]\gt ; The lVIeasurement of Dielectric Constants by the Osciltations of Ellipsoids and Cylinders in a Field of Force .
By W. M. THORNTON , D.Sc .
, D. Eng. , Professor of Electri Engineering in Armstrong College , Newcastle-upon-Tyne .
( Communicated by Sir Oliver Lodge , F.R.S. Received February 22 , \mdash ; Read March 11 , 1909 .
) ( Received in revised form April 28 , 1909 .
) CONTENTS .
PAGE 1 .
Introductory 422 2 .
Theory of Experiment 424 3 .
Description of Apparatus 426 4 .
Turned Ellipsoids 429 5 .
Quartz 430 6 .
Flint Glass 436 7 .
Relation of Dielectric Constant to DenSity and Refractive Index of Flint Glass 437 8 .
Moulded Ellipsoids 438 9 .
Paraffin Wax 439 10 .
Bees 439 11 .
Shellac 440 12 .
Sealing-wax 440 13 .
Gutta-percha 440 14 .
Chatterton Compound .
441 15 .
Dielectric Cylinders 441 16 .
Ebonite ; Amber ; Ivory ; Canada Balsam ; Resin ; Indiarubber ; Sulphur ; Gutta 445 17 .
Olive oil 445 18 .
Heavy Paraffin Oil 446 19 .
Water 446 20 .
Collected Results 446 1 .
This paper deals with the measurement of dielectric constants by a method which , though simple , does not appear to have been previously employed in same form .
Ellipsoids and cylinders accurately cut from crystals and other insulating material are suspended between flat plates connected to an alternating transformer supplying a sinuous electrification of 1500 to 2000 volts amplitude , and the period of small swings of the suspended rod or ellipsoid , with and without the field , is observed .
From these measurements the dielectric constant can be found .
The method is capable of a order of accuracy , and the results obtained agree closely with the best of previous determinations , especially The Measurement of Dielectric those made by Hopkinson for different varieties of glass .
It also lends itself to work with liquids , and some examination can be made of the polarisation of dielectrics in a unidirectional field .
The several methods which have been used in the measurement of the dielectric constants of insulators are fully described by Gray* and by kelmann .
A complete bibliography of the subject is given by the latter and by Landolt and Bornstein .
In most of those methods recorded , observations have been made of the change in the electrified state of plates when the dielectric substance is introduced between them .
Boltzmann determined the constants of solids easuri the translational force on a small sphere suspended in a field of force of known intensity ; and Quincke those of liquids by the tension and oaused in them by electrification .
In the oing the dielectric was in every case at rest ; and with the exception of Boltzmann 's experiments , the effects obselved have been external to the cell containing the dielectric .
The method now to be described differs from these in that the dielectric is ellipsoidal or cylindrical in form , and makes small oscillations about a position of symmetry in line with the field .
The apparatus is really a condenser , with a portion of its dielectric free to move .
The electrostatic field was maintained in sine wave alternation at a frequency of about 80 a second ; the oscillations of the rods never exceeded one a second .
By the use of an field , the influence of " " elastic afterworking\ldquo ; and of conduction currents are avoided , and the conditions are more nearly those for which the constants are usually required .
After some preliminary experiments , a short reference to the method was made in an inaugural address .
S It was found later that Beaulard was at the time of the early experiments investigating the dielectric constant of water by enclosing it in a hollow glass ellipsoid suspended in a field maintained by the usual spark device in a state of oscillation at two million alternations a second .
lIis value is very interesting , and is probably of the right order , though the experiment did not aim at a degree of accuracy The differences between the recorded values of the constant axe not on the whole to be explained by differences of composition .
In view of the importance of its accurate determination under alternating fields of low frequency or infinite wave-length , the following measurements were ' Absolute Measurements in Electricity and Magnetism , ' vol. 1 , chap .
VII , S 2 .
'Handbuch der Physik ' ( Auflage ) , vol. 4 , ( 1 ) , pp. 92\mdash ; 150 .
I Landolt and Bornstein , ' Phys. Tabellen , ' Tab. 240 .
S 'Journal Inst. Elec .
Engrs 1905 , Part 176 , vol. 36 , p. 186 .
'Science Abstracts , ' No. 2097 , November , 1905 ; ' Revue Electrique , ' August 30 , 1905 .
Prof W. M. Thornton .
