]\gt ; [ 329 ] IV .
On the ofPressure on gnetisation of Cobalt .
C. CHREE , ) of 's Colleye , icated by Professor J. J. THOMSON , F.R. Received November 22 , \mdash ; Read , 1889 .
[ PLATES 15 , 16 .
] S 1 .
IN August , 1888 , I commenced an investigation into the effects of pressure on the magnetic properties of cobalt , at the suggestion of Professor J. J. THOMSON .
In his : ' Applications of Dynamics to Physics and Chemistry he has rrived at certain conclusions as to the relations of magnetisations and mechanical strains , and the primary object of the following investigations was to obtain ental data , whereby these conclusions might be tested .
the process of the experiments , numerous points presented themselves whose elucidation seemed of importance for a satisfactory comparison of theory and experiment .
The experiments have thus included a wider range than was originally intended , and have occupied a considerable amount of time .
Phenomena resembling very closely in some points those to be presently discussed have been noticed by previous observers in iron and nickel .
As frequent references to these will be found essential for an adequate discussion of the phenomena presented by cobalt , a brief outline of the more prominent characteristics is here interposed .
S 2 .
If a bar of some magnetic metal , subjected to a con stant lonitudinal stress , be exposed to the action of a magnetising current in a surrounding spiral it becomes magnetised , and in general simultaneously alters its length .
If the current cease to flow the bar retains in general part of its magnetisation , and it may not return to its original length .
On the other hand , if the bar while under the action of a current in the magnetising spiral , be subjected bo a longitudinal stress , it in general simultaneously alters its netisation .
Also , if the stress cease to act the magnetisation may not return to its original value .
Phenomena similar to the last are also shown by the magnetisation which is residual after the break of the current .
These phenomena all depend on the intensity of magnetisation in the bar .
The mode of this dependence will discussed in considerable detail presently , but erroneous ideas are not unlikely to arise unless the following practical considerations be kept clearly in view .
3.30 MR. C. CHRLE ON THE EFFECTS OF PRESSURE When bar whose diameter of cross section is comparable with its length is magnetised in a uniform field , there seems good reason to believe that the magnetisa .
it possesses is very decidedly less in the interior near the axis than it is near the surface , and when the bar is possessed merely of residual magnetism , the difference between the interior and surface distributions seems completely proved .
When , as usual , the magnetising field is not uniform , there is in general still less reas to expect a uniform intensity of magnetisation .
Thus , in general , a bar may be regarded as composed of a large number of elements , throughout each of which the magnetisation may be treated as sensibly uniform , but which may differ widely amongst themselves .
The term intensity applied to the magnetisation of such a bar merely means some sort of of the tensities throughout the several elements .
Now the application of given stress to such a bar may affect very differently the magnetism of different elements .
It may , for instance , increase the magnetisation of some of them and diminish that of others .
Experiments as a rule give merely a sort of integral of the effects on the different elements , and this ought to be clearly recogllised .
In particular I would point out that the magnetic moment of a bar , as determined by some definite method , may have the same value when the bar is exposed to the action of a certain current in a spiral , as it has after bhe break of a larger current in the spiral , and that notwithstanding tbere may be an importanb ence between the distributions of magnetism in the two cases .
There is thus no a priori reason to expect the same phenomena to follow the application of a given longitudinal stress in the two Effects of gnetism on the Length of Bars . .
Such effects have been observed by JOULE , Professor BARRETT , Mr. SHELFORD BTDWELL .
S JOULE'S experiments dealt with iron and steel .
His method of experiment consisted in running a current through a netising spiral and observing the instantaneous changes in the length of the bar accompanyiug the make and break of the cul'rent .
Observations were taken with several currents gradually rising in strength .
Apparently JOULE did not demagnetise his rods in the course of his experiments , nor does he seem to have examined the of making and breaking the same current more than once .
He concluded that in iron or soft steel bars free from stress the institution of a current in a surrounding spiral is followed by an increase in the length of the proportional to the square of the intensity of the magnetisation * ' Phil. Mag vol. 30 ( series ) , 1847 , pp. 76-87 and 225-241 .
Tbid .
, vol. 46 ( 4th series ) , 1873 , pp. 177-201 .
, vol. 47 ( 4th series ) , 1874 , ; also ' Nature , ' vol. 26 , 1882 , p. 585 .
S 'Phil .
Trans , vol. 179 , 1888 , pp. 205-230 .
Also 'Roy .
Soc. Proc vol. 40 , 1886 , pp. 109-133 and ON THE MAGNETISATION OF 3.31 induced in the bar , and that on the of the current there remains an increase in length proportional to the square of the intel ) sity of the residual magnetisation .
JOULE also found that iron and soft steel under tension lengthened or shortened when magnetised according as the sion was small or great .
He believed the tening proportional to the product of the \ldquo ; magnetisation into the strength of the magnetising current .
This last usion , however , is certainly not w , arranted by the results of his experiments .
On JOULE'S p. 237 , are given the results of an experiment , numbered 23 , in which a soft steel wire exposed to a certain tension showed an increased while exposed to a certain current , lengthened further when the current was broken , and then shortened on the make of a stronger current .
This suggests that for the wire in question under a given load there existed a critical intensity of magnetisation , and that the bar lengthened or shortened according as the magnetisation was less or greater than the critical value .
The true interpretation of this experiment seems to have been entirely missed by JOULE , and it would not appear to have attracted the notice of subsequent observers .
It is thus only within the last few years that the existence of a critical field , or state of magnetisation , has been brought to light by the decisive experiments of Mr. SHELFORD BIDWELL .
S4 .
IAYER experimented with rods free from stress , and he gives the results of the make and break of only one strength of current .
The current was , however , twice made and broken in some at least of the experiments , the results of both the magnetic cycles are recorded .
Some of the resuIts , given in a table on his p. 200 , are so snggestive in the light of our present knowledge that I reproduce them here in a form slightly altered to suit the inquiry .
The numbers in the first column distinguish the rods used , all the other numbers give the lengthening , or when with a minus sign the shortening , of the rods in terms of an itrary unit .
, but this seems a int .
MR. .
CHREE ON OF PRESSURHi breaking second ts the rods all returlled to the length they possessed on the break of the first currents , so that the changes in the lengths of the rods became cyclic after a single make and break of the current .
The last three experiments are , it will be observed , very far from agreeing with JOULE'S laws .
The experiments suggest that it is mainly on the intensity of the existing netisation that the phenomena depend , but the divergence between the figures in the second and fourth columns indicates that the previous history of the material may , in some cases at least , be of considerable importance .
In all the iron rods the induced netisation is clearly much below the criticaI .
In 7 , the softest of the steel rods , the induced magnetisation though below is clearly approaching the critical .
In rod 8 , the critical magnetisation lies between the induced and the residual ; and if a critical magnetisation existed in the case of rod 9 it must have been less than the residual .
5 .
BIDWELL'S experiments those described in the ' Philosophical ' appear the most complete .
In these , he first demagnetised the specimen , and then sent through a surrounding spiral a series of currents gradually rising in strength .
For each strength of current the alterations in the length of the rod accompanying two successive makes were observed and recorded .
The Tables III .
and IV .
, pp. 220 and 221 , for an iron rod show a small but unmistakeable difference between the effects of the first and second makes of a current of giyen strength .
In Table .
p. 222 , for a cobalt rod the differences might well be attributed to experimental errors .
Mr. SHELFORD BJDWELL found critical fields both for iron and cobalt .
Iron he found to lengthen or shorten when magnetised according as the magnetising field was below or above the critical , while cobalt presented exactly the opposite phenomena .
A nickel rod shortened in all fields up to 1400 C.G.S. units , and the experiments left the existence of a critical field an open question .
The critical fields for several specimens of iron varied from about 270 to 380 C. .
units , and the maximum elongations appeared in fields of about 80 C.G.S. units .
In one cobalt rod the maximum shortening occurred in a field of about 400 C.G.S. nnits , and the critical field considerably exceeded 800 C.G.S. units .
A second softer specimen showed a maximum shortening in a field of about S00 C.G.S. units , and had a critical field of 750 C.G.S. units .
In these experiments , the rods were free from stress .
Mr. experiments , however , described in the ' Proceedings of the Royal Society , ' show that the critical fields for iron are lowered by tellsion , a result in exact accordance with JOULE'S experiments .
S 6 .
Professor BARRETT found a cobalt rod to lengthen when exposed to a comparatively weak field .
The data he supplies are too limited to enable one bojudge whether his later experiments on the subject were wholly free from the objections which he believed to have invalidated the conclusions based on his earlier experiments .
I scarcely think his results afford any reasonable ground for doubting ON THE MAGNETISATION OF COBALT .
.3.3.3 the accuracy of Mr. BrDWELL'S conclusions , at least for the rods of his experiments .
Effects of Stress of Rods TVires . .
A preliminary general idea of these effects may be obtained consideration of the phenomena presented by a soft iron wire stretched beyond its original limit of perfect elasticity , and then subjected , in a weak netic field .
to a stress cycle consisting of the application and of a definite tension for which it is perfectly elastic .
The first application of the tension causes a large increase in the tion of the wire , and on the removal of tension this increase nearly remains , may even be added to .
The next one or two tension cycles cause a distinct ressive increase in the netisation .
Presently , .
to the tension cycles , there appears a nearly cyclic in the \ldquo ; in which the ximum sation appears when the tension is " " on Also the Ilitude of the cyclic effect remains , at least approximately , stant for , all evcnts , a large ltnber of tension cycles .
Subtracting the effec of the final cyclic change of magnetisation from the change accompanying the first applicution of tension , we get what be fairly } as that portion of the effect of the first tension which is of a permanent and non-cyclic character .
This quantity I have here termed the ' ' \ldquo ; If current be broken , the first application of tension or of pressure has a shockeffect , which invariably consists in a diminution of the residual Jnagnetisation .
The effects of beating , other mechanical agitation of a rod , are in mauy respects very similar .
The experiments on the effects of stress on magnetism which are of most importance for our present purpose are those of VILLARI , Sir W. THOMSON , Mr. SHIDA , and Professor EWING .
All these observers have , at least in practice , recognised the distinction between the effect on a bar 's magnetisation of the first few applications of a given longitudinal stress and its subsequent applications .
S8 .
VILLARI unfortunately gives no clue to the strength of his magnetising fields , save the number and relative size of the DANIELL or BUNSEN elements he employed .
He discovered that the non-cyclic effects of the first tension and the cyclic effects of the later tension cycles in bars\mdash ; or rather stout wires\mdash ; of iron in weak netic fields are of the character indicated in the last paragraph .
In stronger fields he found * See ' ' vol. 8 , p. 666 , \amp ; c. POGGENDORFF'S ' Annalen , ' vol. 126 , 1865 , pp. 87-122 .
'Phil .
Traus 1879 , pp. 55-85 ; or 'Mathematical and Physical Papers , ' vol. 2 .
S 'Roy .
Soc. Proc vol. 35 , 1883 , pp. 404-454 .
'Phil .
1885 , pp. 523-640 ; 1888 , A. , pp. 325-337 .
3.34 MR. C. CHREE ON OF PRESSURE that , the first application of the tension still increased the bar 's magnetisation , the cyclic eflect had changedsign .
On his p. 91 , in experiment 3 , is given an instance where a soft iron bar , exposed to one of his strongest fields , lost magnetisation on the very first application of the tension .
From the data supplied , however , it will be found tha .
the shock-effect of this first tension was still an increase of tion .
Similar cases are also referred to on his p. 97 , but no data are given from which the sigll of the shock-effect can be deduced .
From VILLARI'S experiments it obviously follows that a critical field , or intensity of magnetisation , must exist where the cyclic ects of tension vanish .
The field in which the effect vanishes is by Sir W. THOMSON the " " VILLARI critical point He and Professor EWING have made it clear that the field in which this phenomenon occurs depends on the amount of the tension which is applied and removed , and also on the existence of permanent load .
Professor EWING also prefers to define the VILLARI point by reference to the magnetisation existing in the rod , and not to the field producing it .
Some of VILLARI'S steel bars behaved like iron , but in others a critical field , if existent , was lower than the lowest field he employed .
VILLARI also observed the fall in the residual magnetisation caused by the first tension after the break of the current .
He further observed that after several tension cycles there appeared a cyclic change in the residual ma.gnetisation .
Without making any restriction as to the strength of the pre-existing fields , he lays down the law , that in soft iron the residual magnetisation in the cyclic state is greatest when the tension is " " on while in hard iron and steel , it is greatest when the tension is " " off I can find no clue as to the actual strength of the pre-existing field or fields he employed .
Fitting an iron core inside a coaxial iron ube of the same length , VILLARI found that in fields between certain limits , the core and the tube showed , in response to cyclic changes of tension , cyclic changes of fnetisation of opposite signs .
The phenomena , as VILLARI himself pointed out , are exactly in accordance with the view , stated in , that the intensity of magnetisation is greater on the surface of a bar than in its interior .
S 9 .
Sir W. THOMSON experimenting on a soft iron wire permanently stretched , and then exposed to cycles " " on\ldquo ; and " " off\ldquo ; of some weight causing no further permanent extension , found a critical field which was in general higher the smaller the weight employed in the tension cycles .
The critical fields obtained by the magnetometric method\mdash ; calculated presumably for the centre of the magnetising coil\mdash ; averaged about 20 C.G.S. units , and the fields which the cyclic effect was a maximum , were roughly about a quarter of the critical .
By the ballistic method with an induction coil at the centre of a much longer magnetising coil , critical fields such as C.G.S. units were obtained .
This Sir W. THOMSON apparently assigns to the experimental wire having its magnetisation greatest at the centre of the magnetising spiral , so that this portion of the wire would ON THE MAGNETISATION OF COBALT .
be brought into the critical state by much weaker currents than the portions under the ends of the spiral .
In his S 240 , Sir W. THOMSON states that a critical point was found in the central portions of a nickel bar , but from a note to S 239 , it would appear that with the magnetometer opposite an end of the bar , no trace of a neutral field was obtained .
With a magnetometer as usual opposite one he foumd in nickel and cobalt a cyclic effect the reverse of that occurring in iron in fields below the VILLARI point , i.e. , the magnetisation was least when tension was " " on S 10 .
