]\gt ; On the Electrical Resistance of Moving Matter .
By Professor F. T. TROUTON , , and A. O. RANKINE , B.Sc. , University College , London .
( Received February 20 , \mdash ; Read March 5 , 1908 .
) The question of relative motion between the earth and the neighbouring ether has been under discussion for many years .
It has , from time to time , been the subject of important ations ; but these have all resulted negatively .
The experiment about to be described is not different from the1n in this respect , , as it does , no definite information on the main point .
It was ested and commenced by one of us some years ; but the serious ditticulties which invariably attend measurements of such delicacy have delayed its completion till the present time .
Indirectly , the aim was to measure ths direction and magnitude of etherdrift ; the actual method having been to attempt to demonstrate the existence of the Fitzgerald-Lorentz shrinkage which has been supposed to mask the effect in the direct experiments of Michelson and Morley , and of Trouton and Noble .
It may be as well to say at once that if such a shrinkage be real , it is in this experiment obscured by some other exactly compensating change or changes , no effect approaching that to be otherwise expected having been observed .
The principle of the measurement is a very simple one .
Imagine a uniform wire AB ( fig. 1 ) of length and crosssectional area moving through the ether in a direction parallel to its length with a velocity .
Let be its specific resistance and its total electrical resistance .
The relation ( 1 ) On the Electrical Resistance of is then true .
logarithmically , we obtain .
( 2 ) Now suppose that the wire AB is turned through a right angle , so that its is perpendicular to the velocity to the FitzgeraldLorentz hypothesis , the length of the wire will be thus creased a small amount , such that , where is the velocity of md all powers of than the second have been neglected , upposed very small compared with .
Writing for , we have his is not the only change in dimensions to be expected .
In the first position AB both dimensions of the cross section are perpendicular to the direction motion , while in the second position one remains perpendicular , but the there becomes parallel .
A decrease in this latter dimension of the relative will cause a diminution of in the same ratio .
Therefore .
in ( 2 ) , it follows that it be supposed that the specific resistance of the material forming the wire independent of the direction of motion , , and therefore Hence , on the above assumptions , it is to be expected that the resistance of a wire with its length perpendicular to the ether-drift will be greater than when varallel in the ratio .
On the other hand , should there be no change total resistance , two ernatives present themselves .
Either there is no rlteration of length of the kind supposed , or the specific resistance changes such a way as to compensate it .
In the latter case of change would be that the specific resistance of a material to a current parallel to the -drift reater than that at right angles to this direction in the ratio .
The present investigation was , however , based upon the assumption that the specific resistance was constant ; and the object in view was to detect a variation of the resistance of a wire with direction .
The method used was the ordinary Wheatstone method of comparing resistances , specially adapted , of course , to the particular requirements of this case .
With certain modifications , to be described later , the arrangement was as follows :Four coils of wire , each wound upon a flat rectangular frame , Prof. F. T. Trouton and Mr. A. O. Rankine .
[ Feb. 20 , formed the four arms of a Wheatstone bridge .
By suitably adjusting the position of any frame , the wire on it , with the exception of the small part used in turning the corners , could be made to take uP any desired direction .
The frames were arranged horizontally on a stand so that the wires forming opposite arms of the bridge were parallel , and those adjacent arms perpendicular to one another .
The arrangement is shown diagrammatically in .
The lines marked 1 , 2 , 3 , and 4 , must be taken as representing the ection of the major part of the wire on the corresponding coil .
If the resistances of 1 , 2 , 3 , and 4 are equal , there will be no current through the galvanometer .
Suppose that the coils 1 and 3 are parallel , and the coils 2 and 4 perpendicular to the ebher-drift ; also that balance is obtained , and the resistance of each coil is equal to R. If , now , the stand be rotated through Position 1 .
Position 2 .
FIG. 2 .
a right angle , so that 1 and 3 become perpendicular , and 2 and 4 parallel to the drift , it is to be expected , on the previous assumptions , that 1 and 3 will each increase in resistance by an amount , and that 2 and 4 will each diminish by an equal amount .
This would result in the balance being destroyed ; and a current of magnitude would flow through the galvanometer in the direction indicated by the arrow .
Here represents the E.M.F. of the battery , and the galvanometer resistance ; and the internal resistance of the battery has been ected .
Or , if for some reason it has been impossible to obtain perfect balance in the first instance , and a small current flows in the galvanometer all the time , the change to be expected upon rotation in the nitude of this current is measured also by .
In other words , and remembering that the expected value of 1908 .
