So much attention is now being paid to the determination of high temperatures in industrial processes where a rough guess was formerly considered good enough for practical purposes, that it is well to call attention to the two different methods of measuring temperature electrically, namely, the e.m.f. method and the resistance method. These are totally different and depend upon different properties of metallic substances. The e.m.f. method depends upon thermoelectric action. According to the electronic theory, if a certain metal, say copper, possesses a given average number of free negative corpuscles per cubic centimeter, while at the same temperature iron possesses a larger number of such free corpuscles, then, when these two metals are brought into contact the iron will tend by surplusage to diffuse fret corpuscles across the mutual bounding surface at the junction, and the copper will be made negative by the reception of negative electricity, while the iron will correspondingly become positive. Since the rate of diffusion of free corpuscles increases in each metal with the temperature, so, also, will the e.m.f. of the junction. In order to have a resultant thermal e.m.f. in a closed metallic circuit, a difference of temperature is, of course, necessary between the two junctions. Conversely, by the measurement of this resultant e.m.f., the difference of temperature becomes known. The resistance method depends upon the fact that metals increase in resistivity with temperature. According to the electronic theory, the mean free path of the corpuscles in metal diminishes as the temperature rises. This has the same effect on the electric conductivity as though the length of path remain unchanged, but the number of free corpuscles diminished. But however we try to form a mental picture of the action, the fact is perfectly definite that at the standard temperature of melting ice, pure metals in crease in resistivity about four-tenths of one per cent, per degree Centigrade. Moreover, the rate of increase follows a straight-line law with respect to temperature up to at least 100 degrees C. For high temperatures there appears to be a deviation from the straight-line law, in some metals at least.

An interesting paper on the measurement of temperature by electrical means was read at the Milwaukee Convention of the American Institute of Electrical Engineers by Edwin F. Northrup. The paper is addressed almost entirely to the resistance method, and suggests a number of important details in the technique of the manufacture, use and maintenance of resistance thermometers, such as go to make the difference between satisfaction and dissatisfaction in their application. In the first place, a special brand of nearly pure platinum is recommended for its uniformity and relatively high temperature coefficient. Then precautions have to be taken to prevent the platinum from dissolving or absorbing metallic vapors at high temperatures. Pure nickel wire is also recommended for moderate temperature thermometers. The beautiful principle of the Kelvin double Wheatstone bridge is also recommended for use with low-resistance thermometers. In such cases the resistance coils can be made very compact, and of dimensions comparable with those of a thermo-junction. For direct reading purposes a special form of D'Arsonval differential galvanometer, called a ratiometer, is described, in which the control is magnetic instead of elastic. That is, the coils lie flat in a dissimmetrical magnetic field, such that the differential magnetic actions. cause the movable system to seek a definite magnetic displacement. The three leading-in wires are arranged to exert jointly a negligible torque about the axis of pivot rotation. Some of the results reported with the various types of instruments described constitute jointly a valuable research on temperature measurement and its automatic recording, and will prove directly useful to those responsible for the operation of plants in which it is desirable to maintain temeperatures of a uniformly high degree. The enormous waste of fuel caused by overheating, apart from defective characteristics of the products of high-temperature operations not under close thermometric control, make this subject one of much technical importance.-"The Engineering Record."