180 
MR. ARTHUR SCHUSTER ON THE 
would cause a retardation of about one hour if the amplitude of oscillation is that 
deduced from the barometric variation. If the amplitude in the conducting regions 
is greater, the effect of self-induction is correspondingly less, because a smaller 
conductivity would then be required to account for the magnetic change. 
A few words should be said on the uncertainties of the data which serve as a test 
of the proposed theory and which are derived from my previous calculation of the 
variation potential. In deducing that potential I was practically obliged to confine 
myself to the records of four observatories (Bombay, Lisbon, Greenwich, St. Petersburg), 
all four being situated in the northern hemisphere ; and the year 1870 being the only 
one for which records were available at all four stations, I had to base my calculations 
on the figures for that year. Unfortunately, 1870 was a year of unusual sunspot 
activity and the magnetic records for that year cannot be taken as quite normal. It 
is probable that if the average of a number of years were taken, the phases of the 
components and their relative amplitudes might be somewhat altered ; but I do not 
think that, as far as the averages for the whole year are concerned, the results of the 
present investigation would be materially altered. A renewed discussion is, however, 
very desirable, especially if observations in the southern hemisphere could be made 
use of. In my previous calculations I separated the summer from the winter months, 
and assuming what is known to be ajDproximately the case, each hemisphere to 
behave alike when the sun occupies corresponding positions, I was enabled to form an 
expression of the potential applicable for the whole world simultaneously. But this 
is admittedly a defective process, and in drawing the equipotential curves I was 
careful for this reason to use only the averages taken over the whole year and to 
make no attempt to separate the two hemispheres. Yon Bezold, who later, basing 
his calculations on my figures, effected this separation, is often quoted as having thus 
completed my own investigation, but his extension of my work, for the reason given, 
seems to me to be deceptive and to push too far the observed approximate symmetry 
in the two hemispheres. What I now regret, however, is that I did not divide the 
year into four parts instead of two, as Dr. Chree’s results seem to show that the time 
of equinoxes deserves special consideration. 
If the views here brought forward are correct, all peculiarities of the barometric 
variation should be reproduced in the magnetic effect, though we must remember 
that the converse does not hold, and that peculiarities of the magnetic effect 
depending partly on variations of electric conductivity need have no counterpart in 
the barometric changes. Thus the greater amplitude of the magnetic variation 
between summer and winter has already been ascribed to increased conductivity 
of the atmosphere during the summer. The close relationship between the two 
phenomena is confirmed by the increased amplitude observed in both near the time 
of equinoxes. The diurnal period of barometric pressure is known to have maxima 
at these epochs, and the valuable researches of Dr. Chree have shown that these 
maxima are also found in the diurnal variation of the magnetic element. If we take 
