334 



measurements of r, of dt, of the pressure H, of the reduction-factor 

 of the observed pressure on the manometer to mercury-pressure, of 

 the measurement of temperature, of the depth i, etc., — the last 

 mentioned appeared to be the most significant. If all these errors are 

 assumed to be cumulative, the total effect upon the reproducibility 

 of the results, even at 1650° C. is still within about 1 '/, of the 

 true value of */, and at lower temperatures about 0.6 7o of ^hat value. 

 With many molten salts, where H is very great and the viscosity 

 very small, the percentage error appeared to be even less than this, 

 not exceeding 0,47„. For our purpose this degree of accuracy may be 

 considered a very satisfactory one considering the enormous difficulties 

 of measurements at those extreme temperatures. It is also question- 

 able, whether it will be possible to exceed this accuracy at such 

 high temperatures in the near future. And if this could be 

 done, it is very problematical whether much would be gained 

 for the purpose proposed. For experience teaches us, that at those 

 extreme temperatures all compounds are in a state of more or less 

 advanced dissociation, and it can hardly be of any significance to 

 express the surface-energy / of such compounds in tenths of Ergs, 

 when the uncertainty in the values of /, caused by the inevitable 

 admixture of the dissociation-products, will surely be larger than 

 the correction-factors following from this increase in the accuracy 

 of the measurements. 



§ 4. In this and the following papers we will successively give 

 an account: (I). Of the experimental arrangements and the manner 

 of procedure, including some instances, illustrating the general adapta- 

 bility of the method employed in different cases. (2), The results, obtained 

 between -80° and + 270° C. in the study of a great number of 

 carboncompounds, in connection with their atomic constitution and the 

 validity of Eötvös' theoretical views, (o). The experiments made to 

 determine the free surface-energy of molten salts, by means of the 

 method here developed. In this connection we will also discuss 

 more in detail the earlier attempts to solve the problem by the 

 method of capillary ascension in glass-tubes. (4). Finally a discussion 

 of the results obtained and a number of considerations of a more 

 general kind will be given, which are suggested by the study and 

 comparison of the data now available. 



§ 5. Apparatus and Experhnental Equipment, 

 a. In all the measurements pure, dry nitrogen, free from oxygen, 

 was used, be(;ause even at the highest temperatures this gas appeared 



