As the results obtained in the work relative to rigidity have the dual character indicated, it is convenient to touch upon their economic and scientific bearings consecutively. In studying hardness it was necessary to obtain some exact knowledge about temper, and to discover a means of distinguishing with precision different states of tempered hardness in steel. It was desirable, furthermore, to express the relations of the different iron carbides to each other from a physical point of view, and to construct a diagram for the classification of these products. This part of the work was carried out by Dr. Barus and Dr. Strouhal.1

The investigation began with results previously obtained by Barus, which showed (a) that the electrical resistance of steel could be increased considerably over threefold by changing its temper from soft to hard, and (b) that its thermoelectric power was subject to like variation. Thus it was made possible to construct a scale of tempers for steel by which the hardness of any given sample could be expressed with an accuracy of one in one thousand. Availing themselves of this method of experimental attack, the authors proceeded to an exact development of the laws by which any given degree of temper may be produced. They showed definitely in what manner the diminution of the hardness of a steel rod depends on the temperature to which the rod is exposed and on the time of exposure. An important relation having been discovered in the interdependence of the two electrical constants employed, it became feasible to define the properties of the steel rod whose hardness in the given scale is zero.

After discussing a number of subsidiary data, such, for instance, as the effect of temperature on the electrical constants when temper does not vary, the corresponding variations of hardness and specific volume in steel, the analogies which obtain between tempering and alloying and between temper and a mechanical strain, the effect of magnetism in changing the electrical scale of hardness, etc., the authors made another important application of their method by establishing the exact Conditions under which the magnetization of steel varies with

results are detailed in Bull. No. 14, U. S. Geological Survey, pp. 1-238, 1885.

temper. The effect of temperature on the magnetism of steel was similarly treated and rules were given for the construction of magnets of exceptional degrees of magnetic permanence. In a great variety of magnetic instruments it is essential that the magnets used shall have the greatest possible power of resisting the effects of wear and tear, as well as of atmospheric agencies, and that the result shall be secured at the least possible sacrifice in the amount of magnetization. Thus the practical construction of exceptionally permanent, highly magnetized magnets is an outcome of this research.

The authors also discussed a scheme for the physical classification of the iron-carbon products. This classification is based on the important observation that the difference between the electrical constants for the soft state and the hard state of a given iron-carbide increases with the number of states of mechanical hardness in which the sample is capable of existing. Thus, for wrought iron, which can not be tempered, the difference in question is nearly zero, and for cast iron it is smaller than for steel. This difference, however, would not necessarily imply the changes of mechanical properties which are observed in passing from iron through steel into cast iron, even in the soft states; and the electrical expression for this difference is also a continuous increase of the constants. Hence the ratios of the electrical constants for the hard and the soft states of any given iron carbide were selected as a basis of classification. Experiments then proved that on passing without break of continuity from pure iron to cast iron a unique iron carbide is always encountered in which the ratio of the electrical constants is a maximum. This singular iron carbide is characterized by the notable property of existing in the greatest possible number of mechanical states of hardness. The metal thus definable with almost mathematical precision is steel.

In a series of subsequent researches Dr. Barus and Dr. Strouhal developed the practical side of the inquiry with much greater fullness.' Thus the density of steel in its relation to

'Bull. No. 27 (pp. 30-61, 1886), Bull. No. 35 (pp. 1-62, 1886), and Bull. No. 42 (pp. 98-131).

14 GEOL- -10

As the results obtained in the work relative to rigidity have the dual character indicated, it is convenient to touch upon their economic and scientific bearings consecutively. In studying hardness it was necessary to obtain some exact knowledge about temper, and to discover a means of distinguishing with precision different states of tempered hardness in steel. It was desirable, furthermore, to express the relations of the different iron carbides to each other from a physical point of view, and to construct a diagram for the classification of these prodThis part of the work was carried out by Dr. Barus and

ucts.
Dr. Strouhal.1

The investigation began with results previously obtained by Barus, which showed (a) that the electrical resistance of steel could be increased considerably over threefold by changing its temper from soft to hard, and (b) that its thermoelectric power was subject to like variation. Thus it was made possible to construct a scale of tempers for steel by which the hardness of any given sample could be expressed with an accuracy of one in one thousand. Availing themselves of this method of experimental attack, the authors proceeded to an exact development of the laws by which any given degree of temper may be produced. They showed definitely in what manner the diminution of the hardness of a steel rod depends on the temperature to which the rod is exposed and on the time of exposure. An important relation having been discovered in the interdependence of the two electrical constants employed, it became feasible to define the properties of the steel rod whose hardness in the given scale is zero.

