]\gt ; [ 109 ] III .
Influence of Molecular Constitution and Temperature on Magnetic Susceptibility .
A. E. OXLEY , B.A. , M.Sc .
, Coutts Trotter Student , Trinity College , Cambridge .
Communicated by Prof. Sir J. .
THOMSON , O.M. , F.R. Received December 17 , 1913 , \mdash ; Read January 29 , 1914 .
INTRODUCTION .
THE present communication is a continuation , and in some respects a completion , of several preliminary researches which have been published from time to time during the past three years .
On the hypothesis that a molecule may be distorted by the near approach of its neighbours , it has been shown*how the presence of molecular groups , which vary in 'Proc .
Camb .
Phil. Soc vol. XVI .
, p. 486 , 1912 .
separately , , 1914 .
complexity with change of temperature , may modify the temperature coefficient of the magnetic property of a substance .
The large change of molecular constitution during the transition from the liquid to the crystalline state should therefore be accompanied by a sudden change of the magnetic property .
The relation between molecular complexity and specific diamagnetic susceptibility may be written , the above hypothesis , in the form ( 1 ) where is the total diamagnetic moment produced in a complex of type by the application of an external magnetic field H. is the number of complexes of the type per gramme .
During a transition from one degree of complexity to another , several types of complexes may coexist .
The summation is extended over all such types .
The corresponding expression for the specific paramagnetic susceptibility is ( 2 ) where is the Curie constant per particle of type .
Since , in the above equations , is a function of the temperature , characteristic for each substance , it is probable that the irregularities of the temperature susceptibility curves obtained by Prof HONDA , Prof DUBoIsJ and Mr. OWENS are due to the presence of this variable factor .
The application of ( 1 ) and ( 2 ) to the case of aqueous solutions of salts of the ferromagnetic elements , for which is approximately a linear function of the temperature , gives indirect evidence of the.existence of molecular complexes in solution ( see also footnote , p. 134 infra ) .
It is intended in the present research to test more fully the hypothesis stated above , which is involved in ( 1 ) and ( 2 ) , by investigating the magnetic susceptibility of a number of substances over a wide range of temperature .
To simplify the problem , the substances chosen are all diamagnetic and consist chiefly of organic compounds .
The small electrical conductivity of the latter allows us to neglect any effect which may arise from the presence of moving intermolecular charges .
The diamagnetic property will therefore be a concern of the internal structure of the *Loc .
, p. 489 .
'Ann .
der Phys IV .
, 32 , p. 1027 , 1910 . .
Kon .
Ak . .
Wetensch Amsterdam , XII .
, p. 596 , 1910 .
S 'Versl . .
Ak . .
Wetensch Amsterdam , , p. 637 , 1912 .
'Proc .
Camb .
Phil. Soc vol. XVII .
, p. 65 .
Sir J. J. THOMSON , ' Rapports du Congres Int. Paris , ' 1900 , vol. III .
, p. 148 ; also SCHRoDINGER , ' Sitz . .
Akad .
Wien , ' CXXI .
, p. 1305 , 1912 .
[ See , however , the article by H. A. LORENTZ , ' Mathematische Vorlesungen an der Universitat Gottingen , ' VI .
, p. 189 , 1914.\mdash ; Note added March , 1914 .
] CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
molecule and of such modification of that structure as is produced by the mutual actions between the molecules .
Thus stated , the diamagnetic property of a substance depends upon the molecule as a whole .
The above hypothesis postulates no definite structure of the atom or molecule beyond that it is capable of assuming a slightly distorted configuration due to the influence of neighbouring atoms or molecules .
The interpretation of the experimental facts ( described in Part I. of this communication ) by means of the electron theory of magnetism will be withheld until the results have been discussed from the molecular standpoint .
PART I. ( 1 ) THE MEASUREMENT 0F MAGNETIC SUSCEPTIBILITY AT AIR TEMPERATURE .
The magnetic potential energy of a body placed in a magnetic field may be written where is the specific susceptibility of the substance , the mass of the body , and the intensity of the magnetic field .
For a small displacement of the body , If the body be so suspended that its displacement causes work to be done on the suspended system against a mechanical force , , acting through a distance , we have Therefore . . . . . .
( 3 ) In the present experiments the body , which was spherical , was placed with its centre at ( fig. 1 ) with regard to the pole pieces and B. The mechanical forces acting on the body are therefore by ( 3 ) ( 4 ) * It should be noted that the term involving is large in comparison with the terms involving Hy and MR. A. E. OXLEy ON THE INFLUENCE OF MOLECULAR the resistance to motion in the directions Oy and Oz is great , the disturbance of the equilibrium of the suspended system , due to these forces , may be neglected .
The method of measuring the force X , or rather the method of comparing X for different substances , is shown in fig. 2 .
The torsion arm , ABCD , is made of aluminium .
The angle through which the torsion arm was twisted when the magnetic field was excited was measured by the double suspension mirror DME .
The dash-pot contained a mixture of paraffin-oil and heavy mineral oil in which four vanes , , hung .
The arm OF is the extremity of a three-dimensional adjustment stage .
The brass case , which surrounded the torsion arm and suspension mirror , was rigidly attached to a brass carrier which was provided with a screw motion and slid upon rails .
These rails were screwed to a heavy slate bed which was supported by two large blocks of stone .
To prevent the transmission of vibrations to the suspended system from outside , two layers of indiarubber were inserted between the stone blocks and the slate bed .
The rod GH provided with the cranks , and , was used to rest the torsion arm while the substance under investigation was being attached to the end ( see plan and end elevation ) .
To preserve balance of the torsion arm when different masses of substance were used , rider was provided which could be moved by the rod into different positions along ths torsion arm .
A very fine phosphor-bronze wire , attached to the projection suspended the movable parts of the system .
CONSTITUTION AND ON PTIBILITY .
The displacement of the substance , due to the action of the force X , was usually about mm. This will not appreciably alter the sensitiveness of the bifilar Now , where is the deflection on the scale , and is some constant , provided the deflection of the torsion arm is small .
Hence from ( 4 ) ( ) if the phial be placed in the same position relative to the pole pieces .
Here is the mass of the substance taken , that of the air which fills the same volume , is the VOL. CCXIV.\mdash ; A. mass of air which fills the volume occupied by the water .
In all the experiments described below the current was kept constant at amperes , to within 1 part in 500 .
Hence the factor involving the magnetic field on the .S .
of and is constant , and we get by division . . . . . .
The specific susceptibilities of water and air were taken as and ( at C. ) respectively .
( 2 ) RIMENTS DOWN To THE TEMPERATURE 0F LIQUID The apparatus designed for the measurement of susceptibility at low temperatures is shown in fig. 4 .
The phial , supported from the torsion arm at , was surrounded by a thinwalled glass tube , the latter being sealed at the lower end and supported by the cork .
Conduction of heat from or to the torsion arm was prevented by inserting a strip of wood between the phial and the projection of the torsion arm .
A copper cylinder , which just fitted into the glass tube , served to keep uniformity In the last two experiments the current was increased to 9 amperes .
CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
of temperature throughout the volume occupied by the phial .
A layer of phosphorus pentoxide was placed at the bottom of the glass tube to absorb any moisture and thereby prevent the deposition of water on the phial during cooling .
The liquid air tube was brought up from below into the position shown .
