]\gt ; Conductivity and Dnelectric Constant of ) space which does not contain any such extraneous feature .
But also , viewed from the .
side , in extensional analysis of this type appropriate to isotropic Euclidean manifolds , the vector operator and its powers are the fundamental ones , so that it is not really very surprising that gravitational analysis can be linked up with a theory of deformable space .
Finally and again , this re-statement of theories of relatiyiGy as relations of correspondence in space and time , by aid of uniform auxiliary manifolds of higher dimensions , may appear retl.ograde : in the earliest phase relativity was just such correspondence .
* But it has the advantage of getting rid of the very puzzling auxiliary apparatus of local timekeepers , and their of rate when moved about .
And , moreover , it is not , in fact , possible to do without a scheme of space and time ; relativity merely asserts in arious ways that its final specification so far eludes our powers a large nuulber of partial modes of specification can be employed indifferently over a wide range of problems .
On the Variation with equency of the ductivity and Oielectric Constant of Dielecfrics cquency O tions .
By G. , D.Sc .
, ( Communicated by Prof. J. A. Fleming , F.R.S. Received June 12 , 1919 .
) 1 .
Introduction .
Although we have a certain amount of knowledge regarding the val.iabion of the couductivity of dielectrics with frequency for tively low frequencies , within the telephonic range , say , up to per second , where the conductivity is in ) eneral a linear function of the frequency , it cannot be that any information exists at present as to what happens when we extend the range of frequencies up to those employed in work .
That energy is dissipated in used in oscillation circuits has been known since 1861 , when W. Siemens pointed oub that the glass of a Leyden jar became heated on charge and discharge .
Threlfall , extending *Cf .
'Ether and Matter , ' chap .
xi ( 1900 ) .
Here in cognate manner the fivedimensional space-time foundation is introduced in order to provide the necessary standards of time and space , which , even though provisional , are indispensable .
'Berlin Akad .
Monatsber October , 1861 .
'Phys .
Rev vol. 4 , p. 57 , and vol. 5 , pp. 21 and 65 .
VOL XCVI.\mdash ; A. 2 Dr. G. E. Bairsto .
Conductivity Dielectnc the early experiments of Arno , working with a rotating electrostatic field , found that umder these conditions there was no hysteresis loss at per second in the dielectrics he employed : ebonite , lass , and sulphur .
At somewhat lower frequencies of the order of a million a second , several observers have made measurements of the energy dissipated , and find that condensers have an appreciable decrement .
Reference may be made to the following : W. Hahnemann and L. Adelmann , .
Dupreux J. J. Stockley , U. Wien , S J. A. and G. B. Dyke .
W. AusCin,1 E. F. W. Alexanderson .
Most of the messurements at working voltages , so it is impossible to say how much of the energy Joss is due to brush dischsrges and how much to a true dielectric conductivity .
Moreover , the measurements have generally been confined to soDle particular frequency .
The object of the experiments to be described ) elow was to measure the conductivity of the dielectric over a wide of frequency , employing continuous oscillations of sine wave form , and of low voltages .
Apart from the importance of the subject in connection with the properties of dielectrics in general , measurements of this kind are of value in another direction , , in the theory of the propagation of electromagnetic waves over the earth 's surface .
Before the theory as developed by Zenneck and others can be applied to practice , we require to know the values of the dielectric constants and conductivities of the constituents of the earth 's crust , and the explanation of many anomalous effects which at present are but little understood , would in all probability be cleared up if we were in possession of these data for certain regions of the earth 's surface .
Loewy , with this object in view , has made an extensive series of such measurements , but only with direct currents ( D.C. ) , which we now know ives values of the conductivities which bear no relation to conductivities as " " Losses in Condensers , and Damping in Wireless Telegraph Circuits ' Electrotech .
Zeit vol. 28 , pp. 988 and 1010 ( 1907 ) .
