the Viscosity of Liquids. 455 



sphere M exceeds a certain limiting value. This also is 

 obviated by determining its speed indirectly as (%v%) y , since 

 each individual portion travels so much more slowly. 



There is another reason why better results may be expected 

 by this process of subdivision. Mercury is liquid, and a 

 sphere would be deformed if it passed rapidly through the 

 viscous medium. Its speed would be diminished, and it could 

 not be dealt with directly by the fundamental equation con- 

 necting fju and V. But by subdivision the masses and speeds 

 dealt with are reduced to such an extent that we may con- 

 fidently assume the shapes to remain spherical. If the 

 deformation thus produced were appreciable, it could again 

 be manifested by the application of the formula 



V*=S(t*). 



The calculated V would be greater than the observed V, the 

 error being of the same sign as that produced by the neglect 

 of any possibly existing sliding-friction. 



The error produced by the sides of the burette, which may 

 be so near as to retard the downward motion, is also diminished 

 by the subdivision of the original mass. This also would 

 have the same sign as the other errors specified. The simplest 

 way of testing its existence is to insert into the upper portion 

 of the burette-tube a thin glass cylinder of smaller diameter, 

 and to send down a sphere along the axis of these two con- 

 centric cylinders. In the upper portion of its path the sphere 

 is at a smaller distance from the boundary of the liquid than in 

 the lower portion. If the sides actually affect the motion, 

 the speed of the sphere will suddenly change when it enters 

 into the wider portion of the liquid column. If the change 

 of speed is not appreciable, it may be assumed that the 

 diameter of the burette is sufficiently large. 



It has hitherto been assumed that we have the means of 

 keeping the temperature of the liquid constant. This is of 

 the utmost importance, inasmuch as the temperature-variation 

 of viscosity is rarely inconsiderable. The viscosity of glycerine 

 changes from 45 to 8 in the range of 16° from 4° to 20° C. 

 At a temperature of 15° C. there is a 10-per-cent. variation 

 per degree change. This implies that to get the viscosity 

 correct to four significant figures the thermometers employed 

 should read to '01° C. 



In the case of olive-oil, using Osborne Reynolds's empirical 

 formula for the viscosity at any temperature between 15° and 

 50 C, 



