144 BRIDGMAN. 



The intermediate point was put to ground. This is the regular method 

 of avoiding capacity effects in the telephone. R5 and Re were so large 

 that there was no serious diversion of current from the bridge. 



The D.C. source was a storage battery connected in series with 

 a General Electric Co. ballast lamp (iron filament in hydrogen) of 

 1.4 amp. capacity. In series with the lamp was a commercial ammeter 

 with which the constancy of the input current was checked. Any 

 desired fraction of the output of the battery could be diverted from the 

 bridge by a variable shunt between the lamp and the bridge. The 

 actual current into the bridge or through the sample was not measured 

 directly, but was computed from the ammeter reading, and the resist- 

 ances, which were measured with the requisite accuracy. If a heavier 

 current than 1.4 amp. were needed, two ballast lamps could be used 

 in parallel. The dimensions of the sample were such that in almost 

 all cases the maximum current that it could carry without burning 

 out was not over I amp. 



The source of A.C. was a vacuum tube oscillator. The riiethods of 

 connecting this were the canonical methods, and need not be gone into 

 here. I am much indebted to Professor L. E. Chaffee and Mr. S. 

 Ballantine for assistance and advice in setting up this circuit. For 

 the first readings a Western Electric Co., hot lime transmitting tube 

 Type VT2 was used, but this soon was burned out, and for most of 

 the work a G. E. transmitting tube, type T Pliotron, was used. 



Not only does the present differ from preceding attempts in the 

 method of measurement, but also in the form and dimensions giv^ 

 to the metallic resistance carrying the high current density. In all 

 preceding work the metal has been in the form of a fine wire, of 

 diameter of the order of 0.001 inch or more. An elementary discus- 

 sion will show that a wire of these dimensions will carry only a limited 

 current of the order of 10^ amp/cm^ in the most favorable case. The 

 limit is reached at a rate of heat input so high that the interior of the 

 wire is at the melting point while the outer surface is at 0°, the thermal 

 conductivity of the metal just sufficing to carry off the heat input 

 under the temperature gradient so produced. It is possible to gain 

 somewhat by rolling the wire flat as a galvanometer suspension, but 

 not a great deal. An elementary dimensional discussion will show 

 that the only way to gain on the upper limit of current density is by 

 decreasing the thickness of the specimen, so that a given difference of 

 temperature between interior and exterior will give a larger tempera- 

 ture gradient, and therefore a greater heat dissipation. The thinnest 

 metal that can be obtained is in the form of beaten leaf, and it was 



