[PLASKETT-DELURY] THE SOLAR ROTATION 43 
km. on the average. It is an open question, as these measures were 
made at different epochs, whether this difference is to be ascribed to 
the use of the mask or to a change in the habit of measurement. The 
measures were made with great care by both observers and in precisely 
the same way:—four settings on the line in the centre strip and two on 
each of the outside strips with the screw moving alternately forward 
and back, and, after all the lines were measured, repeated with the 
plate reversed on the carriage. Moreover, as the measures are purely 
differential, the displacement of one absorption line with respect to 
another precisely similar absorption line, the presence of this compara- 
tively large systematic difference between the two observers is not 
readily explainable. Different methods of measurement and various 
comparisons were made in an attempt to explain or overcome the 
difficulty, but the difference still persisted practically unchanged in 
magnitude and sign throughout. It is proposed * by De Lury in order 
to obtain absolute values of the displacement, which are uncertain 
under present circumstances, to impress upon the spectra, in addition 
to the rotation displacement, an arbitrary displacement of say the 
order of a millimetre in magnitude. This would be effected by using 
a double or broken slit, the central section (of the width of one of the 
spectral strips) being displaced laterally any desired distance with 
respect to the body of the slit. If a rotation spectrum be made through 
such a slit the displacement will be s + r, where r is the rotation 
displacement and s the displacement due to the slit. If a spectrum 
of the limb at the pole where there is no rotational displacement be 
made through this slit the displacement will be s. The measured value 
of these displacements will be s + r + e and s + e where e is the error 
of measurement, varying with different observers, yet which should 
(for each observer) have the same value in the mean of a large series 
of measures, as the two displacements are relatively of nearly the same 
magnitude. The true value of the rotational displacement will then 
be 
shor +e (Serre 
and in this result personal habits of measurements should be eliminated. 
Besides the mechanical and observational difficulties in the way of this 
proposal, however, there is the further one that the accidental error 
of measurement would be increased and the amount of measuring 
required doubled. Furthermore, as these spectra could not be taken 
under identical conditions, the possibility of instrumental errors affect- 
ing the results is rather a serious one. Even with rotation spectra 
—— nS — nn 
* Jour. Roy. Astron. Soc, Can, 5, 495. 
