246 A. Durward — Temperature Coefficients of 



hardened steel, where some abnormally large temperature 

 coefficients were explained by the fact that the steel proved to 

 be soft in places. Local softness is very difficult to avoid in 

 re-hardening steel rods which have once been made glass-hard, 

 and one must expect to find that an occasional specimen of 

 once hardened steel is soft in places.* 



If great accuracy is required it is of course necessary to 

 determine the coefficients of a given magnet by direct experi- 

 ment, but the average of determinations made on a number of 

 magnets of a certain size seems to show very approximately 

 what may be expected of normal specimens. 



The straight bar magnets experimented on were of several 

 kinds of steel, and varied in length from 3 cm to 20 cm . Most of 

 them were circular cylinders, though there were about 40 of 

 rectangular cross section and a few were what are commonly 

 called " hollow cylinders." All had been heated to a cherry 

 red in a special gas heater, then immersed in a large tank of 

 rapidly stirred, acidulated or salted, ice water, and finally 

 exposed for a long time to a temperature of 100° C. After 

 the rods, which were still glass-hard, had been magnetized to 

 saturation between the poles of a soft iron yoke in a long sole- 

 noid, they were again exposed for hours to the temperature of 

 boiling water. After standing for several months, most of the 

 magnets were reboiled, re-magnetized in the same direction as 

 before, and then boiled again. In some cases this process was 

 repeated several times. 



Besides a number of twice hardened magnets, and a few bar 

 magnets of rectangular cross section, I tested nearly 120 round, 

 straight magnets made of once hardened tool steel, from # 32 cm 

 to l-ll cm in diameter. I am indebted to Prof. B. O. Peirce, 

 and also to Mr. C. Gr. Persons of the staff of the Jefferson 

 Physical Laboratory, for additional tests made on about 200 

 magnets. 



It was desirable first to find out how nearly constant the 

 temperature coefficients of such magnets are between ordinary 

 room temperatures and 100° C. To test this the magnet to be 

 experimented on was mounted horizontally in a non-magnetic 

 holder, fixed in a wooden tank, so as to be in Gauss's A posi- 

 tion with reference to a mirror magnetometer which rested on a 

 stable support of its own. The tank could be moved eastward 

 or westward on a track and could be clamped to the track if 

 desired. Its distance from the magnetometer could be accu- 

 rately adjusted by the help of a slow motion screw. By means 



*The permanent magnetic moment of one of two hollow magnets made by a 

 well known firm of instrument makers for use in the Jefferson Physical Labora- 

 tory, and supposed to be as nearly alike as possible, proved to be nearly twice as 

 great as that of the other, and this difference was due to difference in harduess. 



