120 PROCEEDINGS OF THE AMERICAN ACADEMY. 



capable of being rotated together about a horizontal axis perpendicular 

 to the pole axis, so as to reverse suddenly the sign of the magnetization 

 in the test piece. Each specimen was 28 millimeters long and 3 milli- 

 meters in diameter. To make sure that the lines of induction in the 

 test piece were throughout parallel to each other, Gumlich sometimes 

 used soft iron rings slipped over the specimen. Gumlich's value of I^ 

 was 1725. 



In December, 1910, Messrs. Hadfield and Hopkinson printed the 

 results of a very carefully carried out and very elaborate investiga- 

 tion into the question whether in such combinations of iron and less 

 magnetic substances as are in practical use, the specific magnetism 

 of any piece of the material multiplied by the mass of the piece is 

 simply equal to the sum of the products obtained by multiplying the 

 mass of each constituent in the specimen by its specific magnetism. 

 They came to the conclusion that although this rule seems not to hold 

 in certain alloys of iron, nickel, and manganese, it is really fulfilled 

 in many practical cases. 



They used a modification of the Isthmus Method very skilfully, 

 employing an electro-magnet like, if not identical with, the magnet 

 which Ewing and Low had, and which was made for the first isthmus 

 experiments, under the direction of W. Low, Esquire, of Bahnakewan. 



Hadfield and Hopkinson had at command a large number of alloys 

 specially made at the Hecla Works, for research purposes, and tlie 

 analyses of their test pieces are therefore beyonfl question. They 

 found that in their annealed iron-carbon steel, where other elements 

 were nearly absent, the specific magnetism was less than for their 

 standard iron by a percentage equal to about six times the percentage 

 of carbon. In such a case they assumed that there are two constituents, 

 pure iron, and iron carbide (FeaC) in mechanical mixture, the percent- 

 age of the carbide present being 15.5 times that of the carbon in the 

 steel. The "pure iron" used as a standard was a sample of Swedish 

 iron (Maker's mark "S. C. I.") containing less than 0.2 per cent of 

 impurities. Of this they used two specimens: one was 6.26 mm. 

 long, and 3.18 mm. in diameter and weighed 0.385 grammes; the other 

 was 15.92 mm. long and 3.19 mm. in diameter, and its weight was 

 0.99 grammes. Both yielded the same value (1680) for the specific 

 magnetism. Table III, obtained from measurements of one of the 

 curves given by Hadfield and Hopkinson, reproduces their results 

 sufficiently well. Most of their pieces of steel were slightly less dense 

 than the S. C. I. iron and their final values of I^ give the magnetization 

 vector per unit volume of matter of the same density as the iron. 



