102 THE INTERFEROMETRY OF 



It thus appears that in case of the solution the equation n=A-\-B/\z is 

 far from sufficient up to the Fraunhofer F line. Nevertheless even here the 

 mean of the first two results may be regarded as holding in the region of the 

 D line. The large value of B as compared with the glasses is particularly 

 noteworthy, and from this a reason for the poor ellipses obtained with dilute 

 solutions is suggested. If for two media the AN/e are identical, as at a and 

 b in figure 67, this implies merely that 



Hence, when the coefficients B differ as largely as is the case for the solution 

 and the glass, the indices ju are far from equal. In the above case, if fj.fjf = 

 0.071, the solution should show no displacement at the D line when the glass 

 is submerged. But this difference in the indices of refraction of the glass and 

 the solution is enormous, and if the submerged body is a lens, the correspond- 

 ing images in the telescope will be thrown quite out of focus. Thus the el- 

 lipses are necessarily washed. Conversely, when the indices of lens and solution 

 are nearly equal at the sodium line, the displacement is 2(eBeB')/\ z , and 

 therefore considerable (0.035 cm. above, in the example taken) ; but the ellipses 

 are now sharp and strong. Unfortunately, therefore, displacement is no 

 criterion for equality of n, and mere dependence on the sharpness of fringes 

 is insufficiently accurate. The only resource left is to compute B and B' for 

 two spectrum lines and adjust the solution for this displacement. Again, 

 this is not convenient, particularly when but a single spectrum line is at hand. 



54o Further observations. A somewhat wider trough was now constructed 

 of the same plate-glass as above. The following dimensions were found by 

 calipering: Glass wall plates, top 0.290 cm., bottom 0.283 cm.; internal 

 thickness (liquid plate), top 0.564 cm., bottom 0.563 cm. As the light passed 

 through near the bottom of the trough, the second data are to be taken in 

 each case. The observations were made with sunlight and at each of the 

 Fraunhofer lines B, C, D, E, b, F. In case of a hazy sun the lines B and F 

 were often less strong than desirable ; but in other respects the work through- 

 out progressed smoothly, showing magnificent ellipses beginning at the B 

 line with the horizontal axis longer and ending at the F line with the vertical 

 axis longer. Circles occur earlier as the refraction is greater. NN a in 

 table 2 7 is the coordinate referring to the difference of micrometer reading for 

 the presence and absence of the plate under observation. If the glass walls 

 of the trough alone are to be taken, the N a refers to the half-silver plate alone. 

 If the solution is in question, A 7 ",, refers to the half-silver and the trough walls 

 taken conjointly and in place. The trough is, of course, not to be moved; 

 but some adjustment is needed when the liquid is introduced, which mars 

 the absolute result. In series 5 a glass plate = 0.293 cm - thick was sub- 

 merged in the solution without readjustment. Hence of the resulting re- 

 fraction 293/563 = 0.5024 belongs to the glass and 270/563=0.4796 to the 

 solution surrounding it. Table 16 contains ten series of results with the 



