]\gt ; [ 33 ] II .
of Hysteresis .
G. F. C. SEARLE , M.A. , Peterhousc , Lecturer in Physics , and T. G. BEDFORD , M.A. , Sussex College , at the Cavendish Laboratory , Cambridge .
icated by J. J. THOMSON , Received May 2 , \mdash ; Read June 6 , 1901 .
Section .
Introduction . . . . .
34 4-6 .
Approximate Theory of the Authors ' lIethod of Iagrletic Hysteresis . . . . .
37 7-12 .
Complete Theory of the Method . . . . . .
40 13-15 .
Determination of the Correction for the Finite Conductivity of the Secondary Circnit , and of the Energy dissipated by Eddy Currcnts 46 16-20 .
Application of the Method to Rods ]inite Length . . . . .
49 21-35 .
Description of the Apparatus . . . . .
52 36-37 .
Practical Lxample of the Method . . . . . . . . . .
Cl 38-41 .
Tests of the Accnracy of the Method . . . . .
( i2 42 .
nergy dissipated by Cnrrents . . . . .
66 43-45 .
Complete Cyclcs and Semi-cycles . . . . . .
68 46-49 .
Effect of Continued Reversals . . . . .
70 50 .
Curves for Zero Stress .
73 51-52 .
Effect of Tension . . . . .
73 53-58 .
Effect of Torsion within the stic Limit . . . . . .
75 69-61 .
Effect of Torsion beyond the Elastic Limit . . . . .
82 62 .
Infiuence of Permnnent upon ths Lffects of Cycles of C3 .
Development of Cyclic after ) ' nent Set . . . .
87 .
Relation with . . . . .
89 70 .
flect of an ; . . . . .
93 nclusion . . . . . . . . . .
AIIED1X I. 72-73 .
On the Heat prodnced by -ddy rents i Tiods of [ FJliptic.all Sections . . . . .
96 74-79 .
On Demagnetising Force due to of initc L ) .
VOL. CXCVIII .
] .
) -L .
G. .
C. AND T. DFORD Introduction .
S1 .
Two ideal physical processes have been deyised as the foundations of two methods of deducing mathematical expressions for the energy dissipated in magnetic material through magnetic hysteresis ; these processes are due to Professor E. WARBURG and to the late Dr. J. HOPKINSON .
In the specimen , in the form of a slender wire , is placed in a netic field due to a pair of permanent magnets so arranged as to produce netic force parallel to the length of the specimen .
The mechanical work spent moving these magnets through such a cycle of changes of position , that the iron is subjected to a cycle of netic changes , is clearly equal to the energy dissipated on account of magnetic hysteresis in the specimen .
In terms of the magnetic quantities the dissipated per cub. centim. per cycle is or ergs , where is the magnetic force and I the intensity of magnetisation .
Professor J. A. EWING has applied the principle involved in WARBTJRG'S theory to the design of a simple instrument by which the hysteresis of any specimen of sheet iron ( for the range of induction to C.G.S. units approximately ) is determined by comparison with two standard specimens supplied with the instrument , and previously tested for resis by the ballistic method .
The principle has also been employed by W. S. FBANKLIN , H. S. WEBB , S and by G. L. W. to obtain absolute determinations of hysteresis .
The theory of the late Dr. JOHN proceeds in a different manner .
The specimen now takes the form of a fine wire bent into a large circular ring , the ends of the wire bein welded together ; the length of the wire is centim. , and its cross-section A .
centim. Let this ring be uniformly overwound with insulated wire at the rate of turns per centim. so that the total number of turns is Nl , and let the wire be without resistance .
Then if be the current at any time , the magnetic force acting on the iron is .
If be the netic induction in the iron , the number of linkages of lines of induction with the electric circu1t at any instant is AINB , and hence , when changes , there is by FARADAY'S law a voltage between the ends of the coil .
We have supposed here that wire is closely wound the iron .
The power spent in driving the current against this voltage is ergs per second .
Using the relation , and noticin that is the volume ( v ) of the iron , the expression becomes HdBfdt .
yied .
Ann vol. 13 ( 1881 ) , p. 141 .
Iagnetic Induction in Iron other Metals , ' 3rd , revised , S 199 . .
S. KLIN , ' Physical Review , ' vol. 2 , p. 466 .
H. , ' Review , ' vol. 8 , p. 310 .
Scien stracts , ' vol. 1 ( 1898 ) , p. 413 .
' ' vol. 176 , 1 ) .
466 .
ON THE MEASUREMENT OF HYSTERESIS .
The work spent per unit volume during any finite change is thus ( 1 ) , the expression found by HOPKINSON .
When the change is cyclic , so that and have the same values at the end as at the beginning of the cycle , we can throw the expression into a different form .
For since we have .
But when goes ( 1 ) through the cycle or ( 2 ) goes from , then , and thus we cover WARBURG'S expression In the present paper we are only concerned with the work spent in causing a complete cycle of netic changes .
We shall always use to denote the energy dissipated by hysteresis per cub. centim. per cycle of magnetic changes , and we shall express in ergs per cub. centim. per cycle .
We have thus ( 2 ) .
To the value of by means of this expression , it has been usual to construct a cyclic B-H curve , best by the method described by and to find its area .
This process is easy enough , but since it involves the vations necessary to find at least ten or twelve points on the B\mdash ; H curve , and the subsequent estimation of the area of the curve after it has been plotted , quite an hour is required for each determination of W. S 2 .
We have given the sketch contained in S 1 for the purpose of contrasting the physical ideas involved in two mathematical methods by which the formulae and have been obtained , and of the manner in which the subject presented itself to one of us in 1895 .
The remarks of S 1 refer only to ( processes and not to the ) lnental methods of studying the effects of resis as exhibited in the relation of I or to H. To the experimental knowledge of the subject the first contributions were made by the independent and nearly simultaneous papers of WARBURG and EWING .
In addition to the theory noticed in S 1 , WARBURG gave magnetometric observations of cyclic I\mdash ; H curves , but his observations were few .
EWING made a much more systematic attack on the subject , using the ballistic as well as the magnetometric method , and determining the values of for a graded series of I\mdash ; H curves for the same specimen of iron .
An account of the subsequent development of the subject , in which EWTNG has had a great share , will be found in his book on ' Magnetic Induction in Iron and other Metals .
' We owe much to Professor wING , It was his hysteresis tester which formed the initial centive to the research described * EWING , ' Magnetic Induction in Iron and other Metals , ' 3rd , revised , S192 .
in the present paper , and , further , much of the general knowledge of magnetism needed in carl.ying out that research has been gained from his writings and from conversations with S3 .
It occurred to one of us some years ago that just as EWING had , in effect , applied WARBURG'S theory to produce a practical hysteresis tester , so it might be possible to apply HOPKINSON'S theory to the design of a method which should give absolute determinations of the energy dissipated through hysteresis as quickly and as accurately as changes.in magnetic induction are foumd by the aid of a ballistic galvanometer .
It was evident that if this could be attained it would be possible to investigate the effects of various physical stress , temperature , the of an electric current , \amp ; c.\mdash ; upon the hysteresis , with a comparatively very small expenditure of time .
A preliminary account of the theory of the method was published in and since then much time has been spent in working out some details which make the method practical ; as only a few weeks in each year have been available for the work , progress has been slow .
In essentials the method is of course well known .
In one of its forms it is in constant use among electrical engineers in testing by means of a watt-meter the energy dissipated in a transformer when -its primary coil is traversed by an alternating current .
In this case there is a deflexion of the watt-meter , and thus the watt-meter method is convenie1lt in commercial work .
From the scientific standpoint , the watt-meter method has the disadvantage that it is not possible to find the limits between which the netic induction alternates without the use of reyolving contact-makers , or oscillographs , or other appliances .
The " " effective\ldquo ; voltage is indeed easily measured , but unless the wave-form of the curve of voltage is known , the limits of the induction cannot be found .
With transformers of commercial dimensions the " " effective\ldquo ; voltage is considerable , } ) when the iron is reduced to a single wire only 1 or 2 sq .
millims. in section the " " effective\ldquo ; voltage for 100 alternations per second does not exceed a few tenths of a volt , eyen when the secondary coil contains 1000 turns of wire .
We know of no method by which so small an alternating voltage can be measured with any accuracy .
The idea which occurred to one of us in 1895 was to use a single revel.sal of the current to produce throw\ldquo ; of a ballistic electro-dynamometer instead of an alternating current to produce a steady deflexion of a watt-meter .
The present paper contains an account of the nent of this idea and of its applications to magnetic research .
So far as we know , the ballistic method of measuring hysteresis is novel .
In the endeavour to make it a practical method , we have met with many difficulties , and the main part of the work has been devoted.to overcoming those difficulties .
* G. F. C. SEARLE , " " A Method of Measuring the Loss of in steresis , \ldquo ; 'Proc .
Camb .
Phil. Soc vol. 9 , Part I. , Noyember 11 , 1895 .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
Of the many advantages of the ballistic method , two may be mentioned .
Thus the induction can be measured simultaneously with the hysteresis far more simply than when an alternating current is used .
Further , the ballistic method enables measurements to be made so quickly as to render experiments easy which would otherwise be practically impossible on account of ths very great time required for the numerous determinations of hysteresis necessary in investigations on the effects of stress or of temperature .
It will appear from the paper that several of the effects of physical changes ) hysteresis which we have studied presented themselves to us as we worked out the method .
We found afterwards that some of these effects had already been discovered by EWING or others , or might have been deduced from their experiments .
These cases we refer to in foot-notes .
The effects of the various physical changes upon the induction have received so much attention from others that we have not thought it necessary to poin out how much of that part of our work has consisted merely in going over old ground .
Theory of the Method .
S4 .
Let an iron ring of section A sq .
centim. and mean circumference centim. be wound with turns of primary windin per centim. , so that the number of turns is Nl .
The current , which passes through and )gnetises the iron , producing the magnetic force , passes also round the fixed coils of a sensitive electro-dynamometer .
A secondary coil of turns is wound oyer the iron , and is connected in series with the suspended coil of the dynamometer , and with an earth inductor , the total resistance of the secondary circuit being S. In finding the total induction through the secondary circuit , we must remember that the secondary will not generally be closely wound upon the iron .
A certain number of lines of induction will in consequence pass through the secondary circuit due to the magnetic force produced by the primary current .
The total number of linkages of lines of induction with the secondary circuit is thus where is some constant .
Let the primary current change gradually ( 1 ) from and back again , so as to complete a cycle , or ( 2 ) from , making a semicycle .
During the change the voltage MdCfdt is set up in the secondary circuit .
If the " " time constant\ldquo ; of the secondary circuit be very small compared with the " " time constant\ldquo ; of the primary circuit , the effect of the selfinduction of the secondary circuit may be neglected , and the current in the secondary circuit may be taken to be IESSRS .
G. .
C. SEARLE AND T. G. BEDFORD Let the couple experienced by the suspended coil , when the currents in the fixed and coils are and respectively , be dyne-centims .
Then since , when the magneti force due to the secondary current is negligible , we have for the couple at any instant Couple If the time of vibration of the moving coil be so great compared with the time occupied by the cycle or semi-cycle , that the cycle or semi-cycle is completed before the coil has sensibly moved from its equilibrium position , the angular momentum acquired by the coil is ( 3 ) , where is the moment of inertia of the coil , and is the angular velocity imparted to the coil by the electro-magnetic impulse .
Now when goes through either a cycle or a semi-cycle vanishes , and thus ( 4 ) .
Let be the greatest angular displacement or " " throw\ldquo ; of the coil , and the restoring couple exerted by the suspension per radian of displacement .
Then by the principle of the conservation of energy , we may equate the initial kinetic energy to the potential energy at end of swing and thus obtain or . . . . .
( 5 ) .
The three constants , and are eliminated , and the " " constant\ldquo ; of the dynamometer is determined in the following manner .
Let a constant current flow in the primary circuit through the fixed coil , and while this current is passin let the earth inductor be inverted so as to produce a known change , , in the number of linkages of lines of induction with the secondary circuit .
The timeintegral of the current thereby produced is , and hence the initial angular momentum of the suspended coil is . . . . .
( 6 ) .
If throw produced , have ( 5 ) , and hence by ( 6 ) and ( 3 ) , ( 7 ) .
Thus by ( 4 ) ( 8 ) .
ON THE MEANT OF Since it is only the ratio of the angles and which appears in the formula for , we may take for the ratio the ratio of the two " " throws\ldquo ; of the spot of light along the scale , provided that the " " throws\ldquo ; not so that differs appreciably from The effect of damping has so far been neglected .
If be the logarithmic decrement , we must write ( 5 ) , where is now the observed " " throw The " " throw\ldquo ; must be treated in the same way , and hence the ratio is unaffected and the expression ( 8 ) remains true .
S 5 .
If , instead of making the primary current go through a complete cycle , we make it go through a semi-cycle from or vice versed , we can very conveniently combine the measurement of the hysteresis by the dynamolneter with the measurement of the induction by a ballistic galvanometer .
if a ballistic galyanometer be connected in series with a second secondary coil wound upon the specimen , the throw of the galvanometer giyes the value of , where and Bg are the algebraical values of corresponding .
In the ideal case , and then ) .
Whether this ideal state obtain or not , we shall denote by it the mean maximum induction .
When the specimen takes the form of a ring , the ballistic galvanometer only enables us to find ; it gives no information as to or .
This knowledge could , however , be obtained in the case of a long straight specimen by slipping the secondary coil off the specimen .
When , as in our case , there are two observers , the observations and caIl be made simultaneously .
S 6 .
To take the specimen through a complete cycle , we must make the current go through the cycle and then from - .
If the iron has reached a cyclic state the throws of the galvanomeler will be the same in magnitude , ough opposite in direction for the two reversals of the current .
But it will often happen that , on account of the previous treatment of the iron , differs consider , and in this case the throws of the dynamolneter differ in magnitude for the two reyersals , they are in the same direction .
A little consideration will show us how to proceed in this case .
Let ( fig. 1 ) be a B\mdash ; H curve , and let be drawn parallel to OH .
By reversal of rent , let the iron be caused to go ough the changes represented by the curve abca ' .
From to is positive and is negative , while from to both and are negative .
Hence the value of for the semi-cycle to is given by area .
Simi- larly for the cycle - to is giyen by area area .
The of for the whole cycle is thus the area of the B\mdash ; H curve If the two throws of the dynamometel for the two semi-cycles , we have and hence , since is the sum of the two ntities on the left sides of these equations , .
