[PLASKETT-DELURY] THE SOLAR ROTATION 69 
Again, it will be noticed that in the final values no mean algebraic 
is one-third as large as the mean numerical residual, even though in 
three cases the algebraic mean for one of the latitudes is nearly as large 
as the corresponding numerical mean. At the foot of the table are given 
the mean residuals obtained from Adams’ 1908 values and indicated 
directly above by the letters L and H those lines which Adams claimed 
gave lower or higher values than the general reversing layer. It will 
at once be seen that the results obtained from the 72 plates measured 
by Plaskett do not agree with those of Adams, but are generally of the 
opposite sign. It seems to us therefore that the only safe conclusion 
to be drawn from the evidence at hand is that any differences found 
in both Adams’ and Plaskett’s values are not real differences of velocity 
but are, if not wholly accidental, rather some personal effect in the 
measurement due possibly to the character of the line. It is unfortu- 
nate that no plates containing He and Ca 4227 were obtained here 
in order to compare the rotational values obtained from these lines 
with the general reversing layer, as was done by Adams; but it seems 
likely that personal differences, at least as high as those occurring in the 
general reversing layer, would be present in the measures of these 
broad and difficult lines. 
SUMMARY. 
32. The principal conclusions reached from this investigation 
may be briefly summarized as follow:— 
(a) The Ottawa values of the solar rotation may be represented by 
the formule 
V = (1.483 + .532 cos? @) cos g dE 
£ — 10°.32 + 4°.05 cos? g Plaskett 
V= (1.448 + .532 cos? @) cos g bk 
£ — 10°.04 + 4°.00 cos @ De Lury 
which are in remarklaby good agreement with Dunér’s and Adams’ 
1908 results except for small and nearly constant differences, and 
which probably represent very closely the law of variation with latitude. 
(b) The absolute velocity of the solar rotation seems to be un- 
certain by the small differences above referred to, of the order of two 
or three per cent. which is apparently due to personal differences in the 
habit of measurement of the rotational displacements on the plates. 
(c) The tabulation and discussion of about 3,000 residuals from 
different lines and elements in the regions measured, show that no 
