EFFECT OF PRESSURE ON CONDUCTIVITY OF METALS. 123 



flows through the bar and out at the other end, and only a compara- 

 tively small part is lost laterally to the surroundings; whereas with 

 Lussana all the heat input flows out laterally. In Lussana's method 

 the correction for the efi^ect of pressure on the transmitting medium 

 affects directly the entire heat input, whereas in my method the pres- 

 sure correction is to be applied only to that part of the heat input 

 which escapes laterally. 



My most serious criticism of Lussana's method concerns this cor- 

 rection for the transmitting medium. The magnitude of the correc- 

 tion is about 30% per thousand kg., whereas the order of magnitude of 

 the changes of thermal conductivity of the metals is at most only 3%, 

 or one tenth of this. This demands that the effect of pressure on the 

 transmitting medium be known ten times well as the final result for 

 the metal. Nevertheless, Lussana determined the correction for the 

 liquid to only one significant figure; as a matter of fact there is a mis- 

 print in his paper, which made the correction appear to be at the rate 

 of 300% for one thousand' kg. I inquired about this in a letter to 

 Lussana, and he told me that the decimal point had been displaced one 

 figure, and that the correct result was 30%, agreeing with my own 

 results as far as order of magnitude goes. Having determined the 

 correction to one significant figure, not even noticing the departure of 

 the effect from linearity with pressure, Lussana gives his coefficient for 

 metals to three significant figures. Three significant figures for the 

 metal would have demanded at least four significant figures in the 

 correction. 



Lussana states that his results were computed from the observa- 

 tions by the method of least squares ; he does not anywhere reproduce 

 a single set of observations, nor does he state the probable error of his 

 results, surely a significant omission considering the method of 

 computation. There is no clue in his paper to the accuracy to be 

 attached to his results. 



There seems to be almost no correlation between Lussana's results 

 and my own. In only one case, that of zinc, do we find the same sign 

 for the change produced by pressure in the Wiedemann-Franz ratio. 

 It seems to me that for the present we are justified in assuming that 

 there are large errors in Lussana's results. 



Discussion. 



Probably the most significant theoretical conclusions from the 

 above data may be derived from the pressure coefficient of the Wiede- 

 mann-Franz ratio. The classical electron theory would lead us to 

 expect that the coefficient would be zero, since the ratio is the same for 



