EXPEEIMENTAL EESEAECHES ON DEAWN STEEL. 19 
of one relatively to the other. Thus there is a small value of the negative coefficient 
in the calculated curve between intensities of 800 and 900, and a maximum about 
an intensity of 700 ; the counterpart to. these occur in the experimental curve 
between intensities of 400 and 500, and at less than 100 respectively. Still more 
interesting is the large displacement of the zero coefficient. In the curve the 
zero state occurs about an intensity of 560, and in the a curve probably at an 
intensity of only a few units. Although I have attempted to obtain experimentally 
the zero temperature coefficient in iron, I have not succeeded unmistakably, partly 
because of the general difficulty of working at very low intensities, and partly 
because of the special difficulty of clearing out all traces of pre-existing magnetisa¬ 
tion, which is a very necessary precaution, and of operating in a field of no force. 
But I have ascertained that at a very feeble intensity the negative coefficient becomes 
decidedly less negative, and that the curve tends towards zero at some extremely 
low magnetisation. 
Ewing, however, has obtained at a very early stage in the magnetisation of iron a 
positive, and at a higher but still a very low intensity a zero coefficient.* Although 
his experiment was not performed on residual magnetism alone, as the vertical 
component of the earth’s force was always in operation in such a way as to tend to 
increase the magnetisation, and at low intensities its efiect would be considerable, 
yet there is little doubt that at some very low residual intensity iron yields a 
zero coefficient. There is thus a satisfactory correspondence between the results 
calculated from the relation of the curves of residual intensity when hot and when 
cold and the results exjDerimentally found for the temperature coefficient of residual 
magnetism. Every magnet therefore may have a positive, a negative, or a zero 
coefficient, unless the hot and cold curves happen to be coincident throughout. 
The changes which take jilace in the magnitude of the permanent loss and gain 
of magnetism due to a series of headings and coolings, are shown graphically for both 
drawn steel and iron on the lower part of Diagram V. 
16. The experiments I have selected for description throw light, I think, on many 
of the numerous results which have been published on the effects of cyclic clianges 
of temperature on magnetism, f and also afford some guiding rules for the construc¬ 
tion of magnets of high permanence and wuth small temperature coefficients. 
* ‘Eoy. Soc. Phil. Trans.,’ vol. 176, p. 633. 
t Eeferences to pajjers on the “ Influence of Changes of Temperature on Magnetism ” and allied 
subjects:— 
Faraday, ‘Phil. Mag.,’ vol. 8, p. 177, 1836; Kater, ‘Eoy. Soc. Phil. Trans.,’ 1821; Barlow and 
Bonnycastle, ‘Eoy. Soc. Phil. Trans.,’ 1822; Eiess and Moser, ‘Pogg. Ann.,’ vol. 17, p. 425, 1829; 
Kupfeer, ‘Kastner’s Archiv,’ vol. 6; Lamont’s ‘ Magnetismus’; Scoresby, ‘ Edin. Phil. Trans.,’vol. 9, 
p. 254; Wiedemann, ‘Pogg. Ann.,’ vol. 103, p. 563, 1858; Mauritius, ‘Pogg. Ann.,’ 1863, ‘Phil. 
Mag.,’ 1864; Gore, ‘Phil. Mag.,’ 1869 and 1870; Gordon and Newall, ‘Phil. Mag.,’ vol. 42, p. 335, 
1871; Whipple, ‘Eoy. Soc. Proc.,’ 1877; Eowland, ‘Phil. Mag.,’ vol. 48, p. 321, 1874; Faye, ‘C.E.,’ 
vol. 82, p. 276, 1876; Gaugain, ‘C.E.,’ vol. 80, p. 297, vol. 82, p. 685, vol. 83, p. 896, vol. 85, pp. 219, 
D 2 
