We have already alluded to the difference of level between the Atlantic and Pacific waters. It is well known that the currents in the Spitzbergen and Greenland seas is to the southward, and that Parry, in his attempt to reach the pole, was foiled by this very current, frequently setting him back in twenty-four hours more than his party could travel in the same time over the ice. Through Baffin's and Hudson's bay the northern waters are also continually bearing their frozen freight southward. We are, therefore, entitled to ask, what supplies this immense drain? Behring's straits are only about sixty miles wide, and twenty-five fathoms deep; the supply, therefore, through this channel is totally inadequate, yet there is no other channel into the Arctic sea where the current is inward. We have already explained the reason why the current through Behring's straits is an exception to the general rule, yet still confirming the principle by referring it to the configuration of the land enclosing the Pacific ocean. The whole south Pacific lies open to the pole, and the inertia of the immense mass of mobile waters pressing northward, and continually contracted by the form of the American and Asiatic coasts, is not balanced by a contrary impulse of the waters of the north Pacific, inasmuch as this ocean becomes narrower as it extends northward, and the only passage to the frozen ocean is through the narrow straits of Behring. The axifugal force of rotation due to the northern waters is, therefore, overborne by the vast preponderance due to the southern waters, and, hence, the northern Pacific may be considered as relatively at a higher level, and there will be a current northward through Behring's straits, as we find it. The same cause accumulates the waters under the equator, thus giving a higher level to the Pacific than to the Atlantic at the isthmus of Panama, where the difference of level is found by actual measurement to be five or six feet. This fact has never before been explained; but the cause is too obvious to admit of question.