{ " " Oscillatory Discharges of Condensels Iectrician , ' vol. 38 , p. 107 ( 1909 ) ; or ' Science tracts , , No. 906 ( 1909 ) .
" " Eli.ect of Tempetature on Dalnping in ss ondeusers , \ldquo ; 'Schweiz .
Electrotech .
Zeit vol. 6 , p. 309 ( 1909 ) ; or ' Science ) , No. 1441 ( 1909 ) .
S " " Damping in Oscillatory Circuits ' -A .
der Phys vol. 29 , 4 , p. 679 ( 1909 ) ; or Science acts , .
" " Energy in High cuits , \ldquo ; 'Proc .
Phys. Soc vol. 2:3 , p. 117 ( 1911 ) ; or ' Science Abstracts 14 , No. 384 ( 1911 ) .
" " Condenser Losses at High requencies , \ldquo ; ' Journ. Wash. Ac .
Sci vol. 1 , p. 143 ( 1911 ) ; 01 ' Science Abstracts , ' 16 , No. 199 ( 1912 ) .
' ' tric Hysteresis at Radio-frequencies , \ldquo ; 'Proc .
Inst. Badio .
' vol. 2 , p. 137 ( 1914 ) .
'Ann .
der ' vol. la , p. 126 ( 1911 ) .
Constant of Dielectrics for High-Frequency Oscillations .
365 measured with alternate currents ( A.C. ) .
Even at telephonic frequencies the alternate conductivity is much greater than the D.C. conductivity ; at radiotelegraphic frequencies the differences become enormous .
As a small contribution to the subject , the author has therefore included two typical constituents of the earth 's crust in measurements , , slate and marble .
2 .
atns and Nethods of surement .
The method of measurement adopted is based upon the resonance a leaky condenser .
Two condensers , one the unknown lealey condenser under test , and the other a standard air condenser shunted with a valiable high resistance , are alternately placed in series with an inductance , and each adjusted for resonance with a primary source of oscillations , and also for the same maximum current in both .
Under these circumstances , the capacities of the two condensers will be the same , and the shunted resistance will be equal to the reciprocal of the conductance for the frequency used during the experiment .
Now it is exceedingly difficult to obtain a perfectly source of undamped alternating currents of high frequency .
The condition of equality was therefore determined by .
means of the specially ldesigned commutator shown in fig. 1 .
This commutator , driven by an S .
SlndinS for HiSh ency Mmenfs .
electric motor , alternately , and for equal times , inserts either of the two condensers in series with the inductance , forming a secondary to the primary .
Two thsrmo-junctions , and , are arranged , one in each circuit , in such a manner that the 's produced by them oppose one As this work was carried out in 1913-1914 , three electrode valves of sufficient power were not available .
Dr. G. E. Bairsto .
Conductivity and Dielectric another .
When the galvanometer placed in the circuit commo to both gives no deflection , then we know that the two conditions mentioned above are fulfilled .
This arrangement being a null reading one , makes it independent of any irregularities in the source of the oscillations .
It also enables us to approach zero from either side , for it is apparent that , according as one conductance is greater or less than the other , so will its corresponding resonance current be less or greater than the other , and therefore one E.M.F. to preponderate over the second .
For ease of construction , the commutator was built up of six brass rings , mounted on red fibre discs supported on a steel spindle .
The two end rings as well as the two middle ones were split in half .
One half of each split ring was electrically connected to the neighbouring whole ring in the manner shown in fig. 1 .
Six small gauze brushes lead the current from the various circuits , and in such a manner once during each revolution of the conimutator , and for equal times , each condenser is inserted in the circuit .
The dielectrics were made up into small condensers with tinfoil interleaved between the sheets , each condenser having a capacity of from 300 to 2000 .
The condenser under test being of a constant .
value , a series of inductances from to millihenry were used , depending at what particular frequency measurements were being made .
The adjustable standard of resistance being very high , of the order of half a megohm , a wire resistance was quite out of the question , especially at high frequencies .
