ON COMPARING AND REDUCING MAGNETIC OBSERVATIONS. 87 
(1) should be calculated for each month from the separate results of all 
the years that are available, and curves be constructed to represent these 
average monthly variations according to time-scales and force-scales which 
would be marked on the curve-forms. It would be convenient that the 
curves should appear, for any one station, in a row, beginning with 
January, on a long narrow slip of thick paper, so that the sets of curves 
for any one station might be placed close under those of any other 
station for easy comparison. For preservation, the slips of paper would 
be kept in a portfolio, not bound into a book. Curves on a less elaborate 
scale, as would be suggested by the meagreness (or fulness) of the 
materials collected, might similarly be constructed on slips to represent 
the variations (2) and (3). Such series of curves, to the extent to which 
data for them would be found easily accessible, would, I imagine, con- 
stitute a conclusive answer to those who doubt the utility of extending 
investigation in the same direction; but, taking continuity of change of 
character of the variations in passing from place to place as a criterion 
of the value and importance of the results obtained, they would also 
serve the further purpose of suggesting whether and where Sabine’s 
methods are exact enough, or to what extent the application of even more 
laborious processes of reduction would be justified. These curves should 
be lithographed on thick slips of paper, and distributed amongst the 
directors of observatories and other students of terrestrial magnetism ; 
and, as little in the shape of description or comment need accompany 
them, the originals could be produced by agency of an order that should 
be readily obtainable, and that would require but little supervision, from 
some specialist member of the Committee. 
The curves might, with advantage, be accompanied by a table of the 
absolute values of the elements declination, horizontal force, and vertical 
force for each station; and also by tables of ranges of the solar-diurnal 
variations of each element on the average of each full year. 
7. It has been well established by Broun and myself that the so-called 
lunar-diurnal variation is a function both of the season of the year and 
of the age of the moon, and there is reason for believing that the bulk of 
the phenomena is really a part of the regular solar-diurnal variation, a 
part that reverses its character four times in the course of the lunation. 
Now the adoption, by Sabine’s process, of a uniform solar-diurnal 
variation for the whole of a calendar month, whilst perhaps accurate 
enough for the determination of the general character of the disturbance 
laws, leaves much to be desired when the object we are in quest of is a 
minute variation which has, in the case of the declination, a less range 
than a single minute of arc, and which is subject to variation of character 
with change of season. Here we require that a mean solar-diurnal 
variation should be calculated for each individual day, in order that the 
elimination of mean solar effect should be nearly complete; and knowing 
that either a part of the solar-diurnal variation, or the bulk of the lunar- 
diurnal variation, runs through a cycle of change in a lunation, the best 
period for which to calculate the daily means is a mean lunation, or the 
nearest odd number of mean solar days to a mean lunation—that is to say, 
twenty-nine days. The importance of this period should be kept in view 
from the first, whether or not there is any immediate purpose of investi- 
gating the lunar-diurnal variations, and my present object is not so much 
to advocate the inclusion of such investigations in the first general 
scheme of operations as to explain why the period of twenty-nine days 
