﻿Measurements of Pendent Drops. 423 



Suppose, e. p., that b and y are measured to one per cent., 



so that 

 we have 



Bz _ -0% -12fty 



and, since, in general, % is small compared with unity, the 

 percentage error in the computation of ( -d~ + tT~ ) i s practi- 

 cally equal to the percentage error in the estimation of b. 



The quantities actually measured are, of course, x and y ; 

 and since 



we have 



b = 



Sb By 8% 



showing that if x and y are subject to the same percentage 

 error, the error in b has numerically the same value as the 

 error in measurement of either x or y. 



Finally, in the equation lor the computation of T, viz. : — 



T (i + iH A 



none of the above quantities involved are raised to a power 

 which unduly multiplies the errors of observation. 



The accompanying figure (fig. 3) shows on a reduced 



Fisr. 3. 



/<] 



B 



scale the outlines of the two photographs of a drop a 

 measured ; the figures relating thereto are given 

 Figures A and B on PL VIII. (tig. 6) "are r 

 reproductions of the photographs, figures C and U 



dually 

 below. 

 educed 

 on the 