A water resistance was therefore used .
This consisted of a block of paraffin wax , in which a circular channel was cut , about 15 cm .
in diameter , the area of channel about 1 sq .
cm .
A plug of wax was then made in the channel near one terminal ( fixed ) ; the other terminal consisted of a small thin plate of copper attached to a radial arm .
This enabled one to obtain a continuous range of resistance from zero up to a maximum depending upon circumstances .
Either distilled water or slightly acidulated water was poured into channel , the maximum resistance upon the particular dielectric under test , and on the frequency used during the test .
The highest resistance was about 2 megohms .
The ement which was kept a glass case was calibrated at the beginning and end of each day 's work .
The thermo-junct.ions were of the type devised by Dr. J. A. Fleming .
It is necessary that they should be both of exactly the same resistance in the heater wire , in order that the total resistance decrement of the two circuits should be the same , and what is more , the magnitude of the heater resistanco should be as small as possible , consistent with sensitiveness , in order to cul Constant of Dielectrics for High-Frequency Oscillations .
867 down the resistance decrement .
It may be pointed out that if we are dealing with a constant .
induced in a resonating circuit having only resistance decrement , the deflection of a thermal ammeter is inversely proportional to the effective resistan ce of the circuit .
As the effective esistance of the inductance coil was very small , we may take as being equal to the resistance of the heater in the thermo-junction .
Then if I be the resonance current , we have ( when the circuit is tuned ) and heat deyeloped heat developed deflection Hence the snlaller make , the greater will be the deflection .
There are , , two factors which limit the bove .
cannot be made too small , ' this means a thick wire and therefore a larger surface for radiation and a lower temperature of the thermo-junction .
Again , a small means a voltage across the condense ] , and this we decided must be kept low in order to prevent any ility of brush with the thin dielectrics used in making up the condensers .
In the above we have ected the decrement of the condenser itself , but taking this into account , it would still remain true that the smaller is the greater will be the galvanometer deflection .
It therefore suggests that we should make the heater of the metal having the best electrical conductivity , viz. , silver .
The heaters finally used were mm. diameter , and had a resistance of ohm .
So , at the highest frequency used , , and a of 100.0 micro-micro-farads , the factor of the inductance and heater would be The total faotor of the entire oscillating circuit , including the commutator , and apart from the condenser itself , was approximately twice this value , about .
Since this is much less than the power factors of the condensers tested , it follows that the greater part of the decrement of the oscillating circuits resided in the condensers themselves , and the experiments were therefore conducted under the best possible conditions as regards the production of resonance .
The junctions gave with a Paul 10-ohm galvanometer a maximum deflection for ampere .
Under the above circu mstances the E.M.F. on the condenser would be 8 volts .
At the lowest frequencies , from half-a-million downwards , in order not to increase this voltage above about 10 volts , a second Dr. G. E. Bairsto .
Conductivity Dielectric pair of junctions were used , having each a resistance of about ohm , iving a resistance decrement of approximately the same order of magnitude as before , because we should be a proportionately larger inductance .
A second equirement of the thermojunctions is that they should be exactly equal as regards sensitiveness , , must give equal deflections on the galvanometer for equa.1 currents in either .
With a little manipulation of the solderedjunction it is possible to get them so to within 4 or 5 per ceut .
The remaining differetlce can be compensated for by shunting the more sensitive one with a tance until they give the same deflection ( see fig. 1 ) .
it is possible to obtain no reading on the galvanometer with the commutator running , and definite equal currents in each circuit , it does not follow that there will be a balance for all equal currents , either below or above .
This necessarily follows from the impossibility of obtaining both the heating and radiation constants of the two thermo-junctions exactly similar all temperatures .
Fig. 2 shows how close we may get to perfect equality under all conditions .
The abscissae give the current in either of the thermo .
She0 of-BaIance Deflection on Different Currenrs in the ncfions .
junctions , and the ordinates the out-of-balance deflection on the galvanometer .
