1886.J Electromagnets and the Magnetisation of Iron. 495 



17,500 units. For pure iron he thought it might reach 18,000, or 

 even go above that.* 



Now, the magnetic force used in Rowland's experiments was very 

 small. The highest value — that corresponding to the point X in the 

 curve, fig. 2— being only 65 C.G.S. units. The imaginary part of 

 the curve from X to C therefore corresponds to values of |j, ranging 

 from 65 to infinity. A part of this exceedingly wide gap is filled by 

 my experiments, which include values of |p up to 585. A curve 

 constructed from the fourth and fifth columns of Table II, showing 

 the relation between ja and §5, is given on a small scale in fig. 2, 

 beside that of Rowland. It corresponds, of course, only with a 

 portion of Rowland's curve, the lowest value of § included in it 

 being 7390. But it seems to throw much new light on the subject. 

 Beginning with a rapid descent it turns aside soon after the limit of 

 Rowland's observations has been passedj and ultimately when §J= 

 19,800 it has become almost parallel to the horizontal axis. The 

 latter portion of the curve, from § = 15,100, is given on a larger scale 

 in fig. 3. We may conclude then that if § has any ultimate limit at 



* In his second paper, however, he says that we have at present no data for 

 determining whether $ attains a maximum or only $. 



