1 1 8 MR. F. E. SMITH ON THE CONSTRUCTION OF 



Throughout the whole of the observations on anyone tu)>e the mass of the mercury 

 thread employed was constant. The observations were therefore combined and a 

 sixth value of pf calculated, the formula employed l>eing 



Since the number of observations in this latter case is five times as great as those 

 of each previous calculation, the proba'ble error is less. This value of p.' has therefore 

 been used in the various calculations (see column 4, Table XVIII.). 



The calibration data for the standard W are given in full in Table XIX. Owing 

 to the large amount of space similar tables for the remaining tubes would occupy, they 

 are omitted. 



In Table XIX. the lengths X of the 5-centim. threads are recorded in the order of 

 measurement. These lengths are the sum of those measured from the bases of the 

 menisci, and the equivalent lengths of the menisci. The first column indicates the 

 mean position of the thread. While it is unnecessary to evaluate X to more than 

 three decimal places, it may not be out of place to mention that the reciprocals must 

 be taken to seven or eight places. 



In Table XVIII. the mean value of p.", the conical correction for the 5-centim. 

 lengths, is given in column 5, and finally the product pfp." = p.. Thus the complete 

 correction for conicality is obtained. 



Calibration Curves. 



In these, the extremities of the standard are indicated by a change in the thickness 

 of the curve, while the position of the etched mark is indicated by a short vertical 

 line lettered M. In each case the mean line of the standard portion of the curve is 

 coincident with the line representing unit cross-section. 