The flange of cork was very broad and prevented any convection currents from disturbing the equilibrium of the suspended system .
A copper German-silver thermocouple was used to measure the temperature , one junction being situated just above the phial .
The thermocouple circuit included a delicate D'Arsonval galvanometer .
A slight kick , due to induction effects , was observed when the magnetic field was excited .
This quickly died away and readings of the flection of the phial and temperature were taken simultaneously .
( 3 ) EXPERIMENTS UP To C. The method of suspension used is essentially that devised by CURIE , fig. 5 is a diagram of the apparatus , looking along the the axis of the electromagnet .
The phial which was used in these experiments was open above and closed below , as shown at Q. The suspension rod , ASSS , was made of brass and was 'Ann .
de Chim .
et de Phys VII .
, vol. 5 , p. 298 , 1895 .
surrounded by glass tubes , TTT , to prevent disturbances due to air currents .
screen , formed of two pads of asbestos board , with an intervening layer of cottonwool kept wet by water from the pipe , served to prevent disturbances due to air currents from the walls of the electric furnace F. The temperature attained was measured by the copper German-silver couple , fitted into the top of the furnace , as shown .
A support of wood was inserted between the phial and the main brass suspension to prevent conduction of heat and the accompanying thermal disturbances .
In both series of experiments the whole apparatus had to be surrounded with pads of cotton-wool to keep a steady zero .
( 4 ) DISCUSSION 0F POSSIBLE ERRORS .
In the measurement of magnetic properties over wide intervals of temperature it is necessary to consider the effect of a number of possible sources of error and these will now be considered .
( a ) The Effect of of Volume during or llization on the Volume of Air Displaced by the We shall denote the specific susceptibility of the liquid by , that of the crystals by .
From equation ( ) , by division , where is the additional mass of air displaced owing to the change of volume of the substance in passing from one state to another , the corresponding change of volume , and is the susceptibility of air per unit volume .
For most substances , if , i.e. , for the volume of substance experimented on in this research .
Hence the variation of deflection due to this cause is , or 4 parts in 1000 , and is beyond the limit of experimental accuracy .
In the above it is assumed that there is no appreciable change of temperature of the system while fusion takes place , so that no effect due to variation of with temperature enters .
is taken as a mean value of Vide G. TAMMANN , ' Kristallisieren .
Schmelzen , ' Leipzig , 1903 , pp. 204 et seq. The values of in cubic centimetres per cubic centimetre for some of the substances investigated in this research Benzene , .l0 ; Benzophenone , ; Nitrobenzene , Naphthalene , ; Mercury , CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
( b ) The Effect of Variation of Temperature on and the Influence of over the TVide Range of Temperature used in these Experiments .
Since , and , where the subscript refers to the crystalline- or liquid state , we have , dashing the deflections and for the higher or lower temperature , , which is the ratio of the deflections due to thermal expansions of the substance and air , and the variation of the susceptibility of the substance and air with temperature over an interval .
The necessary data for calculating the magnitude of this correction are not known for most of the substances which have been used in these experiments , but the largest factor involved is that due to the variation of density and susceptibility of the contained air .
( The variation of the susceptibility of the air external to the phial is eliminated since the difference between and is taken .
) For a temperature interval of C. , this factor amounts to 2 per cent. mately ) of the deflection , , due to the substance .
This error is , however , distributed over a range of C. When the data of thermal expansion of the liquid and crystalline states are , there will be two corrections , one for each state , which will be distributed throughout the range of temperature for which that state exists .
It is to be observed that in this and the foregoing correction the thermal expansion of the glass and the variation of its susceptibility with temperature disappear , since they are equally involved in and , the difference of which is taken .
( c ) The effect of displacement of the ( liquid ) , as whole , during cooling was tested in two ways : ( i ) By actually displacing the liquid and repeating the observations ; ( ii ) By usin , in some of the later experiments , a phial in which the displacement was prevented .
*This factor is ( d ) Correction due to Dissymmetry of the Phial .
The phials were nearly spherical and were placed with their centres as nearly as possible at the point ( fig. 1 ) where the force on them is a maximum and error due to displacement a minimum .
The present measurements are differential ones , and provided the phial is replaced in the same position , no appreciable error can creep in from this source .
( e ) Latent Heat and Conductivity .
The possible errors due to latent heat of change of state and thermal conductivity will be considered in more detail when the preliminary experimental curves are discussed .
( f ) of Air ) in tloe Substance .
The error due to the presence of dissolved air in the substance is reduced to a minimum by heating the substance strongly .
In those experiments performed at air temperature , an inappreciable amount of air was evolved on fusion .
The error arising from this cause is negligibly small ; for if a quantity of air equal in volume to that of the phial were expelled on fusion , it would account for only one-third of the actual effect found .
But a small trace of air dissolved in the substance will , in all probability , remain in the substance if a considerably higher temperature is not sufficient to remove it ; and although such a trace of air would have a small influence on an absolute measurement of the susceptibility , it will not appreciably affect the difference of susceptibility , between the liquid and crystalline states .
( 5 ) METHOD 0F EXPERIMENT .
Fig. 6 shows how the value of the force acting on the suspended system varies with , ( a ) when the phial is filled with air , ( b ) when it is filled with water .
In each figure the dotted circle shows the size of the phial , and OM is a distance equal to the radius of the pole piece on the same scale .
It will be seen that if the phial moves a small fraction of 1 mm. either towards or away from , the resultant deflecting force remains very nearly constant , because one-half of the substance and phial has a slightly larger force acting on it , while the other half has a slightly smaller force .
The pole pieces were cylindrical , 8 cm .
diameter and cm .
apart , so that the liquid air tube could be inserted for low temperature work and a small electric furnace for high temperature work .
advantage of large pole pieces is that the size of the phial is of less consequence .
The hysteresis effect was not appreciable and the deflections were very consistent ; a small increase of the current was readily followed by a small increase of the deflection .
A Grassot fluxmeter showed that the field intensity was 3000 gauss for a current of amperes .
Measurements were first made with the phial filled with air , water , and the CONSTITUTION AND TEMPERATUR ON MAGNETIC SUSCEPTIBILITY .
substance , at room temperature .
If the substance .
were a liquid it was cooled by liquid air , and after about 1 hour observations were begun and taken at a large number of temperatures as the substance gradually warmed to room temperature .
A series of such observations occupied a period of 10 hours .
Frequently the heating was too slow to allow the complete interval to be bridged during the day , and in such cases the curve was completed the following day .
It will be observed that the lag in temperature of the substance behind the thermocouple , with such a small rate of cooling , will be small , and this lag will be uniform so long as there is no sudden change of the state of the substance .
At temperatures considerably higher or lower than the temperature of fusion , the lag will generally be greater than it is at the fusion-point itself It can easily be seen from the curves given below that the effect of this will be to lessen the differeIlce of susceptibility between the liquid and the crystalline states .
The actual difference of susceptibility cannot , therefore , be attributed to this source even if the lag were considerable .
A similar series of readings was taken for the phial alone , which , corrected for the variation of the susceptibility of the contained air with change of temperature and subtracted from the readings obtained above , will give the variation of the specific susceptibility of the substance with temperature .
Similar experiments have been made , using hot water and other liquids in the Dewar tube , but it was difficult to take readings at temperatures higher than C. subtracting the ordinates , we get the values of at different temperatures .
