70 REPORT—1885. 
altering the daily mean, but, on the other hand, they are less likely to 
affect the same hour during consecutive days. Were we able to obtain 
daily means of declination, unaffected by disturbance, it would be better 
to adopt this method of treatment, because it would obviate the intro- 
duction of any residual correction due to the progress of secular change 
or annual or semi-annual variation. Now in the force magnetographs 
the case is different. Here there is a certainty that some—perhaps even 
a considerable—change will be produced in the values belonging to a given 
hour in the course of a month from instrumental changes alone, so that 
treating the observations after the manner pursued with the declination 
might lead to erroneous results. 
15. On the other hand, if there were no disturbance, the difference of 
the various hourly observations of a day, from the mean of that day, 
would give us a good indication of the solar diurnal variation, provided 
the diurnal range of temperature was inconsiderable, as is generally the 
case for self-recording instruments. 
16. These remarks render it manifest that some method of obtaining 
probable values of the undisturbed daily means is, in the case of the 
force instruments, of vital importance, and Senhor Capello has adopted 
a method of this kind in his treatment of his force observations. I would 
venture to remark that the most unexceptionable basis upon which to 
determine the undisturbed daily means of horizontal and vertical force 
would seem to be given by the information already assumed to be obtained 
from the declination magnetograph for the same month. 
17. Here, as a result of the application of Sabine’s method, we have 
rejected a certain number of hourly observations as disturbed. Now let 
us reject, as a preliminary step to something more complete, precisely the 
same hourly observations of the horizontal and vertical force as being, in 
all probability, disturbed, and make use of the remainder, or of that part 
of the remainder which represents whole, or nearly whole, days free from 
disturbance, to aid us in determining, by a curve, the most probable values. 
of the undisturbed daily means. I here assume that there is no sudden 
jamp in the month’s readings from change made on the instrument or 
any other cause; if there be such, the portions before and after the jump 
will have different values, and must be treated by appropriate methods 
which need not here be discussed. Suffice it to say that, by rejecting 
frorn the month’s observations those hours which were separated as dis- 
turbed in the declination, and treating the remainder in the manner 
suggested, we obtain, aided, perhaps, by a slight equalisation, numbers 
representing very nearly the undisturbed daily values of the records 
given by the instruments. 
18. Having obtained these, our next operation is to obtain the hourly 
differences from each day’s undisturbed mean. ‘These differences, so 
obtained, we propose to treat in the same manner in which we treated 
actual declination observations. It is, therefore, to these differences that 
Sabine’s process should be applied, so that ultimately, when we have 
applied it, we shall obtain those departures of each hour from the daily 
mean which characterise undisturbed observations—in other words, we 
obtain the solar diurnal variation. 
19. Having obtained this, we have at once the means of obtaining a 
table similar, in all respects, to that which we have recommended for the 
declination. For instance, if the departure of a given hour of a given 
day from the undisturbed mean of that day were +9 whereas, according 