( 9 ) .
Thus to find we have simply to add together the two throws for the two semi-cycles , and then multiply the result by the factor The expression ( 9 ) is the fundamental formula of the method employed by us for the measurement of hysteresis .
From it we see that all that is needed in addition to the electro-dynamometel for the measurement of is an ampere-meter by which to determine , and an earth-inductor or some other means of producing a known change of induction ( P ) through the secondary circuit .
Complete of the Method . .
In the elementary theory of S 4 , the resistance , , of the secondary circuit is supposed to be so large , and the induced current in consequence so feeble , that the agletiC force due to the induced current is negligible in comparison with .
that due to primary current .
It is thus necessary to proceed to a closer examination of the theory in der to find the correcting term which appears when is only finite .
In this examination we take account also of the energy dissipated by the eddy currents , which circulate in the specimen in consequence of the variations of the applied magnetic force .
We use the notation already employed in the elementary theory . .
In the primary circuit let the magnetising coil and the dynamometer coil be placed next to each other in the circuit , and let be the voltage at any instant between the ends of the conductor so formed .
Now the induction through either circuit depends not only upon the magnetic induction , , in the iron , but also upon the force in the space between the coils and the iron , as well as upon the 1lftgnetic force near the coils of the din .
each case the magnetic force depends only , and not at upon , since the iron is formed into a .
TlnIs if esistance of the circuit between the two points between the voltage is , the equations for the primary and currents may be ttcn NlAB ( 10 ) , .
( 11 ) , ON THE MEASUREMENT OF MAGNETIC HySTERESIS .
where is the average value of the induction over the section of the specimen at any time .
Here and are constant , while varies slightly as the suspended coil turns round .
By the principle of the conservation of energy , the work done by the voltage in any time is equal to the energy dissipated in the specimen , together with the heat produced in the resistances and , and the kinetic energy acquired by the moving coil of the dynamometer and the increase in the magnetic energy of the system .
Now let be the energy dissipated by hysteresis per cub. centim. for two semi-cycles , so that the loss per cycle is .
Further , let be the of the energy dissipated by eddy currents , so that the loss per cycle is .
Then the total energy dissipated in the specimen in a semi-cycle is .
If be the deflection of the suspended coil , then the couple tending to increase is dyne-centim .
The rate of working of this couple at any instant is thus or , and hence total work done is .
This is therefore the kinetic energy acquired by the coil .
Then , if be the magnetic energy at the beginning and end of a semi-cycle , we have ECdt \mdash ; T. But from ( 10 ) ECdt NlAB Comparing these expressions , we find , since for a cycle or a cycle , The integrations are to be effected between the limits and , where denotes any instant before the primary current begins to change , and denotes some instant towards the end of the change when the primary current has with sufficient accuracy reached its final value .
It follows that at both limits .
From ( 11 ) we find since vanishes on integration .
Adding this expression to the last one , we obtain * We take the space-average because the rate of heat production is not uniform over the section ; wheIt the section is circular and is very small the rate at any point is proportional to the sqnare of the distance of the point from the centre of the section .
( See Appendix I. ) VOL. CXCVIII .
The in this formula which involve are the complete differential of , and thus vanish on integration , since at both limits .
We thus have CN lA If we multiply ( 11 ) by and integrate we obtain ( ) Thus , by addition of the last two equations , ( ) The first integral in this expression is proportional to the " " throw\ldquo ; of the moving coil , for , by ( 7 ) , if we measure in the right direction .
As regards the second integral , ( ) But since the time of vibration of the suspended coil is comparatively large , the change in is practically complete before the coil has moved far from its equilibrium position .
Hence may be treated as constant for the whole range of integration , and thus , since at both limits , the value of the integral is zero .
As regards the fourth integral , since vanishes at both limits .
Collecting these results , we have The integral in this expression is the correction which makes its appearance when we take account of the finite conductivity of the secondary circuit .
It will suffice to use in it the value of Sc which obtains when is negligible in comparison with ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
We may also treat as constant .
When this is done we obtain If we add together the results for a pair of semi-cycles , the quantities and disappear when a cyclic state has been established , for then the magnetic energy is the same at the end as at the beginning of a cycle .
We may then replace by and by , but we must remember that the integral is to be taken completely round a cycle .
The expression can be put into a more convenient form it we notice that by the rules of approximation is to be taken as zero in the terms under the integral sign .
We thus replace by , and then writing for , we have , on division by .
( 12 ) , the final formula .
Now , unless the specimen be so thick , and the variation of so rapid , that during part of a complete cycle the verage induction for the whole section increases while the applied magnetic force diminishes , is always positive .
Hence , since always has the same sign as , the correcting integral is always positive .
Thus the true value of is less than that calculated from the throw of the dynamometer on the assumption that is infinite .
S9 .
Before we can.apply equation ( 12 ) in finding , we must either know that X * We can easily obtain with rough approximation the condition which must be satisfied in order that may be negligible in comparison with .
From ( 11 ) we have , whence . . .
Thus the approximation is valid if is small in comparison lvith .
A sufficient idea of the magnitudes involved is obtained by treating as proportional to C. In this case must be small in comparison with .
But , if denote the voltage of the battery , the characteristic of is Kd where is now the resistanoe of the whole of the primary circuit , and depends mainly upon the choking coil ( S33 ) in the circuit .
Hence , treating as constant , Thus the condition for the validity of the approximation can be expressed in simple form by saying that must be small in comparison with is negligible else be able to determine it .
In most of our experiments the specimens have been fine wires , and X has been insignificant in comparison with , but with rings of solid metal , such as those used by Mr. R. L. WILLS , *with a sectional area of over 1 sq .
centim. , X becomes of real importance .
Now from the behaviour of the eddy currents in a rod of circular section , we may assert that when the " " time constant\ldquo ; ( inductance/ resistance ) of the primary circuit is large compared with where is the radius of the largest circle inscribable in the section , and and are the permeability and specific resistance of the material , then the eddy current at any point may be on the assumption that the magnetic induction has at any instant the .
same value at all points of the section ( Appendix I In this case the eddy current at any given point is proportional to , and hence , as the method of " " Dimensions\ldquo ; shows , the of the rate at which heat is generated per volume may be written ( 13 ) , where is a constant upon the form of the section .
On reference to the meaning of X it will be seen that the total rate of heat production by eddy currents per unit length of the specimen is .
For a circular section , and for a square section .
( Appendix I. ) We can now write as follows:\mdash ; NC ' ( S13 ) . . . . . . . . . .
( 14 ) .
If the specimen be built up of similar wires so that the total cross-section is then X is times the value for a single wire of section A. The wires must be insulated from each other so that there are no eddy currents from wire to wire .
The quantity , which occurs in the correcting integral , is for a given specimen a nearly definite ( double valued ) function of and therefore of , for given limits , provided that the " " time constant\ldquo ; of the primary circuit is large compared * Mr. R. L. WILLS , of St. John 's College , 1851 Exhibition Scholar , began in 1900 , at the Cavendish Laboratory , a series of experiments on the effect of temperature upon the energy dissipated through hysteresis in iron and alloys of iron , in continuation of his work on the " " Effects of Temperature on the Magnetic Properties of Iron and Alloys of Iron\ldquo ; ( ' Phil. Mag July , 1900 ) .
At the suggestion of Professor J. J. THObISON he employed the method described in the present paper , while we gladly furnished him with some of the apparatus employed by us in our own researches .
Ib was plain that in the specimens used by .
WILLS the energy dissipated by eddy currents was comparatively much greater than in our own specimens , and so might be far from negligible .
We thus led to extend the theory so to take account of the eddy current loss , and to devise a ntethod of determining that loss , ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
with .
( Appendix I. ) Quite apart from the effects of eddy currents , appears to depend slightly upon the rate of variation of , but so slightly that we may for our present purpose consider as having two definite values for any given value of when varies between given limits .
Hence for a given specimen , the correction to be subtracted from the value of calculated from the throw of the dynamometer by the elementary theory will be increased -fold if , for every value of which occurs in the cycle , be increased -fold .
The part of the correction involving the resistance , , of the secondary circuit , is inversely proportional to S. It also depends upon and and diminishes with those quantities .
When the secondary coil contributes practically the whole of and we have approximately . . . . .
( 15 ) , where is the mean area of one turn of the secondary coil .
In this case , unless is many times greater than and are negligible in comparison with the quantities added to them in ( 14 ) , since for iron is generally very large .
Referring to the definition of and in S8 , we see that as far as these quantities are concerned the correction is made as small as possible by winding the secondary windings as closely as possible upon the iron , and by making the self-induction of the rest of the secondary circuit as small as possible .
To secure the latter point the earth inductor employed to standardise the dynamometer should be removed from the secondary circuit , and a non-inductive coil of equal resistance inserted in its place when the dynamometer readings for are taken .
S10 .
For different specimens of iron the correction will depend upon Indeed since and can be made comparatively small , the correction depends mainly upon the of It is , of course , impossible to assign any definite value to for an particular specimen , for the ratio varies very greatly for different parts of the B\mdash ; H curve .
All we can say is that if for any part of the B\mdash ; H curve has a very high value , and if for the corresponding value of the rate is not very small , then this portion of the B\mdash ; H curve may make a considerable contribution to the correcting integral in ( 14 ) .
Now the maximum value of varies greatly in different kinds of iron , being small in steel and very large in good soft iron .
According to EWING'S experiments in the case of considerable magnetic forces , the maximum value of for glasshard steel is about 300 , while for soft-iron wire , suitable for the manufacture of transformers , its value rises to 13,000 .
The maximum value of for the iron is accordingly about 2000 times its value for the steel .
It is thus evident that the method may give fairly accurate results for steel without any special precautions to secure that should never be large , it may give quite inaccurate results for soft iron unless the proper precautions are taken .
A considerable portion of the time spent upon this research was occupied in investigating the precautions necessary to secure accuracy .
S11 .
If we consider the effect of changing the area of section of the iron we shall see that the part of the correction due to the conductivity of the secondary circuit increases as the section diminishes .
For neglecting and , ( 14 ) becomes .
( 16 ) .
Now for a given kind of iron , if we wish to obtain the same throw of the dynamometer , when we halve A we must double .
Thus we must regard An as determined by the sensitiveness of the dynamometer .
Hence the part of the correction which depends upon is proportional to , and therefore inversely proportional to A. This explains the difficulty we met with in the earlier stages of the work .
We were able to get accurate results for specimens built up of fine wires , with a total section of 1 sq .
centim. , though we failed to do so for specimens with a section of 1 or 2 sq .
millims. The part of the correcting term in ( 16 ) depending upon is inversely proportional to .
The circumference of the ring should therefore be large .
When , as in most of our experiments , a straight wire of finite length is used , the secondary coil should be wound upon a long bobbin and not be heaped up at one part of the wire .
S12 .
Effect due to Air alone.\mdash ; If there be no iron in the system everywhere , and ; if , further , there be no metal , , brass , inside either the primary or the secondary coil also .
In this case ( 14 ) reduces to ( 17 ) .
Since is necessarily positive , has the same sign as , and hence the integral cannot vanish .
There will , therefore , always be a small throw of the dynamometer even when there is no metal in the coils .
But since is always positive and is generally large , it is easily seen that unless and ( S8 ) are very large in comparison with the quantities to which they are added in ( 17 ) , the throw , when the iron is out , is small compared with the part of the throw which is due to the correct .
ing integral in ( 14 ) , when the iron is in .
This deduction we verified by experiment , for in many trials we never found any case in which we could detect any throw when there was no metal in the magnetising coil .
Determination of the Correction for the Finite Conductivity of the Secondary Circuit , and of the Energy by Eddy Cnrrents .
S13 .
It is convenient to have a single symbol for the quantity and we shall use for this purpose .
Thus is the value of the hysteresis ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
which is calculated from the throw of the dynamometer coil simply , without any regard to the corrections .
If , further , we denote by the correction due to the finite conductivity of the secondary circuit , we can write ( 14 ) in the form . . . . .
( 18 ) , where Comparing ( 18 ) with ( 14 ) , we see that we can put .
( 19 ) , where ; and depend only upon , and ot upon S 14 .
The form of at once suggests the way to find .
For if correspond to , and to , then since X and are unaffected by the change in \mdash ; .
But , and hence . . . .
( 20 ) .
Thus , by observing the values of found in two experiments with two values for the values of and can be determined .
In this determination it is not necessary to find the throw for the earth inductor for each value of S. If there were no damping we should have , so that could be calculated if were known for the one resistance S. But the logarithmic decrement depends upon , being of the form where is proportional to the current , and hence each throw must be corrected for damping before it is used in the calculations .
The easiest way of making the correction is to add to each throw a quarter of the difference between it and the next elongation on the same side when the dynamometer coil is allowed to continue in vibration .
Since the throws are of very different magnitudes , it is necessary to correct each one for the difference between and , if accuracy be desired .
S 15 .
The correction X is of great importance when the section of the specimen is of considerable area .
In this case the secondary coil will generally be wound so closely upon the iron that and will be negligible in comparison with the quantities to which they are added in ( 14 ) .
Under these conditions the value of can be calculated , for now by ( 16 ) , and hence , when is known , X can be calculated ; for iron , ohm per centimetre cube approximately .
It is not practicable to calculate X from the value of found by varying , for an examination of the numerical magnitudes shows that is now a large numeric\mdash ; in Mr. WILLS'S research something like 1000\mdash ; while experiment shows that is so small as to elude observation .
To determine or , when the section of the specimen is considerable , we must vary in a known manner .
This could be done in the following Let the poles of a battery of voltage be joined to the points ( fig. 2 ) , and let the resistance between and through the battery be T. Let and be also connected through the primary coil , the fixed coils of the dynamometer , and a Fig. 2 .
choking oil , the.resistance of this part being R. Let the steady current , when and are not in contact , be .
Now let and G. be put into contact .
The current in the primary coil then has the characteristic . . . . .
( 21 ) , where depends upon , since it involves the value of for the core of the choking coil ( S33 ) .
After the current has sunk from to some very small value , let and be separated again , but before the separation let be reversed .
The characteristic is now ( 22 ) .
The cul.rent now gradually attains the value .
The whole process is practically complete in a fraction of the time of vibration of the dynamometer key described in carries out this process exactly , provided that the resistances are made zero .
We find that the time for which no E.M.F. acts on the circuit through the coil is about second , while a rough calculation shows that with our } ) aratus the current sinks very nearly to zero in that time .