These , in conjunction with the value of for water at air temperature , give the specific of nitrobenzene at different temperatures .
The two curves in fig. 8 show how the specific susceptibi]ity of nitrobenzene varies with temperature ; curve ( a ) for phial of type ( a ) , curve ( b ) for phial of type ( b ) .
As the crystals were very gradually melted by heat conducted through the walls of the phial , no effect due to latent heat would be expected to enter .
Moreover , if the crystals were taking heat from the phial.faster than it was supplied by the surrounding enclosure , the temperature of the phial would be colder than that indicated by the thermolneter .
The phial could not get hotter than the enclosure unless the latent heat were ative .
The only way in which latent heat might affect the observations , therefore , is by cooling the phial below the temperature indicated by the thermometer .
But this effect would increase the deflection due to the glass and accordingly reduce that due to the diamagnetic substance contained in it .
This result would l1lake any correction due to latent heat of opposite sign to the difference actually found .
A good test of the absence of any error due to this cause is obtained by showing that the change of susceptibility is permanent .
If the effect were due to absorption or emission * The calculation is as follows:\mdash ; for .
water scale divisions .
scale divisions ; scale divisions . .
; .
; density of nitrobenzenc cquation ( ) , we find Also At the fusion-point scale divisions .
Hence 12 per cent. CONSTITUTION AND ON MAGNETIC SUSCEPTIBILITY .
12.3 of heat on change of state , this could not be the case .
The change of susceptibility cannot be ascribed to an alteration of thermal conductivity for , as will be seen later , the same differen ce has been obtained at air temperature between crystals and supercooled liquid , each stood long enough to take up the temperature of the room .
The experiments on about twenty ) will now be briefly sulnmarised , for the most part in the order in which they have been investigated .
In one or two cases , substances , are cbemically closely related , have been classed together .
The results are shown graphically in figs. 8-14 .
A short account of any peculiarity will be given and the curves will be more fully discussed later .
For brevity , in what follows , the specific diamagnetic susceptibility will be referred to generally by , the absolute temperature by , and a curve showing the relation between and 3 will be called a curve .
About experiments have been made and some of the substances have been investigated several times , using both types of phial described above ( fig. 3 ) .
A of Benzene ) the most part , the simpler benzene derivatives have been selected , because in many cases their crystalline forms have been investigated and it was hoped that an interpretation of the results would thereby be rendered more feasible .
Liquid toluene is more diamagnetic than crystalline toluene , amounting to 5 per cent. Two curves for this substance have been obtained and the in the curve occurred in each case at .
The melting poin is at C. fig. .
The liquid is more diamagnetic than the crystals , the change of taking place at C. and per cent. The melting point of crystals is at C. fig. .
The curve for phial filled with cymene was parallel to that for the empty phial over the interval C. to C. ( see note under pyridine below ) .
Over thirty substances ha been investig } .
An account of these additional eXperimentS will be given in a future paper .
CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
as fig. shows , appeared to be less abrupt than usual .
* The melting point , corresponds to the beginning of the dip , is at .
The middle of the dip is at a temperature C. , which is very nearly equal to the value of the obtained by other methods .
The value of is 5 per cent. This substance showed a departure from rule that for benzene compounds the liquid state is more diamagnetic than the crystallin state .
Five experiments have been made with this substance and two of the curves are shown in fig. 10 , Curve I. The liquid , which is oily , is seen to be less diamagnetic than the crystals by an amount 5 per cent. of the value of .
Taking the point where the of begins , as the temperature of fusion , we deduce C. for the melting point .
Other methods give : This substance will be considered later in conjunction with several other compounds showing similar properties .
The first experiment made on substance showed that it possessed the hitherto unique property of constancy of the magnetic susceptibility through the normal fusion point down to the temperature - C. , as shown by the curve ABC ( fig. 10 , Curve II ) .
But on taking away the metal cylinder ( used to preserve a uniform temperature over * This may be an effect of too rapid heating of the e1ldosure combined with low conductivity of the substance .
the space occupied by the substance ) during a repetition of this experiment , it was discovered that the benzaldehyde had not crystallized but had passed into a transparent jelly-like mass .
By plunging the phial containing the benzaldehyde into liquid air , the benzaldehyde again assumed this jelly form , but on warming , followed by a further slight cooling , a white crystalline mass was obtained .
This was now cooled to temperature of liquid air , withdrawn , suspended from the torsion balance and quickly surrounded by liquid air .
An experiment carried out on the benzaldehyde after this treatment is shown in fig. 10 , Curve .
It will be seen that approximately the normal dip , 5 per cent. , was obtained , the liquid being more dialnagnetic than the crystals , but the transition stage was considerably drawn out and appeared to have an intermediate step or stable state .
Shortly after this difference in magnetic property between the gelatinous and the states of benzaldehyde had been established , a research appeared by M. CHAUDIER , had discovered that the specific magnetic rotation of organic substances behaves in a similar way .
He found that if the substance passed into a gelatinous state , on being cooled below the normal fusion point , there was no discontinuity of the magnetic rotation in passing from the liquid to the gelatinous state .
The rotation , on the other hand , disappeared totally if the substance passed into a crystalline form on being cooled .
The two phenomena\mdash ; magnetic rotation and magnetic susceptibility\mdash ; are doubtless closely allied and each is affected by a transition from the liquid to the crystalline state .
An attempt is made in Part II .
to connect theoretically these effects .
It does not appear that CHAUDIER obtained the same substance in both gelatinous and crystalline forms , and it would be interesting to carry out an experiment on the magnetic rotation of a substance capable of existing in either form at a given temperature .
CHAUDIER shows that the discontinuity exists for benzene , , and nitrobenzene on crystallization , and that the rotation remains constant in the cases of amyl-alcohol , glycerine , and essence of terebene , which pass into the gelatinous form on cooling below the normal freezing point .
The phenomenon with the last substance is made complex on account of optical activity .
This substance shows an anomaly different from that of any other benzene compound investigated .
Five experiments have been made and one of the curves is drawn in fig. 11 .
The pit in the curve occurs in every case at the normal fusion point , C. On either side of this point , the curve rises , the distant parts of the curve corresponding to the liquid and crystalline states bein continuous , as shown by the dotted line .
This dotted lin appears to correspond to no real physical state as it was found impossible to make the substance pass along this path .
The depth of the pit * ' Comptes Rendus , ' vol. 156 , p. 1529 , May , 1913 .
CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
second experiment , crystallization took place at a temperature C. below the normal melting point .
In each case , when the crystals were heated , fusion took place at the normal temperature , C. The substance was melted by hot water contained in a Dewar tube , and the temperatures below that of the room were obtained with a tube which had been cooled by liquid air .
It is apparent that with this substance a definite hysteresis effect , due to temperature , is obtained , similar to that which was discovered by the late Prof. HOPKINSON* , in case of nickel-steels .
If we regard the paramagnetic state of the nickel-steel as isotropicf , the temperature hysteresis effects in the two cases are completely analogous .
The paramagnetic solid corresponds to the netic liquid .
On cooling , the nickelsteel passes the critical without crystallizing just as the benzophenone passes the normal fusion point and assumes a gelatinous form .
Continued cooling in each case causes the more stable crystalline form to appear .
