1889.] 



On the Cavendish Experiment. 



267 



fibre is loaded very heavily. At the time tliat tlie experiment was 

 made the instrument had only been set up a few hours. 



The mean decrement of three positive sets was 0*8011, and of three 

 negative sets, 0*8035. The observed mean period of three positive 

 sets was 79*98, and of three negative sets, 80*03 seconds, from both 

 of which 0*20 must be deducted as the time correction for damping. 



The deflections obtained from the six sets of observations taken in 

 groups of three, so as to take into account the effect of the slow 

 change of the position of rest, were as follows : — 



From sets 1, 2, and 3 17*66 + 0*01 



2, 3, and 4 17*65 + 0*02 



„ 3, 4, and 5 17*65 + 0*02 



„ 4, 5, and 6 17*65 + 0*02 



An examination of these figures shows that the deflection is known 

 with an accuracy of about one part in two thousand, while the period 

 is known to the four thousandth part of the whole. As a matter of 

 fact the discrepancies are not more than may be due to an uncertainty 

 in some of the observations of half a millimetre or less, a quantity 

 which, under the circumstances, is hardly to be avoided. 



The result of these experiments is complete and satisfactory. As a 

 lecture experiment the attraction between small masses can be easily 

 and certainly shown, even though the resolved force causing motion 

 is, as in the present instance, no more than the one two hundred- 

 thousandth of a dyne (less than one ten- millionth of the weight of a 

 grain), and this is possible with the comparatively short half period 

 of eighty seconds. Had it been necessary to make use of such half 

 periods as three to fifteen minutes which have been employed hitherto, 

 then, even though a considerable deflection were produced, this could 

 hardly be considered a lecture experiment. 



The very remarkable agreement between successive deflections and 

 periods shows that an absolute measure made with apparatus designed 

 for the purpose, but on the lines laid down above, is likely to lead to 

 results of far greater accuracy than any that have been obtained. 

 For instance, in the original experiment of Cavendish there seems to 

 have been an irregularity in the position of rest of one-tenth of the 

 deflection obtained, while the period showed discrepancies of five to 

 fifteen seconds in seven minutes. The experiments of Baily made in 

 the most elaborate manner were more consistent, but Cornu was the 

 first to obtain from the Cavendish apparatus results having a precision 

 in any way comparable to that of other physical measurements. The 

 three papers, published by him in the ' Comptes Rendus,' of 1878, re- 

 ferred to above, contain a very complete solution of some of the problems 

 to which the investigation has given rise. The agreement between the 

 successive values, decrement, and period is much the same that I have 



