]\gt ; INDEX SLIP .
RUTHERFORD , E.\mdash ; A Magnetic Detector of Waves , and some ot lts Applications .
Phil. Trans. , lS97 , vol. 169 , pp. 1-24 , Electrical Discharges , Oscillating\mdash ; Absorption of Energy by Conductors ; Investigation by Magnetized Needle ; llagnetization by .
Rutherford , E. Phil. Tlans .
, 1897 , vol. 189 , pp. 1-24 .
Electromagnetic Waves in Air , Detection of , by Nagnetized Needle .
Rutherford , E. Phil. rans .
, 1897 , vol. 180 .
pp. 124 .
bIagnetization by Osoillating Electrical Discharges .
Rutherford , .
Phil. , 1897 , vol. 189 , pp. 124 .
Specific lctive Capacity\mdash ; Plate-glass , Ebonite .
Rutherford , E. Phil. TIans .
, 1897 , vol. 189 , pp. 124 .
PHILOSOPHICAL I. A Magnetic Detector of Waves some of its E. RFORD , , 1851 Exhibition Science Scholar , New Zealamd sity , Trinit .
College , Cambridge .
by Professor J. J. THOMSON , F.R. Received June ll , \mdash ; Read June 18 , 1896 .
INTRODUCTION .
THE present paper deals with the subject of the magnetization of iron by highfrequency discharges , and the uses of netized steel needles for detecting and measuring currents of very great rapidity of alternation .
It will be shown that these magnetic detectors offer a very simple means of investigating many of the phenomena connected with high-frequency discharges , and may be used over a wide of periods of alternation .
Not only may detectors be used in ordinary Leyden jar circuits , but they also offer a sensitive means of investigating waves along wires and free vibrating circuits of short wavelengths .
were also found to be a sensitive means of detecting electrical radiation from Hertzian vibrators at long distances from the vibrator .
In the course of the paper the following subjects are investigated:\mdash ; I. Magnetization of iron by high-frequency discharges and the investigation of the effect on short steel needles .
II .
Magnetic detectors and their uses . .
Detection of electro-magnetic radiation in free space .
Waves were detected over half-a-mile from the vibrator . .
Waves along wires . .
Damping of oscillations .
Resistance of iron wires .
tion of energy by conductors . .
Determination of the period of Leyden jar discharges and the con stants of the discharge circuit .
The magnetization of steel needles when placed in a spiral through which a Leyden jar discharge was passed has long been known .
NDCCCXCVII.--A , 22.2.97 In 1842 Professor HENRY was led to ct from the anomalous agnetization of steel needles that the denjar discharge was cillatory .
Professor HENRY , ABRIA , and several others , used steel attempts to determine the direction of induced currents in secondary and tertiary circuits , when the Leyden jar was discharged through the primary , but conflicting results .
Lord YLEIGH ( Phil. Mitg vol. 39 , 1870 , p. 429 ) made use of steel needJes in a netizing spiral in investigating the ) aximum current of a break for ordinary induction circuits .
general subject of the magn etization of iron , for rapid atory currents , has been worked at by many different experimenters ; Lord RAYLEIGH , using oscillatory currents of a frequency up to 1050 a second , showed that iron wires showed considerable increase of resistance , and deduced the value of the permeability of the wire .
TROWBRIDGE ( ' Phil. 1891 ) has shown that iron wires rapidly damp down the oscillations of the Leydenjar discharge , and from his results deduced a rough value for the permeability of the specimens tested .
V. BJERKNES ( ' Electrician , ' November 18 , 1892 ) found that the damping out of oscillations in a Hertzian resonator takes place much more rapidly in a resonat , of iron than when it is made of material .
Magnetization of Iron by Leyden discharge .
If a piece of steel wire , several centimetres in length , be taken and placed in a solenoid of a few turns , on the passage of a discharge the wire will be found to be magnetized .
The magnetization is , in general , small , and increases slightly in amount when a succession of discharges are passed in the same direction .
Fig 1 .
shows the arrangement .
A and are the poles of a Wimshurst machine .or an induction coil , the condenser , the air-break , and the solenoid in which the steel wire is placed .
The charging-up of the pndenser bt , fore the spark passes woe found to have no effect in magnetizmg the needle .
In all experiments to follow , the magnetization of the needles was tested by means ELECTRICAL WAVES AND SOME OF ITS APPLICATIONS .
.3 of a small mirror .
