]\gt ; Qualities of Porcelain , uith special reference to Dietectric Losses .
By H. F. HAWORTH , Ph.1 ) .
, M.Sc .
, B.Eng .
, ] ( Communicated by Professor W. E. Ayrton , F.R.S. Reccived May 7 , \mdash ; Read June 4 , 1908 .
) The following research was undertaken to determine some of the electrical properties of porcelain , and their variations with respect to potential , temperature , and time .
The electrical properties ated in this paper are as follows:\mdash ; .
1 .
The of of a porcelain condenser .
2 .
The of a porcelain condenser as a function of the potential .
3 .
The lfluence of sudden cyclical of potential on the charge of a porcelain condenser .
4 .
constant of porcelain as a function of the temperature , and its value .
B. surements .
5 .
The apparent conductivity of porcelain as a function of the potential .
6 .
The ) conductivity of porcelain as a function of the time of electrification .
7 .
The apparent conductivity of ) as a function of the temperature , and value .
C. Didectric Loss 8 .
Contact lnethod , variable frequency , and constant potelltial .
9 .
Thermo-electric method : Constant frequency , potential .
10 .
Thermo-electric method : Variable frequency , constant potential .
11 .
Historical .
12 .
Summary .
porcelain was in the of discs , out .
in and cm .
, from the Fabrik Hermsdorf , .
The iscs had a circular tinfoil electrode of .
diameter xed 011 them on each side .
.\mdash ; Current were taken with a Siemens anonlcter , the usual ilit of which was about tl11pere VOL. LXXXI .
Dr. H. F. Hawortb .
ay 7 , per deflection at 2 metres .
The deflections were read by means of a telescope and scale , and be estinlated to 1/ 10 .\mdash ; For potentials up to 200 volts a battery of 100 accumnlatol'S was used .
For potentials ) to volts a battery of 2500 small tors used .
These cells were made sections of 50 in two rows of eacb , fixed into a paraffin wax base which rested on a ] plate backed with wood .
Contact was made with mercury cups .
1 .
The rate of charge of a conl ( enser .
Two connccted in parallel used , and they were and dischar means of a affin nlerctly switch .
To secure accuracy , and to avoid ) other tfects , the method was adopted:\mdash ; ( 1 ) plates were for then the galvanometer , the swing ( 2 ) They were then for , ( isch , and the noted .
The plates were for a time to , reater than , the charge time .
About 70 of these double were .
of one and seconds .
The of these resnlrs then culated , and from their difference the percentage ' of ( the time was incleased one to two The was also then on curve for th plates were then ) ' !
) ) tlnoe in the same manner , then for three and fonr seconds , foul( six , etc. It was assunled in next , that potential charge curye for porcelain was a line , therefore timecharge curve may be continued , the three secon calculated for the o1iginal voltage .
The curve was continued in this nlanler ) .
Beyond 90 seconds a increase of the will noted , for may be due to a temperatule effect ; practically the is fully in one minute , if take the first galvanometer as a of the pacity .
The results raphically represented in Curve 1 , unlel i in Table I. In the and 30 minute the ried S , it on acconnt of temperature so soul r re readings were , and tions in were reduced to per del o ) , the average result per cent. Electrical Qualities of Porcelain .
Table I. so CURVE 1.\mdash ; Charge of a Porcelain Condenser as.'a Function of the Time of Charge .
2 .
The charge of a porcelain condenser as a function of the potential .
The two plates used in the previous experiment charged for half a minute , discharged for one minute , and the of the galyanometer and the potential were noted .
Readings up to 200 volts were taken in steps of 20 volts , and 40 readings were taken at .
Temperatures were noted at the beginning , middle , and end of each series , and a estimate of the percentage variation of per rrade was made ; it was about per cent. Up to 200 volts the points lie along a line , and any small variations from it followed the temperature variations .
Dr. H. F. Haworth .
[ May 7 , The potential charge cve was now constructed , using the high-tension up to volts .
The results ) in Curve 2 , and from it will be seen that the is directly proportional to the charging potential .
