]\gt ; The Diurnal Variation of lIagnetism .
By GEORGE W. WALKER , , F.B.S. , formerly Fellow of Trinity ( Received October 15 , \mdash ; Read November 13 , 1913 .
) The diurnal variation of terrestrial gnetism has been the subject experimental study for many years , and at a considelable numbel of observatories scattered all over the earth .
But the co-ordination of the results , and the theoretical ation of the se or causeh of the phenomena , have not made progress seems to bear a ) proportion to the vast amount of observational that has muLtted .
As far as I know , Dr. Arthur Schuster 's memoirs*titute the most systematic attempt to reduce this matter to scientific law order .
Although I have found it convenient to depart from the pursued by Schuster , thus of ) ocednre in no suggested by various remarks in his llemoirs .
The comparative lack of interest in the sub , ject arises , I believe , in measure from the difficulty ( commented on ) Schuster ) of the experimental data in a fornl conyenient for rational arison .
tories still continue to reduce their observations to publish .
results in a variety of ways , and , unless one ) undertake a amount of tedious ithmetical conn ) nffitiun at the very outset , it tically impossible to obtain a comprehensive of the The Advisory Committee of Eskdalennnir recommended the instruments should be arranged so to record directly the , components of netic force , the resnlts for began to take definite shape , it was , pel.haps , natural I ) old ] ) interested in comparing the results with those the unly other which at the time also recorded the components , Potsdam ( Seddin ) .
The I made in the light of conclusions , brought out points of such interest that I induced ttenlpt to collect data from other observatories ) the values the eographical components .
Following Schuster , the lata I desired to were the Fourier coefficients iu the components .
I myself to the 24-hour and the 12-hour , because I share the entertained ) many of those realise the err that arise on the ' Phil. Trana , ' 1899 , vol. 180 , and 1908 , vol. 208 .
Mr. G. W. Walker .
experimental and side , that the shorter period terms are of very donbtful accuracy .
further confined my attention to the yearly value , not because seasonal variation is } ) ortant , but because I felt that there was a .
danger of coming to rief in a multitude of facts .
The data I have been able to obtain are collected in tabular form .
If they appear somewhat meagre , 1 .
would point out that I had perforce to confine attention to atories whose results could , without very much arithmetical labour , be put in the form desired .
I hope that the information disclosed by the table will encourage others to make a contribution to it , and thus to the progress of knowledge of the phenomena .
In the table the quantities are those in the equivalent formulae and , where is the local mean time at the station and the unit of nitude is or auss .
From information very kindly supplied by the directors of the various observatories , the statement may be made : 1 .
Practically all days are used , not selected quiet days .
2 .
The curves are not smoothed , except in the case of the results for Pola .
7 ) .
The original curves gave records of and except at Eskdalemuir and Seddin , where and were directly recorded .
4 .
The hourly values are those at the exact hour , except at Seddin , where the hourly value is the estimated mean for an hour centering at the exact hour .
1 need perhaps hardly say that one could wish for data obtained in recisely the same way , and that before entering on a minute arithmetical computation , all data should refer to the same year .
It was soon obvious from my inquiries that one would have to wait several years before such data could be obtained .
, I think certain broad inferences may be drawn from the data collected in the table , and these seem to me of vital importance .
It will be convenient to review briefly the main points in Schuster 's memoirs .
If we grant the main proposition , that it is correct to represent the variations by a potential metion , the problem may be divided into two parts : The empirical determination of the potential function which represents the observations .
The Diurnal oflIacynetism .
Mr. G. W. Walker .
( 2 ) The theoretical ation of the physical causes that give the potential function so determined .
These two divisions are substantially represented in Schuster 's first and second memoirs .
But the two divisions are not entirely independent , ( 1 ) must borrow from any hint gested in ( 2 ) , and ( 2 ) must be kept within the indicated by ( 1 ) .
In the first memoir , Schuster dealt with observations from the four Pavlovsk , Greenwich , Lisbon and Bombay .
Observing that in the west component the Fourier terms depended substantially on the local time ( a feature which is in the main confirmed by our table ) , he showed how the potential could be calculated .