[ Apr. 28 , made , in which the solid ellipsoids and cylinders used were worked into shape with all possible care .
Great pains were taken to obtain a steady voltage , and to constantly compare it with a standard unidirectional voltage .
The points in the paper which perhaps call for special note are the small mass of substance used , in the solids not exceeding a third of a gramme and in liquids a tenth of a gramme ; the close agreement between theory and experiment in S 5 ; and the practical method adopted of dealing with cylindrical specimens .
2 .
The restoring couple upon a dielectric ellipsoid with its axis inclined to an electrostatic field of force is a quadratic function of its dielectric constant .
* When the ellipsoid is freely suspended by an elastic fibre , the couple may be measured by the rotation of a torsion head required to bring the body to any fixed angle with the direction of the field , or by the period of scillations about the position of rest .
The motion in the latter case is that of a " " quadrantal pendulum the restoring moment of which reaches a maximum at an angular displacement of 45 degrees .
The general theory of this motion has been given by Thomson and Tait , and a solution large amplitudes by Greenhill , in the above case there is , in addition to the polarisation couple , that of the elastic control .
From measurements of the couple , the dielectric constant can be found in terms of the strength of the field , the dimensions of the body , the constant of the surrounding medium , and the time of oscillation or angle of twist of the suspension fibre .
When the medium is air or , the method is absolute in the sense that the constant is determined from one set of measurements .
The equation of motion of a quadrantal pendulum under torsional control and with negligible damping is where I is the moment of inertia about the axis of suspension , the restoring moment of the suspension per unit twist , and the polarisation couple , being the angle made by the longer axis with the undisturbed field .
In the case of an isotropic dielectric whose constant is in a medium of constant , ( 2 ) where is the volurne of the body , the intensity of the field in electro*Maxwell , ' Electricity and Magnetism , ' vol. 2 , SS437 et seq. 'Natural Philosophy , ' vol. 1 , S322 .
'Applications of the Elliptic Functions , ' A. G. Greenhill , p. 103 .
CbCt , static units , and coefficients depending only upon the shape of the body .
In the case of a prolate ellipsoid of eccentricity , the transverse coefficient , , the longitudinal coefficient .
( 3 ) When the oscillations are so small that the square of the amplitude may be neglected , The period is therefore ( 5 ) and The free period , so that and 6 ) Hence from equation ( 2 ) , ( 7 ) which gives ( 8 ) the equation by which is determined when ] is known .
In this is the term derived from the movement of the ellipsoid , the other terms are found from its dimensions and mass .
When the amplitude is so large that the period is sensibly increased , this must be reduced to the case of infinitely small swings .
Writing , ( 9 ) and with an amplitude , we have Expanding , and rejecting terms from upwards , we have for the period ( 10 ) where , .
( 11 ) Vide Maxwell , Prof. W. M. Thornton .
[ Apr. 28 , Writing , .
( 12 ) where , and For indefinitely small amplitudes , and therefore , are zero , and in this case .
( 13 ) For finite amplitudes in which , and the square of the frequency of oscillation reduced to indefinitely small amplitudes is therefore which for most practical purposes may be written .
( 16 ) The value of upon which depends , itself requires a knowledge of If it is impossible to work with small amplitudes this is a real difficulty .
but if , as in the following experiments , the swing rarely exceeded 5 degrees , the reduction to zero amplitude may be omitted in a first approximation , and calculated from equation ( 6 ) .
This value may then be used to find and the reduction made as above .
If a still closer approximation is required the process may be repeated .
The logarithmic decrement of the motion was in most cases .
The correction of the square of the frequency for this increases it in the ratio , that is The experiment then consists in observing , and , correcting for amplitude and damping , and inserting the value of in equation ( 8 ) , to calculate the value of K. Arrangement of eriment .
3 .
The ellipsoids were suspended by a quartz fibre , about cm .
diameter and 30 cm .
long , between square brass plates held apart by glass distance pieces at their corners ( see fig. 1 ) and enclosed in a glass vessel having a thick glass top carrying a suspension tube , torsion head , and circular scale .
To reduce damping and to accelerate the drying of the specimens the vessel was in most of the determinations exhausted to 1909 .
] Measurement C'onstants .
a vacuum of 700 mm. and a dish of phosphoric anhydride placed at the bottom .
The size of the plates and their distance apart were chosen after trial with the stream line method* to determine the best proportions of the field under the distolting influence of the ellipsoids .
