PROFESSOR TAIT ON THERMAL AND ELECTRIC CONDUCTIVITY. 731 
Rates of cooling of the same bar, simultaneously indicated by thermometers 
whose bulbs had been immersed for different periods. 
Temperature- excess. A B C 
210° C. 5:15 
200° "4:98 
190° 4:75 
180° 4:42 4:10 
eOr 4-06 3°89 
160° 3°70 aol 3°28 
150° 3°33 3°28 3:18 
140° 2°96 2°91 2°88 
Nott.—For this experiment the bar, which was much discoloured, was not polished previous to 
heating ; so that the numbers are necessarily larger than those in §11 above. This does not affect the 
relative results. 
In each of these columns the differences are obviously least at the top, and 
the corresponding points of inflection in the curves of cooling are obviously at 
temperatures which are the lower, the colder was the bar when the thermometer 
was inserted. Also, it will be observed that the thermometers arrive more 
quickly at the true temperature the lower it is—z.e., the shorter the column of 
mercury in the stem. Another experiment gave analogous results with a 
copper bar. Thus the main difference between Forses’ results and mine is 
fully explained. 
One result of this discussion is that in heating the short bars it is more 
important to prevent oxidation than to secure absolute uniformity of heating. 
Another is that the hypothesis of uniform temperature in the cross-sections of 
the long bar is practically very near the truth. 
§ 12. In the treatment of the Statical Curves I have always used, as ForBES 
did. the formula 
Ba 
Tae 

logo = A 
It is easy to work with, and its results are usually accurate within the unavoid- 
able errors of other parts of the determination. 
Where, as with the iron and the German silver bars, the nature of the problem 
admitted it, I have constructed graphically each of two curves of statical dis- 
tribution for the same metal (with the solder at very different temperatures), 
: d 
and, to the same abscissz as the values of v, the calculated values of = One 
of these drawings was on tracing paper, and was superposed upon the other, 
with the view not merely of detecting possible errors in the calculations, but 
also of testing how far the results might be trusted. On this point I have 
2 d 
no remarks to offer further than this, that the values of for the lower tem- 
peratures, must, when they are small, as a rule be determined graphically. 
