91 



These weights were afterwards compared with each other with a 

 balance of extreme delicacy procured from Mr. Barrow. In its con- 

 struction it nearly resembled the balances of the late Mr. T. C. Ro- 

 binson. The beam is made sufficiently strong to carry a kilogramme 

 in each pan. Instead of having an index pointing downwards, as is 

 usual in balances of this description, a thin slip of ivory is affixed to 

 one end of the beam, a little more than half an inch long, divided 

 into spaces of about O'Ol inch each. This scale is viewed through a 

 compound microscope having a single horizontal wire in the focus of 

 the eye-piece. A screen was interposed between the observer and 

 the front of the balance-case, having a very small opening opposite 

 to the eye-piece of the microscope. , 



In making a large number of comparisons, the weights compared 

 are exposed to the risk of being injured by wear. In order to ob- 

 viate this danger, two light pans were used of very nearly equal 

 weight, each of which has a loop of wire forming an arch, the ends 

 of which are attached to the pan at opposite extremities of a diameter 

 of the pan. To the upper point of the loop of wire is affixed an iron 

 hook. Each pan is suspended by a wire of suitable length bent into 

 a hook at either end, from the ring attached to the agate plane rest- 

 ing on the knife-edge at either end of the balance. 



Calling the weights of the pans X and Y, and the weights to be 

 compared P and Q, P was placed in X and Q in Y, and P+X com- 

 pared with Q+ Y n times ; then P was placed in Y and Q in X, and 

 P-l- Y compared n times with Q + X. The weights were thus ex- 

 posed to the wear of two ordinary comparisons only in the course of 

 2 comparisons. The mean of the 2n comparisons gives the differ- 

 ence between P and Q unaffected by the very small, but unknown 

 difference between X and Y. This contrivance was found to be 

 especially useful when either of the weights to be compared consisted 

 of several parts. 



In using the method of double-weighing, the counterpoise was 

 placed in the left-hand pan of the balance, and the detached pan X 

 containing the weight P, and the detached pan Y containing the 

 weight Q, were alternately suspended from the right-hand end of the 

 beam, and the positions of equilibrium deduced in each case from the 

 extreme positions of the beam at the beginning of each of three con- 

 secutive oscillations (usually twenty times). The weights were then 