Mr. SHIDA obtained VILLARI critical fields of 15 and 10 C.G.S. units respectively for two different specimens of iron wire , which had been pel.manently stretched , and on which the tension cycles were performed with weights not much less than those causing the permanent extension .
With a pianoforte steel wire , he found that in the cyclic state tension " " on\ldquo ; co-existed with a minimum of magnetisation in all his fields .
S 11 .
The first thing in Professor EWING'S researches that concerns us is his method of demagnetising wires .
This consists in subjecting them to the action of a series of rapidly reversed currents diminishing in intensity .
In S 19 of the first of his papers referred to above , he states that his wires were thereby reduced to a standard condition\mdash ; different possibly , he admits , from their condition previous to their first magnetisation .
In his S 107 Professor EWING defines the VILLARI critical point as " " that value of the magnetisation at which reversal occurs in the sign of the lnagnetic difference produced by two ( assigned ) states of longitudinal stress He uses the term as applicable ( 1 ) to that value of at which no magnetic change accompanies the alternation from one to the other of two assigned states of stress , ( 2 ) to the value of answering to the intersection of the two curves giving the relations of to \mdash ; the field , \mdash ; when is gradually raised from zero , and separate experiments are completed with the wire in each of the assigned states of stress .
He also proposes that one of the assigned states of stress should be that answering to no load .
It will be noticed that Professor EWING'S first usage of the " " VILLARI critical field\ldquo ; is that which accords with Sir W. THOMSON'S .
In origina ] lie soft annealed iron wires stretched beyond the limit of perfect elasticity , Professor EWING found in every case distinct VILLARI points of the first kind .
These existed in lower fields the greater the weight producing the second assigned state of stress .
If in a certain field the cyclic state has been reached for the loads and , for which the magnetisation has the critical value , and the load be gradually raised from , then there ensues a rise in the magnetisation until the load reaches some value less than , and then an equal as the load is incre sed to Q. Even in the weakest fields Professor EWING seldom found the magnetisation increase continually when the load was raised nearly to the limit of perfect elasticity .
From the figures on Professor EWINe'S Plate 63 , it will be seen that the tirst application of load has eflects very similar to those just described .
It will appear , however , from fig. 42 , 88 , that the critical fields at which the first application of given load has no effect on the magnetisation are distinctly higher than the VILLARI points ) the cyclic applications of these loads .
This figure also shows that the first application of a moderate load produces a very large increase of the magnetisation in fields , whereas in the field of 34 C.G.S. units the effect is trifling .
From fig. 40 it will be seen that in the cyclic state the effects of tension cycles on the magnetisation residual after a field of C.G.S. units were of the same general character as appeared with an equal induced netisation .
Soft annealed wires , when none but very small loads were used , were found by Professor to possess in weak fields a maximum of tisation with tension ' ' on agreeing so far with stretched wires .
With greater loads the effects become reversed after the first application of the load .
Professor EwlNG attributes this to hystcresis , and shows that when tapped these wires behave as stretched wires with low VILLARI points .
From , Plate 64 , it will be seen that soft wires respond to the first application of tension in a precisely similar way to stretched wires , but have much lower critical fields .
The phenomena accompanying the application of load cycles to a soft annealed wire possessed of the magnetism residual after the break of a field of 34 C.G.S. unit , discussed and shown in fig. 39 .
They are unquestionably of the same general character as those occurring with an equal induced magnetisation .
Iron wires , both stretched and annealed , and a pianoforte steel stretched after annealing , were exposed by Professor wING to gradually increasing magnetising currents , and the curves connecting and are given in his Plates 64 and 67 .
The specimens all showed VILLARI points of the second kind , which agreed with tbose of the first kind in being as rule lower , the heavier the load .
These critical fields , or of the second kind were distinctly higher than the corresponding ones of the first kind .
Simultaneously observations were taken of the residual magnetisation whose results are shown in figures in Plates 67 and 68 .
In all the specimens the magnetisation residual after weak fields , and also its ratio to the induced magnetisation , were found to be largely increased by the presence of a load , the increase being greatest with the heaviest loads used .
As the fields were raised the influence of the loads diminished , and in the annealed iron and the steel wires the residual magnetisation and its latio to the induced became eventually less under the heavier loads than in the absence of a load .
S 12 .
In his papers in ' Phil. .
Trans A. , 1888 , Professor EWING found that in fields from to 116 C.G.S. units , presence of a tension on a thin nickel wire diminished its induced and residual , and also the ratio of the residual to the induced .
Also cyclic changes of tension " " on\ldquo ; and " " ofl were accompanied by ON MAGNETISATION OF COBALT .
cyclic changes of tVnetisation , in which the magnetisation , whether induced or , was least when the tension was " " on The effects of pressule on a nickel bar were exactly the opposite of those of tension , and the increase in the susceptibility was extremely large , especially near the " " Wendepunkt\ldquo ; \mdash ; or point where is a maximum .
Professor EWING looked for a VILLARI point in low fields but found no trace of its existence .
Considerations .
S13 .
The experimental results already stated will enable Professor J. J. THOMSON'S theoretical conclusions to be understood .
Let denote the strain in an isotropic elastic cylinder parallel to its axis , and the strains in the cross section perpendicular to and along the radius vector .
Also let and denote respectively the strength of the magnetic field , the intensity of magnetisation , and the coefficient of magnetic induction , assumed eyerywhere the same throughout the cylinder .
In the case of a uniform stress Thus , assuming ( ) , Professor THOMSON obtains for this case , on p. 51 , two equations ( 4S ) , of which the first may be put in the slightly altered form ( 2 ) .
Here denotes a positive constant depending only on the elastic properties of the cylinder , and is POISSON'S ratio .
All differential coefficients must of course apply to some one definite state of the cylinder , and in determining and the strength of the field must be kept constant .
It is pointed out by Professor THOMSON that according to EWING'S experiments is positive for iron .
It is obviously positive for any magnetic metal in any field exceeding that which answers to its " " Wendepunkt It also appears to be positive for nickel and cobalt throughout the experiments of Professor ROWLAND Thus it is probably safe to regard it as essentially positive .
Under a uniform longitudinal stress * ' Phil. Mag vol. 48 , 1874 , p. 321 .
If then , through a mall increase in the stress , there be increases Se , Sf , Sg in the strains , we must have SgSf Se . . . . .
( 3 ) .
If be the consequent increase in the netisation , the field remaining constant , @ @ and so by ( 1 ) and Thus from ( 2 ) ( 4 ) ; where and Se denote the small increments in the netisation and longitudinal strain which follow a small change in the longitudinal stress .
If ( 4 ) were a complete representation of the state of matters , then and should always have the , and vanish for the same value of the magnetisation .
It should be carefully noticed that in employing ( 4 ) we ] assume the changes in the strains to be very small , and thus the value of should be obtained by experiments in which the change in the longitudinal stress is very small .
Thus , in pal.ticular , the magnetisation at which vanishes is the as determined by ents in which the difference between the two assigned states of is very small .
And thus , in accordance with the experiments of Sir W. THOMSON Professor EWING , this theoretical VILLARI point may well be very much higher than those hitherto found by experiment .
It may also be as well to point out that the magnetisation answering to the vanishing of is not the critical magnetisation observed by Mr. SHELFORD BIDWELL where the rod resumes its original , but the much lower where the rod , if iron , its greatest extension , if cobalt , its greatest shortening .
Comparison of Theory xperiment .
S 14 .
The already recorded sufficiently show that in iron , there is in weak fields , and , in very strong fields , an agreement in sign between and , both expressions changing from positive to negative as the strength of the field is raised .
It will be noticed , however , that the magnetisations obtained experimentally for the VILLARI point are much ] lower than those obtained for the point of maximum lengthening of rod .
This difference is certainly in some measure accounted for ON THE MA iTISATION OF COBALT .
by the lowering of the VILLARI point by the comparatively large stresses practically enlployed .
Professor THOMSON also has pointed out that MAXWELL'S distribution of stress ] produce an independent system of strains in the bar , the strain in the direction of the lines of force being an extension .
Though a very small efFect this would tend in iron to raise the magnetisation where vanishes somewhat over the VILLARI point .
In nickel we see from the experiments of Sir W. THOMSON , Mr. SHELFORD BIDWELL , and Professor EWING , that } and are both tive in all ordinary fields .
If the results obtained by Sir W. THOMSON in the central portions of a nickel bar , in a strong field , be accepted , then there does actually exist a very high critical gnetisation at which , vanishes .
There should accordingly be , not in fields where Professor EWING looked for it , but in very high fields , a clitical magnetisation where a nickel vould cease to contract .
Such a phenomenon cannot be said to have its existence ted by Mr. BIDWELL , but at the same time it is certainly not disproved by his experiments .
In cobalt ; according to Sir W. is negative in weak fields , and so also is if we accept the results of Mr. SHELFORD BIDWELL .
According to the latter observer , however , changes sign in fields much higher , it is true , than the corresponding fields for iron , but still easily obtainable .
Further , the not recoyers from the very considerable shortening it has experienced , but thens considerably as the field is raised .
It would thus appear impossible to assign the phenomenon in any essential degree to the MAXWELL effect .
If then Professor theoretical condusions are sound , should change sign from negative to positive in a cobalt rod for a magnetisation quite within the reach of experiment .
of Theory in Cobalt .
S 15 .
The question whether or ot a point of the above kind exists in cobalt was the primary object of the following experiments , and a decided answer in the affirmative was obtained .
Under the moderate stress employed , was negative in weak fields and positive in strong , and an keable VILLARI point appeared in a field of about 120 C.G.S. This , it will be noticed , is a much lower } those which .
SHELFORD obtained for the vanishing of in ] of the specimens he employed , and the MAXWELL effect would here tend to lower these latter fields .
In my specimen , howeyer , as the field was raised from 120 to C. .S .
units the increased by less than 20 per cent. The accordance of theory and experiment would thus , in reality , appear to be ] osel in cobalt than in any recorded experiments on iron .
MR. C. CHREh ON THE EFFECTS OF Preliminary Sketch of Phenomena Observed .
S 16 .
As the phenomena observed are somewhat complex , their mutual relationships might be hidden under the multiplicity of facts , which the discussion of each separate phenomenon introduces .
A brief outline of some of the more important and certain results will , it is hoped , serve as a key to the subsequent more complete discussion .
It is important to notice that all the phenomena were observed in a given specimen of cobalt , magnetised by a current in a given coil , and that it was found by repeating several of the sets of observations at various periods of the research , that the character of the specimen had not to all appearance been altered during the process of the experiments .
The importance of this , when an attempt shall be made to connect the phenomena together , and explain them consequences of one or more fundamental principles , will be at once obvious to any one who has noticed the variety in the phenomena observed in different specimens of iron .
At the same time , when various specimens of the same magnetic metal\mdash ; or even of different gnetic metals\mdash ; are exposed to gradually increasing fields , and curves drawn in which the abscissae give the strength of the field and the ordinates the intensity of magnetisation , experiment shows the fields in which corresponding of the several curves occur may widely differ , but thab there exists a general resemblance in the form of the curves , which may be made much closer by properly altering the scale of bscissae .
Tbus , while the particular field at which a certain phenomenon presents itself in a particular specimen may so largely depend on the individual peculiarities of the specimen that it may be of little value as an isolated fact , yet the determination of the point on the curve where the phenomenon occurs , and the position it occupies relative to the points where other phenomena occur , may lead to the recognition of general Observed in the Induced gnetisation .
S17 .
The " " Wendepunkt or ) oint where the coefficient of induced magnetisation is a maximum , is a point in the curves that can easily be recognised .
Very probably the most satisfactory comparison of the phenomena observed in different specimens would be obtained by expressing the magnetisation of each specimen in terms of the gnetisation it possesses at its Wendepunkt as a unit .
The Wendepunkt of the specimen employed occurred in a field of about 35 C.G.S. units .
For the reasons just stated , the relations of both the induced and residual magnetisatlons to the strength of the field , though not a primary object of investigation , were observed under various conditions as to pressure .
'Ihese relations , in the case of the induced magnetisation , are shown in figs. 1-4 ( Plate 15 ) .
The difference between the effects on the induced netisation of the rirst few and ON THE MAGNETISATION COBALT .
subsequent a , pplications of longitudinal stress , referred to in as occurring in iron , is equally characteristic of cobalt , so long as the magnetisation is not much in excess of that of the critical or VILLARI point .
In the present experiments the applied was a pressure , and so Se being always negative , and the metisation in the weaker fields greatest when the rod was under pressure , was , as already stated , negative in fields below the critical .
The effect of the first application of pressure may , as explained in S7 , be regarded as the sum of the cyclic effect and of a non-cyclic or shock-effect .
In all fields below 200 .
units the shock-effect an unmistakeable increase of magnetisation .
In all stronger fields , up to at least 725 C.G.S. units , the nitude of the shockeffect was so small that its sign even was doubtful matter .
Thus , in cobalt , the shock-effect , so long at least as it is an appreciable quantity , gives an increase of netisati o. This consequence of the shock-effect , even in fields much above the VILLARI point , is known from VILLARI'S experiments , recorded in S8 , to be also characteristic of iron .
The absolute measures , both of the shock-effect of the first pressure of the cyclic effect , possessed maxima in the neighbourhood of the Wendepunkt .
Relative , howeyer , to the magnetisation existing prior to pressure , both these effects increased continually in importance as the strength of the field was reduced within the actual hmits of the experiments .
The total effect of the first application of pressure had a critical field of about 160 C.G.S. units ; which , it will be noticed , is decidedly higher than the critical field for the cyclic effect .
As stated in S11 , in discussing Professor EWING'S nents , the existence of a higher critical field for the first application of stress than for the cyclic application , is equally characteristic of iron .
As it is important to know how far , if all , the nature and magnitude of the cyclic effect of pressure depend on the circumstances attending the process of magnetisation , the rod lvas sometimes subjected to pressure before and during its introduction into the magnetising spiral .
It was found that there was no material alteration in the cyclic effect .
The mode of variation of the effects of the first and of the cyclic application of pressure with the strength of the field is exhibited in figs. 5-8 ( Plate 15 ) .
Phenomena Observed in the S18 .
