164 BRIDGMAN. 



ties without the specimen burning out. If the maximum current 

 density without burning out is fixed by the amount of heat that can be 

 carried away by the cooHng water, and if this is independent of thick- 

 ness, as we have seen it is approximately, then the maximum current 

 density for a piece of twice the thickness shoukl be inversely as V2. 

 This relation is approximately satisfied. It is to be noticed that the 

 effect is greater for the thicker gold; this would be expected from 

 theoretical reasons. 



In addition to these two thicknesses of gold, I tried several 

 samples of the next size gold that I could obtain in the market, namely 

 5 X 10~^. Results with this were unsatisfactory. The curvature of 

 the relation between A.C. and D.C. difference and reciprocal frequency 

 continues to much higher frequencies than for the thinner specimens, 

 and in fact is quite marked over the entire frequency range, so that 

 it was not possible to extrapolate. It can be seen at once that the 

 inequalities of temperature in a leaf of five times the thickness are 

 twenty-five times as great for the same current density. 



The results for silver 2 X 10"^ cm. thick are shown in Figure 12. 

 Measurements were made on five different samples, of widths varying 

 from 0.072 to 0.165 mm. The results are more scattered than for the 

 thin gold, but again there can be no question of the existence of the 

 effect, and the fact that there is no correlation with the breadth. 



In addition to these measurements on gold and silver I attempted 

 to measure the effect in aluminum. The thinnest aluminum leaf that 

 can be obtained in the market is 5 X 10~^ cm. thick, and I found the 

 same trouble with it that I did with the thickest gold, namely that the 

 curvature persists to such high frequencies that the extrapolation 

 cannot be made with any assurance. 



Other Possible Explanations of the Effect. 



Skin Effect. It is natural to search for explanations of these 

 results other than a departure from Ohm's law. One of the first 

 that suggests itself is the "skin effect." The resistance of a con- 

 ductor is higher to alternating than direct currents because the 

 alternating current tends to collect in the surface layers, not having 

 time to soak into the interior of the conductor. This eft'ect is verv 

 important at wireless frequencies with conductors of the ordinary 

 dimensions. It increases rapidly with increase of frequency, and is 

 greater in large than in small conductors. An upper limit for the 

 conductor used in these experiments may be found by applying 