] On the Electrical of Moving Matter .
is , a variation of current of value is to be looked for .
Since this expression contains so small a factor as , it is obvious that the measurement will be a difficult one ; and this was indeed found to be the case , many effects , usually negligible in resistance measurements , now .
comparatively large .
Their elimination\mdash ; or rather , partial elimination , for they were to the last not entirely absent\mdash ; was vely tedious , and times appeared almost it was at last effected in the malluer about to be described .
The first difficulty was due to the presence of thermoelectric currents , which , at first , varied so rapidly as to make determinations impossible .
It was very soon found that junctions , even between pieces of the saJne wire , which were originally in the resistance arms , and the key in the neter arm produced disturbances of such a serious character that they had to be omitted .
, it was at first intended to effect balance by the movements of a slider on a thick copper rod , the other end of the galyanometer arm being permanently attached to the junction between the opposite arms .
It was also hoped to be able to re-balance after rotation by a further movement this slider ; this method , however , had to be abandoned , owing to the impossibility of the slider without producing further heating .
Another objection to the use of this thick copper rod lay in the fact that it was the cause of a difference of temperature between the two ends of the galvanometer arm .
Practically the same current flowed in this rod and in the thin wires of which the coils were made , and to which the other end of the galvanometer wire was attached .
Owing , therefore , to the heating effect of the current itself , a permanent difference of temperature became established at the terminals referred to .
To effect the removal of these disturbances , the following means were adopted .
The four bridge arms were made of two unbroken pieces of uniform wire soldered together at the points at which the current was led in from the battery .
Here , of course , small variations in potential were ineffective , producing , in the case of perfect balance , no current through the galvanometer , and , even when a small current was flowing , causing changes of the second order only in it .
There were [ lnctions at all in the whose resistances were being compared .
The ometer was inserted by means of a slider ( as indicated in ) , which joined through the former the mid points of the unbroken wires previously referred to .
Contact was made by simple pressure between crossed wires .
It was , of course , impossible to avoid two junctions here ; but , by them very close together , and because they were now equally heated by the current , the thermoelectric effects were reduced practically to zero .
Prof F. T. Trouton and Mr. A. O. Rankine .
[ Feb. 20 , 1908 .
] On the of Moving Matter .
Another and more persistent disturbance arose from unequal heating changes of resistance in the coils .
It soon became evident that it would be impossible to use uncovered wire ; but even when the wire used copper was thickly covered with gutta-percha , the.effect of heating was too to make definite measurements .
The coils were arranged above the other on a stand , and were turned about a vertical axis inside an cnclosure made of wood and felt .
The temperature of the air varied from point to point of this enclosure ; and , upon rotation of the stand , of resistance occurred , .
to the coils occupying different positions in it .
Moreover , even when the coils were not rotated , the behaviour of the current in the galvanometer indicated a radual icrease of resistance in the upper coils relative to the lower ones .
This was , doubtless , due to the warming of the air in the enclosure b.y the currents which were ; and , the warm air , the upper resistances increased more rapidly than those lower down .
With the exception of this latter effect , the disturbances were removed by rotating the ellclosure itself with the stand , thus the temperature distribution of the air round , and by making the coils with manganin wire instead of copper , on account of the much smaller temperature coefficient of the former .
The relative increase of resistance of the upper coils was thus made much more gradual , but it has been found impossible to entirely eliminate the effect , and it has been necessary , even in the final form of the apparatus , to take time of the current for the various positions of the stand .
The variation of the current in the meter , due to this cause , is now , however , sufficiently slow to make it quite easy to uish from it the immediate genuine effect which is looked for .
A further spurious effect was that due to alterations in resistance which were brought about by stresses introduced in rotating the stand .
As it happened , the magnitude of this effect was just of the order of that expected ; and this at one time led us to suspect a positive result .
The apparatus at that time was not in its final form , and was not adapted for rotations other than a right angle ; so it was impossible to make an absolutely test .
The balancing ( shown in fig. 3 ) was not then rotated with the rest of the apparatus , and thus there arose a possibility of strain in the wires forming parts of the esistances which were being compared .
This difficulty was finally surmounted by rotating the whole of the apparatus bodily , with the exceptions of the galvanometer and battery .
This removed the strain to the wires leading to the two latter , i.e. , to places where small changes of resistance were unimportant .
Finally , it was necessary to remove an effect which can hardly be called a disturbance .
As has been already pointed out , the use of a key was Prof F. T. Trouton and Mr. A. O. Rankine .
[ Feb. 20 , dispensed with in the galvanometer arm .
