190 



brass scale beneath, and on the sleeve was a fiducial line, in the form of a 

 flue black wire. Three indeiDendent settings were made on each nodal point, 

 the mean being taken as the position of the node. Since the figures were 

 formed at temperatures different from those at which they were measured, 

 the corresponding corrections for the expansion and contraction of the 

 glass tubes were applied. 



Fig._3. 



From such a series of measured inter-nodal distances the most proba- 

 ble value of the wave length was calculated from the formula, 



(n — 1) (a. — Oi) + {n — 3) (aj-i — a_,) • 



n{n^- — l) 

 1. J. 3 

 where // is the number of settings, and (ir-(u-(i~, ai'e the respective settings. 

 The writer is indebted to ^Ir. P. P. Koch for a complete calibration of 

 the brass scale used, in terms of the standard meter bar oelonging to the 

 Institute. Corresponding corrections have been computed and applied to 

 all the measurements of both series. 



