40 Rubens and Trowbridge — Dispersion and Absorption 



tangent to the curve a through r ; then the abscissa of the 

 point of contact o of this tangent gives the corrected position 

 of the maximum. 



The construction given is only admissible if the curve a in 

 its middle portion represents a comparatively clean spectrum ; 

 that is, if the width of the central image is considerably less 

 than that of the diffraction image. It is apparent, on consider- 

 ing this point, that this construction may be applied in all cases 

 in which the size of the correction does not equal half the dif- 

 ference between the width of the central image and that of 

 the diffraction image. In making the measurements given 

 below, this limit was never exceeded. The size of the correc- 

 tion differed for each image," corresponding to the different 

 width of each. On the right image it was only a few seconds ; 

 on the left in some cases more than a minute. If the slit s x 

 were infinitely narrow, s and s 2 having a finite width, both 

 refraction images would appear of equal width, and the correc- 

 tion would be in both cases the same ; that is to say, it would 

 be, approximately, the arithmetical mean of the corrections 

 applied by us on the right and left image. We have convinced 

 ourselves of the correctness of this explanation by observations 

 in which, under otherwise similar conditions, the width of the 

 bilateral slit s x was varied between the limits 0'2 mm and 0*5 mm . 

 Here no systematic difference in the results was noticeable. 



At the beginning of each series of observations the slit s 

 was illuminated with sodium light and the mirror e Q was 

 adjusted so that the sodium line fell exactly in the middle of 

 the slit s x . As the accuracy of the objective adjustment was 

 found to be insufficient, we substituted an ocular of high 

 power for the slit s 2 , and by this subjective method, were able 

 to adjust the sodium line on the slit s 1 with an accuracy of two 

 seconds of arc. After this adjustment, the mirror e was 

 turned by means of the slow motion screw, through an angle 

 ^~i(7D~~7^) which corresponded to a change in the mini- 

 mum deviation of a^—a^. The wave length was then meas- 

 ured in the manner given above. 



The following two tables give the results obtained with a 

 rock salt prism (angle 6 — 10° 53' 0") and a sylvine prism 

 (<f) — 12° 39' 10") together with several indices of refraction 

 and wave lengths taken from the two communications already 

 cited.* The figures of the first column represent the differ- 

 ences between the minimum deviation a, for the wave length 



in question X, and the minimum deviation of the sodium line 



* The series marked with ooe asterisk * are taken from the work of Rubens; 

 those marked with two asterisks from that of Rubens and Nichols. 