With dielectric ellipsoids2 cm .
long a convenient distance is 6 cm .
apart , with plates 10 square .
At this distance the feeble induced on the suspended body were , as shown by the results obtained for flint glass and quartz , insutficient .\mdash ; Testing Cell .
to change the distribution of electricity on the plates by any appreciable amount .
At 4 cm .
apal't the period is increased by a few hundredths of a second .
When , however , a metallic ellipsoid of the same size is placed ill the field it is necessary to space the plates 10 .
apart to reduce its reaction on them to less than a measurable amount .
The motion of the specimen was observed by reflexion from a mirror into a telescope having an eye-piece with five cross wires radiating from the centre at of * Heal Shaw and Hay , ' Phil. Trans vol. 195 , ser. , pp. Prof W. M. Thornton .
[ Apr. 28 , 10 degrees apart .
were set with regard to the field by focussing the central one upon the ellipsoid brought to rest in a strong field of force .
The potential difference between the plates could be raised to 10,000 volts .
obtained from a small oil-cooled transformer .
Iu order to get very steady voltages the steam end of a 24-kilowatt Parsons turbogenerator was dis- connected from the generator so that the latter could be run independently as a motor .
SIip rings were fitted over part of the commutator , and the machine driven from large storage cells as an inverted converter fig. FIG. 2.\mdash ; Arrangement of Apparatus .
B. , storage battery .
T. , step-np transformer .
S. , motor starter .
H.R. , 99,000 resistance .
M.G. , motor generator .
G. , hot wire galvanometer .
A.A. , alternating current slip rings .
R. , standard resistance , 100 , switch of current regulator .
T.C. , testing chamber .
All slight .
of the alternating voltage which had previously given trouble was in this way avoided , the maximum variation of pressure during several minutes\mdash ; the time of observation of the period of the ellipsoids\mdash ; , as a rule , one part in 800 .
Difficulty was for some time experienced in obtaining readings of the voltage consistent to per cent. Several forms of voltmeter were tried without success .
The means eventually adopted was to connect a high resistance across the terminals in series with a Duddell twisted strip ammeter which was enclosed in a metal case surrounded by flanneL It was found necessary , on account of the influence of slight changes of temperature upon the readings , to calibrate with direct current before and after each set 1 ne of observations with alternating current .
When this was done very consistent were obtained .
Working by the method of oscillations the period was increased by air resistance to the extent of about 4 parts in 1000 .
The in every case was so small that a correction could be made by observation of the logarithmic decrement as indicated in S2 .
Taking this into account no measurable effect was found from the of the ionised state of the surrounding air when the pressure was reduced to 755 .
of mercury , below which it was found not advisable to go on account of the occurrence of luminous discharge between the plates .
In working , a torsional swing was given to the ellipsoid , from the period of which and the known moment of inertia the elastic constants of the suspension were found .
The voltage was then applied and the new period found .
The ratio of the square of the intensity of the field to the difference of the squares of the frequency of oscillation with and without the field was the object of the experiment .
The glass and quartz ellipsoids and cylinders were suspended by forming a small drop of fused canada balsam on the quartz fibre , slightly warming the surface of the solid and quickly melting the drop into contact with it by the approach of a hot rod .
If the ellipsoid were then out of balance , as tested on a levelled mirror , the attachment was broken and the process repeated , usually a dozen or more times before a perfect suspension was obtained .
The other substances were hung in a cocoon silk saddle .
4 .
The substances selected for trial of method were optical flint glass of densities , and , and quartz cut parallel and perpendicular to the optical axis .
The former were chosen because their constants have been very fully determined , especially by Hopkinso the latter on account of the calculations of their dielectric constants which have been made , from very accurate optical observations , upon the netic theory of dispersion .
Ellipsoids of reyolution of each of these substances were made by Messrs. Hilger , and , tested by optical enlal.gement , were found to agree so closely with the true elliptic section that no difference could be measured between a photographic enlal.gement , by ) , of the ellipsoids , and a calculated ) , having the same major axis , 10 inches .
The ellipsoids were carefully measured , their volumes and densities determined , and their moments of inertia compared with a standard cylinder of heavy glass also made by Hilger , which ether with a shorter cylinder was used in the determination of the influence of the ends of cylinders upon the polarisation torque .
The measurements are iven in Table I. It can be 'Original Papers , ' vol. 2 .
