ATMOSPHERIC MAGNETISM — ST. HELENA — CUMULATIVE EFFECTS. 
Ill 
3051. At 1 o’clock the needle begins to turn from extreme east, i. e. sooner than 
in the former months, because the magnetic meridian is sooner passed, and follows 
the sun until 4’*, when it stops, and then the evening or night action due to cold ap- 
pearing in the east, carries the needle back eastward till 10’', and then as it ad- 
vances in the quadrant lets the needle return back again (3004.) until between 12’' 
and 13’', when the latter is in its mean position. The cold region then appears to 
draw it westward until 16’', when its distance increasing it releases the needle, and 
lets it return east until 18 o’clock, w'hen the latter is still west of its normal posi- 
tion, and then the sun region rising up, helped perhaps by the cold that immediately 
precedes it, which is probably now over or beyond the magnetic meridian, sets it 
toward the west prior to the sun-swing. 
3052. In December and January the sun is south of the station. This makes no 
difference in the general character of the curve for these months ; nor should it accord- 
ing to the hypothesis, except in this point, that though the sun is very near to St. He- 
lena and has the cumulative effects of the preceding months (3050. 3053.) added to 
its own effect at the time, still it is in weaker parts of the quadrant, and whilst in the 
chief segment is almost up in the corner and near the place where the two neutral 
planes cross each other; hence its effect ought to be less, and so it is; for the sun- 
swing of November and February is larger than that of December and January. 
The sun-swing happens in December at the same time as for October, though in the 
latter month it crosses the magnetic meridian after, and in the present before mid- 
day: still there is only half an hour’s difference from one to the other, and the ob- 
servations are perhaps not close enough to allow one to separate its peculiar effect 
out of an interval of four hours. Besides, accumulative causes may interfere : the 
places of the December curve are altogether a little more west than those for Oc- 
tober. 
3053. The cumulative effect of preceding months is very important and well-shown 
at St. Helena (3050.). Thus, taking the September curve and comparing it with 
that for October or the following month, we have a great difference of a certain 
kind ; then again comparing September with the month in which the sun is returning 
from the southern tropic instead of proceeding to it, and has arrived at the same 
position as it had in October, another striking difference appears. March is tlie 
nearest month for the second comparison. Up to 20’' its curve changes like the Octo- 
ber curve, but the upper end of the needle is all the time about half a minute east of 
its place in October. At 20’' the needle in October begins to swing from west, and 
reaches extreme east at 24’': in March it westens until 2^, then returns and reaches 
extreme east at U ; so that the swing is an hour later, and during that time the end 
is from half a minute (of space) to a minute more west than in October. This differ- 
ence I believe to be due to the cumulative effect of the months between October and 
March, during which time the heat has been diminishing in the northern hemisphere, 
and increasing in the south. Similar results in other months make it probable that 