Suppose now that and are all increased -fold .
Then remains unchanged , and for any given value of , during the whole time of variation , is also increased -fold .
Thus , if the ratio be kept constant as well as the maximum current for any value of is proportional to E. Hence . . .
( 23 ) , where is a quantity independent of E. A method of finding similar to that employed for is now available .
For if correspond to , and to , then remains unaltered , nnless the hysteresis depends upon the which the magnetic occur : But , and hence . . . .
( 24 ) .
Thus , by observing the values of found in two experiments with two values for the values of and can be determined .
Practical examples of these two methods are given in SS36 , 41 , 42 .
ON THE MLASUREMENT OF MAGNETIC Application the Method to Rods of Finite Length .
S 16 .
In many experiments it is convenient , and in most of our experiments it was necessary , that the specimen should be a straight rod of finite length instead of a rin .
We must therefore consider what modification of the theory is necessary in order to make it fit this case .
The magnetising solenoid will not be infinite in length , but the correction due to the finite length is very small when the diameter is small compared with the length .
For if the mean radius of the windings is , and the whole length of the solenoid is , then at a point within the solenoid whose distances from the central plane and from the axis are and , the components of the magnetic force are . .
( 25 ) , . .
( 26 ) .
The magnetic force is thus very nearly constant in the central parts of the noid , and it is practically suffic ent to substitute for in the expression the quantity In our experiments we had approximately , and thus the magnetic force at any point within the centra120 centims. of the solenoid did not differ from that at the centre by more than per cent. S 17 .
When the specimen is a finite straight rod instead of a ring , " " poles\ldquo ; are developed upon it , and these give rise to a demagnetising force , , thus causing the magnetic force at the centre of the rod to differ from that calculated from the currents in the primary and secondary circuits .
As we are now dealing with a correction it will suffice to find the effect of the netising force on the assumption that the elementary theory is applicable so that the force due to the secondary current is negligible .
We further suppose that the secondary coil , which is placed round the centre of the rod , is short compared with both solenoid and rod ; we can then treat the demagneti , force , , as well as the magnetic force , , due to the solenoid , as constant within the secondary coil .
The rod may be either longer or shorter than the solenoid .
Let A be the area of the section of the iron and the mean area of one turn of the secondary coil .
We only restrict A to be constant in the neighbourhood of the secondary coil ; the section may , if convenient , increase or decrease considerably at a distance from the secondal.y coil .
If be the magnetic at the centre of the specimen , we have ( 27 ) .
VOL. The secondary current is given by . . . .
( 28 ) .
Substituting for from ( 27 ) we find .
that for a cycle or a semi-cycle , we have , on using ( 7 ) , ( 29 ) .
Hence the throw of the dynamometer is larger than that corresponding to and thus if we calculate from the throw of the dynamometer we must subtract a correction depending upon the integrals and .
The correction can be found only when is known both as a function of and also as a function of H. S 18 .
Lord RAYLEIGH* was the first to show how to correct an I\mdash ; H curve or the demagnetising force .
He supposes that the specimen is an ellipsoid , and that the applied magnetic force is constant throughout its volume ; the magnetic quantities , and are , in this case , constant throughout the specimen , and thus we can write , where is a constant factor ( usually denoted by N ) .
This relation between and I allows the I\mdash ; H curve for an infinitely long ellipsoid to be deduced from the curve for a short ellipsoid by " " shearing\ldquo ; through a distance everywhere proportional to I. Some investigators , supposing that the demagnetising force , , at the centre of along cylinder is the same as that for the ellipsoid inscribable in the cylinder , have applied Lord RAYLEIGH'S construction to the case oflong cylinders .
Others , such as Dr. H. BOIS , though ayoiding this error , have assumed that can be expressed in the form , where ) is a function of the ratio of the length to the diameter , but is independent of I , the intensity of magnetisation at the centre of the cylinder .
But it is easy to see that , quite apart from the influence of hysteresis , cannot be constant , since the permeability of iron is not independent of the magnetic force .
For , on account of the demagnetising action of the ends of the rod , the magnetic force near the ends differs from that near the centre of the rod , and thus the rod has , in effect , different values of in different parts .
If were constant for each part , would still be constant , but in the actual case , when the applied magnetic force is " " On the Energy of Magnetised Iron ' Phil. ' 1886 , vol. 22 , p. 175 , or ' Scientific Papers , ' vol. 2 , art .
139 .
The lsIagnetic Circuit in Theory and Practice , ' p. 41 .
Dr. BOIS , however , describes ( p. 123 ) the experiments of LEHbIANN upon the magnetisation of a toroid lvith a radial slit , and points out that LEHbIANN'S results show that the demagnetising factor increases , but only gradually , as I increases up to about half its maximum value ; beyond this point the increase is more rapid .
The experiments were not arranged so as to show the effects of hysteresis .
ON MEASUREMENT OF MAGNETIC RESIS .
1 changed , the value of for each part of the rod changes , the change being greater for some parts of the rod than for others .
The magnetism which appears at any part of the rod is thus not proportional to the value of I at the centre of the rod , and accordingly is not constant .
The curves obtained by Mr. C. G. show to how great an extent the distribution of magnetism depends upon the applied magnetic force .
But , further , if the rod be put through cycles of magnetic changes , as the ends of the rod are approached the range of the magnetic force changes , on account of the demagnetising action of the ends .
Hence , since the behaviour of iron in regard to hysteresis depends upon the range of the magnetic force , the demagnetising force , at the centre does not depend simply upon the intensity of magnetisation , I , at the centre of the rod , but depends also upon the manner in which that value of I has been reached .
Thus will exhibit hysteresis with respect to I and hence also to and , each quantity referring to the centre of the rod .
S 19 .
We give in Appendix II .
a simple experimental method of both as a function of and as a function of , the magnetic force due to the solenoid .
Since we may replace by in the second correcting integral in ( 29 ) .
When has been found in terms of and by this method , the values of and can be found by measurement of the areas of the h\mdash ; B and h\mdash ; H curves .
Now in ( 29 ) the integral has the factor , and hence disappears when , i.e. , when the secondary coil is wound infinitely closely upon the iron .
In most cases it will be convenient that the secondary coil should be wound upon a tube large enough to pass easily over the rod , and will then be considerably larger than , though need not exceed 10 or 20 except for very thin rods .
Now for a given step the corresponding step is always by comparison small , since is very small\mdash ; perhaps \mdash ; in the steep parts of the cyclic B\mdash ; H curve , and never rises above 1/ 100 unless the iron is well " " saturated Hence the term will generally be negligible in comparison with For a bundle of ten iron wires 47 centims. long and sq .
centim. in total area of section , placed in a solenoid 47 centims. long , and furnished with a secondary for which sq .
centim , so that , we found ( S 78 ) In this case the second integral introduces a correction of about 1 per cent. , and the third one is negligible .
The limits of , and those of .
The value of at these limits * ' On the Distribution of Magnetic lnduction in a Long Iron Bar ' Proc. Physical Society , ' vol. 16 , p. 509 , or ' Phil. Mag September , 1899 .
SSRS .
G. F. C. AND T. G. DFORD For a single wire 47 centims. long a\amp ; nacute ; d 00412 sq .
centim. in section , placed in a solenoid 60 centims. long , we had , and found ( S79 ) , The limits of , of , and of S20 .
These corrections are also applicable to the area of a cyclic B\mdash ; H curve drawn by the aid of a hallistic anometer .
With respect to the maximum induction , when the current is reyersed , the throw of the galvanometer measures , where it will sufficient to take as , since is small compared with .
But in finding the maximum netic force , the correction may not be yible , and must be subtracted from the quantity .
The value corresponding to is easily found by the method described in Appendix II .
DESCRIPTION 0F THE APPARATUS .
S 21. .
We now pass on to describe the instl.uments employed in the experiments .
It will perhaps conduce to clearness if we leave till last the description of the special appliances which were designed in order to secure accuracy .
The S22 .
In all except the preliminary part of the work a Weston direct reading ampere-meter was used for the measurement of the primary current .
The instrument only read up to amperes , but when the largest number of windings on the solenoid ( S 24 ) was employed this current gave a magnetic force of about 108 C.G.S. units .
The instrument reads in one direction only , and thus it is necessary to join of its terminals directly to one pole of the battery , so that the current always flows through it in the same direction .
The Earth Inductoz S 23 .
The earth inductor , which served throughout our experiments to produce a standard change of induction , was a simple coil of .
100 turns , a mean radius of about centims. ; its resistance is ohms .
Using a Chrk cell as a standard of , the change in the number of linkages of lines of induction with the circuit when the coil is turned over through 18 about a horizontal axis was found to be C.G.S. units .
We prepared a resistance coil of copper wire with the same resistance as the earth inductor .
We could thus substitute this coil for the earth inductor when we desired , for the reason given in S 9 , to get rid of the self-induction of the earth inductor and ON THE MEASUREMENT OF HYSTERESIS .
5.3 at the same time to keep the resistance of the circuit unchanged .
Since both coils were of copper , their resistances were equal when their temperatures were equal .
The Magnetising Solenoid .
S24 .
In our experiments the specimens of iron have taken the form of straight wires , and in consequence the magnetising coil has been a straight solenoid .
We used a straight wire instead of a ring in order to be able to apply tension or torsion to the specimen .
The solenoid is formed of several independent layers of wire wound upon an ebonite tube 47 centims. in .
It is essential that the tube should be of non-conducting material , for otherwise the currents induced in it would cause the magnetic force to differ lerably from that calculated from the formula ; this effect would cause a considerable error in the measurement of the hysteresis .
By combining the independent layers in different ways we could vary the magnetic force due to unit C.G.S. current by steps of about 50 from to C.G.S. units .
The magnetic forces due to unit current in each of the four coils are as follows .
AA , , BB , , MN , The resistances of the coils are , 54 , , and ohms respectively .
The Coils .
S 25 .
The secondary coils were formed of fine insulated copper wire wound on narrow tubes of ebonite or glass , through which the specimen passed .
Insulating material was used for the bobbins to avoid the induction of currents in them .
In our earliest experiments it was found that metal tubes caused very large errors by the action of the currents induced in them .
Several coils were used with the dynamometer , the number of windings varying from S00 to 1285 ; the coil of 1285 turns had a bobbin centims. long , thus conforming to the recommendation of S 11 .
The mean area of this coil was785 sq .
centim. The Ballistic lvanomete : S26 .
The ballistic galyanometer was made by one of us , with some assistance from .
W. G. .
The magnet system consists of two vertical magnetised wires , each about centims. long and 1 millim. in diameter .
These are fixed parallel to each other at a distance of 1 centim. , the north pole of one magnet being opposite the south pole of the other .
When parallelism is secured , the magnetic system is necessarily astatic , however much the magnets may vary in strength .
The system is suspended from a torsion head by a single phosphor-bronze wire inch in diameter and 12 centims. in length .
The torsion of this bronze wire supplies the restoring couple .
There are four coils , each containing about 250 turns of No. 18 B.W.G. copper wire , the total resistance of the coils in series being ohms .
The coik are arranged so that the upper end of the magnet system is at the centre of the upper pair of coils and the lower end is at the centre of the lower pair of coils .
The motion of the magnets is observed by the aid of a lamp and scale .
The galvanometer is very sensitive , and thus measurements of for quite thin iron wires can be made with a comparatively small number of turns of secondary winding ; this is often a point of some convenience .
We have found the instrument very efficient .
The time of vibration , 12 seconds , is long enough to enable the ballistic throws to be read with ease .
Since the needle is practically astatic , the zero depends only on the action of the bronze wire and not upon the earth 's magnetic field .
The result is that the zero is remarkably constant , often not changing by more than one-tenth millim. during several hours .
The only disadvantage is that there is so little damping that , to bring the needle to rest in any reasonable time , it is necessary to use a coil of wire placed near the galvanometer in conjunction with a Leclanche ' cell and a tapping key .
A little practice enables the observer to bring the spot quickly to rest .
The .
restoring couple varying as the angle of deflexion instead of as its sine , the time integral of a transient current is proportional to the angle of throw instead of the sine of half that angle .
We verified by experiment that this is accurately obeyed .
With rise of temperature the netic moment of a magnet diminishes , and we consequently found that with the same resistance in circuit the throw due to a given change of induction was rather less on a hot than on a cool day .
The logarithmic decrement depends upon the resistance in circuit with the galvanometer .
But in every case the resistance of that circuit was kept constant during a set of observations , and thus all error due to this cause was avoided .
When we desired to draw a cyclic B\mdash ; H curve for a specimen of iron , we practically followed the method described by Professor J. A. The Electro-dynamometer .
S27 .
The electro-dynamometer employed in the later experiments has a pair of fixed coils , each formed of 250 turns of No. 20 B.W.G. cotton-covered wire wound on ebonite bobbins .
The mean radius of the coils is about 5 centims. , and their resistance in series about 4 ohms .
In order to avoid induced currents there were no large pieces of metal near the coils .
In the central space hangs the suspended coil .
consists of 190 turns of No. 40 B.W.G. silk-covered wire .
The mean radius of the coil is centims. and its resistance about 28 ohms .
The coil is attached along a diameter to a stiff brass wire , whose upper end carries a mirror .
The mirror is placed so far above the centre of the coils that the beam of light from the lamp passes above the outside of the fixed coils .
The moving coil is suspended by a phosphorbronze wire inch in diameter , and about 4 centims. long .
The other connexion " " Magnetic Induction in Iron and other Metals 3rd Ed. Revd .
, S I92 .
ON THE hIENT OF MAGNETIC HYSTERESIS .
with the moveable coil is formed by a second piece of the same bronze wire , about 6 centims. long , the total resistance of these two bronze wires being about 6 ohms .
The time of vibration of the coil so suspended is about seconds .
It was found best to allow the lower bronze wire to be quite slack .
The plane of the suspended coil then takes up a definite position due to the control of the bronze wires , and is unaffected by slight tiltings of the instrument .
If the lower wire is pulled tight , and if the centre of gravity of the suspended system does not lie on the line joining the points of attachment of the bronze wire , a slight tilt of the instrument causes the coil to turn through a considerable angle .
The dynamometer was placed so that the axis of the fixed coils was at right angles to the magnetic meridian .
The earth 's magnetic force had in consequence no action upon the suspended coil in its equilibrium position .
By diverting the primary current from the fixed coils and allowing the secondary current to flow through the suspended coil , we found that the earth 's magnetic force produced no deflexion of the suspended coil .
