452 Serviss — Internal Temperature Gradient of Metals. 



The Apparatus. 



Dr. Thwing's general method was followed ; viz., the sub- 

 stance to be tested is used in the form of two equal cylinders? 

 set in an air bath, one on top of the other, with a thermopile 

 consisting of a large number of junctions wound radially 

 around a ring of mica, between them to measure the mean 

 gradient along the radii. Certain modifications in the details 

 were, however, made ; these will be pointed out as the appa- 

 ratus is described. There may be some further departures 

 from Dr. Thwing's construction — his paper is unfortunately 

 brief, especially in the description of his apparatus. 



The Specimens. — For these experiments cylinders of lead 

 and of iron were cast, about 12 cm in diameter and 4*5 or 5'0 cm 

 thick, and turned down to 10x2*5 cm . As the castings appeared 

 less porous on the bottom and consequently that end of the 

 cylinders seemed likely to be more uniform, the lower end of 

 each casting was turned down until a satisfactory surface was 

 obtained and then the top cut off until the cylinder was of the 

 desired thickness. It was necessary to oil the cutting tool to 

 obtain smooth surfaces, so after the cylinders were taken from 

 the chuck they were washed with naphtha to remove the oil, 

 and placed in a cabinet until tested. The top side of each (as 

 cast) was marked ; when the specimens were not in use, the 

 unmarked base was exposed to the air, and when tested, placed 

 next to the thermal couples. The weights of the cylinders 

 after turning indicated that they were solid. The specimens 

 were of ordinary commercial purity. The lead was cast in 

 the laboratory by melting clean weights which were no longer 

 needed. The iron came from a local foundry and was fine and 

 homogeneous. 



The thermopile consisted of 100 copper-constantan thermal 

 couples, made from wires - 020 em in diameter (No. 32 on the 

 Brown and Sharpe gauge). All the wire of each kind was cut 

 from the same strand. Pieces of the constantan 1*5 meters 

 long were annealed by clamping the ends to wide U-shaped, 

 rigid, wooden frame and heating to incandescence in free air 

 by a current of 3*5 amperes from a direct current dynamo. 

 The current was of uniform strength and was maintained for 

 about three seconds in each case. The ends of wire were held 

 at the same level, so that while hot, it hung in approximately a 

 catenary. When the wire was clamped on the frame it was 

 drawn just taut, and after cooling returned to its original ten- 

 sion, so it was not permanently extended to any appreciable 

 degree. 



Two lengths were selected at random and laid aside for 

 calibration. The remainder was cut into pieces exactly 3 cm 



