tion, except with a variation in the intensity of the vertical component of terrestrial magnetism.
In making the observations, care must he taken that the temperature be kept as nearly constant as possible, a magnet losing about one ten-thousandth part of its power for each degree Fahrenheit of increase in temperature; also, that no magnetic bodies are present to influence the magnets; that the instruments be secured from all mechanical interference; and that all the adjustments be true.
Let us see how these requirements are met. The observatory has been built underground, and has double walls and roof, there being a space of two feet between the outside and inside walls. A differential thermometer placed within shows a daily variation in temperature of but 1° or 2° Fahr. In case of artificial heat being required, a brick stove has been built, with which it would be impossible to cause a sudden change in temperature.
The observatory has been placed without the influence of iron water or gas pipes, and in its construction and furnishing no iron has been employed, all metallic supports, mountings, etc., being of brass, copper, or zinc. The reservoirs of the lamps used have to be taken outside to be filled, since to bring a so-called tin oil-can within the observatory would seriously disturb the instrument. As before mentioned, visitors to the observatory leave outside whatever of iron they may have about them. An abnormal variation in the movements of the magnets at the Key West Observatory is thought to have been caused by the landing of some heavy guns in the vicinity, and their subsequent transportation past the observatory.
The mounting of the instruments upon heavy blocks of stone and their close incasement reduce the chance of mechanical interference to a minimum.
To still further guard against errors of observation, there are special adjustments in the several instruments.
The magnet of the declinometer is suspended by a skein of one hundred fibres of silk, the utmost pains being taken to reduce the torsion to a minimum. The length of skein is at least three feet, so that any residual torsion has the less effect. It seems impossible, however, to entirely get rid of this disturbing element. The records of the Key West observations show that the torsion of the suspension skein changed rapidly during the first five months after the suspension of the magnet, and did not become constant even after six years.
As variations in temperature do not affect the direction of the line of action of the magnetic force, no temperature adjustment is required for the declinometer.
The magnet of the bi-filar magnetometer is rigidly connected with a small glass rod of the same length as the magnet. Over the ends of this rod slip two zinc tubes, of such length as to reach within about five millimetres of its centre. At the inner end of each tube is at-
