REVERSED AND NON-REVERSED SPECTRA. 



99 



Apart from measurement of the thickness e, therefore, the method is guaranteed 

 for fourth-place work. 



52. Observations. At short distances the ellipses are always rounder and 

 present in all focal planes of the telescope. Doublet lens compensators of 

 relatively large focal power are available. Using the CS 2 -sweet oil solution, 

 the submergence of convex and concave lenses in it made but very little dif- 

 ference in the position and definition of the ellipses. There must, therefore, 

 have been approximate equality of the conditions of refraction. With glass 

 in water this was naturally not the case. 



A number of experiments were made with the well-known mercury-potassic 

 iodide solution as obtained from Eimer & Amend. The index of refraction 

 in the concentrated state (floating glass) exceeds 1.7. The slight straw- 

 color is no disadvantage. 



A plane-parallel trough of ordinary plate-glass, 0.293 cm. internal width, 

 was now constructed, only just large enough to receive a glass plate or a con- 

 vex lens. With the plate or lens submerged centrally, therefore, the excess 

 of path-difference over the empty trough would be nearly constant. The 

 liquid was then gradually diluted until the excess of path-difference of the 

 liquid over the glass content passed through zero into a deficiency. Con- 

 siderable stirring was needed to insure homogeneity on each dilution. The 

 results are given in table 23 for the plate and in table 24 for the lens. So 

 long as the diluted solution was effectively more refracting than the glass, 

 the ellipses were nearly circular and very clear both for the submerged plate 

 and lens, as well as for the liquid, but when the solution began to effectively 

 refract less than the glass, the ellipses were washed and could not be obtained 

 strongly. On submerging the lens in these cases it was necessary to so adjust 

 its position horizontally and vertically that the white images (obviously in 

 different focal planes) coincide as nearly as possible. This insures a sym- 

 metric position for the lens, after which the spectrum is to be examined and 

 the ellipses placed by moving the micrometer. Unfortunately, as neither 

 the trough nor the plate was optically plane parallel, some readjustment 

 for this was necessary at each observation, an operation which introduces 

 the error in the results shown in tables 23 and 24, so far as absolute values 

 are concerned, which has been alluded to above. 



TABLE 23. Submergence of plate-glass (thickness 0.284 

 cm.) in mercury-potassic iodide solution. Thickness inside 

 of trough, 0.293 cm. Sodium line. 2B/\ Z = K. Glass, 

 5=4.5Xio~ u ; 2S/X 2 = 0.0262. 