This agrees with J. Curie 's on crystals , he found that the charging current is exactly tional to ) potential when the potential changes are slow .
It will be seen latel that this does not hold if the changes are rapid .
400 Izoo CURVE } Condenser . .
The effect of sudden of on the of a porcelain was now Three plates cnclosed in a phospl ) oric acid as ( , for one minute in the usual manner .
The tcred by ) roximRtely 4 each time b limits [ cievcral } , and the charge plotted ) ) in .
The values are reater t ones , a small dielectric loss even very slow lates o field .
point is dealt with 1norc fully ' ) \ldquo ; -k J. Curie , ' dc Docto ] at , ' June , 1888 .
The El ectrical Quaiities of CURVE Curve for Porcelain Condenser , Potential varied cyclically .
4 .
The variation of the dielectric constant with temperature was investigated in following manner:\mdash ; Three porcelain plates were placed in a copper vessel containing a drying agent .
This copper vessel was heated by means of an oil bath , the oil of which was first heated to the required temperature by a Bunsen burner .
The temperature was then main ( iained constant by means of all electric heater , which consisted of a framework of asbestos insulated iron wire wound on the \mdash ; er copper vessel .
The heat ladiation from the apparatus was exactly balanced by the current sent through the resistance framework .
The oil was ci1culated by means of a pump and stirrers worked by band .
The temperature was measured by a thermometer resting on the porcelain .
The porcelain was cred and in the usual mannel , and discharge being each of one minute 's duration , the potential being reversed each time to neutralise polarisation effects .
The oinc and was continued until the alvanomeber swings and temperature were constant for at least one hour .
The results of the experiment are given in II , and are hically shown in Curve 4 .
Dr. H. F. Haworth .
[ May 7 , Table II .
4.\mdash ; The Dielectric Constant of Porcelain as a Function of the Temperature .
From the results it will be seen that the ) erature has a reat influence on che dielectric constant .
At telnpent to C. increase of city is much smalier than at mpcraturcs .
the capacity increases 54 per ccllt .
, and } and increases 108 per cent. ) , a ninc1 times as great a rate of growth .
From to the ctric ) is function of temperature , tion b of the .
where dielectric constant at temperature dielectric c temperature .
1908 .
] The Electrical Qualities of Porcelain .
From to 10 the equation of the curve is where dielectric constant at temperature and , From Table II , and the above equation , we find that the dielectric constant for this kind of porcelain at C. is B. Apparent By the apparent conductivity is meant the conductivity as expressed by the ratio of current to potential .
5 .
The apparent conductivity of porcelain as a function of the potential .
A porcelain plate was placed in an glass vessel containing a drying , and a current was sent it as well as through the galyanometer from the -tension battery .
At the end of minutes the deflection was noted , and it the conductivity per centimetre cube was calculated .
The potential was increased by steps of 110 volts to volts ; the current curves obtained were ncave towards the voltage axis , but they did not remain constant , they varied daily .
The decreased greatly with increase of potential .
On the with potentials the results obtained were widely different from those obtained with increasing potentials , and on putting the porcelain through a cycle of electrification the curve obtained for current with respect to the potential formed a closed curve as shown in Curve 5 .
On putting the porcelain through a number of continuous ] of electrification the same curve was traced out each time , thus that a steady state had been reached .
The porcelail ] , on having a otcntial innpressed on it , ates a back E.M.F. which varies for different cycles of electrilication .
In other wordb the dielectric becomes polarised .
If we consider the dielectric as being made up of electric doublets ( particles matter having electric poles ) , they are normally in a heteroeneous condition , but when an electrostatic field is impressed on the dielectric , the ) rticles tend to orientate with their electrical axes the lines of .
The number which would do so would vary directly with the impressed voltage , and their electrostatic potentials would add to resist the flow of rent t dielectric .
free poles of the ) ticles in contact with the .
electrodes would constitute the charge .
Dr. H. F. Haworth .
[ May 7 , for , the tential being varied cyclically .
