220 Experiments on the Temperature Coefficient of a Kew Collimator Magnet .
By G. A. Shakespeare , M.A. , D.Sc .
( Communicated by Prof. J. H. Pointing , F.R.S. Received June 20 , \#151 ; Read June 26 , 1913 .
) It is usually assumed that the relation between the moment mt of a collimator magnet used in determining the horizontal component of the earth 's magnetic field is dependent on the temperature , the relation being given by the equation mt = m0 ( 1 \#151 ; qt\#151 ; q't The evidence , however , upon which this assumption is based not appearing conclusive , the writer was led to make an independent investigation of the subject , and the present paper is an account of the experiments carried out with that object on a Kew collimating magnet of the ordinary type about 20 years old .
One of the difficulties in any magnetometer observations arises from the variability of the earth 's field .
Variations may be either those of a widespread nature , such as are generally dealt with under the name of diurnal variation , or those of purely'local origin , due to the proximity of electric trams , dynamo-electric machinery , electric mains , or like causes Among these disturbances we may include those arising from earth currents , since these currents , to whatever cause they may be due , are often of much more considerable magnitude that is usually suspected .
It is therefore necessary to eliminate the errors which such disturbances might produce .
The method of attaining this end , which suggested itself to the writer , is the use of what may be called a compensating magnetometer .
The principle of the method is simple .
Let A and B be two plane mirrors .
Then , if a ray of light falling on one of these be reflected on to the other in the way indicated in the diagram ( fig. 1 ) , the direction of the emergent ray will make with that of the incident ray a certain angle dependent on the angle between the planes of the mirrors .
If , now , each of the mirrors be rotated through the same small angle in the same sense , the deviation of the ray brought about by the first mirror will be corrected by the second , and the emergent ray will still form the same angle with the ray incident on the first mirror .
Hence , if the ray comes from a very distant scale , and is received in a telescope , there is no displacement of the image of the scale in the field of view .
In practice it is often inconvenient to have the scale at a great distance , and to overcome this difficulty a collimator may be used , having a scale at Temperature Coefficient of a Kew Collimator Magnet .
221 the focus of the collimating lens .
A more convenient device , however , is to use an ordinary scale at a moderate distance .
But in this case there will be a movement of the image of the scale in the field of view , depending on the distance between the mirrors , and , to secure absence of such movement , the mirror B must he made to move through a slightly greater angle than the mirror A. If A and B be the mirrors attached to two magnetometers , this end can be attained by slightly weakening the controlling field of the magnetometer B by the use of a small subsidiary magnet so arranged as to weaken the field without altering its direction , in a manner to he described later .
In this way the image of a scale viewed with a telescope in the manner indicated may be made to remain stationary , though the direction of the field change considerably .
It is assumed that the disturbing force may be regarded as a second uniform field superposed on the general field of the earth .
But it is useful to note that if the disturbing field be not uniform ( as in the case of a neighbouring dynamo ) , it is still possible to arrange the strength of the field at the compensating magnetometer in such a way as to eliminate disturbance , provided the relative magnitudes of the disturbing field at the two magnetometers remain constant .
But if the field be thus arranged to compensate for a non-uniform field , which varies in a given ratio at the two magnetometers respectively , it will not be such as to compensate for a uniform disturbing field .
It is , for instance , possible to compensate for the effect of a dynamo at a distance of , say , 10 metres , but not at the same time for electric trams at a distance of 400 metres .
In the present experiments compensation was made for uniform disturbing fields , thus eliminating the effect of the trams and of the diurnal change .
Experiments were suspended while dynamos close at hand were being used , though the disturbances due to those were very greatly diminished .
Arrangement of Magnets and Magnetometers.\#151 ; A differential method was 222 Dr. G. A. Shakespeare .
Experiments on the used , in which the effect of the magnet to be tested was balanced by means of a compensating magnet at ordinary temperature .
Several dispositions of the apparatus were tried , which will be understood from the accompanying figure ( fig. 2 ) , where K and C represent the Kew magnet and compensating magnet respectively , and M and M ' the primary and compensating magnetometers .
Of these ( a ) was most satisfactory and was used exclusively in the later work , so that the description given must be taken to apply to this case unless otherwise stated .
Heating Arrangements.\#151 ; The Kew magnet was contained in a brass tube , being fixed therein by its middle point by means of a collar of brass provided with two projecting teeth at the bottom and a central screw at the top , as shown in section at P ( fig. 3 ) .