Our experiments are therefore free from any error arising from the action of the earth .
The motion of the coil was observed with the aid of a lamp and scale in the ordinary way .
A little care is required in soldering the phosphor-bronze wire to its attachments .
If the soldering bit is too hot , and if it comes into contact with the bronze , the solder alloys with the bronze to such an extent that the latter becomes very weak .
Following a suggestion of Mr. W. G. , we found that a strong joint is easily made in the following manner .
The stout wire to which the fine bronze wire is to be soldered has a fine hole drilled along its axis ; this hole is filled with solder which is kept melted by applying the soldering bit to the side of the wire .
The fine bronze wire is then inserted into the melted solder , and the soldering bit is once removed .
The result is a very satisfactory joint .
The Reversing Keys .
S 28 .
In the earliest experiments an ordinary ( mercury ) rocking key was used .
Its defect is that , when contact is broken , the high resistance of the spark causes the current to sink to zero very rapidly .
When contact is re-established the current rises at a rate depending upon the self-induction and the resistance of the circuit .
S 29 .
To avoid all sparking , and to cause a very gradual change of the current , we next tried a liquid commutator similar to that used by Professor EWING .
drum of insulating material , carrying two copper plates , , revolves between two other plates , , in a vessel containing a solution of .
copper sulphate .
The plates are connected to the battery , and the plates to the primary circuit .
The current flows in one direction in the circuit when A is close to , and in the opposite direction when A is close to D. The reversal is thus very gradual .
But the key was not altogether convenient , and its use was abandoned .
S 30 .
In a third arrangement an ordinary ( mercury ) rocking key was used , a noninductive coil of 100 ohms being put in parallel with the primary circuit , , as in fig. 3 .
When the self-induction of the primary circuit is large the current does not stop very rapidly when contact is broken , for the primary current can still flow on through , and thus the decay of the current is much less rapid than when it has to overcome the high resistance of the spark , as in S28 .
When contact is re-established the current rises in exactly the same manner as with the simple rocking key .
The practical objection to the method is that it does not allow of the current being directly measured by the convenient Weston ampere-meter , for the current must always flow.in the same direction through this instrument .
A Kelvin graded galvanometer or a shunted mirror galvanometer , MG , was accordingly used with this key .
Fig. 3 .
Fig. 4 .
S 31 .
The fourth reversing key used is shown in fig. 4 .
The battery is connected through the Weston ampere-meter WA with the terminals , on an arm working about a pivot at O. The terminals and are connected to two brass springs , one at each end of the arm .
One spring slides over a series of studs , while the other slides over two brass sectors .
Half the studs are connected to the terminal by resistance coils in the manner shown in fig. 4 , the remaining studs being connected to the terminal .
The sectors , are connected to the minals ; the primary circuit , PC , joins the key at and .
The resistance coils which connect the studs have resistances of 40 and 200 ohms .
The two stops serve to limit the motion of the arm , so that in its extreme positions the sprin connected with presses upon the stud nearest to either or .
The width of the is sufficient to ensure that before the spring leaves one stud it is in contact with the next one .
When the arm is in either of its extreme positions the total resistance of the circuit is , where is the esistance of the battery and ampere-meter and is the resistance of the rest of the circuit .
When the spring connected with comes on to the next stud the resistance is , and one more step makes the resistance .
When the spring comes to the next stud the E.M.F. acting on the circuit is reversed .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
Since the spring connected with does not leaye any stud before it touches the next one , as the arm is moved from one side to the other , the primary circuit is never broken except possibly in the central position of the arm .
The circuit will be broken for an instant in the central position unless the spring connected with touches the two central studs at the same instant that the spring connected with touches both the sectors .
It is more or less a matter of chance whether this break of the circuit occurs , but if it does occur it is only after the current has been reduced to a small value by the introduction of the large resistance of 240 ohms .
Except for this uncertainty we may say that the current is reversed in several steps which ( except possibly those occurring in the uncertain part of the motion ) are not sudden because of the great self-induction of the choking coil ( S 33 ) .
( with a possible exception ) there is no sudden change in the current , the rate of variation of the current is no doubt much greater at some stages of the change than at others .
The key generally worked well ; it was , however , subject to slight uncertainties .
S 32 .
A fifth key was designed in 1900 .
Our aim was to ensure that the primary circuit should never be broken , and also that the resistances introduced into the circuit should be as small as possible , so that the rate of change of the current should , at every part of its variation , be as small as possible .
The battery is connected through the Weston ampere-meter , WA ( fig. 5 ) , with the terminals on an ebonite arm working about a pivot at O. The terminals are Fig. 5 .
colmected to two brass springs , , one at each end of the arm .
These springs slide over a series of sectors , out out of a brass ring .
The sectors and and VOL. CXCVIII.\mdash ; A. MESSRS .
G. F. C. AND T. G. BEDFORD the sectors and are conI ) ected by the resistan , while the sector is insulated .
The spaces between the sectors , about millims. wide , are filled with ebonite so that the springs pass smoothly over them .
The primary circuit PC joins the sectors .
The two stops , , s*arrest the motion of the arm when has got well on to either or , so as to be clear of or .
To avoid any possible break in passing from one sector to the next , the spring is made in two portions , , each of which presses upon the sectors .
The spring is arranged so as to touch the sectors only at its ends , The lengths of the sectors and of the springs are arranged so that the key performs the reversal in the following manner : the start is on , and is on ; both parts of reach before reaches .
While is on the primary current falls from towards the limit , where is the voltage of the battery , the resistance of the battery and ampere-meter , and the resistance of the rest of the circuit .
Before leaves reaches , and then there is no E.M.F. acting round the primary circuit except what arises from the very small internal esistance of the key , and then also the resistance offered to any current flowing in the primary circuit is reduced from to R. When is on one pole of the battery is insulated , and there is no E.M.F. at all acting on the primary circuit , whose resistance remains R. When reaches , and is still on a very small E. F. , due to the internal resistance of the key , acts in the reversed direction on the primary circuit .
The resistance offered to a current in the primary circuit is still R. While is on and before it reaches leaves ; the resistance of the circuit is now raised to , and the voltage is introduced .
When reaches the resistance is diminished to , the voltage remaining E. The resistances need not be large ; they are only used at all iu order to save the battery and the ampere-meter .
If the resistances were made very small , and if were small also , a large current would flow through the ampere-meter when is on on , and on , causing damage to the meter .
The resistances need only be great enough to prevent the ampere-meter from being damaged by too strong a current .
We have only had opportunity to use this key in a few experiments , but as far as we have tested it we are satisfied with its action .
The motion of the spot of light on the scale of the dynamometer is now quite free from the sudden jumps which it exhibited when the key of S31 was used .
The Choking Coil .
S33 .
The " " coil which was inserted in the primary circuit to introduce great self-induction , has a core built up of 149 armature rings an average thickness 0714 centim. , the inner and outer radii being and centims. respectively .
The mean circumference is thus centims. and the cross-section sq .
centims. It the core had been solid and not laminated , the eddy currents induced in it would *By oversight the wire from / to WA has ) drawm ON ASlJ ROF have rendered the coil ineffectivo in " " choking\ldquo ; any sudden variation of the current due to a sudden change of E.M.F. or of resistance .
The iron , which was supplied by Messrs. CROMPTON , was tested for magnetic quality with the following results:\mdash ; In magnetic quality , the iron is very nearly the same as the soft thin sheet iron ( Ring V. ) tested by Professor EWING and Miss KLAASSEN .
* The core was wound with three independent layers of cotton-covered copper wire as follows:\mdash ; Stout wire was used in order to avoid any considerable increase in the resistance of the primary circuit .
This resistance must be kept low if the choking action is to be efficient .
S 34 .
This choking coil is perhaps unnecessarily large .
It might be better to use a core smaller both in diameter and cross-section and to form each of the coils on it with wire of the saroe gauge as that used for a corresponding coil on the solenoid .
By this expedient , if we always employ corresponding coils on the choking coil and the solenoid , the magnetic force in the choking coil has the same value as in the solenoid .
Thus , if the iron used in the choking coil is of approximately the same quality as the specimen under test , the coil will most effectively choke the primary current approximately at the time when the choking is most needed , viz. , when the value of for the specimen has its greatest value , as was explained in S10 .
If the iron plates used in the core of the choking coil are of good quality it is very unlikely that any specimen will be so " " soft\ldquo ; in the magnetic sense as to have a value of so many times greater than the value of for the core as to lead to inaccurate measurements .
If the specimen is of " " hard\ldquo ; iron or steel the value of for the specimen is by comparison so small that it is of little consequence if the most efficient choking action takes place when for the specimen is not at its " " Magnetic Qualities of Iron ' Phil. Trans , vol. 184 1893 ) , p. 1003 .
maximum value .
It would be necessary to make the core smaller than in the coil used by us , so that in spite of the smaller section of the wire the resistance might still be small .
ngement of the Apparatus .
S 35 .
After this description of the apparatus we proceed to explain , by the help of fig. 6 , how it was arranged .
The battery SC , consisting of one or more small storage cells , was connected through the ad.justable resistance and the Weston amperemeter WA to the special reversing key .
An adjustable length of German-silver wire included in the circuit enabled us to keep the current to a definite value for any length of time .
On the key , the current passes round .
the fixed coils of the electro-dynamometer , round the magnetising solenoid MS and the choking coil , and finally round a compensating coil F. By adjusting the position of the small effect upon the ballistic galvanometer of the current passing through ED and MS was completely annulled .
The circuit of the ballistic galvanometer BG contains a resistance box and the earth-inductor , as well as the secondary coil and the key The circuit of the suspended coil contains the resistance box , the secondary coil V , the key , and the resistance coil X formed of copper wire adjusted to have the same resistance as the earth-inductor , as described in S23 .
To bring the suspended coil to a current of the order of 1/ 200,000 ampere was sent through the suspended coil by depressing the tapping key , a Leclanche cell providing the current .
In the actual experiments a system of shunts was instead of the single high resistance , which , for the sake of simplicity , is represented in the diagram ; the effect is , however , the same in both cases .
When ON THE OF we .
desired to damp the more violent motions of the coil , was diminished so that a current of 1/ 20,000 ampere could be sent through the suspended coil , the -key being opened to prevent any of this stronger current from passing round the secondary coil and affecting the magnetisation of the specimen .
The final damping was always done with closed and at its large value .
We verified by experiment that currents many times larger than those actually employed for this purpose had no appreciable effect upon the hysteresis of the specimen .
When the dynamometer is to be standardised the resistance coil X is taken out of the circuit and the earth-inductor is put in its place .
Practical of the Method .
S36 .
To illustrate the working of the method , we now give a set of observations made to determine the hysteresis loss , , as well as the mean maximum induction , for a definite range of magnetic force , for a specimen of soft iron : Area of cross-section ( circular ) . . . .
sq .
centim. Change of induction due to earth inductor Magnetic force per unit C.G.S. current Maximum current . . . . .
C.G.S. Maximum magnetic force . . . . .
Number of turns on secondary coil . . .
Throw of galvanometer due to earth inductor . . . . .
S centims. Mean throw on reversing primary current centims. Mean maximum induction Current when earth inductor is turned C.G.S. Throw of dynamometer due to earth inductor . . . . .
centims. Number of turns on secondary coil . . .
Resistance of secondary circuit . . . .
ohms .
Length of secondary coil . . . . .
centims. Mean area of one turn of secondary coil sq .
centim. Throw , to ( volts ) centims. Throw , from \mdash ; 15 ( volts ) centims. Hence ergs per cub. centim. per cycle .
In order to find , the observations for and were repeated with other voltages ( S 15 ) , with the result 8 12 16 volts .
From and 860 From and 840 Mean value of ergs per cub. centim. per cycle .
Hence ergs per cub. centim. per cyde .
In this case and are negligible , and thus , as in S 15 , Taking as ohm per centimetre cube , and putting , since the section is circular , we have .
But , and hence ergs per cub. centim. per cycle .
The value of is so small in rison with that it could not be determined satisfactorily by varying ( S14 ) .
A series of experiments was made in which was varied , but the small irregularities rendered the observations useless for the determination of so small a quantity as is in this case .
[ See S41 ] S37 .
A word should perhaps be said as to the accuracy aimed at in our experiments .
The throws were all recorded to millim. , the throws themselves varying from 25 centims. to 1 centim. The difference between and , seldom amounting to more than 1 per cent. , and usually much less , has generally been neglected .
The calculations were mostly effected by a 10-inch slide-rule , and the numbers recorded in the tables are the numbers read from the rule .
Tests of the the Method .
S38 .
We may now pass on from the theory of the method , and tho description of the apparatus used for applying it in actual measurements of hysteresis , to an account of the tests which we have made in order to determine if , in our experiments , the theoretical conditions are so nearly satisfied that the dynamometer yields accurate measurements of hysteresis .
S39 .
Test by Comparison with the Curve.\mdash ; We were without the guidance of the completed theory till 1900 , and thus till that date we did not know what measurements were required for the determination of the correction arising from the eddy currents and the finite conductivity of the secondary circuit .
Under these conditions the only way of testing the accuracy of the dynamometer method was to make a cyclic B\mdash ; H curve by the use of the ballistic galvanometer , to calculate from the area of the curve , and to compare this value with the value of ( S 13 ) found by the dynamometer for the same range of magnetic force .
We made this test on many ON THE MEASUREMENT OF MAGNETIC ESIS .
occasions with different specimens and with various keys .
We now give the results in the following table , where the third and fourth columns give respectively the value of found by the ballistic galvanometer and the value of found by the dynamometer ; in the last column the section of the specimen is given .
The numbers in this column , such as ( 3 ) , refer to the first column . .
It will be seen that there is very fair agreement between the values found for and , except in the cases of ( 1 ) and ( 2 ) , when there was no choking coil in the circuit , and only an ordinary mercury key was used .
Except in ( 12 ) and ( 14 ) the agreement is perhaps as close as could be expected , when we consider how much the behaviour of iron , specially under small magnetic forces , depends upon its previous history .
In making a cyclic B\mdash ; H curve , if we change from to , and find that changes from to , then is the value assumed by after has acted for a finite time\mdash ; about one-quarter of the period of vibration of the galvanometer needle .
But in the dynamometer method does not halt for any appreciable time at any intermediate value as it changes from to , and thus the value of corresponding to may somewhat from the value found by the galvanometer .
The experiments of SS45-49 below show that , for a given value of may vary considerably from causes depending upon the magnetic history of the iron even when its temperature and strain remain constant .