It will be seen that we have a zero deflection for 63 milliamperes , which was the maximunl safe current in the junction in question .
This corresponds to a ranometer deflection of 100 divisions .
At 40 milliamperes we have a maximum out-of-balance current of } divisions ; this , however , is very small compared with the total deflection , and only introduces a possible error of between 1 and 2 per cent. The junctions could stand an overload current up to 80 milliamperes .
As the source of oscillations a graphite arc in air , shunted with inductance and capacity , was employed .
Above about 200,000 per second , it was Constant of Dielectrics for equency Oscvllations .
369 found impossible to obtain sufficient power in the secondary circuit without making the coupling too .
Placing the arc in compressed air at about atmospheres pressure enabled a much greater power to be taken out and increased the available frequency to about a million or so .
Hydrogen was also tried as recommended by Poulson and others , but while the oscillations are stronger , they are much less steady , and , what is much more unsatisfac tory , it gives a very impure wave , and the circuit is exceedingly difficult to tune , as the frequency varies with the arc current .
The latter point has also been observed by L. W. Austin .
* The graphite arc in compressed air , on the other hand , for some reason which is not very plain , tends to suppress all the higher harmonic components of the main oscillations and gives a nearly pure sine wave .
The limit at which one can obtain steady oscillations with the Duddell arc is about 500 , second .
Beyond this the arc behaves very erratically and constantly fluctuates ; at 2 , oscillations are only intermittent .
The author has succeeded in obtaining fairly steady continuous oscillations up to about per second by means of the arrangement of circuits shown in fig. 3 .
The circuit A is tuned to a frequency at which steady oscillations can be obtained , i.e. , from 500,000 to , and circuit , placed so that it can retiits A a placed so as no mutual action on each other .
FiS5 .
Scheme of Parallel to Si})e HiSh Frequency enfs .
have no mutual action on , is tuned to a harmonic , either second , third fourth , or fifth of circuit L. Under these ustances , oscillatory currents of these different frequencies can easily be obtained , circuit A acting with a steadying influence .
If circuit A is tuned to a frequency of and to , and the oscillations started , it is found that if the lower frequency circuit A is cut out , the oscillations are immediately stopped in the higher frequency circuit B. On making circuit again , the oscillations in P ) ' Bull .
Bureau tandards , ' vol. 3 , p. 325 .
Dr. G. E. Bairsto .
onductivity Dielectric immediately restart .
It was only by the use of this arrangement that it was possible to reach the extreme limit of per second , to which the obseryations were for the last three dielectrics mentioned below .
The actual method of procedure followed in making the measurements was follows:\mdash ; With the commutator at rest each condenser was adjusted for .
maximum resonance current .
The commutator was then set running and the out-of-balance current flowing the galvanometer brought to zero varying the resistance shunted across the standard air condenser .
This adjustment in the of the ] er dielectrics would need a slight alteration in capacity of the standard air oondenser to bring the circuit into resonance again .
A final readjustment of the shunted resistance would then bring the circuits into perfect equality .
If and be the capacity and conductance of the unknown condenser under test , the capacity in parallel with it to produce resonance , the capacity in the standard circuit to produce resonance , and the shunted the , then and From and knowing the of area to thickness the dielectric can calculate and the specific altel.nate current conductivity , and the dielectric constant .
The frequency from 400 , second and upwards was measured with a Fleming cymometer .
The lower range was determined with another made-up cymometer , of a variable air condenser , reading from 200 and 3300 micro-micro-farads combined with a single layer solenoidal coil .
The inductance of the latter was calculated by formula and found to be millihenry ; ured value at telephonic frequencies by the Anderson bridge method was found to be equal to millihenry .
This value requires a correction when used at high frequencies .
This colTectionvas applied by Heaviside 's formula:* microhenly , where equals number of turns , equals radius of coil , equals diameter of wire and equals the of coil .
