Dispersion and Selective Absorption of Fuchsin. 223 



Some idea of the accuracy of the measurement can be ob- 

 tained from the following set of measurements made upon 

 one of the films, in which is included every measurement that 

 was made upon that film. 



Waye- 

 length. 



661 



661 



667 



667 



672 



672- 



, 677 



i 677 



i 



Half the rp, . , 

 measured shift, -l^l^^*^^'^^^^- 



Wave- Half the 

 length, measured shift. 



Thickness. 



36 

 46 

 57 

 60 

 70 

 67 

 92 

 89 



625 

 615 



610 

 606 



602 

 604 



585 

 588 



660 , 37 

 660 , 48 

 665 i 67 

 665 60 

 671 ! 70 

 671 72 

 676 1 89 

 676 ! 82 



623 ! 



612 



608 



605 



601 1 



604 



587 



594 



I 



This makes the probable error of a single observation 8 fi/ju 

 and of the mean 2 /x/x, which is about the order of accuracy 

 that interference methods usually give. 



There is some little uncertainty sometimes as to whether 

 the shift refers to a fraction of a band, or to one or two whole 

 bands plus a fraction of a band. In the above instance the 

 shift is about eight-tenths of a band, and the total displace- 

 ment one and eight-tenths, giving a thickness of about nine- 

 tenths. If the displacement had been assumed to be '2'8 

 bands the resulting thickness would have been 1*4 wave- 

 lengths. To decide this point definitely a number of films 

 were made of diff'erent thickness. The first of this series was 

 made from a saturated solution and the rest of the films were 

 made, each succeeding one thinner than those which pre- 

 ceded it, by continually diluting the solution from which 

 they were deposited. Starting from the thinnest (which was 

 320 yLt/A thick), the thickness of each succeeding one was 

 measured till the thickest was reached {617 fifi). The con- 

 tinuity of the series furnished data from which the thickness 

 was determined definitely, and observation upon the films 

 with a spectrophotometer confirmed the conclusions. 



The same trouble was experienced in the measurement of 

 the retardations with the interferometer, but when the thick- 

 ness was definitely known a comparison with a film sufficiently 

 thin to show bands all through the spectrum cleared up this 

 point, for with the very thin film the retardation was never 

 so much as a whole wave-length, for reasons already mentioned. 

 The matter might have been cleared up by viewing the fringes 

 of white light in the usual way and observing the shift of 

 the central fringe produced by the fuchsin, but this was found 

 impracticable on account of the selective absorption of the 