The needle was either fixed in position in the solenoid and the magnetometer placed beside it , or the needle removed and tested after each experiment .
If wires of the same length but of different diameters be takeo , it will be found that the magnetization is roughly proportional to the diameter of the wires .
This is to be expected , since the magnetizing forces are confined to a thin skin on the surface of the needle , and so the amount of magnetization depends more on the surface than on the sectional area .
In order to determine accurately the way in which a piece of steel was magnetize , from the surface inwards , recourse was had to a method of solution of the iron in acid .
The needle to be tested was fixed in a glass vessel before a dead-beat magnetometer .
Dilute hot nitric acid was poured in and kept at a constant temperature .
As soon as the needle was covered it commenced to dissolve , and the variation of the deflectiou with the time was noted .
In this way the amou1lt and stages of the magnetization of the iron could be completely determined .
From preliminary experiments on uniformly gnetized needles , it was found that under the action of the acid the diameter of the wire decreased uniformly with the time .
Let I represent the intensity of magnetization of a thirl circular shell distant from the centre of the needle , and the deflect , ion of the magnetometer at any instant .
I. is pop.tional to therefore Ir is proportional to Let be the radius of the wire at first .
It has been shown that is proportionaI to , the time of action of the acid .
Therefore where is a constant , and therefore I varies If a curve be plotted whose ordinates represent the deflection and the abscissae the time of action of the acid , at any point is equal to the tangent of the angle which the tangent to the curve at that point makes with the axis of .
The variation of I can thus be completely determined from the experimental curve .
The following urve I is an example of the magnetization of a piece of pianoforte wire , 4 centims. long , 08 centim. in } eter , placed in a solenoid of two turns per centim. The frequency of the discharge was about 3 million per second .
4 MR E. HERFORD ON A MAGNETIC DETECBOR or The ordinates of the curve represent the deflection of the magnetometer and the abscissae the depth to which the iron has been dissolved by the acid , measuring from the surface inwards .
Each division of the ordinates corresponds to a depth centim. Ihe deflection of the magnetometer at first was 85 divisions .
As the needle commenced to dissolve , the deflection increased rapidly to 156 , remained nearly steady for a short time , and then rapidIy diminished to zero ; when this was the case the diameter of the needle was 032 centim. , so that the magnetization had penetrated to a distance centim. If the variation of the value of I , the intensity of magnetization from the surface inwards , be deduced from this curve , it will be seen that the surface layer is magnetized in one direction and an interior layer in the opposite direction .
This apparently gives evidence of two half oscillations , in opposite directions , in the discharge .
A large number of needles were dissolved after magnetization under various conditions , and the same peculiarity was observed , although , from other evidence , it was known that there were a large number of vigorous oscillations before the discharge was much damped down .
When a needle magnetized to saturation was subjected to the discharge , the of tho needJe was always diminished , and on solution of the iron the same effect was observed , , a surface layer magnetized in the opposite direction to the internal magnetism .
Since Leyden jar discharge iu general gives several complete oscillations before it is greatly damped down , it would be expected that the surfuce layer of a uniformly magnetized needle would either be completely demagnetized or show evidence of several oscillations in opposite directions .
The effect observed may be explained when the demagnetizing force of the euds of a short needle on itself is taken into account .
The first half oscillation that teuds to demagnetize the ELECTRICAL WAVES AND SOME OF ITS APPLICATIONS .
needle has the demagnetizing force of needle assisting it , while the return oscillation has it in opposition .
The retul.n oscillation will therefore not be able to completely remagnetize the surface layer already affected , but a thin layer will be left in the interior .
After the passage of the next oscillation another layer will be added in the same direction , and so on , till the final effect will be that the surface of the needle will be magnetized in the opposite direction to the interior .
If strongly-magnetized needles of the same diameter , but of different lengths , are taken and placed in the same solenoid , it will be found that the reduc.tion of magnetic moment of the needle , due to the discharge , is greater the shorler the needle .
This effect is due to the demagnetizing influence of the ends , which is greater the shorter the needle .
It was also found that if sive d be passed , the reduction of deflection gradually increases , till it reaches a steady state , so that the passage of any further number of discharges has no apparent effect on the magnetism of the needle .
The following table shows the effect of varying the length of the needle , the diameter being kept constant , and also the effect of successive discharges in each case .
Needle08 centim. in diameter ; frequency about 3 millions .
