PROFESSOR TAIT ON THERMAL AND ELECTRIC CONDUCTIVITY. Tan 
somewhere about 10° at 250° C., and varying for other temperatures as above 
stated. I have preserved all the notes of experiments, as well as the thermo- 
meters, as it may ultimately be possible to get an air-thermometer which will 
enable me to reduce the determinations to a more accurate standard; but 
until that can be done it seems hopeless to expect to improve (in this par- 
ticular) the method I have employed, however important might be the results. 
§ 11. There is one respect, and one only, in which my results have been 
found to be not quite consistent with those of Forsrs. This is in regard to 
the law of cooling of the short bar in terms of the temperature. FORBES, in 
fact, called special attention to this question, and he evidently felt considerable 
surprise at the result he obtained, for he tried it over and over again with the 
same conclusion. Although he pointed out that the initial uniformity of tem- 
perature of the heated bar would tend to produce the appearance of such a 
result, Forbes expressed himself as convinced that the curve representing the 
rate of cooling of the short bar in terms of the temperature begins to be straight 
about 150° C., and then bends over so as to become convex upwards. I have 
carried it considerably farther, in fact, up to estimated temperatures of at least 
300° C., without finding the slightest trace of convexity. It is obviously essential 
that this discrepancy should be explained ; and I think it depends on the fact 
that Fores did not heat his short bar much above the temperature (200° C. or 
thereabout) at which the readings commenced. Under these circumstances 
the flow of heat from the interior of the bar is for some time retarded; in fact, 
till a state of things is arrived at in which the temperatures at different 
distances from the axis or from the ends of the bar cease to undergo a rapid 
relative change, the inserted thermometer does not indicate the true loss of 
heat by the bar. I think that this explanation is borne out by the fact that 
Forses’ results, with a bar of smaller section and length (in which the abnormal 
state is of shorter duration), agree more nearly with mine, so far at least as 
change of rate of cooling is concerned. 
I easily reproduced Forses’ results by heating the bar only to the tempe- 
rature at which the readings commenced. But to avoid this source of error I 
always, when it could be done, raised the temperature much above the point at 
which readings were to begin, so as, in fact, to read only when the normal state 
of cooling had been arrived at. In some of my experiments with iron the bar 
was heated to such an extent that mercury boiled furiously when put into the 
hole—and I had to employ fusible metal instead. In all cases I obtained 
results resembling those of Forses during the first few minutes of cooling. 
The following short table illustrates this difference, as well as the fact stated 
in § 9 that my numbers are all a little too high. The first column gives the 
temperature-excess over the air; the second contains the rate of cooling as 
given by Forses; the third column contains results obtained (for the same 
