Serviss — Internal Temperature Gradient of Metals. 465 



the greater will be the fraction of the total difference occur- 

 ring at the surface. The computations were made for spheres 

 rather than for a cylinder, because of the greater simplicity 

 of the formulas. The gradients in the plane through the 

 center and parallel to the bases of a cylinder 10 om in diameter 

 and 5 cm tall will be similar to those in the spheres but the 

 part of the drop which is in the metal will be smaller, because 

 of the greater surface in proportion to the volume and the fact 

 that the center is nearer the surface along the axis than before ; 

 hence the relative discontinuity at the surface will be even 

 greater for the cylinder. 



Suppose, then, that the cylinders are cooling in air, either 

 with or without a generation of internal heat ; the gradient is 

 CA-BD. My inner junction at S 2 will come to the tem- 

 perature of the adjacent metal, unless there is some small 

 gradient in the thin coating of asphaltum, and this is certainly 

 slight, as the quantity of heat transferred to maintain equality 



is small. So long as the sphere is cooling, i. e. so long as the 

 gradient CA exists, heat will flow into this junction and down 

 the constantan wire, as well as through the cylinders them- 

 selves. Again, the outer junction comes very nearly to the 

 temperature of the adjacent metal ; there may be a gradient 

 through the insulation here as at the inner junction, and this 

 gradient may be from wire to cylinder or from cylinder to 

 wire, depending on the relative ease of conduction and radia- 

 tion from the cylinders and from the thermopile to the sur- 

 rounding air. So the difference of temperature between the 



