Gaseous Ammonia, Pliosphine, and Arsine. G03 



phenomena of the mercury pellet. Quite noticeable changes 

 in the curvatures of its ends were observed, not only when 

 the gas was changed, but also when the temperature of the 

 same gas was altered. This was confirmed by discordances 

 in the results. In earlier work with this apparatus it has 

 been assumed that the effect of capillarity was constant, and, 

 in a pellet about 4 cm. long, equivalent to a reduction of 

 driving pressure of approximately 4 per cent. Our present 

 observations show that this is not quite true, and that 

 variations of the driving pressure of as much as 2 per cent., 

 or in extreme cases even more, may arise from changes of 

 curvature of the ends of the pellet. It appears probable that 

 the effect is due to change of the angle of contact of the 

 mercury with glass, rather than to an actual variation of 

 the surface tension. The effect has been eliminated by a 

 method already suggested by Knenen and Visser *, which 

 consists of taking observations in every case both with the 

 pellet intact and broken into segments. By this means the 

 consistency of the results is much improved and the reliability 

 of the method definitely increased. 



If we assume that the capillary effect is doubled in a 

 pellet broken into two segments, and trebled when the 

 segments are three in number (as indeed proves to be the 

 case experimentally), we may denote it as a certain fraction 

 x of the full hydrostatic pressure difference p due to the 

 weight of the mercury. If £ l5 £ 2 , and £ 3 are the observed 

 times of descent of the pellet when it is in one, two, and 

 three parts, respectively, in driving equal volumes of gas 

 through the capillary tube, we have 



(i-i)* l =(l-2*)« 8 =(i-3a?>3, . . . (1) 



from which x can be derived. 



Equality of the two values of x thus obtainable, viz., 



^2 — 1~\ h t\ 



will indicate that the capillary effect, is additive; and if t 

 denote the time of transpiration we might expect in the 

 absence of surface tension, i. e. when the full hydrostatic 

 pressure of the mercury would be operative, then 



t=*i(i-*) (2) 



This time t is then strictly proportional to the viscosity o£ 



* Knenen and Visser, Communications 1'livs. Lab. Leiden, No. 13S. 

 p. 3 (1913). 



2 S 2 