Since the west component is the form is the co-latitude and the longitude , it is only necessary to the proper expression for in Tesseral and then the simple integration with respect to gives , when multiplied by , the function His conclusion reached was that principal parts of the mean diurnal variation for the year could be expressed by a potential metion of the form where is the local time or the equivalent Greenwich time plus the longitude , say .
In the second memoir Schuster proposes to explain these terms as arising from electrical currents in the atmosphere , set up by the joint action of the ermanent part of the earth 's magnetic field and the mechanical oscillation of the atmosphere .
Thus the barometric variation is associated with the llagnetic diurnal variation .
Certain i.ormidable difficulties occur and are carefully noted by Schuster , but I need not comment on them .
In attempting to deal with the seasonal variation Schuster supposes that the conductivity of the air depends on the sun 's zenith distance , and , assuming a simple and tentative expression for the conductivity , investiates o potential function that arises .
The analysis is necessarily exceedingly complicated .
Let us now return to the empirical expression obtained Schuster .
A potential of the form 1nust give components to north and to wesb ; and again the form Th of zetism .
gives components , sin2 .
The following Table II facilitates the parison of the formulae just obtained with the data collected in Table Table II .
If ow we examine the data and consider only , then , allowance for the that the are not all for the same year , we have , I thinl 's expression .
Batavia is , however , bnormal .
, taking the values for only we have strong support for a formula of the type iven by Schuster .
But clearly for any one station values of and should the same constants in the formula .
is not the case .
The phase are not in agreement , and what is even more serious is fhat the amplitudes of the nolth component and all too boreat as compal.ed with the values computed from the potential function which ) resents t ) west componenlt .
The matter looks still more difficult if we compare the of with the and of W. I confess that the result was ) one could see no flaw in method , and the west values did , on the whole , depend ' on local time .
There , of course , the qibility of a higher zonal harmonic , but this did not promise much help , and I was certainly to enter on much arithmetical computation .
There remained the possibility of obtaining function which would contribute to without to .
This practically meant a term on time from some fixed meridian , as Schuster indicated .
After carefully studying Schustel.'S theory of the cause , with its complicated analysis , it to that one : process , that is to say , one may in perfectly elleral 1luine 384 Mr. G. W. Walker .
potential functions which are differentially related to opposite sides of the earth .
The process may stop as soon as we get the terms which represent the data , and then we may seek for the physical meaning of the terms .
In mathematical terms we start with the primary potential l. Any space derivative of this is also a potential .
Thus a slJace derivative fixed with to the earth gives a contribution to the fixed part of the earth 's field , while a space derivative fixed with to the sun gives a contribution to diurnal change .
In this simple way expressions were formed and it was found that the terms sought for could arise .
I think will be convenient to postpone consideration of the manner in which they were obtained and to give first the results .
For the 24-hour term the form of potential function at which I arrived may be written A , wherein is the local mean time and are the co-latitude and longitude of the station , while may .
be regarded as the lopgitude of some arbitrary meridian .
Since the data do not all refer to the same year a minute numerical analysis would be out of place .
Thus only round numbers were selected for the constants .
By trial the following form was finally selected:\mdash ; .
This gives in units of the following values for the north and west components : numerical values computed are shown in Table III .
Although the observed values are given in units of , it must be remembered that the accuracy in most cases does not exceed 1 I think it will be admitted that we have , on the whole , got a substantial representation of the data .
The better agreement that might have been obtained for the European stations had , of course , to be sacrificed somewhat to get the cnrious features exhibited by the Bombay and Batavia data .
The gnetism .
Table sion .
Helwan Tha potential function obtained must also account for values of We are at liberty to suppose that any term in arises from an external or interior source , and further that these contributions may differ in phase .
We may , therefore , introduce a division of the constants that will the correct value of at the surface and at the same time account for the observed values of .
In fact , if we could depend on the values of we have the important means of determining the external and internal proportions .
But unfortunately the obseryations of are poor and everyone who has really faced the experimental difficulties admits that the results are most unsatisfactory .
In particular I know that the Eskdalemuir results for are quite unreliable , and , from the other data , I think they must be regarded with some suspicion .
The Potsdam and De Bilt results do , however , agree very well , and so I take them as a sort of standard .