Comparing figs. 1 and 9 , it will be seen that the residual magnetisation approaches " " saturation\ldquo ; in much lower fields than does the induced .
The intensity of the residual , as of the induced , magnetisation was found to depend on the treatment of the rod during the flow and the of the current .
When the rod was free from pressure at the instant the current was broken , the residual magnetisation in all fields below 120 or 130 C.G. .
units was increased by the apply .
C. CHREF ON THE FFECTS OF tion of pressure cycles during the flow of the current .
In stronger fields no certain effect could be attributed to the pressure cycles .
The effect , as ma be seen from fig. 10 , was greatest in weak fields .
The existence of pressure during the break of the current proved to have a considerable influence on the intensity of the residual magnetisation .
As might have been anticipated it increased the residual in weak fields , bnt the effect changed its si in fields fsr below VILLARI point , and showed signs of vanishing even in a field of 400 .
units .
Ths tios of the magnetisation to the strength of the pre-existing field and to the intensity of the induced mngnetisation under various conditions as to pressure are shown in figs. 11 and 12 ( Plate 16 ) .
From these it appears that in the absence of all pressure intensity of the residual magnetisation is extremely small in weak fields .
In this case the phenomena closely resemble those noticed by many observers in iron .
The only essential difference apparently is that the fields at which certain phenomena appear in cobalt are much higher than the fields at which the ponding 1henomena appear in ordinary iron .
This is in exact agreement with the views already expressed as to the rela , tive positions of the Wendepunkt .
The application of pressure cycles during the flow of the current increased immensely the intensity of the residual ation in weak fields .
The increase , in fact , due to this cause in the residual ) etisation in the weakest experimental fields , was even greater proportionally than in the induced , so that the ordinates of curves and of fig. 12 continually increase as the strength of the field is reduced .
The effects of the application or removal of pressure in shaking out the residual magnetisation were observed , and are shown in Curve II .
, fig. 13 , and in figs. 14 and 15 .
From the first of these curves it appears that the percentage of residual magnetisation shaken out by a series of pressure cycles continually diminished as the strength of the field was raised .
When the rod was free from pressure during the break of the current , the percentage of the residual magnetisation , shaken out by the first application of pressure , was in weak fields decidedly dimini .
by the previous application of pressure cycles during the flow of the current .
In fields over 70 or 80 C.G.S. units , however , the effect of previous pressure cycles was extremely small .
In weak fields the removal of a pressure that had existed during the break of the current was fully as effective in shaking out residual magnetisation as was the application of a pressure when the rod , during the break of the current , remained free from pressure .
In fields over 30 or 40 C.G.S. units , however , the of pressure became decidedly ] effective than the application .
As has been stated in SS 8 and 11 , VILLARI and Professor EWING found that cyclic applications of stress were followed by cyclic changes in the residual magnetisation of the same character as those produced in the induced magnetisation .
As it seemed important to test the generality of these conclusions , the cyclic effects of pressure on ON THE MAGNETISATION OF COBALT .
the residual magnetisation were here observed when the treatment of the the flow and break of the current was varied .
When pressure cycles were applied during the flow of the current , and pressure was " " on\ldquo ; during its break , the cyclic effect had .
invariably the same sign while the strength of the field was raised from 11 to 400 C.G.S. units .
The effect in this case was in the same direction as in the induced netisation below the point , i.e. , the magnetisation was greatest when pressure was " " on When , however , the rod was free from pressure during the break of the current , critical field was found to exist .
In the magnetisation residua ] after fields above the Cl.itical , the magnetisation was least when pressure was " " on Tbis critical field was raised by the application of pressure cycles during the flow of the current , but it was in any case much lower than the critical field for the induced magnetisation .
It follows , of course , that the intensity of the critical residual mag.netisation was very much ] less than that of the critical induced .
The amounts of the cyclic effects under the various conditions are shown in the fifth columns of Tables II .
and VIII .
, and in the sixth columns of Tables VI .
and VII .
These results show conclusively that for cobalt , and so in all probability for iron , the relations between stress and residual magnetisation must be largely dependent on circumstances other than the so-called intensity of netisation .
This , as pointed out in S2 , was iori far from improbable , and should act as a warning ainst extending to residual magnetisation laws proved only for the induced .
The fact that in fields over 60 or 70 C.G.S. units both the intensity of the residual magnetisation and the character of the cyclic effect of pressure are very little affected by the application of pressure cycles during the flow of current , while markedly influenced by the existence of pressure during the break of the current , seems worthy of special notice .
All the phenomena observed in the residual magnetisation apparently altered but little in magnitude , as the strength of the field was raised from 100 to 400 C.G.S. units .
This emphasises the difference between them and the phenomena observed in the induced magnetisation .
The Apparafus .
S19 .
The rod imented on was supplied by Messrs. JOHNSON and MATTHEY .
The mean of a series of measurements gave .
for its length and sq .
cm .
for the area of its cross section .
During the experiments the rod was contained in a brass tube , the closed end of which was sufficiently thick to afford an unyielding resistance to the rod when under pressure .
The rod could be slid in or out of the tube , but the fit was tight enough to prevent any lateral movement .
On the outside of the brass tube , at its open end , a screw of small pitch was cut , answering to the screw cut on the inside of a brass cap. The cap carried a projecting arm , forming a diameter of a cross section , by means of which a considerable couple could be exerted .
The pressure was applied to the cobalt by means of a sort of ram-rod rigidly attached to the inside of the brass cap. When the cap was screwed on this ram-rod advanced into the brass tube , and came to bear on the end of the cobalt rod .
Drawing of the brass tube containing the cobalt rod , removed from the coil .
A Closed end of brass fube .
B Pin which can be slipped through the holes in the cheeks attached to one face of the coil and , as in the figul'e , through the diametral hole bored in the .
Qolid end of the brass tube .
C Screw cut on the outside of the brass tube at its open end .
cap with scrcw cut on its inside to , and carrying ram-rod arrangement for applying pressurc to the cobalt rod .
FiFi Projecting arm , by means of which the cap is screwed on .
In the original experiments , conducted in August , 1888 , the brass tube was firmly imbedded in a block of wood , fixed inside a magnetising coil .
This coil had a length of , and consisted of five layers of very thick copper wire .
A current of one ampere produced a field varying from C.G.S. units at the centre of the rod to un its at its ends .
This , it must be admitted , is not a very uniform field , but the fall occurred mainly within a centimetre or two of the ends of the rod .
In ) this preliminary investigation the observations were taken somewhat roughly .
reason they are referred to here is that they clearly showed the existence of a cyclic and a non-cyclic effect of pressure , the existence of a VILLARI point for the former effect .
Further , the result of experiments on several different occasions reed in giving 120 C.G.S. units as a close pproximation to the strength of the field due to the current at the centre of the rod when the VILLARI point appeared .
The mode of variation of the several effects with the strength of the field showed also a general agreement with the subsequent more accurate observations taken with a different coil .
This seems to warrant the belief that the phenomena observed do not owe any of their essential features to peculiarities of the apparatus .
When it was attempted to obtain accurate results with the origiral apparatus various difficulties were cncountered .
In overcoming these I am much indebted to the ingenuity of Mr. BARTLETT , the assistant at the Cavendish Laboratory , who constructed the new apparatus .
ON THE MAGNETISATION O COBALT .
The new coil had a length of cm .
, an in ternal eter of 1.4 cm .
, and an external diameter of 7 cm .
It consisted of thirteen layers of copper wire of 24 cm .
diameter covered .
In the solid end of the tube holding the cobalt a diametral hole was bored at angles to the length of the tube , itnd sponding hole was made in each of two cheeks projecting from one of the of the coil .
By slipping a pin through these holes the tube was in a fixed posi relative to the coil , ftlld when the pin was withdrawn the tube and its contained cobalt could be removed and replaced without the least risk of disturbing the remainder of the apparatus .
A current of 1 ampere in the new coil produces a field varying from C.G.S. units at the centre to units at the ends of the cobalt rod .
Throughout the greater portion of the rod , however , the field is vely nearly the same as at the centre .
These figures refer entirely to the action of the current , no attempt having been ) to allow for the action of the rod itself .
In all the ] owing calculations the values to the field are calculated from the action of the current alone at the centre of the rod , and so are higher than the meaIl values of the actual fields .
difficulty of determining the actual fields at the different elements of the rod would be very great even ) the permeability as constant , a most erroneous supposition ; and , the differences to be expected between different specimens of the same metal , an tttelll ) at great accuracy in this department see1ns entirely supererogatory .
S20 .
Throughout the whole investigation the ordinary magnetometric method employed .
The axis of the coil was perpendicular to the magnetic mel.idian , and magnetometer was situated in the direction of this axis produced .
The lnagnetometer needle carried a mirror which formed on a millimetre scale , set parallel to the gnetic meridian , an image of a vertical wire placed across a slit in the centre of the scale behind which a lamp stood .
The direct action of the coil curTent was dised by a compensating coil traversed by the same current .
The strength of tlJe c recorded by a Thomson graded ameter in circuit with the coil .
The current was derived from the storage cells of the laboratory , and its strength varied by changing the number cells and the resistance in a wire bridge circuit with the coil .
S21 .
In order to render apparent the relative magnitudes of the several phenomena some common system of magnetic measurements was essential .
Thus all the scale readings have been reduced , and in the following tubles and diagrams only C.G. units appear .
From the previous remarks as to the method of calculating it will be clearly understood that the values given for the engtbs of the fields , and consequently for the coefficients of magnetic induction , cannot rigidly be ]leld to ) to the actual condition of the rod , in which and varied point to point , but must be regarded as giving an approximation to the mean state of the roo1 , and mole indicating by their variation the amounts of the changes actually occurrinl c It should also be understood that in ulculating the intensity of magnetisation the rod was treated as magnetised solenoidally .
In ] some of the experiments , , those magnetism , a direct calculation on this basis might .
introduced a considerable , owing to the small ance of the rod from ] the magnetometer .
The actually dings , however , were first reduced by direct comparison of the readings obtained in an independent set of experiments , in which the rod in a given netised condition was placed first of all in the positions it was about to occupy in the course of the main observations , and then in a certain standard position , where its distance from the magnetometer was so great that the precise position of the " " poles " " was of comparatively little moment .
As stated in S2 , in doubtless ries f point to point of the rod , and the values here recorded indicate lnereJ sort of average .
S22 .
the experimenbs in detail it will be as well to refer to disturbing agencies , the one of importance mainly in weak fields , the other in The the residual effects of previous netisations , affected very some of the earlier experiments here recorded .
In weak fields so important was this that the rod was sometimes found , after showing a considerable induced magnetisation , to possess , on breaking the circuit , residual magnetism of opposite The application of pressure in general increased this residual magnetisation of opposite sign , sometimes to a very considerable extent .
Or , supposing the residual magnetisation to be at first of the same sign as the preceding induced , it might nange s on the application of pressure .
These effects were produced at pleasure with the greatest when the rod was treated in accordance with the following hypothesis:\mdash ; Calling the ends of the rod A and , a weak current makes , say , a north ) and on breaking the current A is left with a quuntity , , of northern arity .
smaller reverse current shakes out only so much of this residual magnetism , quantity , , in , which exists alongside of a larger quantity , , of southern polarity .
The end A thus appears a south pole of strength .
On breaking this second current , A is left with a quantity , less than of residual southern netism , and a quantity , probably less than , of residual northern netism .
Whether A appears a south or a north pole depends on whether or is the greater .
On applying pressure , a considerably greater proportion of is out of and the polarity of the end A may thus change sigll .
Similar phenomena , proceeding doubtless from the same cause , are described by WIEt)EMANN as occurring in iron .
They fit in well enough either with WEBER'S theory or with the view that the deep-seated netic molecules are less affect by weak fields than are the molecules .
ON THE MAGNETISATION OF COBALT .
The total removal of this disturbing agency would require the complete neti sation of the rod .
The method of demagnetisation employed at first consisted of applying a few reverse currents gradually intensity .
The left in the rod seemed infinitesimal , but the nnsatisfSctory character of the result will be easily seen on reference to Tables I. and II .
The method finally adopted was to expose the rod to a succession of diminishing reverse currents\mdash ; the first current in case of weak field considerably exceedin that in existence during pressure cycles\mdash ; and then to tap it vigorously .
The first attempt was by no means always successful , and the loss of time was a distinct objection to the method .
It is , of course , impossible to be sure that the rod was ever demngnetised , in the sense of being restorod to its condition prior to its first tion , but it was , all events , reduced to such a condition that successive experiments made with the sante field showed an excellent agreement .
In .
fields tisation appeared to be , at for many purposes , of very little importance .
The second disturbing agency , the heating of the ence of the rod , by means of current , was troublesome only in fields over 400 C.G.S. units .
The heating of the coil wires increases their resistance , and so tends to diminish the strength of the field ; while the heating of the rod increases its permeability , as been shown by Professor ROWLAND For a rise of temperature of C. he found the maximum value of the coefficient of induced magnetisation to increase by about 70 per cent. , and to appear in a distinctly lower field .
This explains , in some of my stronger fields , the scale reading kept aJtering in a direction indicating a ssive increase in the rod 's netisation .
This did ) of course , annul the cyclic ) angss accompanying cyclic changes of pressure , but it rondered impossible any very great accuracy in the determination of their magnitudes .
The determination of the effect of the first application of pressure was found particularly difficult .
Lest it should be supposed that the cyelic changes in the netisation m in reality be due to the heating and cooling of the rod produced by the application and removal of pressure , I would here point out that the cyclic changes of magnetisation alter in sign in a comparatively low field , while the increase in the perme bility accomp nying heating was observed by ROWLAND in fields of all strengths , from 50 to 1470 C.G. .
units .
First Series of )'iments .
S24 .
With the new apparatus four principal series of experiments were performed .
first series took place in December , 1888 , and the results given in Tables I. and .
The general order of conducting these experiments as follows:\mdash ; The rod , after being demagnetised by reverse currents alone without was removed to a distance .
The number of storage cells and the bridge resistan ce were * ' Phil. Mag 4th series , , 1874 , .
C. ON OF then adjusted to give the field desired .
When the ammeter and the scale became steady , the rod was brongbt up and pushed rapidly into the coil , and the secured .
