360 
REPORT — 1854. 
and a difference of longitude amounting to nearly half the circumference of 
the globe. It is difficult to resist the conclusion that an effect so general 
and so simultaneously produced points to a direct action of the sun upon 
the magnetism of the earth. 
Now as to the precise epoch when the phenomena pass from the repre¬ 
sentation which is given by one of these curves into the very dissimilar phe¬ 
nomena of the oilier semiannual curve: — The separation of March which h 
in the one group from April which is iu the other, and of September in the 
one group from October in the other, suffices to point generally to the suns 
passage of the equator as the epoch of transition. Hut by taking our mean 
representations for shorter and more critical periods than monthly one, 
(seeing that the equinoxes occur nearer the middle than the end of their 
respective months,) we arc able to press the identification still closer, aud to 
ascertain that the mean of the three weeks from the 1st to the 21si of Sep¬ 
tember, being the three weeks immediately preceding the day ot the .vp- 
teraber equinox, scarcely differs sensibly at any hour of the day from the 
semiannual mean of the preceding half-year, and consequently that the phe¬ 
nomena of that semiannual group have undergone no sensible change whtu 
a mean is taken corresponding to the tenth or eleventh day before the up 
nox. On the other hand, when a mean is taken corresponding to the three 
weeks succeeding the day of the equinox, namely the last nine days in Sep¬ 
tember and the early days of October, of which the mean corresponds to me 
10th or Uth day after the equinox, the transition from the character ol 
the preceding six months has not only commenced, but has already advanced 
very far towards its completion; and by the middle of October is quite com- 
plete, as the line corresponding to the middle of October exhibits no trace 
whatever of the semiannual characters which ten days before the equuiox 
had undergone no modification. 
Io recapitulate; — the mean line corresponding to the ten or eleven »)* 
beforelhe equinox exhibits no trace whatever of the commpnceroentofacba»gf* 
1 he mean line corresponding to ten or eleven days after the equinox exli‘ lb 
the change as having commenced, though not yet complete. The mean 
corresponding to twenty-five days subsequent to the equinox (viz. the ™ 
hue for October) exhibits the transition fully completed. Such are the 4 
at the September equinox. A similar investigation in the case ol the Ja |( 
equinox shows in the same way ho trace of change ten days before the < T l 
nox; and the change commenced, hut incomplete, at an equal perioiU 
the equinox. Apparently the progress of the change is somewhat w* 
ardy in the March than in the September equinox In both " 
have the indication that the commencement of the change synchro^; 
wnh the equinox, but that a certain interval of time, a few days more orkj 
rGt * u ‘ re( ? . for , lts completion. This appears analogous to the change nW* 
is operated in the induced magnetism of a ship, when she changes her r 1 ' 
graphical portion and by so doing changes her position also relatively* 1 
It hu d hZn e F aUB S f u 1 ” 0 ! 1 itt the easc of tl>e ship is the terrestrial magnet^; 
beirint. ° und that in such cases the induced magnetism ot the > f 
tervalVf 0 timers?• ■" f ? chttn 8 e iu her Position commences; but iWJ* 
ronnectld Wit h » < variabl V P«“tly with the rapidity and other 
the iron she emu *• cb “ n 8 e of position, and partly alsowiththedegreem v 
the change is comnWrl 5US ° e P tib . lc of immediate magnetic change) 
its inducing cause *. ,nduced magnetism is again u accoid ^ 
# ' * e have no reason to suppose that the effect o 
ibid. 1846, art? xviibnn* su i P ,o' l® 2 ,’ i. 5 ^ ; ibid - 1844, art. vii. pp. 88, 99, and lh' 11. 
