78 
PROFESSOR HUGH L. CALLENDAR OX 
(‘ Phil. Trans./ A, 1900, vol. 194, p. 59), though they do not appear to have appre¬ 
ciated its full significance. In order to reduce the temperature correction, they 
selected a resistance box with manganin coils made by Messrs. Crompton in pre¬ 
ference to one with platinum-silver coils annealed at a red heat after the pattern 
made by the Cambridge Instrument Company. In attempting to determine the 
temperature-coefficient of their box by raising the temperature for a short period, 
they noticed that the resistances did not return at once to their original values on 
cooling. They also observed a considerable increase of resistance in all the coils, 
which was most rapid at the outset, and which they attributed to the effect of 
recovery from the soldering. We have seen that No. 4086 underwent a similar 
increase of resistance at the outset. -Messrs. Harker and Chappuis do not seem to 
have observed the long-period effect of hysteresis, which is more insidious. The 
correction for this cannot be accurately applied, but it appears that it might explain 
some of their errors of observation. 
In order to reconcile the observations of Barnes and Fraser with each other 
and with those of Glazebrook, it is necessary to suppose that the value of the 
temperature coefficient '000254 assumed by Barnes for the McGill Pt-Ag standard 
ohms is a little too small. According to my own experiments with these particular 
standards, in 1893 and 1894, made by heating the coils in a water-bath, the value 
should be ‘000275. I tested only two of the coils, but it is clear from the agreement 
of the relative differences observed by Barnes and Fraser that the temperature 
coefficients of all the coils are nearly the same. The changes in the relative 
differences observed by them are in no case greater than can be explained by the 
uncertainty of temperature, as the coils were not immersed in a water-bath in their 
comparisons. 
Adopting the value ‘000275 for the temperature coefficient of the 10 Pt-Ag 
standard ohms, we find the following corrected results in place of (5) and (6), 
Barnes, No. 4086 at 21°‘74 C. = 1-00048 ohms .... (8). 
Fraser ,, „ „ 14°‘8 C. = 1‘00035 ohms .... (9). 
The difference between these two values gives a temperature coefficient for No. 4086 
of ‘000020, which agrees with their other results supposing that Fraser’s obser¬ 
vations were taken in the basement within a few days of those of Barnes. In 
comparing these results with those of Glazebrook, it must be remembered that 
owing to the solidity of the construction and the perfection of the heating arrange¬ 
ments in the Physics Building at McGill College, the temperature of the Heat 
Laboratory very rarely fell below 15° in winter (excejff in a few corners near the 
windows), or rose above 22° in summer. No. 4086 was never exposed at Montreal to 
so great changes of temperature as it experienced in the temporary observing room 
at Kew. The results of the comparisons as regards the absolute values of the 
