The time on board was evidently kept by hour-glasses of half or a whole hour. Thus under date of Tuesday, January 22, when homeward bound, his log reads: “They made 8 miles an hour during five glasses ... afterwards they went N.E. by N. for six glasses.... Then during four glasses of the second watch N.E. at six miles an hour.” But the reader must be cautious not to accept the speed given as conclusive. One of the greatest drawbacks to navigation in those days was the absence of any instrument which would record the speed through the water. The log had yet to be invented, and the mariner could only make a conjectural estimate of the ship’s speed by looking over the side and noting the time it took the bubbles to come aft from the bow, or by throwing a piece of wood overboard from the bows and noticing how long it took for the stern to be abreast of that object. Many a steamship traveller gambling on the ship’s speed does the same thing to-day; many a fore-and-aft sailorman with no patent log still employs a similar method. Columbus’s journal shows the kind of helmsman which he had to put up with. On September 9, when the ship’s course was west, the narrator on board wrote: “The sailors steered badly, letting the ship fall off to N.E., and even more; respecting which the Admiral complained many times.” On September 13 Columbus observed a variation in the compass. “On this day, at the commencement of the night, the needles turned a half point to N.W., and in the morning they turned somewhat more N.W.” For up till then no one had observed the variation of the needle. [Illustration: FIFTEENTH-CENTURY CARAVEL. Drawn from a woodcut after a delineation by Columbus in the Latin translation of his letter dated March 1, 1493, to Don Raphael de Sanxis (Treasurer of the King of Spain), in the Library at Milano. (See next plate.)]