The collar was brazed to the brass tube at the t^ZZZZTSTZZZZZZTZZZSTi middle of its length , an aperture B being left above the screw to allow the latter to be screwed up when the magnet had been inserted in its place , by means of a gauge bar , through the open end A of the tube .
A short tube from this aperture to the exterior was closed with an indiarubber bung .
This magnet tube was fixed by means of two perforated annuli of brass , D and D ' , in the middle of a large tube EE , which served as a water jacket , through which water could be allowed to flow by means of inlet and outlet tubes as indicated by the arrows .
Brazed on the underside of the outer tube at the middle point was a short rod of brass ( F ) with a conical point .
At G two other hemispherical feet were attached .
These three feet rested on a brass plate which had a hole to receive the point of the cone F and a V groove to take the hemispherical end of one foot G , the end of the third foot resting on the plane surface of the plate .
This elaborate arrangement was necessary to secure that the centre of the magnet should neither approach nor recede from the magnetometer in the course of heating or cooling , for an extremely small movement would entirely mask the true nature of the effect to be observed .
The water jacket was covered with thick felt ; and a lump Temperature Coefficient of a Kew Collimator Magnet .
223 of lead ( L ) was fixed between the feet for greater stability .
The brass plate was cemented down to a slate slab on which the magnetometer and compensating magnet stood .
As an illustration of the necessity for elaborate precautions to avoid small changes of distance , it may be mentioned that the tilting of the floor due to change of position of the observer may be quite sufficient to produce an appreciable change of scale reading .
In fact the changes in humidity and temperature caused minute tilting of the wooden table and wooden magnetometers at first used , thereby producing irregularities which seemed inexplicable .
These troubles , however , were overcome by building up a table made of a slate slab on masonry supports standing on the concrete floor , from which the wooden blocks had been removed , and by making the magnetometers of brass .
The water supply from the main passed through a gas heater on its way to the water jacket and the temperature of the latter could be adjusted to any value between about 9 ' C. and 60 ' C. , and kept constant to about 0'01 ' C. The temperature of the magnet was given by the thermometer T which projected into the middle of the magnet through the rubber bung R. The range of temperature used in most of the experiments was about 10 ' C. to 35 ' C. , i.e.about such variations as a Kew magnet in ordinary use , and occasionally handled , might be subjected to .
In some of the experiments however , the range was considerably greater .
Compensating Magnet.\#151 ; The compensating magnet was contained in a brass tube fitted into a brass box , in which the tube was surrounded by about a litre of water .
In this way the temperature of the compensating magnet changed very slowly with the change of the room temperature .
It was necessary , as will be seen later , that the compensating magnet should be of as nearly as possible the same length and moment as the Kew magnet .
To secure this , the magnet was built up of a number of thin magnetised rods , each of the same length as the Kew magnet , the number being chosen to give approximately the same moment as that of the Kew magnet .
These rods were arranged in the form of a tube inside a brass tube and a thermometer projected into the middle of the compound tubular magnet thus formed .
Magnetometers.\#151 ; Each of these consisted of a box with three sides of glass , the framework and the remaining side being of brass .
This last side had a large circular opening .
When the magnetometers were in position , these two apertures faced one another and were connected by means of a cardboard tube covered with felt , so that a beam of light could pass from the mirror of the one instrument to that of the other without traversing any unnecessary 224 Dr. G. A. Shakespeare .
Experiments on the glass on the way .
The mirrors were sextant mirrors about 2'5 x 2 cm .
, silvered on both sides .
Each was fitted with eight small magnetised needles about 1 cm .
long , mounted on a small slab of cork through which passed a thin copper wire depending centrally from the frame of the mirror .
By this means the needles could be turned to any desired orientation with respect to the mirrors .
The copper wire was continued below the needles and w'as bent into the form of a horizontal ring about 1 cm .
beneath them .
This ring dipped into a dash-pot of oil , the size of* the ring being such as to make the motion of the magnetometers very nearly dead-beat .
The mirror and needles were suspended with a single fibre of cocoon silk about 10 cm .
long .
Adjustment.\#151 ; The two magnetometers were placed in the required position in the same N.S. line .
The mirror and telescope were then adjusted so that the scale was seen by reflection in both mirrors as shown in fig. 4 .
For making the final adjustment it was necessary to be able to superpose on each of the magnetometers equal fields transverse to that of the earth ; for this purpose a long bar magnet was placed at a distance of about 5 metres on the line bisecting the distance between the magnetometers at right angles .
By means of telescopes viewing a scale by reflection from the backs of the mirrors the position of the bar magnet was adjusted so that on rotation of the magnet through 180 ' about a vertical axis , equal deflections were produced in the two magnetometers .
