Mr. R. H. M. Bosanquet on Electromagnets. 535 



approach very closely to 1. The values throughout the middle 

 range, for both bars, are well represented by the formula 



/=-650 + -44()£. 



J r 



It is only for small inductions, and again for over 19,000, that 

 this fails. It holds throughout the range of ordinary practice. 

 In the following comparisons the last entry in each table is 

 beyond the range for which it holds. 



r 

 / = 64966, log C=T-64373. 





Bar 



I. r=-9379. 





M 



P- 



/. 



Calc- Obs. 



Calc. 



Obs. 



2,467 

 10,210 

 17,214 



17,858 



•23784 

 •23296 

 •29928 

 •33107 



•76161 

 •75932 

 •79046 



•80538 



•76734 

 •75417 



•79787 

 •80815 



-•00573 

 + •00515 

 -00741 



-•00277 



19,289 



•44536 



•85908 



•83466 



+ 02442 





Bar 



II. >■ = • 



4800. 





5,675 

 11,113 

 14,093 



18,200 

 19,411 



•08504 

 •08216 

 •08363 

 •13384 

 •18011 



•72796 

 •72531 

 •72666 

 •77271 



•81515 



•72626 

 •71846 

 •73590 

 •79298 * 

 •81437 



+ •00170 

 +•00685 

 -•00924 

 -•02027 

 + •00078 



19,927 



•26328 



•89143 



•86892 



+ •02251 



Generally, we may chiefly note that the course of the values 

 does not at all indicate any definite limit to 3Bf. The only 

 limit that stopped me from forcing it up higher in both cases 

 was that the coils became hot. In one case the coil of I. was 

 actually charred. 



It appears then, so far as the iron I am working with is con- 

 cerned, that the existing idea of saturation is as complete a 

 mistake as that of the existence of unalterable permanent 

 magnets. And the idea of constructing measuring-instru- 

 ments of an accurate character based on either principle is 

 fallacious. 



* There is evidently a discontinuity in this value, 

 t It is commonly assumed that 33 cannot exceed about 17,500 or 

 18,000 (Rowland, I. c.) JEnc. Brit art. Magnetism). 



