210 ROYAL SOCIETY OF CANADA 



of using a wide slit is therefore evident and the purpose of the original 

 and of the present investigation is to determine liow wide the slit may 

 be made before the accidental and systematic errors become prohibitively 

 large. 



There are three ways in which an increase in slit width may lead 

 to loss of accuracy in velocity determinations : 



(a) By the loss of purity and consequent difficulty of identification 

 of lines and determination of the blended wave-lengths. 



(6) By the increased breadth and diffuseness of the spectral lines 

 and consequent probable increase of the accidental errors of measurement. 



(c) By systematic displacements of the lines as a whole^ with con- 

 sequent error in the velocity, due to asymmetric position of the nucleus 

 of the star image within the slit opening. 



The only method of determining these errors is by actual measure- 

 ment of a number of plates of the same star at a number of different slit 

 widths and by comparison of the resulting probable errors. The investi- 

 gation was limited to spectra of early type where the lines are single and 

 where loss of purity will have little effect and case (a) may be omitted 

 from consideration. The star chosen was /? Orionis which contains 

 several lines of only moderate sharpness, which is bright thus requiring 

 only short exposures, and which was, when the investigation was begun, 

 supposed to be constant in its velocity. The variability in velocity is, 

 however, over such a small range and long period that the change 

 in velocity, occurring during the hour or less required to obtain a set 

 of plates at one slit width, is inappreciable. 



In the original investigation the test was completed for three dif- 

 ferent dispersions of our spectrograph, which has a collimator of 35 mm. 

 aperture and 525 mm. focus. 



I. One prism form, Brashear "Single Material" camera objective 

 of 525 mm. focus, linear dispersion at H;- 30.1 tenth-metres per 

 millimetre. 



II. Three j^rism form, Zeiss "Chromât" camera objective of 

 272 mm. focus; hnear dispersion at H^ 10.2 tenth-metres per mili- 

 metre. 



III. Three prism form, Eoss "Homocentric" camera objective of 

 275 mm. focus; linear dispersion at H;- 18.2 tenth-metres per milli- 

 metre. 



In the present investigation two dispersions were tested, one of the 

 new single prism spectograph, and one of a new short focus camera ob- 

 jective on the three prism form, of the old spectrograph. The Eoss 

 "Homocentric" lens used previously was imperfect and not much confi- 



