]\gt ; [ 519 ] XIII .
Etectrometers .
W. , AYRTON , J. PERRY , W. E. SUMPNER .
Received May 19 , \mdash ; Read June 4 , 1891 .
[ PLATES 9-12 .
] TABLE 0F CONTENTS .
Section .
Page .
I. General account of the experiments . . . . . .
519 II .
Effect of varying the distancs between the fibres . . . . .
526 III .
Effect of varying the distance between the quadrants . . . . . .
533 IV .
Cause of the peculiar action of the White electrometer . . . . .
539 V. Motion of the electrical zero . . . . .
644 VI .
Electrometer used to measurs powel . . . . . .
545 VII .
Improved quadrant electrometer , construction . . . . .
548 VIII .
ResuIts obtained with the improved quadrant electrometer . . . . . .
554 IX .
Sketch of the mathematical theory of the White quadrant electromeber . . . . . . .
561 I.\mdash ; GENERAL ACCOUNT 0F THE EXPERIMENTS .
IN 1886 .
on continuously charging up the needle of Sir WILLIAM THOMSON'S bihlar quadrant electrometer , No. 5 , made by Messrs. WHITE , of Glasgow , and in use at the laboratories at the Central Institution , it was noticed that the deflection of the needle , when the same P.D. ( potential difference ) was maintained between the quadrants , instead of steadily increasing , firsb increased and then diminished , so that both for a large charge on the needle as well as for a small , the sensibility of the instrument was small .
A similar effect had been described by Dr. J. HOPKINSON in the ' Proceedings of the Physical Society , ' vol. 7 , Part I. , for the previous year , and the explanation he gives of this curious result is that if the aluminium needle be below the centre of the quadrants the downward attraction of the needle , which increases with the square of the needle 's charge , increases the pull on the bifilar suspension , and so for high charges more than compensates for the increased deflecting couple due to electrical action .
On raising , however , the needle of our eIectrometer much above the centre of the quadrants , the anomalous variation of sensibility of the instrument , with increase of charge in the needle , did not disappear , and when the was raised so that it was very close to the top of the quadrants , and when , 2 !
IIESSRS .
W. E. AYBTON , J. PERRY , AND W. E. SUMPNER if Dr. HOPKINSON'S explanation correct , the sensibility ( or deflection corresp ing with a given P.D. between the quadrants ) ought to have been very great large rge on the needle , it was , on the contrary , found to be small .
As the needle had been somewhat damaged while the instrument was in possession of the late Mr. CROMWELL , before it was so kindly made over the Central Institution by his son , Mr. VARLEY , Junior , together other apparatus belonging to his father , a new needle .
platinum wire , and were obtained from Messrs. WHITE .
On suspending the new needle with the platinum wire and weight , as received from Messrs. WIIITE , the needle was found have two modes of vibration , one the ordinary slow one , when the platinum turned in the acid at the bottom of the Leyden jar , and the other a very quick one , due to the twisting of the wire tself without the weight moving .
This vibration was removed by replacing the fine platinum wire supplied by Messrs. with a somowhat thicker one .
The quadrants were adjusted for symmetry , the fibres tightened so as to have equal tension , as shown by the sensif ) for a given potential of the needle , being , minimum , and the electrometer again tried , but it was still found that when the quadrants were close to one another , and when , therefore , as the needle was best shielded from external action , the standard formula for the electrometer might be to be most nearly fulfilled , the sensibility as before , first increased as the charge in the needle was increased , and then steadily diminished for further increase of charge on the needle .
It was now , however , observed that , if the distance separating the quadrants was increased to inch , the sensibility tended to a limit for a large charge instead of first increasing and then diminishing .
Before replacing the new needle , it was carefully weighed with the platinum wire attached and the weight dipping into the acid , and a calculation was made as to the magnitude of the effect that should arise from the change of the pull of the fibres due to any upward or downward attraction of the needle by the quadrants .
This calcuIation showed that for a P.D. of 3000 volts between the needle and the the amount of such attraction was quite unable to account for the observed diminution of sensibility with large charges in the needle .
Dr. HOPXINSON says in his paper , " " Increased tension of the fibres from eIectrical attraction does not therefore account for the whole of the facts , although it does play the principal part The experimenb that we made at the end of 1886 and beginning of 1887 , confirmed by the calculation above referred to , proved that , at any rate in our specime\amp ; nacute ; of the quadrant electro meter , the principal part of the anomalous action was not caused by an inereased tension of the fibres , and that , therefore , some other cause must be looked for explain the observed results .
The following investigation was , therefore , to ascertain the cause , and the means of removing it , and we are happy to say ' We do not know whether the weight , with the fine wire attached , supplisd us by Nessrs .
exactIy the same as is supplied generally by them with the quadrant electrometer , if so erience a very cessary difficulty from the Blackbupendulum-like vibration of the needle .
ON QIiADRANT ELECTROMETERS .
resulted in our finding out that a quadrant electrometer , as made by Messrs. WmTE can , by a specia ] adjustment being giveo to its quadr , , be made to fulfil the law given in text books the quadrant electrometer ; secondly , that this adjustment may be dispensed with if the construction of the instrument be altered ; and thirdly , that when this alteration in the construction is carried out , it is possible to obtain an instrument having a sensibility many times as great as that obtainable with the White form of construction .
AB long as an eIectrometer is only employed to measure a steady P.D. between the quadrants , 'it may not be of much interest to know the way in which ] the sensibility varies with the charge in the needle , unless great sensibility be required ; in which case it is important to know for what charge on the needle maximum sensibility is obtained ; in fact , for the measurement of steady P.D 's .
the main thing to be sure of , is that the deflection is proportionaI to the P.D. between the quadrants when the charge in the needle is constant .
But when the eIectrometer is employed in conjunction with a non-inductive resistance to measure the power given by an aIternate or an intermittent current , which may be any function of the time , to a circuit that may possess capacity , or an E.M.F. , or self or mutual induction , or all four ( in accordance with the method devised by Professor FITZGERALD and ourselves during the meeting of the Electrical Congress in Paris in 1881 , and published the same autumn by M. POTIER ) , it is absolutely necessary that the electrometer should exactly fulfil the law\mdash ; , where is the torque exerted at any moment on the needle by electrical action , and , and the potentials , at the same time , of the opposite pairs of quadrants and of the needle respectively relatively to the outside case of the instrument , and is a constant .
[ See Section VI .
" " Electrometer used to Measure Power Much has been written during the last three or four years by MM .
GouY , LEDEBOER , and others regarding departure of the action of the quadrant electrometer from the law given above , but we are not aware that hitherto the cause of the discrepancy has been pointed out , or the means of removing it and obtaining an instrument that fulfils this law .
We thought at first that the curious results might be due to some capillary action between the platinum weight and the sulphuric acid which varied with the potential of the weight , but , experiments having shown that no such action existed , it was thought desirable to measure the P.D. between the needIe and the outside case of the instrument by means of one of Sir WILLIAM THOMSON'S absolute electrometers ( made especially sensitive by thinning the little coach springs carrying the movable aluminium disc ) , and so ascertain completely the laws connecting the variation of sensibility of the quadrant electrometer with the potential of the needle , the distance between the silk fibres , the distance apart of the quadrants , \amp ; c. The investigation has taken some years to complete , and in carrying it out this quadrant has had to be taken entirely to pieces many times .
The labour that this has will be appreciated by anyone who has had experience in taking a White to pieces .
In the ordinary form of the quadrant electrometer as constructed by Messrs. it is not possible to get at the guard-tube or the lower part , either of the of the quadrants , without taking the whole interior mechanism bodily out of Leyden jar .
Since in doing this there is danger of breaking the silk fibres which hangs the minium needle with its attached platinum weight , as the pull on the fibres is , of course , much increased when the weight is lifted out the dense sulphuric acid , it may be advisable to clamp the needle and take stress off the fibres before lifting out the interior of the instrument .
