

of Light at Certain Metal-Liquid Surfaces. 97 



a second lens which could be inserted behind the analyzer, and 

 an extra draw tube, the eyepiece could be focused on the com- 

 pensator fringes. Since these are not definite objects on which 

 to focus, a small circular diaphragm was inserted behind the 

 last lens of the ocular in order to avoid a small residual paral- 

 lax which was otherwise necessarily present. These modifi- 

 cations of the apparatus as originally received from the Societe" 

 Genevoise, and also the holders for the glass tubes and the 

 surface-cleaning apparatus described above, were made in the 

 instrument shop of the Sheffield Scientific School. The divided 

 circle of the spectrometer table could be read to 20", and those 

 of the nicols to 6' of arc. The source of light used was that 

 obtained from the sodium flame of a small blast lamp fed with 

 oxygen. 



Measurements. 



The preliminary adjustments of the apparatus, i. e., making 

 the axes of collimator and telescope to intersect the axis of the 

 spectrometer table orthogonally, and the adjustment of the axes 

 of the nicols to parallelism with those of the collimator and 

 telescope, were made in the usual manner. The positions of 

 the polarizer giving light polarized in the plane of incidence 

 were determined by reflection from water at the polarizing 

 angle. The water surface was cleaned in the same manner as 

 described above for the mercury. This plane was determined 

 several times in the course of the investigation and with several 

 minor modifications of method. Thus, sometimes the extinc- 

 tion position of the analyzer was determined with the reflec- 

 tion taking place at the polarizing angle, and then that of the 

 polarizer by getting its " crossed " position with reference to 

 the analyzer when the two were put in line ; or, sometimes the 

 extinction position of the polarizer for the reflection at the 

 polarizing angle was determined directly, the beam of light 

 being sent through the apparatus in the reverse direction : 

 another modification was the use of glass of known index in 

 place of water. None of these various determinations yielded 

 a position of the polarizer differing from that of any other 

 determination by an amount as large as the probable error of 

 the readings. The result of all these determinations gives a 

 value of 156'58° for the position of the polarizer. The proba- 

 ble error is less than 0-003°. 



The calibration of the compensator was made with the polari- 

 zer in all four of the positions yielding light vibrating in a 

 plane making an angle of 45° with the plane of incidence. For 

 each position the reading for the center fringe and the first 

 fringe on either side of the center was determined as the mean 

 of fifteen settings. Thus each position of the polarizer yields 