The difference between the scale readings when stationary , before and after the introduction of the rod , supplied the data for culating t initial value of induced isation .
A con siderable number of cycles of pressure " " on\ldquo ; and " " off\ldquo ; were then performed , the scale , when become stationary , being usually taken after each " " on\ldquo ; and " " off\ldquo ; Sometimes when readings had been taken of the effects of the first four or five pressure cycles , some five or six cycles were applied in rapid succession the taking of once more esumed .
When a sufficient number of readings been taken for the ation of the effect of a pressure cycle , the circuit suddenly broken , the rod always free from pressure during the break .
When ) scale r had become stationary after the break and had been observed , some six pressure cycles were applied , readings being taken after each " " on " " and off\ldquo ; A series of six pressure cycles were then applied whose effects were not observed .
These were followed by some six or seven more ) ressure cycles whose effects were rved .
Finally the rod was removed to a distance and the magnetometer taken .
The difference between the scale reading taken immediately after the break of the current and that taken after the removal of the rod supplied the data for calculating the initial value of the residual magnetisation .
TABLE I. * The values employed herc and are derived from subsequent tables .
ON THE MAGNETISATTON OF COBALT .
\ldquo ; Cyclic \ldquo ; off ' S In : Table I. is the strength of the field calculated as is explained S19 , is the intensity and the coefficient of in:duced magnetisation in the rod previous to the application pf any pressures ; while is :he and the coefficient of induced magnetisation in the rod free from pressure after all the cycles have been applied .
Thus is the increase in the magnetisation taki , ng ce during the appiication of the pressure cycles .
Their application o , ccupied a considerable time , during which absolute steadiness in the current could hardly be expected ; and so is obably in no case an absolutely exact measure of the increase in the magnetisation due to the pressure cycles .
By first " " on ' ?
is meant the increase in the mngnetisation caused by the first application of pressure .
Cyclic " " on\ldquo ; \mdash ; " " off\ldquo ; gives the average oraic ecess of the existing ) pressure is " " on\ldquo ; over ttJat existing when pressure is off As already explained the is the gebraic excess of the first " " on\ldquo ; over cyclic " " on\ldquo ; \mdash ; " " off The second last column gives the ratio of the to the intensity of the magnetisation prior to pressure , while the last column gives the ratio of the cyclic " " on\ldquo ; \mdash ; " " off\ldquo ; to the intensity of gnetisation in the cyclic state .
Thus the figures in these columns may fairly be regarded as measuring the importance in the different fields of shock-effect and the cyclic effect respectively .
In weakest field the last three havs no entries , because the experiments determining the cyclic effect of pressure failed to indicate a clear esult .
The entries in the two last fields indicate , as they do elsewhere , that no observations of the corresponding quantities were obtained .
In these fields when the compensating coil exactly balanced the coil current so that in absence of the rod the spot was at the centre of the scale , the introduction of the rod drove the off the scale .
The compensating coil was thus moved so as to neutralise part of the .
CHRE ON ECTS OF PRESSU R ( , and so bring the spot centre of the scale .
In the absence of the ) now off or neal .
lie off the scale , so that accurate measures of the induced gnetisation were impossible .
The effect of the changed position of the compensating coil was allowed for it ] reducing the observations .
A similar nse of the compensating coil was subsequently in some of the fields .
It had the disadvantage of making the scale reading more disturbed by slight irregularities in the current .
In these strong fields , however , the cyclic changes were so small compared to bhe total induced magnetisation that the alternative of increasing the distances of the coils from the magnetometer possessed greater ' In Table II .
, is the field prior to the break of the current .
is the residual netisation prior to ) ] ication of any pressure , that finally existing fter v-hat was approximately a constant number of pressure cycles .
minus sign r as elsewhere , a diminution in the tion , and when placed a the head of a column it applies to all the entries in that polumn .
The is obtained by subtr cting from the algebraic value of the total effect of the first pressure the value of the cyclic " " on\ldquo ; \mdash ; " " off The second last column gives the ratio of the shock-effect to the intensity of the residual magnetisation prior to pressure , while the last column gives the ratio of the cyclic " " on " " \mdash ; " " off\ldquo ; to the intensity of ) residual magnetisation in the cyclic state .
S27 .
In both the bless I. and II .
the letter signifies that a certain end of rod , which we shall call , was a north pole ; ) it was a south pole .
The numbers in brackets in the first columns give the order in which the experiments took place .
It was intended to apply as nearly as possible the same degree of snre in every single case .
Still , as the pressure was applied by hand , a certain amount of irregularity was bound to exist .
The smalJness in the variations shown by the individual observations of the cyclic effect in each separate field , and the smoothness of the several curves obtained prove that throughout any single series of experiments a pretty uniform must have been maintained .
As a considerable interval of elapsed between some of the successive series of experiments , notably those made in nber and January , the standard probably varied somewhat from one series to nother ; fig. 7 , at all events suggests that in February the standard pressure was somewhat less than in December . .
The results of Table I. and II .
will be presentiy discussed along with the corresponding results from subsequent tables , but certain peculiarities illustrative of the first difficulty stated in S22 , claim a special attention : A glance at the values of and in Table I. , shows that the supposed demagnetisation the strong field ( 7 ) had not sufficed to remove all its effects .
comparison of the fields ( 14 ) and ( 15 ) , or of ( 12 ) and , shows a much smaller susceptibility for currents making A a south pole , than for those making it as in ( 7 ) north pole .
From the results subscquently obtained with a more perfect system of ON TH MAGNLTISATION OF COBALT .
.30- 1 demagnetisation , and ntained iu Tables IX .
and X. , it ) that in those of the eight weakest fields of Table I. , in which A was a north pole : the values obtained , and the cyclic effect are all undoubtedly somewhat too great .
The effects of the residual trity a still more decidedly in II .
in the case of field ( 14 ) .
Here , what must be arded as an abnormally large amount of residual magnetism , was inally present .
very trace of this was , howevel remoyed by a single ssure , and with subsequent pressures gnetism of opposite sign became apparent .
An evenl more striking case , that of field ( 13 ) , does llot appear in Table II .
because no absolute measures were taken .
In case , the residual netism on the break even of the current was of opposite sign to the induced , and it increased ou the application of pressure .
Second Series of iments .
In the next principal series of experiments\mdash ; taken in January , 1889\mdash ; the method of taking the obseryations was much the same as in the first series , but the distances between pieces of the apparatus were varied .
Thus the distance of the centre of the rod from the netometer needle was cm .
, 50 cm .
, or ouly cm .
, according as the quantity under examination was the intensity of the induced netisation , the effect of pressure on the induced magnetisation , or the residual magnetisation .
The compensating coil and the scale remained screwed to the beam carrying the upparatus , while the three of the magnetising coil and the corresponding positions of the netometer were indicated by pencil llarks on the beam .
There was no shaking of the rod , variation in the current caused by the movements of the coil and magnetometer .
The fields in this series of periments varied from to 725 C.G.S. units .
Preliminary IJents in the weakest of these fields , showed that the rod had no appreciable influence on the magnitude of the cyclic effect of pressul.e .
here was thus in none of the recorded experiments , as given in .
and IV .
, any attempt at tion ; but the weaker fields were taken first so as to ayoi the .
influence of any elfects that might originate with currents .
ON MAGNETISATION OF COBALT .
during which an increase of units of netisation appeared , the current was broken , and after a short interval -made , before the rod had cooled much .
a second almost equally numerous series of pressure cycles an increase of only unit of magnetisation.appeared .
At the commencement of the second of the experiment most of the coil was doubtless colder than the contained rod , so that the temperature of the rod would not alter much for some time after the re-make of the current .
This last experiment indicates , I think conclusively , that in fields of this strength the permanent change in the magnetisation brought about pressure cycles must be small , whatever its sign may be .
The values given for the First " " on\ldquo ; in the strongest fields cannot claim great accuracy .
The rod began to get heated before the oscillations of the magnetometer needle following the introduction of the rod into the coil had sufficiently subsided to permit of a reading being taken .
The progressive increase of magnetisation was , in fact , going on most apidly when the first pressure was applied .
There had thus occurred a considerable progressive change before the next scale-reading could be taken , so that it was very difficult to deduce the true effect of the first pressure .
The values ined in these high fields for the intensity of the induced tion must also have been to some extent unduly raised by heating , so thab the curve of fig. 1 is , in its higher portions , not so flat as it ought to be .
S31 .
Table IV .
giyes merely certain average results , calculated from the large number of obseryations actual ] made on the residual magnetisation .
On comparing the results after the conclusion of the experiments , it became apparent that the magnitudes of the several phenomena varied so little with the strength of the preexisting field within the limits of the observations , that trustworthy deductions as to the exact modes of their variation were rendered impossible by small variations in the distances of the pieces of the pparatus which the method of experiment was sure to introduce .
The surements of the effects of pressure are , in addition , exposed to possible irregularities in the magnitude of the pressure , but this ought not affect sensibly the average results recorded in the Table .
Third Series of Experiments .
S82 .
In the next series of experiments , extending from January to February , the residual magnetisation alone was under investigation .
Every single piece of apparatus was bolted or screwed to the supporting beam , so as to prevent any relative movement .
The magnetometer needle was distant 82 cm .
from the scale and 38 cm .
from the middle of the rod .
First of all a set of observations were taken of the initial amount of the residual llagnetisation as the strength of the pre-existing field was raised by steps from to C.G.S. units .
The rod was apparently very nearly demagnetised by reversed currents previous to the first experiment , but no subsequent demagnetisation was performed .
No presstlre or tapping was applied to the rod throughout these observations .
The current hvays in the direction making the end A the north pole .
The results , as recorded in Table V. , are deduced from the scale taken ( 1 ) with the rod in its ) ] inside the coil immediately after the break of the curl.ent , ( 2 ) with rod removed to a distance .
In general , two observations were taken for each strength of field .
The mean of the two observations is the of the able .
By is meant the of the field due to the current just broken , and by the .
ratio of to .
The mode of variation of with is shown in cve ; while the mode of variation of , which is subsequently termed the , is shown in curve Il .
Table V. SS3 .
In continuation of these experiments a much more elaborate set of observations were taken , with the object of whether the amount of the residual magnetisation or its properties depended on the treatment of the rod during the flow or break of the current .
In one set of experiments , spoken of in future as the type , no pressure at all was applied the flow of the current .
In a second set , type , six cycles of pressure " " on\ldquo ; and " " off\ldquo ; were applied , the current being broken when pressure was " " \ldquo ; In a third set , the type , a pressure was applied after six pressure cycles , the current being broken when pressure was " " on In these , as in all the other experiments , the rod was introduced into the coil after the scale reading showed that the current had become steady .
The experiments of the several types were not conducted separately , but on the following plan .
Starting with a ain number of stol.age cells and a certain resistance in circuit , complete observations were made for the corresponding field with each of the three types in the order , say .
with a greater number of cells , or a reduced resistance , observations were taken for the next higher field , in the order and so on , in cyclic order .
Care in fact was taken that the experiments of any given type should not unduly often be either the first experiment of the day or the first ON THE MAGNETISATION OF COBALT .
experiment with a given strength of field .
This variation the order was adopted lest during the application for several hours of a succession of pressure cycles , interrupted only by the demagnetisations , the rod become less responsive to pressure , developing a sensible amount of what may be termed .
Under the actual conditions , if this did happen , it would not affect the experiments of one type more than those of another .
The strength of the pre-existing fields was raised step by step from 1 to 400 C.G.S. units , until fields of 350 units , or were reached , the rod was netised , before each single experiment , by reversed currents and vigorous .
Tn the stronger fields the rod was mel'ely exposed to a preliminary reverse current , of the same strength as that about to be used in the experiment .
During the actual observations the end A was invariably the north pole .
In each experiment of each of the three types , with the exception of some of the weakest fields , the following operations were conducted .
When the spot on the had become stationary , after the break of the current , a reading was taken .
The pin was then withdrawn and the rod carefully removed , being kept at right angles to the magnetic meridian .
The scale reading when stationary been obseryed , the rod was gently restored to its place and the pin secured .
With a little practice this operation was accomplished without shaking out any sensible a1nount of netism , as the coincidence of the scale readings before the rod 's removal and after its restoration sufficiently testified .
Pressure was then applied once and removed in ths experiments of types and , being simply removed in those of type , and the corresponding scale readings observed .
Twelve pressures cycles were then applied , the effects of the last six only being observed .
The rod was then removed and the constancy of the magnetometer zero tested .
The various observations obviously suppIied sufficient data for calculating the residual magneti , ation , the } " " on\ldquo ; or first " " off\ldquo ; of pressure , the cyclic effect of pressure , and the final esidual magnetisation , i.e. , the residual a definite number\mdash ; 13 for types and M\mdash ; of pressure cycles .
S34 .
The results of the experiments of type are given in Table .
All the have been already explained , except next " " off This means the change in the magnetisation accompanying the removal of the first pressure applied after the break of the current .
In the two weakest fields the amount of the initial residual magnetisation alone observed , and in the three next weakest fields the effects of only one pressure cycle were taken .
Thus no data existed for calculating the cyclic effect in these fields ; and the first three results given in the last column are unduly large , because the values there assigned to are the existing after the application of only a single pressure cycle .
In the fields oyer 90 C.G.S. units the changes in the properties of the residual netism accompanying the in the fields are so small that the small irregularities in the magnitudes of the pressures applied become important .
Thus the ON THE MAGNETISATION COBALT .
3 Cl S36 .
In Table IX .
the first six have been already explained .
The letter ' after the numerical measure of the field the first column nifies that the rod was under pressure previous to and during its introduction into the coil .
In such a case the entries under and refer to the tion existing before this pressure was removed , and so are on a different footing from the other entries under these headings .
The entries under the other headings also in such a case doubtless indirectly affected , but to a much smaller degree .
In the last four headings snflix indicates which end of the rod was the north pole when the magnetisation in question was measured .
The results under these headings are deduced from the readings obtained by relnoving and replacing the rod to the application of the pressure cycles , during which application it be remembered was always the north pole .
The colnmn gives the excess of the netisation existing when the rod was under pressure when reintroduced into the coil over that existing when it free from pressure when reintroduced , A being in each the north pole of the reintroduced rod .