And , if , example , it , be desired to the aluminium wire relatively to the guard-tube which it hangs , or to remove this guard-tube so as to be able to get at the attachment of the aluminium wire , or to straighten the needle if slightly bent , or , generally , to make any adjustment liable to break the silk fibres , it is certainly advisable to clamp the needle before removing the interior of the instrument .
Under these stances the following operation has to be performed : .
Clamp the needle , the interior of the apparatus out of the Leyden jar , and place it on some support provided with levelling screws by which the instrument can be levelled .
2nd .
Make the necessary adjustment , and unclamp the needle to see whether adjustment has been correctly made .
3rd .
Replace the interior of the apparatus position inside the Leyden jar .
The method ordinarily adopted for clamping the needle consists in passing a through a loop in the aluminium wire connecting the mirror with the needle , screwing this screw into the insulated framework supporting the apparatus .
There is not much difficulty in putting in this screw by means of the square screwdriver provided with the instrument , but there is the greatest difficulty ON QUADRANT ELECTROMETERS .
moving the screw without breaking the fibres .
In fact , the danger of breaking the fiku when making one of the adjustments above referred to without previously damping the needle is hardly greater than the danger of breaking them when removin the clamping screw .
We , therefore , had fitted to the electrometer the device which is seen in fig. for easily raising and lowerin the needle .
is a circular piece of metal turning eccentricaly about a pin fixed to a small plate , which can slide up and down at the back of the support .
Ths plate carries two little hooks , which , on being raised by turning the eccentric ( by means of the square screwdriver provided with the electrometer ) , lift up the -piece at the top of the suspension , and take the weight of the mirror , the needle , and its attachment , off the silk fibres .
When then it is desired to remove the interior of the electrometer out of the Leyden jar , all that is necessary to be done is to turn the eccentric one way to raise the needle , and the other way to lower it , either of which can be done quickly without the slightest risk of breaking the fibres .
If the instrument is to be packed for travelling it will be well to also clamp the needle by means of the screw inserted as usual in the loop on the aluminium wire , but as the eccentric lifting arrangement which we have added to the electrometer enables the screw to be inserted and removed , if it be desired to use it , when there is no tension whatever on the fibres , the clamping and unclamping by means of the screw can be done with perfect safety .
We have to thank Mr. H. B. BOURNE for helping us in designing this eccentric attachment for the White electrometer .
We may take this opportunity of mentioning another modification that we have * In order that the mirror may be clearly seen , one half of the horizontal guard-tube which surrounds the mirror has been omitted in fig. 1 .
introduced , with the assistance of Mr. BOURNE , for the purpose of the good working of the quadrant electrometer .
In the instrument as constructed the framework of replenisher is always in connection witk sulphuric acid which forms the interior coating of the Leyden jar , and electric leakage can always take place over the framework of the replenisher .
avoid this we altered the replenisher in one of our quadrant electrometers , so by raising a knob on the top outside the instrument the replenisher could be raised up until the platinum wire attached to it waS lifted out of the acid .
think that the following very simple device which we have attached to electrometer is in every way better .
Instead of the replenisher being electr connected with the interior of the Leyden jar through the acid , as is the case Messrs. WHITE'S instruments , we arrange that it can be connected with , or nected from , the guard-tube ( fig. 2 ) , which is itself always in connection with acid .
This is done thus : , the insulated inductor of the replenisher , is by means of a flexible wire to a springy wire carried at the end of an ebonite .
By turning the milled head this springy wire can be made to touch , not to touch , the guard-tube , and thus II can be connected with the interior of Leyden jar , when the replenisher is to be used , or left wholly disconnected from Leyden jar , so that there is absolutely no leakage over the replenisher frame , the replenisher is not in use .
With this device we can dispense altogether with the clamping arrangement that is usually employed for holding the rotatory spindle that supports the carriers of enisher in such a position that the carriers are not in contact with the inductors .
In the ordinary use of the quadrant electrometer the charging rod is kept disconnected from the Leyden jar , and therefore the insulation of the charging rod need not be very perfect , since leakage 'an only occur over it during the time the Leyden jar is being charged ; in these experiments , however , as the interior of the jar of the quadrant electrometer and the needle had to be permanent , lie connected with the lower movable disc of the absolute electrometer , it was necessary to keep the charging rod of the quadrant electrometer always in connection with the Leyden of this instrument .
This made the experiments very difficult to perform on damp indeed on some days hours have been spent without a single trustworthy result obtained .
Hence the results given in the following tables , which are the trustworthy ones , only represent a small portion of the work that has been actually carried out .
As the two eters are .rmanently fixed on two brick pillars , some 14 apart , in the laboratory , the absolute electrometer being close to the light , the reflectin quadrant electrometer with its scale well away from the light , the connection the lower disc of the absolute and the charging rod of the quadrant was made by a long fine wire supported on a well insulated glass rod , shielded dust and kept dry by sulphuric acid .
The outsides of the two instruments were connected together by an uninsulated copper wire .
ON QUADRANT ELECTROMETERS .
In the earlier of the experiments the Leyden jars of the two electrometers were oharged quite irrespectively of one another , sometimes , and sometimes negatively , but as the investigation advanced both jars were always charged positively , the P.D. between the inside and outside of the .
jar of the absolute electrometer being maintained at volts , the force required to bring the movable aluminium disc to the sighted position was always the same , as the reading of the upper micrometer screw was not varied .
The attracted aluminium disc was brought to the sighted position by moving the lower plate only , and the potential of the needle of the quadrant electrometer was measured by showing the difference between the readings of the screw moving this lower plate ( 1 ) when the lower plate was connected with the outside of the absolute electrometer ; ( 2 ) when the lower plate was disconnected from the case and connected with the needle of the quadrant electrometer .
The absolute electrometer behaved well when the upper micrometer screw was not touched , the earth reading for a fixed P.D. between the inside and outside of the Leyden jar varying but very slightly from day to day .
A series of calibrations of this absolute electrometer carried out in the ordinary way by placing known weights on the movable aluminium disc to reproduce the effect of electric abtraction , showed that with the position we gave to the upper micrometer screw , and taking " " \ldquo ; as 30 true ohms , a motion of the lower plate through one division of the vertical scale corresponded with volts .
In , 1888 , several large P.D. 's were measured by means of this meter ( using the constants that we had determined for this instrument and also by means of one of Sir WILLIAM THOMSON'S commercial ' electrostatic voltmeters \ldquo ; reading to 20,000 volts , kindly lent us by Messrs. ELLIOTT BROS .
The result of these comparisons led first to a correction in the constants that had been sent out with the electrostatic voltmeters from Glasgow , and secondly to a new determination of the value of For Sir WILLIAM THOMSON had calibrated these voltmeters by galvanometer method , based on the value of the electro-chemical equivalent of silver , as determined by Lord RAYLEIGH and Mrs. SIDGWICwhile we had checked the calibration of the electrostatic voltmeter , by comparing this instrument with the absolute electrometer .
The value of " " \ldquo ; thus obtained was 298,000,000 metres per second .
The scale employed with the quadrant is a circular one at 5 feet distance from the mirror .
There are 290 divisions in foot length of scale , and as both scale and electrometer are permanent]y screwed down in position , one of the scale always corresponds with radian deflection of the mirror , or about of a degree .
The scale is so placed that the mechanical zero of the electrometer corresponds with on the scale , that is to say , the spot of light is at when the needle adjusted as symmetrically relatively to the quadrants as it possible to do with the $ ' Electrometric determination of Brit. .
Report , ' lS88 .
See also ' Electrician , ' p.448 , vol. , 1889 , the correction for the effective area of the attracted disc in the absolute electrometer .
MESSRS .
W. E. AYRTON , J. PERRY , AND W. E. SUliPNER eye , the whole instrument being discharged .
The electrical zero is next ma\amp ; coincide with the zero , that is to say , on the needle whenfour quadrants are connected with the outside case of the instrument , the quadrant is slightly moved until the spot of light is brought back to the same that it occupied on the scale when the needle was discharged .