The position of the magnet was then marked and the magnet removed .
The Kew magnet was next placed in position to the W. of the primary magnetometer so as to lie in an E.W. line passing through the centre of that magnetometer , but so as not to produce any deflection of the compensating magnetometer .
The compensating magnet was now brought into position to restore the primary magnetometer to its original direction , while at the same time giving no deflection of the compensating magnetometer .
This was rendered simple by the equality of its moment and length with those of the Kew magnet .
On now bringing into position the disturbing bar magnet , the two magnetometers , being in controlling fields of unequal magnitudes , were deflected through different angles and the image of the scale seen by reflection in both mirrors moved .
In the case represented in the diagram ( fig. 4 ) , the controlling field at M ' is too small .
A small subsidiary magnet , M " , of steel wire , about 3 cm .
Temperature Coefficient of Kew Collimator Magnet .
225 long , fixed in a brass support , was therefore brought to a suitable position S. of the compensating magnetometer to give an increase in magnitude of the controlling field ( without change in direction ) , so that on rotating the disturbing magnet no deflection was observed in telescope T. In this way the deflections due to the disturbing magnet were reduced to less than 1/ 400 of that which would have been given without the compensating device .
It is evident that if the moments of K and C remain constant , small changes in the direction of the earth 's field will not alter the scale reading .
But if the temperature vary , the moments of the two magnets will in general vary to differing extents .
It is necessary , therefore , to determine the change of scale reading due to each degree change of temperature of the compensating magnet , the Kew magnet being kept meanwhile at constant temperature .
This was done separately for each arrangement of the apparatus .
Course of an Experiment.\#151 ; The water was turned on through the water jacket of the K magnet and when the temperature indicated by the thermometer remained steady the scale was read and the temperatures of both magnets were taken .
The gas was now lighted in the heater through which the water passed on its way to the magnet .
The temperature of the magnet could thus be gradually raised to any desired temperature up to about 60 ' by adjusting the gas tap .
Headings of scale and thermometers were taken at suitable intervals of temperature ; the temperature of the magnet being-allowed to become steady before taking a reading .
To avoid rapid fluctuations in temperature the water after leaving the heater passed through a spiral of metal tube immersed in a vessel of water , to serve as a sort of temperature flywheel .
The temperature of the magnet thus could not be suddenly changed .
Value of Sw/ m.\#151 ; So far we have dealt with changes in the moment for a given change in temperature .
It is necessary to know the ratio of such changes to the original value of the moment .
Several ways of doing this were tried , but the following proved most satisfactory .
The primary magnetometer was fitted with a coil consisting of two single turns of insulated wire , one on each side of the needle , the common axis of the turns passing through the centre of the needle system ( as seen in fig. 6 ) , so that , when an electric current passed , a field was produced perpendicular to the horizontal component of the earth 's field and in the same direction as that due to the Kew magnet .
The Kew magnet and compensating magnet having been adjusted in position as already described , the Kew magnet was removed and a current was sent through the coil so as to restore the needle to its original position .
( The effect of the coil on the compensating 226 Dr. G. A. Shakespeare .
Experiments on the magnetometer was balanced by means of a single turn of wire in series with the coil , not shown in the diagram .
) The magnitude of the current C was measured by means of a potentiometer .
The current was then stopped and the Kew magnet was replaced in position and heated through any desired range of temperature , the small current SC necessary to restore the needle to zero position being determined .
It was assumed that Bm/ mwas equal to SC/ C. In this way the value of each scale division could be determined in terms of the total effect due to the Kew magnet alone .
In some of the experiments the results were obtained by means of the current at each temperature , in others by observing the scale deflection and deducing the ratio jm from the value of each division determined at the beginning of the experiment .
The advantage of this method lies in the elimination of the uncertainties introduced by torsion of the suspending fibre .
Separate experiments devised for the purpose showed that the effect of the current in the coil upon the Kew magnet and compensating magnet was negligible .
Results.\#151 ; The moment of the magnet is not simply a function of the temperature .
The relation is complicated by the fact that time enters into it .
Moreover , the moment at any given temperature depends on the previous history of the magnet ; e.g. , in general , the moment at a given temperature depends to some extent upon whether that temperature was preceded by a higher or a lower temperature and also upon the extreme temperature to which it has immediately previously been subjected .
The general effect may be illustrated by reference to two curves shown in fig. 5 .
These two curves are for two consecutive temperature cycles , the second covering a greater range than the first .