Thus , unless it is possible to give an accurate account of the history of the iron and of the manner in which the magnetic force changes from to , it is impossible to assign any definite meaning to and hence , in the absence of such an account , it is useless to attempt a very close parison between the value of found by the dynamometer and the value of found from the cyclic B\mdash ; H curve .
But the reement between the two values as recorded in the table will perhaps suffice to give general confidence in the method .
S41 .
Test Variation of the Resistcance of the Secondctry Circuit.\mdash ; The process of comparing the value of found by the dynamometer with the value of found from the area of the cyclic B\mdash ; H curve is laborious , and after all does not furnish an absolute test of the equality of for that particular law of variation of the magnetic force which is obeyed when the measurement is made , as was explained in S40 .
When , as was the case in most of our experiments , the specimen is a fine wire not exceeding 2 millims. in diameter there is a second method , described in S \mdash ; it may be termed , in contrast with the first , a self-contained method\mdash ; which is easy of application , enabling us to test the accuracy of the dynamometer measurements by the dynamometer itself If , in any case , we can find the effect of , we have a superior limit to the effect of X , for by ( 14 ) .
We now give the results of some applications of this second method .
It will generally suffice to take two values of , one double the other .
Since , when is varied , ( 1 ) is inversely proportional to , it follows that is proportional to , provided that and are corrected for damping ( S 14 ) .
Hence , in testing the accuracy of the dynamometer measurements , it is sufficient to compare the values of for the two values of S. If be doubled the consequent diminution in is , by S 14 , equal to the amount which must be subtracted from the product for the larger resistance in order to obtain the value of the product corresponding to We will first refer to the results obtained by one of us in November , 1895 , for an iron wire .
centim. in section , with and , the value of being 17780 when ohms , while was found from a cyclic B\mdash ; H curve to be 7409 .
A simple mercury rocking key was used , and the primary circuit had only the self-induction of the dynamometer and the solenoid .
As rose from to 584 ohms , fell from 17780 to 14550 , being closely represented by .
From this we find by ( 20 ) that .
when ohms , 3720 .
In these experiments centims. , sq .
centim. , and hence if and ( S11 ) be neglected , when ohms .
Thus , when is ohms .
Subtracting this from the correON THE MEASUREMENT OF MAGNETIC HySTERESIS .
sponding value of , viz. , 17780 , we obtain .
Thus , when the proper corrections are applied , the method yields approximately correct results even with an ordinary key and without a choking coil .
It was a little disheartening to find so great a discrepancy between and as the numbers 7409 and 17800 indicated , and we were glad to find five years later that the discrepancy is satisfactorily accounted for by the more complete theory .
Using the key of S31 as well as the choking coil , we found ( August 8 , 1899 ) for a soft iron wire 0324 sq .
ceutim .
in section , with , after correction for damping and for finite arcs , ohms , centims. , ohms , centims. , For the secondary coil ; also , and thus , when ohms , .
But the effect of , the correction arising from the conductivity of the secondary circuit , in causing a change in the product , is quite insensible , and hence the eddy current effect , X , is negligible .
In this case , as closely as we could measure , With the same key and choking coil we found ( August 7 , 1900 ) for a bundle of ten pianoforte-steel wires , of total section 0255 sq .
centim. , with ohms , centims. , ohms , centims. , Here , so that when ohms .
Here we have divided by 10 , since the specimen is formed of ten wires ( S9 ) .
In this case X and are both negligible .
To illustrate the method of finding the correction by doubling the resistance , we take a test made on a bundle of ten iron wires coated with shellac varnish to prevent eddy currents from wire to wire .
The total section was sq .
centim. With the key of S 31 , but with an inefficient choking coil , we found ( August 9 , 1900 ) , when ohms , centims. , ohms , centims. , The difference between the products is 12 , and thus the correction to be subtracte from 391 is 24 .
We had centims. , so that since there are ten wires , when ohms , .
But we make only a small error in taking , and thus the total correction to be subtracted from 391 is or 28 .
Hence the value of to be used in finding is 36,3 .
VOL. CXCVIIL \mdash ; A. It was only in the later experiments that we were guided by the complete theory to test the accuracy of the measurements by varying , but from the tests just described we may conclude that the value of , obtained from the dynamometer throw , was in all our experiments very nearly equal to W. Energy dissipated by Eddy Currents .
S42 .
In our experiments the ratio was small , so that , bsing small in comparison with was small also , and thus and were nearly equal .
We now consider the case in which the section of the specimen is large , so that has to be made large and small , in order to reduce the sensitiveness of the apparatus .
Under these conditions is large , while is now so small in comparison with that it eludes observation , and thus cannot be determined by varying , the resistance of the secondary circuit .
These conditions have prevailed in the experiments made by Mr. R. L. WILLS with some of our apparatus ; in these experiments the eddy current loss was so large that systematic measurements were made to determine X for eyery value of employed .
Mr. WILLS used the key described in S 32 , and found that for a given total resistance of the primary circuit the " " throw\ldquo ; of the dynamometer is practically independent of the resistances denoted by in the description of the key .
This key divides the primary circuit into two portions ; the resistance of the portion which includes the battery is denoted by , and the resistance of the other portion by R. When the voltage was to be changed in order to make the observations suggested in S 15 , Mr. WILLS changed the total resistance , , of the primary circuit , mainly by changing R. The resistance was small compared with , and was used as a means of obtaining an exact adjustment of the current to definite values .
No attempt was made to make bear any fixed ratio to R. The reversal of the current does not take place quite in the manner described in S 15 .
In addition to the effect of the resistances , there is the further point of difference that although is adjusted to ' be accurately proportional to , the voltage driving the current , yet is not accurately proportional to because does not bear any fixed ratio to R. The value of and , consequently , the power absorbed by the eddy currents while the primary current is sinking to zero is thus not quite proportional to for a given value of C. During the subsequent rise of the current the whole resistance comes into play , and , the resistances being comparatively small , the value of for a given value of is nearly proportional to E. Now the increment of current contributes to X a quantity proportional to .
Thus although , for a given valu of is much greater during the fall than during the rise of the current , because for the iron of the chokin coil is much smaller during the first than the second stage , yet , ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
for specimens of high permeability , is enormously great , er during the second than during the first stage .
Thus , for a given value of , the product during the first stage is small compared with its value during the second stage .
Hence by far the greater part of the eddy current loss occurs during the rise of the current .
In fact , if the choking coil have a core of the same iron as the specimen and be similarly wound , and if the current be reversed in the manner described in S15 , the eddy current losses during the fall and subsequent rise of the current are proportional to area and area respectively , the areas being shown in fig. 1 .
Thus the main part of the eddy current loss occurs during the rise of the current ; this part is very nearly proportional to E. The eddy current loss during the fall of the current is indeed only roughly proportional to but it is small in comparison with the loss during the rise of the current .
Hence the total eddy current loss during a semi-cycle is nearly proportional to , and thus can be deternuned approximately by the formula , as explained in S 15 , since is now negligible , and X is thus sensibly equal to As illustrations we now give two examples kindly furnished us by Mr. R. L. WILLS .
In the first example the specimen was a portion , centims. in length , of a circular tube of radii and centims. plane through the axis forms a rectangular section of the ring , the sides of the rectangle being centim. Thus sq centims. , , while the mean circumference is centims. Hence , treating the ring as a straight rod , we find by Appendix I. , .
Now the resistance , , of the secondary circuit varied from 23 to 523 ohms , while , the number of turns , was 50 , so that varied from 356 to 8300 .
Hence was negligible in comparison with X , was thus sensibly equal to For each value of three determinations of were made with batteries of 8 , 16 , and 24 volts .
The value of was deduced from the three values of by the formulae which follow from S 15 .
In the last column we give the value of X , the space-average of the eddy current loss corresponding to volts , X as as being expressed in ergs per cub. centim. per cycle .
The agreement between and is satisfactory ; it may be taken as evidence that the theory sketched in S15 is practically applicable to Mr. WILLS'S experiments .
The ratio ranged from 1040 to 66,000 , and thus no appreciable error was introduced by neglecting the correction due to the conductivity of the secondary circuit .
In the second example the specimen was a ring similar to the last , the numbers now being centims. , sq .
centims. , .
The material was an alloy of iron and aluminium , and the experiments were made when the specimen was at C. We do not know ; if it was , then varisd from 430 to 1380 .
The results are shown in the following table , where it will be seen that there is again good agreement between and The agreement is the more significant because the eddy current loss , X , now forms a very considerable part of the whole energy dissipated in each cycle .
Complete Cycles and Semi-cydes .
S43 .
The theoretical investigation of S8 shows that the throw due to a complete cycle should be equal to the sum of the throws due to a pair of semicycles .
We have not spent any considerable time in testing this point , for the arrangements used by us have not been well adapted for that purpose .
In order that shall be sensibly equal to , it is necessary , when the specimen is a thin wire of soft iron , that the self-induction of the choking coil should be very large , and in this case , after the current has been reversed by the key , it does not at once rise to its full strength .
Thus , if the key be worked very rapidly so as to make a pair of semi-cycles in quick succession , the current may never rise to its full value in the middle of the cycle .
On the other hand , if a definite halt be made after the current has been reversed for the first semi-cycle , the dynamometer coil will have moved considerably from its zero position when it receives the impulse due to the second semi-cycle .
It is perhaps for these reasons that we have not found good agreement between the throw due to a complete cycle and the suln of the throws due to a pair ON THE MEASUREMENT OF MAGNETIC HySTERESIS .
of semi-cycles .
We have not taken any special steps to secure this agreement because ( 1 ) we are satisfied that the sum of the throws for a pair of semi-cycles giyes a correct measure of the hysteresis , and ( 2 ) it was more convenient to take a pair of semi-cycles than a complete cycle , since with the semi-cycles we could measure and simultaneously .
We found that the throw for a complete cycle depended a good deal upon the way in which the key was manipulated .
On the other hand , the throws for each of a pair of semi-cycles were generally very regular .
S44 .
In S6 we show that , when , on account of the previous magnetic history of the specimen , the two throws of the dynamometer for a pair of semi-cycles are unequal , the sum of the throws still furnishes a correct measure of the energy dissipated in the complete cycle , if .
This inequality in the two throws gave us much trouble .
It persisted in some cases in spite of very many reversals of the magnetic force .
As tested by the ballistic galvanometer , the iron had reached a steady state , but the galvanometer only shows the change of induction on reversing , and not the actual values of corresponding to or to .
In the belief that after many reversals the cyclic B\mdash ; H curve would lie symmetrically about the axes of and , we naturally did not look to the iron for the cause of the inequality in the throws .
We spent some weeks in making changes in the dynamometer and other parts of the apparatus , but of course without result .
At times the inequality would almost disappear ( probably on account of a large number of reversals ) , only to reappear for some apparently slight cause , such as changing the range of the magnetic force , or using another specimen , and this uncertainty made the matter very perplexing .
But though we felt that we had failed to discover the of the inequality , yet the agreement between the value of , found by the dynamometer , with that of deduced from the cyclic B\mdash ; H curve , made us certain that the method was giving at least approximate results , and we began to make experiments on the effects of tension and torsion upon hysteresis .
During these experiments we found that the inequality varied as the stress was applied , and we were thus led to see that the origin of the inequality lay in the iron .
When once we suspected the cause , it was easy to make experiments to satisfy ourselves that our suspicion was true .
We were then able to employ the method without hesitation .
S 45 .
To study the inequality in the two throws for a pair of semi-cycles in a definite manner we made the following experiments : freshly annealed iron wire , which had not been magnetised since the annealing , was placed in the solenoid when no current was flowing , and the magnetic force was then applied for the first time .
We found that the throw of the dynamometer for the first reversal from to was greater than for any subsequent reversal in either direction , and that up to 100 reversals the throw is greater when the magnetic force changes from to than when it changes from to .
In these experiments the value of was and the value of , after 100 reversals , 9990 ; the section of the wire was sq .
centim. An ordinary mercury rocking key was used , and the choking coil was inserted in the circuit .
We now give the results of one of these experiments .
The numbers heading the columns indicate the cycle to which the pair of semi-cycles in any column belong .
In this example the inequality is greater and more persistent than in most cases .
The inequality is generally much more marked for small than for large magnetic forces .
It will be noticed that the hysteresis diminishes with continued reversals .
* Efect of Continued Reversals .
S46 .
In the course of the experiments undertaken in the hope of finding the cause of the inequality of the two throws of the dynamometer , and , for the two semicycles belonging to a single cycle , we had occasion to put the specimen through many cycles .
We then discovered that the hysteresis dimimishes very considerably with continued reversals of the magnetic force .
To investigate this matter more completely a systematic set of experiments was made in 1899 with the object of determining how the effect depends upon the limits of the magnetic force .
The experiments were carried out in the following manner : specimen , demagnetised by annealing or " " by reversals was placed in the solenoid when no current was flowing .
The magnetic force was first applied in the positive direction ; the magnetic force was then reversed from to and the throws of the dynamometer and the galvanometer were read simultaneously .
The next reversal to was observed in like manner .
These two reversals constituted the first cycle .
The observations were repeated for other cycles as shown in the tables , where in each case the first column shows the number of the cycle .
The two throws of the dynamometer , and , sometimes showed an inequality which was rather persistent , though it was never greater than that recorded in S 45 .
We have therefore thought it sufficient to give the value of the hysteresis deduced from each complete cycle .
The mean maximum magnetic induction , , which diminished in much the same manner as the hysteresis , is also recorded in the tables .
The initial asymmetry of the hysteresis loop and its gradual shrinking with continued reversals are well shown in fig. 154 in Professor EWING'S Magnetic Induction .
3rd Edition .
Much labour would have been saved had we realised the significance of Professor EWING'S curves .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
The experiments of Professor J. A. EWING* on soft iron show that after the few dozens of reversals which are necessary to avoid the effects of the initial diminution of , the values of and of remain constant , even after 70,000,000 of cycles of reversal when the temperature of the iron is not allowed to rise above that of the atmosphere .
In his experiments the maximum magnetic force in these cycles was about 4 , and the maximum induction , , about 8000 .
It is now well known that if iron be raised to a temperature only C. above the atmospheric temperature , and be maintained in that state for many hours , a large increase in the hysteresis often results .
S 47 .
Soft Iron demagnetised by Annealing.\mdash ; The specimen was a wire of soft iron sq .
centim. in section .