The result of permanent cont ct was to produce an induction current in the alvanometer when the apparatus was turned round .
This would not have mattered if it had not been necessary to take time readings on of heating effects .
eadings were eventually taken every quarter minute , and the galvanometer was not sufficiently damped to make this possible when the throws were .
In the whole region of space occupied by the stand , the magnetic field of the earth was reduced practically to zero by suitably disposing 16 permanent magnets in the neighbourhood .
The temporary induction effect upon rotation died out then completely in about five or six seconds after that rotation .
The final ement of the apparatus is shown diagrammatically in fig. 3 .
The foul coils , 1 , 2 , 3 , and 4 , are on a stand as before indicated , and above them ( and also fixed to the stand ) the balancing bridge A. This latter merely consists of about 5 or 6 cm .
of bared wire drawn taut a wooden stand .
The wires are here parallel and about a centimetre apart , and the slider , through which wires lead to the galvanometer , is movable along their length by means of a screw D. The slider consists of an arrangement by which the two wires from the galvanometer are pressed down by , one on each wire of the bridge , and balance is obtained by using the screw D. The whole of the apparatus , with the exception of the galvanometer and battery , is encased in a cubical doublewalled enclosure , which is fixed to a horizontal turntable , the interspace between the two walls of the enclosure being filled with cork dust for purposes of thermal insulation .
The screw-head projects outside the enclosure , so that adjustments may be made without opening the latter , and the wires to the battery ralvanometer are led out through a small hole at the top on the vertical axis of rotation .
The battery used is a cell .
The galvanometer is of the Du Bois type\mdash ; a low-resistance suspended needle galv nometer , trebly shielded with soft iron .
These shields are very effective in removing magnetic disturbances such as those caused by the hbouring electric railway , and they are found to be very necessary in delicate work of this description .
The behaviour of the needle is examined by using a Nernst lamp and scale at about metres distant , and , in its most sensitive state , .scale deflection of about 4 cm .
can obtained with a current of ampere .
In the actual experimental work the field about the needle was found to be variable , and the sensitiveness increased with the scale .
This is shown in , where the deflections produced by an additional ampere are plotted against the scale reading .
The expected deflection upon rotation , 1908 .
] On the Electrical of Moving therefore , varied with the scale reading at that time , and use was made of this calibration in calculating the results .
With to the dimensions of the apparatus , the coils were made of gutta-percha-covered 24 manganin wire , each of them of 16 turns round a 6-inch square flat frame .
Almost exactly 1/ 7 part of the wire was not horizontal , i.e. , the parts used in turning corners , and in leading to the .
This fraction was , therefore , not expected to contribute to the calculated chance of resistance .
The resistance of each coil was ohms and that of the galvanometer 10 ohms .
The E.M.F. of the storage cell was slightly variable , but was taken as having an value of volts , and its internal resistance has been arded as negligible .
Scale reading in Sensitiveness of galvanometer .
To calcuJate the expected variation in current upon rotation , therefore , we substitute the above values in the formula This must , however , be reduced by 1/ 7 part , in order to allow for the noncontributing parts of the wire .
We have ampere ampere , approximately .
If the earth 's orbital motion only be taken into account , the value of is VOL. LXXX.\mdash ; A. 2 Prof F. T. Trouton and Mr. A. O. kine .
[ Feb. 20 , approximately , and in this case the expected change in current would be ampere .
Such a current would produce , in the neighbourhood of the scale reading ( fig. 4 ) , a deflection of mm. , but the particular deflection to be expected depends on the part of the scale at which the determination is made .
Moreover , if the sun 's proper motion be allowed for , the value of is dependent upon the time of year , and a special calculation is required in each case .
This has been done for the results given later in Tables II and IV .
The values of , the time of horizontal drift , and its azimuth when horizontal , were obtained from the values given in the paper describing the ether-drift experiments of Trouton and Noble .
* The method of taking observations was as follows : time was chosen when the calculated direction of the total drift was horizontal .
By means of the slider ( fig. 3 ) , an attempt was made to reduce the current in the galvanometer to zero .
This , however , very difficult , and not essential .
Usually , the spot of light , whose position indicated the magnitude of the galvanometer current , was merely brought somewhere within the limits of the scale , and possibly there would be already a current of about ampere .
The spot of light would be found to slowly creep in the direction indicating a relative increase of resistance in the upper coils .
Its velocity would , however , become much smaller after the current had been flowing for some time .
( It may here be pointed out that , as a rule , the battery was , on this account , connected up with the bridge some hours before taking an observation .
