[PLASKETT-DELURY] THE SOLAR ROTATION 65 
The trend and magnitude of these mean residuals in Plaskett’s 
measures for the different latitudes and the ratio of the mean algebraic 
to the mean numerical residual, which is except in one case less than 
one-third, do not indicate any systematic differences for the different 
lines. If any lines or elements gave a different velocity to the mean 
reversing layer, then the mean residuals for the different latitudes 
should be of the same sign, should diminish as the latitude increased 
and should vanish at the pole; but we find, on the contrary, that none 
of the lines fulfils this condition, but that the residuals bear the appear- 
ance of being quite accidental in character. Even take the case of 
the Na line 5682.869 which gives a strong negative residual, we find 
no decrease with higher latitudes and the mean ‘residual for the pole 
is much higher than the average, showing that the difference is probably 
due to something in the line. Again, if this sodium line did give a 
lower value of the velocity, the other sodium line, the last on the list, 
should also give a negative residual, whereas we see its residuals are 
entirely accidental. The same condition of affairs is shown by the 
tabulated residuals from De Lury’s measures of Series II in which the 
mean algebraic is always less than one-fourth the mean numerical 
residual, although the numbers are higher owing to his higher probable 
error of measurement. 
These considerations form sufficient grounds for the statement that 
in the region around À 5600 none of these lines or elements shows any 
differences of velocity from that of the reversing layer other than can 
readily be accounted for by accidental errors of measurement. 
31. The same thing appears to be the case in the A 4250 region. 
The following table contains the residuals in metres per second from 
the 15 lines measured on 24 plates at the equator. 