In the above table of the needles was placed at such a distance from the magnetometer to give the same steady deflection of 250 .
The vertical colulnns show how the deflection fell after the passage of the different numbers of discharges .
The vertical columns correspond 10 needles , 32 , , 75 centim. respectively .
For the needle centims. long , the deflection fell from 250 to 188 , while for the short needle , 75 centim. long , the deflection fell from 250 to 50 , although all the other conditions were precisely the same for each .
It will be observed from the above that the first discharge is mainly instrumenta ] in reducing the deflection , and that after ten charges have been passed , the deflection has nearly its final value .
Whenever a gnetized needle is placed in a enoid and a discharge passed , there is always a reduction of the , the amount depending , for any given size of needle , on ihe intensity of the magnetic force in the solenoid and on the period of the oscillation .
This is the case whether we are dealing with uyden jar circuits , or the free vibrations , suoh as are set up in Hel'tzian reoeivers .
An uninagnetized needle , on the other hand , is not appreciably magnetiaeedd when placed in a circuit where the damping is small , for each successive oscillation destroyi the effect of the previous one .
Soft iron wires exhibit a similar to steel , only it is difficult to use wires of sufficient length to retain their magnetization .
The effect on the needle is in general a purely surface one , and the amount of demagnetization does not bear any simple relation to the magnetizing force acting on it .
After every experiment , the needle was removed and placed in a solenoid and a steady current passed , sufficient to saturate the steel .
In this way the needle could always be quickly reduced to the same state afcer any experiment , and it was found that , hard steel , the results of successive experimente were very consistent with one another .
Passage of Discharge through a Magnetized Wiri ' .
If a piece of pianoforte wire , several centimetres in length , be taken and placed in series with the discharge circuit , in the sage of a discharge , the magnetic moment of the needle is diminished , due to the " " circular\ldquo ; magnetization of the wire .
If the needy be dissolved in acid , it will be found that there is a thin skin , apparently magnetized in opposition to the original magnetization , due ) the resultant action of the demagnetizing force of the needle and the magnetic force due to the current in the wire .
The magn etic force acting at any given point in the wire is given by where is the current through the conductor flowing internal to the circle described through the point , and concentric with the surface of the conductor .
The value at the surface of the wire is given by where is the radius of the wire .
Assuming permeability of the iron wire , as constant , the maximum value of the current at any point of the conductor decreases in geometrical progression as the distance from the surface inwards increases in arithmetical progression .
As will be shown later in the part on " " Resistance of Iron Wires the current falls off even more rapidIy than the theoretical law , on account of the increase of the value of : as the amplitude of the current diminishes in intensity from the surface inward\amp ; For thin wires the magnetic force at the surface of the wire is much greater than for thicker ones .
We should , therefore , expect a thick magnetized wire conveying the to be affected to less depth than a thin one , and this is found to be the ELECrRICAL WAVES AND SOME OF ITS APPLICATIONS .
A thin stoel wire , 025 centim. in diameler , was completely demagnetized by a discharge .
Iu this case the maximum value of the current through wire about 100 amperes , and the value of the magnetic force at the surface of the wire was , therefore , about 1600 C.G.S. units .
A hard steel wire , 08 centim. in nneter , was only partially demagnetized , the deflection being reduced from 250 to 116 scale divisions .
The following are examples of a few of the experiments on the letization of iron wires when the frequency of the discharge was about million and the value of the maximum current about 100 amperes:\mdash ; 1 Thin steelThin swire diameter , 025 centim. : completely demagnetized .
2 .
teel wire : diameter , 08 centim. : fall of deflection from 250 to 116 .
3 .
Steel wire : , , 16 centim. : , , , , 250 to 184 .
4 .
Steel wire : , , 25 centi .
: , , , , 250 to 216 .
5 .
Long hollow soft iron cylinder , millim. thick and diameter : fall of deflection from 250 to 230 .
The same condenser discharging current were used for all the specimens tested , and it is of interest to observe the depth to which the netism of bhe iron was affected by the discharge , assuming that the final deflection is that due to the mass of iron not magnetized .
Experiments of this kind show to what a small depth the current penotrates into the iron wire .
Very large momentary currents are conveyed through a surface of the conductor , and the intensity of the current diminishes rapidly inwards .
A thin magnetized steel wire was placed in the circuit of a small Hertzian plate vibrator .
The deflection due the needle fell from 300 to 250 after a succession of discharges .