Prof W. M. Thornton .
[ Apr. 28 , hown*that , on account of the low inductive constants of dielectrics , the influence of slight deviations from the true form does not rise in magnitude in the result .
Table I.\mdash ; Measurements of Glass and Quartz Ellipsoids .
Substance .
Mass .
Volume .
Density .
Length .
Diameter .
L. N. I. Quartz .
5 .
When a uniaxal crystalline ellipsoid is suspended in a field of force , the restoring couple upon it depends upon the inclinations of the optic axis both to the direction of the field and to the axis of suspension .
The general expression for the couple upon an aeolotropic ellipsoid is , with the previous notation .
( 17 ) For an ellipsoid cut perpendicular to the optic axis this will be greatest when the optic axis is parallel to the , suspension .
In this case and are each equal to , the coefficient of dielectric polarisation at right angles to the optic axis .
When the ellipsoid is cut parallel to the axis , the coefficient of longitudinal polarisation is KII , the transverse coefficient We proceed therefore to find iirst the dielectric constant for the ellipsoid cut perpendicular to the optic axis .
Quartz , perpendicular to the optic axis\mdash ; Ellipsoid.\mdash ; Length cm .
, diameter cm .
, mass gramme , volume , density gm.-c A series of observations was made with the optic axis inclined to and parallel to the axis of suspension , iving the following values , each of which is the mean of from 20 to observations .
is the angle between the two above axes in degrees .
Vide Chrystal , ' Encyc .
Britannica , ' ed. 9 , Art .
" " Magnetism p. 245 .
' Encyc .
Britannica , ' .
cit. , eq. ( 82 ) .
1909 .
] The surement of Dielectric Constants .
A curve of these is drawn in fig. 3 , the dip in which points to a change in the apparent electrical behaviour of the crystal wlJen the polarisation is everywhere exactly at right angles to the optic axis , , the condition under which the ordinal'y ray is transmitted .
The couple on the ellipsoid is a maximum at this point , but the approach to it is so steep that a very angular displacement from the axis of suspension causes a marked difference in the period .
A calculation of the variation of the couple on the ellipsoid with the angle shows a dip of this nature but not quite so steep as in uaytz between axis of Sion 3.\mdash ; Change of horizontal couple with vertical inclination of optical axis .
The , value is mean of two series taken with diflerenl suspensions , the care being taken to have the axes in line .
The average value of the first series was 138 and of the other .
It was impossible to obtain perfect alignment , a difference of mlll .
on the circumference corresponding to an angular displacement of 4 degrees .
Prof W. M. Thornton .
[ Apr. 28 , The determining equation ( 8 ) for this ellipsoid is from which Taking for a vacuum of 700 mm. The following observations were made on the ellipsoid of quartz cut parallel to the optic axis over an interval of six months with the same suspension in each case .
Each is the mean of ten , but the pressure was not at this time very steady:\mdash ; Table II .
Here is the hot-wire galvanometer deflection ; the corresponding current ; P.D. the potential difference across the plates ; the electrostatic intensity of the field calculated from the voltage , assuming " " ; the frequency of swing .
In order to reduce to the frequency for zero amplitude , the leading nearest to the mean is taken , that is the second row in the table ; we have then from equation ( 6 ) and from ( 11 ) , Thus and by ( 16 ) , 1909 .
] The Measurement of Dielectric that is .
This is increased to allow for in the ratio finally .
The corresponding value of is .
The influence of the correction has been to reduce the above quantity in the ratio , that is .
Applying , now , this reduction factor to the mean value , we obtain finally Measurements of the ellipsoid .
Thus , from equation ( 8 ) , from which finally , with and KI I It is interesting to compare these results with hose obtained from optical data by the Ketteler and Sellmeier dispersion formula\mdash ; ' where is the dielectric constant for exceedingly long waves ; the index of refraction at wave-length , constants at the wave , at which selective absorption occurs .
These constants have been found for quartz by Bubens and Nichols , *using a bolometer as a detector of waves coming from an Aver burner , reflected several times from the surfaces of quartz blocks , and separated by a grating .
FroIn their results the values wele obtained\mdash ; the wave-lengths being in If is taken for the I ) , and for the ordinary ray the formula ives ( ord. ) The above values of the constants usually quoted as being appropriate to the ordinary ray .
The constants for a similar dispersion fo1mula for the extraordinary rlty have not been calculated , but if is taken for the line and for the extraordinary ray we as an estimate ( ext .
