While the observations of the early navigators in respect to latitude were usually accompanied by errors, which were of no considerable extent, their determinations of longitude, when attempted at all, were almost always wide of the truth,[383]—so far, indeed, that their observations helped them but little then to steer their courses, and are of small assistance now to us in following their tracks. It happened that while Columbus was at Hispaniola on his second voyage, in September, 1494, there was an eclipse of the moon.[384] Columbus observed it; and his calculations placed himself five hours and a half from Seville,—an error of eighteen degrees, or an hour and a quarter too much. The error was due doubtless as much to the rudeness of his instruments as to the errors of the lunar tables then in use.[385] [Illustration: THE JACKSTAFF.] The removal of the Line of Demarcation from the supposed meridian of non-variation of the needle did not prevent the phenomena of terrestrial magnetism becoming of vast importance in the dispute between the Crowns of Spain and Portugal. It characterizes the difference between the imaginative and somewhat fantastic quality of Columbus’ mind and the cooler, more practical, and better administrative apprehension of Sebastian Cabot, that while each observed the phenomenon of the variation of the needle, and each imagined it a clew to some system of determining longitude, to Columbus it was associated with wild notions of a too-ample revolution of the North Star about the true pole.[386] It was not disconnected in his mind from a fancy which gave the earth the shape of a pear; so that when he perceived on his voyage a clearing of the atmosphere, he imagined he was ascending the stem-end of the pear; where he would find the terrestrial paradise.[387] To Cabot the phenomenon had only its practical significance; and he seems to have pondered on a solution of the problem during the rest of his life, if, as Humboldt supposes, the intimations of his death-bed in respect to some as yet unregistered way of discovering longitude refer to his observations on the magnetic declination.[388] The idea of a constantly increasing declination east and west from a point of non-variation, which both Columbus and Cabot had discovered, and which increase could be reduced to a formula, was indeed partly true; except, as is now well known, the line of non-variation, instead of being a meridian, and fixed, is a curve of constantly changing proportions.[389] [Illustration: THE BACKSTAFF.]