of the time reckoned at the two meridians at the same absolute instant." If a chronometer be regulated to the time at any place A, and then transported to a second place B, and the local time at B, be determined at any instant, and at that instant the time at A, as shown by the chronometer is noted, the difference of the times is at once known, and that is the difference of longitude required. The principal objection to this plan is that the best chronometers vary. If the variations were constant and regular, and the chronometer always gained or lost a fixed amount for the same interval of time, this objection would disappear. But the variation is not constant, the rate of gain or loss, even in the best instruments, changes from time to time from various causes. Some of these causes may be discovered and allowed for in a measure, others are accidental and unknown. Of the former class are variations due to changes of temperature. At the Naval Observatory, chronometers are rated at different temperatures, and the changes due thereto are noted, and serve to a great extent as a guide in their use. But the transportation of a chronometer, even when done with great care is liable to cause sudden changes in its indications, and of course in carrying it long distances, numerous shocks of greater or less violence are unavoidable. Still, chronometric measurements, when well carried out with a number of chronometers and skilled observers have been very successful. Among notable expeditions of this sort was that undertaken in 1843, by Struve between Pulkova and Altona, in which eighty-one chronometers were employed and nine voyages made from Pulkova to Altona and eight the other way. The results from thirteen of the chronometers were rejected as being discordant, and the deduced longitude was made to depend on the remaining 68. The result thus obtained differs from the latest determination by 0.2.
The U.S. Coast Survey instituted chronometric expeditions between Cambridge, Mass., and Liverpool, England, in the years 1849, '50, '51 and '55. The probable error of the results of six voyages, three in each direction, in 1855 was 0.19, fifty chronometers being carried.
Among other methods of determining differences of time may be mentioned the observation of certain celestial phenomena, which are visible at the same absolute instant by observers in various parts of the globe, such as the instant of the beginning or end of an eclipse of the moon, the eclipses of Jupiter's satel-