This is accompanied by a large rise of the paramagnetism of the nickel-steel ( ferro-magnetism ) and by a fall of the gnetism of the benzophenone .
If now the nickel-steel be heated , it loses its ferro-magnetism at the critical temperature\mdash ; a temperature which is characteristic for a given alloy and is lower the greater the proportion of nickel in the steel\mdash ; while the benzophenone regains its larger diamagnetic property at the normal fusion point\mdash ; a temperature which is lower the greater the impurity present .
Further , the heat produced ( recalescence ) when the netic state appears on crystallization of the nickel-steel is analogous to the heat produced ( heat of formation of the crystals ) when the benzophenone crystallizes .
The molecular changes are of the same nature in the two cases ; the only difference is in the character of the magnetic property possessed by the molecules and by which such molecular changes are indicated .
S 'Roy .
Soc. Proc vol. XLVIII .
, p. 1 , 1890 ; or EWING'S 'Magnetic Indnction in Iron and other Metals , ' p. 184 , ei seq. HOPKINSON referred to the paramagnetic state above the critical temperaturo as non-magnetizable state .
[ At least we may regard it as an allotropic modification of the alloy .
See the article by GUILLAUNIE , ' Rapports du Congres Int. Paris , ' 1900 , vol. 1 ; also DUMAS , ' Journ. of the Iron and Steel Institute , ' 68 , p. 255 , .
It should be mentioned that the results described in this communication concerning change of state are in accordance with the work of TAMMANN , 'WIED .
Ann LXII .
, p. 285 , 1897 .
See the article by SPRING , 'Paris Reports , ' 1900 , and WHETHAM'S 'Theory of Solution , ' Chap. II .
, where additional ences to TAMXANN'S work are given .
added March , 1914 .
] It would be more correct to speak of the ' range of temperature of transformation\ldquo ; instead of " " critical temperature [ .
M. OWEN has observed that the specific susceptibility of super-coo ] galhum , between C. and C. , is constant and equal to that of the liquid above the normal fusion point .
The crystalline state is much more diamagnetic and remains nearly constant until the fusion point , C. , is reached .
My attention was drawn to this point of priority , after my work had been completed , by the curve given in OWEN'S , ' Ann. der Phys , 37 , p. 693 fig. , 1912 .
The English translation of this paper , 'Versl .
Kon .
Ak . .
Wetonsch Amsterdam , , p. 637 , 1912 , which first came to my notice , contains no diagrams and I had overlooked the brief reference there made , at the foot of p. 642 , to the behaviour of gallium.\mdash ; Note added March , 1914 .
] VOL. CCXIV .
( b ) A Study of Naphthalene Compounds .
In each case the liquid state is more diamagnetic than the crystalline state .
The change of susceptibility on crystallization was per cent. for each substance , and took place at C. with naphthalene , at C. with -naphthylamine fig. , and at C. with -bromonaphthalene .
The rather low value of for naphthalene and its derivatives may be due to the two carbon atoms " " condensed\ldquo ; in the naphthalene nucleus , which differ considerably in their chemical nature from the remaining carbon atoms of the naphthalene system .
( c ) ercury .
The liquid state was found to be more diamagnetic than the crystalline state , the value of amounting to 14 per cent. The curves are shown in fig. 14 .
These were obtained by first cooling the mercury with liquid air and observing the susceptibility as the temperature gradually rose .
In the curve ABCD , when the point had been reached , the mercury was re-cooled to a temperature just below that of the freezing-point determined from position and afterwards to C. , and the points marked @ were obtained .
Between C. and C. the diamagnetic property remained constant .
Experiments on Substances which show Liquid State .
Cholesteryl Chloride , , and Para-azoxyanisol , In each case the diamagnetic property decreased during the transition from the CONSTITUTION AND TEMPERATUR ON MAGNETIC SUSCEPTIBILITY .
1.31 true crystalline to the true liquid state .
The value of for para-azoxyanisol was 6 per cent. A smaller and less definite value was obtained with cholesteryl chloride , and after the liquid state had appeared the susceptibility continued to decrease , probably owing to further dissociation of molecular complexes .
M. PASCAL* has recently investigated para-azoxyphenetol , and has discovered that the red liquid form is less diamagnetic than the yellow crystalline form , the value of being 18 per cent. This substance appears to behave like para-azoxyanisol , but gives a larger value In the case of para-azoxyanisol , the -3'curve for temperature was similar to that for descending temperature , but the two curves were slightly displaced owing to thermal lag .
Fusion took place at The sign of for each of these liquid crystal compounds is the same as that for benzoyl chloride and phenylhydrazine .
Through the kindness of Prof. POPL , I have been able to examine some crystals of the last substance under the crystallization microscope , but no trace whatever of a doubly refracting layer at the boundaries of the crystals could be discerned .
It may perhaps be supposed transition of all substances , from the true liquid to the true crystalline state , takes place over a finite though small interval of temperature .
This does not imply that all substances should show the liquid crystal state , but merely that a labile state exists , which may or may not be doubly refracting , in which the molecular grouping is determining how the substance will crystallize on further cooling .
( 7 ) GENEBAL DISCUSSION 0F THE XPERIMENTAL RESULTS .
The dependence of diamagnetic property o11 crystallization readily follows as a result of the production of eneously close-packed assemblages of atomic spheres of influence .
The difference of structure of the crystalline and amorphous states of a substance is due to the distribution of the restraining forces acting on the molecules of the crystalline structure , which will necessarily produce a small distortion of each molecule ; while in the amorphous or gelatinous structure , the molecules will possess a more symmetrical shape on account of the withdrawal of the restraining forces .
This deformation effect in every molecule of the crystalline structure is in accordance with the hypothesis stated at the beginning of this communication\mdash ; a deformation of the atom presumably producing a modification of the internal structure which will in turn modify the diamagnetic property .
In Part II .
, an attempt will be made to interpret this change of diamagnetism in terlns of the change of self induction of the electron orbits contained in the atom .
The inappreciable deformation of the molecules when 'Ann .
de Chim .
et de Phys VIII .
, vol. 25 , p. 375 , 1912 .
BARLOW POPE , ' Trans. Chem. Soc vol. LXXXIX .
, p. 1675 , 1906 .
MR. A. .
OXLEy ON THE INFLUENCE OF MOLECULAR the substance assumes a gelatinous or liquid form is in accordance with the constancy of the magnetic property which is observed when the substance super-cools or passes into a jelly .
Until we postulate some definite internal structure of the atom or molecule , it is impossible to say what the precise effect of the adjustment , from the symmetrical shape of the molecule of the gelatinous state to the non-symmetrical shape of the crystalline state , will be , either as regards magnitude or Providin the molecular complexity does not change , the diamagnetic property does not depend upon the temperature , and it can therefore be represented by a series of straight lines parallel to the temperature axis .
Continuity between these straight lines is represented by the equation ( 1 ) During the transition from one state to another , different types of particles co-exist , characterizing the growth of the number of particles of type , the diamagnetic moment produced by the applied field , in a complex of type , is dependent upon the molecular distortion of the complex .
In the crystalline and liquid states , where and are the diamagnetic moments produced in a distorted molecule of the crystalline and liquid states respectively , and is the number of molecules per gramme of the substance .
