﻿Convective Cooling in Fluids, 935 



correction to be expected is sensible, and is of the sign and 

 order of magnitude required to bring the results for the 

 two wires into even closer agreement. In fig. 2 single 

 curves have been drawn representing mean values for each 

 of the liquids. 



Although our own figures for gases fall on a very distant 



-eoion of the 



grapn 



and so are not shown, a curve has been 



given to represent gases. It was obtained from the paper 

 already mentioned, where cooling was studied for a wide 

 range of wire diameter and of temperature excess, for 

 various gases — oxygen, hydrogen, and air — over a thousand- 

 fold range of gas-pressure. 



Fia-. 2. 



: 2 ad 3 c9/k 2 



It is seen that the curves for the different fluids form a 

 family of more or less parallel lines, with cv/k as parameter. 

 The magnitude of the effect of cv/k is not great, for 

 although glvcerine has a value more than 10,000 times that 

 for air, the value of H/& for a given value of c 2 gd*a6/k 2 is 

 only reduced in the ratio 3:1. It is satisfactory that 

 toluene and carbon tetrachloride, with values of cv/k so 

 nearly the same, should be represented by one line. The 

 curve for diatomic gases (cv/k = 0'74:) may be shown to 

 agree with published results for cooling of wires in C0 2 

 («//& = 0*83), so tnat Dere a g am i fc appears that cv/k 

 determines the position of the curve. 



A logarithmic plot of the relation between H/k and cv/k 

 for various constant values (1, 10, and 100) of c 2 gd' 3 ad/k 2 

 yields more or less linear relations; and it appears that within 



