]\gt ; On Electrostatic Solid Insulators .
By Prof. A. WlLSON , F.B.S. , , London .
Received Read The following paper contains an account of a series of experiments on the variation of the capacity of condensers having solid dielectrics with the potential difference and the time .
The influence of temperature on the results has also been investigated .
It is not proposed to any detailed account of previous work on this subject here .
A very full list of papers is iven in a paper by Schweidler entitled " " Studien uber die Anomalien im Verhalten der Dielektrika ' Ann. der Physik , ' No. 14 , 1907 , p. 711 .
Let A and be two guard ring parallel plate condensers , the dielectric in A being air and in any insulator .
Let the two guard be permanently connected to earth , and the two small plates permanently connected together .
Suppose initially the large plates and small plates are connected to earth , the small plates then disconnected , and then the large plate of the condenser raised to a potential and that of A to a potential .
After a time , measured from the instant at which the plates were charged , the total on the small plates will be given by the equation where is the capacity of the condenser that of the capacity of the backs of the small plates and their connecting wire , and the potential of the small plates .
The charge will be very small if the dielectric in is a good insulator , and we may put it equal to , where is a constant approximately .
Hence .
( 1 ) Suppose , now , another experiment is done in which both large plates are charged to the same potential , and let the resulting potential of the small plates be just after charging the large plates , then .
( 2 ) is a function of , but it does not vary very much , so that if ' is large compared with we may take to be constant without sensible error , in which case ( 1 ) and ( 2 ) give .
( 3 ) When is large will vary very slowly with the time , so that then can Prof H. A. Wilson .
On Electrostatic [ May 18 , be found from the variation of with the time .
Then knowing can be found as a function of the time .
If during an experiment is altered to , then we have where is the potential of the small plates after the change in C. Hence Thus so long as is small compared with , the difference between and will be independent of .
This enables a change in to be allowed for very easily .
For a particular value of equation ( 3 ) shows that depends on so that if is independent of then should be independent of If two condensers are of nearly equal capacities , will be small compared with , so that we may make large , say 500 volts , and measur6 with a quadrant electrometer to , say , volt .
This will enable variations in of 1 part in 10,000 or more to be measured .
The experiment described provides therefore a sensitive method of determining small variations in the capacity of B. The apparatus used is shown diagrammatically in the figure .
Alarge battery of small secondary cells was connected through a reversing commutator to two megohm resistances and in series .
The guard rings of the two condensers were connected to the junction between and .
The large plates of the condensers were connected as shown respectively to the battery ends of the resistances .
The small plates were connected together and to one pair of quadrants of a quadrant electrometer E. The other pair of quadrants was connected to the guard rings and to the case of the instrument .
The two pairs of quadrants could be connected together by means of a wire dipping in a mercury cup P. The point of junction between and was connected to earth through the gas mains , while the battery and the other parts of the apparatus were well insulated .
The potential difference due to the battery was indicated by a Kelvin 's multicellular electrostatic voltmeter The condensers were contained in a water-jacketed copper oven , provided with a temperature regulator , which maintained the temperature constant within one-tenth of a degree .
The air condenser was specially made to fit into the oven .
Its guard ring was a square brass plate , which fitted into the oven like a shelf .
The large disc was mounted on the guard ring , and could be moved up and down parallel to itself by means of a screw and nut .
The nut could be turned slowly by means of a worm worked by a rod passing 1909 .
] Induction through Insulators .
through a hole in the oven door .
The other condenser was placed a few centimetres below the guard ring , so that the two small plates were opposite each other , and the wire leading to the electrometer was passed through an ebonite plug fitted into the door of the oven .
The electrometer was put up close to the oven door , and the wire leading to ' it and the mercury cup key were surrounded by metallic screens .
The method of making an experiment was as follows : two pairs of quadrants were connected together , and the battery disconnected so that .
everything was at zero potential .
The apparatus was left in this until all residualcharge from previous experiments had disappeared .
this was the case , on insulating the small plates and their pair of quadrants the electrometer needle remained steady .
