20 
MR. J. REGINALD ASHWORTH; 
When the magnet is short relatively to its thickness, the self-demagnetising force 
will have so iireponderating an influence, that it will be advisable, in order to reduce 
its effect to a minimum, to choose a material of small susceptibility ; a hard steel is 
thus preferred. If, however, the magnet is long and thin, attention must be paid 
chiefly to the quality and treatment of the material, so that it may develop that 
condition in which the hot and cold curves of its magnetisation aie sepaiated as 
little as possible. 
And further, it has been shown that drawing influences the disposition of the 
hot and cold curves in such a way tliat it affords an effecti've method of regulating 
the magnitude and sign of the temperature coefficient. 
Time Tests 
Oil the Constancy of Magnets ivith Neghgihle Coefficients. 
17. At the conclusion of a previous paper"^ reciting experiments upon the consti notion 
of magnets with zero temperature coefficients, a brief note was added on the question 
of the constancy of the zero state with lapse of time. This is obviously important in 
the application of such magnets to the work of an observatory, and it has therefore 
received some attention. 
In May, 1897, a magnet was constructed of a piece of H 30 wire, 0T87 centiin in 
diameter, and from a previous series of experiments upon this kind of wire it wms 
calculated that it should be cut to a length of 8 centims., having a dimension ratio of 
42'6, in order to yield a zero coefficient. It was then magnetised and heated and 
cooled about twenty times so as to reduce the magnetism to a settled state. The 
intensity before heating and cooling was 474-4 c.g.s. units, calculating this here, as 
elsewhere, as the magnetic moment per unit volume and, alter heating and cooling, 
the intensity was 28-1 per cent, less; the coefficient, a, w^as very small and positive, 
its value being + 0-000015 per degree centigrade. This magnet was tested at 
intervals for the next three years, and its history is given in Table VIL, and also 
in the diagram constructed from this table (Diagram VI.). After the fiist test iu 
615, 1014, and vol. 86, p. 5.36; Jamix and Gaugaix, ‘C.R.,’ 1876, ‘Phil. Mag.,’ 1876; POLOXI, ‘Mied. 
Beibl.,’ 1878; Wassmutii, ‘Wien. Per.,’ 1880-02; Baur, ‘ Wied. Ann.,’ vol. 11, 1880^ Browx, ‘Phil. 
Mag.,’ vol. 23, pp. 293, 420, 1887 ; Ciieesmax, ‘Wied. Ann.,’ vol. 15, p. 204, vol. 16, p. 712; Barus and 
STi-rouilAL, ‘Wied. Ann.,’ vol. 20, p. 662, 1883; ‘Bulletin U.S. Geol. Survey,’ No. 14, 1885; Barus, 
‘Phil. Mag.,’Nov., 1888; Gray, . ‘ Phil. Mag.,’ vol. 6, p. 321, 1878; Bosanquet, ‘Phil. Mag.,’^vol. 19, 
p. 57, 1885; Caxoani, ‘ Atti della R. Acc. dei Lincei’ (4), vol. 3, p. 501, 1887 ; Morris, IX K., ‘Phil. 
INlag.,’vol. 44, p. 213, 1897 ; DuRWAim, A., ‘Am. Journal of Sci.,’ April, 1898 ; Pierce, 1>. 0., ‘Am. 
Joimial of Sci.,’May, 1898; Ewing, ‘Roy. Soc. Phil. Trans.,’ vol. 176, 1885; also Canton and Hall- 
strOm, Coulomb, Hansteen, Christie, Lloyd, Ciiree, Ac. 
* ‘Roy. Soc. Proc.,’ vol. 62, p. 210. 
