68 REPORT—1885. 
APPENDIX. 
Suggestions for the Committee on Magnetical Reductions. 
I. By Professor Barour Srewart, F.R.S. 
1. The following suggestions are founded on the methods proposed by 
several magneticians, including Sabine, Broun, Lefroy, Capello, and Buys 
Ballot. To Senhor Capello I am especially indebted for the trouble he 
has taken in explaining his views, with which these suggestions are 
almost identical. 
2. The measurements derived from self-recording magnetographs may 
be used for two purposes, the first being to ascertain the solar diwrnal 
variation, by which name we designate that variation which is exhibited 
by comparatively undisturbed observations. The second of these pur- 
poses is to ascertain the laws which regulate disturbances. Now disturbances 
may act in two ways. Jirst, they may exhibit a diurnal variation different 
from that of the undisturbed observations, which we may call the dis- 
turbance diurnal variation; and, secondly, they may exalt or depress the 
day’s value of the particular element in question. 
As a matter of fact I believe they act in both these ways. It appears 
to me that it is of very great importance that these two effects of dis- 
turbance should be exhibited and studied together, and yet not impro- 
perly mixed up with one another. 
3. Let me explain my meaning with reference to the method of Sabine, 
which I believe to be, in many respects, an excellent one. Sabine did a 
very great deal in finding out and exhibiting the diurnal variations of the 
disturbed and undisturbed observations, but he did not greatly study, 
along with these, the effect of disturbances in altering the daily mean 
values of an element, so that it was reserved for Broun to discover that 
there were changes in the daily values of the horizontal force which were 
practically simultaneous at the various stations of the globe. Let us first 
of all consider the hourly values of declination, as this element presents — 
fewest difficulties. 
Declination, 
4, Here, I imagine, the first thing is to determine the solar diurnal 
variation, or that presented by the comparatively undisturbed observations, 
and for this purpose I fail to see a better plan than that proposed by 
Sabine. This method may be described as follows: 
5. Suppose that we have hourly observations at a station, then, first 
of all, we should arrange these into monthly groups, each hour by itself. 
We should then reject, as disturbed observations, all those which differ 
by more than a certain amount from their respective normals of the same 
month and hour, the normals being the hourly means in each month after 
the exclusion of all the disturbed observations. For the purpose of 
ascertaining the true solar diurnal variation, it seems probable that a 
considerable choice might be allowed in selecting the separating value 
implied in the above process, one value serving, for this purpose, probably 
as well as another a little above or below it. 
6. Perhaps under ordinary circumstances a value which will exclude 
as disturbed about one-twentieth of the whole body of observations will be 
found convenient. 
7. Let us now imagine that we have determined by this process 
