80 
PROFESSOR HUGH L. CALLENDAR ON 
accurate work, which is not I think sufficiently appreciated. In the second place, it 
is possible that at some future time the absolute values of the electrical standards 
may be more accurately determined, in which case it is desirable that the apparent 
discrepancies in the comparisons of the standards employed in the present investiga¬ 
tion should be clearly explained, in order that the correction may be suitably applied. 
The above explanation of the discrepancies observed in the comparisons of the 
manganin standard No. 4086, was submitted to Mr. Glazebrook for his approval in 
January, 1901. He considered, however, that the measurements in his possession 
at that time were not conclusive evidence of hysteresis, as four other manganin coils 
of German make, which had been tested at the same time, had shown similar though 
not identical variations. He was himself inclined to explain these irregularities by 
uncertainties of temperature or of temperature-coefficient of the Pt-Ag standards, 
especially at the higher temperature in the July observations. The large value of 
the coefficient ‘000047 for the manganin coil, deduced from the July and December 
observations, might be explained by supposing that the value of the coefficient 
assumed for the Pt-Ag coils was too large. This, however, would not explain the 
smaller values of the coefficient for the mana;anin coil deduced from the observations 
from day to day, in which the same value of the Pt-Ag coefficient was assumed. 
In order to settle the question, a further series of comparisons of No. 4086 was 
made in June and July, 1901, at the National Physical Laboratory, which com¬ 
pletely confirmed the observations made in the previous July, showing that they 
were not merely accidental results due to erratic change of No. 4086 or to uncertainty 
in the temperature conditions prevailing at that time in the observing room. 
As a further test of the possibility of hysteresis theory already advanced, a special 
comparison was made, in which No. 4086 was heated through a small range of 
temperature and allowed to cool slowly, being tested against a manganin standard 
kept at a steady temperature throughout the comparison. This test gave a very 
small coefficient, but the curve representing the observations gave unmistakable 
indications of hysteresis. It was confirmed by a second test, in which No. 4086 was 
cooled, and allowed to warm up slowly. The curve representing the observations in 
this case had the opposite curvature. The two curves combined gave a figure 
similar to one of the familiar cycles of magnetic hysteresis. These effects could not 
be explained as due to a real lag of temperature, since the rate of change of tempe¬ 
rature was very slow, and the case was filled with oil in direct contact with the wire 
and also with the thin metal tube on which the wire was wound. With this direct 
evidence of the existence of hysteresis over short periods and small ranges of 
temperature, it becomes easier to admit the possibility of larger variations of long 
period, such as those indicated by the July and December comparisons with the 
platinum-silver standards. 
My assistant, Mr. Eumorfopoulos, is now engaged in testing the conditions of 
existence of the long-period effect, by a differential method designed for the purpose. 
