88 REPORT—1885. 
enters into modifications that I would suggest of the procedure proposed 
by Dr. Balfour Stewart in dealing with the horizontal force tabulations, 
but which modified process should, I think, be applied also to the de- 
clination tabulations. 
Tt is not a general rule that the hours at which the bulk of disturb- 
ance occurs are the same for both the elements declination and hori- 
zontal force; and hence—though it is highly probable that disturbance 
of some degree in one element occurs on the same day as disturbance of 
another degree in the other—we cannot with safety allot the disturbances 
to identical hours. 
8. First, I would substitute for Sabine’s classification of disturbances 
as ‘larger’ and ‘smaller,’ a division into those that are without the 
limits set by the normal + the separating value, and those that are 
within those limits; and instead of rejecting disturbed observations I 
would, at such step of Sabine’s process for separating the larger dis- 
turbances, replace each disturbed entry by the same number minus the 
disturbance without the limits—as apparent at that stage. The dis- 
turbances without the limits would be separated and the laws of their 
variations determined by the methods that Sabine applied to his larger 
disturbances, but the disturbances within the limits would remain in- 
volved with the regular variations until a late stage of the investigations. 
9. Secondly, as regards progressive change in the readings, both of 
the declination and horizontal force instruments, it would, I think, gene- 
rally suffice to treat that change as uniform during the course of a month. 
Having entered the hourly tabulations for a given month on a table 
(A call it) having the hours marked at the top of the columns and the 
days of the month in the first or left-hand column, and having taken 
daily means, I would take the mean of the first fifteen of those daily 
means and the last fifteen of the preceding month’s table A as the mean 
number for the beginning of the month ; and similarly the mean number 
for the end of the month would be the mean of the last fifteen daily means 
of that month and the first fifteen of the next following month. Change 
at the uniform rate indicated by the mean numbers! for the beginning 
and end of the month I would eliminate from the original hourly tabula- 
tions of table A, and enter the new number on a new table (B), to which 
I would proceed to apply Sabine’s (modified) process. This would lead 
to a general knowledge of the regular solar-diurnal variations for each 
month, and of the laws of the disturbance variations ; and here a rest- 
ing-place might be found if it were desired to compare results from 
different stations before proceeding with more elaborate reductions. 
10. To proceed, however, I would next, having obtained the amounts 
of disturbance without the limits, eliminate these amounts from the re- 
spective disturbed observations of table A, calling the table thus altered 
(A’), and this table should form the basis of discussion in respect of the 
regular solar-diurnal variations for each day, the lunar-diurnal variations, 
and the laws of variation of disturbances within the limits. 
From table (A’), and the corresponding table of the preceding and 
following months, I would construct another similar table (C), each entry 
in which would be the 29-day mean of the numbers for the same hour 
1 The effects of disturbances without the limits on the daily means I would take 
to be sufficiently indicated by the departures of those means from corresponding 
daily means, as calculated from the mean numbers for the beginning and end of the 
month, with a uniform rate of change from one to the other. 
‘ Ma 
