In 1761 Harrison produced a watch which was tested on a British ship in a trip to the West Indies in that and the succeeding year, and which proved to be a time-keeper of hitherto unexampled accuracy. The inventor had calculated that the watch, when carried into the tropics, would vary its speed by one second per day with each average rise of ten degrees of temperature. Making allowance for this predicted alteration, it was found that the watch was far within the limits of variation allowed by the conditions of the test above outlined. It had varied indeed only five seconds during the journey across the ocean. On the return trip the watch was kept in a place near the stern of the ship, for the sake of dryness, where, however, it was subjected to a great deal of joggling, which led to a considerably greater irregularity of action; but even so its variation on reaching British shores was such as to cause a maximum miscalculation of considerably less than thirty nautical miles. Although Harrison seemed clearly enough to have won the prize, there were influences at work that interfered for a time with full recognition of his accomplishment. Presently he received half the sum, however, and ultimately, after having divulged the secret of his compensating balance and proved that he could make other watches of corresponding accuracy, he received the full award. Minor improvements have naturally been made in the watch since that time, but the essential problem of making a really reliable portable timepiece was solved by the compensating balance-wheel of Harrison. The ship's chronometer of to-day is merely a large watch, with an escapement of particular construction, mounted on gimbals so that it will maintain a practically horizontal position. Modern ships are ordinarily provided with at least three of these time-keepers in order that each may be compared with the others, and a more accurate determination of the time made than would be possible from observation of a single instrument; inasmuch as no absolutely accurate time-keeper has ever been constructed. Two chronometers would obviously be not much better than one, since there would be no guide as to whether any variation between them had been caused by one running too fast or the other too slowly. But with a third chronometer to check the comparison, it is equally obvious that a dependable clue will be given as to the exact time.