128 PROCEEDINGS OF THE AMERICAN ACADEMY. 



tances were obtained with the help of a set of micrometer screw gauges 

 by Brown and Sharpe. The smallest one of these was used for deter- 

 mining the diameters of the specimens and of the coils wound upon 

 them. The accuracy of this gauge was tested by a comparator (which 

 had a screw by Gaertner), and by another comparator (b}^ Zeiss) 

 which reads directly to microns. 



An illustration, the case of a specimen of American Ingot Iron, will 

 show how much error would be introduced into the value of the 

 specific magnetism of the iron by a given error in measurement of its 

 dimensions. The length was 100 cm., the diameter of the bare iron, 

 1.278(5) cm. and the mean diameter of the test coil, 1.326 cm. The 

 coil consisted of 100 turns of copper wire triply covered with white 

 silk, and as the dimensions show, the flux through it was 128.32 B + 

 9.8 H. The last term which shows the correction for the air flux 

 linked with the coil, is relativeh* small at feeble excitations, and even 

 when // rises to 2800 and B becomes about 24500, the whole term is less 

 than 1% of the flux through the iron. Moreover the value of the term 

 may be found to within one twenty -fifth of its value without trouble. 

 An error of 0.001 in measuring the diameter of the iron might make 

 an error of three units in the last place in the value of the specific 

 magnetism and this makes it desirable to use exactly round rods. 

 The piece here described was cut out of a large bar with great skill, at 

 the works of Messrs. Barbour and Stockwell. 



At high excitations, the corrections for the effect of the ends of the 

 cylindrical test pieces are, of course, much less than those which 

 according to theory and to the formulas of DuBois and of Shudde- 

 magen are necessary in low fields. I shall hope to discuss this matter 

 at length in another paper, and need only state here that the correc- 

 tion for a piece of the dimensions used was practically negligible in 

 fields of strength above 2000 gausses. 



The rods to be annealed were first packed tightly in fine iron filings 

 in a piece of pipe the ends of which were closed by screw caps, and the 

 whole was carefully supported perpendicular to the meridian in a 

 special gas heater where it would be exposed to several hundred flames 

 dri\en l)y a power compressor. In this manner a piece 150 centimeters 

 long could be heated very uniformly. After the specimen had been 

 kept for perhaps an hour at a temperature considerably above the 

 critical point of the iron it could be then allowed to cool very slowly 

 in situ, protected from magnetic action. 



If a slender rod of iron be placed inside a long solenoid which is 

 in the secondary circuit of a powerful open-core transformer, and if 



