118 THE ROYAL SOCIETY OF CANADA 
In discussing these measures and probable errors it will be noted 
that the probable error of one observer is considerably the greater, 
this being probably due to less experience in spectrum measurements. 
Considerable more weight should be given to the results with the lower 
probable error. 
In the first place, the mean displacement given by the measures 
of all lines on all plates is practically the same for the two observers, 
the difference being too small to be systematic. In this connection 
it may be worth noting that measures of lines 4, 5, 6, 7, 8, probably the 
five best lines in the fifteen, on one of the plates by 6 observers showed 
considerable differences in the mean displacements, there being a dif- 
ference of about -003 mm. between the highest -0707 mm. and lowest 
-0674 mm. The trend of the measures seemed to indicate that the 
greater the experience of the measurer the higher the measured dis- 
placement. It is difficult to assign any reason for such apparently 
systematic differences, and although not sufficient data are at present 
available, it is evident that some attention must be bestowed upon this 
phase of the question. Such a difference in the measures of a rotation 
plate would correspond to a difference of about 3% in the value of the 
rotation at the equator. , 
In the second place, considering the differences in displacements 
given by different lines, they appear to be on the whole accidental in 
character, though, in the case of Plaskett’s measures, there are at least 
two lines, C 4197-257 and Se. 4246-996, one asymmetrical and the 
other very broad, showing systematic deviations from the mean greater 
than attributable to accidental error. There are even greater differ- 
ences in De Lury’s values, but owing to his larger probable error they 
can not probably be definitely assumed to be systematic in character. 
There is thus seen to be a possibility that the character of the lines 
themselves may lead to differences in the measured displacement, of 
about the same amount as those found by Adams in the rotation values. 
It is questionable whether the differences here found are sufficient to 
account for all the discrepancies found by Adams, as they do not apply 
to all the lines, and the question can only be definitely settled by the 
employment of much more material, such as that to be secured by the 
committee, than is at present available. 
The probable errors of measurement of an average line by Plaskett 
is + -0006 mm., equivalent with the dispersion used to + +013 km. in the 
rotation value, which is of about the same order as those obtained by 
Adams. 
It is only necessary to say a few words in this paper about the 
procedure followed in making the rotation plates, as that will be dis- 
cussed at greater length in a future paper when definite results will be 