This shows that the iron was unaffected below a depth of about centim. For rough comparisons of the intensity of currents in multiple circuits , the use of the " " longitudinal\ldquo ; detector is often erable to placing the needle in a solenoid .
A thin magnetized steel wire , placed in series with the circuit , is a surprisingly sensitive detector of oscillatory currents of small intensity .
In practice copper wires soIdered on to the extremities of the steel nffile , which is plaoed in position before a magnetometer .
A solenoid is wound over the needle , and after every experiment a steady cnrrent was sQnt through in order to qaturate the needle .
Both the " " longitudinal\ldquo ; and " " soleuoidal\ldquo ; detectors may be readily used to compare the intensities of currents in multiple circuits when the period of oscillation is the same for each .
The best form of the solenoidal detector is explained Iater , and it has the advanta ge of being able to distinguish between the intensity of the first and second half oscillations .
Detection of in Free It has been shown the amount of demagnetization of a magnetized needle depends on the fineness of the wire and the number of turns per centim. on the nagnetizing solenoid .
If a short piece of thin magnetized steel wire be taken , and a large number of turns wound over it , it is a very sensitive means of detecting electrical oscillations in a conductol when the amplitude of the lationS is extremely small .
It was on this principle that a detector for electrical waves was devisd , which proved to be a sensitive means of detecting Hertzian waves at considerable distances from the vibrator .
About twenty pieces of fine steel wire 007 centim. in diameter , each about 1 centim. long , and insulated each other by shellac varnish , formed the detector needle .
A fine wire solenoid was wound directly over it , of two layels corresponding to about 80 turns per centim. As the solenoid was of very small diameter , about 15 ntim .
of wire served to wind the coil .
This small detector was fixed at the end of a glass tube , which was itself fixed on to a wooden base , the terminals of the detector coil bein brought out to mercury cups .
Fig. 2 .
( fig. 2 ) represents the detector needle and the solenoid wound over it .
A and are the mercury cups .
CA and BD were two straight rods which served as receivers , one end of each being placed in the mercnry cups .
The ector needle was strongly magnetized and placed before a small magnetometer , the deflection due to the needle being compensated by an auxiliary magnet .
If the receiving wires parallel to the electric force of the wsve from- the ELECTRICAL WAVES AND SOME OF ITS APPLICATIONS .
vibrator , oscillations were seb up in the receiver circuit , the surface layers of the needles were demagnetized , and there resulted a co1Tesponding deflection of the magnetometer needle .
The amount of the deflection , of course , depended on the amplitude of the oscillations set up in the receiver , and , therefore , on the distance from the vibrator .
Long ts .
When a Hertzian vibrator was used with plates 40 centims. square , and a short discharge circuit , quite a large defleotion was obtained at a distance of 40 yards , the waves passing through several thick walls between the vibrator and reoeiver .
Further experiments were made to see how far from the vibrator electromagnetic radiation could be detected .
For the long distance experiments , the vibrator consisted of two zinc plates , 6 feet by 3 feet , and separated by a short discharge circuit of about 30 centims. When Iarge plates were used , a Wimshurst machine was equally efficient as a Ruhmkorff coil for exciting the vibrations .
The first experiments were made over Iesus C , the receiver being )laced in one of the buildings on Park Parade .
Quite a effecb was obtained at a distance of a quarter of a mile from the vibrator , and from the deflection obtained it was probable that an effect would have been got for severaI times that distance .
When the vibrator was set up in the top floor of the Cavendish LaboratoIy , small , but quite marked effect was at Park Parade , a distance of over half a mile in the direct line .
In this case the waves , before they reached the jeiver , must have passed through sevelal brick and stone walIs , and many large blocks of buildings intervened between the vibrator and receiver .
The length of wave given out by the vibrator was probably six or seven metres , and a wave of that length seemed to suffer very little loss of intensity in passing through ordinary brick walls .
From au experiment tried in the Cavendish Laboratory it was found that the effect of six solid walls and other obstacles between the vibrator and receiver did not diminish the effect appreciably .
When the vibralor was working in the upper part of the Laboratory , a large effect could be obtained all over the building , notwithst , anding the floors and walls intervening .
A large number of experiments were made on bhe effect of varying the length and of the receiving wires .
If a fairly dead-beat vibrator were used , e.g. , plates with .
short inductance , it was found that the deflection gradually increased with increase oflength of the receiving wires , reaching a maximum which was unchanged by any further of length .