The arrows indicate the branches of the curves corresponding to rising and falling temperatures respectively .
The range of temperature in these curves is somewhat greater than in the majority of cases investigated , and in consequence the effects to be noticed are considerably greater , and thereby made clearer .
The dotted lines are straight lines .
It will be noted that the curve for falling temperature in each case lies below that for rising temperature .
The curve AB is represented by the equation mt = mn { 1-3-292 x 10~4(*-17)-l\L5 x 10~6(7-17)2 } .
The moment at 17 ' C. has been taken for convenience as 1000 .
In general , if the cycle has been performed at about the same rate throughout , the sagitta of the cooling branch is less than that of the warming branch .
In fact , the cooling branch is often indistinguishable from a straight line , the actual shape depending largely upon the rate of Temperature Coefficient of a Kew Collimator Magnet .
227 cooling at different parts of the range of fall of temperature .
If the first part of the fall is made rapidly and the last part slowly the curvature of the branch may be reversed , so as to be concave upwards .
It will be seen that in the case illustrated there is a subpermanent loss TEMPERATURE of moment on cooling to the final temperature in each cycle , the loss due to the second being very slightly less than that due to the first .
But if the second cycle had been for the same range of temperature ( instead of a greater ) the second loss would have been considerably less than the first .
228 Dr. G. A. Shakespeare .
Experiments on the For the experiments show that this subpermanent loss is dependent on the highest temperature reached .
In successive cycles of the same range this subpermanent loss rapidly diminishes , so that if , for example , we take three cycles in quick succession , the loss for the second is less than that for the first , while that for the third , is less than that for the second .
The subpermanent loss , however , gradually disappears after the cycle is ended if the temperature is kept constant .
Thus , in the case shown , after a rest of about 48 hours , the finishing point E would have risen so as to lie very nearly on the original curve AB , though it would still be slightly below .
In fact , the experiments show that it is highly probable that any cycle results in a very small permanent loss of moment , even if the range be only one of a few degrees centigrade ; but the higher the extreme temperature the greater the permanent loss .
In the case shown the curve CD lies entirely below the curve BC , but this is not always so ; the curve for the second heating may lie largely above that for the first cooling , especially if the interval of time between the two cycles be somewhat long ( say two or three hours ) .
To illustrate the gradual recovery of the subpermanent loss the result of a cycle in which the range of temperature was from 7 ' to 31 ' C. may be quoted .
At the end of the cycle the moment had at 7 ' C. the value which it previously had at 7'9 ' C. on the warming branch of the curve .
But in 48 hours it had recovered to such an extent that at 7 ' C. its value was that which it previously had at 7T ' C. If the magnet has been heated to , say , 60 ' C. and cooled quickly to , say , 15 ' C. and then quickly heated again to , say , 25 ' C. the recovery is facilitated , so that at the end of the second cycle the moment may be greater than at the beginning of that cycle , though less than at the beginning of the previous one .
In general we may say that the subpermanent effect Qf a cycle is always greater for the first cycle after a period of rest of several days than for succeeding cycles following at short intervals , the range of temperature in each case being the same .
In fact there is a striking analogy between this effect and the permanent lengthening of a wire which has been stretched with a load far below that required to reach the elastic limit .
For it has been shown that a wire does not accurately recover its length after being stretched with a comparatively small load and this subpermanent residual stretch is greater for the first load after a period of rest than for succeeding loads following at short intervals .
This fact suggests that part at least of the effect of a change of Temperature Coefficient of a Kew Collimator Magnet .
229 temperature on the moment of a magnet may be ascribed to the same kind of action as occurs in the stretching of a wire , i.e. to some change in the relative positions of the crystals of which the steel is composed .
Such a change would occur if the crystals had a coefficient of expansion slightly different from that of the matrix in which they were embedded , with a consequent minute amount of rupture on heating ; and the gradual recovery with time after cooling would be accounted for by the gradual growing together again of the crystals .
There can be little doubt that the magnetic properties of the steel depend to some extent on the crystalline structure of the metal .
It may be noted that permanent effects on the moment of the magnet tend to be masked by permanent effects of changes of temperature on the compensating magnet , but if the latter is kept at a fairly constant temperature the compensating device enables the experiments to be carried on continuously for months without appreciable change of zero .
Effect of the Magnetic Field , in which the Magnet lies , on the Temperature Coefficient.\#151 ; Experiments with different arrangements ( ) ( b ) ( c ) ( d ) , fig. 2 , suggested that the temperature coefficient might be slightly affected by the field in which the magnet lay .
