102 REPORTS ON INVESTIGATIONS AND PROJECTS. 



(20) Preliminary report on the ternary system CaO-Al 2 03-Si02 : A contribution to the 



study of the constitution of Portland cement clinker. E. S. Shepherd and 

 G. A. Rankin. With optical study by Fred. Eugene Wright. Am. Journ. Sci. 

 (In press.) 



In this paper the general limits for the different solid phases of the ternary 

 system have been established and a few of the quintuple points located. A 

 new form of calcium orthosilicate has been discovered. From the ternary 

 diagram it follows that relatively slight changes in the initial composition of 

 the raw material will bring about relatively large changes in the composition 

 of the clinker. That is, the crystal phases present in the clinker will be very 

 different, depending on the initial concentration of the charge from which it 

 is made. It remains to be shown whether these different varieties of clinker 

 will show appreciable differences in properties when used for making 

 cements. 



(21) Some calorimetric methods. Walter P. White. Phys. Rev. 31, 545. 1910. 



The work on silicate specific heats has led to the development of new ap- 

 paratus, and of methods, some of which differ radically from those hitherto 

 generally accepted as best. A description of these new features, an exami- 

 nation of the principal errors in calorimetry, and of the accuracy attained 

 by different methods, together with a report of an experimental test confirm- 

 ing the conclusions reached, are given in this and the following three papers. 



1. The oft-noted uncertainties in the calorimetric cooling correction are 

 usually due to changes in room temperature and to evaporation. For the 

 greatest accuracy, and usually for convenience also, it is well to have : ( I ) a 

 jacket whose temperature is measured directly, completely surrounding the 

 calorimeter; (2) the calorimeter cover in contact with the water, so that all 

 bodies at uncertain temperatures are avoided; and (3) evaporation made 

 very small. The difficulties of the cooling correction then largely disappear. 



2. With this system the method of computing the cooling correction often 

 requires but one cooling period, and about one minute's arithmetical work 

 after observations are finished. The method is especially advantageous 

 where bodies are dropped into the calorimeter. 



3. Ordinarily, in accurate calorimetry, a very small temperature interval 

 is used, in order that the observer may have the advantage of assuming New- 

 ton's law of cooling in his computations. This magnifies the effect of ther- 

 mometric errors. To make a correction for the variation from Newton's law 

 increases accuracy in most cases without increasing appreciably the labor of 

 computation. The procedure is this: Since the cooling rates are not pro- 

 portional to the temperatures, the temperatures are simply corrected so as 

 to be proportional to the rates ; all the ordinary formulae can then be applied. 

 For ordinary accuracies, the correction can be disregarded up to 8°. With 

 the correction, deviations from Newton's law cause no error in 20 intervals. 



4. If a small interval is necessary, the preceding method can not be ap- 

 plied. The required absolute accuracy in temperature measurement can then 

 usually be obtained by electrical methods, but with these, as usually con- 

 ducted, the electrical quantity measured corresponds to a relatively large 

 temperature interval (20 to 300 ), so that great precision is required. This 

 difficulty is avoided if, with the thermo-element, the customary ice bath is 

 replaced by a comparison body in a water-jacket. The correction for its 

 change is only a cooling correction, and does not materially affect the total 

 error, since the regular calorimeter necessarily has a larger and more uncer- 

 tain correction of the same kind, or something equivalent. The net gain is 



