333 



passing a series of mere equilihria ; that Is : the method required to 

 be made practically a static one, the tinal niaxinmm-pressure beinjï 

 independent of the special way, in which the pressure in the growing 

 gas-bubble is gradually augmented. 



Thus a verii slow rise of pressure in the growing gas-bubble is 

 necessary, and only in this way does it appear possible to eliminate 

 the small differences of pressure in the long connecting tubes of the 

 apparatus. For it is well known, that the adjustment of such small 

 pressure-differences takes a considerable time, if the connecting tubes 

 are relatively long. 



If the radius of the capillai-y tube is r (in cm.), the specific 

 gravity of the liquid d, and the observed maximum-pressure (in 

 Dynes) is H, then the surface-energy x (in Ei'g. pro cm\) is calculated 

 from Cantor's expression (loco cit.) : 



_ r// 1 ^ 1 (Pr"^ 



The last two terms of the second member of this equation are 

 iisiiall}^ so small, that they can be neglected in comparison with the 

 experimental errors, as being corrections of the secondary order. 



Nevertheless it has become clear, that a special correction requires 

 to be applied to the values calculated in this manner, because of 

 the fact, that in the theoretical deduction of this relation, a simpli- 

 fication is used, which cannot be considered quite legitimate. We will 

 advert to this correction lateron. (Vid. : VI ; under general remarks). 



With this limitation extended experience in the use of the method 

 leads to the conviction, that in the form it is used here, one can 

 obtain reliable and, within narrow limits, reproducible results. It 

 has the advantage, that the surface-layer of the liquid is continually 

 renewed, thus the often-observed and troublesome phenomenon of 

 the alteration of this layer, need not be feared. Furthermore one 

 can vary the flowing gas at will with the different liquids, to prevent 

 eventual oxidations or reductions ^). With these precautions the 

 results can be considered as accurate at 1650° C. as at ordinary 

 temperatures, if only no abnormally high viscosity iB encountered 

 in the liquids ; for this will destroy to some extent the reliability of 

 the measurements. The influence of the viscosity will be discussed 

 lateron in more detail. 



Of all sources of error to be considered : inaccuracy in the 



1) As long as the gas is indifferent, i.e. as long as it does not react with the 

 liquid, tlie results will be quite comparable, because experience teaches, that the 

 differences in the values of x> measured with different gases, are vanishingly 

 small in comparison with the experimental errors. 