The column differs from the seventh only in that the end was in each case the north pole of the reintroduced rod .
ninth colulnn gives the excess of the magnetisation found when A was the north pole of the reintroduced over that found when was the north pole , the rod in each of its reintroductions being from pressure .
The tenth ) differs the ninth only in that the rod during of its trod uctions was under pressure .
S37 .
In Table X. " " on ' ' is , as previously , the total in the tion\mdash ; cyclic and non-cyclic\mdash ; due to the first.pplication of pressure .
the effect when the rod , mmediately after netised , has been exposed first time to the field in question .
gives the effect of the first pressure after the of the rod in a state free from pressure , and with the elld A a north pole .
In the case of the results under , there have intervened tion a complete series of pressure cycles applied with A a north pole , nd at least one emoval and reinbroduction of the rod .
In some cases there were several removals and reintroductions , and the mean of the is recorded .
Under we get the when , on its reintroduction , the rod has for its north pole .
In this case it must be borne in mind that the pressure cycles applied since the ) demagnetisation occurred when the opposite end A was the north .
pole .
off\ldquo ; is the total effect of the removal cf a pressure existing previous to and during the introduction of the rod into the coil .
refers to the case when tisation has immediately preceded the observation , and to the cases when a complete series of pressure cycles , with A the north pole , and at least one reintroduction have interyened since demagnetisation .
The distinction and is the same as in the case of the First " " on Cyclic\ldquo ; on\ldquo ; \mdash ; ' ' off\ldquo ; has its usual meaning .
Under the next two the non-cyclic portions of the effects of the first application of pressure and of .
C. CHREE ON THE OF removal a pre-existing pressure .
and refer to the results recorded in the first two columns under " " " " and " " off The last heading has been already explained .
As usual . .
indicates that no observations were taken .
When no algebraic is attached to a figure understood . .
To a c.asual observer the fluctuations that appear in the results of the foregoing tables seem excessive , It should , however .
be borne in mind that most of the pressure effects ctre very small compaled to the total amount of the in uced \mdash ; sometimes as little as one-thousandth part .
the small irregularities ) the gnetising current which produced by the slightest want of in the cells , or by the vibrations communicated to the resistance wires from shakin , of floor , produce commensurable effects .
irregularities were frequently through a continual quivering motion of the spot on the scale , while the ammeter reading steady enough .
Of course , when the unsteadivery marked , observations were suspended , but comparatively little of the done vould have been accomplished in the time if observations had been taken only when absol steadiness .
A separate discussion of the results of each table would involve a deal of repetition .
It has thus appeared best , as a rule , to embody the most trustworthy results in curves , and to discuss particular points in connection with the features of the individual curves .
In every curve the horizontal coordinate gives the strength of the field , , calculated after the manner explained in S 19 .
In the case of the residual magnetism , is of course the strength of the pre-existing field .
In fig. 1 ( Plate 15 ) , the ordinates give the initial magnetisation when the rod is intro- dnced into ) magnetising coil free from pressure .
In the ordinates give the coefficient of induced magnetisation under the same conditions .
The first uions of both curves , in which the individual observations are indicated by dots , are based ) Table IX .
, while the second portions , in which the individual obseryations are cated by crosses are based on Table III .
These curves are of the same general character as those obtained by many observers for iron .
In fields below 8 or 9 C. .
units , the magnetisation increases comslowly as the strength of the field is raised .
There then ensues , as is most clearly shown by the steepness of the commencing portion of the curve of fig. 3 , much more apid increase of magIJetisation .
The rate of increase attains a as shown in fig. 3 , in a field of about 35 C.G.S. units ; which , accordingly , is the for the specimen .
The rate of increase of the netisation then falls off , but at first in a comparatively gradual manner .
There is thus no very clear indication of an approach to " " saturation\ldquo ; until the strength of the field approaches 200 C.G.S. nits .
Even in the strongest experimental fields the rate of increase is by no means infinitesimal , though , as already stated , this is probably in part accounted for by the heating of the rod .
S40 .
The smallness of the scale of fig. 1 , does not allow the effect on ON MAGNETISATION OF COBALT .
36,3 the induced ) to be shown , but the enlarged scale of fig. 2 shows the effect in fields below 30 C.G.S. units , where it is ofJnost importance .
The thick line , with the individual ations indicated by dots , is merely iniiial portion of fig. 1 on an enlarged scale .
The dotted line , with the individnal observations indicated by circles , gives the duced magnetisation after the application of the pressure cycles .
The curves , and , are based on the sari ) ) eriments , and the difference between their ordinates answers to the values in the colunlIl of Table IX .
in those experiments in which the rod was originally free from pressure .
The third curve , with the individual observations indicated by crosses , gives the induced netisation prior to the application of pressure cycles , in these experiments denoted by a in Table IX .
, in which the rod was nder pressure when introduced into the coil .
In fig. 4 the ordinates give the coeflicients of netic induction in the three cases of fig. 2 .
The data are taken from the -ame table , and the letters , \amp ; c. , have the tions .
The difference between the curves and the very effect of the pressure cycles in increasing the magnetisation in the weaker fields .
It should remembered that between the corresponding pairs of observations on which these curves are based there intervened no break of the , nor any change in the resistance , or in the position of the ammeter .
Thus any error in the zero or orientation of the ammeter would affect each curve alike , and leave the difference of the ordinates pract unaffected .
Variations , it is true , in the strength of the current during the application of pressure cycles , would affect curve without cting curve .
Such val.iations , however , cculd escape notice in the weaker fields only when very small , because the ammeter was then in its most itive position .
The comparison of the curves wibh the others is not so satisfactory .
During the time of most of the experiments the ammeter was being used for other purposes , and so had be set up afresh every other day .
Thus , though its position was carefully adjusted , small differences were certain to occur .
For these and other reasons it be unsafe to draw any conclusions from ] crossing of the curves in fig. 4 .
It may , however , be regarded as pertectly certain in the weaker fields the curve lies between the curves and , and that within the limits of , 4 the curve lies very distinctly above the curve The form of all these curves of fig. 4 , under all conditions as to the coefficient of magnetic induction becomes extremely small in very weak fields .
There may , however , be a point in one or all of the curves in fields -er those experimented S 41 .
In curve of fig. 5 the ordinates give the total .
in the produced by the first Qsure in those experiments of ] IX .
in which the free from pressure when introduced into the coil .
The chance in is measured with the pressure ' on and so is really the algebraic uf th MR. C. CHREL ON THh ' OF PRESSURE effects .
The individual tions are indicated by dots , whose disbances the curVe are no ca , se The curve shows the total eflect of ] first pressure attains a in a field of about 45 C.G. units\mdash ; which somewhat exceeds the Wendethen diminishes somewhat rapidly as the strength of the field is raised .
An unnistakeable critical field , where the effect vanishes , appears at about 160 C.G.S. units , and in fields the total effect of the first pressure is a diminution of etisation .
From the form of the curve it appear that the diminution does not increase indefinitely as the strength of the field is raised .
But the experiments leave it uncertain whether this diminution of magnetisa.tion attains a maximum in a field of about .
C.G.S. , or whether it continually approaches an asymptotic value . .
In curve of fig. 5 , the ordinates give the loss in the induced magnetisatiou accompanying the removal of the rod from the coil , after the application of the pressure cycles , and its reintroduction .
The curve is based on the column of Table IX .
The ordinate corresponds more or less closely to the non-cyclic of the sum of the pressure cycles applied during the flow of the current prior to the rod 's removal .
There are two reasons however , why it CiIlnot be regarded an act measure of this effect , more especially in the weaker fields .
In the first place , even when no pressures are applied it is well at tlJe magnetisation of iron is exactly the same on the second exposure to a certain field as it is on the first .
Some occasional observations on the cobalt itself gave a somewhat greater magnetisation on a second exposure thau on the first .
The difference was , however , unimportant , and in fields over the Wendepunkt it was , if existent , extremely small .
the second place part of the effect of the pressure cycles unquestionably survives a removal of the rod from the netic field .
As will be seen from , this residual effect is of considerable portance in fields below 30 units .
ng to both these , causes ffect of the pressure cycles in fields below 30 or 40 units must be decidedly than the curve would indicat , .
No tions exist to show whether or not the curve eventually crosses the axis of abscissae .
Not improbably this would be a very difficult point to settle owing to th heating of the rod .
reason for drawing a curve for in preference to one for , is that a few minutes elaps.ed .
ween the taking the readings giving and , while the reading which gives might have been taken over an hour previously .
Thus under ordinary circumstanc.es the riation , of the current could but be utterly significant in the one instance , whereas in the other it occasionally reached a quantity .
In Table .
omitting of course the fields marked , the values of and a very fair agreement .
In fields below the Wendepunkt the former quantity , as the above reasoning gests , distinctly the larger ; but in stronger ON COBALT .
3 b5 fields the one is sometimes ) and bometimes the other .
In no case a tive value ined for either quantity , thouoho ' in fields over 200 C.G.S. units , bot , become unquestionably very small .
The only safe conclusion appears to that if a field existed for the noncyclic or shock-eHect of the cycles it exceeded 275 C.G.S. units .
CoIlsidering , however , that the critical for cyclic effect of the pressure cycles is only about 120 C.G.S. units , this seems important result . .
In the curve of fig. 6 , which is based on Table X. , the ordinates the ratio of the increase in the induced tion 1 ) roduced by the non-cyclic part of the first pressure relative to the amount of tlJe induced fnetisation existing .
to the pressure .
The ordinates may thus be held to measure the relative importance in fields of various strengths of the non-cyclic portion of the effect of the hrst pressure .
The curve shows in a manner how the relative importance of the -cyclic portion of the eHect of the first pressure diminishes as the strengtb of the field increases .
In the weakest field employed , viz. , C.G.S. units , the first pressure permanently increases the gnetisation 1 one half its original value , whereas in a field of 160 C.G.S. the increase is certainly less than one in two hundred .
In stronger fields , as appears from the ninth and tenth columns of Table X. , this shock-effect becomes extremely all , and continues to be so within the .
of the experi 1nents .
The experiments cannot be said to settle conclusivel.y the even of the the evidence is decidedly in favour of its remaining positive in fields of at least C.G.S. units .
It will , in , be observed that in the tenth column of Table X. , a decrease of netisation w in 110 single obtained .
Now in the higher fields the results in this column are calculated the of two or three tions , whereas those in the column swer , of course , to only one L'ffects of the of .
It might appear at first sight that the increase in the invariably to accompany the application of the sum of pressure cycles in fields up to 270 C.G.S. units is in itself sufiicient proof that the non-cyclic effec , of the first pressure must be within the same limits an increase of neti sation .
The increase , however , in the former case be due to the remova ] not the applications of pressure .
The removal of pressure from a rod occasions relative displacements of its parts to pretty much the same extent as does the application of pressure , and so it too be expected to have some permanent influence on the nctisation .
effect was ctually found to exist , and its magnitude relative to the original netisation 1 .366 MR. C. CHREE ON THE ECTS OF given in the eleventh and twelfth columns of Table X. Comparing these with the two previous , it will be seen that in weak fields the non-cyclic effect of the moval of the original , though much less importarlt than the effect of the application of the first pressure , is by no a negligible quantity .
It may , however , be easily overlooked , because in the weaker fields , where its importance is greatest , it is in the opposite direction to the cyclic effect of the removal of pressure .
Also in fields between 25 and 95 C.G.S. units , as appears from the hirteenth column of Table X. , the cyclic part of the is numerically the larger , and so between these limits the remo of an original pressure appears to be accompanied by a diminution in the magnetisation .
In fields over 120 C.G.S. units the non-cyclic effect has the same sign as the cyclic , and so the increase in magnetisation accom- panying the removal of all original pressure is fairly conspicuous .
of Pressure iloe etisalion . .
In fig. 7 the ordinates the cyclic change in the induced magnetisation the pressure cycles .
The thick line , in which individual observations are indicated by dots , is based on Table X. , the broken line with individual observations irldicated -by crosses , on I. , and the dotted line with individual observations indicated by circles , on Table III .
These curves agree in showing a critical or VILLARI field of about 120 C.G.S. units .
The is greatest or least when the rod is under pressure according as the strength of the field is less or greater than that of the critical field .
It will be seen from fig. 1 that the tion of the rod in the critical field is about 520 units .
ding to both the curves , and , the absolute magnitude of the cyclic attains a nlaximum in a field of about 35 C.G.S. units .
This , it will be remembered , is the field found for the Wendepunkt in S39 .
In general form the two curves and l ) could hardly agree better than they do , the difference in the absolute lengths of their ordinates might wel ] be due to a ifference between the standard pressures adopted at the times .
Both curves , it will be noticed , pass through almost all their experimental .
Near the critical field all three curves would lie so close together that the experimental points of are there shown .
The highest { ield in Table IX .
is 351 C.G.S. units , and the form of the curve leaves it uncertain whether the cyclic effect attains an algebraic minimum in a field somewhat higher than this and then diminishes numerically , or whether it continually approaches an asymptotic value .
The furmer alternative is unquestionably supported by the form of the curve , which extends to a field of 725 C.G.S. units .
The deviations of the ) erimental points from this curve are , however , so large that it would be rash to attach much weight to its precise form .
Probably all we are entitled to infer is that , if an asymptotic value exists for the cyclic effect , its numerical value cannot much exceed that found in a field of 400 C.G.S. units ; whereas , if a second COI3 critical field exists , where the effect a second time , its vadue musb exceed 700 C.G.S. units .
It should be noticed the ordinates in fig. 7 are drawn on double ) of those in fig. 5 .
Thus , comparing in those ures the two curves , which are based on the series of nents , we see that , in fields below the ) oint of fig. 7 , the total effect of the first pressure is invariably very iderably larger ) the cyclic effect of pressure .
The maximum ined by the nler effect is about thrice that attained by the latter . .
In fig. 8 the ordinates give the ratio of the cyclic increase in neti ation accompanying pressure " " on\ldquo ; to amount of the induced etisation existing when the cyclic state is ched .
They thus show proportion of the tion takes part in the cyclic change .
The curves and are derived from the same as the corresponding curves and of fig. 7 .
The form of , near the vertical axis , how much the elative importance of the cyclic effect increases as the strength of the field is reduced .
