136 CARNEGIE INSTITUTION OF WASHINGTON. 



the quantitative system of classification, a tabulation of its divisions and 

 names, a description of the method of calculating the norm, and tables giving 

 the molecular numbers of the chemical constituents, and the percentage 

 amounts of the mineral molecules, of igneous rocks. A list of the corrigenda 

 to Professional Paper 99 is also included. 



(25) Manual of the chemical analysis of rocks. Henry S. Washington. (Third edition, 



John Wiley & Sons, New York, 1918.) 



This is a much enlarged and entirely revised edition of a book originally 

 published in 1910. Particular attention has been paid to the discussion of the 

 sources of error, both operative and methodic, and the various operations and 

 methods are described in greater detail than formerly. The revision embodies 

 the results of six years' experience gained in the Laboratory. 



(26) The calculation of the rational analysis of clays. Henry S. Washington. J. Am. 



Ceram. Soc, 1, 405-421 (1918). 



This paper discusses briefly the factors that render the so-called "rational" 

 analysis of clays uncertain, erroneous, and of little or no value for any pur- 

 pose. A method for calculating from the chemical analysis the mineral com- 

 position, generally quartz, feldspar, and kaolin, is suggested, which is an 

 application of the principles and methods of calculating the "norm" of 

 igneous rocks. In the case of clays the procedure is of great simplicity and 

 accuracy, is very expeditious after the chemical analysis has been made, 

 and jdelds results of great reliability. 



(27) The conditions of calorimetric precision. Walter P. White. J. Am. Chem. Soc, 40, 



1872-1886 (1918). 



In a calibrated calorimeter practically all the errors come in temperature 

 measurement, and the most, though often not the greatest, of these come in 

 the "cooling correction," that is, the determination of the effect of the thermal 

 leakage between calorimeter and environment. 



This leakage effect is equal to K^xT, where T is time, </)x is the thermal 

 head (difference of calorimeter and environment temperature) for the ex- 

 perimental period, and K the thermal leakiness, or leakage modulus, of the 

 calorimeter. If any of these three quantities is diminished its own errors are 

 usually little changed, but the errors of the others now have a smaller multi- 

 plier. Thus : 



1. By diminishing K (as by means of a vacuum) we diminish the effect of 

 errors in 0x, which lie (a) in getting the environing (jacket) temperature 

 sufficiently uniform, (6) in getting the calorimeter temperature uniform with- 

 out too great an error from heat of stirring. This latter difficulty grows in 

 importance as measurements become more delicate. Usually, errors in <(> 

 may easily be made negligible, and great diminution of K is then of second- 

 ary importance. 



2. A diminution in ^, as by the adiabatic method, or by methods similar 

 to that of Rumford, diminishes the effect of errors in K, such as variation 

 from Newton's Law. But owing to the peculiarities of the computation, it 

 does not much affect the main error connected with K, that in determining K 

 by means of the cooling rate, unless methods can be used which presuppose 

 an accurate knowledge of the heat of stirring. For this and other reasons, 

 the diminution of 0, though the commonest of operative devices, is in practice 

 largely illusory as far as concerns the end chiefly sought. The determination 

 of K generally results in about doubling the accidental thermometric error, 

 but this is usually the largest thermal leakage error there is in well-planned 

 work. This error can be diminished by devices some of which are adapted 

 to commercial work and which save time aswell as error. 



