Passing Isothermally from Solid to Liquid. 



11 



while to construct an apparatus for r = u ; for although ordi- 

 narily the correction for surface conduction is nearly enough 

 given by the computed value, there are cases of exceptionally 

 bad conduction (for instance in gases) where this is no longer 

 fully the case. Suppose, however, the system be duplicated 

 as in figure 2. Let the 



Fig£. 



F 



«: 







~~ 



s 



F 



G 



GT--' 



Fig. 2. — Duplicate apparatus for heat 

 conduction. 



u , and therefore as regards the 



A'' & 



lower pair of plates, D, E, 

 be like those of the above 

 figure. Let the upper 

 pair of plates D', E f , be 

 somewhat larger, and the 

 conductor D', somewhat 

 thinner and shall owly bell- 

 shaped, so as to surround 

 ' the plate D on all exposed 

 sides. Then if JEpc/ 

 M lGi = AF'p'c'/M\c\ the 

 constants q and q' for the 

 two systems will be iden- 

 tical to a second degree of 

 approximation. Hence u ■ 



plate D, t = u at all times. 



In using this device for measuring conduction in gases, the 

 environment ABC, figure 1, would have to be sealed hermeti- 

 cally, and the gas to be studied, introduced into the whole 

 interior space, at any temperature (environment) or pressure 

 desirable. In such a case, since d = J', p — p', c = c', c= <?/ 

 the simple conditions are 



FjF'= mjm; 



(4) If the environment be at a temperature intermediate 

 between the initial and the final values u of a given pair of 

 observations, then 1 — r/u passes from positive to negative 

 values, and the corrective may also vanish ; but this case is 

 superfluous here. 



Returning for a moment to case (2), it is clear that the radia- 

 tion corrective is least when the temperature gradient of the 

 layer is steepest. So long therefore as the thermal variation 

 of conductivity is not fully known, it is advisable to prefer 

 small values both of TJ and J. 



Experimental Results. 



12. Method of observation. — If after the lapse of 60 minutes 

 the plates were heated uniformly, the cold water faucet was 

 suddenly opened; but observations of temperature (it) and 

 time (t) were not commenced until a minute later. A good 