In comparing the relative magnitude of the cyclic effect and of the shock effect of the first pressure , it should be noticed that the ordinates of drawn on scale ten times that on which the ordinates of fig. 6 are drawn .
Thus , the weakest fields , shock effect is fully ten times the cyclic effect .
Residual etisation .
S4 In figs. 9 and 10 ( Plate 16 ) the ordinates give the amount of the gnetisation existing immediately after the break of the current .
The cu rves are shown in botb figures , but on a different scale .
The curve is based on Table V. The rod wss here demagnetised before exposure to the weakest field , but not subsequently , and remained entirely free pressure .
The curve is based on Table VI .
The rod was here demaguetised by reverse currents and vigorous tapping before each introduction into the coil , but no pressures were applied while it was under the influence of the magnetising currents .
The curves and are based on Tables VII .
and VIII .
respectively .
In both cases the rod was demagnetised , as in case of curve , but subjected to six pressure cycles while under the influence of the magnetising current .
In the case of pressure was " " off\ldquo ; when the current broken , whereas in the case of the pressure was " " on The individual vations are indicated by dots for the curves , by circles for the curve , and by crosses for the curves and .
With the scale of the curve would be indistinguishable from the curve in weak fields , and from the curve in strong fields .
Effof Pressure th Amount of the ) S48 .
The curves and of agree in showing only a small nount of residual magnetisation on the break of weak fields ; but in fields C.G.S , uIlits .
C. ON OF the ordinates of are at least double those of difference is surprisingly great , considering the rity of the conditions under which the corresponding experiments conducted .
Ib , presumably , be mainly attributed to the residual effect of the shocks ) loyed in completing the demagnetisation in the experiments on ictJ the curve l ) is based .
In weak fields , as the difference between the curves and on the one hand , and the curves and the other abundantly , the effect of pressures applied during the flow of current in the residual ' is simply enlorlnous .
It priori probable that the application of shocks immediately before the starting of the current should have some appreciable tendency in the same direction .
Fortunatcly I happened to observe the induced as well as ) residual tion for the field C.G.S. units in the experiments on which curve is based .
The value of the induced etisa t was 8 which is almost exactly the value obtained by interpolation for the same field from Table IX .
Consequently , in this field , and so presumably in other weak fields , the ence between the curves and cannot be ttributed , in any important degree , to am increase in the induced netisation brought about by the process of netisation employed in the case of curve As the strongest field in Table V. is only 79 C.G.S. units , no data exist for drawing curve in higher fields .
It is only , however , in fields below 45 or 50 C.G.S. nits that there is any clear difference between the results of Tables V. and VI .
We thus probably justified in conclu Iing that the process of demagnetisation has an appreciable effect on the amount of the residual magnetisation only in fields below .
units .
As already stated , the results of Table V. are each the mean of two observations taken in close succession with the same current .
Ahmost invariably the second ervation showed a slightly larger amount of residual magn tisation than the first .
This property has been noticed by several observers in iron .
The difference between the curves and of fig. 10 , which are based on experiments in which he same process of demagnetisation was applied , shows how effective pressure cycles are in increasing the amount of the residual magnetisation in weak .
In stronger fields thaIl those of the cnrves actually approach one another , and in fields exceeding 120 or 130 C.G.S. units they cannot with certainty be said to differ .
Thus , in fields below 120 or 130 C.G.S. units , the application of pressure cycles during the flow of the current increases the amount of the residual magnetisation , but in stronger fields no effect can with certainty be said to exist .
In fields below.30 , or at all events 20 C.G.S. units , the difference between the curves and is , though small , perfectly clear and incontestable .
The existence of pressure during the break of the current is the ] point in which the experiments on which curve is based differed from those on which curve is based .
As the effect of pressure on\ldquo ; during the flow of current is an increase of induced magnetisation ON SATION OF COBA r in fields below 120 C.G.S. units , an excess in the ordinates of curve over those of curve in weak fields might reasonably have been expected .
The difference , however , as will more fully be seen presently , is somewhat greater than might have been anticipated .
The crossing of the curves in a field of from 35 to 40 C.G.S. units , and thus far below the VILLARI field , appears a somewhat striking fact .
In fields over 120 C.G.S. units the curve , if shown in fig. 9 , could not be distinguished curve , and the difference between its ordinates and those of curve seems then truly remarkable .
In fields over 150 C.G.S. units this difference remains nearly constant , and amounts to about 4 per cent. of the ordinates of .
If it be remembered that .
to the last column of Table X. the percentage of the magnetisation , which is cyclic for the same pressure " " on\ldquo ; and ' ' off is in fields between 150 and 400 C.G.S. units never in excess of 4 , the significance of this result will be more fully understood .
S49 .
ratio of the residual netisation existing in rod immediately after the break of the current to the numerical value of the strength of the pre-existing field will , for shortness , be here termed the snsceptibility .
It is denoted by in Tables V. to VIII .
, and its values are represented graphically by the ordinates of the curves of fig. 11 .
In this and in the subsequent figures the letters denote curves obtained from the same series of nents as are the respective curves of fig. 10 .
In fig. 11 and the subsequent figures , however , the vidual observations are indicated by crosses for curves and , by dots for curve and by circles for curve In the absence of pressure , as shown by curves and , the residual susceptibility is very ] in weak fields , but increases rapidly , attaining a maximum in a field a little over 20 C.G.S. units .
This , it will be noticed , is not much over half the field which gives the Wendepunkt for the induced magnetisation .
greatest value actually observed for in the case of curve , was only .
After the maximum is passed there is a continuous , but gradual , minution in the residual susceptibility .
The difference between the commencing portions of curves and shows , even more clearly than in fig. 10 , the large effect in weak fields of the process of demagnetisation .
In fields over 50 C.G.S. units the two curves practically coincide , and are drawn as one .
The crossing points for the various curves in are , of course , the same as in figs. 9 and 10 .
Within the range of the figure the curve lies distinctly above the .
In fields oyer 55 C.G.S. units , the curve is not drawn , as it could hardly within the remaining limits of the figure be distinguished from curve with the strongest experimental , the ordinates of the curves and increase gradually as the strength of the field is reduced , and show maxima values in somewhat lower fields than do the curves and .
As the strength of the fields is further reduced , the ordinates of and pass distinct minima values , and then increase rapidly as the strength of the fields oaches the lowest experimental value .
value of cannot , of course , exceed that of ; so either the ordinates of and must be rapidly approaching second maxima , or else the ordinates the curves and of fig. 4 must possess minima values in fields lower than were nented on .
$ 50 .
In comparing the amounts of the residual and induced magnetisations , the employment of the term retentiveness in an exact way will be found serviceable .
It is here used in accordance with the following definition:\mdash ; is the ratio of the to the induced magnetisation , both quantitics being with the rod exposed to one the same stress , the stress remaining during the interval that between the ?
neasut ements .
In fig. 12 the ordinates give the retentiveness of the rod in the cases illustrated by the curves .
As the residual magnetisation alone was observed in the experi .
ments recorded in Tables VI .
, VII .
, and VIII .
, the necessary values of the induced netisation.were derived by interpolation from Tables IX .
and X. In curve the induced magnetisation is that existing prior to the application of pressure , or is the of .
Table IX .
In curve the induced magnetisation is ; for after six pressure cycles the the netisation accompanying further pressure cycles is , for our present purpose , quite negligible .
In the experiments answering to curve , the rod , after experiencing six pressure cycles , was under pressure when the current was broken .
Thus , for curve the induced magnetisation is got by adding to the algebraic value of the cyclic " " on\ldquo ; \mdash ; ' ' off\ldquo ; of Table X. The curve shows that , in the absence of pressure , the retentiveness attains a maximum in a field of about 15 C.G.S. units , a field somewhat below that at which the maximum residual susceptibility occurs in the corresponding curve of fig. 11 .
great similarity exists between curve and the curve for zero load in Professor EWING'S fig. 57 .
The principal difference is that the field at the maximum retentiveness appears is considerably higher in cobalt than in iron .
The curves and of fig. 12 bear a considerable resemblance to the corresponding curves of fig. 11 ; but there are no longer distinct maxima or minima .
Comparing these curves with curve we see that in the weakest experimental fields pressure cycles increase the retentiveness in the ratio of four or five to one .
The tendency to becorne tangential to the vertical axis shown by the and d\mdash ; which cannot , however , go on indefinitely as the fields are further reduced\mdash ; is not exhibited by any of the curves for the retentiveness of loaded iron-wire occurring in Professor EWING'S fig. .
Possibly if he had employed lower fields he might have found similar phenomena , so it would not be safe to assume that iron and cobalt actually differ in this particular .
The ordinates of curve being greater in weak fields than those of curve it follows that in fields below 30 C.G.S. units , where the curves cross , the retentiveness of the rod is greatest when it is under pressure while the current is broken .
In fields ON MAGNETISATION OF COBALT .
over 30 C.G.S. the retentivelless is least when the current is broken with pressure on and in fields over 50 units the difference is not inconsiderable .
In fields over the VILLARI point , of 120 C.G.S. units , the induced nelisation is greatcr for curve than for curve , so that in such fields the of the ordinates of curve over those of curve would , ively to their absolute magnitude , be somewbat less in the case of fig. 12 than in case of fig. 11 .
Similar considerations lead to conclusion that in fields over 130 C.G.S1 .
units the curve would be , if anything , slightly above the curve .
In other , in fields over 130 C.G.S. units the effect of pressure cycles is , if anything , slightly to diminish the retentiveness .
This effect is , howeve ] , insignific nt compared to that proceeding from the existence of pressure " " on\ldquo ; during the break of the current .
The curves and cross in a field of ) C.G.S. units .
The curve would then become the lower , but it would be for a time so close to that it is not attetnpted to show it separately in the The conclusions which the preceding considerations lead to are the In fields below 30 C.G.S. units the application of cycles of " " on\ldquo ; and " " oli during the flow of the current , and the existence of pressure " " on\ldquo ; during the break of the current , both tend to increase the retellliveness , the former agency being in fields below 10 C.G.S. units the more important .
In fields between 30 and 120 C.G.S. units , or thereby , the application of cycles of pressure still increases tlJe etentiveness , but the effect continually diminishes as the of the field is raised .
fields over 130 C.G.S. units up to at least 300 or 400 C.G.S. u1lits the effect of pressure cycles is very LQmall , but is , if anything , a diminution of the retentiveness .
In fields from 30 C.G.S. units up to at least 300 or 400 C.G.S. units the tence of pressure " " on\ldquo ; during the break of the current diminishes the retentiveness , and the effect is of the same order of lnagnitude as that due to the cause in fields below 30 C.G.S. units .
In fields between 30 C.G.S. units , thereby , the increase in the retentiveness occasioned by the pressure cycles exceeds the diminution occasioned by the existence of pressure " " on but in stronger fields the latter effect is the larger , and its superiority appears colltinually to increase with the strength of the field .
S51 .
Curve of fig. 12 is repeated in I. of fig. , in order that its form in fields over 75 C.G.S. may be seen .
There appears a continual but very slow diminution in the entiveness as the field is raised to 400 C.G.S. units .
Curve II .
of fig. 13 will be discussed presently .
S52 .
In fig. 14 the ordinates give the absolute amount of the residual tion which seems disappear with the first alteration of stress aftel ' the break of the current .
Comparing the curves and , in which the loss of magnetisation accolnpanies the application of pressure , with the curve , in which it accompanles removal of the pressure existing during the break of the current , we that the effect must be to a Alll , C. CHREI ON THE ] PRESSURL extent } ) endent of the precise character of the of stress , and would doubtless follow more or less any lnechanical agitation of the rod .
The curve is drawn through nearly all the experimental points , but in the stronger fields of , especially the latter , the experimental points are somewhat widely scattered .
These latter curves accordingly cannot be trusted in their minute details .
The curves are bu , sed on data in the fourth columns of Tables VI .
, VII .
, and VIII .
These tables give only certain verage results for fields over 90 C.G. .
units , but they indicate that the ordinates of and would attain maxima values in fields somewhere between 100 and 200 .
units .
Thereafter the ordinates of would apparently diminish the more rapidly of the two , so that in fields of 400 .
units the curve would have approached pretty close to the curye , which remains apparently nearly horizontal . .
The application of cycles of pressure causes cyclic changes in the magnetisation , whose character will ) esently be discussed .
In consequence of this the effects of the first change of stress on the residual magnetisation are only , in part , of a noncyclic character .
To get the exact quantity of magnetisation whose permanent disappearance is secul.ed by the first change of stress , the cyclic effect must be allowed for .
uppose , for instance , a field of 50 C.G. .
units is broken when pressure is " " off\ldquo ; The cyclic efiect is then diminution of magnetisation when pressure is " " on Consequently the loss of magnetisation accompanying the first application of pressure exceeds the tity whose permanent disappearance is secured by the amount which takes part in the cyclic change .
Subtracting from the total loss the numerical of the cyclic effect , we get what may be regarded as the true loss of residual magnetisation due to the first application of pressure .
This quantity is that here termed the shock-effect of the first pressure .
In the case of residual magnetism the cyclic effect is in general small compared to the shock-effect , so that a considerable error in the cyclic effect will seldom seriously affect the calculation of the shock-effect .
S54 .
In fig. 15 the ordinates represent the ratio of the shock-effect of the first alteration of pressure to the amount of the pre-existing residual magnetisation .
They may thus be regarded as measuring the efficacy of the first change of pressure in shaking out the residual magnetisation .
Unfortunately no readings were taken of the cyclic effects of pressure in the weaker fields in the series of experiments on which the curves are based .
Thus up to the points where the transverse lines are drawn , the ordinates represent the ratio of the total effect of the first change of pressure to the initial residual magnetisation .
Consequently , as will be seen presently , in the commencing portions of the curves the ordinates of and are probably underestimated , and those of on the other hand overestimated .
corrections could , however , hardly modify the relative positions of the curves .
The curves all show that the efficacy of the first change of pressure in out ON THE MAGNETISATION OF the residual magnetisation , as measured by the percentage rid of , continually ishes as the strength of the field is incl.eased .
According to the seventh of Tables VI .
and VII .
, and the sixth column of Table VIII .