On Curve 6 the is plotte } } lespect to the okntial .
It ticed t with ] dielectric , the conductivity is fairly constant ) ) ( A to I ) on ( the potential the conductivity ) idly f to the ] ) tion of the ( to C ) , and fact the goes off to a ltegative infinity siml ) lncans that .
of ) celain is greater than the implcssed cm on account of ) back , then the conductivity then eases l sl lie ( 1 ) to ) , a fairJy level value at the position of Dtaxinllll , and 011 the h. it a infinity again ( to ]ting a from positive infinity ) .
A similar curve would be obfained if we plotted the of ainst ttetonlotive force .
Current ( curves were ) lotted for cycles , and volts with sinnilar results .
1908 .
] The Electrical lities of Porcelain .
CURVE G.\mdash ; Appaxent Volt ( Curve for Porcelain .
The back 's for zero current , points where the curves cut the voltage axes , were as follows:\mdash ; curve and volts , , : It will be noticed that the positive values greater than the negative ones .
The of these baclk .M.F. 's bear the following relationship to one another:\mdash ; The back .
is not pp.tionate to the voltage applied .
It would seem as if the number of doublets iuto line per unit fall of potential was decreasing , or in other words the meability of the dielectric decre ses with increase of cation , or the dielectlic tellds towards electrical tion .
The currents voltage , i.e. the points where the cut the current axis , were as :\mdash ; volts curveaIld - volts , : , , 9.0 Dr. H. F. Haworth .
[ May 7 , .
The negative values are greater than the positive ones .
The current due to the polarisation of the dielectric does not increase proportionately to the amplitude of the cycle but varies with the back E.M.F. The currents at volts were : and The apparent conductivity increases with decrease of negative values here are greater than the positive , the sum of these variations that the current curves are depressed htly into the .
ative quadrants .
6 .
The variation of the apparent uctivlty of porcelain with the time of electrification now rated .
Fifteen plates , connected in parallel , were placed in a lass vessel containing a drying agent and a thermometer .
The plates were placed in a on porcelain feet were connected in parallel with strips of tinfoil .
The glass vesscl was paraffin lvax , and the connecting wires were out tubes filled with raffin 1 .
CurIent as supplied by a batter ) of 100 cclls , the constar]t ( 200 ) throu hout the experiment .
The rent cctivity plotted inst l time which it will be seen that the conductivity falls very rapidly the first 10 minutes , and then more gradnally for hour .
After 18 the was considerably snlaller at tho ) hour , and C5 hours the conductivity an value .
This was tbly a peratnre effect .
Aftcr 60 hours 3 minutes odes -circuited a current } given in everse oction the ] ) ) urrent measured the cells tlot the , as galvanontetcr uited while this lace .
current } minutc for over hours , and it fcll the manner the Gllt 1 ) th F. .
fell .
results arc1 in ; they the dielectric tinte to ] } ) nt that in so it stolcb ) trds be cycle is , the ) reelaiil d likc the current voltage ) of would altcr with the tinlt1 .
the in vise I8 .
] The Electncal Qualities of Porcelain .
CURVE 7.\mdash ; A. Conductivity of Porcelain as a Function of the Time of Electrification .
B. Short Circuit Current from Porcelain .
direction with increased time of the cycle on account of the lower conductivity due to the increase of polarisation with tinle .
This point was investigated with the plates and -volt battery .
The voltage was raised or lowered by steps of 40 ettC time .
The time of the step in the first series was one minute , for the second sclies two minutes , for the third se1ies three minutes , series iive minutes , fifth series seven minutes .
The curves flatten out and turn in a ockwise direction the of the cycle increases .
The areas of the loops bear the following ratio to one another:\mdash ; : 2 : The maximum currents the ratio:\mdash ; : -( : : The apparent lctivity has with the increased tinle of tlJe cycle , thus viscosity in the 7 .
The conductivity of porcelain as a function of th nperature .
Fourteen porcelain plates were placed in a pile in the inner ) vessel used in the previous temperature experiment , with drying Dr. .