In each case it was annealed before the observations for any given value .
of were made ; the preyious magnetic history of the iron was thus wiped out .
The key of S 31 was used .
It will be seen that the effect of continued reyersals of in producing a diminution of and rapidly decreases as increases , both and becoming sensibly constant when reaches the value .
The following table illustrates this fact , and shows that the percentage change of is always greater than the percentage change of .
The change recorded is that which occurred in the first 41 cycles:\mdash ; The Electrician , ' vol. 34 , nuary 1 , 1895 .
Percentage change in . . . . . . . . .
Percentage change in 27 5 S48 , Soft Iron demagnetised " " by \ldquo ; \mdash ; The same piece of wire and the same key were used as in S47 .
The wire was very completely demagnetised by reversals before the tests for and were made ; a mirror magnetometer served as indicator .
The results are shown in the table below:\mdash ; For and the values of and are constant within the errors of observation .
For there is a considerable diminution in both and during the first 41 cycles , but much less than in the case of the wire demagnetised by annealing , the changes in the first 41 cycles being now for per cent. , and for per cent. The wire had been annealed before the tests for in S 47 were made .
Without further annealing , the tests of the present section for and were made in the order written .
This treatment has had the result of dimimishing the initial value of and raising its value after 41 reversals ; the value of is at the same time considerably increased .
S 49 .
Steel Rod.\mdash ; A few experiments were made upon a steel rod sq .
centim. in section , using the key of S 31 .
We endeavoured to demagnetise it by reversals before the tests for and were made , but we were only partially successful .
The dynamometer was not sensitive enough to enable us to go to small magnetising forces .
The induction remained practically constant for both values of ; in 41 cycles the hysteresis fell by 10 per cent. when and by per cent. when * See EWING , 'Magnetic Induction .
3rd Edition , fig. 165 .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
, the changes in being roughly the same as those for the iron wire , demagnetised by annealing , under approximately the same magnetic forces .
Curves for Zero Stress .
S 50 .
The energy dissipated through hysteresis can be varied in many ways .
We describe , in the sequel , many experiments in which the hysteresis loss for constant values of was caused to vary by varying the stress applied to the specimen ; Mr. has varied the hysteresis loss for constant values of by varying the temperature of the specimen .
But perhaps the most natural , and certainly the most usual , way of varying is to vary , while the stress is kept at a constant zero value .
The curve representing as a function of under these conditions is the curve which is useful to engineers when designing transformers .
To distinguish it from the curves , which represent as a function of , when is kept constant and is varied by varying the stress , we call it a curve for zero stress .
It is the curve for which Mr. C. P. STEINMETZ has proposed the formula Effects of Stress .
S 51 .
We now pass on to describe a number of experiments in which we studied the effects of tension and torsion upon the mean maximum magnetic induction , and upon the energy dissipated by hysteresis when the magnetic force ranged between definite limits .
In each case the same series of stresses was applied to the -specimen for each value taken for Effect of Tension on Soft Iron Wire .
S 52 .
The first systematic experiments on the effect of stress upon the energy dissipated in hysteresis were made in 1898 upon a soft iron wire .
VOL. CXCVIII .
The wire was placed horizontally , perpendicular to the magnetic meridian , and the tension was applied by a flexible silk cord which passed over a pulley and supported a weight .
The reversing key described in S30 was used , the current being measured by a shunted d'Arsonval galvanometer .
We satisfied ourselves by the comparison of the values of found ( 1 ) from cyclic B\mdash ; H curves , ( 2 ) by the electro-dynamometer , that the method was yielding at least approximately exact values of W. Before the observations corresponding to any given value of were made , the wire was subjected to several cycles of loading and unloading , the maximum load being 24 kilogrammes .
The magnetic observations were taken only as the load was being increased .
The section of the wire was 00708 sq .
centim. , so that a load of 1 kilogramme gives a tension of ynes per sq .
centim. The results are given in the following table and in fig. 7 .
( See also S67 .
) Fig. 7 .
Observations on a soft iron wire during both loading and unloading showed that the curves for loading and unloading are not quite identical , though the difference between them is not large .
ON THE MEASUREMENT OF MAGNETIC HySTERESIS .
In order to save space on the diagram , the zero for the four upper curves differs from that for the two lower ones .
The numbers on the diagram will prevent any confusion .
In each case , as the tension increases , both and rise to maximum values , the tension corresponding to the maximum values diminishing as increases .
Next to the similarity between the curves for and for a given value of the most striking feature is the great increase in both and occasioned by tension when is small .
Thus for a pull of 14 kilogrammes raises from 1233 to 5920 , and from 494 to 3770 .
The ma,.netic force was not carried to values high enough to obtain the Villari reversal of the effect of tension .
Effect of Torsion within the Limit .
S53 .
In July and August , 1899 , we made a long series of experiments on the effect of torsion upon the magnetic qualities of iron and steel .
The arrangements for applying torsion were very simple .
A wooden wheel centims. in diameter was mounted on a brass tube as an axle , and this tube revolved in a bearing .
The wire under test passed through the solenoid and through this tube , and was clamped at one end to the wooden wheel , while the other end was held in a vice .
The vice the bearing of the wheel were mounted on a stiff wooden beam so that the wire was parallel to the beam , which was placed at right angles to the magnetic meridian .
The magnetising solenoid was so fixed that the wire passed along its axis .
To apply a torsional couple to the wire , a weight was hung from the edge of the wheel by a flexible string wrapped round the wheel .
When we wished to give the wire a definite twist , the wheel was clamped in the desired position .
Experiments on Stee , S54 .
Curve for Zero Stress.\mdash ; The first set of experiments was made upon a steel rod sq .
centim. in section .
To find how and depend upon , when there is no torsion , we made a series of observations for and , using the key of S81 , and varying from to ; the results are given in the table below .
The electro-dynamometer was not sensitive enough to allow us to work with magnetic forces less than 5 units .
These observations will serve as an example of the application of the method described in this paper to determine the manner in which depends upon , when is made to vary by changing , the stress being constant .
The curve representing these observations is shown in fig. 8 .
The numbers placed along the curve show the values of for which the values of and were found .
The curves ( 1 ) to ( 4 ) are considered in S66 .
Fig. 8 .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
S55 .
Effect of Torsion , within the Elastic was now subjected to torsion in the way 2000 grammes was hung from the rim of the by steps of 100 or 200 grammes , care bein with as little jerking as possible .
For each cycle of loading nd unloading the maximum magnetic force was kept at a constant value .
In every case , except possibly that of , the wire was put through several cycles of twist in which the load varied between the limits grammes , and then the magnetising current was put through 20 cycles of reversal before the obseryations for and were made .
The positive and negative signs indicate that the weights were hung on the right-and left-hand sides of the wheel respectively .
In the actual experiments on the effect of torsion on and the torsion was always positive , the load being initially zero , then increasing to 2000 grammes , and then decreasing to zero .
In the ideal case of pelfect symmetry about the state of no torsion , the part of the curve for negative torsions would be exactly similar to the part of the curve for positive torsions .
The results are given in the following table , and are exhibited in fig. * 1000 grammes .
600 grammes .
From to the curves for unloading agree , within the errors of experiment , with the curves for loading , and only a single line is shown in the diagram .
In the cases of and the curves for differ from those for loading .
The curve for could not be conveniently shown on the diagram .
In these two cases there is also an evident want of symmetry , for , if the symmetry were perfect , the lines for loading and unloading would cross each other on the line of no load .
It is interesting to notice how closely the curves for imitate those for .
We consider this similarity in S 66 below .
Experiments on Soft Iron .
S 56 .
A much more extended series of experiments was made upon soft iron wire , 0324 sq .
centim in cross-section .
The wire was " " galvanised\ldquo ; when bought , but by heating it to bright redness in a large blowpipe flame all the zinc was burnt off the wire and at the same time the wire was annealed .
The wire was twisted by means of the arrangement described in S 53 .
The reversing key described in S 31 was used in these experiments .
The first step was to test by the method of S41 whether X and could be neglected .
The numbers for this specimen , recorded in section ( b ) of S 41 , show that X and were negligible .
S 57 .
Curve for Zero Stress.\mdash ; To gain a general idea of the magnetic character of the wire , a set of observations was made to determine and for a series of values of , the wire having been annealed and being free from strain .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
The magnetic force was put through about 20 cycles of reversal before the tests for and were made .
The results are given in the following table , and the curve representing them is shown in fig. 10 ; the curves ( 1 ) to ( 5 ) are considered in S65 .
Fig. 10 .
Effect.of Torsion within the Elastic Limit .
S 58 .
The wire was found capable of sustaining 300 grammes hung from the edge of the wheel without acquiring permanent set .
We therefore decided to subject it to cycles of torsion in which the load varied within the limits grammes .
The changes in the load were made by steps of 50 grammes .
Before the cycle of loading and unloading , during which the magnetic observations were made , the wire was subjected to several cycles of positive and negative torsion , the load having the limits grammes .
In every case , before the observations for and were made for any particular value of the torsion , the magnetic force was put through about 20 cycles of reversal ; in this way we avoided the major part of the indeterminateness due to the diminution of and which occurs ] with continued reversals of .
( S46 .
) The results of these experiments are given in the table below and are exhibited in fig. 11:\mdash ; Loa4 Fig. 11 .
suoo u S1 S1HI s. op aoafqns } pug } } uauI S Xug ; op SAtOJ m S fqns u SU01$ o$ sn ?
nIIL U01SJ0 } } qPIHIIqtE eu1I eu1I } asoq } S1 ql onpuI uInuIIXBUI U ( Itoq uoIS .
} I8 llJiSXH ffII NO asuring hysteresis , for the obseryations for the first four values of the magnetic force , involving fifty-two determinations of hysteresis , were easily taken in one day , though much time was spent in changing load , in putting the current through 20 reversals before the magnetic tests were made , and in bringing the suspended coil of the dynamometer to rest .
of beyond the Elastic Limit .
S 59 .
So far our experiments were made for torsional strains in which the elastic limit was not overpassed at all , or in any case was not exceeded to such an extent that the wire took a noticeable permanent set .
We thought it would be interesting to trace the effect of great torsional overstrain upon the induction and the hysteresis for a constant maximum magnetic force .
In 1899 we made experiments in this direction upon two specimens of the same soft iron wire as was used in the experiments of SS57 , 58 .
Both specimens were heated in a large blowpipe flame to burn off the zinc coating and to anneal the wire .
The arrangement of S 53 was used for applying torsion .
S60 .
periments on Soft Iron Wire ( 1 ) ; 5.\mdash ; In these experiments the wheel was turned , always in one direction , through the desired angle , measured from its position when the wire was unstrained , and was then clamped .
A mechanical " " counter\ldquo ; served to record the number of revolutions made by the wheel .
When the wheel had been clamped in any position , the magnetic force was put through 20 cycles of reversal , and then readings for and were made .
The wheel was now turned still further and again clamped , and then the magnetic observations were repeated .
This process was continued till the wire broke .
The fracture occurred when the wheel had been turned through 104 revolutions .
Since the length of the wire was centims. , this angle corresponds to a twist of turns per centim. The observations were too numerous to be recorded in a table ; we therefore only give a diagram in fig. 12 , distinguishing the curves for this specimen by the mark ( 1 ) .
In these experiments the first result of the torsion was to cause a very rapid inution of both and W. The curve for falls continuously till the wire breaks , but the curve for shows a well-marked rise and fall , with a maximum at about revolutions .
The values of and were determined immediately after the wire broke , the tests showing that , when the stress was relieved by the fracture , both and fell to about half the values they had just before the fracture .
Thus fell from 2950 to 1510 , while fell from 2410 to 1044 .
We found that when the wire had been twisted dimimished considerably with continued reversals , even after 20 cycles of magnetisation .
S61 .
Experiments on Soft Iron Wire ( 2 ) ; .\mdash ; The second specimen ( 2 ) was treated rather differeutly , The el was turned through a definite number of ON THE MEASUREMENT OF MAGNETIC HySTERESIS .
revolutions , measured from its position when the wire initially free from torsion the number being read on the " " counter The wheel was then clamped in this position , and , after 20 cycles of magnetisation , observations were made for and W. The wheel was now unclamped , and the wire was allowed to untwist so as to rid Fig. 12 .
itself of torsional stress ; after twenty more cycles of magnetisation , the new values of and were then determined for this condition of the wire .
The wheel was then turned still further , and fresh were made , the process being repeated until the wire broke .
The results are shown in fig. 12 .
The curves for and when the wire was under torsional siress are marked , and those for zero stress are marked .
For the sake of clearness , the origin for these four curves is at the side of the diagram .
The wire in untwisting turned the wheel back through an angle upori the twist from which it was to rid itself\mdash ; an angle which increased with that twist .
After a twist of one revolution the wire turned the wheel back through , retainin a " " permanent set\ldquo ; of , while , after a twist of 100 reyolutions , the wheel was turned back through .
In the curves the points for and hich placed on the ordinate to any number , , of complete volutions , indicate the values found for and when the wire was allowed to free itself from stress after the wheel had been turned through revolutions .
The curves show that , as in the case of specimen ] , ] application of torsion causes a continued decrease of both and W. The decrease is at first very rapid , falling from 6520 to 4750 and from 4380 to 3180 for the first half revolution of the wheel .
The curves show that after the first half revolution of the wheel both and diminish contin , but at first much less rapidly than is the case for the curves .
The initial parts of the curves are an exception to the rest , for when the wire was allowed to untwist after the wheel had been turned through the first half revolution both and were greater than before the wire was twisted at all , rising from 6520 to 6720 and from 4380 to 4590 .
Up to about revolutions of the wheel , the relief of torsional stress was followed by an increase of both and , but beyond this poin the relief caused a decrease of both and W. The wire broke at 101 reyolutions of the wheel , , at practically the same twist as wire ( 1 ) ; the length was very nearly the same , cerltims .
Just before fracture the values of and under the torsion due to 100 reyolutions were , and the wire had freed itself from stress , .
After fxacture the values fell to , the changes being much smaller than ) for specimen ( 1 ) .
It seems probable that the diminution of and which occurs on fracture is due to the violent jar which attends the fracture .
The values of both and in and are for large twists very much smaller than the corresponding values in ( 1 ) .
It is true that the permeability of ( 2 ) before it was strained , viz. , for , is much less tluan the value for ( 1 ) under the same nditions , But the difference between ( 1 ) and ( 2 ) in to and for large twists can hardly be due to this fact .
It is more likely due to the untwisting of the wire at the various staoeso of the process .
