A STANDARD OF ELECTRO-MOTIVE FORCE. 
795 
The coefficient '00038 is so close to that given by Alder Wright, viz., '00041, 
that one cannot help suspecting that the cells used by him for this purpose may have 
been unsaturated, or possibly supersaturated (without deposit of lower hydrate). 
Over the above range of temperature a linear expression represents the E.M.F. 
sufficiently well. When, however, the values recorded for (36), and (37) for tem¬ 
peratures from 32° F. to 67|° F. were plotted, a distinct curvature was apparent 
indicating a lower coefficient at the lower temperatures. Table XVI. exhibits the 
observed values, and (for comparison with them) numbers calculated from quadratic 
expressions. For (37) the expression is 
'9985(1 — '00037(^-521) — '0000025(f-52|) 2 } ; 
the equivalent of which in Centigrade degrees is 
•9985(1— 000666(^—11'4) — '0000081^ —11*4) 2 }. 
According to this, the change for one degree Cent, is the following linear function 
of temperature 
•000674 + '0000162(« —11-4) ; 
so that the temperature-coefficient ranges from '000489 at 0° C. to '000813 at 20° C. 
At 15° C. it would be '00073. 
For the other cell (36) the observed values of E.M.F. are pretty well repre¬ 
sented by 
'9991(1 — '00042(^ — 53) — '0000028(£ — 53) 2 }, 
from which are deduced the numbers in the column headed “calculated.” In Centi¬ 
grade degrees this becomes 
•9991(1 — -000756(< —11*6) — '0000091 (£ —11'6) 2 } ; 
giving for the temperature-coefficient at f Cent. 
•00076 5+ *000018(^ — 11'6). 
At-15° the value from this formula is '00083. 
It would seem that these two cells have temperature coefficients which differ 
sensibly. But this way of presenting the matter is apt to give an exaggerated 
impression. The difference in the coefficients indicates a separation of electro¬ 
motive forces at the rate of y ooo o on ^J P er degree Cent., so that the whole relative 
change for ordinary indoor variations would not exceed t~ oo o of the whole. It will be 
seen from Tables XIV., XV. that, through a more limited range of temperature, a large 
number of various cells are satisfactorily represented by the coefficient (intermediate 
between those just found) '00077; and I believe that the adoption of this number* for 
cells with saturated solutions can lead to no appreciable error in ordinary use. For 
* It should be mentioned that ( loc. cit.) Dr. Fleming found for a Clark cell the coefficient '00082 at 
15° C. 