) The was then rotated until one pair of coils became parallel to the drift , and a reading was taken at a particular instant .
The turntable was then turned at once through a , and a further reading taken after 1 seconds .
Immediately the turntable was restored to its original position , another reading following after 10- seconds , and so on for about 20 reversals .
Thus a set of 20 readings , at half-minute intervals , was obtained for each of the two positions of the stand .
Unfortunately , owing to mechanical of the galvanometer disturbance which is never absent in London except in the early hours of the morning ) , it was impossible in daytime to take readings nearer than 1 mm. , although the optical definition was otherwise good to admit of estimation to 1/ 10 mm. The following set of observations is typical:\mdash ; ' Phil. Trans , vol. 202 , pp. 165\mdash ; 181 .
On the Electrical of Moving Table I. Date , December 16 , 1907 .
Time , 4 . .
Azimuth of horizontal drift , measured eastwards from meridian , Readings on scale ( measuring the values alvanometer current ) It will at once be apparent that there is no general tendency .
for the numbers in one column to be in excess of those in the other by the expected amount , viz. , mm. It is somewhat difficult , however , to determine the best method of interpreting them , for the purpose of discovering the limits of measurement of the apparatus .
The form of the general time variation in current ( so called to distinguish its immediate effects attributable to rotation ) is unknown and not necessarily linear ; hence , to take the difference of the means of the numbers in the two columns is only approximate .
This latter was the method at first adopted , but although the results were satisfactory enough where the general current variation was practically linear , in cases where this condition did not exist discordant values of the difference vere o , according to the number of observations utilised .
Eventually the method about to be described was adopted as the most consistent results .
The following is an ideal set of those for one position of the coils , and those for the other position:\mdash ; Prof. F. T. Trouton and Mr. A. O. Rankine .
[ Feb. 20 , lleadings .
Mean of differences in first column .
Mean of differences in last column .
The expression in the first column represents the difference between a particular value of and the mean of the values of , just before and just after , while that in the third column represents the difference between the mean of two successive values of and the intermediate value of .
The means of the differences on the two sides respectively are given below , and they are to be expected to be equal to one another .
This turns out to be very nearly true , and the final mean of these two numbers has been recorded in Tables II , etc. , as measuring the observed excess of the " " \ldquo ; column over the\ldquo ; \ldquo ; column .
For instance , take the numbers given in Table I. Here Mean of differences in first column mm. Mean of differences in last column mm. The final mean is therefore mm. , and this measured the .
in current caused by the rotation of the coils in this particular case .
Now , by reference to the magnitude of the ether-drift on this particular day , and to the curve of sensitiveness of the galvanometer ( fig. 4 ) , it will be seen that the expected difference of scale reading is mm. This certainly does not exist ; and , in view of the fact that readings were made correct to 1 mm. only , there is reason for supposing that the observed difference is due to error in observation .
1908 .
] On the Electrical Resistance of Moving Matter .
The following tables are records of the other observations taken:\mdash ; Table II .
These observations were made at the best times according to the calculations of Trouton and Noble , i.e. , when the resultant drift is horizontal .
Date .
Time .
Magnitudeof drift.zimuthof 1 Observed .
Table III .
In the following cases rotation was through , so that no effect is to be expected .
Table The following three observations are tests for the earth 's orbital motion alone , no attention being paid to the effect of the sun 's proper motion .
Prof. F. T. Trouton and Mr. A. O. Rankine .
[ Feb. 20 , It will be noticed that the observed difference is sometimes of the same sig as that calculated , and at other times of opposite sign .
We are , therefore , inclined to attribute it , as before gested , merely to error of observation ; however , even supposing it to be a real effect , its maximum value is less than 2 per cent. of that looked for .
It may be objected that the above method of is not the correct one\mdash ; that , to be quite conclusive , no assumption as to the direction of the ether-drift should be made .
With a view to settling this and , incidentally , making use of the increased accuracy of reading possible at night time , the following sets of observations were made throughout the early morning hours of January 18 last .
The freedom from vibration made estimation to 1/ 10 mm. as easy as 1 mm. readings in the daytime .
The observations were spread over the whole time from 12 midnight to 4 .
and were , in a sense , a search for ether-drift .
The results are calculated in the way previously indicated , and tabulated in five sections , each section containing the results of exactly similar treatment as ards rotation .
Thus , Section I contains the three cases in which the first azimuth of 1 and 3 was and the second azimuth Table In interpreting the results above recorded , careful attention should be paid to he treatment of the coils in any particular case .
