ON PBACTICAL STANDARDS FOR ELECTRICAL MEASUREMENTS, 115 



constructed. The 10-ohm build-up is here described. In this the coils 

 are of nominal value, 1, 1, 2, 2, 5 ohms, and may be described as 

 la, 1/3, 2f(, 2/3, and 5. Each coil is of manganin, is immersed in oil, and 

 connected by two copper posts to massive copper blocks, the blocks being 

 provided with side terminals and mercury contacts. The coils la and 1/3 

 are evaluated by the Kelvin double bridge as described for standard unit 

 coils. The leads to the bridge are from the mercury contacts, and the 

 connections with the shunt coils a and /3 are from the side terminals. 

 The resistance thus measured is that between two points lying centrally 

 under the mercury contacts in the copper blocks. The value of the 

 5, 2/3, 2a, and 1/3 in series will, therefore, be exactly equal to the sum of 

 their individual values. The coils la and 1/3 ; la, 1/3, and 2a ; 2a and 

 2/3 ; and 1^8, 2«, 2/3, and 5 are compared by forming a simple bridge, the 

 coils in the other arms being of 10 ohms resistance. A reversal in 

 position of the two coils enables the difference to be accurately found. 

 Finally the 5, 2/3, 2a, and 1/3 are employed to evaluate a 10-ohm coil. 

 100 and 1000 ohms are built up in a similar manner. 



Let the constructional errors of the 10, 100, and 1000 ohms build- 

 up boxes be a, b, and c respectively. Then, if we neglect the errors of 

 observation, which ai'e small, the error of a 10-ohm is a, of a 100-ohm 

 (a + b), and of a 1000-ohm (a-^b + c). If the 100 and 1000 ohm coils 

 are evaluated by a Wheatstone bridge using the 10 to 1 ratio, then the 

 error of the 100-ohm is '2a, and of the 1000-ohm 3a. Hence, if in 

 practice a =6, and 2a=6-f-c, the probable error of the built-up values 

 must be veiy small. Observations show that the differences 2a — {b + c), 

 3« — 36, ifec, are not measurable with certainty, for not only are the 

 observed differences very small, but often the sign changes. The dif- 

 ferences resulting in one set of observations is given in the following 

 table : — 



Jtine 11, 1906. Observed Values in Int. Ohms at 17° C. 



The values given on the first and fourth lines are from the three 

 build-up boxes. The second values of 2450 and 2449 are obtained by the 

 bridge, using the 1 to 10 ratio from the first build-up. The third value 

 of 2449 results from the 1 to 10 ratio from the second build-up, and the 

 probable error is therefore 36. 



Low Resistance Standards. — A large number of methods have been 

 suggested for the measurement of small resistances, and as many of these 

 are known to be in use, it may be of service to point out the advantages 

 and disadvantages of each. 



(a) Matlhiessen and Hockin's Method (fig. 7). — By adjusting the 

 resistances R and S, a balance is obtained with the galvanometer arm 

 connecting R-!S with each of the potential points of P and Q in succession. 

 The value of R-f S is kept constant. The ratio of R to S is necessarily 



i2 



