300 PROCEEDINGS OF THE AMERICAN ACADEMY. 



a new series of experiments with same cast iron disk would probably 

 give results much more concordant than those set down in this paper. In 

 wrought iron the temperature coefficient of electrical conductivity is 

 much greater than in cast iron, and if the tempei*ature coefficient of 

 thermal conductivity is correspondingly large in wrought iron, fairly ac- 

 curate measurements of this latter coefficient should be attainable with 

 this material. A disk of wrought iron will probably be put to the test 

 before long. The disk of mild steel used in the experiments described 

 in a preceding paper was very like wrought iron in many respects ; but 

 it has already been stated, in the first part of this paper, that the experi- 

 ments with this disk were not entirely satisfactory, the disk itself and its 

 copper coverings being too thin for the best effect. 



The experiments of this paper have given a larger value of K, for the 

 piece of cast iron dealt with, than was expected. It is much larger than 

 the value, about 0.105, found some years ago for two specimens of cast 

 iron near 115°C. by one of the authors * of this paper, using the method 

 of Forbes. It is much larger than the values found by Kohlrausch and 

 by AViedemann and Franz for soft steel near 15°. Nevertheless, there 

 seems to be no good reason for doubting the substantial accuracy of the 

 value of K found in this paper. The most novel, and perhaps the most 

 doubtful, feature of the method here described is the use of the iron 

 itself as part of a thermo-electric element. How carefully the thermo- 

 electric behavior of the iron with respect to copper has been considered 

 will be apparent to the reader of Appendix I. 



Another subject of possible doubt is the amount of error caused by 

 neglect of radiation or convection between the water jacket, Figure 2, and 

 the apparatus surrounded by it. The value found for K is affected, 1st, 

 by such interaction as occurs between the jacket and the disk ; 2d, by 

 that between the jacket and those surfaces which lie above the disk and 

 below Jx and J<,_. The mean temperature of the curved surface of the 

 disk was probably four or five degrees below the temperature of the 

 jacket when the warm stream ran above, and nearly an equal amount 

 above that of the disk when the cold stream ran above. The area of 

 this surface between the two hard rubber rings h h and //// was about 

 50 sq. cm. Preston, "Theory of Heat," p. 4G1, gives, as found by McFar- 

 lane for a blackened sphere suspended within a water jacket 5° cooler 

 than itself, ^''heaf. emitted per second, per degree difference of temperature^ 



* E. H. Hall, in these Proceedings, 1892, p. 202. 