To test this point the primary magnetometer alone was used .
This magnetometer was supported on a vertical rod attached to a fixed horizontal slab .
The Kew magnet was fixed on a second slab which was capable of rotating in a horizontal plane about the vertical rod .
The Kew magnet was first laid in an E.W. direction , as indicated at K , fig. 6 , in such a position as to produce no deflection of the magnetometer .
The slab was then rotated through 90 ' so as to bring the magnet to position K ' , when its field would of course be perpendicular to that of the earth .
The compensating magnet was placed at C , fig. 6 , to restore the magnetometer to zero ; the Kew magnet was then removed , and the current required in the coil to restore the magnetometer to zero once more was observed .
The Kew magnet was then replaced at K ' and heated through a certain range of temperature and the small current which was necessary again to restore the magnetometer was measured .
The whole experiment was then repeated with the Kew magnet reversed , end for end , i.e.with its S.-seeking pole to the northward .
Thus the mean coefficient over a given range of temperature could be found when the magnet was lying in fields differing by about 2 H. The results showed that for the magnet under investigation the temperature coefficient is not sensibly affected by changes of field of the order of H. VOL. lxxxix.\#151 ; A. T 230 Dr. G. A. Shakespeare .
Experiments on the Conclusions .
0 We may now give a summary of the conclusions to which the experiments lead:\#151 ; ( 1 ) For a steady rise of temperature , the relation between the moment mt at a temperature t ' above some standard temperature , 0 , may be approximately represented by an equation of the form mt = m0(l \#151 ; qt where q and q are constants , and for a subsequent steady fall a similar equation holds , but with different constants .
( 2 ) If the rise and fall of temperature be not steady , the relation cannot be represented by any such simple equation .
( 3 ) There is always , even for a change of temperature of only a few degrees , a residual weakening of the magnet , which diminishes with time , if kept at constant temperature , until after the lapse of about 24 hours the original value is very nearly ( though perhaps never perfectly ) regained .
( 4 ) It appears that it is easy to over-rate the degree of accuracy with which the moment of a collimator magnet at any temperature can be deduced from that at some other temperature , the application of the ordinary quadratic formula being likely to give an apparent degree of accuracy which is quite illusory .
Perhaps the following way of dealing with the question would be fairly satisfactory .
Suppose the magnet is generally to be used at temperatures in the neighbourhood of 15 ' C. Let the magnet be heated to , say , 30 ' C. , and the mean coefficient \#171 ; over this range be observed .
Then let it be cooled to , say , 0 ' C. , and the mean coefficient j3 over this range be ascertained .
How , let us suppose that we are doing magnetometer observations , and having done the oscillation experiment at temperature 0 we proceed to the deflection experiment .
If now the temperature has risen to d + t , we may put for the moment m0+\lt ; = me(l \#151 ; at ) If , on the other hand , the temperature has fallen to 6\#151 ; t , we may put me-t = me ( 1 + Probably some such method as this would lead to rather more accurate results than would be attained by the simple application of the quadratic formula .
But if the collimating magnet is to be regarded as an instrument of precision its temperature must not be allowed to vary more than a few degrees , and after being handled it should not be used for some hours , as Temperature Coefficient of a Kew Collimator Magnet .
231 such handling may give rise to a subpermanent change of moment which will not immediately disappear .
It is , moreover , necessary to take precautions to secure that the temperature of the magnet is known , especially as , in the ordinary use of a Kew magnetometer , an error of even half a degree in the estimate of the temperature is by no means impossible .
( 5 ) For differences of field of the order of +H the temperature coefficient is not sensibly affected by the field .
In conclusion , I wish to thank my assistant , Mr. E. Simpson , for help in construction and adjustment of apparatus throughout the work .
Note.\#151 ; The principle of the compensating magnetometer may be applied to needle galvanometers , and I have constructed an instrument consisting of two galvanometers to test the utility of such an application .
I have not yet had an opportunity of fully investigating its possibilities , but it seems likely that a high degree of sensitiveness may be attained with a fair immunity from effects of small local magnetic disturbances .
Moreover , for a given resistance , two galvanometers with this arrangement can be made more efficient than one .
On Light-Sensations and the Theory of Forced Vibrations .
By George J. Burch , M.A. , D.Sc .
Oxon , F.R.S. ( Received April 19 , \#151 ; Read June 26 , 1913 .
) [ This paper is published in Series B , vol. 86 , No. 590 .
] VOL. lxxxix.\#151 ; A. u