Since the readings are now made to 1/ 10 mm. , we think that a difference which affects the first place of decimals measures a real effect produced by rotation .
Thus , in the first section , the differences are of this magnitude and of same sign .
That they are not , however , due to an effect of ether-drift is proved by the 1908 .
] On the Resistance of Moving Matter .
observations taken at intermediate times and recorded in Section 2 .
Here the coils were rotated in the opposite direction through also , and no real effect was produced .
It is obvious that , for a genuine ether-drift effect , the direction of rotation through is indifferent ; and the fact that the observed differences of depend on the direction of rotation removes the possibility of them , small as they are , to ether-drift .
It should be noticed , too , that the observation recorded in Section 5 shows a difference of the same order , and that here also it must be due to a cause other than motion the ether , because rotation is through .
In the other sections , with the exception of the second observation in Section 4 , the differences are not large enough to justify any meaning being attached to them .
On whole , therefore , this set of readings points to the conclusion that at no time during the on they were taken there a of resistance comparable with that looked for .
We have , however , been unable up to the present to account for the small spurious effects observed .
Several suggestions have presented themselves , but none appears to be valid .
It was crht that possibly the twist on the galvanometer caused by the rotation produce a sufficient change of resistance there to effect the small alteration in current .
Calculations show , however , that a change of resistance of about 100 per cent. due to twisting copper wire through would be necessary for this to be the case ; so that the observed effects cannot be attributed to this cause .
A second idea was that the relative change of resistance of the coils was brought about by the alteration of their distribution with respect to the magnetic field in which they stood .
That magnetic field , as has been already pointed out , was very small , .
been taken to reduce it , as nearly as be , to zero .
This point was tested for by making the field purposely , in the hope of the effect ; but to no purpose .
Finally , a small direct action of the rotating coils on the galvanometer was looked for when a larger current than usual was passed them .
Here , , there was no observable This question must therefore be .
left undecided .
It does not really affec the main aim of these experiments .
With regard to this we consider ourselves justified in making the following assertions:\mdash ; 1 .
The total electrical resistance of a wire is not altered by an exceeding of the whole amount by any change of its position relative to its motion through space .
2 .
On the assumption the -Lorentz is a real efiect , the specific resistance of a material is dependent upon direction of flow of the current , being greater to a current parallel to the velocity Prof F. T. Trouton and Mr. A. O. Rankine .
[ Feb. 20 , of the material through space than to a current in a perpendicular direction .
The magnitude of this change of specific resistance is shown by the experiments to be certainly within 2 per cent. of being sufficient to compensate the change of length .
Note.\mdash ; In view of the very general acceptance of the Fitzgerald-Lorentz shrinkage theory , the negative results of these experiments will probably be attributed to a dependence of specific resistance on direction of current flow .
In connection it is worthy of note that certain independent considerations point to the same conclusion .
The electronic theory of metallic conduction to the result*that the specific conductivity of a material is measured by the expression where is the number of electrons per unit volume , the mass of an electron and the charge upon it , the mean velocity , the mean free path ( i.e. , the mean distance traversed by an electron between successive collisions with atoms ) , and a numerical constant .
It is not here of importance whether this expression is absolutely correct or not , provided that it represents the facts dimensionally .
The specific resistance is the reciprocal of the above quantity , and we therefore have Of these quantities and are independent of the motion through space .
The number of electrons per unit volume may also be supposed unaltered by changes of azimuth of the conductor , because the latter has the same volume in all azimuths .
The changes of , therefore , depend on the variations of the quantities , and .
Let us denote by the suffix the values measured parallel to the drift , and by the corresponding values in a direction at right angles .
Hence , .
( 1 ) On the shrinkage theory , we expect and , following Lorentz * See J. J. homson , ' The Corpuscular Theory of Matter , ' p. 53 .
'Amsterdam Acad. Proc 1903\mdash ; 04 , p.809 .
This value gives complete compensation , while value , , does not .
We have consequently taken it in our suggestion of the direction in which to look for the mechanism of compensation .
1908 .
] On the Electrical Resistance of Moving Matter .
The only remaining ratio to be determined is Now it is to be expected that the average kinetic energy of the electrons should be independent of the direction of motion ; or , in other words , the total kinetic energy associated with any particular direction should be the same .
On this assumption we obtain , ; hence , or Returning to equation ( 1 ) , it follows that since is a very small quantity .
That is to say , the specific resistance parallel to the ether-drift is greater than that at right angles in the ratio This corresponds exactly to the conclusions respecting specific resistance arrived at in the experiments above described .

