

PROFESSOR TAIT ON THERMAL AND ELECTRIC CONDUCTIVITY. (ae 
increasing ratio, so that the assumed correction for oxidation was easily made, 
and probably pretty accurate. I cannot, however, feel certain that I have in 
all cases applied it rightly. It is not at all easy to pronounce on an equality 
of oxidation of two bars (so far as our present purpose is concerned) unless 
both be employed for the cooling experiment. 
Forses expressed an opinion (which I do not share) against electro-plating 
the bars to prevent oxidation. I intend to try this method; and also, if 
possible, the wrapping of the bar in thin sheet iron, so as to employ ForBEs’ 
bath of solder. I have made several experiments with bars smoked. The 
method promises well, except perhaps in the case of copper, but the calculations 
are not yet effected. 
§ 14. The Statical Curves of Cooling were constructed exactly as described 
by Forzes. But there are two remarks of some importance to be made upon 
the mode of obtaining their areas. 
In the first place, they are not even approximately logarithmic, except for 
small intervals. And even then the axis is not usually the asymptote. Their 
area between two ordinates is usually greater than that of a logarithmic with 
the same axis and passing through the two points. 
Secondly,—It is a matter of great difficulty to determine what to allow for 
the portion, in theory infinitely long, but finite in area, which extends beyond 
the point of lowest observation of temperature on the long bar :—except in the 
case of the copper bars, where the temperature was kept at the further end 
lower than that of the surrounding air. The end of the bar was introduced into 
a large vessel of gutta-percha full of water, which was constantly renewed from 
below by means of a pipe connected with a large cistern. Thus the values of 
d 
were never very small at any observed part of the bar. 
The question here raised is a very important one. It is not at all probable 
that the thermal conductivity should, in all the substances I have examined, begin 
to change very much more rapidly below 50° C. than it had been changing 
during the whole range to that point from 200° C. or even from 300° C. Hence, 
when I found the conductivity to be well represented between these limits (in 
terms of the temperature) by a straight line, I have ignored (as almost 
certainly due to errors inseparable from the method employed) the some- 
what marked and rapidly increasing curvature, which is indicated in many 
cases, for the lower 20° or 30° of observed temperatures. I justify this pro- 
ceeding on the ground that (in addition to the fact that the areas, the 
smaller ones especially, are underrated by treating the curve as logarithmic) 
very slight differences in the quantity allowed for the infinitely prolonged area 
‘a quantity whose value we can only guess at) make all the difference between 
a rapidly increasing curvature and a rapidly diminishing one (sometimes even 
VOL. XXVIII. PART III. 9F 
