]\gt ; Diurnal riation of Terrestrial Magnetism .
Specification of Field at Angular Radius On a New Expression for the Representation of the Components of the Variation of Terrestrial lllagnetism .
By GEORGE W. wALxBIt , A.R.C.Sc .
( Communicated by Prof. J. H. Pointing , F.R.S. Received January 8 , \mdash ; Read January 30 , 1913 .
) In any enquiry as to the cause or causes that contribute to daily or seasonal change of a periodic character in any observational quantity , the primary step is the determination of a simple and comprehensive expression for the dominant features of the phenomenon .
The periodic character of the variations of an element of terrestrial magnetic effect , such as declination , horizontal force , or the equivalent geographical components of force , is evident on almost every daily record obtained .
When the hourly values are set out and properly cleared from non-periodic chan , ( a problem of considerable subtlety ) , the historic method is to compute the Fourier harmonic components .
Another method that appears to possess great power is that so successf.ully carried out by Dr. W. N. Shaw in representing the daily and seasonal changes of meteorological elements by means of " " isopleths If a Fourier analysis reveals the existence of a limited number of dominant terms , that is so far satisfactory and provides definite material on which tloe theorist may work .
But , in the case of the terrestrial magnetic elements , there abundant evidence to show that even four Fourier terms give a very inadequate representation of the facts , and , with the exception of Dr. Schuster 's valuable memoirs ased on consideration of the first two terms in the diurnal variation , the subject is in rather a dismal stnte .
The problem has attracted my attention for several years , and the following at once occurred as going a good way towards a rough representation .
The Curve I on the diagram ( fig. is aotually We note that as is required .
To plot the corresponding curve an obvious step .
In order now to make we require to shift the origin of , and so Curve II represents FIG. l.\mdash ; Graphs of the functions .
The curve III is obtained by the simple addition of I and II .
We have only to look at the curves*to see that III represents the characteristic features of the declination or west component for either Kew , Potsdam , or Paris .
' The essential feature of a declination curve is that the morning minimum is less in magnitude than the afternoon maximum , and that the curve is flatter at the minimum than it is at the maximum .
: The turning points be shifted by changing and the relative propor$ tions of I and II .
A great variety of curves can be obtained in this way , and : I am satisfied that any component , west , north , or vertical , can be imitated in essential features .
The general form suggested is thus wherein and are at our disposal and the origin of is also arbitrary .
It seems to me rather remarkable that curves essentially unsymmetrical can be imitated in this way .
Doubtless there are many other expressions of a similar type which could be used and the above is probably only the principal part of the correct formula .
It is not at once clear how a term in the denominator can arise physically , but the following tentative suggestion is worth considering .
There are some reasons for supposing that the magnetic field of the earth is in some way connected with the intensity of solar radiation in the upper regions of the atmosphere .
Now the intensity of radiation is approximately represented by a term of the form , where is the zenibh distance of the sun , and as , where is the latitude , the sun 's declination , and the hour angle , the occurrence of the term is suggested , but , of course , careful examination of local and seasonal effect is required before it can be accepted .
The curves of diurnal variation of the magnetic elements always indicate * Arrhenius , ' Cosmical Physics , ' or Chree , ' Phil. Trans 1903 , , vol. 202 .
194 Variation of Terrestrial twice ihours , imum aained bnother tithSented thief features bpresent tusion iated mhose different constants .
We may now indicate how the constants in the formula would be chosen so as to get the closest representation of the experimental curve .
The general scale of the curve fixes the most important relation of all between the constants .
It would be most natural to take the extreme range as fairly accurately known .
Again , the times of occurrence of zeros are given by and those of the maximum and minimum by We shall not be able in general to satisfy all these relations , for when the scale is prescribed only three more conditions can be satisfied .
In any particular case we should have to decide what points are most accurately given by the curve .
For instance , with declination in these latitudes there would be little hesitation in taking the maximum and the minimum with the zero occurring about noon as much more accurately determined than the zero towards midnight , and so we should take these three to determine the constants and let the remaining zero take care of itself .
The divergence will in fact be a measure of the correctness of the representation .
I had intended to reserve these considerations for inclusion in an extended examination of magnetic diurnal variation which I have in progress ; but as this method of considering functions that recur only once in 24 hours , and are capable of representing unsymmetrical curves , is of considerable interest , and I believe novelty , this account is now .