NDCCCXCVII . .
, : 10 MR. E. RUTHERFORD ON A MAGNETIC DETBCTOB or The effect on the detector was found to be practically independent of the sectir1d area of the receiving wires .
A thin wire and a thick rod of ffi same longth had equal effects ; a plate of metal , 6 centims. wide , produced the same defleotion as a thin wire .
If two wires instead of one were used in parallel the effect was the same as one , though the wires were some distance apart .
Any number of wires in parallel had the same effect as a single wire or plate .
No difference could be detected whether the first half oscillation in the reoeiver tended to magnetize the needle or the reverse .
Since the vibrator used was nearly dead-beat , this shows that the damping of the oscillations in the receiver is small .
On introducing a short carbon rod in the circuib the deflection was greatly reduced .
It was found impossible to magnetize soft iron or steel when placed in the receiving circuit on account of the slow decay of the amplitude of the oscillations .
detector needle be kept in posibion for a succession of observations , provided the currenb in the receiving circuit is steadily increasing for each experiment , otherwise the detector should be remagnetized and placed in position eaoh tion .
The deflection found to be very constant for a series of experiments under the same conditions .
The connection between the intensity of the electric force at the receiver and the deflection of the maguetometer needle can be easily determined by swinging the receiving wires through angles .
When the receiver is placed symmetrically with regard to the vibrator , the deflection was a maximum when the receiving wires were parallel to the axis of the vibrator , and the intensity of the electric force acting along the receiver varies as the cosine of the angle from the maximum position .
With plate vibrators the deflection was found to be nearly independent of the degree of brightness of the spark terminals and remained sensibly constant for long intervals .
In the case of the small cylindrical vibrator used by HERTZ with the parabolic reflectors , the deflection coutinually varied with the state of the sparking terminals , and such small vibrators cannot be relied on for metrical experimenffi Some experiments were made to see if the magnetic force in the wave front couId be directly detected .
A collection of thin wires , insulated from each other and magnetized to saturation , were used and placed in the direction of the magnetic displacemenb , but the values of the magnetio force were too small to be observed , except quite close to the vibrator .
along Wires .
It was found that the use of a detector , composed of fine ulatd wiroe , was quite delicate enough to investigate waves along wires .
when there was only one turn of ELECTRICAL WAVES AND SOME OF ITS APPLICATIONS .
wire round the detector } edle .
Since the reduction of magnetic moment was nearly proportional to the of { he current , the intensity of currents at various points could be very approximately compared .
The ordinary Hertz arrangement ; set up for obtaining free vibrating circuits .
A and were two plates set up ertically .
Beside them were two small plates and , and long wires led from these plates .
A fixed bridge was placed at EF , the ends of the wires , and a detector placed at the middle point of , with a small magnetometer fixed in position .
A sliding bridge , , was then moved till the fall of the deflection of the detector needle was a maximum .
This position of the bridge could be very accurately determined , for a movement of the bridge through 1 centim. altered the deflection considel'ably .
The detector was then placed in various parts of the circuit CE , and the amplitude of the current at the different points determined .
It was found that the current was a maximum at and the middle point of .
A well-defined node was found at the middle point of CE .
The length CEFD was thus half a wave-length .
Since the use of the bolometer has been the only means of accurafely investigating waves along wires , it was interesting to observe whether the magnetic detectors were of the same order of sensitiveness as the bolometer .
One turn of was wound round each of two glass tubes , sliding along the wires CE and , as in experiments .
Instead of the fine bolometer wire , detector needle , with several turns of wire around it , was placed in series with the two turns of wire .
The charging and discharging of the small condensers , formed by the straight wires and the small coils around them , was quite sufficient to almost completely demagnetize the needle .
By this method the movement of small quantities of electricity could ) detected and the sensitiveness was quite comparable with that of a delicate bolometer .
No appreciable damping could be detected for the long wire circuits , showing that they were probably vibrating almost independently of the primary vibrator .
If a metre or two of wire was fixed to the pole of a Wimshurst machine , on the of a spark .
there was evidence of a rapidly-oscillating current set up in the wire .
By using a sensitive detector , with a few turns round it , the variation of current , along the wire could readily be determined .
If short lengths of wire were any portion of a Leyden jar , on the passage of a discharge , there was always evidence of a rapid oscillation set up in the wire .
Each of the short circuits had a tendency to vibrate in its own tural period , but the results were icated by the oscillatio .
of the main circuit .