The magnitude and sign of the change of susceptibility will be determined by the nature of the crystalline structure , while the shape of the curve of transition*depends on the variation of with temperature .
Below the melting points of benzaldehyde , benzophenone , and phenylhydrazine , two distinct curves have been obtained .
In each case the velocity of crystallization of the super-cooled liquid was small , for observations showed that , hours after cooling , the susceptibility of the gelatinous state was indistinguishable from that of the true liquid .
For benzene derivatives , the diamagnetic property of the crystalline state is in general less than that of the liquid state ; but in addition to those compounds which show a liquid crystal state , there are two important exceptions , viz. , benzoyl chloride and phenylhydrazine , in which the crystals are more diamagnetic than the liquid .
It is interesting to notice that there exists for benzene and its derivatives an alternative kind of packing which is found in the crystalline structure of some of the halogen derivatives .
This alternative or cubic system is highly symmetrical for In the experiments which have been described the shape of this curve is masked by the effect of low thermal conductivity of the substance .
CONSTITUTION AND TEMPERATURE ON MAGNETIC 1,3 .
' } benzene itself , but the presence of a halogen atom destroys the symmetry of the complex , the latter being sheared in various directions .
lt is by a process of shearing , similar to that which corresponds to the change from the cubieally to the hexagonally complex , that we may account for the different diamagnetic properties accompanying polymorphism and by which we may , at any rate provisionally , explain the abnormal results obtained with benzoyl chloride and phenylhydrazine .
matically , we may represent such a shear by the accompanying figures ( a ) and ( b ) , where the black dot represents the substituent of the benzene complex .
During crystallization it frequently happens that pairs of molecules assume mirror-image positions\mdash ; ( a ) and ( b ) are types of this arrangement- .
In the case shown at ( c ) , the mirror-image formation is absent .
The forces of crystallization will be in the three cases .
On the whole , the resultant of the forces between the molecules will act inwards on each ring system in case ( a ) , and outwards in case ( b ) .
Each of these types differs from ( c ) where the forces on the whole may act neither inwards nor outwards on ally given ring system .
Hence we may regard the substituent as a directive agent , and in so far as it influences its own complex and the neighbouring complexes , there is a reversal of the effects produced by the forces of crystallization in case ( b ) as compared with case ( a ) .
This reversal of the interaction of substituent and nucleus with respect to any particular molecule will produce a distortion of the internal structure of an opposite kind .
For convenience and clearness the disposition of the six carbon atoms of the benzene ring is represented as uniplanar .
This need not be the case .
It is possible now to explain the two abnormal cases of benzoyl chloride and without having to assume that the liquid state possesses a greater molecular complexity or a closer packing of molecules than the corresponding crystalline state .
I have had the opportunity of discussing , with Prof. POPE , this question of a reversal of the change in magnetic property due to crystallization , and he informed me that the demands of this explanation are in accordance with the modern views of chemists and crystallographers .
The nature of the irregularity of the curves does not appear to be directly MR. A. E. OXLEy ON INFLUENCE OF MOLECULAR connected with the change of volume on crystallization ; for in the case of water*the effect is small , and is normal as far as sign is concerned , while a large expansion takes place on crystallization .
In the case of benzene there is a fall and a rise of on crystallization , while the change of state involves a contraction simply .
With regard to the crystalline forms of the elements , BARLOW and POPE point out that those elements which show allotropic modifications crystallize in forms of low symmetry , while other elements not so easily obtainable in allotropic forms crystallize in forms of high symmetry .
It seems necessary , therefore , to distinguish between the forces holding the atoms in position in the molecule and the forces holding the molecules together in the crystal structure .
The atoms of an element are assumed to be identical , and the molecular aggregate to which they give rise and the accompanying distortion depend upon the kind of packing which these equal atomic spheres of influence assume .
The only element which has been investigated in this research is mercury .
The large change of on freezing indicates a large change of molecular configuration .
The allotropic forms of an element , which possess characteristic crystalline structures , will have different molecular distortions and consequently different magnetic properties .
Moreover , as these forms are of different degrees of stability , the disintegration of the structure with change of temperature will in general be different for each form .
The curves may therefore be expected to show irregularities .
Reference has already been made to the work of BOIS and the magnetic properties of the allotropic modifications of phosphorus , antimony , carbon , and tin .
But these authors used the irregularities which their observations disclosed as an argument against the truth of the foundations of LANGEVIN'S electron theory of diamagnetism instead of interpreting them as due to the variation of molecular complexity .
The experiments described above show that the peculiarities of the curves can in all cases be traced to the variation of molecular complexity .
In the liquid state most of the compounds that have been investigated are unassociated , and the absence of any variation of with is conclusive proof of the truth of the Curie law of diamagnetism when the molecules are not appreciably distorted .
This law is true for distorted molecules , if the distortion is not a function of the temperature , as the parallelism of the curves for the various crystals and for the empty phial shows .
It is clear from the experimental results that diamagnetism is not wholly an atomic property , for the near approach of the molecules , which takes place on crystallization , * PICCARD , ' Comptcs Rendus , ' vol. , p. 1497 , December 23 , 1912 , and WEISS and PICCARD , ' Comptes Rendus , ' vol. 155 , p. 1234 , December 9 , 1912 .
These authors find that the variation of the diamagnetic p1operty of water with temperature can be interpreted as due to the co-existence of two types of complexes , one type growing at the expense of the other as the temperature is varied .
This explanation is in accordance with equation ( 1 ) which I had published earlier .
'Trans .
Chem. Soc vol. LXXXIX .
, , 1906 . .
Kon .
Ak . .
Wetensch Amsterdam , XII .
, p. 601 , 1910 .
CONSTITUTION AND TEMPE ON MAGNETIC SUSCEPTIBILITY .
is able to modify the susceptibility appreciably .
From the decrease of diamagnstism accompanying the crystallization of benzene derivatives , it appears that the process of crystallization implies the utilization of the residual forces between the molecules which tends to make them more normally saturated .
The value of is not constant for the different members of the same class of compounds , and in benzene derivatives it is a function not only of the rin system but also of the substituent .
Before the extent of this adjustment of the diamagnetic property\mdash ; due to the juxtaposition of unsaturated groups\mdash ; can be of value , it is necessary to examine different classes of compounds each containing a large number of members .
It will have been seen that a satisfactory interpretation of the variations of the diamagnetic property found in this research can be given in terms ofmolecular theory .
All that has been assumed concerning the structure of the molecule is that it is a unit which can be distorted by the near approach of other units .
In order that we may inquire more fully into the nature of the distortion and the forces which cause it , we must postulate some definite structure of the molecule .
We shall assume , in the investigation which follows ( Part II that the molecule consists of a congeries of electrons ; and that the positive charges , which keep the electrons whirling in their orbits , are fixed in the molecule .
For the present , looking at the problem from a purely molecular point of view , the hypothesis stated at the beginning concerning the molecular distortion which accompanies crystallization , has been found justifiable , and has led to results in harmony with the views of chemists and crystallographers .
Little has been said in this communication co1lcerning the Curie law of paramagnetism , which may be written in the form The recent work of Prof. KAMERLINGH ONNES , the variation of paramagnetism at low temperatures , shows how we may interpret departures from the simple Curie law , , by a modification of the Curie constant .
The above expression for shows that if the change of aggregation or the process of crystallization modifies the structure molecule , then for every type of complex , we shall have a new Curie constant per particle of this type .