These quantities were respectively and and therefore gave a correction of millihenry .
The value of the coupling inductance at high frequencies was therefore taken as being equal to or ( Electrical Papers , vol. 1 , p. 356 .
Constant of Dielectrics for -Frequency .
371 3 .
Discussion Results .
In the Tables are given the experimental results obtained for the following dielectrics :Blotting paper , lass , guttapercha , vulcanised indiarubber , marble and slate .
The general nature of the results is to show that the conductivity increases enormously as the frequency increases .
The linear law found by Dr. Fleming and .
Dyke*connectin the two quantities at telephonic frequencies is now olonger obeyed , but the tends to a and theu seems to decrease We will now discuss the behaviour of each particular dielectl.ic in turn .
( a ) \mdash ; Table I gives the results for a condenser made up of white .
paper which had been previously dried in an oyen .
The first three columns give the values of the measured conductance capacity at given frequencies .
The last three colulnns the values of the dielectric constant , specific conduclivity , and ( approximately proportional to the power-factor ) as calculated from these quantities and C. The observations are depicted in fig. 4 .
It will be seen that the conductivity at first proportional to the frequency quickly rises to a maximum at about 600 , second , and then decreases .
In the showing the variation of dielectric constant with frequency we have a sharp fall at Table I.\mdash ; Dielectric : Dried Blotting Paper .
Area sq .
cm .
Ihickness cm .
The numbers in the brackets tlrc the values of the factor ( alculated by subtracting the ordinary direct current from the total conductance .
' Journ. Inst. Elec .
Eng. , vol. 49 , p. 323 ( 1912 ) .
Dr. G. E. Bairsto .
Conductivity and Dielectric telephonic frequencies and then a gradual decrease as the frequency increases .
the curve we have , for the telephonic range of frequencies , ency of he Consfanf of BIofinS Paper .
separated out the ordinary direct current component of the conductivity which is superposed on the true alternate current conductivity .
Thus if total measured conductivity , and that due to the direct current component , then represents the power-factor due to the alternate current conductivity alone .
These values are given in the brackets in column 4 .
At the higher frequencies is too small compared with to have any influence on .
The result of these calculations is to give an intercept on the axis of conductivity indicating that from up to about per second we have a constant power-factor ( see fig. 4 ) approximately equal to .
Above this there is an increase in the power- factor proportional to the frequency , that is to say , may be represented as the sum of two quantities:\mdash ; or , what is the same It is apparent , therefore , that we are dealing with two separate effects , one implying a loss proportional to the frequency , and the other a loss proportional to the square of the frequency .
At telephonic frequencies the first term is the only one of importance .
At higher frequencies the second has the more influence .
The first implies a loss which is independent of the time taken for taking the dielectric through a complete cycle , and we will , therefore , tiyely designate it a ' dielectric hysteresis\ldquo ; loss .
Constant of Dielectr for High-Frequency Oscillations .
373 The second which is a loss depending upon the time taken to complete the cycle will , for reasons to be given later , be called a " " viscous\ldquo ; loss .
A comparison of the different quantities for low for high frequencies is instructive .
Thus , the maximum power-factor is times as great as the power-factor at telephonic frequencies .
The calculated value of the D.C. component is mhos per cubic centimetre .
Therefore the maximum conductivity at high frequencies is , or 1020 times as great as the conductivity with steady clirrents .
( b ) Crown Glass.\mdash ; The condenser used for these experiments was the thin -glass used for the cover-gasses of microscope slides .
The results are given in Table II and are plotted in fig. 5 .
The general nature of the curves is similar to that of the ones for blotting-paper .
The linear part of the conductivity curve extends over a wider range , and the point of conductivity is a little lower , viz. , 510,000 per second .
The calculated value of the D.C. conductivity is , and therefore the maximum alternating current conductivity is 30,500/ 12 , or 2500 times as large .
The curve is slightly different .
