Section III., 1910. (Aa TRANS. R.S. C: 
XI.—Probable Errors of Radial Velocity Determination. 
By J. S. PLASKETT, B.A. 
(Read 28th September, 1910.) 
The magnitude of the probable errors attending the spectrographic 
determination of stellar radial velocities has always been with me a 
question of much interest and considerable work! along the line of the 
dependence of probable error upon the width of the slit employed has 
already been accomplished. It is proposed in this paper to give a gene- 
ral discussion of the probable errors of radial velocities as affected by 
changes in the dispersion of the instrument, and in the type of stellar 
spectrum observed. 
It may not be amiss to point out that in measuring stellar spectra, 
as in practically all scientific measurements, we have two classes of 
errors to deal with or guard against: first the accidental errors of setting 
upon the lines of the spectra due partly to imperfect definition, and 
partly to the unavoidable differences in successive settings which are 
always present even with the most careful observers; second the syste- 
matic errors, due in this case generally to instrumental conditions, which 
give rise to spurious relative displacements of star and comparison lines. 
Among such conditions may be cited flexure or change of temperature 
of the spectrograph, non-uniform illumination of the collimator, prisms, 
and camera by the star or spark light, faulty focal adjustments of 
collimator or camera objectives, and so on. The former can be readily 
evaluated from the measures of the plates themselves, but no evidence 
of the latter appears in such measurements, and its magnitude can only 
be determined from the comparison of a number of plates of the same 
star. In the latter case, however, the errors so obtained will not be 
entirely systematic but will be affected by the accidental errors present 
in the measured velocities. In the discussion to follow, relative mea- 
sures of the accidental errors are given by the probable errors of single 
lines or regions on a plate, while for the systematic effect the probable 
error of a single plate, obtained by the discussion of several plates of the 
same star, is probably the best that can be done although, as stated 
above, such result has also included in it the effect of the accidental 
errors. 
In considering the effect of change of dispersion, one would naturally 
expect to find the actual linear errors of the same magnitude for all dis- 
persions and, as errors are always expressed in velocity values, the 


? Astrophysical Journal XX VIII, 259 and Trans. Roy. Soc. Can. 1909, p. 209. 