But it was found if this were done for some particular potential of the needle , a fresh adjustment of fourth quadrant had to be made for each potential to which the needle was or , in other words , it was practically impossible so to adjust the quadrants that out further touching them the electrical and mechanical zeros should be the same a considerable variation of the potential of the needle .
Hence , in what follows ths position of the electrical zero is in each case give1J .
The question of the electrical and mechanical zeros differing from one another will be further dealt with in Section V , , entitled " " Motion of the Electrical ero \mdash ; EFFECT OF VARYING THE DISTANCE BETWEEN THE FIBRES .
If be the P.D. in volts between the needle and the outside case of the electrometer , and the deflection of the needle in microradians produced by a P.D. equal to V between the opposite pairs of quadran , one pair being connected with the outside case , the ordinary formula given for the electrometer reduces to , where is usually regarded as being constant both for variations of V and of , the distance between the quadrants and the distance between the fibres being ) constant .
The first thing tested was the constancy of for a fixed value of and for varying values of V. The relative values of V were measured by means of a Thomson 's and Varley 's potential divider , which enables a P.D. to be accurately subdivided to the part if required , and it was found that for any large and fixed charge on the needle less than the value which produced instability , and for any fixed position of the quadrants and distance between the fibres , the deflection of the needle was proportional to V to within 2 per cent. , that is to say , when V and alone were the variables was always constant to within 2 per cent. In the great majority of cases the constancy of was much closer than this , but , as it was found necessary to vary the position of the fourth quadrant in order that the deflections for all the different values used for V should be on the scale when the charge in the needle was vari,4 the want of constancy of for unsymmetrical positions of the instrument was sometimes nearly 2 per cent. On Sheet I. ( Plate 9 ) are seen two curves connecting the double defle tion of the ON QUADRANT ELECTROMETERS .
electrometer needle with the deflecting P.D. between the quadrants .
One set of results was obtained when the P.D. between the needle and the outside case of the instrument was maintained at 1793 volts , and the other when the needle 's potential was 2510 volts .
In each case the curve is practically a straight line , and it will be observed that in spite of the needle 's potential having been increased about 40 per cent. , the combination of these two lines is practically the same .
The next point tested was the constancy of when was varied .
When is large the preceding formula is approximately so that In the following tables is the deflection of the ray of light in divided by the product of bhe deflecting P.D. in volts maintained between the quadrants into the P.D. in volts between the needle and the case of the electrometer .
is thus the sensibility factor of the instrument for a given position of the quadrants , and for a given distance between the fibres ; and the different values of not only show how the behaviour of this electrometer departs from the simple mathematical law , but enables its sensibility to be compared with the sensibility of any other quadrant electrometer which may be tested absolutely in this manner .
On November @ , 1887 , the fibres were adjusted so as to be parallel , and at about 2 millims. apart , and the results of the experiments are given in Table I. The values for the double deflection per one Leclanche cell for the different values of are also plotted in the curve on Sheet .
( Plate 9 ) , corresponding with ' ' Fibres Parallel It is interesting to notice that while is far from bein constant for different values of it has the same value for the same value of , although V and are very rent , thus confiming the experiments previously made with fixed values of and diiferent values of V. TABLE TI .
( continued).\mdash ; December 4 , 1887 .
Some Experiments repeated to the Accuracy of those previously made .
From the dates given for the tests contained in Table II .
, it will be seen that this group was made at intervals lasting over one month .
The sensibility factor is very far from being constant , varying from about 14 when the potential of the needle was about 3000 volts to about 50 when it was about 300 volts ; but nevertheless for each potential of the needle with the constant distance between the quadrants , and between the fibres , the sensibility has practically a perfectly constant value .
The following examples selected from Table II .
illustrate this : The small differences between the values of the sensibility factor on November 4 , and on the later dates for the same potential of the needle , arises probably from the fact that in the interval the fibres were separated far apart at the top to enable next set of experiments to be made , and approached again on December @ , until ON QUADRANT ELECTROMETERS .
they were again , as far as the eye could judge , in the same position as on November 4 .
In changing the distance between the fibres , as well as in alt , ering the tension of the fibres to adjust in each case for minimum sensibility ( corresponding with equal tension ) only one of the rollers on which the fibres are wound was touched , in order that a position could be reproduced as nearly as possible .
But it was , of course , impossible to exactly reproduce a previous position of the fibres , and the sensibility factor for a fixed potential of the needle , and for a so-called fixed position of the fibres was , consequently , not absolutely constant .
In view , however , of the extreme delicacy of the tests and the great variation of the sensibility factor with a variation of the potential of the needle , the wonder is that for the same potential of the needle the sensibility factor was as constant as it was .
On November 8 , 1887 , the fibres were separated until were far apart at the top , the tension adjusted for minimum sensibility , and experiments were made on November , 8 , 10 , 11 , and December 2 , to determine the values of the sensibility factor .
The results are given in Table III .
, and the values of the double deflection per Leclanche cell for the different P.D 's .
between the needle and the outside case are plotted in the curve on Sheet II .
( Plate 9 ) , corresponding with " " Fibres far apart at the Top ON QUADRANT ELECTROMETERS .
From the Tables of Results and from the curves on Sheet II .
( Plate 9 ) , we see that when the fibres are parallel the sensibility factor falls off somewhat as the potential of the needle is increased , whereas with the fibres far apart at the top the sensibility factor first falls off , and then increases again as the potential of the needle steadily increases .
With the fibres near together at the top the curve is perhaps the most interesting , as here the sensibility factor faUs off so much for high potentials of the needle that the deflection of a fixed P.D. betweeen the quadrants is actually smaller for a high potential of the needle than for a lower potential .
III.\mdash ; EFFECT OF VARYING THE DISTANCE BETWEEN THE QUADRANTS .
The reasoning which leads to the ordinary formula for the quadrant rometer ( where is the deflection produced by a P.D. equal to V between the opposite pairs of quadrants , one pair being connected with the outside case of the instrument , and the P.D. between the needle and the outside case ) , shows that is proportional to the variation of capacity of the needle per radian deflection .
Consequently it might be expected that since the needle is less covered by the quadrants when they are drawn out , and , since the variation of capacity of the needle per radian deflection is when the quadrants are pulled out than when they are in , that the instrument would be less sensitive when the quadrants are drawn out than when they are in .
But those who are accustomed to use the quadrant electrometer know that the opposite is the case .
Here again then there is a marked difference between the actual behaviour of the instrument , and behaviour that the simple formula would lead one to expect .
In all the preceding experiments three of the quadrants had been left untouched with their edges at from to 3 millims. apart , and the fourth quadrant , the one to which the micrometer screw is attached , bad alone been moved .
But at the beginning of 1888 experiments were made to determine the way in which the sensibility factor varied with the potential of the needle for different distances between the quadrants and for different distances between the fibres .
When the quadrant under the induction plate was moved far out , it seemed quite possible that the latter would exert some influence on the curves .
The deflections , however , were found to be practically unaffected by disconnecting the induction plate from one pair of quadrants , and connecting it with the other pair .
As a rule the induction plate was kept electrically oonnected with the quadrant immediately underneath it .
The distances between the An examination of the preceding results , and of the curves on Sheet III .
( Plat corresponding with them , shows that when the quadrants are near together the bility factor diminishes at first slowly and afterwards rapidly as the potential of needle is steadily increased .
As the distance between the fibres increases , the rate diminution of the sensibility factor , with increase of the potential of the needle , less until , when the quadrants are about 4 Inillims .
apart , the sensibility factor actua increases with increase of potential of the needle , so that the curve connectin deflection for a constant P.D. between the quadrants with a varying P.D. betw the needle and the outside case , becomes convex instead of concave to the axis which the potential of the needle is reckoned .
HeIlce , with our specimen of the White electrometer , when the distance between the quadrants is about millims. the sensibility factor is constant , the ) for the quadrant electrobecomes the true formula .
