Much the most convenient means of measuring temperature is the common glass tube filled with mercury. This metal is chosen because a liquid, and because it varies extremely in bulk when warmed or cooled. Materials of parallel susceptibility are adopted for instruments which measure the intensity of magnetism or of electricity, the working core of the instrument being made of a substance highly responsive to magnetism or to electricity. A mercurial thermometer, for all its convenience, has its accuracy assailed on more sides than one. When the barometric pressure rises, the bulb is compressed; when the barometer falls, the bulb enlarges by virtue of the diminution in atmospheric pressure. Further, when its graduated tube is upright the mercury exerts a distending pressure which introduces error. At all temperatures the metal is giving off a vapor which has tension, in its upper ranges entailing marked inaccuracies. The glass itself of which the instrument is made, when of ordinary composition, spontaneously undergoes changes of volume. While this is a minor source of error it may be almost completely avoided by using a boro-silicate glass from the factory of Schott & Genossen, at Jena. Other substances than mercury are employed in thermometers with gratifying results. Hydrogen gas is found very suitable within the interval from -30° to 200° Centigrade. Pentane serves in temperatures reaching down to -180°. But it is in alliances with electricity that the measurement of heat has its broadest scope and utmost exactitude. It was long ago remarked that heating a metallic conductor increases its resistance to the flow of an electric current; to measure that resistance in a platinum wire serves, therefore, to measure its temperature. An instrument on this principle is the bolometer of the late Professor S. P. Langley, of Washington. Through a strip of platinum barely 1/500 inch in width, and less than 1/5000 inch in thickness, a current of electricity flows continuously. When radiation, visible or invisible, on occasion from a star, falls upon it, the strip when warmed by as little as one millionth of a degree duly records the fact. An instrument, modified from the Crookes radiometer by Professor E. F. Nichols of Columbia University, New York, is more sensitive still. An exhausted hollow metal block has a window of fluorite, a mineral transparent to ether vibrations of a long range of frequencies. Suspended inside the block is a fine quartz fibre supporting a horizontal bar, at the ends of which are attached thin plates of mica, blackened on one side. Rays passing through the fluorite window strike the blackened side of the mica, which is parallel and opposite to it. The resulting rise in temperature causes the vane to revolve against the torsion of the quartz fibre. The angle of torsion when thermal equilibrium is reached, measures the intensity of the incident radiation.