It will be seen that it has a relatively larger intercept on the axis of co-ordinates , implying that hysteresis contributes more to the conductivity than viscosity , for the maximum is only 50 per cent. greater than the constant corresponding Table II.\mdash ; Dielectric : .
Area sq .
CIII .
Thickness cm .
thickness ctric ctivi.ymicro .
* The numbers in the brackets are the values of the power-factor calculated by aoti the ordinary direct current component from the total conductance .
Dr. G. E. Bairsto .
and Dielectric to low frequencies .
The further course of the curve is interesting for it eventually falls to a value only half that of the intercept for , and FiS5 .
ifh ency or the ) , a Constant of Crovfn GIass .
is still rapidly decreasing .
This indicates that the hysteresis loss per cycle which is constant at low frequencies , decreases and to become zero at highel frequencies , viscosity only being of importance .
During the observations on glass , means were taken to see whether the alternate current conducCivity varied with the voltage .
This was done by the commutator for some particular coupling of the pl.imary and secondary and then altering it by either increasing or decreasing the distance between them .
No alteration of the balance was required although the voltage conditions were such that at the lower limit only about 2 volts were across the condenser , and at the limit a perceptible noise inside the condenser indicated that had just set in .
This confirms for high what B. Mollasch found at telephonic frequencies\mdash ; the energy loss in a dielectrio is strictly proportional to the square of the voltage , provided brush discharge is eliminated .
* ( c ) Vulcanised Indiarubber.\mdash ; This dielectric is one.of the two ( guttapercha being the other ) out of all those tested by .
Fleming and .
Dyke , whose conductivity they fouud was incapable of being represented within the telephonic range of frequency the linear law " " Dielectric Losses ' Anu .
der Phys vol. 22 , p. 905 ( 1907 ) ; or ' Science Abstracts , ' 10 , No. 897 ( 1907 ) .
Constant of Dielectrics for High-Frequency .
S75 increasing much more rapidly .
This is illustrated by the first three rows of figures in Table III ; the power-factor for is , whereas for it is nearly twice as great , .
The reason for this is clearly seen on inspection of fig. 6 , for there we see that Table III.\mdash ; Dielectric : Vulcanised Indiarubber .
Area sq .
cm .
Thickness cm .
ecQnd .
14 Fig 6 .
of he a befric canisod Iarobber .
Dr. G. E. Bairsto .
Conductivity and Dielectric the intel.cept on the co-ordinate axis is very small , so that it is over a very short range , say up to or per second , that can be called even approximately constant ( by interpolation being about ) .
Ths viscous loss is , in fact , of so much greater importance , that at 350,000 per second , the power-factor is , or 23 times as large as at very low frequencies .
Hysteresis is therefore , comparatively speaking , absent .
For a very considel.able range , from up to per becond , is exactly a function of the f.requency , or The conductivity has a nlaximum value at about 560 , second , and increases so rapidly with frequency that it is 16,000 times as great as at .
There is also a subsidiary maximum at 350,000 per second , a hump on the curve , and indicating the existence of components in the dielectric .
It remains now to point out a remarkable reement between the naximum power-factor observed above on the power-factor-frequency curve and the maximum power-factor observed by Fleming and Dyke for vulcanised indiarubber at certain low temperatures .
They found that at about C. the conductivity at a yiven frequency rose very rapidly and reached a maximum of considerable value , fourteen times that at ordinary temperatures .
The power-factor at this point was , and was independent of the frequency .
Now , the power-factor observed above for the first hump on the conductivity-frequency curve is , which agrees fairly well with the former figure .
If we allow for the slight change of capacity with frequency and temperature , the agreement is still closer .
( d ) \mdash ; This dielectric is another material that does not follow a linear law in the variation of conductivity with frequency within the telephonic range .
In the last case ( vulcanised indiarubber ) , it was due to the fact that the hysteresis loss was negligi'ole in comparison with the viscous loss .