In addition to this , the above expression provides a continuity of the magnetic property during the transition where there will be a co-existence of various types of particles .
The passage from the liquid to the crystalline state modifies the paramagnetic as well as the diamagnetic property .
ONNES has shown that there is a sudden change of the value of during the crystallization of liquid oxygen .
Versl .
Kon .
Ak . .
Wetensch Amsterdam , XV .
, p. 322 , 1912 .
'Versl .
Kon .
Ak . .
Wetensch Amsterdam , XIV .
, p. 674 , 1912 .
It appears therefore that these two expressions for the diamagnetic and netic properties are generally applicable and explain the variation of susceptibility with temperature in terms of two Curie laws .
These laws are exact to within the limits of experimental accuracy , and any variation of the magnetic property which is apparently not in accordance with them can be attributed to variation of the lnolecular complexity .
PART II .
( 1 ) A OF THE PHENOMENA FOUND IN THE ABOVE EXPERIMENTAL RESEARCHES 1N TERMS 0F THE ELECTRON THEORY 0F MAGNETISM .
The theory of LORENTZ , which postulates no free magnetism , but attributes magnetic phenomena to the motion of electric charges , accounts admirably and in a simple way for the main characteristics of the Zeeman effect and for the rotation of the plane of polarisation in a magnetic field .
Each of the latter phenomena appears to be closely related to the diamagnetic property of a substance ; for LANGEVIN has shown that the magnetic moment produced in a molecule by the application of an external magnetic field is such as would lead to diamagnetism by a process which is compatible with the simple explanation of the Zeeman effect given by LORENTZ .
Further , the fact that the magnetic rotation produced by the ferromagnetic elements is proportional to the intensity of magnetization , and not simply to the intensity of the applied magnetic field , shows that we must take into consideration the magnetic property in this case also .
No theory has yet been devised , however , which is completely consistent with the principles of mechanics and the phenomena\mdash ; optical and electric\mdash ; of the luminiferous medium , and in adopting the theory of LORENTZ , we are merely seeking to obtain a fuller interpretation , on certain hypotheses , of the experimental facts obtained in Part I. But it is evident that whatever theory be adopted with regard to the origin of the diamagnetic property , any distortion of the molecular system , such as is produced by crystallization , must be accompanied by an adjustment of the magnitude of that property .
To obtain a satisfactory interpretation of the phenomena which have been observed in the experiments of Part I. , we must take into consideration the forces which the systems of rotating electrons in different molecules exert on each other .
LORENTZ has shown that the force exerted on an electron ( surrounded by neighbouring systems ) by an electric force is not simply , where is the charge on the electron .
A term must be added , where is the electric polarisation of the medium , and the function characterizes the grouping of the molecules for any particular state of the substance .
The total mechanical force acting on an electron is therefore CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
We may regard the crystalline state as isotropic to a first approximation , since the crystals formed in the experiments have all kinds of orientations .
The introduction of the term will then represent the effect due to the modification of the internal motions of an atom or molecule , by the process of crystallization , and will no longer characterize the different properties which the substance really possesses along its crystalline axes .
We shall follow the electron theory of M. LANGEVIN* and refer the electron system of any molecule to axes fixed in direction relative to the aether , but whose origin always coincides with the centre of gravity of the whole system of electrons .
Let be the co-ordinates of a particular electron .
Then , and since the medium is -assumed to be isotropic , Let be the co-ordinates of the centre of gravity of a molecule .
Then , since the medium is isotropic , and The co-ordinates of an electron , with respect to the centre of the molecule in which it is situated , will be denoted by .
The above equations give The areal velocity of an electron , with reference to the centre of gravity of the molecule , will have a component along Oz given by , . . . . . . . . .
( 1 ) and , assuming that we can interpret the magnetic effect of the electrons spinning in their orbits by the magnetic moment where .
( 2 ) we may write from ( 1 ) and ( 2 ) .
( 3 ) 'Ann .
de Chim .
et de Phys VIII .
, vol. 5 , p. 70 , 1905 .
For clearness I have borrowed the treatment up to equations cctly irom M. LANGEVIN'S paper .
Sir JosEPlI LARMOR Phil. Trans , has shown that this relation implies the existence of a constraint to provent the accumulation of displacement during the action of a steady magnetic field on a system of spinning electrons .
See also TRBM , 'Roy .
Soc. Proc vol. p. 31 , 1913 .
VOL. CCXIV.\mdash ; A. MR. A. E. OXLEy ON THE INFLUENCE OF MOLECULAR In addition to the forces due to external sources acting on the electrons contained in a given molecule , there will be internal forces having components X , , which are determined by the configuration of the molecule .
If the origin move with velocity the velocity of any electron will be and LANGEVIN writes the equations of motion in the form ( 4 ) where and are the electric force and magnetic force and is the mass of an electron .
If we wish to take into account the forces due to neighbouring molecules , when crystallization takes place , we must add the term to these equations .
Now , in a crystalline structure , the motion of a molecule will consist of oscillations about a centre , and , therefore , if we take a mean value of this polarisation term and add it to the equations ( 4 ) we shall include the effect of the presence of neighbouring molecules .
Assuming that the medium is isotropic ( which is permissible since the crystals will have all kinds of orientations , and we are seeking the effect of close approach of the molecules only and not any particular property in a specified direction ) the new equations are ( 4 ) On account of the small dimensions of the elementary system considered , the electric force and the electric polarisation term will be nearly constant throughout its extent , so that denoting the value of and at the centre of the elementary system by and , we have , expanding and neglecting higher powers than the first , and ( 5 ) Calculating from ( 3 ) , ( 4 ) , and ( 5 ) , and using the relations given on p. 137 for an isotropic medium , we find writing CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
The last term is zero if each molecule has no initial moment , as LANGEVIN'S theory of diamagnetism requires .
Leaving aside the subscript and usin the equations of the electromagnetic field and we get , following .
Integrating this equation from time to time we find where is the magnetic moment produced in the molecule during this interval .
The second term on the R.H.S. of ( 6 ) will depend upon the molecular configuration of the substance and implies a modification of the electron circuits which will change their self-induction .
Now any change of the self-induction may be represented by a small change of the intensity of the applied magnetic field .
We may write where " " \ldquo ; characterizes the groupin of the molecules and therefore , where is the elementary change in the external field during a small interval of time Therefore .
, ( 6 ) where is the total variation of due to establishing this magnetic field when the electron orbits are modified by the presence of neighbouring molecules .
LORENTZ , ' Theory of Electrons , ' p. 137 .
If be the number of molecules in one gramme of the substance we may write the specific susceptibility ( 7 ) and it is clear that the value of is modified by crystallization , for the polarisation term varies with the state of the substance .
If , ( 7 ) reduces to the Langevin expression for diamagnetic susceptibility in which the mutual influences of the molecules .
are neglected .
The variation of on crystallization may be written ( 8 ) where and are the electric polarisation constants for the crystals and liquid respectively ; is of the order 1/ 3 for most liquids .
* It now remains for us to examine the amount of this change and to do this we must determine .
The value of the polarisation constant for any substance is positive and may be very large compared with the value 1/ 3 which it usually possesses for the liquid state ; but its exact determination in any particular case is difficult since it depends upon the actual distribution of the molecules , about which we know very little .