) The above formula has recently been slightly modified by lIaclaurin who has worked out the constants for rocksalt and sylvin but not quartz .
It 'Wied .
Ann vol. 60 , p. 418 , 1897 .
' Roy .
Soc. Proc , vol. 81 , p. 367 , 1908 .
Prof W. M. Thornton .
[ Apr. 28 , is probable that the value 46065 would be leduced in conformity with his treatment by as much possibly as per cent. In comparing results obtained from optical data with those derived from direct electrical experiment , the electric polarisation must be in the same direction in cases .
Since the electrical polarisation in a light wave is transverse , values of calculated by the above formula , using the index of for the ordinary ray , obtained by transmission of light along the optic axis , correspond to those found directly by exposing the quartz to an electric field at right angles to the axis .
Thus ( ord. ) as calculated should with , and ( ext .
) with KII .
We have Observed .
Calculated .
ord. 4578 ext .
It may be remarked that the mean of the observed values of which is , agrees better with the value of the corresponding constant in the dispersion formula than There are two further bearing upon the agreement of the observed and calculated values which must be considered .
The expression " " at right angles to the axis\ldquo ; does not specify the precise direction the crystal in which the cutting was made .
The-value is therefore subject to a possible variation of a small fraction of 1 per cent. from the mean .
Again , the composition of quartz is known to , but fine clear crystals are pure silica .
Both Nichols and Rubens , and also Hilger , remark that the crystals used by them were very fine , and the agreement indicates that the molecular state was very nearly the same in both cases .
There is no recorded investigation of the influence of change of composition on the dielectric constant of quartz , but judging by the small difference between the constants KII and , it would be small .
At the end of the paper , values of obtained by Fellinger , Ferry , Curie , and Romich and Nowak are given , the mean of which for KII is and for Fused Qnartz .
The density of fused quartz is about that of crystalline .
In order to see how the dielectric constant depends upon density as distinct from structure , an ellipsoid was made of quartz , carefully fused and perfectly clear .
It The Measnrement of Dielectric lad the following dimensions : diameter , cm .
; length , 2003 cm .
; mass , ramme ; volume , .
; , density , The mean value for was , from which iving and The ratio of this to the mean of I and is .
The in lielectric constant is from this about 4 per cent. less than the corresponding in density : in keeping with the fact that the influence of the crystalstructure in quartz upon the dielectric constant is comparatively small .
6 .
The dielectric constant of flint glass has been very fully ted by fopkinson , *using a } condenser .
It is common in modern optical work to use a heavy flint lass of densit ( bout .
Its departure from Maxwell 's law is great , but its large lielectric constant renders it a suitable material for a standard dielectric llipsoid to be used in alternating fields of low frequency .
In order , however , to make a comparison over the whole of fopkinson 's research three ellipsoids were made of densities , and , the measurements of which are iven in Table I. For the first , diameter cm .
; length , cm .
; mass , : rolume , ; , the mean of many )bservations of corrected for amplitude and the ralue We then which In this experiment the vessel was 1lot exhausted , so that , and .
In order to fix the position of the curve between the densities and , the ellipsoid of density was made , the measurements of which Volume .
The mean value of , corrected as before , was at a frequency of .
Thus .
'Phil .
Trans Part I , 1878 , pp. ; Part , 1881 , pp. VOL. LXXXII.\mdash ; A. 2 Prof. W. M. Thornton .
[ Apr. 28 , The ellipsoid of light flint lass , of density , melting point 67 C. , had Volume The mean value of was at a frequency of 86 .
Thus for this lass from which .
Comparing these values with HopkinsoIl'S , there is seen to be very close agreement in the dense glasses , but llot with the light glass , the difference in that case eing 2 per cent. In working with light flint glass the composition and melting point must be known in order to make a fair comparison .
because of the possibility of obtaining the same density by a variation of the proportions .
The composition of the heavier glasses is more closely fixed by the proportion of lead lequired to give the gl.eater density .
The symmetry of the curve of fig. 4 shows that the variation of density FIG. 4.\mdash ; Variation of the dielectric constant of flint ] with density .
was obtained by a systematic of the proportion of lead , while the ctlrve is valuable in itself , especially near the origin , there must always yide Hopkinson , .
cit. 1909 .