But clearly the position is a weak one , and therefore I do not feel justified in giving more than a general indication of what our formula would do .
I therefore assume that the outside and inside contl.ibutions agree in phase , and as regards the second order harmonics I adopt SchusCel.'S result that the internal contribution is one quarter of the external .
In the first order harmonic I assume that the contribution from source is nil .
We thus get the formula for The computed values , marked provisional , are entered in Table III .
The results are , on the whole , in the right direction , and might be brought closer by taking a smaller proportion from an internal source , but I do not think it worth while to force the matter until better data are available .
VOL. LXXXIX.\mdash ; A. 2 Mr. G. W. Walker .
We now consider the 12-hour terms .
The form which I found it convenient to try was The first four terms involve harmonics of order 3 but the last is of order 1 .
I finally dropped the term in , made and C. It also seemed an advantage to make Thus the empirical form adopted was so that The components to north and west thus become cos2 The computed values shown in the followin Table Table W. V. ( Provisional .
) Again , I think these numbers give substantial agreement with the data .
Turning to the data for , the numbers look more hopeful than in the case of the 24-hour terms , and the experimental errors that affect the 24-hour term are perhaps not so serious in the 12-hour term .
Thus I was tempted to push the agreement somewhat further .
The Diurnal Variation of Terrestrial qnetism .
387 The data gested that the main part for the northern stations arose from the term in , but was chiefly iu the cosine term and practically nil for the sine term .
This can be met a } ) has difference between the outside and inside contril ) utions .
Thus assumed .
Hence we have , Asin To get the results for Potsdam , we have the additional equations Hence , , which Tees with Schuster 's result thab is about 1/ 4 of A. Next the term in .
the main of the terms at Batavia , and assuming for the 1noment that the harnlonic was entirely of order 3 , I took the form Hence The values at Batavia gest taking , Hence This makes about one-fourth of A. But the harmonic was really made up of a main part of or and a minor part of order 1 , and the correction to is thus found to be sec45o where we have assumed that the first order term was entirely external .
The net result is The computed values in Table to very satisfactory agreement with the data .
I do not , of course , that Chis empirical representation of the data in Table I is unique or final , but I do think it is clear that the Schuster terms alone will not co-ordinate the data , and the form we have obtained does go a very conside1able way vards meeting the difficulties .
388 .
G. W. Walker .
We may now consider how the terms were arrived at , in such a way as to 6 provide a clue to the physical meaning without binding oneself to any special theory .
In the figure let , be rectangular axes through the earth 's centre in the directions , earth 's way , radius to the sun , and perpendicular to the plane of the ecliptic .
Let OX , , be fixed rectular axes in the , but not rotating with it , OZ the axis of rotation , and OX in the plane and OZ .
is thus the obliquity of the ecliptic and an angle expressing the time of the year .
The direction cosines of OX , OY , , referred to , are OX ; ; OZ Let be the polar co-ordinates of a point fixed on the earth , and let refer to any other point on the earth .
Suppose start with a potential and differentiate along the direction .
We get a potential function Since remains constant while the earth rotates , we simply have a contribution to permanent field of the earth , representing a doublet with its axis in the direction , ] us differentiate along .
We get where is the local time at the point of observation reckoned from midnight .
The term in contributes nothing to diurnal variation , but contributes a seasonal term to the general field .
It has a maximum in summer , minimum in winter , and vanishes at the equinoxes .
Variation of Terrestrial Jfcxgnetism .
389 The term in contributes to the variation .
The amplitude has a semi-annual seasonal variation , to its maximum value 1 at the equinoxes , and falling to its minimum value at the solstices .
Ayeraged for the year it contributes a term to the ) diurnal variation .
Such a term is indicated by the data .
We may , if we like , picture it as a uniform field in the direction of the , within ich the earth rotates .
In a similar way we may differentiate , and so a term Let us now carry the process of forming functions further by differentiating the direction .
We may write the result in the form where is Greenwich mean time , the co-ordinates of a point fixed with respect to Greenwich , and the co-ordinates of the station referred to Greenwich .
The first two terms contribute nothing to diurnal variation , but only annual seasonal terms to the , eneral field .
