210
subsequently adopted as a better means of carrying the largeballs, we have endeavoured in the following diagram,—Fig. LIX.,—to show them, first, in the positions a 2 and Bj; andsecondly to indicate that a, is moved horizontally to a,, and Bjto b 2 . The corresponding positions of the small balls are, forthe one, a x and a 2 ; for the other, b 1 and b 2 . The normal posi-tion of the rod which carries these latter, when unaffected bythe attraction of the large balls, would be half-way betweenthe directions and a. 2 bc. One-half, therefore, of the anglebetween these two directions is that through which the largeballs twist the rod from its original position when brought
Fig. LIX.—Relative positions of the large and small halls in the Cavendish experiment.
within such a distance from the small ones as is indicated inthe diagram.
By taking the mean result of a number of such experiments,in which the rod, which bears the little balls at its extremities,is first twisted in one direction and then in the opposite, it ispossible to determine very precisely how far the attraction ofthe large balls, when at a given distance from the small ones,is able to move the latter. Very accurate observations arealso made of the time occupied by the rod, which carries thesmall balls, in swinging horizontally between the extremepositions to which it moves, when it is disturbed by theattraction of the large balls upon the little ones.
We need not go into the details of the various methodswhich were actually adopted in practice to determine the time of