It appears , however , that we can deduce an approximate value of the magnitude of from CHAUDIER'S researches on the change of magnetic rotatory power with change of state .
This deduction leads to a reasonable value .
It has been shown by BECQUEREL , BICHAT , and later by CHAUDIER , that the magnetic rotatory power of the liquid and vapour states is not generally the same , and it is reasonable to suppose that a similar but more pronounced effect will be found in the passage from the liquid to the crystalline state , but the phenomenon will usually be complicated in the latter case owing to the appearance of double refraction .
The researches of and CHAUVINS show that in the case of Iceland spar the effects of double refraction and magnetic rotation are simply super osed .
Now CHAUDIER has established that the magnetic rotatory power disappears completely during the crystallization of aniline , benzene , and nitrobenzene ( all of which have been investigated in Part I If this is a true discontinuity of the magnetic rotatory power , apart from any effect due to double refraction , then we can use this fact to obtain an estimate of CONSTITUTION AND TEMPERATURE .
ON MAGNETIC SUSCEPTIBILITY .
the value of for the crystalline forms of these substances .
The value deduced* ( see footnote ) on these assumptions must be at least of the order and may be larger .
Now reverting to equation ( 8 ) , we see that the experimental value of , approximately 5 per cent. , will be in accordance with the above theory of the effect of crystallization if the change of self-induction of an electron orbit of a substance in the liquid state , due to the presence of neighbouring molecules , be such that The value of this ratio will be still smaller if possesses a greater value than .
In the crystalline state , the modification of the -electron orbit will be determined by and since is large , this term accounts for the variation , , found in the experimental investigation of Part I. In the case of liquids , where , this term produces no appreciable effect on the value of , as equation ( 7 ) shows .
( 2 ) THE MEAN MOLECULAR FIELD 0F A DIAMAGNETIC CRYSTALLINE SUBSTANCE .
The large value of the electric polarisation constant is compatible with the considerable rigidity possessed by crystalline structures .
The term therefore be important in the crystalline state .
For a super-cooled liquid we may write ' * An extension of the theory of magnetic rotation has just been published by G. H. ( ' Phil. Mag vol. 26 , p. , 1913 ) , in which the influence of the electric polarisabion due to neighbouring molecules is taken into consideration .
LIVENS shows that the rate of rotation of the plane of polarisation by a solution may be written where is the frequency of the light wave , the restoring coefficient due to displacement of an electron , the constant of electric isation , and , and have their usual interpretation .
The extends to all the electrons in unit volume .
If we get the result of DRUDE'S theory , where the influences of the molecules neglected ; is of the order for the liquid state , and if is reduced to 1 per cent. of this value when crystallization sets in , the value of must be at least of the order 10 .
Now E.S.U. , , and .
As there are about electrons per cubic centimetre , must be of the ordor least .
We have assumed that the medium is isotropic , but as the crystals are differently orientated this assumption will not change the order of .
This ratio corresponds to a small aIteration of the area of an electron orbit to the maximum extent of 1 in 2,000 , a magnitude which is quite missible .
This equation is in accordance with the observations of M. HAUDlER , who has shown that , if a substance super-cools or assumes a gelatinous form , the magnetic rotation is the same as for the normal liquid state gelatinous state ) .
neglecting the term since 1/ 3 .
For the crystals , at the same temperature , we have Hence or I shall call the mean molecular field of the diamagnetic crystals .
for a given change of the electron orbits due to crystallization , will be proportional to Hence we may neglect the mutual forces between the diamagnetic molecules when the substance crystallizes providing we supply a molecular field which is proportional to , and therefore to , or to the intensity of magnetization of the substance ( equation ( 6 ) ) .
For the liquid and gelatinous states this molecular field is negligibly small , because is small , but for the crystalline state is large , and the molecular field is comparable with the external field This is exactly analogous to the theory of ferro-magnetism developed by M. WEISS , in which the molecular field is , where is the magnetic moment produced by the external field and is a constant determined by the intermolecular forces of the crystalline ferro-magnetic material .
corresponds to the polarisation constant of the diamagnetic theory .
We ought , however , to compare with , where the molecular field .
N. , the density of the substance , is of the order 1 .
If , this gives since Assuming , we find .
Hence the constant of the diamagnetic field is .
The corresponding values of given by WEISS and BECK* are for iron , for nickel , and for magnetite .
The mean molecular fields are enormously greater in the cases of ferro-magnetic substances on account of the large value of the intensity of rletization as compared with that for diamagnetic ubstances .
The values of ( the saturation molecular field ) are 6 , 530 , 000 gauss for lron , 6 , 350 , 000 gauss for nickel , and 14 , 300 , 000 gauss for magnetite , while the corresponding value of for a diamagnetic substance is of the order 150 gauss for an applied field of 3000 gauss , the field used in the above experiments .
To ount for the phenomena of the Zeeman effect , magnetic rotation of the plane of polarlsation , and diamagnetism , it is necessary to assume that an atom contains ourn .
de Phys ser. , vol. 7 , p. 249 , 1908 .
CONSTITUTION AND TEMPERATURE ON MAGNETIC SUSCEPTIBILITY .
oppositely rotating systems of electrons .
A diamagnetic molecule , on LANGEVIN'S theory , has no initial magnetic moment but there can be complete compensation of the electro-magnetic forces due to the spinning electrons only at points considerably removed from the molecule .
Although these forces are compensated at an external point , and the magnetic property observed corresponds to diamagnetism , yet between the molecules the forces will be large and the local intensity of magnetization may be comparable with that of ferro-magnetic substances .
* It is interesting that the value of , which defines the crystal structure of a diamagnetic substance , should be at least of as high an order of magnitude as the constant of the ferro-magnetic field .
It is probable that these large values of the coefficients of the molecular fields are the origin of the stability and rigidity of crystal structure in general .
Since we may interpret the effect of the luolecules by means of the term , it follows that if from any cause the polarisation suffers a reversal of sign , then the sign of will change , i.e. , the variation of on crystallization will correspond to the abnormal cases referred to in Part I. The mechanism by which this reversal of is brought about may be attributed to a directive influence of the substituent ( or other dissymmetry of the complex molecule ) which will produce an electric polarisation effect of opposite kind in a neighbouring nucleus , according to the direction of the force between the substituent and the nucleus .
The liquid state will be more diamagnetic than the crystalline state if is negative , i.e. , if the process of crystallization involves a small reduction of the selfinduction of the electron orbits .
This corresponds to the majority of cases investigated in Part I. In the above theory no has been taken of the small reaction forces which will arise from the production of the diamagnetic moment .
Since , however , ( and therefore ) does not vary by more than a few per cent. as the substance passes from the liquid to the crystal state , these small reaction effects will appear in both states approximately to the same extent and therefore the value of will not be affected on their account .
SUMMARY 0F CONCLUSIONS .
The experimental results described in Part I. have established an almost complete parallelism between the dependence of and ferro-magnetic susceptibilities upon molecular constitution .
The molecular changes are of the same nature and the only difference between the two classes of magnetic phenomena lies iu the character of the magnetic property possessed by the molecules and by which the variation of molecular complexity is disclosed .
* See Note added , p. 145 .
LANGEVIN , ' Ann. de Chim .
ct de Phys VIII .
