LINES OF MAGNETIC DECLINATION IN THE ATLANTIC. 
185 
From the small amount of the mean value of A, we may infer that the iron which 
affected the compass was distributed systematically, or nearly so, on either side of the 
midship line ; the variations in the values at different stations are greater than could 
be wished, but they have no regular appearance, and may probably be due to acci- 
dental circumstances, which in such experiments cannot possibly be wholly guarded 
against. Considering the small mean value of the coefficient, and the extent of its 
variations on the different occasions, we may dispense altogether with its further 
consideration. A similar remark will apply to E; since if its mean value were 
employed, the maxitnum effect on the correction would in no instance exceed two 
minutes. But it is otherwise with regard to D, which has a very sensible value in 
respect to the whole amount of the correction, especially in low latitudes ; and the 
deductions in regard to it are tolerably consistent at the different stations ; I have 
taken its mean value at -\- 22 '. 
For the variable coefficients B and C, we have not the advantage of possessing the 
experimental determinations at sea, which have been pointed out as possible to be 
made on future occasions with the deflecting apparatus ; and we must therefore obtain 
these also from the observations in harbour in the best manner that circumstances 
will admit. 
Commencing with B as the more important, we have the following values at the 
four stations for the passages between which the corrections are required. (The 
values are expressed by the sines of the respective arcs) : — 
Dip. B. 
Gillingham, September 20, 1839 -1-69 05 -l-‘0675 
Port Praya, November 18, 1839 -4-45 32 -l-'0324 
St. Helena, February 8, 1840 —20 06 -l-'0073 
Cape of Good Hope, April 4, 1840 —53 02 —•0219 
If from these values of B we seek intermediate values corresponding to interme- 
diate dips or times, we are obliged, for the reasons already stated, to have recourse to 
some more or less arbitrary supposition. It has been already shown that the inter- 
mediate values cannot be computed directly from the observations of the dip ; and if 
the explanation which has been proposed be correct, it may not be unreasonable to 
regard the variation of this coefficient as a function of the time elapsed rather than of 
the change of dip. In two of the three passages at least, viz. from the British Channel 
to Port Praya and from St. Helena to the Cape, this might be the more safely assumed, 
because the ship’s progress with respect to the terrestrial dip was uninterrupted, in 
the first case to diminishing north dips, and in the second to diminishing south dips. 
In the first case we have a change in B of *0351 in fifty-nine days, or *00059 per diem, 
and in the second of '0292 in fifty-six days, or *00052 per diem. In the voyage from 
Port Praya to St. Helena the progress in respect to the change of dip was uninterrupted 
from the period of departure from Port Praya on the 21st of November 1839 to the 
2nd of January 1840, the dip having diminished in that interval from -1-45° 32' at Port 
Praya to between —29° and —30°. But from the 2nd of January the ship, in beating 