] The Measuremeni of Dielectric Constants .
be apossible variation of several per cent. in the value of the dielectric onstant of light flint glass from different makers at any CJiVen density .
7 .
The curve in fig. 4 is parabolic and agrees very well with , ( 18 ) where is the density , the dielectric constant , and The density , , is that of nearly the htest flint lass made .
From his to the value the of the dielectric constant is } ) ortional to he square of the change of density .
The relation between and which is shown by is also a qua atic unction expressed by , ( 19 ) 11 which .
5.\mdash ; Dielectric constant and index of refraction .
lint glass .
following table gives values of , and , and the eIived values of Prof. W. M. Thornton .
[ Apr. 28 , Table III .
From ( 18 ) and ( 19 ) , , ( 20 ) .
in which The value has been taken rather than , for the zero of the curve of density is nearer the former value , and if Hopkinson 's experimental determinations of in the empirical relation* are drawn with respect to the density , the density corresponding to upon a mean line is The value of the coefficient was calculated from Hopkinson 's experiments , being the slope of the , but from 18 , 19 , and 20 , .
If for the last figure in the foregoing table is taken , Moulded psotds .
8 .
Many substances for which the dielectric constant is required in practice cannot be conveniently worked when hard , but may be moulded .
Paraffin wax , shellac , and gutta-percha are examples .
To deal with these a mould was made by preparing two blocks of lead cm .
Between these a steel ellipsoid , 3 cm .
long , was placed , and the blocks crushed together in a verful press .
A semi-ellipsoidal cavity was in this way formed in each .
.Castings could not be made by pouring into the mould on account of the smallness of the cavity and the difficulty in removing air bubbles .
The substances moulded were all of a kind which becomes plastic on heating .
They were therefore warmed until , when placed between the plates and pressed , they took the form of the mould , any excess being squeezed out between the blocks .
These were then separated and the excess 'Roy .
Soc. Proc No. 182 , pp. 1\mdash ; 8 , 1877 .
1909 .
] The surement of Dielectric Constants .
removed , the blocks warmed , and the process repeated until no excess was found .
On being removed they were then very fair ellipsoids , but in every case required trimming by scraping away of some excess along the joint .
Substances which , like sulphur or resin , have a relatively high melting point , solidify quickly , and adhere to the mould even when smeared with oil or glycerine , could be more conveniently prepared by into cylinders , as described in SS15 and 16 .
In every case bc ; reat care had to be taken to dry the specimen and the vessel .
The latter was kept warm , and when it was necessary to admit air in order to remove the specimen a test , this done through tubes of phosphorus pentoxide .
WheIl moisture was suspected the specimenoe were always allowed to stand for several hours in the alternating field and vacuo , in order by the internal heating caused by dielectrio hysteresis to help to dry the surface .
Paraffin 9 .
Moulded ellipsoid of pure clear wax of density 0914.\mdash ; Melting point C. , diameter cm .
, length cm .
, volume c.c. , moment of inertia by comparison with standard glass cylinder .
cm.2 , mass gramme , eccentricity , temperature C. , vacuum 700 , frequency 86 .
The free period was seconds .
Period seconds in a field units .
The determining equation is and For freshly-moulded wax the highest value found was the frequency from GO to 84 decreased by per cent. Bces- ?
uax .
10 .
Ellipsoid.\mdash ; Dialneter cm .
, length cm .
, mass rramme , volume , density , temperature C. Free period 73 seconds .
seconds in a field units .
Here and so that This wax , from a cake , one of the Royal Society 's prize winners , vorked in warmed moulds , ) dried over phosphorus pentoxide for a week .
Prof W. M. Thornton .
[ Apr. 28 , 11 .
Ellipsoid.\mdash ; Diameter cm .
, length cm .
, volume , mass gramme , density , moment of inertia temperature C. Free period seconds .
Period seconds in field units .
Thus from which The shellac was dark orange flake before meIting , and was carefully heated until only the most minute bubbles of air were found in it when solid .
Like paraffin , shellac polarises quickly , the higher values which have been obtained by previous experimenters may have been in some cases from too contact .
12 .
I. Ellipsoid.\mdash ; Diameter cm .
, length cm .
, volume moment of inertia .
cm.2 , mass gramme , .
Free period seconds .
seconds in units .
The determining equation is from which , at a density In making this ellipsoid , a stick of the best sealing-wax was melted once , and from the plastic mass sufficient taken to form the ellipsoid between hot moulds .