When the wire been twisted we found just as for ( 1 ) that diminishes considerably even after 20 cycles of magnetisation .
We took 20 cycles in each case before making the tests for and , but if is still diminishing with continued reyersals , it is hardly possible to assign any very definite value to it .
The wire , initially quite , was very stiff after it had been broken by twisting .
Its ngth originally was centims. , and this was increased by the torsion to centims. The mean diameter from 0796 to 0793 inch .
Thus the volume of the wire increased by about 1 per oent .
The smallest values reached by and , viz. , those found after fracture , are very small compared with those found before the wire was strained .
Initially , and finally THE OF NETIC H Influence of manent Set the flects of Cycles of S62 .
The experiments of S61 made on specimen showed that up to twists giving a permanent set of about 13 revolutions in a length of centims. both and become greater when the torsional stress is relieved , and that beyond the limit of 13 revolutions the reverse is the case .
This observation led us to inquire into the forms taken by the two curves connecting and with a cyclical torsional couple applied to a piece of soft iron wire , which had been previously twisted so as to have a considerable permanent set .
The actual specimen was that used in S58 without any further annealing .
In carrying out these experiments , the wheel was turned through a definite number of revolutions , measured from its position before the wire was strained , thereby giving the wire a permanent set .
The wheel was then set free so as to allow the wire to untwist itself as far as possible , and then about 10 cycles of twisting were given to the wire , the load producing the torsional couple varying between the limits of grammes .
After these cycles of twisting , a set of magnetic observations was taken as the next cycle of twisting was gone through .
In every case the magnetic force was put through 20 cycles of reversal before the magnetic observations were made .
One effect of the cycles of twisting was to diminish the " " permanent set\ldquo ; of the wire .
Thus , reckoning from the position of the wheel when the wire was set free after a twist of 50 revolutions , 10 cyoles with the load between the limits of grammes caused the wheel to turn back through revolution .
It will be noticed that the curves in figs. 18 , 14 do not always form closed figures .
This is probably in the main due to the fact that the preliminary cycles of twisting were gone through in a few minutes , while the one during which the tests for and were made occupied about an hour .
We have often obseryed a similar effect when a copper wire is put through cycles of loading and unloading .
There is in this case so much " " creeping\ldquo ; when a load is applied , that , if a cycle of loading and ullloading with a given range of load is performed slowly after a number of cycles of loading and unloading performed comparatively quickly with the same range of load , the curve connecting the elongation and the load is not closed .
The curves exhibit in a highly developed form the want of symmetry which was rudimentary in the experiments of S58 ( fig. 11 ) .
In every oase the maximum magnetic force was kept at the constant value , so that the experiments might be comparable with those of S61 .
The length of the wire , oentims .
, was about the same as in the experiments of S61 .
The results are given in the following table and are exhibited in figs. 13 , 14 .
The sign prefixed to a load in the table indicates that the couple due to it has the same sign as the couple which gave the wire its permanent set .
In order to show all the curves on a single diagram have taken a different position of the zero line for , and in each curve .
The numbers marked near each curve will enable the values of or to be read off for any point on the curves .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Fig. 13 .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes .
Load , mmes . .
Fig. 14 .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
The curves tend to confirm the result of S61 , viz. , that for small amounts of permanent set both and increase when the couple is removed .
The sixth curve shows that when the permanent set is large the removal of the couple causes a decrease of both and W. In these experiments the change from increase to decrease occurs somewhere between 25 and 50 revolutions , as against about 13 in the experiments of S61 .
The wire of S 61 was not , however , subjected to cycles of twisting .
The curves for zero permanent set correspond to the curves for found in S58 and shown in fig. 11 .
But instead of cnrves for ondy half a cycle of twisting we have now curves for a complete cycle .
The mean values of and for zero couples are very nearly the same in the two cases , but the couple due to a load of 300 grammes produces a decrease of and rather greater than was found in the experiment of S58 for .
The specimen had , however , been put through many cycles of twisting in the interval .
The observations could not be extended beyond 50 turns of permanent set , for when the attempt was made the wire broke at 54 .
The similar wires used in the experiments of SS60 , 61 did not break till 104 and 101 revolutions respectively .
It is possible that the difference , if not due to a flaw , is due to the effect of the cycles of positive and negative couples .
The curves of figs. 13 , 14 show in a very striking manner the close correspondence between and when is kept constant and is made to change by varying the stress .
of a Cydic State after Initial Permanent Set .
S 63 .
In the last series of experiments made in 1899 we studied the manner in which a wire , after torsional overstrain , gradually settles down to a definite cyclic state as the applied couple goes through a series of cycles .
The specimen was a piece of the same annealed soft iron wire as that used in SS 58 , 60 , 61 , and had the same length as the wire used in S 60 , centims. In these experiments the wire was strained far beyond the elastic limit by turning the torsion wheel through a definite number of \mdash ; in our ents 6 , 20 , and 50\mdash ; from the position for zero strain .
The wheel being now clamped , a weight of 300 grammes was hung from the right edge of the wheel , producin a couple having the same sense as that which strained the wire .
The wheel was then unclamped , and the wire , being allowed to untwist , turned the wheel back until the couple due to the elasticity of the wire balanced the couple due to the 300 grammes .
After 20 cycles of magnetisation , varying between the limits and were determined .
The load was then reduced to 200 grammes , and the magnetic tests were repeated .
When by two more steps of 100 grammes the load had been reduced to zero , loads of 100 , 200 , and 300 grammes were hung in succession from the left edge of the wheel to produce negative couples .
The process of changing the load from to grammes and back again was continued till it appeared that the curves connecting and with the load had nearly settled down to a permament form .
In the first experiments , when the permanent set was about 6 reyolutions , the first application of a couple due to 200 grammes caused the wire to yield somewhat , and when the load of \mdash ; 300 grammes was reached , the yielding was great enough to cause the wheel to turn gradually through half a reyolution .
In the next cycle of loading this effect of a negative couple was small , and it became much smaller for each successive cycle .
With the large permanent sets of 20 and 50 revolutions the wire became so hard that the yielding under a negative couple was very small eyen in the first cycle .
The positive couples caused little yielding even in the first cycle with the permanent set of 6 revolutions .
The observations were too numerous to be recorded , except in the form of the curyes in figs. 15 , 16 .
In ordel to show all the curves on a single diagram , we have taken a different position of the zero line for and in each curve .
oduclng 1 Fig. 15 .
Fig. 16 .
In these experiments it was found that after the first application of a negative couple due to 800 grammes the iron had roughly reached a cyclic state , and thus , to avoid confusion , we give only cycles as continuous cllrves .
For a permanent set of 6 turns were actually gone through , and in each of the other cases ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
23 cycles .
In each case the last cycle is indicated by isolated points ; when ths positive load is increasing a , and when the positive load is decreasing is used .
In their general forms the curves due to the last cycle for 20 and 50 turns of permanent set resemble those obtained after several comparatively rapid cycles of torsion for 25 and 50 turns of permanent set in S 62 .
Any close comparison is impossible on account of the fact that the curves of S 62 do not form closed figures .
In the case of 6 turns of permanent set , the curves for the last cycle differ widely from those for 5 turns of permanent set in the experiments of S 62 .
The part of the curve which corresponds to a decreasing positive couple now lies always above that part which corresponds to an increasing positive couple instead of crossing it .
The two parts of the curve show , however , a tendency to approach each other with repetition of the cycles of torsion .
In this case there is a very great recovery of magnetic quality in the first cycles of torsion .
When the load was reduced to zero for the first time was 5190 , at the second zero load we found , and at the third zero load 8050 .
Since , the permeability rose from 1038 to 1610 .
Relation connecting noith S 64 .
A cursory examination of the curves recording the effect of strain upon the hysteresis and the induction , for given values of the magnetic force , is sufficient to make it evident that the changes in , due to strain , are very closely followed by the consequent changes in W. This correspondence is so close that it invites the attempt to express it mathematically .
We have here the means of analysing and determining it as a function of and ; this is impossible with the curve for zero stress , since we cannot , without strainmg the specimen , change without changing .
Our plan has been to plot curves showing in terms of for definite values of , the variations of and being due to strain .
S 65 .
The first example is taken from the experiments on the effect of torsion upon a soft iron wire , the corresponding values of and for definite values of being recorded in S 58 .
Five curves connecting and were plotted for five values of upon fig. 10 , where the curve for zero stress is also shown .
The points corresponding to increasing torsion lie so closely on the same curves as those corresponding to diminishing torsion , that we have marked both sets in the same manner .
For the larger values of each curve consists of a straight lin with a small hook at one end\mdash ; the end corresponding to zero strain .
The straight parts of the curves all pass , on prolongation , through the point on the axis of , and hence over the greater part of the variation of , caused by torsion under a constant maximum magnetic force , the value of is given by VOL , CXCVIII .
where is a function of .
To determine in terms of we plotted against , and found that is nearly proportional to as the following table shows:\mdash ; Hence , with fair accuracy , the formula represents as a function of and , over a large range , provided that the load producing torsion exceeds 100 grammes .
The hooked part of each curve corresponds to small torsions due to loads of 100 grammes and under .
The value of for zero strain , recorded in S57 , is given with considerable accuracy by over a considerable region in the neighbourhood of the maximum permeability .
S66 .
The second example is furnished by the experiments on the torsion of a steel rod , described in S55 .
The curves connecting and are shown in fig. 8 along with the curve for zero stress .
We again find that for the larger values of those portion of the curves which correspond to the larger stresses are straight lines radiating from a single point\mdash ; in this case the origin\mdash ; and thus obtain .
The values of and are given in the table .
Thus , approximately , This ion would naturally fail to represent facts when The value of for zero stress , recorded in S54 , is given closely over the whole range by S67 .
We now take the experiments on the effect of tension upon a soft iron wire described in S52 .
The curves connecting with are shown in fig. 17 along with the curve for zero stress .
Each of the curves , which show the effects of stress , is again made up of a straight part and a hook , the straight parts radiating from the point , For the slop - of the lines we hay ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
so that the straight parts are given by The curves just described differ in two important particulars from the curves for torsion in fig. 10 .
In the case of torsion the straight parts of the curves correspond to large stresses and the hooks to small stresses , but in the case of tension the straight parts correspond to small stresses and the hooks to large stresses ; the hook is moreover much more developed in the curves for tension than in the curves for torsion .
The second point of difference is that the tension culyes lie to the right of the curve for zero stress whereas the torsion curves lie to the left .
' .
MESSRS .
G. F. C. SEARLE AND T. G. BEDFORD S68 .
So far the strains have been practically within the elastic limit .
We now examine some cases in which this limit was much exceeded .
In the curve , plotted from the experiments described in S 60 , the points appear to be irregularly placed till the torsion wheel has made about eight revolutions .
For greater strains the points lie well on the straight line , Since this may be written In the experiments of S 61 the points corresponding to both the curves and cluster round a single curve , which when exceeds 2500 is represented by .
S69 .
We are not prepared to offer any physical explanation of the formula .
But as it has a rational appearance it seemed worth while to test it on some results obtained under zero stress .
We plotted against for several of the tables given by Professor J. A. EWING and Miss KLAASSEN , as well as for the results for cobalt obtained by Professor J. A. In every case the curve was straight over a considerable range of .
For small values of the value of given by is too small , while for large values of it is too large .
The value of begins to be too large when with increasing the permeability begins to fall rapidly below its maximum value .
When , with decreasing falls much below its maximum value , becomes too small .
It also appears that when is less than , the value corresponding to the maximum of may , without causing serious error , differ much more from its maximum value than when exceeds .
The following examples from the experiments of Professor EWING and Miss KLAASSEN will serve as illustrations .
[ November 4 , 1901 .
Mr. WILLS has obtained from his experiments a series of curves showing how , for constant values of , depends upon , when is varied by varying the temperature .
These curves exhibit in a striking manner the characteristic features of our own curves .
Thus the curve for a given value of consists of a straight line and a hook , while the straight lines , corresponding to different values of , all radiate from a point on the axis of W. Mr. WILLS , however , finds that the index of instead of .
The relation deduced from his experiments is thus ] " " On the netic Qualities of Iron ' Phil. Trans , vol. 184 , p. 985 .
' The gnetic Hysteresis of Cobalt ' Proc. Physical Soc vol. 16 , p. 519 .
ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
Effect of Electric urrent upon Hysteresis .
S70 .
If an electric current be sent along an iron wire it produces a circular magnetic force which tends to link together the magnetic molecules in circular chains .
We may expect that this linking of the molecules will make them less susceptible to the influence of a longitudinal magnetic force , and that in consequence , for a given range both the range of the longitudinal component , , of the magnetic induction and also the energy dissipated by hysteresis in each cycle would be diminished .
The effect deserves a systematic investigation , but this up to the present we have not been able to carry out .
We must content ourselves with recording some qualitative experiments which show that the expected effect of an electric current actually occurs .
The experiments were made in March , 1896 ; by one of us with the help of Mr. JOHN TALBOT .
An iron wire about millim. in diameter was used , and an alternatin current was employed in the plimary circuit , giving rise to a steady deflexion of the dynamometer coil proportional to the energy dissipated by hysteresis in each cycle , In an experiment made on March 9 , a steady current varying from to amperes was sent through the wire , with the result that the hysteresis was diminished .
No observations were made to determine , or in absolute measure , and thus we can only represent by means of the deflexion of the dynamometer .
The table shows the result of the experiment , the third column recording the percentage diminution in the hysteresis occasioned by the passage of the current .
It will be seen that the strongest current diminishes the hysteresis by nearly one-quarter . .
cit. , p. 995 .
, p. 1002 .
The experiments of GEROSA and FINZI are described by Professor EwING , Magnetic lction iu , \amp ; c 3rd Edition , p. 330 .
MESSRS .
G. F. C. SEARLE AND T. G. BEDFORD In an experiment made on March 7 the wire was subjected to tension due to a load varying from to 20 kilogrammes .
The deflexion due to hysteresis was observed ( 1 ) when no current flowed through the wire , and ( 2 ) when a current of definite strength flowed through the wire .
The current was furnished by a single Daniell cell , and , judging from the last experiment , was about amperes .
In the table the last column shows the percentage diminution of the hysteresis due to the passage of the current .
The numbers in this column are rather irregular , but they show quite clearly that the effect of the current diminishes as the tension increases .
The initial increase of and its subsequent decrease , noticed in detail in S52 , are well shown in this experiment .
S71 .
