GEOPHYSICAI, LABORATORY. I05 



andalusite and cyanite pass slowlv into sillimanite on being heated above 

 1300° C. 



(2) There are four definite compounds of lime with alumina, namely, 

 3CaO.Al203; 5Ca0.3ALOo. melting-point 1387° C. ; CaO.AUOg, melting- 

 point 1587° C; 3Ca0.5Alob3. 



(3) 3CaO.Al203 and 3Ca0.5Al203 have no true melting-point, but the 

 former will be completely melted at about 1550° C. and the latter at about 

 1725° C. 



(4) Two of these compounds, 5Ca0.3Al203 and 3Ca0.5Al203, have an 

 unstable form each, while 3CaO.Al203, and probably 3Ca0.5Al203, are 

 unstable at the melting-point, i. e., do not produce a maximum on the liquidus. 



(5) There is one compound, MgCALOg, between magnesia and alumina. 



(6) There is reason to believe that the system MgO.CaO is a eutectic 

 series with no compound, and little, if any, solid solution. The temperature 

 range is too high for satisfactory investigation. 



Having established the nature of the binary systems, experimental study 

 of the ternary system CaO-Al203-Si02 is now under way. 



Complete optical data are given for all the preparations studied, and in 

 conclusion a brief discussion of the geologic significance of the various 

 kinds of measurements, chemical, thermal, and optical, which can be compe- 

 tently undertaken in the laboratory. 



(10) Specific heats of silicates and platinum. Walter P. White. Am. Journ. Sci. (4), 



vol. 28, 334. 1909. 



For determining the specific heats of silicates up to 1500° C, the method 

 of mixtures in which the heated substance is dropped from a furnace into a 

 calorimeter at room temperature was selected as the most accurate. The 

 chief source of error is in the lack of uniformity of furnace temperature. 

 This has been diminished in some cases by special forms of furnace. The 

 error introduced in the process of transferring from the furnace to the calo- 

 rimeter is negligible. An electrical method of dropping greatly reduces the 

 time required, and determinations tnade with the containing crucible alone 

 eliminate what heat loss there is. 



All temperatures, including that of the calorimeter, were read by thermo- 

 elements. By this means rapidity and simplicity of manipulation were 

 secured. 



The calorimeter was completely inclosed by its water-jacket. An unusual 

 temperature rise (sometimes 23°) was successfully employed to increase 

 accuracy. Some simple ways of treating specific heats mathematically are 

 given. 



A consideration of the various sources of error indicates for the mean spe- 

 cific heats a final accuracy of better than 0.5 per cent at most temperatures. 

 This conclusion has been strengthened by the results obtained with a special 

 furnace. The true specific heats, derived from these, are less accurate. Sili- 

 cate specific heats show, after a considerable increase with temperature up to 

 700°, a tendency to diminish at higher temperatures. 



(11) On the dependence of valence upon volume in certain trivalent elements. A. Lud- 



wig. Journ. Am. Chem. Soc, 31, 1 130. 1909. 



A study of the effect of pressures from 6,000 to 17,000 atmospheres upon 

 bismuth, antimony, and aluminium in the presence of water vapor near their 



