ON THI-: SI'KCIFIC RKSisi'AXCE OF MERCURY 359 



Hence, 



Volume of thread at = W//j. 



Volume of thread at = VV(1 " < " y<) 



P 



W (I 4- yf ) 



Mean section of tube at t = - 



pi (1 4- W) 



\\- /i . *\ 

 Mean section of tube at = 



Length of tube at = ^ '* 



(1 ~r 39 1 ') 



Hence, the value of W/plL corrected for temperature is 



_W (1 + yt) (1 + Iff?) . 



P lL(l 4-W 



and, if W be the weight in vacua, a- the density of air, and p of the brass weights 

 used, 



1:1 ' W-W ,- 



. 



= W (1 -000062], 



taking dry air at 10, the mean temperature of the weighings, and putting p' = S'l. 

 Hence, finally we get, introducing all the corrections, 



{ 1 - "' + f - -000002 - ^ ( ' - l) 

 [ L t L \/w, / 



' ' H-(r-r/-/>)<-(6-i^K; .... (s) 



r - 



and, if we put in numerical values for the one unit tubes, we have 



1M j t _ .,, . + . 42 - .000002 - '-if-' (1 - l) 



p/i /L 1 L n / L V! / 



r = 



! 

 + -000149 1 -000009*'! ' ( y ) 



while, for the other tubes, the third term is '46a/?. We may write this 



r = 1 -^(l+A), ........ (10) 



where A is a small fraction, being the sum of all the correcting terms with their 

 proper signs. 



The methods used in finding L and I differed but little from those employed by 

 Lord RAYLKIQH. The tubes were supplied by Messrs. POWELL and Sox, and a 

 number were roughly calibrated. The best of these were selected and were cut so as 



