The Thomson Effect and Thermal Conduction in Metals.

The Thomson Eflect.-In these PROCZZDINGS for March, 1920, is a paper in which I undertook to account for the Thomson effect values, as found by Bridgman for many metals, in accordance with the hypothesis of dual electric conduction. Using data obtained by the methods of that paper, I was able. as I stated at the Washington meeting of the National Academy in April, 1920, to account quantitatively for the values of thermal conductivity found in a large number of metals at room temperature. I stated further, however, that the temperature-coefficient of thermal conductivity indicated by my calculations was far too large to accord with the obvious facts, and that a revision of my formulae was therefore necessary. This revision, which has now been made, enables me to deal pretty satisfactorily with thermal conduction, if my assumptions are granted. It involves important changes in most of the numerical tables in the paper referred to, but no great change in the method there followed. The revision begins with equation (11), which is changed by the substitution of s for 2.5, so that it now stands x=i' + sR T, (1)