The next set of experiments was made with the fibres far apart at the top , the fibres being carefully adjusted for equality of tension after having been separated .
The sults are given on Table V. , and they are also plotted on Sheet IV .
( Plate 10 ) ; it will be seen that the general shape of the curves for the various distances between the quadrants is the same as was obtained when the fibres were near together .
The curves contained on Sheets III .
and IV .
( Plate 10 ) show that this electrometer may be adjusted so that the variation of sensibility with the the needle may be made to follow one or other of three taws .
If the rants be near together there are certain limits between which the potential of needle may yary without producing more than a small change in the corresponding with a fixed P.D. between the quadrants ; for example , when quadrants were about millims. apart , and the fibres near together at the top , deflection produced by a P.D. of volts between the quadrants only varied per cent. when the potential of the needle was increased from 896 to 3586 volk When the fibres were far apart at the top it was when the quadrants were aboub 1 millim. apart as seen in Sheet III .
( Plate 10 ) that a similar flatness was obtained in the curve connecting deflection with potential of the needle .
In this case the deflection of the needle was practically quite constant when its potential varied between 2152 and 3227 volts ; and even when the potential of the needle was increased from 1434 to 3407 volts , that is , by nearly 2000 volts , the deflection only increased by less than 9 per cent. This arrangement of the quadrants gives but a comparatively small sensibility , but , where great sensibility is not required , it would be a convenient one to employ , as leakage of the Leyden jar , or loss of potential of the needle due to the rapid absorption that occurs when thejar is first charged , would only slightly affect the deflection for a fixed P.D. between the quadrants .
When the quadrants were at about millims. apart the deflection for a giveri .
between the quadrants was almost directly proportional to the potential of the needle .
This then would be the arrangem.ent to employ when the electrometer is used with alternating P.D 's .
And , lastly , when the quadrants were 4 or more millims. apart the deflection increased much more rapidly than the potential of the needle , so that maximum sensibility , bordering on instability , is obtained with this arrangement of the quadrants .
For each distance between the quadrants less than the distance at which instability begins to occur , the shape of the sensibility ve is perfectly definite , but its shape alters very much for an exceedingly small alteration in the position of the quadrants when the distance separating them is some three or more millims. It might have been anticipated that the conventional mathenlatical law for the quadrant electrometer , viz. :\mdash ; ( where is the deflection , and , and the potentials of the two pairs of quadrants and of the needle relatively to the outside case ) would have been most nearly fulfilled when the quadrants were pushed far in , so that the needle well under cover of them .
But the preceding results show that this is very far from being the case , and that constancy in the sensibility factor , for different values of ON QUADRANT ELECTROMETERS .
obtained only when the quadrants are so far separated that the sensibility of the instrument for the particular adjustment of the distances between the fibres is fairly large .
\mdash ; CAUSE 0F THE PECULIAR ACTION 0F THE WHITE ELECTROMETER .
Towards the end of 1887 the cause of the irregularity in the action of the Thomson quadrant electrometer , as constructed by Messrs. WHITE , began to dawn on us .
The wire supporting the aluminium needle , as well as the wire which connects the needle with } sulphuric acid in the Leyden jar , is enclosed in a metallic guard-tube to screen the wire from external action .
But in order that the needle may project outside the guard-tube , , openings are made in its two sides , as seen in fig. 3 .
Hence the moment the needle is deflected from its zero position each half of the needle becomes unsymmetrically placed relatively to the two bits I and I ' of the guard-tube .
Therefore , in spite of the needle and the guard tube being always maintained at the same potential , there is a repulsion between the charges on the two bits I and I ' of the and the charges on the two halves of the needle ; and this repulsion might not only have the defect of seriously diminishing the sensibility of the quadrant electrometer , as made by Messrs. WHITE , but might cause the variation of sensibility of the electrometer with variation of the P.D. between the needle and the outer coating of the Leyden jar to follow a far more complicated law than that expressed by the conventional formula just given .
To test this theory , that the peculiarities in the law of the quadrant electrometer are due to the electric action of bhe guard-tube on the needle , in consequence of the special shape of the former , the want of symmetry of the guard-tube was intensified and varied by attaching a piece of thin aluminum foil to it above and below the needle , and experiments on the variation of sensibility of the electrometer with variation of the potential of the needle made for the positions of this piece of aluminium foil , , shown in figs. 4 and 5 .
These figures show a section of the guardbube through the needle , I and I ' , figs. 4 and 5 giving the sections of the connectlng pieces I and I ' seen in fig. 3 .
The difficulty connected with these experiments was MESSRS .
W. E. AYRTON , J. PERRY , AND W. E. SUMPNElt very great , arising from the almost impossibility of attaching the aluminium foil without slightly altering the position of the tube .
And as the alteration of the position of the guard-tube considerably altered the electrical was necessary in each case to readjust the movable quadrant in order that electrical zero might coincide with the mechanical zero .
This , in itself , altered sensibility curves slightly , and the effect due to this cause must be kept in mind comparing the sensibility curves made before attaching the foil , for the positions of the foil , and after removing it .
These experiments on the guardwere more troublesome than any of the others , and very great care was expended them .
Many sets of experiments were made and curves drawn to the but , since in the majority of cases it was considered that too much disturbance of position of the guard-tube had been produced in the act of attaching the foil or altering its position , the majority of the results were rejected .
The following sets , given in Tables VI .
and VII .
, were obtained on December 6 , 7 , and 8 , 1887 .
They correspond with positions of the adjustab]e quadrants varying between and , and , as subsequent experiments made on December 9 ( the results of which are given in Table IX .
) confirm the idea that this amount of variation of the adjustable quadrant alone would not account for the variations in the results that were obtained on attaching the thin piece of aluminium foil to the side of the guard-tube , and on moving this foil from the side to the back of the guard-tube , the results given in Tables VI .
and VII .
may be regarded as correct .
They may , there- fore , be compared with those contained in the latter part of Table II .
, and obtained on December 3 and 4 before the foil was attached , as ] as with those contained in Table VIII .
, and obtained on December s , after the foil had been removed .
For in all these experiments the variation in the position of the adjustable quadrant was kept between the same limits and , the position of the other quadrants remained the same , and there was the same distance between the fibres , which were adjusted for minimum sensibility when the potential of the needle was about 1800 volts .
The results are all plotted on Sheet V. ( Plate 11 ) , and it is important to that : 1 .
The curves obtained with ' no foil\ldquo ; were practicaIIy the same before attaching and after removing the foil , and had ordinates much higher than when the foil attached .
2 .
The curves obtained with ' foil the side of the guard-tube\ldquo ; had the higher ordinates .
ON QUADRANT ELECTROMETERS .
3 .
The curves obtained with " " foil at the of the -tube\ldquo ; had the lowest ordinates .
We may , therefore , conclude that the attachment of a small piece of foil to the guard-tube not only altered the shape of the curve connecting the sensibility of the electrometer with the potential of the needle , but , whether the foil was put at the side or at the back of the guard-tube , it diminished the sensibility for high potentials of the needle , in fact , exaggerated the very peculiarity in the behaviour of the instru- which is under investigation .
This makes it almost certain that the reason why , under ordinary circumstances , a high potential of the needle does not produce a high sensibility , lies in the fact that the guard-tube is not continuous all round in neighbourhood of the needle .
The up which may be seen at the ends of the curves on Sheet V. ( Plate 11 ) , as well as at the ends of many of the curves on the preceding sheets , is probably due to a slight tilting of the needle at high potentials .
Such a tilting brings the ends of the needle near the quadrants , and much increases the sensibility .
( See Section IX .
, " " Sketch of the Mathematical Theory of the White Electrometer TABLE VI .
\mdash ; December 6 , 1887 .
Thin Piece of Aluminium Foil at the side of the Guard-Tube .
ON QUADRANT fERS .