, on which the curves are based , this diminution most probably goes on up to fields of 400 C.G.S. units at least .
The difference , however , between the percentages shaken out in the case of fields of 100 and 400 C.G.S. is under all conditions very small .
Comparing curves and , in figs. 14 and 15 , and the seventh columns o ables V and , we see that while in fields below C.G.S. units au absolutely amount of residual magnetisation is shaken out by the first pressure when pressure cycles have been applied during the flow of the current , still in these and in all higher fields within the experimental limits a greatel percentage is removed by the first pressure when pressures have been applied during the current .
We thus see that in fields below 100 C.G.S. units , pressure cycles during the flow of the current diminish simultaneously the percentage of the induced magnetisation got rid of by breaking the current , and the percentage of residual magnetisation got rid of by the subsequent application of a pressure .
In fields from 130 to at least 400 C.G.S. units pressure cycles during the flow of the current appear slightly to increase the ) ercentage of induced magnetisation got rid of by breaking the while simultaneously diminishing the percentage of residual magnetisation got rid of by the subsequent application of a pressure .
In these strong fields , however , both these effects are comparatively insignificant .
the experiments which the curves and are based the effects of removing the first pressure applied after the break of the current were ] observed .
The result was a loss of magnetisation , and its magnitude is recorded in fifth columns of Tables VI .
and VII .
In considering how much of this loss is nent , the cyclic effect of the change of pressure must be allowed .
This been done in calculating the eighth columns of Tables VI .
and , which give the relative importance of the first pressure and its removal in permanently reducing the magnetisation .
When no pressures have been applied dlJring the flow of the current we see from able V. thab , in fields between 18 and 75 C.G.S. units , the removal of the first pressure produces uniformly about one-ninth of the effect of its application .
In the higher fields the relative importance of the removal of pressure shows a distinct diminution .
Comparing ables V. and .
we see in fields below 50 C.G.S. units the of the removal of the first pressure aftel the break of the current relative to its application is largely increased by the application of pressure cycles during the flow of the current .
In fields over 50 C.G.S. units the effect of bhe removal bearF to that of the application of the first pl.essure in the of Table .
the apparently nearly constant ratio of 1 : 9 , which still somewhat exceeds the ratios found in the corresponding fields of Table ON OF fact that the importance of the emoval of the first pressure relative to its pplicatiou ieatest in weak fields must , I think , be connected with the fact that in veak fields , as shown by curves and , fig. 15 , a larger proportion of the residual magnetisation is shaken out by the removal of a pressure existing during the break of the current , by the pplication of equal pressure when the rod is free from pressure during the break of the current .
In the weak fields pressure cycles , which exist in the case of curves and , cause a large permanent increase in the induced magnetisation , and , of this , a much larger proportion is due to the application than to the removal of pressure .
Thus , the phenomena suggest that the residual magnetisation arising from the portion of the induced magnetisation which the application of pressure enables the rod to acquire is more easily shaken out by the removal of pressure than is the remainder of the residual magnetisation .
It may also be worth mentioning that by increasing the magnitude of the first pressure after the break of the current the effect of its application was increased , but the effect of its removal diminished absolutely as well .
relatively .
Cyclic effect of Pressure on the Residual .
The cyclic effects of pressure cycles on the residual magnetisation during the on which the curves are based , are shown in the sixth columns of Tables VI .
and VII .
and the fifth column of Table VIII .
The cyclic effects were also observed in the December experiments , and are recorded in the fifth column of Table Il .
In the experiments of Table II .
the demagnetisation was incolrplete , but otherwise the conditions were identical with those of the experiments of Table VII .
The cyclic effect was found very difficult to measure accurately , partly on account of its smallness , and partly , doubtless , on account of its sensitiveness to variations in the treatment of the rod during the flow , and especially the break of the current .
In the weaker fields the cyclic effects were observed only in the experiments of Table II .
, which were inferior in accuracy to the later observations .
For these reasons the numerical values attributed to the cyclic effects in the tables cannot claim to be more than somewhat rough approximations .
Thus , while believing the results to represent correctly the general features of the phenomena , I have not embodied them in curves , as the precise forms of these might have owed too much to the imagination .
S 57 .
first the case when pressure cycles were applied during the flow of the current but no pressure existed during the break , we see from Tables II .
and that the cyclic effect in the residual magnetisation is an increase or a decrease of magnetisation when pressure is\ldquo ; on according as the pre-existing field is below or above a certain critical field .
So far the parallelism between the residual and irduced mag- netisations seems complete .
The critical field , however , for the residual magnetisation is accol.ding to both tables only about 80 C.G. .
units , and so only about a quarter of ON THE NFTISAT I OF .375 that for the induced netisation , and the intensity of the critical magnetisation in the former case is less than an eleventh of that in the latter .
Table .
shows no sensible variation in the magnitude of the effect as the field is raised from 70 to 400 C.G.S. units .
Takingnext the case when no pressures were applied the flow of the current , we see from Table .
that in a field of 18 C.G.S. units\mdash ; the lowest in which observations of the effect were taken\mdash ; no sure cyclic effect was detected .
Thus a critical field must exist at or very close to this field .
In higher fields there is a clear diminution of netisation accompanying pressure on and in fields over 100 C.G.S. units the results as closely with those of Table .
as they well could .
Thus , the only conspicuous difference in the cyclic effect produced by pressure cycles the flow of the current is a rising of the critical field .
Table VIII .
, however , shows that a truly remarkable difference occurs in the phenomena when pressure exists during the break of the current .
In this case , in fields from to 400 C.G.S. units , the residual magnetisation in the cyclic state is invariably greater when pressure is " " on\ldquo ; than when it is " " off\ldquo ; The magnitude of the cyclic effect apparently dimimshes at first as the field is raised , but in fields of from 70 to 400 C.G.S. units , it is at least approximately constant .
It thus appears that the sign even of the cyclic of the residual netisation accompanying cycles of pressure , is altered by such a seemingly trifling circumstance as the existence of pressure during the break of the pre-existing current .
It seems almost unnecessary to point out that the facts stated in this raph convincingly show that the effect on the magnetisation of cyclic applications of pressure is not determined solely , or in some circumstances even principally , by the measure of the rod 's magnetic moment , or its so-called intensity of magnetisation .
S58 .
As the conclusions of the last paragraph as to the cyclic effect seem of considerable importance , it may not be out of place to supply data by which the value of the evidence on which they are based may be fairly judged .
The precautions taken in varying the order of the experiments have been already mentioned in explaining the Tables VI .
, , and VIII .
Now in fields stronger than the critical field of Table VII .
, there were , in all , 46 separote experiments of the types , and N. Answering to pressure " " on there appeared a minimum of magnetisation in each of the 31 experiments of types L and , a maximum in each of the 15 experiments of type N. As the diiference was wholly unexpected , and the observer had no preconceived ideas on the subject , it is difficult to conceive how evidence could be stronger .
S59 .
The ordinates of Curve .
of fig. 13 , give the fraction of the residual tion , which is removed by a definite number of pressure cycles in the experiments of Table VI .
, or type L. The percentage removed continually diminishes as the strength the pre existing field rises .
The rate of diminution in the percentage removed is very rapid as the .376 MR. C. ON OF SSURE strength of the field is raised to or units , but very slow when the field is ised oyer 150 C.G.S. units .
A of this curve with the curves of figs. 4 and 6 brings out clearly the fact that the residual magnetisation , which is most easily removed by pressure cycles , is that left on the break of those fields in which pressure cycles have the greatest effect on the induced magnetisation .
The data in the last columns of Tables VII .
and VIII .
would lead to curves closely resembling Curve II .
Thus the preceding remarks apply equally to the experiments of types and N. S60 .
Comparing Curves I. and II .
of fig. 13 , we see that in fields over 30 C.G.S. units , the increase in the percentage of the induced magnetisation removed by breaking the current , and the diminution in the percentage of the residual magnetisation remove by a definite number of pressure , go hand in hand .
It has been already mentioned that the application of a particularly severe first pressure after the break of a current lessened the effect of the removal of the pressure , and it also lessened the joint effect of succeeding pressure cycles , in which the pressure wlis of the normal intensity .
the close resemblance of Curves I. and II .
in the stronger fields suggests that the sudden break of a current acts in some respects as a mechanical shock , and that the magnitude of this shock-effect continual]y increases as the strength of the field is raised .
Separation of the Effects of the Existence of Pressure .
S 61 .
There are still some points connected with the fundamental character of the effect on the rod 's magnetisation produced by the application of pressure , on which Tables IX .
and X. throw some light .
We have seen that in general the first application of pressure after the rod 's introduction into a given field , causes a change in the magnetisation .
Further , as a rule , the greater portion of this change survives the removal of the pressure , so that for its continued existence the existence of pressure is not essential .
It thus seems of importance to determine whether the change depends solely on the existence of pressure for a finite time during the flow of the current , or is due in whole or in part to the actual application of pressure .
Taking WEBER'S hypothesis for illustration , the change must consist either in a sub-permanent increase in the general mobility of the ultimate magnetic molecules , or in a swinging round of some of them into extreme positions , where they are kept molecular friction even after the removal of the pressure .
The application of pressure produces doubtless molecular movements , during which the rotations of the ultimate magnets be expected to go on more freely , so that on this theory we sbould be disinclined to attribute the entire effect to the mere existence of pressure .
ON THE MAGNFTISATION OF COBALT .
In practice it is difficult if not impossible wholly to separate the effects of the application of pressure from those of its mere existence .
When the rod is put under pressure before its introduction into the magnetising coil , and not introduced until sufficient time has elapsed for any molecular motion to die out , the magnetisation might at first sight be supposed to differ from that possessed by the rod when the pressure is applied after its introduction by an amount which exactly measures the effect of the actual application of the pressure .
It must be remembered , however , that sudden magnetisation , aJtering as it does the rod 's length , must act in part as a mechanical shock and set up molecular movements , whose character and amplitude would naturally depend to some extent on the state of the rod as to pressure .
In the following remarks the difference between the magnetisations of the rod under pressure in the two circumstances of its introduction is entirely attributed to the effect of the application of pressure , but the previous statements should be taken as a warning against accepting this as more than an approximation to the truth .
S 62 .
Certain experiments bearing on the question have been already referred to in S 40 in cussing the curves of fig. 4 .
These experiments were limited to fields under 60 C.G.S. units ; but within that range they showed conclusively that in a freshly demagnetised bar the existence of sure applied before the rod 's introduction into the magnetising coil increases the susceptibility .
In the weaker fields of fig. 4 the curves show a greater susceptibility when the application of the pressure has succeeded than when it has preceded the introduction of the rod into the coil , and the evidence for this was considered satisfactory .
In fields from 30 to 60 C.G.S. units the reverse is the case according to the curves , but reasons were given for regarding the evidence on this point as insufficient .
In these experiments when pressure existed during the introduction of the rod into the coil it had always been applied some minutes previously , so that any molecular movements set oing by the process had presumab]y pretty well subsided .
Further , any such movements must have been inconsiderable compared to those occasioned by the vigorous tapping which accompanied the demagnetisation of the rod shortly before its introduction in both sets of experiments .
We are thus in all probability entitled to conclude from these experiments that in a freshly demagnetised rod the mere existence of , apart from the consequences of its application , has , in fields up to at least 60 C.G.S. units , the effect of permanently increasing the induced magnetisation , while in the weakest fields can be no doubt that the actual application of pressure has an independent effect in the same direction .
S 63 .
The results of amore complete series of observalions bearing on the same point given in the seventh column of Table IX , under the heading " " on off This signifies , as already stated , the excess of magnetisation possessed by the rod when introduced into the coil under pressure , over that it possesses when introduced free from pressure .
The circumstances of the rod are , except as regards the pressure , .
C. CHREE ON THE ElCTS OF PRESSURE in the two cases .
The rod was not , as in the experiments discussed in the graph , in a fi.eshly demagnetised condition , but had just been exposed to numerous ) ressure cycles in a field of the same strength and sign , colu1nn yrees with fig. 4 in showing in weak fields an increase of magnetisation due to the existence of pressure .
An unmistakeable critical field , however , ensues , the effect eventually changing .
Thus the character of the effect of the mere existence of pressure varies in the same way as does that of the joint effect of the application and existence of pressure , which we have already discussed in S 41 .
In this case an exact basis of comparison exists , because in the third column of X. , under the heading First " " on\ldquo ; , we have the combined effect of the application and existence of pressure when the previous treatment of the rod was the same as in the experiments of the seventh column of Table IX .
comparison of these two columns shows that the fields in which the effects attain their , and the critical fields must be nearly the same , if not absolutely identical in the two cases .
The magnitudes of the effects are , however , decidedly different .
There can be no question that in the weaker fields the effect is very considerably greatest when th application of the pressure succeeds the introduction of the rod into the coil , and this effect of the application of the pressure clearly continues until the field of 128 units is reached .
From this to somewhat over the critical field neither the application nor the existence of pressure has effects large for satisfactory determination .
In the fields of 198 units and upwards the diminution of magnetisation due to pressure was in every case found greatest when its application preceded the introduction of the rod .
us in fields up to 128 C.G.S. units the actual application of pressure during bhe flow of the current produces in the condition of the rod during these experiments an keable increase in the magnetisation , and not improbably this tendency to increase the gnetisation exists in all the fields within the range of the experiments .
In the eighth column of Table IX are a few results from experiments on the same point .
The circumstances of these experiments differed from those of the preceding column , only in that the previous fields of equal strength in which pressure cycles had been applied , were of opposite sign to those in which the effects were observed .
The phenomena are clearly of the same general character as those of the previous column , and the existence of a critical field is unmistakeable .
The observations are insufficient to determine the exact position of the critical field , but it is obviously not far from the critical field of the preceding column .
The or Non-polar of the Effects of S64 .
Another very important question is whether the effects of the application and removal of pressure on the magnetisation are of a polar or a non-polar character .
The meaning of these terms will be best understood from an illustration .
ON THE MAGNETISATION OF COBALT .
Let us suppose the induced magnetisation increased by a series of pressure cycles applied during the flow of a current , which makes A north pole .
The effect is non-polar if the rod be brought into the condition of an unstrained rod of different material possessed of a higher susceptibility .
