332 



Mr. J. W. Gififord. 



[May 7, 



higher temperatures, recourse was had to the table of temperature 

 refraction coefficients. These coefficients are the mean of a separate 

 series of measurements carried out at temperatures at least 10° C. 

 apart. 



4. Material. — The material of which the prisms were made is quite 

 normal. The fluorite is from Germany, the quartz from Brazil, the 

 calcite from Iceland. 



5. Standard Wave-lengths. — Rowland's wave-lengths have been 

 adopted wherever available. As a specimen of the work in fullest 

 detail the determination of the index of line C for fluorite is given. 

 (See Appendix I.) 



6. Curves. — A very severe method of testing refractive indices, when 

 they are fairly close together, is to calculate the form of a thin lens 

 from two of the substances, so as to be achromatic for two definite 

 wave-lengths, and then focal lengths calculated by the formula 



<>->HB 



for the different wave-lengths should form the ordinates of a smooth 

 curve. The focal length 6985 mm. was taken as being ten times that 

 required for the lenses of a large spectrometer now just completed. 

 The curves shown (p. 333) are for (A) fluorite and quartz, (B) fluorite 

 and calcite, (C) quartz and calcite. It appears that the focal length 

 of the fluorite and quartz combination is very nearly independent 

 of the wave-length. A list of such focal lengths is given in 

 Appendix II. 



7/ Measure of Error. — An approximate estimate of the error in the 

 deviations may be made as follows : — -In reducing a set of observations 

 three means, called " group deviations " (see Appendix I), are taken. 

 Let x, y be values of the two of these which show the greatest differ- 

 ence, then the quantity a = may be taken as a measure of the 



o 



error. There are 118 measurements in the table for 



33 of which a = less than § " ; corresponding variation of index = '0000023 

 89 „ = „ li"; „ „ . „ = 0-0000034 



31 „ - „ 2£" ; „ „ „ = 0-0000084 



15 „ = more than 2§" ; „ = 



1 only where a = as much as 5£"; „ „ = 0-0000150 



Optical Correctness of Prisms. — In order to test whether the faces of the 

 quartz and calcite prisms were parallel to the optic axis, each of the 

 angles of the prisms was measured, and from these measurements and 

 the deviations proper to them separate indices were calculated for each 

 of the three angles (a) for the ordinary ray, (b) for the extraordinary 

 ray. Owing for the most part to errors inseparable from measurements 



