line on one side, both microscopes were voail and the telescope was moved to the 

 other side and set upon the same line and the reading taken again. The 

 temperature of the grating was noted after each setting. This was always 

 repeated several times and then the circle was undamped and turned through an 

 angle of from 20" to 60°, the grating was again made normal to the collimator 

 and the angular measurements resumed. Tlie case with which the grating could 

 be adjusted by means of the collimatiug eye-piece reduced the labor of working 

 over difl'erent portions of the circle very greatly and, besides, it was easy at any 

 time to examine this adjustment so as to be sure that the grating had not been 

 disturbed from its position. At frequent intervals during the measureuunits the 

 instrument was readjusted througliout, in order that no constant errors due to 

 false adjustments might influence the results. 



THE GRATING. 



The University possesses four ditYraction gratings ruled by Jlr. Cliapnian. 

 Three of tlicse are upon metal and one is on glass. I am iidbrmed by Mr. 

 Chapmaii tliat the cmnposition of the metal is of copper and tin in the ratio of 

 17 parts of the former to 8 of the latter. 



The metal gratings are .«qnare, eaeli side being IS) cm. in lengili. The 

 tliicknoss is 7.7 mm. The ruled surfaee covers about 20 sipiare centimeters. 

 Tlie finest of llie three consists of about :5l)000 lines, nominally ruled at tlie rate 

 of 1729G to one English im-h. This grating has great dis|)ersi\(' [lower, therefore, 

 separating tlie D lines in the s|)ectrum of tlie second order a triHe less than 5', 

 ami in addition to tiiis the definition is most excellent. In conseijiicnce of these 

 facts tliis grating was selected as bein<i the most desirable for the work of precise 

 measurement i)f wave-lengths and, with the single exception noted in the results 

 given, tlie measurements were made njioii the s[)ectra of the sec()nd order, 'i'lieie 

 are one or two |M.'ciiliaritit's about tlie spectra jiroduced by this grating, found 

 also to exist in a less degree with the others, wliich are worthy of a moment's 

 notice. Tlie most notable of these is the iueiinality in briL;htiiess (jf spectra of 

 the Nirne order on opposite sides ot tlie normally iiieideut rays. This is notice- 

 able in Bpectru of both the first and second orders but is much more marked in 

 the latter. 



Tlie intensity of illumination at any point in a plane at riu'ht angles lo the 

 lines of a dilTraction grating is gcner.dly expresseil in the shape ot' a function 

 ciinsisting of the jiroduct ol three terms. When a certain relation exists between 

 the second and third tenns, spcclm of a certain order will disappear entirely anil 

 when this relation exists approximately, tliese spectra will Ix' correspondingly 

 faint. 



The onler of the spectra tlnis alVected is det4,'rniined by the relation lietween 

 the wliole grating »piue and the reflecting surface wliich remains at Icr the groove 

 is cut out, or Ijetween tlio opaque and transparent spaces in a ruling upon glass. 



