CONTRIBUTIONS FROM THE JEFFERSON PHYSICAL LABORATORY, HARVARD UNIVERSITY. In this paper data are given sufficient for a thermodynamic specifi- range of this paper is from 0° to 200°, and from 1 to 12000 kgm. The measurements to be given here must be regarded as only the beginning of an attack on an immense field. The most important immediate task would seem to be the collection of a large mass of data, so that we may become familiar with the general types of phenomena. About the only discussion that can be attempted is a thermodynamic one, and even from this narrow point of view the measurements are not sufficient to completely determine the behavior of the several phases. We can go only a little way toward the solution of the general problem, which is to predict from the properties of any one phase when to expect new polymorphic phases, and what their properties will be. We may perhaps expect more definite results when methods such as are used here are taken in conjunction with recent X-ray methods for determining crystal structure. Thermodynamically, the transition between two solids is characterized by the same elements as determine the melting and the vaporization curves, and from this point of view the discussion of previous papers is applicable. As a matter of fact, however, the character of the solid-solid curves may vary much more widely than that of the melting or vaporization curves. For instance, all vaporization or melting curves, absolutely without exception, either rise or fall over their entire length, while solid-solid curves may have either a maximum temperature or pressure. In the previous discussion it was not necessary, therefore, to consider certain special relations between the thermodynamic constants at the maximum points, but now these relations become of importance. A discussion of these is given in the following. Another matter that needs reconsideration is that of the effect of impurities at a point of maximum pressure. Evidently the usual statement of the effect of impurity as causing a displaced temperature of equilibrium will not serve here. In the following are given the slight modifications necessary in the usual discussion to find the pressure shift of the equilibrium line. It has been possible to give a much more thorough investigation of the difference of compressibility, thermal expansion, and specific heat between the several phases than was possible in the case of melting. The reason for this is the much greater reliability of the experimental measurements of the difference of compressibility between the separate phases, because, except in those cases where the impurity forms mixed crystals, there is absolutely no rounding of the corners of the volume 1 P. W. Bridgman, Phys. Rev. 3, 126-203 (1914), and 6, 1-33 (1915). parts of the mass. It might at first thought seem that the steel shell would produce irregularities by tending to hinder the change of volume accompanying the reaction, but no irregularities of this nature were ever found. Sometimes, when the substance could be melted without danger of decomposition, it was melted, instead of hammered, into the shell. Under these conditions the shell was, of course, closed at the bottom end. If the substance were one soluble in kerosene, two slightly different methods were used. By the first method, the salt was formed into a cylinder of suitable dimensions, either by ramming into a split mound or by melting into a form, and then submerged under mercury in a steel shell, being prevented from rising by a clip at the upper end. Or it might be hammered or melted into a steel cylinder, which was then inverted below the surface of mercury as shown in Figure 1. This method is applicable if the substance A melts somewhere within the range of the experiment, and was used in all such cases. The air, of course, was exhausted in this case. The chief trouble with this method of the inverted cup is that the cup is very likely to be ruptured by the change of volume during either melting or the change from one solid to another. Substances are strikingly different in their rigidity and the readiness with which they rupture the cylinder. It is significant that the shell was never ruptured while determining transitions in the low pressure apparatus; this is another bit of evidence showing the greatly increased rigidity, or better, viscosity, produced by high pressures. The shells finally had to be made of hardened chrome nickel steel of the dimensions shown, and even then they were sometimes broken. The dimensions of the free space allowed for the mercury are of importance, since the kerosene must not come in contact with the substance in consequence of the changes of volume brought about by the transitions or the compresrange of the exper- sibility. The proportions shown sufficed for all substances investigated here up to 12000 kgm. B The FIGURE 1. modified form of container for those substances which melt within the iments. In other respects the apparatus was of the same design as that previously used, but accidents necessitated the renewal of various parts. Three lower cylinders were ruptured, two by amalgamation, and one by a violent explosion. This explosion also ruined |