126 Barrett, Brown & Hatfield — On various Alloi/s of Iron. 



at and at the middle of the first division to the riglat of OB at the point 16000. 

 It will be seen that, practically speaking, this correction only affects the values of 

 the residual induction and permeability shown on our plates and tables, and 

 mainly of those specimens having a high permeability. Thus the residual induc- 

 tion of iron will be raised in round numbers from 7000 to 10,000, or about 40 per 

 cent., whilst that of a specimen of lower permeability, like 1267 B on Plate IV., 

 will be raised only 2\ per cent. The values given in our tables are deduced 

 from the experimental results, without any corrections being applied ; for the 

 correcting factor really relates to long ellipsoids, and it is only as a first approxi- 

 mation that it can be applied to cylindrical rods such as ours.* 



As already stated on ji. 100 we have had some of our specimens made in the 

 form of rings ; these were turned out of a solid block of the material in as nearly 

 as possible the same physical state as our annealed rods. The experiments on 

 these rings are still in progress, and when completed we shall be able to obtain a 

 more accurate knowledge of the demagnetizing factor for specimens having a wide 

 range of permeability. 



It is perhaps desirable to repeat that we have used the word " retentivity " in 

 our tables as a brief expression for the residual induction after a maximum 

 mao-netising force of 45, but the true definition of retentivity is that given by Du 

 Bois in his recent work on the Magnetic Circuit, viz. the ratio of the residual to 

 the maximum induction. 



Note B. 



Owing to difficulties in exact magnetometric measurements in an unsuitable 

 laboratory, we have found, since this paper was in type, that the values assigned 

 to the induction B in our Plates and Tables must be reduced about 2 per cent., 

 but this only affects the absolute values, and not the comparative results. 



A recent repetition of the magnetic cycle with our best annealed iron rod S.C.I. 

 turned down till its length was 255 diameters, together with a redetermination of 

 the Earth's horizontal force in a better place of experiment, gave a maximum 

 induction (for 45 C.G.S. units) of 16750. This is about four per cent, lower than 

 we have given in our tables for our standard iron rod marked B, but this latter, 

 thouo-h less permeable in low fields than S.C.I., has a higher induction at 45, 

 probably due to its slight difference in chemical composition or annealing. 



* It appears from some recent experiments by Mr. C. G. Lamb, m.a., " On the distribution of magnetic 

 induction in along iron bar," tbat the employment of a constant demagnetizing factor is not quite correct. 

 Mr. Lamb concludes that "the magnetometric method with long cylindrical rods, although extremely 

 useful for comparative work, must be used with much caution in determinations of an absolute character." 

 — Proceedings of the Physical Society, December, 1899, p. 517. 