VIII.\mdash ; December 8 , 1887 .
Aluminium Foil Removed from the Guard-Tube .
The two last sets of experiments made with the aluminium foil at the back of the guard-tube , and after the foil had been removed , show that no considerable change is made in the double deflection by turning the micrometer screw attached to the adjustable quadrant through 50 divisions , that is , turning it through Some experiments were therefore made to obtain further information on this point , especially to see how a motion of the adjustable quadrant affected the sensibility for different potentials of the needle , and the results are given in Table IX .
It will be observed that , for a given potential of the needle , the effect produced on the sensibility , by attaching or altering the position of the foil , is much greater than the effect produced by the corresponding motion of the adjustable quadrant , so that we are justified in regarding the greater part of the difference between the curves on Sheet V. ( Plate 11 ) , as really due to the presence and position of the foil on the guard-tube .
MESSRS .
W. E. AYRTON , J. PERRY , AND W. E. SUMPNBR TABLE IX.\mdash ; December 9 , 1887 .
Effect of Varying the Position Quadrant .
Double Deflection for a Deflecting P.D. of Volts .
V.\mdash ; MOTION 0F THE ELECTRICAL ZERO .
It was found to be practically impossible to adjust the needle , quadrants , and other parts of the electrometer , so that the electrical and mechanical zeros should be the same for all P.D. between the needle and the outside of the instrument , although , when the controlling force was made somewhat large by separating the suspendin fibres at the top , it was possible with a careful adjustment to render the difference between the electrical and mechanical nearly nought for all potentials of the needle up to some 1400 volts .
For example , on January 30 , 1888 , when the fibres were separated as far as is possible with the ordinary construction of the WHITE form of quadrant electrometer , and the quadrants and needle very carefully adjusted for symmetry , the electrical zero only differe by three scale divisions from the mechanical , when the needle was gradually charged up from to 1434 volts .
On another occasion this motion was reduced to two divisions for the me limits of the potential of the needle , and in a third case there was no perceptible motion of the spot of light on charging up the needle to this potential .
But even in these there was a considerable motion of the electrical zero for still higher P.D. between the needle and the outside case of the electrometer .
On one occasion , when the fibres were very near together , .
and the coritrolling force therefore weak , the electrical zero only moved through ten divisions the mechanical zero when the needle was charged up to as high as 2869 volts .
This was , however , a totally exceptional result , and was obtained , if by accident , after very many trial adjustments .
The adjustments of the quadrants , however good they may seem to the eye , are not accurate enough to cause the electrical and mechanical to remain constant for a weak control of the needle as the needle is steadily up .
It does not seem to be sufficient to have only one quadrant adjustable by ON QUADRANT of a micrometer screw , and we would suggest that , if two quadrants were provided with a micrometer screw motion , much time would be saved .
It is , of course , possible , by moving the adjustable quadrant , to make the electrical zero coincide with the mechanical zero for some one potential of the needle , but although the coincidence is made to occur when the potential the needle is , say , 2000 volts , it by no means follows that the two zeros will agree when the needle 's potential is 1000 volts .
And the experiments , of which some of the results are given in the following Tabie X. , and plotted on Sheet VI .
( Plate 12 ) , show that when the four quadrants are adjusted symmetrically , as far as the eye can judge , following result is obtained .
First , the electrical zero moves away from the mechanical zero as the potential of the needle is increased ; then , having reached a certain distance from the mechanical zero the electrical zero returns towards the mechanica ] zero as the needle 's potential is still further increased ; next it passes through the position of the mechanical zero , and finally it moves away farther and farther down the other side of the scale .
TABLE X.\mdash ; November 8 , 1887 .
Motion of the Electrical Zero .
Position of the Electrical Zero Measured from the Position of the Mechanical Zero .
ibres near at the top .
Quadrants about milliJns .
apart .
LiECTROMETER USED TO MEASURE When the power given by any current to any circuit ( fig. 6 ) has to be measured by means of a quadrant electrometer the method is as follows : series with is connected a non-inductive circuit , of resistance ; the opposite pairs of quadrants are joined respectively to and , and two observations are taken , one when the needle is joined to , and one when it is joined to .
Then the difference between these readings divided by is proportional to the power given to .
A second method consists in connecting the opposite pairs of quadrants respectively with and and to take two one when the needle is joined to , and one when .
joined to .
Then , as before , the difference between these two readings divided by is proportional to the power given to ab .
If be the primary coil of a transformer , the P.D. between the terminals of will be 1000 or 2000 , or more , , volts , the first of the two methods just described of joining up the electrometer will be inapplicabJe , for , with the ordinary adjustment of this instrument , the spot of light would fly off the scale if one pair of quadrants were connected with and the other pair and the needle with , and the P.D. between and were some thousands of volts .
Hence the second method of joining up apparatus is alone applicable in such a case .
When the quadrants are .
joined to and , the needle to , and the outside case of the electrometer to , the instrument is being used idiostatically .
The text-book law where is the deflection , , and the potentials of the opposite pairs of quad- ants , and of the needle relatively to the outside case of the instrument , reduces to when the needle is connected with one pair of quadrants , and the outside case with the other pair .
Now a P.D. of only some 100 volts between the needle and the outside case will deflect the spot of hght to the end of the scale , when the electrometer is used idiostatically .
we see that the first part of each of the curves on Sheets II .
, , and IV .
is straight , that is , that the text-book law holds for the quadrant electrometer made by Messrs. WHITE when the P.D. between the needle and the case does not exceed a hundred yolts or so .
Hence we should anticipate that when this instrument is used idiostatically the deflection would be directly proioned to the square of the P.D. between the quadrants for any fixed position of the quadrants .
And experiment proves this to be the case , for a number of tests made to check the accuracy of the square law showed that the numbers obtained by dividing the deflection by the square of the potential of the needle did not differ from one another by more than one per cent. for deflections in all parts of the ] he various P.D.@ were obtained and measured by comlecting up 200 silver chloride cells with the terminals of a Thomson-Varley sliding P. L ) .
divider , the ' zero\ldquo ; point being joined to ON QUADRANT one pair of quadrants and to the outside of the electrometer , while the " " index\ldquo ; was joined to the other pair of quadrants and to the needle .
The square law obtained on using the quadrant electrometer idiostatically is never quite true for small P.D. , since , as pointed out by Dr. HOPKINSON and the authors some years ago , the contact P.D. between the needle and the quadrants auses the deflection to be different for the same P.D. between the quadrants , depending on whether the needle is charged positively or negatively .
When making the second observation with the second method of joining up the apparatus for measuring the power given to ab , we have the pairs of quadrants joined respectively to and , between which there may be a P.D. of 1 or 2 volts , and the needle joined to , the potential of which , when is the primary of a transformer , may differ by some thousands of volts from that of the quadrants or of the outside of the Leyden jar .
In other words , we are now using the electrometer heterostatically in exactly the way in which it is ordinarily ] oyed , excepting that , if an alternating or intermittent current be passing through the circuit abc , all the potentials will be varyin Now we have shown that when the P.D. between the needle and the outside case is some hundreds of volts or more , the formula does not hold unless the quadrants be in one definite position .
Hence the deflection will not measure the mean value of the expression on the right hand side of the last equation unless the quadrants be in this definite position .
But the electrometer method of measuring power is based on the assumption that the deflection measures the mean value of Hence it follows that the electrometer , if measuring the power given to the primar coit of an ordinary nsformer , will not give correct results unless the be in the one special position we have pointed It might be thought that , although the electrometer did not obey the usual law given for it , it would be possible to calibrate it empirically as a watt meter .
As regards one of the observations , viz. , that made when the needle and one pair of quadrants are joined to ( fig. 6 ) and the outside case and other pair of quadrants to , it is possible to calibrate the electrometer empirically , whatever its law may be .
For any special deflection can , for a given adjustment of the instrument , be ] produced by one special P.D. between the quadrants .