Haworth .
[ May 7 , .
The heating arrangements were as before , but the oil was maintained at a temperature inclined propeller driven by a small motor .
The plates were connected in parallel strips of tinfoil , and the terminals consisted of two mercury cups into rlass tubes which passed ( holes in cover of ) vessel .
The mercury excluded air from the vessel , and contact was ] with the pol.celain ping ires into the mercul.y .
was tested for by just the wires out of the mercury ( ( notin the .
In all cases , however , the leakage was practically top and bottom porcelain plates had no electrodes and served as .
The connections were as follows:\mdash ; H. ing battery . .
Battery hing .
F. reve1 switc .
A B and ng By connecting A to current was sent ) ) celain , and by with C conductivity after ) electrification by switcb ]loti The was the1l ) , to get rid of polarisa ; hc ) ttery and , and the .
At each nperitttue of the , in 1( when both ( temperature least one hour , the of these } txnd ctivit ) calculated .
The heated live or taking the final it to electrical corresponding lesults of the ) eriment i given in II1 and Curve 8 .
It seen th of luclivity with ature i , , .
08 .
] The Qualities of Porcelain .
Table III and the above equation the specific conductiviCy of celain , calculated at C. , was mhos per centimetre cube .
Table III .
Specific resistance , Temperature , C. ohms per cm .
2 .
141.0 51.5 50.8 42.3 20 .
35.5 30 .
34 .
37 .
8.24 , , 6.25 , c 8.\mdash ; Apparent Conductivity of Porc Temperature , Specific resistance , ohms per cm.3 40 .
43 .
4.85 47 .
4 .
50 .
2.64 54 .
1.78 66 .
1.44 58 .
1.11 59 .
1.08 62 .
0.77 64 .
0.15 Dr. H. F. Haworth .
[ May 7 , .
8 .
The foregoing results show that there is a of energy in putting a porcelttin condenser a of ctrification , and it is desirable to find on what ctors this loss depellds and the laws connecting it with variations of voltage , frequetlcy , tcmperature .
Another point which ] ( itself is : lJoes the ) acity o the condenser alter under if so , how ?
To ' the } ) ence of the dielcctric loss on the frequency ol the cycle of electrification , the ) eriments were made : A porcelain plate was placed in a sealed lass V , with sulphuric acid as and was connected across secondary of a giving about 2000 volts .
The current and voltage waves were then plotted with the Joubert contact method .
The of the curre1lt wave then calculated for erent values of the time , and so the quantity electricity on the condenser was obtained for that time , The at this time WttS knowlI curve , and plotted ainst V. The resultant curve enclosed an arca similar to Curye 3 The area of this curve represents the required to put the condense1 a cycle of electrihcaCion at that frequency and voltage .
A nunlber of these curves were ucted at various frequencies , and the results are given in results secm to show that the loss cycle is a constant quantity .
Results , and 9 vory consideral ) from the average .
to erl.ors of measurement , also the process of obtaining final loss rather a complicated one , and e1rors . .
could easily creep in .
Table I Cycles per cond .
] ) Temp , , whole plate ) .
130 2(i80 1SO 8 1$ II .
1908 .
] The Qnalities of To hese points further , the foJlowing method was adopted , which could be worked more rapidly , and by one person alone , whereas the previous method equired two observers .
The porcelain was placed in a large lass case containing some strong huric acid .
Pressure was supplied with a Ferranti transforrner , transfrom 150 to 40,000 volts .
In series with the porcelain condenser was placed a condenser of large capacity with that of the porcelain , and a non-inductive resistance of 40,000 ohms .
The current the porcelain was measured by the voltage across the resistance , and the voltage across the was measured by the same Grument ( an Ayrton ) erry Quarter Cylinder Electrometer ) placed across part of a non-inductive resistance , hich itself was placed across the primary of the transformer ( see diagram of connections ) .
Connection to the bottom electrode of the porcelain was made through the brass case of the thermopile .
The current passing had no effect on the thermopile .