The is polar if the susceptibility , while increased in fields making A a pole , is equally diminished in fields of opposite sign\mdash ; supposing for the moment that an opposite field could be introduced without altering the rod 's temporary properties .
If the susceptibility be simply altered by different amounts in fields of opposite sign the may be regarded as a combination of a polar and a non-polar effect .
It is probably impossible to arrive at a complete knowledge of the character of the effect , because the introduction of an opposite field implies a break and make of the magnetised current affecting the temporary properties of the rod .
It is only in so far as the effects of pressure cycles are residual that their polar or non-polar character admits of direct investigation .
If , after the application of pressure cycles in a certain field , an increase of susceptibility , not otherwise accounted for , be found in fields both of the same and of opposite polarity , the existence of a non-polar effec would seem to be established .
The same conclusion follows if there be a decrease of susceptibility irrespective of the sign of the field .
If the susceptibility be found less in fields of bhe same name as that existing during the pressure cycles , than in fieldQ of opposite name , it is difficult to think of any probable explanation other than a true polar effect .
If , however , the susceptibility be greater in fields of the same sign as that existing during the pressure cycles than in fields of opposite sign , the phenomenon be explained without assuming a true polar effect .
The reasons for these statements will probably be most clearly presented by embodying them in language literally applicable only to WEBER'S theory of magnetism .
According to this theory , the only way in which we can imagine the application or removal of pressure to produce a permanent change in the tion is either by endowing the ultimate magnetic molecules with an increased or diminished mobility , or by enabling them to swing round into extreme positions , where they stick .
The change in the molecular mobility is non-polar if the molecules turn as freely in fields of one sign as in fields of the opposite sign , otherwise the change is polar .
Now when the susceptibility is found increased in fields both of the sftlne and of opposite sign to that existing during the pressure ] , or when it is ninished irrespective of the sign of the field , or , lastly , when it is less in a field of the same sign as that during which the pressures weYe applied than in a field of opposite sign , it is clear that the phenomena cannot be attributed to the magnetic molecules havin stuck in positions into which they swung while under the directiye in fluence of the original field .
Thus , in these cases , the phenomena must be scribed to change in the mobility of the magnetic molecules , and so there must be a true or non-polar effect .
It is obvious , however , that if the magnetic molecules swung round and stuck in the way imagined , then the magnetisation would naturally be greater in a field of the same sign as the original than in a field of opposite sign .
This is not regarded here as a true polar effect , though producing in the induced magnetisation phenomena identical with those which would arise from a true polar change in the molecular mobility .
The turning round of the molecules has been already discussed in SS22 and 28 , where its occurrence was demonstrated .
The effect on the induced magnetisation will here be referred to as a quasi polar effect .
S65 .
The ninth column of Table IX .
may be regarded as measuring the combined true and quasi polar effects of a numerous series of pressure cycles , in so far as they survive the break of the field during which the pressures were applied , and the make of an equal field of the same or opposite sign .
The exact quantity measured is the algebraic excess of the magnetisation in a field of the same sign as that existing during the pressure cycles over that in a field of opposite sign , the rod being in each case introduced free of pressure into the second fields .
The results obtained with the weakest field of the table , C.G.S. units , differed so much from the others that it was inconvenient to include them in the table .
They were as follows:\mdash ; After pressur .
Kept inFirst intro.ductionOut , cedOut , introducedKept iintroducedCircumstances o End a north pole .
Pressureoffonoifoonff .
There here two or three instances of an increase of susceptibility of a non-polar character , and also clear indications of a polar effect .
Considering how small the total induced magnetisation is , both the polar and non-polar effects are of considerable importance .
In the next field , C.G.S. units , the rod was originally under pressure , and yet exceeds .
Thus , as is only , there occurred a very decided residual increase of susceptibility , mainly of a non-polar character .
In all the fields , from to 357 C.G.S. units , the residual effect is largely of a polar character .
Take , for instance , the field .
Here the pressure cycles raise the magnetisation by .
Of this , survives the break and re-make of the current , and so only about the half of the increase is lost .
ON THE MAGNETISATION OF COBALT .
The difference , however , in this case between and amounts to no less than , and thus the susceptibility , when is the north pole , is lowered -by nearly as much as the susceptibility , when A is a north pole , is increased .
The residual effect is , in fact , almost entirely of a polar character .
This appears to be bout the strength of field at which the polar effect is largest .
In the fields from to 128 C.G.S. units the results show too large fluctuations to lay any claims to great accuracy .
They may , however , be regarded as satisfactorily proving that within these limits the polar residual is small , and that it has the same sign as in the weaker fields .
In the field .
of 174 C.G.S. units no residual polar effect could be detected , and in stronger fields the effect changes sign .
There is thus a neutral field , which if not identical with , is certainly very near that at which the total effect of the first pressure on the induced magnetisation vanishes .
In weaker fields the susceptibility of the rod is greatest when the field is of the same sign as that existing during the pressure cycles , in stronger fields the susceptibility is in these circumstances least .
In fields above this neutral field there is , according to the previous reasoning , a true polar effect .
In fields again , such as C.G.S. ullits , a quasi polar effect unquestionably existed , because there could then be obtained the phenomenon of the reversal of the rod 's polarity on the break of an equal reverse current described in S22 .
This phenomenon could not be obtained in a field of C.G.S. units .
There is thus positive proof that a true polar effect exists in strong fields and a polar effect in weak fields , and most probably in all but certain critical fields they exist simultaneously .
S66 .
We have seen that the residual polar effect has practically the same critical field as the total effect of the first pressure , or First " " on\ldquo ; of Table X. Also both these effects are relatively very large in weak fields , and it might thus be supposed they stood in some intimate relationship to one another .
In fields of 30 to 50 C.G.S. units , however , the residual polar effect is very small ] , while the " " \ldquo ; is about its maximum .
An explanation of the discrepancy is supplied by a consideration of the curves and of fig. 12 .
The difference between the ordinates of these curves in a field of 8 C.G.S. units is great to the difference in a field of 30 C.G.S. units .
in the former field pressure cycles are much more eflective in increasing the retentiveness than in the latter .
A consideration of the numerical data shows that of the additional magnetisation which the pressure cycles enable the rod to acquire the loss on the break of the current is proportionally much ] in a field of 8 than in a field of 30 C.G.S. units .
Thus the quasi polar residual effect must be much smaller in the latter field than in the former , which doubtless accounts for the rapid falling off in the value of the total polar effect as the field is raised .
In the last column of Table IX .
there are a few results bearing on the same point .
These are derived from experiments in which the rod was under pressure when reintroduced into the coil after being subjected to a series of pressure cycles in a previous field of the same strength .
So far as they go the results are very similar to those of the previous column , from which it follows that the residual polar effect is , at least in its general character , independent of the state of the reintroduced rod as to pressure .
S67 .
Further results bearing on the polar character of the residual effects of pressure are given in the second , third , and fourth columns of Table X. These give the total algebraic increase of the induced magnetisation ue to the first application of pressure .
The headings show the condition of the rod as explained in S37 .
Comparing the columns headed and , we see that in fields below 50 or 60 C.G. uIlits the increase in the magnetisation caused by the application of the first pressure is very considerably greater when the rod is in a ] demagnetised condition than when it has recently been subjected to a numerous series of pressure cycles in a field of the same strength and sign .
In fields between 60 and 90 C.G.S. units the effect of pressure is still distinctly larger in the freshly demagnetised rod , but whether considered absolutely or relatively the difference between the results in the two columns is small .
In stronger fields the numerical value of , whether it signify an increase or a decrease of magnetisation , appears almost invariably greater than that of .
Both quantities , however , are then small , and in the strongest fields the effects of the first pressure were difficult to measure exactly , as the irregularities in the column headed would suggest .
It would thus be unsa , fe to build too much upon the differe1lces between the two columns in fields over 90 C.G.S. units .
The evidence , however , clearly supports the conclusion that even in the strongest fields the numerical value of the change in the magnetisation due to the first pressure is , ceteris , greatest in a freshly demagnetised bar .
The observed differences between the columns headed and are exactly of the character we should have expected from the already discussed experiments on the residual effects of pressure .
In the weaker fields , we saw that the residual effect is very considerable .
Thus the rod on its second exposure to a given field , under the circumstances of the experiment , is more or less in the state of a rod already subjected to pressure .
Thus its condition , when the pressure is apphed , is somewhat intermediate between its conditions before and after the application of the pressure Now the effect of the second pressure , after the introduction of the rod into the coil , was invariably , in the weaker fields , small compared to the effect of the first pressure .
Consequently , in such fields , the pressure would naturally have a smaller effect than the pressure , which is precisely the phenomenon observed .
Comparing the fourth column of Table X. with the previous two , we see that in fields up to 60 or 70 C.G.S. units , the increase in the magnetisation of the reintroduced rod , due to the first pressure , is distinctly greater when the field is of the ON THE MAGNETISATION OF COBALT .
opposite sign to that existing during the pressure cycles , than when it is of the same sign .
In stronger fields , the sign of the field existing during the pressure cycles appears practically immaterial .
A difference between the effects of the pressures and , or and , might be due either to a polar or a non-polar residual of pressure , but a difference between the effects of the pressures and must be due entirely to a polar residual effect .
Thus the experiments on the first pressures afford independent proof of the existence of a residual polar effect in fields up to 60 or C.G.S. units .
This is in accordance with the results of S 65 .
There would also appear to be evidence of the existence of a non-polar effect the weaker fields .
In such a field , we have already shown the existence of a very considerable quasi polar residual effect .
Now the application of a pressure in a field of opposite sign , would unquestionably shake out part of this quasi polar effect , and be expected to produce , in addition , a polar of its own , equal or nearly equal to that of the first pressure in the original field .
Thus , if there were not a non-polar residual effect , we should expect the pressure to have , in weak fields , a larger effect than the pressure A This does not , however , appear to be the case .
The positions of the critical fields for the vanishing of the total effect of the first application of pressure , indicated by the second , third , and fourth columns of Table X. are in good agreement .
Consequently , it would appear that the position of this critical field depends but little on the treatment of the rod since its last demagnetisation .
S 68 .
The results given under heading First " " off\ldquo ; in Table X. , show that in the strongest fields of that table the residual polar effects , due to the application of pressure cycles in pre-existing fields of the same strength , have no appreciable influence on the nitude of the increase of magnetisation which accompanies the removal of the pressure existing during the re-introduction of the rod into the magnetising coil .
Combining this phenomenon with the close agreement in the stronger fields of the magnitudes of the effects of the first pressures and , we seem entitled to conclude that the residual polar effect of pressure cycles in a strong field , of whose existence the two last columns in fable IX .
give clear proof , is not sensibly diminished by the application or removal of a single pressure in a subsequent equal field , whether of the same , or of opposite sign .
between the Phenomena shown by Iron Cobalt .
S 69 .
Most of the following resemblances between the phenomena shown by the cobalt rod and those shown by the stretched iron wires of Professor EWING'S experlments , have been already incidentally noticed , but a formal statement and comparison may be useful .
In the following statements , stress signifies tension in the case of the iron wires , and pressure in the case of the cobalt .
1 .
The strengths of the fields in which the changes in the magnetisation produced by the application of a given stress , or by its cyclic repetitions , bear the greatest Jatios to the pre-existing magnetisations , are very low , if indeed they possess a finite value .
2 .
The fields in which the changes in the magnetisation produced by the first application of a given stress , or by its cyclic repetitions , are absolutely greatest , are not very far from the region of the Wendepunkt .
3 .
The maximum cyclic effect appears , however , in a distinctly lower field than does the maximum effect of the first application of stress .
4 .
Both the cyclic effect and the effect of the first application of stress possess critical or VILLARI points , and the critical magnetisations are both considerably greater than that answering to the Wendepunkt .
5 .
The critical field , however , for the cyclic effect is dedly lower than that for the first application of stress .
6 .
The residual tion approaches " " saturation\ldquo ; in much lower fields than does the induced .
7 .
The residual susceptibility\mdash ; or the ratio of the residual magnetisation to the strength of the pre-existing field\mdash ; when the specimen is free from stress , attains a maximum when the pre-existing field is very considerably lower than that answering to the Wendepunkt .
8 .
The retentiveness\mdash ; or the ratio of the residual to the induced magnetisation\mdash ; in the absence of stress , attains a maximum in a field decidedly lower than that in which the susceptibility is a maximum .
9 .
Stress increases largely the residual susceptibility and the retentiveness in weak fields , and decidedly lowers the fields in which these magnitudes present their maxima .
S70 .
If the tion answering to its Wendepunkt be taken as the unit for each individual specimen , the following Table gives approximately the induced magnetisations existing during certain critical events in the magnetisation of the cobalt rod , and three samples of stretched iron wire examined by Professor EWING .
TABLE XI .
Cobalt . . .
1st iron wire This will serve to show the kind of evidence on which the above statements as to the relations between cobalt and iron are based .
Considering also that the intensities ON THE MAGNICTISATION OF COBALT .
of the magnetisations answering to the Wendepunkte of the stretched iron wires were on an verage nearly thrice that erirl to the Wendepunkt in the rod , that the wires were practically of infinite length , and that the critical fields observed in them answer , as a rule , to stresses considerably more severe than the normal stress applied to the cobalt , it seems not unlikely that by taking the of the Wendepunkt in the absence of stress as a standard for comparison , it may be possible to trace general laws underlying the varied phenomena occurring in the different magnetic metals , and thus throw light on the material changes in which the process of magnetisation consists .
NOTE .
[ The referees to whom the paper was submitted having suggested the desirability of an examination into the chemical character of the rod , I add the following analysis , for which I am indebted to Mr. R. H. ADIE , of Trinity College , Cambridge , who kindly undertook the work : Percentages .
Cobalt . . . . . . . . .
] Nickel . . . . . . .
Manganese Iron .
] Carbon , \amp ; c. 42 The rod is thus not a pure specimen of ) .
The principal impurity , however , is nickel , which closely resembles cobalt in its netic properties , though in general regarded as less strongly magnetic .
The effects of pressure on the magnetisation of nickel are of the same general character as in the case of cobaIt in fields below the critical , though more intense .
One would thus expect the critical field to be slightly raised in consequence of the presence of the nickel.\mdash ; April 16 , 1890 .
]

