it forward; but there was this remarkable distinction,—in the northern hemisphere it was the north end of the needle which was attracted, and in the southern hemisphere it was the south end. In the instance off the Start before cited, when the ship's head was West, the north end of the needle had been drawn forward, or to the left of North, nearly 4°, and the west variation thereby increased to 29½°; with the head East, it would be drawn to the right of its natural position, and the variation diminished to about 21½°; but at North, the attraction in the ship was in the same line with the magnetic poles of the earth, and would therefore produce no change. The same thing took place at South, for the two attractions were still in the same continued line, though on opposite sides of the compass; and throughout the voyage I found, that variations taken with the head at North and South agreed very nearly in themselves and with the observations on shore near the same place, when such observations were not affected by local attractions.
But although the errors were always the same way in the same hemisphere, when the head was at West, and when it was East they were always the contrary, yet the quantities varied with the situation of the ship, being greater in high, and less in low latitudes; and yet they did not increase and diminish in proportion to the latitude. After much examination and comparison of the observations, and some thinking on the subject, I found that the errors had a close connexion with the dip of the needle. When the north end of the needle had dipped, it was the north point of the compass which had been attracted by the iron in the ship; and as that dip diminished, so had the attraction, until, at the magnetic equator, where the dipping needle stands horizontal, there seemed to be no attraction. After passing some distance into the southern hemisphere, and the south end of the needle dipped, our observations again showed errors in the compass; but the west variation was now too great when the ship's head was eastward. These errors increased as the dip augmented; and in Bass' Strait, where the south dip is nearly as great as the north dip in the English Channel, the attraction produced almost as much error as when we left England, but it was of an opposite nature. On turning northward again, along the east coast of New South Wales, the dip of the south end of the needle and the attraction of the iron upon the south point of the com-