It was intended to measure the capacity of the porcelain by the pressnre across the condenser placed in series with it , according to the law , where small the capacity could be assumed to remain constant , but on connecting the condenser to the electrometer a snlall deflection was first obtained which rapidly grew larger until it was beyond the range of the electl'ometer .
It was found that this deflection was due to polarisation , for , when a battery was placed in series with the ele , ctrometer , the deHection always increased when the pole of the battery was con nected with the earthed side of the condenser , thus thnt the current was through , or over , the porcelain from low to high tension side .
It was that perhaps the ) rush discharge was effect , so an earthed guard was put round the bottom electrode , but the result was the same .
This condenser was then dispensed with , and capacity calculations were made from the values of the current , voltage , and frequency .
This was allowable as the E.M.F. wave was a pure Wave .
A considerable alteration had to be made in the apparatus on account of the large quantities of ozone and oxides of nitrogen which were formed ) the , and which rendered the roonls very objectionable to work in .
The polcelain plate was placed on one end of a thermopile ( used for tenlpera ( measurements ) , the whole in a raphic dish , and was covered with a bell-jar .
Air was sucked into the apparatus two towers containing calcium chloride , and a wash-bottle containing huric acid ( see diagram of Dr. H. wortb .
[ May 7 , ratus ) .
The was sealed at , bottom with ordinary machine oil .
The ozonised air led out a tower manganese to split the OIle a ottle c acid , to prevent tIty w ) ) into the ratus .
air current was produced by an dinary pump , and the air was dried before ring the } ) paratus , in order to diminish the brush of Ci cctric lobs was measured ] ) the CIlCC of tcml ) which could ) ined ) centre of the surface of th above ) tcml ) eraturc .
This teml ) creature difference ) a The Electrical Qualities of Porcelain .
few degrees , it follows from Newton 's laws of cooling , that the heat lost is proportional to this difference ; and as the heat lost is equal to the heat gained , if the temperature is constant , a thermopile placed with one end in contact with the porcelain enerate an E.M.F. which is directly tional to the dielectric loss .
This E.M.F. was measured by connecting the thermopile with a low resistance galvanometer through a reversing switch , and noting the direct and the reversed deflections to neutralise the effect of local E.M.F. 's , etc. The losses were first measured at a constant frequency of 50 per second , with pressures up to 38,000 volts ( R.M.S. ) .
The results are graphically represented in Curve 9 .
The equation to the curve is expressed by Loss per cycle Joule per cubic where is the B.M.S. radiant per centimetre .
CURVE 9.\mdash ; Potential-Loss Curve for Porcelain .
Frequency Constant Loss The losses were then detel.lnined at constant pressul.e ( 30,000 volts ) , and for frequencies between 108 and 200 .
The points obtained lie practically on a straight line which does ) pass the ( Cve 10 ) , so we have the equation for the dielcctric loss as ectric 1imetre iiJne t VOL. LXXXI.\mdash ; A. Dr. H. F. Haworth .
[ May where is the potentiaJ radien of cycles per second . .
\mdash ; Frequency-Loss for elain .
Constant B. Lobh constant is probably to the fact that porcelain is not a perfect insulator , and that a certain current would flow if a direct pressure 30,000 volts was applied to it , rise to an ordinary loss .
We have here a striking si1nilarity between netic and dielectric losses .
They each independent of the time in which the cycle is completed ( i4 we ( lect the due to the rfectio of the dielectric ) , and arc proportional to si1nilar powels of the amplitude of the cycle .
May we not extend this similarly to the internal actions ?
Let us the dielectric to consist of electric instead of netic doublets : these particlec would oscillate or evolve under the influence of an and energy would be absorbed in ovelconling their nloleculal friction .
In 1861 , W. Siemcns ished a paper on the of the Leyden jar , pointed out that a condenser became heated on and .
Many attcnll ) have been made to measure the loss of in dielectrics subjected to a field .
Monatsl)'October , 1861 .
1908 .
] The Qualities of Porcelain .
