720 PROFESSOR TAIT ON THERMAL AND ELECTRIC CONDUCTIVITY. 
the various bars and thermometers obtained for the present inquiry. In Ane- 
strom’s method it is so much more easy to calculate out the results, and derive 
the conductivity from the experiments, than in that of Forsgs, that I have already 
—in 1872—73*—communicated to the Society the results obtained by this method, 
though I had years before made most of the experimental determinations 
required by Forses’ method, whose numerical consequences are only now pro- 
duced. But my thermometers, though excellent for the use of ForsEs’ method, 
were not nearly delicate enough for the proper application of that of ANesTROM. 
It requires, for its proper carrying out, the very accurate reading of small 
changes of temperature. Hence the results of 1872-73 can be looked upon as at 
best but very rough approximations. One great defect of Anestrom’s method, 
as compared with that of ForsEs, lies in the assumption (which forms part of 
its necessary basis) that the rate of surface loss is proportional directly to the 
excess of temperature over the surrounding air. Even for the moderate range 
of temperature employed in AnastROm’s experiments,t this is not nearly correct. 
Hence, and for other reasons (for instance, his equations being formed as if & 
were constant), I do not accept his statement that the thermal conductivity 
of copper falls off as the temperature rises, as one which his method was 
competent to decide. Even with Forsrs’ much superior method, a range of at 
least 100° C. is absolutely necessary to settle such a point. 
I have had several reasons for delay in publishing the results of these ex- 
periments. For the most part, the experiments themselves were made eight or 
nine years ago, but for the delay with regard to the calculations I am not 
wholly responsible. Since I obtained the assistance of Mr Evans, however, 
there has been no unnecessary delay in the computations. Experimental diffi- 
culties of various kinds were, however, constantly.cropping up. Besides the 
difficulty already alluded to, of maintaining a steady temperature of the source 
of heat, a very peculiar difficulty arose from the behaviour of the thermometers. 
These, after being exposed to high temperatures and cooled, showed a gradual 
rise of the zero points ; and, in some of those which have been most frequently 
exposed to the highest temperatures, the zero point has risen as much as about 
five degrees. There were also very great difficulties about the heating of the short 
bar for the cooling experiment. Here my results were very different (at high 
temperatures) from those of ForBrs. Again, the lead and copper, and sometimes 
(in extreme cases), even the iron and German silver, when highly heated, become 
oxidised, and the coating of oxide on the surface promotes radiation, if not also 
convection ; and as the surface becomes oxidised to different degrees at different 
temperatures no one set of experiments with the short bar is strictly comparable 
with anything but one part of the long bar. That difficulty is not so much felt 
FPTOC mS ou ke + Pogg. Ann., Band 118, 1863. 