But on making the second observation when the ants are joined respectively to and and the needle to the same deflection may be obtained with many different potentials of the points enoe it follows , that unless the deflection is connected with the P.D. the fingers .
To do this , however , the ebonite tube must be tightly grasped in fingers , and as they are necessarily chemically dirty , there is the possibility Leyden jar being partially discharged in the act of disconnecting the charging This objection is entirely overcome with the arrangement , seen enlarged in Fig which we have adopted for the charging rod in the improved quadrant electro At the top of a glass rod is cemented a cap , to which is attached a piece of the lower end of which is fixed a pivot , round which the charging rod can The ball at the upper end of this rod is made of lead , and its weight causes the rod to emain in the position shown by the full line , or by the dotted line .
When in the position shown by the full line , the ball is in electric contact with the needle , guard-tube , and the interior of the Leyden jar , the area of contact being increased by the bending of the little piece of flat spring fastened to the ball .
The Leyden jar can now be charged , and at the moment that the right amount of charge has been given , the charging rod is tilted over into the position shown by the dotted line , and the electric connection is broken .
This tilting over is easily done by pushing the lower end of the rod with the eleclrophorus , if an electrophorus be used to charge with , or by means of a glass rod , or by slightly pulling the wire attached to the binding screw at the lower end of , if such a wire be used to charge through .
The slide is now closed to exclude dust and moisture .
When this electrometer is employed for alternating current work it is desirable to connect the binding screw by means of a fine wire with a binding screw attached to the interior of the instrument , so that the electric contact between the charging rod and the interior of the instrument ( and which should be very good for very rapid reversals of the charge ) does not depend solely on the pressure of the leaden ball at the top of .
To make such a wire connection would be practically impossible with the WHITE form of the quadrant electrometer , since the charging rod must be bodily removed from the instrument before the part of it inside the electrometer can be got at .
It is also desirable , when an electrometer is employed for measuring a rapidly alternating P.D. , that the electric connection between the electrodes and the quadrants should be far better than can be obtained with the comparatively slight pressure that exists between the ends of the electrodes and the quadrants in the WHITE form of instrument ; hence in the improved quadrant electrometer the connection with the quadrants is made by means of screwed contacts .
Further , for alternate current work it is not only important to ve as good contacts as possible- , but also as little capacity as possible in the electromet , This result we attain by removing the springy wire , and electrically connecting the interior llic coating of the glass shade with the exterior .
The next point to consider in designing this rument wag the size.to give to the needle and quadrants .
Theoretical reasoning , confirmed by some ' roughly madt experiments , show that the sensibility of the quadrant electrometer ; , as mad-e by ssrs .
WHITE , could be much increased if the needle and the quadrants were made ON QUADRANT ELECTROMETERS .
smaLler than in that instrument .
We therefore made the diameter of the quadrants , their vertical depth , and the length of the needle about six-tenths of the size adopted in the White pattern .
Feeling , however , that the needle in the White pattern was not broad enough at the widest part to prevent its edges approaching too near the intersection of the quadrants when the needle was deflected , we thought it better not to reduce the width of our needle in the same proportion as its length .
Therefore while we reduced its length to six-tenths we left its breadth eight-tenths of the sizes adopted in the White electrometer .
Then as regards the controlling force .
In our laboratories are electrometers with a bifilar suspension control , a torsional control , and a magnetic control respectively , and of these three forms we have found that the magnetic control is the best to be used in an electrometer , whose sensibility we require to be able to easily vary .
The bifilar suspension control is unnecessarily troublesome , arising partly from the frequent necessity for adjusting for equal tension the two fibres .
or , in the White form of quadrant electrometer , no alteration in the centering of the needle , or in the height of the needle , or in the distance between the fibres , can be effected without a fresh adjustment being rendered necessary for equality of tension in the fibres .
The magnetic control has no doubt the defect that the position of the needle is liable to be ffected by outside netic disturbance .
But this source of error can be practically annihilated if the magnet attached to the suspension wire be very short and be very ) magmetised , and if the requisite magnetic couple be produced by the use of a very powerful controlling placed near to the suspended system , and we venture to think that it was a neglect of this principle that caused the abandonment of the magnetic control originally employed by Sir WILLIAM THOMSON in his quadrant electrometers .
Another plan that we have found very successful is to fix on the suspension wire an combination of magnets , and to place the controlling magnet nearer one of the little magnets than the other .
But probably the most perfect plan is to use , as the astatic combination , two little vertical magnets , which may be conveniently attached to the back of the mirror , as seen in fig. 9 , since such a cornbination remains as far rotatio round vertical is concerned , no how much one or both of the magnets weaken with time .
Where completeness in the electrometer is of more than the the controlling magnet may be a vertical horseshoe carried inside the instrument from a support attached to the base , as seen in fig. 9 , which shows the system of magnetic control adopted in the improved electrometer .
Two adjustments are provided , one worked by a milled head at the base of the supporting pillar , , which projects about an inch through the base plate to enable this milled head to be easily turned ; the other by a tangent screw fixed under the base plate .
Turning the milled head at the base of the supporting pillar slides the soft-iron keeper , , over the poles of the horseshoe magnet , , and enables the strength of the controlling field to be varied ON QUADRANT ELECTROMETE]RS .
from a screw which can be raised or lowered by turning the nut , a vertical slot in the screw and a horizontal pin projecting from the framework preven ting the screw itself turning round .
The platinum wire at the of the needle dips into some sulphuric acid contained in a leaden cup .
This cup is supported and ] connected to by a stout wire , the upper end of which slides through a damping screw at the back of the lower end of the right-hand brass rod .
The leaden cup can , therefore , be easily raised or lowered and also centred .
Similarly the lower portion of the guard tube is carried by a stout wire passing through a clamping screw at the back of the lower end of the left-hand brass rod .
This portion of the guardtube is hinged and is shown open in fig. 10 .
After being put in position the tube.is osed round platinum wire , which projects down from the needle but so as not , of course , to touch this wire .
The upper portion of the guard-tube is made in two halves which are carried by the arms ( fig. 7 ) attached to collars sliding with a bayonet-joint motion on the rods .
By raising up these collars and turning them round the two halves of the upper portion of the gnard..tube can be opened as seen in fig. 11 , and the mirror and the wire attached to it got at .
In fact , by opening and turning aside both the lower and the upper portions of the guard-tube , the mirror , needle , and its attached wire , can be left hanging freely from the silk fibre suspension , and it can be seen whether they are hanging properly , or whether they require adjustment .
The two upper portions of the guard-tube are made skew , as shown in fig. 11 , in order that when the mirror is parallel to the needle , and the latter symmetrical with the quadrants , the glass supports of the latter may clear the glass supports of To avoid confusion in the illustrations the fine wire spirals the opposite pairs of quadrants together , and with the terminal screws at the tops of the glass rods supporting the terminal rods , have been omitted , We have to thank Mr. MATHER for taking a leading part in designing and constructing this improved electrometer : Fig. 11 .
VIII .
OBTAINED WITH THE I A large number of experiments have been made with the improved quadrant electrometer , and marked differences al.e observed between its behaviour and the behaviour of the White instrument .
First , ] tering the distanco- between the quadrants of our instrument does not materially affect the sensibility .
For example , increasing the distance between adjacent quadrants from to 4 millims. hardly varies the sensibility at all ; whereas with the White electrometer , drawing out the quadrants increases the sensibility ( a result which , as we have pointed out , is quite opposed to the text-book formula for a quadrant electrometer ) , and this increase in sensibility is brought about by a com- plete change in the shape of the curve connecting the potential of the needle with the deflection produced by a given P.D. between the quadrants .
Secondly , whereas with the White instrument it is only with one position of the quadrants that the potential of the needle is even approximately proportioned to the deflection produced by a given P.D. between the quadrants , this result is obtained considerable accuracy with symmetrical position of the quadrants in our instrument .
The following table is a sample of the results obtained , no special adjustment of the quadrants having been made to obtain constancy in the sensibility factor .