In 1890 , J. Swinburne suggested* that the loss was due to want of homogeneity in the dielectric , and imagined conducting channels to run across the dielectric , giving rise to a loss .
This view has been accepted by many subsequent investigators .
In 1891 , Major Cardew attempted to measure the power factor of a paraffined paper condenser ) the " " Three Voltmeter\ldquo ; method , but the results obtained are doubtful .
In the same year Hutin and Leblanc , workin with paraffined paper condensers , also attributed the loss to heterogeneity of the dielectric , and came to the conclusion that they were dealing with tlJe case studied by Poisson of a perfect dielectric , containing many small spheres of conducting nlaterial .
They destroyed the fibre of the paper by strongly heating , and in doing so the dielectric constant fell from 8 to .
The condenser then stood 1000 volts per centimetre without heating , at higher pressures the temperature rose .
Steinmetz , S experimenting on paraffined paper condensers , found the dielectric loss to vary as the square of the applied pressure .
The pressures used varied between 80 and 320 volts at 170 ; at higher pressures he found that the loss increased at a higher power than the square of the applied pressure .
Steinmetz pointed out the between the hysteresis loss in iron and dielectric loss , and so the name " " Dielectric Hysteresis\ldquo ; was loosely given to any kind of dielectric loss .
makes a similar remark , and says that solid dielectrics present phenomena quite analogous to magnetic hysteresis , in consequence of which , at equal pressures , the charge in a condenser is smaller for increasing than decreasing pressures .
Janet , in a later account , experimenting on a mica condenser , obtained a curve enclosing an area for the relation between and .
analogous with a magnetic cycle in iron , but its form may be accounted for by hysteresis , or viscosity , or both .
By a hysteretic loss we mean one which is independent of the time taken for the cycle , and this point has been overlooked by many experimenters .
rees with Poisson in his case of a perfect dielectric containing conducting spheres , and deduces theoretically the existence of residual c and dielectric loss , and shows that the loss would with the results of 'Phys .
Soc. ' vol. 11 , p. 49 .
I. E. E. Proc May , 1891 .
'La re vol. 4 ] , p. 179 , ] , 1891 .
S 'Electrical ' of New York , Malch 16 , 1892 .
'Comptes Rendus , ' Deceml ) , 1892 .
Comptes Rendus , ' vol. 116 , p. , Februan , 1893 .
'La ectrique , ' vol. 46 , p. 402 , November , 1892 .
Dr. H. F. Haworth .
[ May 7 , Steinmetz and others .
Hcss does not think that the result obtained by Steinmetz , for dielectrics , can be considered as analogous to his law for magnetic hysteresis , , because , if dielectric hysteresis existed , it would be entirely swamped by the loss due to the conducting bodies .
Arnot*suspended hollow cylinders of various substances in a rotating electric field , and showed that the cylinders had a tendency to turn ; from this , Arnot calculated the energy expended in the dielectric , and called the ioss hysteretic , although he might just as easily have attributed it to viscosity .
Bedell , Ballantyne and employed a method similar to the one described on p. 234 , and attributed the loss to hysteresis .
and Morris , using very slow cycles of electrification , found the ialion in of a condenser for equal pressures , on decreasing the potential , to be less than one in eight thousand , and so concluded that the effects were due to viscosity rather than hysteresis .
F. BeaulardS found that the energy dissipated in different dielectrics at different frequencies varied with the time of the cycle , and tended to vanish as the time of cycle increased .
According to hinl , the loss was a viscous one .
E. E. Northrup shows that the value of the specific inductive capacity of a dielectric is for a slowly field than for a rapidly varying field .
Experiments of Blondlot and J. J. Tholnso show even greater changes .
H. Pellat tried to show that the loss was due to dielectric olarisation , and that if a solid or liquid dielectric be suddenly placed in an electric field it polarises , the polarisation increasing with time , and tendin to a maximum .
If the field ceases , risation diminishes to zero .
W. suspended ellipsoids of ebonite and bifilarly etween the plntes of a Kohlrausch condensel found the loss from the of the ations .