The remaining terms contribute to diurnal variation , the amplitude a semi-annnal seasonal variation fluctuating between 1 and .
Averaged for the year we get terms in the mean diurnal variation .
If is we have a term , where is the local time .
Similarly , a term arises by differentiating .
We thus get the Schuster terms in the 24-hour term of diurnal variation .
If , we have the term ( A ) Similarly , differentiating along , we get a term Now , in this latter form eplace by are at liberty to do .
We get ( B ) Adding ( A ) and ( B ) , we where is Greenwich mean time .
Mr. G. W. Walker .
Again , deduct from ( A ) and we get These are the terms we found it desirable to introduce .
We have thus obtained the terms required for the data , and clearly we might have taken a llore general form had it been necessary .
We seen that Schuster proposed to explain his term as arising from the permanent field and tidal oscillation .
But there are great difficuJties in the way .
Our analysis confirms the association with the permanent field , but otherwise may silIJply mean differential conductivity of the air as between midday and midnight , and as between sunrise and sunset , the difference arising from the ultra-violet radiation from the sun .
As ards the term depending on mean time , its mode of generation would appear to associate it with a magnetic axis lying in the plane of the equator .
If there is such an axis the intensity will have to be feeble if it is not to produce a very pronounced influence on the eneral field , as indicated by observations .
My main objective has been the mean diurnal variation for the year , letting the seasonal variation take care of itself .
But I anticipated that any correct method of approach would provide by extension for the seasonal appears to be the case in analysis iven .
Seasonal change is provided for , but the terms we have obtained contribute only semi-annual change , not annual change .
The latter is important , but will not of course leave an Huence on the mean yearly value .
Thus it may be expected arise in a way which differs from that we have already used .
But differentiation along supplies the kind of function wanted .
Thus where is the local time and tau The first term contributes to the permanent field .
The second term contributes to the diurnal variation at any particular time of the year .
But the phase passes a cycle in the course of the year , and so contributes when averaged for a year .
seasonal change is , however , beyond the lilnits I had prescribed myself in this paper , and I shall not pursue it further on this occasion .
So far the examination has referred to the -hour terms , but the data indicate the necessity for to the Schuster expression siu2 in the 12-hour terms also .
Our method for the The Ofnetism .
391 24-hour term suggests that we should get new 12-hour terms by a second differentiation along or .
The formulae become more complicated if we are to retain the sun 's annual variation in declination .
We therefore simplify the matter by regarding the sun as remaining in the plane of the equator .
If we thus lose sight of the seasonal variation , we gain in conciseness .
If we take the form ' i.e. , we get as a potential form which leads to The contribution to the 12-hour term is of the form co , but did not find this term of much assistance in the data .
The Schuster term arises by in , which then becomes , and then the , or of If , on the ) , we put in , which then and then form the , or of this , we arrive at 12-hour terms of the forms\mdash ; ( C ) and 3 .
In the latter replace by .
We get 3 eos ) .
( D ) Addition of ( C ) and ( D ) gives the form , while , if we subtract ( D ) from ( C ) , we get the form .
I found it conyenient to add to these forms a term ) may be regarded as a field parallel to the } ) lane of the equator rotating with the angular velocity of the earth but in the ) osite direction .
We have thus formed a simple specification of the terms which seem to fit with the data .
I have no theory to propose as to their , beyond Schuster 's view that they arise from ential conductivity of the regions of the } ) here .
If further data confirm the probable reality of Che terms , it should not be a difficult matter to } ) ress the law of conductivity which uld account for them .
As we have only had to proceed to Cond differentials with r to the sun 's direction , or the earth 's , it does not look as if the differential conductivity required is of a very complex nature .
The Diurnal Variation of Terrestrial Magnetism .
The possibility of the electric and magnetic state of the earth with its translational movement has often attracted attention , although without success .
It may not be out of place here to give a general solution of the electromagnetic equations for a body of conductivity in a straight line with velocity , where is the velocity of radiation .
The tion , which I believe is new , refers to the steady state .
If the components of electric force and magnetic force the equations are and A solution is expressed by ; where is a solution of Primary solution\mdash ; where and other solutions may be obtained by the usual process of differentiation .

