DIURNAL VARIATIONS OF TERRESTRIAL MAGNETISM. 
5 
The seasonal spherical harmonics in the magnetic variations are also considered, and 
it is found that they are relatively twice or thrice as large in the lunar variation as 
in the solar variation, in comparison with the harmonics which persist uniformly 
throughout the year. This suggests that the main 12-hour oscillation in the solar 
case may produce seasonal harmonics which are partly neutralized by some other 
oscillation, such as that of 24-hour period, just mentioned. 
The phase relations of the latter oscillation present some difficulty, and it is 
questionable whether it is connected with the 24-hour barometric variation, of thermal 
origin, which is observed at the earth’s surface. The 12-hour barometric variation, 
on the other hand, is so much more fundamental in character that it is not un¬ 
reasonable to suppose that it persists even up to the high levels here contemplated. 
The magnetic data suggest that its proportional amplitude ( Sp/p ) diminishes upwards 
to the extent of one-half its surface value. 
The upper air currents responsible for the magnetic variations will have a heating 
effect which can be approximately calculated, and it appears that there is a possibility 
of the production, by this means, of an appreciable solar diurnal temperature and 
pressure variation in the conducting layer. It may be that the above-mentioned 
diurnal oscillation, peculiar to the solar diurnal magnetic variations, is to be accounted 
for in this way. 
The questions of phase raised by this discussion prove to be very perplexing. As 
the theory would indicate, the phases of the annual mean harmonics of various periods 
agree amongst themselves, both in the solar and lunar variations. But these phases 
seem to possess little or no relation with those of the solar and lunar semi-diurnal 
barometric variations at the earth’s surface. It would appear that the phase of the 
solar barometric variation diminishes with increasing height, by an amount which has 
been estimated at 80 or 90 degrees.^ But a much larger change of phase with 
height is required, affecting the lunar as well as the solar barometric variation, if 
these are to be brought into simple relation with the magnetic variations. 
The other principal remaining difficulty is that the diurnal North force variations 
do not agree at all well with those calculated from the potential function which 
represents the diurnal West force variations. This does not seem to be explicable, as 
G. W. Walker has suggested, by the presence of magnetic variations depending on 
the time of some fixed meridian. 
In order that the present paper may be the better understood, it has seemed well 
to indicate the existing state of the problem, by giving a brief critical account 
(§§ 2 to 6) of the work of the previous writers already named. In the course of this 
historical survey mention is made of the chief modifications of previous methods, and 
of previous conclusions, which are introduced in the present discussion. 
The burden of labour entailed by an investigation of this kind is very great, and could 
* This was not remarked on in Schuster’s second memoir, in which a measure of agreement seemed 
to be indicated between the phases of the barometric and magnetic variations. 