Thi6 sensibility factor , as already explained , is the ratio of the deflection , in microradian* of the reflected ray of to the obtained by multiplying together the ON ELECTROMETERS .
deflecting P.D. in volts , and the P.D. between the needle and the outside case also in volts .
TABLE XI.\mdash ; July 13 , 1888 .
Improved Electrometer .
Further , it is possible to practically obtain far greater sensibility with this improved electrometer than with the specimen we possess of the quadrant electrometer , as made by Messrs. WHITE .
The greatest double deflection for a Leclanche cell that we have been able to obtain , with our specimen of the White pattern , is 386 scale divisions , the P.D. between the needle and the outside case being 2500 volts , the scale being at 5 feet distance from the mirror , and 1 foob length of scale containing 290 divisions .
It would , however , be impossible to work with this adjustment , as the instability was too great .
Practically , therefore , about 250 scale divisions , as a double deflection for a Leclanche cell , or , what is the thing , about 180 di visions per volt P.D. between the quadrants , is all that we can obtain with our specimen of the White electrometer .
With the improved , on the contrary , very great sensibility indeed can be obtained by weakening the controlling magnetic field , the only limit being caused by the increased importance of want of symmetry , the greater disturbing effects produced by capilIary attraction , by the motion of exterior } etic bodies , \amp ; c. , as the sensibility is made greater and greater .
When the controlling field was made very weak it not found advisable to make the P.D. between the needle and the outside case more than 1800 volts .
Under these circumstances it possible to obtain , double deflection , 100 scale divisions for a P.D. between the of only one hundredth part of the E.M.F. of Leclanche ' corresI ) OlJding with a factor of 580 .
The needle , it is true , was very unstable , the deflection was not definite enough for ordinary practical use .
Possibly , however , the instrument might be used very well in this condition for null methods when great sensibility is required .
However , with a somewhat less sensibility , a double deflection of 44 isions 1jeing obtained for one hundredth of the E.M.F. of a Leclanche cell , the improved meter worked more satisfactorily than did the White instrument , when adjusted have its maximum sensibility , so that there is no doubt that the improved is at least ten times as sensitive as our specimen of the White pattern , when the two instruments are adjusted to be in an equally trustworthy condition as regards definiteness of the zero , definiteness of a given deflection , \amp ; c. Tests were also made of the two instruments when used idiostatically ; and , as regards the White pattern , it was found , as stated ) Professor GRAY on p. 29$ , vol. 1 , ' Absolute Measurements in Electricity and Magnetism , ' that " " The quadrant electrometer may be used idiostatically .
the measurement of differences of potential of not less than 30 volts But , on the contrary , we found that three yolts would give a double deflection of about 50 scale divisions with the improved electrometer when used idiostatically .
Hence the improved instrument is also ten times tsitive as the White pattern when used Spring , 1891.\mdash ; In what precedes we have shown that the improved electrometer is free from the defect in the White pattern , which forms the subject of this paper .
We have attributed this defect to the electrical action of certain pieces ( fig. 3 ) , which connect the upper and lower portions of the guard-tube in the White pattern .
But as our instrument differs in many points from the White pattern , not possessing , for example , a bifilar suspension which , according to Dr. HOPKINSON , is a necessary feature in the electrometer in order to explain the major part of the anomalous action of the White instrument , we determined not to publish this paper until we had experimentally proved , beyond all doubt , that our explanation was correct .
This we were unable to do until the year ) , various reasons having prevented our taking up the electrometer investigation again from the point we left it some years ago .
First , as regards a bifilar suspension being necessary , in order that the deflection by a given P.D. between the quadrants should be very far from proportional to the P.D. between the needle and the outside case .
This question is settled by the experiments that we have made this year on a quadrant electrometer made by Messrs. WHITE , in the possession of the Royal College of Science , and the needle of which has been suspended with a single quartz fibre by Professor Boys .
The following were the results that we obtained , the quadrants bein about 2 millims. apart , and the instrument in excellent condition in so that the spot of light returned absolutely to zero , and a deflection could be read with confidence to millim. The slight irregularities in the curve on Sheet .
( Plate 11 ) , which graphically represents the results contained in the following ] from the potential of the needle having been read by means of a portable pivotted electrostatic the suspeosion of which was , of course , far more frictional than the quartz fibre of the quadrant electrometer .
This voltmeter was calibrated by comparison with the absolute ele ometer .
ON QUADRANT TERS .
TABLE XII.\mdash ; February 3rd , 1891 .
White Electrometer with Single Quartz Fibre Suspension .
From the preceding we see that even when the needle is suspended by a single fibre , an increase in the potential of the needle from 260 to 440 volts alters the sensibility factor by 50 per cent. It is , therefore , quite certain that the peculiar action in the White electrometer can occur without a bifilar Next we had made the collar , or saddle , shown in fg .
12 , which could be fastened on the lower part of the upper guard-tube of the improved electrometer , the pieces projecting down into the quadrants on each side of the needle .
As these ons , are at the same potential as the needle , there will be the same sort of electl.lcal action between them and the needle as there is between the pieces ( fig. 3 ) of the White instrument .
Hence , if our explanation of the defect in the White instrument be correct , the application of the collar with its projections ( fig. 12 ) , to the guardtube of the improved electrometer ought to destroy the straight-line-law which the numbers given in Table XI .
show that this instrument possesses .
In Table XIII .
are given the results obtained with the improved electrometer before attaching the saddle piece ( fig. 12 ) , while those given in Table XIV .
were obtained after merely attaching the saddle ( fig. 12 ) , no other change whatever being made in the instrument .
The adjustment of the quadrants , and the position of the controlling magnet were quite different when obtaining the results given in Table XIII .
from what they were when the numbers given in Table XI .
were obtained nearly three years ago with the same instrument .
The sensibility factor had , therefore , quite a different value March , 1891 , from what it had in July , 1888 , but , jusb as a large .
var at on the potential of the needle caused at that date practically no change in the bility factor with the improved electrometer , so now the sensibility factor is practically independent of the potential of the needle before attaching the saddle .
After , however , the saddle has been attached , we see that the oN RANT I HS .
.559 TABLE XIV.\mdash ; March 20 , 1891 .
Improved Electrometer , Saddle Attached .
As the electrical zero altered very much in the above set of measurements made with the saddle attached , experiments were subsequently tried to see whether this alteration of zero affected the sensibility .
In each of the above cases , the controlling magnet was turned so as to alter the direction of the controlling field without altering its strength , and , as it was found that the change of zero thus produced did not materially alter the sensibility for any one of the potentials of the needle given in the above table , it inferred that the mere alterations of the zero mentioned in the table did not materially affect the sensibilities .
[ Lastly , an experiment was made to try how rate the improved quadrant electrometer might be , if care were taken to adjust the quadrants even for symmetry , and if the needle were suspended with a very coarse silk fibre , which had neither been boiled nor subsequently cleaned with absolute alcohol , in accordance with the plan that we usually adopt with silk fibres for suspension .
The keeper , the horseshoe magnet ( fig. 9 ) was placed so that the magnetic control was a maximum , and the instrument was used idiostatically .
In order that , although the electrometer was used idiostatically , the deflection might be reversed and zero errors , the needle and the outside case were attached respectively to two of the terminals of a reversing key , and the pair of quadrants with the other two terminals respecti vely .
or 60 SSRS .
W. ARYTO , J. PERRY , AND .
E. The results obtained are given in Table XV .
, and , considering that the instrum was set up in the rough way above described , the constancy of the sensibiIity is sufficiently striking .
TABLE XV.\mdash ; July 1891 .
Improved Quadrant Electrometer used Idiostatically .
N.B.\mdash ; The fibre suspending the needle was a very coarse one , and the instrument was tested without any previous adjustment of the quadrants even for symmetry .
One hundred and seventy old silver chloride ceIis were connected with a Thomson and Varley potential divider .
