Wave-length of Light. 



355 



Date. 



1887. Nov. 2 . . 



. 36 







27'19 



3 . . 



. 36 







25-87 



4 . . 



. 36 







24-40 



5 . . 



. 36 







24-95 



5 . . 



. 36 







26-83 



9 . . 



. 36 







26-14 



16 . . 



. 36 







27-40 



16 . . 



36 







27-37 



17 . . 



36 







27-57 



22 . . 



36 







25-16 



29 . . 



36 







25-69 



29 . . 



36 







25-99 



29 . . 



36 







25-91 



30 . . 



36 







26-10 



30 . . 



36 







25-86 



30 . . 



36 







25-81 



Dec. 1 . . 



36 







25-68 



1 . . 



36 







25-80 



The last decimal place is retained simply for convenience 

 in averaging. The mean value of </> is 36° 0' 26 //, 07, which, 

 reduced for the error of thermometer at 20°, gives finally 



</> = 36° 0' 25 / '-17. 



The probable error of this value is //, 14. The effect of a 

 small error in (f> on the resulting wave-length is given at 

 once by 



B\= cos <f)8(j). 



In this case the error introduced by an error of 1" in (f> is 

 a little less than 1 part in 250000. 



The mean value of 6 during these measurements was 



6>=6° 59' 58 v -56. 



In case of grating IY. the line selected for observation was 

 one of Rowland's standards at w.l. 591 4*319 of his prelimi- 

 nary list. It is a very close double, the components being- 

 distant from each other something like 7-5^00 of their wave- 

 length. The double deflection was as before 72°, but in the 

 fifth order. As with grating III. eighteen series of obser- 



vations were 

 of <£:— 



obtained, with the following resulting values 