After discussing a number of subsidiary data, such, for instance, as the effect of temperature on the electrical constants when temper does not vary, the corresponding variations of hardness and specific volume in steel, the analogies which obtain between tempering and alloying and between temper and a mechanical strain, the effect of magnetism in changing the electrical scale of hardness, etc., the authors made another important application of their method by establishing the exact conditions under which the magnetization of steel varies with

The results are detailed in Bull. No. 14, U. S. Geological Survey, pp. 1-238, 1885.

each agency (heat, stress, magnetism, time, etc.) being considered in detail. The work showed it to be necessary to distinguish two species of molecular break-up, viz, that in which the molecules pass from configurations of less to configurations of greater stability, apparently without loss of their individuality, and that in which the transfer is accompanied by a disintegration of the molecules. The last of these cases is exemplified by the changes of temper in steel, whether brought about by change of temperature or by time. Thus glass-hard steel is always spontaneously softening, even at ordinary temperatures. This is proved by the changes of the electric constants of steel in the lapse of years. The other case of configurational break-up, namely, that in which the configurations are of a physical character, is apparently the more common. Examples properly belonging here were obtained in abundance by subjecting metals to intense tensile, compressional, flexural, torsional, and other strains, the presence of each of which in the metal was accompanied by marked decrease of viscosity. In spite of the fact, therefore, that the hardness and tensile strength of steel are notably increased by stresses or strains, the metal itself is none the less brought nearer the liquid state, because the strains introduce a greater number of instabilities than were present before stress was applied. On the other hand, slight mechanical strains (twisting, for instance) increase the viscosity, because the instability thereby broken down exceed the number evoked. This effect is therefore analogous to thermal annealing, which is always accompanied by increased viscosity.

At this stage of progress much time was spent in the endeavor to obtain concrete knowledge of the mechanism of viscosity. Thus the effect of mechanical strain on the carburation of steel and on the rate of solution of the metal was considered. The hydroelectric effect of permanent changes of molecular arrangement was measured, and also the amount of energy potentialized by such strains. The most interesting results obtained came from a study of the electrical resistance of glass under stress at temperatures above 100° C. Here stress produced a greater decrease of resistance than could be accounted for by change of dimensions, and

an increased molecular break-up induced by (dilatational) stress was therefore inferred. The results of stress and of increased temperature are similar. In view of the geologic bearing of the effect of stress on the chemical equilibrium of mixed silicates, the question was again attacked by a new method. Glass submerged in hydrocarbon oil was subjected to great pressure and the changes of electrolytic resistance were compared with the corresponding changes of pressure. This experiment confirms the above inference for strained glass. Compression increases the specific resistance and hence decreases the amount of molecular break-up. Researches of the kind here mentioned are exceedingly difficult because of the many sources of error to be avoided. Hence the question whether the permanent retention of strain by a body is accompanied by the formation of a new arrangement of molecules is still open, although the electro-chemical difference between unstrained and strained or even magnetized metal gives additional evidence in favor of the positive answer.

Notwithstanding the geologic importance of the relation of viscosity to pressure and temperature, the first systematic treatment of the subject by the artificial production of high pressures was that undertaken in the physical laboratory. A highly viscous liquid of pitch-like consistency was selected for study, with the object of defining the lines of equal temperature and the lines of equal pressure within a range of one or two thousand atmospheres. From these results the isometrics, or lines of constant viscosity, could be computed and some knowledge gained of the rate at which pressure must increase relatively to temperature in order that the viscosity of the body may remain unchanged. Supposing, therefore, that an aperture of fixed dimensions is given, the isometrics of viscosity express the condition of equal flow for all pressures and temperatures.

The body used in these experiments, however tenacious, must retain the properties of a liquid throughout, i. e., under constant conditions of pressure, temperature, and stress it must show constant viscosity as to time. In solids, however, viscosity under the same circumstances increases markedly with