268 Prof. S. P. Thompson on the 



One-ohm coil made in Physical Laboratory, University 

 College, Bristol, by Mr. C. Swan, 



$==1-14599. 



Shunt required 7 # 85 ohms. Cut off and soldered. Final 

 testing by Foster's method, as follows : — 



U = 0-9890385 Rayleigh ohms. (This is the standard 

 "Cambridge Unit" verified at the Cavendish Labo- 

 ratory.) 



a/= 68*845 (mean of three readings). 



•£=65-280 „ „ „ 



yL6 = 0-0031 ohm per centimetre of the bridge wire. 



b=u-k^-^). 



R= 1-0000900 Rayleigh ohms (temp. 16° C). 

 [Tested again next day this same coil gave 



R = 1*0000435 Rayleigh ohms (temp. 17°*2 C.).] 



5. The amount of the final error caused by any given error 

 in the shunt is easily expressed by a formula. Let r be the error 

 of the shunt, and n the ratio of the shunt to the unshunted 

 resistance. In practice 60 > ft > 10. If the shunt S were 

 without error we should have, from the usual formula, 



$+S ; 

 or, if S were truly equal to n9?, 



ndi 



R = 



1+71 



But in reality S = n9? + r, and consequently the shunted coil 

 has for its slightly erroneous final resistance, 



w _ 9i (n3? + r) 



m+tiat+r 



nSK'+r 



1+n+ E 



Therefore we get as the final error, 

 e = R / -E=- 



(» + l) f + £(»-!) 



9? 



The second term in the denominator is always small com- 

 pared with the first term, and may be omitted, for n is never 

 less than 10, as remarked above, and r is very seldom one 



