SOME MKRCURY STANDARDS OF RESISTANCE, ETC. 103 



standard, with these connections, was measured by the potentiometer. Its value 

 was r00146 2 ohms. 



The potential leads were now removed, and the thick copper leads used for the 

 bridge introduced. These now served as potential leads. After an interval of 

 2 hours an observation was made. The value obtained by the potentiometer was 

 l-00150 ohms. 



A quick change over to the Carey Foster bridge, and the value there was also 

 determined as I '00 150 ohms. After an interval of 12 hours the resistance was again 

 measured by the potentiometer. The value was ro0150 5 ohms. The copper leads 

 were then removed and the platinum wires re-introduced. After 2 hours a measure- 

 ment of 1'00146 5 ohms resulted. The conclusion is obvious. All bridge measurements 

 of mercury standards in which the copper leads have been thick, and, in consequence, 

 the temperature of the mercury uncertain, require correcting. The magnitude of this 

 correction depends on the particular conditions attaining during the measurements. 

 This is also indicated by Mr. GLAZEBROOK, 'Phil. Trans.,' 1888, p. 375. No great 

 importance is, in consequence, attached to the measurements made by the Carey Foster 

 bridge. It is interesting to note, however, that the conduction of heat through leads 

 is easily observable with other resistances than mercury standards. A resistance coil, 

 if cooled below the temperature of the room and then introduced to a bridge for 

 evaluation, must be observed immediately. After an interval of a few minutes a 

 different value may be obtained, owing to the conduction of heat through the leads to 

 the coil. With a difference of temperature of 8 C. and employing a coil of platinum 

 silver imbedded in paraffin wax, a difference in the values of O'Ol per cent, has often 

 been obtained, the temperature of the liquid, in which the coil is immersed, remaining 

 constant during the observations. With coils made of copper, no readable difference 

 of temperature between the coil and the external contacts of the leads is admissible. 

 In such cases, the whole must be immersed in an oil-bath. The influence of this 

 conduction effect is considerable when determinations of temperature coefficients are 

 made, and has probably led to many inaccurate determinations. 



Purification of Mercury. 



The whole of the mercury employed had been subjected to the following treatment. 

 It was taken from the iron bottle in which it was delivered, squeezed through wash- 

 leather, and passed in a finely divided condition through dilute nitric acid, and 

 afterwards through distilled water. The acid and water were contained in glass tubes 

 alxmt 6 feet long. The mercury was then distilled in vacuo, and finally re-distilled 

 in vacuo in an apparatus free from rubber in any form, so that glass alone came into 

 contact with the mercury. In some of the determinations, when the standards were 

 erected by Method II. , the mercury employed had been but once distilled. No 

 difference in the results was, however, found. 



VOL. cviv. A. P 



