_ ON COMPARING AND REDUCING MAGNETIC OBSERVATIONS. 75 
first place, they are to discuss the best methods of reducing magnetic ob- 
servations ; but, before these methods can be put into execution, we must 
secure that the observations taken at different places are sufficiently 
homogeneous to admit of a common treatment. As we have to deal not 
with the individual observations, but with numbers which have already 
been reduced at the different observatories, it is clearly of importance that 
these preliminary reductions should be done everywhere in the same 
manner. Professor Stewart’s suggestions refer exclusively to this point, 
while Sir Henry Lefroy rather discusses the question as to how the measure- 
ments already in existence can be made to yield information of physical 
value, and as they are treating of different matters, there does not seem 
to me to be necessarily any real difference of opinion between them. 
While agreeing entirely with a great many of the remarks made by Sir 
Henry Lefroy, I believe that some common method of reduction like that 
proposed by Professor Stewart is necessary before we can gain any know- 
ledge of magnetical disturbances. With regard to the proposals them- 
selves, the principal question will always be, whether the different heads 
of observatories can be made to agree on a uniform plan. The exact 
nature of the method of reduction is a matter which has to be settled 
chiefly by those who have practical experience in magnetic observatories. 
The method of rejecting disturbed observations, commented upon by 
Sir Henry Lefroy, is, no doubt, open to objection. If it was simply our 
object to gain information on the mean value of magnetic elements, no 
observation however much disturbed ought to be rejected ; but as soon as 
we suspect that the mean value is not the normal value—that is to say, 
that disturbances act more frequently in one direction than in another— 
we are necessarily driven to adopt some method of rejecting disturbed 
observations. The objections raised by Sir Henry Lefroy against the par- 
ticular method employed by Sabine seem to me to be, however, very 
serious, but I can see no difficulty in amending that method so as to 
render it free of the difficulty. 
IV. Letter from Professor G. H. Darwin, F.R.S. 
Cambridge : 
June 10, 1885. 
A pricri I should not have thought of distinguishing between mean 
and normal values, but I suppose that it is desirable to do so. It is 
obvious that if all the observations for a month are analysed, we get the 
mean harmonic constituents. Then if we recompute the values with 
these constituents (which may be done with a tide predicter), and sub- 
tract the hourly values from the observations originally analysed, we get 
a series of residuals. Supposing from those residuals we arbitrarily cut 
out a certain number which are above some arbitrarily chosen magnitude, 
and submit the rest to harmonic analyses, and supposing these present 
us with a new series of constituents with pretty constant phases and 
amplitudes, then it would seem to me that we should be justified in the 
hypothesis that normal and mean are not the same thing. I must suppose 
that some process more or less equivalent to this has been carried out. 
I do not observe that any proposal is made to submit the monthly 
constants derived from harmonic analysis to a further analysis, and thus 
to derive the annual, semiannual, and terannual inequalities of the con- 
