344 White — Specific Heats of Silicates and Platinum. 



atic error can seldom be based on one series of determinations. 

 The results for quartz and platinum are several per cent lower 

 than values given in Landolt and Bornstein. Two values for 

 platinum at 600°, given by Plato, fall within half a per cent of 

 the curve here given.* I have found no published data for 

 the other substances as high as 500°. 



Eighteen earlier determinations on wollastonite and pseudo- 

 wollastonite have been omitted as now worthless and to avoid 

 confusion. All the other observations are given, the rejected 

 ones being bracketed. 



In order to establish as certainly as possible the small dif- 

 ference between wollastonite and pseudo- wollastonite, special 

 precautions were taken against any change in conditions 

 between the two sets of determinations. The wollastonite was 

 inverted to pseudo-wollastonite directly in the calorimetric 

 crucible, so that the same lumps, in the same positions, served 

 for both. As a check on possible systematic errors the pseudo- 

 wollastonite determinations were repeated with half charges, 

 showing very fair agreement. 



The two values for soft glass at 500° first obtained seemed 

 suspiciously low, as they cause a distinct reversal of the curva- 

 ture. Two further determinations were therefore made in a 

 new furnace, where the crucible was surrounded above and at 

 the sides by a stirred bath of lead, and below \>y an electrically 

 heated plate, whose temperature was observed and regulated. 

 The results agreed to 0*1 per cent, and were only about 0*5 per 

 cent above the old ones. This indicates that the peculiarity in 

 the curve of this glass is really due to the substance itself, and 

 also tends to confirm the estimate above (p. 335) as to the 

 systematic error of the older observations at this temperature. 



The accuracy of the true specific heats is less than that of 

 the mean heats, from which they are derived. This is because 

 the true heats are dependent on the differences of the others. 

 The inversion in quartz at 575° renders more numerous data 

 desirable before computing the true heat for it. 



The mean specific heats of platinum and of diopside are 

 expressed by the following formulas : 



Pt -03198 + 3-4 XlO- 6 



CaMgSi 2 6 ._ -1779 + 1-516X 10-^-1-047 X lO- 7 2 + 2'81 X l(r n 3 

 The agreement with the simple curve of platinum is as good 

 as the original data, but the formula is not recommended for 

 extrapolation. With diopside it amounts to only half a per 

 cent, and a curve with one less coefficient is nearly as good. 



For the results at 100° the heating was done in a steam 

 bath. 



*W. Plato, Erstarrungserscheinungen an anorganischen Salzen und 

 Salzgemischen. I, Zeitsehr, phys. Chem., lv, 736, 1906. 



1 