Another ellipsoid was formed from the same wax , but the heating was inadvertently continued for half an hour .
The density of the new ellipsoid was , and , measured in the same way as above , the dielectric constant had risen to .
The wax ltained vermilion , button lac , resin , and " " mineral about half being compounds of mercury and lead and half etable resins .
-percha .
13 .
This ellipsoid as made from IJew thin clear gutta-percha tissue ( Silvertown best quality ) , dried , melted , and moulded .
In melting the tissue to a eneous mass , the heat was continued for about half an hour , and was such that bubbles formed in the plastic substance .
There is reason to believe that this was excessive , but the case is given as an example .
It is possible that something of the nature of carbonisation may have taken place .
The ellipsoid was dried for several months over phosphoric anhydride , and 1909 .
] The of Dielectric Constants .
then exposed to an alternating field of about 300 volts per centimetre for four hours in a vacuum of 700 mm. before a reading .
These precautions were taken , on of the values of obtained , to reduce the risk of the presence of moisture .
The measurements were\mdash ; Diameter cm .
, cm .
, volume , mass ) grammes , density .
Free period seconds .
Period in a field seconds .
) Hence giving ( see S 16 ) .
Chattcrton Co 14 .
The sample was of ordinary comnlercial icompounld .
Its composition was nown , though it apparently consisted of percha and pitch .
Ellipsoid.\mdash ; Diameter cm .
, length cm .
, volume , lmoment of inertia gm .
cm.2 , mass ramme , temperature C. , density .
Free seconds .
Period seconds in units .
Thus so that Didectric 15 .
Many dielectrics are difficult to shape into ellipsoids even mately true in form , which can be cast or turned to very accurate cylinders .
The electrical intensity within a cylinder in a field of force is , however , not uniform , and the reaction coefficients and have , strictly no definite meaning for cylinders .
If not more than three-figure accuracy is required , as in commercial testing , it may be assumed that the polarisation couple upon cylinders can be expressed by , where is of the form giyen in equation ( 2 ) .
If the cylinder is long compared with its meter , we may take the transverse coefficient , the value which it ] ) for infinitely cylinders .
The value of the coefficient , derived from equation ( 7 ) , is and is therefore not independent of the dielectric constant .
By measureProf W. M. Thornton .
[ Apr. 28 , ment of the polarisation couple upon cylinders of known dielectric constant , a series of empirical values of can be obtained , covering the range of required in practice , for all convenient ratios of ] rrth to diameter .
The following measurements wers made for the purpose of finding in this way , and and 7 are drawn from them .
linders ( infinite ) .
Length cm .
, diameter cm .
, .
Free period more than 5 minutes .
Period in a field ; from which and ( b ) Length cm .
, diameter cm .
, .
Free period negligibly great .
Period in seconds .
and ( c ) Length cm .
, diameter cm .
, .
Free period over a minute .
Period in seconds .
and Dense Flint Glass .
( a ) Length cm .
, diameter cm .
, .
Free period seconds .
Period in a field seconds .
Thus ; from which ( b ) Length cm .
, diameter 0225\amp ; cm .
, Free period 1751 seconds .
Period in seconds .
and ( c ) Length cm .
, diameter , cm .
, Free period ( on silk suspension ) several minutes .
Period in seconds .
and liight Flint ( a ) Length cm .
, diameter cm .
, Free period seyeral minutes .
Period in seconds .
Thus and ( b ) Length cm .
, diameter cm .
, Free period 69 seconds .
Period in seconds .
and Lenoth o cm .
, diameter cm .
, period seconds .
Period in , 81 seconds and The surcment oConstants .
Para .
The value previously found for paraffin wax is that for annealed wa free from structure .
It was found to be practically ) to obtain ] inders free from initial strain .
The method adopted to melt the wax and draw it into a short piece of barometer tube fhtly wa1med to prevent the solidifying at the circumference and shearing .
With little practice 11 coefliciel ) ith and K. letter of cylinder .
cylinders were made , the diameter of which did not differ at any point by inch from the mean .
( a ) Length ; cn diameter .
Free period seconds .
eriod in field seconds .
which with oives o ( b ) Length cm .
, .
Free period seconds .
Period in field 17 seconds .
and with ) Prof. W. M. Thornton .
[ Apr. 28 , ( c ) Length cm .
, diameter cm .
, .
Free period 10 seconds .
