Measurements of Precision in Platinum thermometry '. 569 



be about 250 ohms. If the resistance of each lead of the 

 platinum thermometer is of the order of O'lo ohm the value 

 at 0° C. of /3L 3 /(a + /3 + L3 J r L 4 ) is about 0*15, and the value 

 of Z/Lo/fa + ^ + Li + Ls) is about 0*10. With increasing- 

 temperature the value of the first ratio will slightly increase 

 and that of the second ratio will diminish. As R/S = 0*01 



very nearly it follows that if ~ is equal to (a + L 4 )//3 within 



1 per cent., the value of the second term on the right hand 

 side of equation (24) cannot exceed 0*00001 5 ohm, which 

 corresponds to o, 001 5 C. Hence, if the resistance-coils have 

 the values already stated, the lead L 4 may vary in resistance 

 from 0-09 to 021 ohm before a correction equal to o, 001 (J. 

 has to be applied. 



At 0°C. the ratio Q/S is about 05, and it is necessary for 

 Q/S to be equal to ajb within 1 part in 5000 in order that 



the correction-; --. — ^ — |-^_ -J may not be greater 



(a + fc + Lj + LOVS b) \ 6 



than o, 001 C. As Q, S, a, and b are resistance-coils and 



include no leads, such equality in the ratios may readily be 



obtained. 



AVith regard to the last term of equation (24), as the 

 ratio bh 2 l [a + b + L x 4- L 2 ) is about 0'10 and the ratios 

 /3L 3 /S(a-f # + L 3 + L 4 ) an( l L,/& are each about 0'0003, it is 

 necessary for L x to differ from L 3 by 30 per cent, before an 

 error so great as o, 001 C. could be introduced because of 

 the inequality. In practice we may safely assume that the 

 equality will be within less than 1 per cent, and so may 

 dismiss the correction from consideration. 



As a convenient summary it may be stated : — If the 

 resistance of each of the leads of the platinum thermometer 

 be not greater than 0*15 ohm, and if the ratio R/S is equal to 

 (a + L 4 )//3 withm 4 parts in 1000 and the ratio Q/S is equal 

 to ajb within 1 part in 10,000, no greater total error than 

 that corresponding to 0°'001 C. will result by neglecting all 

 the terms after the first on the right side ot equation (16j. 

 The bridge is then used as if it were a simple Wheatstone 

 bridge having the arms P, Q, R, and S. No preliminary 

 setting need be made. The resistances of the leads of the 

 platinum thermometer should be approximately equal, but a 

 dillerence of 30 per cent, produces an error of o, 001 0. 

 only. If the preliminary balancing of the bridge is made 

 there is no error due to any inequalities of leads or ratios. 



The construction of the bridge should be much on the- 

 same lines as that described for the Third Method, there 

 being six dials for the Q and a coils. 