The battery was then connected at a time determined with an accurate chronometer , and the electrometer reading was noted at subsequent times .
The air condenser was adjusted so that the average electrometer deflection during the time in which the observations were made was as small as possible .
When once the air condenser was .
adjusted , a complete selies of observations with different potentials was made without altering the air condenser .
The oven was kept at a constant temperature day and night .
The sensibility of the electrometer for potential difference was determined with a standard cadmium cell .
It was usually about 150 scale divisions per volt .
Prof. H. A. Wilson .
On Electrostatic [ May 18 , iments on Ebonite .
A plate of ebonite was first used for the condenser B. It was 10- cm .
squale and cm .
thick .
Both sides were coated with thin silver foil , and an annular strip of this , 1 mm. wide and 5 cm .
in diameter , was cut off in the middle of one side .
The strip of bare ebonite thus exposed was covered with paraffin wax , which was found to improve the insulation considerably .
Contact with the silver foil was made by means of platinum-tipped springs tightly against it .
The following table contains the results of a series of experiments at C. Column 1 gives the times to the nearest hour at which each experiment was begun .
Column 2 ives the temperature , which remained very constant .
Column 3 gives the potential difference maintained between the ends of the total resistance .
Columns 4 to 12 give the values of at a series of times in minutes .
Thus each horizontal row contains the results of a series of measurements of the potential of the small plates at different times after charging up the lalge plates .
The values of are yiven in volts per thousand volts:\mdash ; Table I. Between 72 and 74 hours the capacity of the air condenser was altered in order to get electrometer deflections when using large potentials on the scale for a longer time .
The alteration diminished by volts per 1000 volts , so I have added this quantity to all the values of after duction through Solid Insulators .
'72 hours .
The results in the table are therefore for a constant capacity of { he air condenser .
The values of are given instead of because nearly all the values of are ative .
An inspection of this , able shows that is nearly independent of ( but it varies with the time in two ways .
It diminishes with , the time in minutes , since the condensers were charged in a particular experiment and also it increases with the time ( given in hours ) , since the series of experiments was begun .
The second variation is very slow , so that it is scarcely appreciable in under about 10 hours .
Each vertical column can be divided into four groups containing results got at times only differing by a few hours , so that we may take the means of the results in each group .
The ving .table contains these means:\mdash ; Table II .
For each value of nearly equally with the time .
The mean increase per hour during the first 19 hours is , during the next 51 hours it is , and .
the next 23 hours it is .
Since this slow increase is nearly the same for the different values of we can separate the two effects and take the means of the numbers in each vertical column in Table I to the variation of with and the means of the numbers in each horizontal row in Table II to give the Qlow variation of with the time .
When is negative the air condenser has the greater capacity , so that it appears that the capacity of the ebonite condenser was slowly tively to the capacity of the air condenser .
The value of was found from the following results:\mdash ; Mean Prof. H. A. Wilson .
On Electrostatic [ May 18 ' , The electrometer gave 342 divisions with volts .
Hence For the purpose of calculating the slow variation of the capacity we may take .
Substituting the mean values from Table II in this formula gives the following numbers:\mdash ; Time in hours 4 .
23 .
74 .
97 .
Thus the ratio of the capacities of the condensers diminished by about per cent. in 93 hours .
This variation was probably connected with the fact that the temperature had recently been raised from about C. C. The variation is more rapid during the first 19 hours than afterwards .
After the temperature had been kept at C. for several days the variation of the ratio at C. was again examined .
The following values of were obtained .
Each number is the mean of several observations with different potentials:\mdash ; volts per 1000 In this case was positive , so that the ebomite condenser had the greater capacity , and the increase of with the time shows that its capacity was increasing relatively to that of the air condenser .
It appears , therefore , that after the temperature has been raised the capacity of the ebonite condenser falls slowly at constant temperature for a long time , while after a fall of temperature its capacity slowly rises .
This effect is probably analogous the slow variations which take place in the volume of a thermometer bulb after a large temperature change .