In the case of guttapercha , however , the divergency results from quite a different cause , being due to the presence of a component in the dielectric having a frequency of maximum conductivity which is only just olltside the telephonic range .
Thus , if we inspect Table , we see that the power-factor increases from at second up to at 15 , second .
Tbis was the upper limit imposed by the limits of audition on measureluents made with the bridge and telephone .
At , however , has fallen to , implying the existence of a slight hump on the conductivity-frequency curve at 50 , second .
It was not possible to make any measurements at lower frequencies , because with the present experimental arrangements they would entail a considerable Constant of Dielectncs for High-Frequency Osciltations .
377 TabIe I : Guttapercha .
Area sq .
Thickness cm .
voltage on the condenser to obtain sufficient .
For this reason , the curve has had to be filled in tentatively by means of the dotted line .
From about 300 , second and upwards , the conductivity and curves ( see fig. 7 ) are precisely similar to those of the dielectrics already considered , 11th equency of fIapercha .
Dr. G. E. Bairsto .
Conductivity and Dielectric in that they rapidly rise up to a maximum and decrease again ; the frequency for maxinlum conductivity in this case is about 750 , second .
By interpolation the power-factor for very low frequencies is , and it will be seen that if we prolong the base of the second hump of the curve backwards in a straight line it will also cut the vertical axis at a point very close to .
In other words , each of the two components in the dielectrics , except for frequencies in the neighbourhood of its own peak , has little influence on the conductivity at other parts of the curve , both with a common arising from hysteresis .
As in the previous case the maximul1l power-factor observed at the first hump on the power-factor curve agrees approximately with the maximum power-factor observed at telephonic frequencies by Fleming and Dyke at a certain low temperature They foumd a maximum power-factor at this temperature , which was pement of frequency and amounted to The maximum power-factor observed above on the conductivity-frequency curve at ordinary temperatures is .
This coincidence is very striking and may be of interest in developing the theory of dielectrics under .
Fs .
( e ) Ncarble.\mdash ; The next two dielectrics to be considered are typical constituents of the earth 's crust .
In the case of marble the curves present some points of difference from those already considered .
The experimental results are given in Table and depicted in fig. 8 .
They refer to a condenser made up of 20 square marble slabs*about cm .
thick , and effective area coated with tinfoil of 2400 sq .
cm .
Only 16 were effective , the remaining 4 being used on the outside of the condenser to carry the lines of force from the outside faces of the last tinfoil electrode .
The whole was bolted ether with two bolts .
As with the other dielectrics , the dielectric constant at first falls rapidly within the telephonic range , and then slowly decreases as the frequency aches those used in radiotelegraphy .
The conductivity-frequency curve however is different , in ihat after the usual linear stage at the beginning , the curve in approaching its maximum gradually bends over , instead of turning upward .
The result of this is that the power-factor throughout the whole range decreases with increase of frequency .
At the very highest frequencies it may be noticed that , as regards powerfactor , marble is better than any of the other the lowest reached being The calculated value of the D.C. conductivity is 310 mhos per have to express my indebtedness to Dr. J. A. Fleming , F.R.S. , for placing these marble slabs at my disposal .
Constant of Didectrics for High-Frequency Oscillations .
379 cubic centimetre , while the maximum A.C. conductivity is 23,000 , the ratio being 74 .
Table : Marble .
Area sq .
cm .
Mean thickness cm .
* The numbers in the brackets are the values of the power-factor calculated by subtracting the ordmary direct current component from the total conductance .
FiS8 Frequency of the } Consianf of Marble .
( f ) Slate.\mdash ; The results for the slate condenser are given in Table VI and depicted in fig. 9 .
The frequency was taken up to 3,000,000 cycles per VOL. XCVI.\mdash ; A. Dr G. E. Bairsto .
Conductivity and Dielectric second , and it will be seen that the conductivity is still rising , although there seems to be a tendency for the curve to flatten and turn over again .