, vol. 6 , p. 88 , 1905 , concludes that such effects would not produce an appreciable change in the magnitude of In the liquid and crystalline states , providing there are no molecular complexes whose constitutions vary with the temperature , the diamagnetic property involves no temperature co-ordinate and is represented by a series of straight lines parallel to the temperature axis .
The work of MM .
WEISS and KAMERLINGH ONNES has shown that ferro-magnetic and paramagnetic properties can be represented by a series of hyperbolae .
In either case , I have shown that if the substance forms aggregates , or assumes a crystalline structure , the transition stage is represented by the equations ( diamag .
) , ( paramag .
) which take into account the coexistence of molecular complexes of different types .
These two expressions are generally applicable and represent the variation of magnetic property with temperature in terms of the two Curie laws , and the superposed effect of the change of molecular complexity with temperature .
For most of the substances examined , the liquid state is more diamagnetic than the crystals .
The cases of benzoyl chloride , phenylhydrazine and the liquid crystal compounds , cholesteryl chloride and para-azoxyanisol , are abnormal .
Nearly all benzene derivatives show a change of susceptibility on crystallization , amounting to 5 per cent. ( approximate ) .
The si gn of this change is attributed to the directive action of the substituent .
The experiments show that diamagnetism is not wholly an atomic property ; for the process of crystallization , involving simply a closer approach of the atomic spheres of influence , is able to modify the susceptibility .
The existence of a gelatinous or super-cooled liquid state , possessing in all the cases examined a diamagnetic property equal to that of the liquid above the normal fusion point , corresponds to the existence of a supercooled paramagnetic state of nickelsteels below the critical temperature .
Definite hysteresis loops with respect to temperature have been obtained , similar to those discovered by the late Prof. HOPKINSON for nickel-steels .
The heat of formation of the diamagnetic crystals corresponds to the heat evolved at the critical temperature of the nickel- steel .
The constancy of the diamagnetic property when the substance supercools or passes into a jelly is in harmony with the constancy of the closely related phenomenon of the magne'tic rotation of the plane of polarisation by that substance when a gelatinous form is assumed .
The process of crystallization is accompanied by a modification of the susceptibility and of the magnetic rotation .
The hypothesis of molecular distortion stated at the beginning of this communication has been justified , and the interpretation of the effect of crystallization as due to a slight modification of the internal structure of the molecule is in agreement with the conceptions of chemists and crystallographers concerning the nature of crystal structure .
CONSTITUTION AND TEMPERATURE ON MAGNETIC PTIBILITY .
In Part II .
, the extension of LANGEVIN'S theory shows that an exceedingly small and reasonable deformation of the molecules in the liquid state is sufficient , owing to the large value of the constant , to account for the change of susceptibility actually observed when the substance crystallizes .
Moreover , the introduction of the idea of a molecular field of the diamagnetic crystals allows us to neglect the mutual influences of the diamagnetic molecules in exactly the same way as the molecular field of WEISS allows us to express the phenomena of ferro-magnetism , without bringing into consideration the mutual influences of the molecules .
Both the molecular field of WEISS and the molecular field of the diamagnetic crystals are proportional to the intensity of magnetization , in the respective cases , and the minimum value of the constant of the latter field , deduced from the variation of magnetic rotatory power when crystallization sets in , is of the same order as that of the ferro-magnetic field .
Although on account of a particular structure of the molecules the magnetic properties of different substances , observed externally , may vary enormously , yet , within the substances , the local intensities of magnetization and the forces between the molecules will be in all crystalline snbstances .
These forces determine the rigidity of the crystalline structures .
These investigations have been made in the Cavendish Laboratory and I wish to thank Prof. Sir JOSEPH THOMSON for placing the necessary apparatus at my disposal and for the intereSt he has shown in the work .
I am , indebted to Prof. PoPIl for supplying me with many substances for investigation and also for a valuable discussion on the subject of crystal structure .
[ It will have been observed that the theoretical treatment given in Part II .
consists of an interpretation of the polarisation term , , ( which represents that part of the diamagnetic moment due to the distortion of a molecule by the influences of neighbouring molecules of the crystalline structure ) by means of a magnetic field .
This field is superposed upon the applied field and it has been shown that we may disregard the mutual influences of the molecules forming the crystalline structure if the resultant of these fields is substituted for the applied field .
This corresponds to the interpretation , given by M. WEISS , of the mutual influences of the molecules of ferro-magnetic substances by a magnetic field , , superposed upon the applied field .
Since the work described in Parts I. and II .
was communicated to the Royal Society , a paper has appeared by WEISS ( ' Comptes Rendus , ' vol. , No. 25 , p. 1405 , December 22 , 1913 ) which deals more fully with the nature of the molecular field in ferro-magnetic substances .
It is there shown that the large forces which act between the molecules are not necessarily magnetic but may be of any other nature .
VOL. CCXIV .
146 MR. A. E. OXLEy ON TlfE INFLU NCE OF MOLECULAR CONSTITUTION , ETC .
consideration of the continuity of magnetic flux has led WEISS to abandon the idea that the magnetic field is actually a magnetic field .
The difficulty may be surmounted by assuming the molecular,.field to be localized .
This localized nature is a necessary property of the molecular field in diamagnetic crystalline substances .
But , as WEISS points out , it is rather difficult to see how the molecular magnets , or the electron orbits , in adjacent molecules , can approach one another so closely that , in spite of thermal agitation , the local field between the molecules should amount to the enormous value of gauss .
In any case , whatever may be the nature of the local forces which hold the molecules of a crystalline structure together , we may interpret the accompanying distortion by means of a magnetic field of such magnitude that it produces in the molecules the same change of moment as the actual forces within the crystalline structure will produce .
This statement holds whether the substance be ferromagnetic or diamagnetic .
The large value of this local magnetic field in diamagnetic substances has been briefly mentioned at the end of Part II .
It can be shown that a local field , comparable in intensity with the ferro-magnetic field of WEISS , will produce , according to LANGEVIN'S theory , a modification of the electron orbits in the molecules of a diamagnetic substance such as would account for the change of observed on crystallization in the experiments of Part I. In a subsequent paper I hope to describe in detail several extensions of the work contained in Parts I. and Il .
, and I should like to make a brief reference here to one or two points because they furnish additional support to the conclusions which have already been mentioned .
Further experiments have been made , similar to those described in Part I. , and the change of susceptibility on crystallization has been correlated with the magnetic double refraction of the same substance by MM .
COTTON and MOUTON .
The large double refraction of a crystalline structure , as compared with that of a liquid Sub.ected to the strongest magnetic field which we can produce in the laboratory , is compatible with the relatively large local magnetic field by which we may express the forces which hold the molecules in position in the crystal .
Perhaps these large local fields between the molecules of a crystalline structure may be identified with the intense intra-atomic fields of RITZ ( ' Ann. der Phys IV .
, vol. 25 , p. 660 , 1908 ) which give us a direct interpretation of BALMER'S and RYDBERG'S expressions for spectral series , or with the magneton of WEISS .
This would imply that the molecules in the crystal structure are exceedingly close together and that their oscillations are of extremely small amplitude .
It is possible that the magneton will prove to be a constituent of diamagnetic as well as of paramagnetic and ferro-magnetic matter.\mdash ; Note added , 1914 .
]