This variation of the ratio of the two capacities might be due to variation in the capacity of the air condenser .
The distance between the plates of this condenser was mm. during these experiments on ebonite .
If this altered by mm. this would account for the per cent. change in the ratio of the capacities .
The condenser was rigidly constructed , so that it does not seem at all likely that this can be the explanation of the variation .
To test the working of the apparatus some experiments were tried with two air condensers .
The second one consisted of a shallow brass box , the top of 1909 .
] Induction Sotid which was closed with a thick ebonite plate .
This plate carried a brass disc inside the box and about 3 mm. from the bottom .
The bottom of the box and the disc formed the condenser plates .
A wire from the disc passed through the ebonite plate and was surrounded by an earthed to preveut from the box to the disc across the ebonite .
This condenser was made specially .
It was put in the oven C. instead of the ebonite condenser , and the adjustable air condenser was adjusted until no deflection was obtained with volts .
There was then a deflection of about three scale divisions in one minute with volts .
The electrometer gave 170 diyisions for one volt .
With volts the deflection was about two scale divisions in one minute .
After 24 hours volts still gave no deflection .
This nent shows that the capacity of the adjustable air condenser remained constant to within about 2 parts in 10,000 during 24 hours , just before the experiment it was cooled down .
variation was at least 50 times less in this experiment than with the ebonite condenser , which seems to show that the variation was really due to the ebonite and not to the air condenser .
I have since repeated this experiment with the two air condensers , and found that the ratio of their capacities did not vary by 1 part in 100,000 in 90 hours at constant temperature .
The variation of the capacity with the time of charging will now be considered .
The mean values of iven iu able I were plotted on paper and a curve drawn through them .
ents were very carefully drawn to this curye , and from the inclinations of the tangents the following values were obtained .
Time in minutes 1 .
2 3 .
4 .
6 .
8 .
10 .
15 .
20 .
30 .
The last row contains by from the values of in the second row , and tiplying by the times .
These numbers for the larger times are nearly constant , so that it appears that is inversely proportional to the time since the condenser was charged .
After a long time the rate of variation would therefore be Prof H. A. Wilson .
On Electrostatic [ May 18 per minute , so that this may be taken to be the effect due t , o the conduction through the ebonite .
Adding to the values of , we get what should be the values of , which would have been obtained if the ebonite had insulated perfectly .
The following table contains these corrected values of in volts per 1000 , and the values of deduced from them by means of the equation which is got from ( 3 ) by putting in the value found for and The fourth column contains values of calculated from the formula .
If the capacity of the ebonite condenser is put equal to 1 when its capacity at any subsequent time is en by the equation , which agrees with the values found to about 1 part in 1000 from 2 to 30 minutes .
The variation of the capacity with the time cannot be represented at all exactly by a formula like , for this makes the rate of variation proportional to instead of to Schweidler*found that the rate of variation of the capacity of some of his condensers was proportional to where was a constant .
His values of were less than unity .
Trouton and Buss found that the current due to the of the residual charge of condensers could be represented by the formula , where and are constants .
This gives for the total charge recovered in a time . .
cit. 'Phil .
Mag May , 1907 .
1909 .
] through Solid lators .
A series of experiments was done on the ebonite condenser at After raising the temperature it was necessary to slightly alter the capacity of the air condenser .
The table contains the mean results obtained. .
The conductivity of the ebonite was not appreciable .
Time ( inminutes .
Found .
( Calculated .
) The last colunm contains numbers given by the equation .
Putting when gives .
It appears , therefore , that the capacity varies with the time times as , rapidly at C. as at.30o C. Another set of measurements was made at C. , but at this temperature the insulation between the small plate and the guard not so that reliable results could only be obtained for times of up to about five minutes .
The following table contains the mean results obtained:\mdash ; Thus at C. the capacity varied as much in 7 minutes as in 15 minutes at C. The last column contains numbers given the equation , which gives , on when .
Prof H. A. Wilson .
On Electrostatic [ May 18 , Wax .
A paraffin wax condenser was substituted for the ebonite one .