The potential of one ndred of these cells was tested by an electrostatic voltmeter , and found to be 99 volts .
One scale division corresponded with an angular movement of the needle equal to radian .
1891 .
] We cannot conclude this account of the rous defects of the Thomson quadrant electrom , as hitherto made by Messrs. WHITE , and of the way to overcome them , without expressing the deep debt of gratitude that we , in common with electricians of all countries , owe to the inventor of the electrometer , for having provided us twenty years ago with an instrument that has been of inestimable value , both for scientific research and for the testing of submarine cables .
ON ADRANT ELECTROMETERS .
\mdash ; SKETCH 0F THE MATHEMATICAL THEORY 0F THE WHITE LECTRObIETER .
Before we were certain of the cause of the peculiarities in the action of the White electrometer , we devoted a large amount of time to a mathematical tion of its behaviour .
This gave us many weeks ' hard work , as the investigation is a complicated one .
For , in order to obtain an equation ( for the deflection of the needle in terrtls of its potential and the P.D. of the quadrants ) that would fit all the various curves obtained experimentally , we found it necessary to take into account : \mdash ; 1st .
The repulsion between the pieces ( fig. ; 3 ) , of the guard-tube and the needle ; 2nd .
The diminution of this as the quadrants were separated sym- metrically ; .
The alteration that the tilting of the needle at high potentials produces on this repulsion ; 4th .
The alteration that the tilting of the needle produces on the rate of variation , with motion of the needle , of the coefficient of induction between the insulated pair of quadrants and the needle .
As , however , subsequently to this mathematical work , we completed the experimental proof already described of the cause of the curious action in the White electrometer , we think that it will be unnecessary to give more than a brief sketch of the mathematical investigation .
Let be the P.D. between the outside of the Leyden jar of the White electrometer and the system of conductors , consisting of the needle , guard-tube , sulphuric acid , \amp ; c. Let be the capacity of this system of conductors .
Let the pair of quadrants that is connected with the outside of the Leyden jar be regarded as being at potential nought , and let V be the potential of the other pair .
Let be the capacity of the insulated pair of quadrants , and the coefficient of induction between this pair and the needle .
Then the energy stored up in the system is therefore , if be the moment exerted by the bifilar suspension , when the needle is deflected through an angle from the equilibrium position , Now , it is usually assumed that , in consequence of symmetry , , a constant ( 1 ) , and MESSRS .
W. li AYRTON , J. PERRY , AND W. E. SUMPNER which leads to the ordinary equation , The repulsion ween the projections 1 , , of the guard-tube would necessarily render assumption ( 1 ) unjustifiable , since the rate of variation of capacity of the insulated pair of quadrants , and the rate of variation of the efficient of induction between this pair and the needle , would not necessarily affected by the action of the guard-tube .
But in all probability this action between the projections , and the needle would affect the distribution of electricity on and on the needle near them as the needle moved , so that the capacity of the system of conductors at potential would vary somewhat as the needle moved .
In other words , would not equal nought , and the equation for the White electrometer , taking into account this action of the guard-tube , would be \mdash ; ( 3 ) , instead of the text-book equation ( 2 ) .
When the fibres are near together so that the control they exercise is smffi , experiments show that does not increase as fast as it ought to do when is increased , and that for moderately high values of actually diminishes instead of increasing .
From this it follows that in equation ( 3 ) is negative , a result that might be expected , seeing that , as the needle is deflected , so that portions of it approach the projections of the guard-tube , it is probable that the action between projections and the needle at the same potential tends to diminish the capacity the needle and guard-tube .
Further , it is clear that diminishes as the quadrants are pulled out symmetricaly so as to increase the distance between them ; consequently , if ( 2 ) were the true formula for the White electrometer , the sensibility ought to diminish as the quadrants are drawn out .
But , as seen from our curves on Sheets III .
and IV .
( Plate 10 ) , and as is probably generally known , the sensibility actually increases as the quadrants are withdrawn symmetrically .
Hence for a given value of must diminish numerically as the quadrants are drawn out , and this diminution in the negative value of must more than compensate for the diminution in .
This is consistent with what we know , since represents the couple exerted between the charge on the needle and the charges on the projections , of the guard-tube , and it might be expected that the repulsion between two conductors at the same fixed potential would be increased by enclosing them in a conductor like the quadrants at a low potential , and diminished as this cover is more and more withdrawn .
Although the electrical zero is made to correspond with the mechanical zero by an adjustment of the fourth quadrant , the position is not necessarily a position in which ON QUADRANT ELECTROMETERS .
the attraction of the quadrants on the needle is zero , but is a position in which the action of the guard-tube on the needle is balanced by the action of the quadrants when all are at the same potential .
Let be the angle between the zero position of the needle , and the position in which the force exerbed by the guard-tube on the needle is nought , then is the angle through which the needle has been moved against the action of the guard-tube on the needle .
This force will vary in sign with the sign of , therefore , be expanded in powers of , it must be of the form so that if be small we may say that approximately , that is , ( 4 ) .
If then represents the double deflection of the needle obtained by reversing ( the P.D. between the quadrants ) , and represents the position of the electrical zero , that is , the position of the needle when all four quadrants are at the same potential , ( 5 ) , ( 6 ) , and equations of this form would probably give the value of the double deflecbion and the electrical zero of the White .
electrometer , if it were possible to hang the needle so symmetrically that there was no tilting even when it was raised to a high potential .
The tilting of the needle will increase the value of , and this probably accounts for the rise which occurs at the ends of many of the sensibility curves obtained with the White electrometer and shown on Sheets II .
, III .
, , and .
Indeed , if there were no action due to the gnard-tube it is probable that for all symmetrical distances between the quadrants and for ] ] distances between the fibres the sensibility would increase more rapidly than the potential of the needle , the ding in the sensibility curve being the greater the more easily the needle could be tilted .
With improved quadrant electrometer there is but little tilting of the needle but in highest part of the curve , obtained with this instrument when the saddle was Sheet VIII .
( Plate 12 ) , there is an indication of the sensibility curve bendin towards the vertical axis along which deflection is measured .
With the White electrometer we can , in spite of the opposing action of the guard-tube projections ( fig. S obtain a sensibility curve concave to the axis along which deflection is reckoned , is , we can make the sensibility in\ldquo ; rease more rapidly than the potential of the in one or other of two ways .
If a bifilar pension be employed , which tends to resist tilting of the needle , then the quadrants must be drawn out so far that , even for high potentiaJs of the needle , the repulsive action of the guard-tube shall be relatively not large .
Examples of this are seen in the curves marked \ldquo ; Quadrants about 4 millilns- .
apart\ldquo ; on Sheets IIL and IV .
( Plate 10 ) .
If a unifilar suspension be employed , which enables tilting to take place much more easily , then a sensibility curve concave to the axis along which deflection is reckoned can be obtained , without drawingthe quadrants out far , if the control be so weak that a deflection to the end of the scale is obtained for a potential of the needle too small for the control due to the guard-tube to be important .
An example of this is seen in the curve on Sheet VII .
( Plate 11 ) , obtained with a White electrometer having its needle suspended by a fine quartz fibre .
And generally it would appear that for a White electrometer to have a sensibility directly proportional to the potential of the needle , it is necessary to separate the quadrants by such an amount that the increase in the repulsion between the guardtube and the needle exactly balances the effect due to increased tilting as the potential of the needle is raised , or , in other words , ib is necessary that the quadrants be at such a distance apart that the increase of ( equation 8 ) exactly balances the of increase of in the expression as increases .
We had , therefore , to express and in equations ( 5 ) and ( 6 ) as functions of , and after much working we succeeded in obtaining expressions for and , each involving the first , second , third , fourth power of , and three constants , which fairly well with the various sensibility curves obtained with the White electrometer and with the curves on heet V. ( Plate 12 ) for the " " Motion of the Electrical Zero\ldquo ; as the potential of the needle and the position of the adjustable quadrant were changed .

