542 MR A. CRICHTON MITCHELL ON THE THERMAL CONDUCTIVITY 



after it has been reached. This means that (besides such minor details as 

 draughts of air, and a steady or a slowly changing temperature in the laboratory) 

 great regularity in gas pressure has to be ensured. 



The gas regulator, devised by Prof. Crum Brown, was employed in this 

 inquiry also. Its working is as effective as its construction is simple ; and its 

 automatic action, hour after hour, relieves the experimenter of what would 

 otherwise be the necessity of constantly noting and regulating the temperature 

 of the melted solder in the crucible. 



The following extract from my note-book shows sufficiently how perfect are 

 the results obtained by employing it : — 



German Silver. 

 July 9, 1886. 



Gas lit 6 - 50 a.m. 



Hour p.m. 2.30 2.45 3.0 3.15 3.30 3.45 4.0 4.15 4.30 



Temp, m hole nearest 1 27Q85 m ^ ^^ ^^ 27o7 mQ m ^ ^^ ^^ 

 not end, ) 



This, showing as it does only an extreme variation in the temperature excess 

 of "09 per cent., bears out what has been said. The room in which the experi- 

 ments were conducted was lighted from the north alone, the south windows and 

 the doors were kept closed, and the air temperature throughout the day did 

 not change so rapidly as to cause any uncertainty in the results. 



A slight modification of the experiment on the long bar was tried, enabling 

 the method to give results with more certainty. Where, as in iron, the flux of 

 heat across any section at the lower temperatures (i.e., below 20°) is small, 

 a small error in the determination of the temperatures makes a comparatively 

 large difference in the estimate of conductivity. In order to avoid this, a 

 suggestion of Prof. Tait's (first given in Trans. R.S.E., 1878, p. 734) was 

 adopted. While the bar was heated in the ordinary manner, a cold water bath 

 was placed halfway up the bar towards the heated end, and through it a 

 stream of water was continually passed from below. This had the effect of 

 "steepening" the temperature-gradient, and thus allowing the measurement or 

 calculation of the tangent to be made with greater exactness. This process is 

 almost necessary in the case of metals the change in whose conductivity with 

 increased temperature is so small that the ordinary experiment of Forbes is 

 insufficient to detect its sign and amount. 



An idea of the effect of this midway cooling on the distribution of tempera- 

 ture along the bar may be obtained from the following table, in which are 

 compared the temperature excesses at the various holes in the bar, both in the 

 ordinary experiment and in that with the cooling bath. 