At the time there are several irregularities on the curve , which seem to indicate that there are several component materials in the slate .
The most noticeable point about the dielectric constant curve is its enormous drop at low frequencies , being about 50 at , whereas at 500 , per second , it has fallen to about 10 .
This large drop in causes the power-factor to be approximately constant over a wide of frequency , whereas if we calculate , which is proportional to the amount of work done per cycle due to the alternating conductivity , it will be seen that this quantity after approximately Table .\mdash ; Dielectric : Slate .
Area sq .
cm .
Thickness cm .
Constant of Dielecfrics for High-Frequency OsciUations .
381 constant within the telephonic range decreases with increase of frequency from about mhos per cycle per second to about 1 Fre ency Million second .
FiS9 .
Freqaency of he , a Diebcfric Constantof SIafe .
The maximum A.C. conductivity is mhos per cubic centimetre , and the calculated value of the D.C. conductivity is , hence the ratio of the two quantities is 1700 .
1 .
Measurements of the conductivity and dielectric constants of dry blotting paper , glass , vulcanised indiarubber , guttapercha , marble , and slate are given for alternating currents of low voltage and sine wave form , and for a wide range of high frequencies .
2 .
It is shown in all cases that there are present two independent sources of loss .
One , a hysteresis loss , is usualiy the only one that is important at telephonic frequencies .
The second , a viscous loss , has more influence at higher frequencies .
The first loss is one which is independent of the time taken for a complete cycle , while the second depends upon this time and at low frequencies gives rise to a loss which varies as the square of the frequency .
3 .
The linear law found at telephonic frequencies connecting with frequency , is no obeyed at high frequencies , but gradually rises to a maximum , and then decreases again .
This rise to a maximum is very rapid for some substances , e.g. , guttapercha and vulcanised indiarubber while with slate and marble , the maximum is approached by the curve bending over instead of turning up .
This leads to a very flat curve for which is practically constant over a considerable range of frequency .
4 .
The maximum A.C. conductivity is very much greater than the D.C. Dr. T. R. Merton .
conductivity .
In the case of glass , it is 2500 times as large , while in the case of vulcanised indiarubber , the maximum A.C. value is even 16,000 times as large as the A.C. value at second .
5 .
The constant hysteresis loss per cycle at low frequencies tends to become zero at very high frequencies .
6 .
The dielectric constant after a rapid drop at low frequencies , slowly decreases as the frequency rises .
In conclusion , I wish to express my indebtedness to Dr. J. A. Fleming , F.R.S. , for placing the resources of his laboratory at my disposal , and for much valuable advice during the course of the work which was carried out the years 1913 and 1914 .
On the pectrum of Hydrogen .
By T. B. MERTON , M.A. , D.Sc .
( Communicated by Prof. A. Fowler , F.R.S. Received August 15 , 1919 .
) [ PLATES 6 AND 7 .
] Theoretical investigations of the origin of spectra in relation.to the structure of the atom have concentrated especially on the spectrum of hydrogen , on account of the supposed simplicity of the hydrogen atom .
They have , however , been confined almost exclusively to the Balmer series , and have ignored the difficulties which arise from the fact that hydrogen possesses another spectrum , usually known as the secondary spectrum , which is of great complexity , and the co-ordination of whose lines into recognised bands or series of lines is still in a very unfinished state .
The investigations of Buisson and Fabry , which the physical widths of spectrum lines were measured with the interferometer , refer at least a of the lines of the secondary spectrum to the hydrogen atom , and the complications introduced into theoretical investigations cannot therefore be } ) artially waived by the assumption that the molecule is concerned in the production of the secondary spectrum .
As regards the relation of the two spectra , there is abundant evidence of a fundamental diffel'ence in their origin .
In many celestial spectra the lines of the Balmer series constitute one of the most prominent features , whilst the identification of lines of the secondary spectrum is at least extremely 'Journal de Physique , ' vol. 2 , p. 442 ( 1912 ) .

