280 REMARKS ON THE TEMPERATURE 
REMARKS ON THE TEMPERATURE COEFFICIENTS 
OF MAGNETS. 
BY G. T. KINGSTON, M.A. 
‘DIRECTOR OF THE PROVINCIAL MAGNETIC OBSERVATORY, TORONTO. 
Ir has been long a matter of notoriety, that the correction to the 
scale-reading of a force magnetometer which is due to a change of one 
degree in its temperature, and which is derived from a comparison of 
the changes in its scale-readings with those of the attached thermo- 
meter, is often greatly at variance with ( z): the value of the correc- 
tion in which q is the temperature coefficient of the magnet found by 
the ordinary hot and cold water experiments, and & the scale co- 
efficient. The discrepancy has been attributed to the action of changes 
of temperature on the supports and appendages of the magnet when 
in adjustment, independent of, and in addition to, the alteration that 
such changes effect in the magnetic moment of the magnet; and as 
these effects cannot be’ determined a@ prior, it is safer, instead of re- 
lying on the experiments, to adopt the practice now almost general, 
of deriving the temperature corrections from the recorded observa- 
tions with the instrument and its thermometer, either by grouping 
them in seasons, as explained in page ii. of the Introduction to the 
third volume of the Toronto Observations, or by comparing groups of 
scale-readings at intervals of a few days, as in pp. xxii. and xxiv. of 
the same volume. 
Since, however, the first of these methods requires an unbroken 
series of two or more years, and the second a series of a year in 
length at least, it may be worthy of consideration whether the method 
of obtaining the correction from the temperature experiments cannot 
be made to yield results in sufficient accordance with the truth to 
serve the purpose of provisional reduction, or to meet the case in 
which the bifilar is needed as an auxiliary instrument to aid m the 
reduction of the absolute determination of the horizontal force, where 
since the range of temperature during the observations is small, the 
effect of a small error in the correction to the bifilar readings will be 
of less moment. 
Let (m) be the magnetic moment of the magnet whose temperature 
coefficient (g) is sought, and which is placed as a deflector with its 