Period in seconds .
with Fig. 6 is a record of these experimental values of , but in finding the appropriate constant to use in any particular case , fig. 7 will be more useful .
It is not necessary , however , to know any of these curves with great accuracy , for in working out the following tests the procedure was to any likely value for , having to the dimensions of the specimelJ 'FIG .
7.\mdash ; Change of with for different ratios of length to diameter of cylinder .
and the nature of the substance , and to calculate the corresponding dielectric constant , then from the curves to find for the new value of , and in this way by successive approximation to reach a point where the became less than experimental error .
Ths use of the curves facilitates the finding of by this method of trial and error .
The value of taken as determining that of was usually found at the third approximation .
When the point for the iven ratio reed with the curves of figs. 6 or 7 it taken as final .
1909 .
] The Measurement of Dielectric 16 .
The cyJinders were shaped by and .
That of , was turned from a sheet 1 cm .
thick of the best material and a value .
A sample of lber , was turned ) holding fine sand-paper between the fingers upon the specimen , and gave .
A cylinder turned from a piece of old ?
, gave .
For , a pnle clear yellow sample melted and cast , cooling slowly , .
For prepared in the same way , a clear light brown sample , .
Sublimed sutphur ; , heated to point and cast in a ftSS tube , slowly with radial structure not quite to the centre , .
A gntta-percl , cylinder , , was prepared gently a small piece cut from a rod of Silvertown " " super\ldquo ; quality warlned over an incandescent lamp , yiving K. A cylinder of pure soft black , obtained from the makers and tested in fields of different intensities to eliminate electrostriction , gave finally Olirc Oil .
17 .
The had a density at .
It was dried by with fused calcium chloride .
In order to eliminate as far as possible the influence of the necessary envelope , a tube was made by rolling a picce of arette paper 2 cm .
square around an oiled steel cylinder cm .
diametel .
The oil caused the three layers used to adhere by capillarity , and the tube was dipped in the oil and ound up tightly on the cylinder .
Made in way a tube has sufficient stiffness to retain its shape perfectly when suspended filled with liquid .
The measurements were as follows:\mdash ; tTth of , diameter cm .
, thickness cm .
, luass of tube gramme , of oil contained gralnme , ratio of length to bore .
Free period of paper tube alone 7 seconds .
Period in field 6 seconds .
Free period full of seconds .
in a field seconds .
The moment of inertia of the cylinder alone and of the oil alone gm .
c The polarisation couple is therefore dyne-cm .
per radian .
The volume of the Assuming by trial a depolarising factor , we have from which Prof. W. M. Thornton .
[ Apr. 28 , The average value of all the determinations quoted by Winkelmann is .
Hopkinson found .
The commonest adulterant of olive oil is cottonseed oil , which has a dielectric constant of Heavy " " White\ldquo ; Paraffin Oil .
18 .
Used for electric condensers , having an exceedingly high resistivity .
Density , at C. Paper tube cm .
long , cm .
diameter , mass of liquid used gramme , ratio of length to diameter .
Free period of tube filled with oil seconds .
Period in a field seconds .
Polarisation couple dyne-crn .
radian , and the equation with giving For this oil 19 .
Proceeding as before , distilled water was enclosed in a paper tube cm .
long , cm .
diameter , or , for ths wettirg of the paper , 0.24 cm .
ass of water in tube gramme .
Free period , full , seconds .
Period in a field seconds .
Polarisation couple dyne-cm .
radian .
From this the cylinder had apparently an infinite dielectric constant , .
which means that at the frequency of 80 a second electrical conduction has time to come into full operation .
C'ollected Results .
20 .
For convenience of reference the results obtained in the paper are collected below , together with some previous determinations at frequencies not widely different from those of the paper .
There is on the whole close agl.eement between the new values and those quoted .
The differences .
between the best previous deterlninations are from 2 to 5 per cent. By the method of the paper accuracy of one-tenth per cent. is obtained without difficulty except that of moisture or other conducting agency .
The work has .
been extended to the slow polarisation of dielectrics in unidirectional fields .
I wish to thank my late students , Mr. O. J. Williams , B.Sc. , and Mr. B. J. M. Lane , B.Sc. , for much help , especially in the development of the method , and my friend , Dr. T. H. Havelock , for several valuable gestions .
1909 .
] The Measurement of Dielectric Constants .
Table .
( SO to S5 .
)