It had a capacity about three times larger .
It was found that with the wax the deflection of the electrometer was simply a linear function of the time after .
This was taken to mean that the capacity did not change appreciably , and that the increase in the deflection was due to a small conduction current .
The increase amounted to about volt per minute with 500 volts , but was nearly independent of the potential difference .
This makes the conductivity of the paraffin with 500 volts about The electrometer deflection for was foumd by the straight line drawn through the points representing the observed deflections .
The following are some of the results obtained in this way at C. :\mdash ; In this case the capacity was independent of the potential to about 1 in 2000 .
At C. similar results were obtained .
The leak amounted to about volt per minute , and again was nearly independent of the potential ditTerence .
The conductivity therefore increased six times for a rise of C. At C. the leak was volt per minute with 60 volts , with 120 , with 240 , and with 480 .
The current was therefore nearly proportional to the square root of the potential difference .
Quartz .
An attempt was made to work with a condenser having a plate of crystalline quartz cut perpendicular to the optic axis for its dielectric .
The quartz , however , did not insulate sufficiently well , although it was very carefully cleaned and embedded in highly insulating paraffin wax .
The leak was through the quartz , not over its surface .
1909 .
] Induction through Solid A series of measurements was made with a sulphur condenser at C. The following are the mean results:\mdash ; The last column contains the numbers given by the formula\mdash ; .
Making when ; this gives .
The capacity of the sulphur condenser increased by I8 per cent. from 10 to 30 minutes .
The ebonite condenser at C. increased in capacity per cent. in the same time .
The conductivity of the sulphur was not appreciable .
Conclusion .
experiments described above show that for a definite tinle of charging the capacity of a condenser is nearly independent of the potential difference .
The variation is not more than 1 part in 1000 , and is probably really much less than this .
Small variations appear to occur , but they are evidently due to slight temperature chances or to the after effects of the previous chargings .
There is consequently no reason to suppose that anything of the nature of true hysteresis exists .
The capacity is a function of the time of charging , and in the case of ebonite and sulphur varies from 5 per cent. to 20 per cent. in the course of half an hour .
large variation of the specific inductive capacity of sulphur and ebonite makes these substances very unsuitable for use as insulators in experiments in which exact electrostatic measurements are to be made , unless care is takell that the total capacity is large compared with the capacity due to the insulation .
The capacity ( C ) at constant potential ifference ( is given vely approximately by the formula\mdash ; where is the initial value of the capacity , the time of charging , and VOL. LXXXII.\mdash ; A. 2 Prof H. A. Wilson .
On Electrostatic [ May 18 , and are constants .
The following table contains the values found for and The constant appears , therefore , to be independent of the temperature .
The charging current or is equal to .
Schweidler* determined the charging current with a galvanometer , and considered that it was proportional to , where was rather less than unity .
I find that his results for glass can be represented fairly well by a formula of the type .
The following are his results for a glass condenser:\mdash ; The currents in the above table are those due to a constant potential difference of 300 volts .
After 900 seconds the condenser was disconnected from the battery and connected directly to the galvanometer , when the following currents were observed : Here is reckoned from the instant at which the condenser was connected to the galvanometer .
If a condenser is charged at constant potential for a 1909 .
] Induction through Solid Insulators .
long time and then discharged through a galvanometer , then the discharge current should , according to Pellat 's theory as generalised by Schweidler , equal to current .
Schweidler 's results show that this is approximately true .
If we assume this , then according to my results we should expect the discharge current to be given by the expression , which is of the same form as that found by Trouton and Buss to represent the results of their experiments on the discharge of condensers .
They did not determine the absolute values of the constants in their formulae .
The apparatus used in these experiments was purchased with a grant from the Government Grant Committee of the Boyal Society , to whom , therefore , I wish to express my indebtedness .
A great many preliminary experiments with different forms of apparatus were tried , and many long series of observations were taken and rejected before the results described above were obtained .
My thanks are due to my assistant , Mr. H. G. Martyn , B.Sc. , for the great pains which he took with these long and tedious experiments .

