426 Prof. D. B. Brace on a New Spectrophotometer 



the different sectors may cover the slit in these successive 

 positions, will vary the intensity of light at the ocular slit in 

 ratios depending on the angle of these sectors. Suppose, for 

 example, the disk is placed before the slit of T, fig. 3 — which 

 may be unilateral — a match having first been obtained and 

 the bilateral slit of T. r read. When it is rotated a new adjust- 

 ment of the slit of T' is made until a match is again obtained. 

 This gives a position representing seven-eighths the full 

 optical value of the slit. This may then be repeated for the 

 other sectors of different angles, obtaining readiugs for six- 

 eighths, five-eighths, down to one-eighth the optical value of 

 the full slit. We have, then, a series of optical values of the 

 slit in terms of the first or of any other, and found by knowing 

 the ratio of the angles of the corresponding sectors. With 

 these data we may interpolate to find the optical values of the 

 slit widths for any other measurement. By shifting the 

 telescope R we obtain calibrations for the specific colours 

 whose comparison we wish to make, these being free from the 

 numerous errors entering into the general optical system. 

 We may also obtain intermediate values by varying the slit 

 of T and repeating the process. We can then remove the 

 disk and place our source to be compared before the same 

 slit, and make all further measurements in terms of our cali- 

 brated bilateral slit. The width of the slit of T being once 

 set should not be disturbed during one series of measure- 

 ments, after which it may be varied and a new series made. 

 After the calibration the bilateral slit should in no case be 

 closed, and in making a setting the screw should always be 

 turned in the same, direction. This calibration assumes a 

 constancy in the radiant before T', but the error arising from 

 the luminosity curve is small in most cases, and as further 

 the form of this curve does not change materially over con- 

 siderable variations in the intensity of the source, such, for 

 example, as filaments or pencils brought to incandescence by 

 the electric current, no sensible error enters into the calibra- 

 tion from this cause. 



When the inclination of the curve of luminosity does not 

 alter rapidly, as in the case of filaments, corrections for 

 colours not calibrated may be interpolated from the calibra- 

 tion curves of adjacent colours. However, with the collimator 

 slit of "25 millim. width or less, and a dispersive power equal 

 to that of the flint-glass above mentioned, the corrections due 

 to the form of the luminosity curve are less than the error of 

 observation when the standardized slit is bilateral as described 

 above. In this case we only need to determine the optical 

 values for two readings of the slit corresponding, say, to the 



