70 Ml: K. K. SMITH ON THE CONSTRUCTION <>K 



that the tube was immersed in water at a temperature considerably lower than th.-it 

 of the room in which it was filled, and the consequent contraction of the mercury first 

 affected the menisci. 



No trouble was experienced from irregular refraction in the glass, except probably 

 in the case of the tube P. A certain part of the tube had been selected and it was 

 found impossible to get consistent values with different fillings, which, of course, 

 occupied slightly different positions in the tube. Another portion of the tube was 

 tried and no further trouble was experienced. 



Determination of W. 



The weights employed were by OEHTLING, and were of brass, platinum, and 

 aluminium. In the case of the platinum and aluminium weights the density of the 

 material was assumed, that of platinum being taken as 21 '5 and of aluminium 2'6. 

 The density of the substance of the brass weights was determined by observations on 

 the weights themselves, and a value of 8 '421 chosen, this representing the mean of 

 several very consistent determinations. 



The standard weight employed was 100 grammes of bronze blanc from Sevres. 

 The value of this was given by the Bureau Int. Poids et Mesures in July, 1 902, as 

 99 '999 143 grammes. Immediately previous to the work on mercury standards the 

 laboratory weights were compared with this. A second bronze blanc weight was at 

 the same time evaluated and despatched to the Board of Trade, where, through the 

 kindness of Mr. CHANEY, it was compared with the International prototypes there. 

 The value assigned to it by the Board of Trade was 100 '00036 grammes. The value 

 assigned to it at the Laboratory by direct comparison with the Sevres weight was 

 before despatching 100'0005 2 grammes, after receiving from the Board of Trade 

 100'0005 n grammes, a maximum difference of 0*0002 per cent. only. Soon afterwards, 

 close examination showed that the Sevres weight was stained slightly by the silk lining 

 of the box in which it was contained. Fearing that some chemical action was taking 

 place, the weight was returned to Sevres for re-polishing and re-evaluation. Its new 

 value, as given by the Bureau, was 99'998824 grammes. The weight previously 

 employed for comparisons was again evaluated, with the result that the value 

 100 '000 5 3 grammes was assigned to it. The small observed differences are clearly 

 beyond the degree of accuracy contemplated with the mercury standards. 



The Laboratory weights by OEKTLING were calibrated three times. First, as a 

 preliminary to the weighings of mercury, secondly, when the work was half 

 completed, and again at the conclusion of the observations. No differences greater 

 than the errors of observation were recorded. 



The balance was also by OERTUNG, designed to carry a maximum load of 

 200 grammes and to respond to increments of a tenth of a milligramme. The 



