc 
rods partly immersed in the paste and partly exposed to the 
air, gave sensibly the same average result as another series of 
observations made with the same cells when the zinc rods were 
pushed down so as to be wholly immersed (the upper end and 
the wire serving as electrode being protected from contact 
with the paste by gutta-percha). 
It is specially to be noticed, in connexion with the question 
of the permanence of Clark's cells, that the cells experimented 
with were only used in connexion with the quadrant-electro- 
meter; so that from first to last they never generated any con- 
tinuous current, nor had any current {however small) sent 
through in the inverse direction — conditions impossible com- 
pletely to realize in practice when working by the " method 
of opposition " or with the potentiometer. 
Effect of Temperature on the E.M.F. of Clark's Cell. 
139. According to Clark (Proc. Roy. Soc. xx. p. 444), the 
E.M.F. of a hot-prepared mercurous zinc-sulphate cell dimi- 
nishes at an approximately constant rate of 0*06 per cent, per 
degree rise in temperature between 15 0, 5 and 100° ; he states, 
however, that this figure might be verified with advantage. 
A number of observations having indicated, as a preliminary 
result, that this value is considerably too high between the 
temperature-range (10° to 25°) most frequently obtaining in 
practice, and that fairly constant results are given with dif- 
ferent cells, the following experiments were made in order to 
determine more exactly the mean coefficient of alteration per 
degree between these temperature-limits, with the result of 
showing that, instead of Clark's number (0*0006) being de- 
duced, a value but little above two thirds of this figure was 
obtained, viz. '000411, as the average often experiments with 
five cells. 
Phil Mag. S. 5. Vol. 16. No. 97. July 1883, D 
