604 Prof. Rankine and Mr. Smith on the Viscosity of 



the gas in the apparatus (except for the small correction for 

 slip, which can be applied independently). 



The justification for this procedure will be apparent from 

 the following set of observations with air at atmospheric 

 temperature. 



Length of whole mercury pellet about 4 cm. 



Mean temperature 16°*8 C. 



*! = 130*65 sec. ; f 2 =137'96 sec. ; £ 3 =146'00 sec. ; 



whence ^-P^ = 0'0503, and A^L =.0-0500. 



Each of the above times is the mean of six observations 

 of the time of fall of the rearmost portion of the pellet 

 between specified marks A and B (fig. 1), consistent among 

 themselves to within 0*3 per cent. It is clear that the 

 capillary effect is strictly additive, at any rate when the 

 number of segments is small and their lengths are con- 

 siderable, although Kuenen and Visser have found departures 

 from this rule with numerous segments of very small length. 

 The procedure which we have adopted, therefore, is to make 

 observations with the pellet intact, and then in two segments, 

 deducing therefrom the value of x from equation (1), and 

 determining t by equation (2). 



The purified gases were introduced in turn to the evacuated 

 viscometer, and observations of the times of fall taken at 

 atmospheric temperature and at steam temperature. In the 

 former case the temperature never varied by more than 

 o, 5 C. during any particular set of readings. Frequent 

 observations with air were also made, so that the absolute 

 viscosities of the gases could be calculated simply by 

 comparison with air as a standard. 



Strictly speaking, corrections are necessary for the expan- 

 sion of the apparatus with rise of temperature and for the 

 exposure of the small parts, C and D, projecting beyond the 

 steam-jacket. These corrections are small, and operate in 

 opposite senses; and since the temperature variation of 

 viscosity for air, as indicated simply by the ratio of the times 

 of fall without correction, has been found to be in excellent 

 agreement with the recognized value, the corrections in 

 question have been regarded as balancing within the limits 

 of accuracy of the experiments. 



The same mercury pellet was used throughout the com- 

 parisons, so that the adjusted time of fall (t) required only 