In the case of parnffin the was pptional to the square of field .
Arnot found that varied as the of field .
The methods have also been dopted to the losses in lectrics : ' Rendiconti .
Accad .
Lincei , ' ) ) ; 'The ician , , ) ) , , 189.3 ; 'La 4 ical P ' vol. 1 , ] , ' .
468 , .
de A ust , ) ) . .
and in Icctl itism , p. , etc. lnal .
, ' vol. 18 , ) } .
) \ldquo ; .
, 189 .
1908 .
] Tloe lities of Porcelain .
Kleiner*used a thermo-couple embedded in the dielectric .
BermischkeT used a bolometer to determine the rise in temperature in a paraffin wax plate , but failed to detect any , and concluded that dielectric hysteresis did not exist .
He attributed the losses to ( 1 ) Joule effect , ( 2 ) residual charges ( viscosity ) , ( 3 ) mechanical lcsses due to yiblation condenser sheets .
Rowland and used the split dynamometer method brought out by Professor Rowland ; they experimented on pa1affined paper condensers , and , found the loss to increase with the frequency of the cycle .
In other experiments they found that the loss per cycle was independent of the time of the cycle .
These results are opposite to those of Porter and Morl.is , and it seems not unlikely that , under certain conditions , viscosity may be the predominating feature of the dielectric , and , under other conditions , hysteresis .
and Smith , S using a resonance method in conjunction with a wattmeter , found that the loss was proportional to the of the current , and hence the loss be attributed to the equivalent resistance of the condenser .
The voltage varied to , and the condenser was of paraffined paper .
Dr. P. Humman using the resonance wattmeter method with different kinds of cables , found that the loss was proportional to 1 .
A porcelain condenser charges at a comparatively rate .
This may account for the dielectric constant , as measured by alternating current , being smaller ( 718 at C. ) than that measured by continuous current ( 8 at 2 .
For pressures up to 1200 volts the charge was directly proportional to the pressure , if the potential changes were made slow enough .
3 .
If the potential changes were made rapidly , the was not quite proportional to the potential .
Ihere was a certain dielectric loss .
4 .
The dielectric constant , measured after one minute 's electrification varied with the temperature according to the laws : Between and C. , Between and 10 C. The ectric constant at was ( Curve 4 ) .
'Wied .
Ann vol. 50 , p. 138 .
Wiener Sitzber No. 102 , vol. .
Mag vol. 45 , p. 66 , 1898 .
S 'Phys .
Review , ' vol. 8 , p. 4 , January , 1899 .
'Inaug .
Diss Bonn , 1896 , Extract , " " El .
Zeitschr 1898 , pp. 435\mdash ; 436 .
Electrical Qualities of .
The apparent conductivity of porcelain varies with the applied pressure and the duration of the application .
The dielectric polarises , or generates a back , when a potential difference is applied to it ( Curves 5 and 7 ) .
The conducting mechanism viscosity .
6 .
apparent conductivity of porcelain , measured after one minute 's electrification , increases with the telnperature to the law:\mdash ; .
urve 8 .
The conductivity at C. is 0 mhos per centimetre cube .
7 .
The dielectric loss varies as the -power of the voltage ( Curve 9 ) , and is independent of time of the .
The dielectric loss at sures and fiequencies ) etel.nled hysteretic .
At very slow frequencies the loss is nlainly loss , this slJows viscosity effects ; losses are swamped at by the dielectric hysteresis ( see Curve 10 ) .
The dielectrlc consbant is not by , electrostatic , or by the frequency alternation of field , within the limits of the ) el.iment .
The continuous Yent ecnls this were carried out at the Institut des technikums , , and 1 have to thank tisor Dr. H. ] .
Wcber for his yaluable ] .
The alternating current ere carried out ( the City and Guilds of tldon Centlal Teclmical ; and 1 indebted to ofessor W. Aylton , nd T. loan and advice .
Kiilnes , , Lefe ( Nolthcute ( students of the Central ) the work required for