In concluding this paper we desire to express our thanks to Professor J. J. THOMSON for his encouragement during the progress of the experiments , as well as for the use of the resources of the Cavendish Laboratory .
We are also indebted to Mr. JOHN TALBOT , of Trimity College and to Mr. W. G. FRAZER , Fellow of Queens ' College , for valuable assistance during the earlier stages of the work , and to Mr. L. N. G. FILON , of King 's College , for help in connexion with Appendix I. We have to thank Dr. R. S. CLAY , of St. John 's College , for some help in the preliminary experiments .
We gladly record obligation to the writings of Mr. OLIVER HEAVISIDE , for it was by the method of operators , so fruitfully used by him , that we first obtained the complete theory of the method .
Our thanks are also due to Mr. W. G. and to Mr. F. LINCOLN , the mechanical assistants at the Cavendish Laboratory , for help and advice on many occasions .
Their mechanical skill has been of great service to us .
ON THE MEASUREMENT OF MAGNETIC HySTERESIS .
On the produced by Eddy Currents in Rod of Circular Section .
S72 .
The problem cannot be completely solved unless the permeability , , is mdependent of the magnetic force , a condition not fulfilled with actual specimens of iron .
Though this is so , we can obtain useful information from the complete solution when is constant .
If the magnetic force be parallel to the axis of the rod , and if be the current at a distance from the axis , and if be the specific resistance , then the field within the rod has the characteristics . . .
( 1 ) , . . . .
( 2 ) , so that .
( 3 ) .
Expressing in the form . . . . . .
( 4 ) , where , .
are functions of only , we see that , since when Using this value of in ( 3 ) , and comparing coefficients of powers of , we find , .
, and thus obtain .
Hence for , the magnetic force at the surface where . . .
so that . . . .
Hence . . .
( 5 ) .
Thus when is known as a function of can be found at any point of the section .
Now , if we had assumed that is constant over the section , we should have found ffom ( 1 ) , since when .
This assumption is thus equivalent to the assumption that the second , third , .
terms in ( 5 ) are negligible in comparison with the first term .
In this case we find for the rate at which heat is generated by eddy currents where has the meaning assigned to it in S 8 .
Here , since is constant over the section , we may put , and thus Hence , with the notation of ( 13 ) S9 , The second term of ( 5 ) will be negligible in comparison with the first , provided that is negligible in comparison with , and the third term will be negligible in comparison with the second if is negligible in comparison with Now , as in S15 , the characteristic of is Hence , supposing that may be treated as constant , Thus we see that the ratio of each term in ( 5 ) to the term before it is small , provided that is small compared with the " " time constant\ldquo ; .
When this condition is satisfied , we may treat and also as constant over the section of the rod , and may then calculate the eddy current from ( 1 ) .
On the Heat produced by Eddy Cur .ents in Rod of Rectangula Section .
S73 .
The section of the rod is supposed so small that the rent at any point may be calculated by FARADAY'S law , on the assumption that has the same value at all points of the section .
We see by the case of the circular rod that this assumption is legitimate , provided that is small compared with being the radius of the largest circle inscribable in the section .
Let , now , be the sides of the rectangulal- .
section , and let the origin be at the centre of the section .
Then , since the magnetic force is parallel to the axis of the rod , we have , under the specified conditions ( 1 ) , when . . .
( 3 ) , ( 2 ) , when ( 4 ) , where are the components of the current , and ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
Now ( 2 ) is satisfied if we write . . .
, while ( 1 ) now becomes . . . . .
( 6 ) .
The solution of ( 6 ) , appropriate to the problem in hand , is ( 7 ) .
This value of satisfies , because , and both rangin from to within the limits Now by ( 5 ) and ( 7 ) with a similar expression for .
These expressions satisfy ( S ) and ( 4 ) , and thus all the conditions are fulfilled .
The rate at which heat is generated is given by . . . . .
( 8 ) .
The necessary integrations are easily effected , for the integral is zero unless and are equal .
When , its value is .
Similar results hold when two sines are substituted for the two cosines .
We thus obtain It is not convenient to calculate from this double series , on account of the slow convergence .
We therefore transform it into a single series .
Now . . .
Differentiating the logarithm of both sides , we obtain .
( from to ) .
* The method of solution was suggested by notes taken by one of us at a course of lectures on hydrodynamics , given by Mr. R. A. HERMAN , Fellow of Trinity College , in November , 1889 .
VOL. CXCVIII .
\mdash ; A. Expanding in powers of and comparing the coefficients of and of , we find Hence Now put , and then Writing this in the form , we have ( 9 ) .
As increases , the expression ( 9 ) very rapidly tends to the limit ( 10 ) since , the error not exceeding 1 in 4000 when is as small as 2 .
We give a table of the values of for small values of ; it will be seen that is rather smaller for a square than for a circular rod .
Mr. L. N. G. FILON has kindly verified our results , employing the mathematical method used by DE ST .
VENANT in finding the torsional rigidity of a rectangular prism .
THOMSON TAIT , Phil Part APPENDIX II .
On the Demagnetising Force due to Rods of Finite Length .
S74 .
In order to find how the demagnetising force at the centre of a long cylindrical rod depends upon the induction at the centre of the rod , the following experiments were made .
The magnetising solenoid was placed at right angles to the [ For an elliptic cylinder , axes satisfies ( 6 ) as well as the condition that no current crosses the bounding surface .
We hence find )\mdash ; December 26 , 1901 .
] ON THE MEASUREMENT OF MAGNETIC HYSTERESIS .
magnetic meridian , and a magnetometer was so adjusted that its magnet was vertically above the centre of the solenoid , and as near to the solenoid as possible .
On sending a current through the solenoid a small deflexion of the magnet ensued , due to the finite length of the solenoid ; by adjusting the wires connecting the solenoid to the rest of the apparatus this effect was easily annulled .
On now placing the rod inside the solenoid there was a deflexion of the magnet , due entirely to the distribution of magnetism on the bar .
When the diameter of the rod and the distance of the magnetometer needle from the rod are both small compared with the length of the rod , the magnetic force at the magnetometer , due to the rod , differs very little from the magnetic force at the centre of the rod due to the rod itself Under these conditions the deflexion of the magnetometer may be taken as a nearly exact measure of the demagnetising force due to the rod .
When the specimen was a single thin wire , a mirror magnetometer was used ; for bundles of wire , a mirror magnetometer had too small a range , and then a magnetometer , with a pointer moving over a circular scale , was employed .
If be the deflexion of the magnetometer , and be the earth 's horizontal magnetic force , then the demagnetising force is given by S .
When iron is tested for permeability , the specimen , previously demagnetised ' ' by reversals is subjected to a magnetic force which is reversed many times between the limits .
The maximum induction is then determined from the galvanometer " " throw\ldquo ; due to a single reversal of the magnetic force between the same limits .
Starting from a small value , is increased by suitable steps , and the value of corresponding to each value of is determined , the process yielding a single point on the diagram for any given value of We made a test of this kind upon an anmealed soft iron wire , determining also , by the method of S74 , the demagnetising force for every value of employed .
The area of section of the wire was sq .
centim. , and its length was centims. As the solenoid was 60 centims. in length and of small diameter , the magnetic force due to the current was very nearly uniform over the whole length of the specimen .
A mirror magnetometer was used to determine , the needle being about 3 centims. from the wire .
The results of this experiment are shown graphically in fig. 18 .
The abscissae of the curves marked and represent the values of , the mean maximum induction at the centre of the wire , while the ordinates represent the corresponding values of and respectively .
The nearly straight line passing through the origin indicates the value would have if the flow of induction from the wire occurred entirely at its ends , in which case the " " poles\ldquo ; would be concentrated at the ends of the wire .
This lin represents , where is the mean maximum of intensity of magnetisation at the centre of the wire , and is the length of the wire .
Initially is nearly proportional to .
Since , however , much of the induction leaks out from the wire by the cylindrical surface , thus developing a distribution of magnetism along the wire , is considerably greater than it would be were the\ldquo ; poles\ldquo ; at the ends of the wire .
As increases , increases less rapidly , reaches a maximum for a value of [ 10300 ] somewhat greater than [ 8400 ] , the value corresponding Naenet io lnduction Fig. 18 .
to the maximum value of , and then diminishes .
As ) increases still further , diminishes rapidly till reaches the stage where begins to diminish comparatively slowly .
After passing a minimum again increases , and in this last stage is less than S 76 .
The experiments of Mr. C. G. LAMB , " " On the Distribution of Magnetic Induction in a Long Iron serve to explain the rise of to a maximum , and its subsequent fall .
For he found that as increases from zero , the percentage of the induction at the centre of the rod which leaks out between the centre and the end of 'Proc .
Phys. Soc vol. 16 , p. 509 ; or 'Phil .
Mag Sept. , 1899 .
ON THE MEASUREMENT OF MAGNETIC HySTERESIS .
the rod , increases until reaches approximately .
Thus as increases we may regard the " " poles\ldquo ; as moving inwards towards the centre .
There is thus a double reason why should increase with during this stage .
When increases beyond , he found that the " " poles\ldquo ; move out again towards the ends of the rod , and we must suppose that this motion more than compensates for the increasing strength of the poles with increasing In our experiments , for the largest values of was less than .
This effect can only arise from the development of a subsidiary pole between the centre and either end of the wire , with sign opposite to that of the pole at the end .
This result is easily seen to be impossible with a bar of uniform material as long as the induction is a definite single-valued function of the magnetic force , which increases as the magnetic force increases .
It must therefore be due to the effects of hysteresis , which becomes an increasingly important factor as the point of maximum permeability is reached and passed .
A different result might perhaps have been obtained if the magnetic force had been gradually increased from zero .
But in our experiments , as in those of LAMB , the magnetic force was put through seyeral cycles between the limits before the observations were made .
S77 .
When iron is tested by the ballistic method for hysteresis , the specimen is subjected to an applied magnetic force which is reversed between the limits until the iron has reached a cyclic state .
This attained , the points on the cyclic B\mdash ; H diagram are found from the throws of the galvanometer which occur when the magnetic force is suddenly changed , by means of a special key , from to a series of values between and We made hysteresis tests by this method upon two specimens , determining also the demagnetising force , , at the centre of the wire , for each value of which was employed in constructing the B\mdash ; H curve .
Thus , the magnetic force was changed from to , and the throw of the galvanometer , giving , was obsexved , and then , without altering the magnetic force from its value , the deflexion of the magnetometer was noted .
We reckon positive when its direction is opposite to that of the induction at the centre of the specimen . .
The first specimen was a bundle of ten iron wires with a total area of section of 0412 sq .
centim. The length of the wire was equal to the length of the solenoid , * Dr. L. HOLBORN , in a paper " " On the Distribution of Induced Magnetism in Cylinders " " ( ' Sitzungsberichte der Akademie der Wissenschaften zu Berlin , ' 17th February , 1898 ) , has obtained a resuIt similar to that found by Mr. LAMB .
He used two secondary coils , one a uniformly wound solenoid closeJy fitting the rod , the a coil wound about the centre of the rod .
The rod and the secondary coils were placed inside a long magnetising solenoid .
By comparing the changes of induction through these two secondaries due to a reversal of the primary current , he found the distance between the " " centres of gravity\ldquo ; of the free magnetism on the two halves of the rod .
He found that ths " " of gravity\ldquo ; move towards the centre as increases , until reaches .
A further increase of caused the " " centres of gravity\ldquo ; to move out again from the centre .
Dr. HOLBORN made similar experiments on ellipsoids , and found that for them was remarkably constant .
47 centims. , and thus since the internal radius of the windings was about 2 centims. the applied magnetic was far from uniform near the ends of the wires .
The results of this experiment are shown in figs. 19 , 20 .
As we pointed out S18 , exhibits hysteresis with respect to both and H. In fig. 19 the straight line Fig. 19 .
Fig. 20 .
through the origin represents , the demagnetising force corresponding to uniform magnetisation , in these experiments being negligible in comparison with B. From the areas of tho curves we found ON THE MEASUREMENT OF MAGNETIC HySTERESIS .
S79 .
The second specimen was a single iron wire 47 centims. in , and 00412 sq .
centim. in section .
The magnetisin solenoid was that described in S75 , and thus the applied magnetic force was practically unif.orm over the whole length of the wire .
We examined the polarity of the eastern end of the wire at every stage by means of a small compass .
We show the results of the experiment in figs. 21 , 22 , and also in the following table , where the last cohlmn indicates the Fig. 21 .
Fig. 22 .
polarity of the eastern end .
Tt will be seen that the polarity of the end of the wire agrees in every case with that corresponding to the direction of the induction at the centre of the wire .
In fig. 21 the straight line through the origin represents , the demagnetising force corresponding to uniform magnetisation , in these experiments being negligible in comparison with B. As Bdiminishes from 9850 , diminishes to a minimum at , and then increases to a maximum at ; after this it passes to its greatest negative value corresponding to without passing through a maximum or minimum .
As increases again to 9850 , goes through a similar set of changes , its values from to 9850 differing but little from those which it had when was diminishing .
From to 4700 , is less than , and from to has the opposite sign to B. Both these cases require that there should be " " poles\ldquo ; on the wire between the centre and the ends , with signs opposite to those of the poles at the ends of the wire .
Our arrangement of apparatus was not well adapted for detecting these subsidiary poles , since the magnetic force due to a uniformly distributed pole would be at right angles to the wire , and therefore parallel to the mag1letic meridian , thus producing no deflexion of the search compass .
Still we were able to verify the conclusion when was , for though the corresponding -magnetism appeared at the eastern portion of the wire , being most concentrated at a point about 7 centims. from the end , yet we found a weak -pole at 9 centims. from the same end .
From the areas of the curves we found These experiments add emphasis to Mr. LAMB'S remark that " " the magnetometric method [ of determining I\mdash ; H curves ] , although extremely useful for comparative work , must be used with much caution in determinations of an absolute character [ December 23 , 1901.\mdash ; C. BENEDICKS ( ' Annalen der Physik , ' 1901 , vol. 6 , p. 726 ) compared the B\mdash ; H curve for a cylinder of steel with the curve for an ellipsoid formed out of the same piece of metal , and deduces that , as increases , rises to a maximum and then rapidly decreases , as in our fig. 18 .
From EWING'S experiments ( ' Phil. Trans 1885 , pp. , 535 on ( 1 ) a ring , ( 2 ) straight pieces of iron wire , he deduces the same result .
Neither experimenter , however , reached the minimum of ]