Damping of use of magnetized needles offers a simple and ready means of determining the damping of oscillations in a discharge circuit .
Let be self-inductance of discharge circuit for rapid currents .
, , capacity of condenser .
resistance of leads and airbreak to the discharge .
, , potential to which condenser is charged .
The current at aIly instant is given by The exponential only inclueles the case of frictional dissipation of energy , and does not take into account radiation into space .
In the experiments at present considered , where the condenser is of the type of a Leyden jar , there can be but very small amount of dissipation of energy due to radialion .
ning R to be constant , the amplitude of the current decays in geometrical progression .
Consider two similar small oppositely wound solenoids A and placed in series in the discharge circuit .
Two magnetized needles are placed in A and , the north poles facing in the same direction .
After the passage of a discharge , it will be found that the reduction of magnetic moment is not the same in the two needles .
Let be the half-oscillations of the discharge in one direction .
, , be the half-oscillations in the opposite direction .
Suppose that the half-oscillation tends to magnetize the needle in the solenoid A still further .
Since the needle is saturated no effect is produced .
demagnetizes the surface skin , tends to remove the effect of , and so on .
In the solenoid the needle , tends to remagnetize it in its original direction , and so on .
Since the maximum value of the current of is greater than the ximum value of , the needle in will be more demagnetized than in A. If , however , we increase the number of turns per centimetre on the solenoid A. until the effects on the two needles are exactly the , then that the value of the current deoreases in geometrical progression , the maximum value of ffi ELECTRICAL WAVES AND SOME OF ITS APPLICATIONS .
netic force due to the oscillation acting on the needle A is equal to the maximum value due to the oscillation on Let be the maximum values of the current in the first and second halfoscillations respectively .
Let be the number of turns per centimetre on solenoids A and respectively .
Then , since the ratio of the second to the iirst half-oscillation is therefore known , and the damping is thus determined .
actual resistance in the circuit may also be deduced .
Now where period of complete oscillation and Therefore , say .
Therefore ( 1 ) .
Since are known from the constants of the discharge circuit , and is determined by experiment , is known .
In practice , in order to avoid the necessity of rmining the constants of the discharge circuit , an additional known resistance is introduced into the circuit .
If an electrolytic resistance of zinc sulphate with zinc electrodes be used , the resistance will be found to be practically the same for steady as rapidly alternating currents , as the specific resistance is very great .
Let be the ratio of the amplitudes of the two half oscillations wheu .
is in the circuit ( 2 ) .
Dividing ( 2 ) by ( 1 ) is therefore determined in term of , a known resistance .
The method of two solenoids was not adopted in practice , but one theoretically equivalent employed .
A narrow piece of sheet zinc ABC was taken ( fig. 4 ) and bent into almost a complete circle of 7 centims. diameter .
This was fixed on a block of ebonite .
At the centre of the circle a thin glass tube OM was placed , which served as the axis of a metal arm , which pressed againsl the circumference of tk de moved round it .
The " " detector\ldquo ; consisted of about thirty ver7 ; 003 inch in diameter , arranged into a compound magnet about 1 long .
the wires were insulated from each other by shellac varnish , and the small needle vas fixed inside a thin glass tube which could be easily slipped in and out of the cenihl glass tube OM .
Fig. 4 .
A divided scale was placed round the circumference ABC , and the whole arrangement was fixed in position before.a small mirror magnetometer .
The magnetized " " detector\ldquo ; needle was placed in position by sliding it in the glass tube , and the deflection due to the needle was compensated by another magnet .
The wires of the discharge circuit were connected to and , and when the arm ML was at no effect was produced on the needle .
When the discharge passed round the circle there was a deflection due to the partial demagnetization of the detector .
The detector was then quickly removed and magnetized to saturation in an adjacent solenoid and then replaced .
It was found that , provided the detector was magnetized in a very strong field , on replacing it in position the zero remained unchanged , and the same deflection was obtained time after time for similar discharges .
Since the magnetic field at the centre of a circle due to an arc of length is given by , where is the current , we see that the magnetic force acting on the needle is proportionaI to the lengh of the arc traversed by the discharge .
A series of observations were made and it was found that the deflection due to the detector was approximately proportional to the magnetic force acting on the needle , provided the magnetic force was well below the value required to completely netizeo the steel .
Curve ( 2 ) represents the relation between the deflection of the magnetometer and the magnetic force acting on the needle .
The curve is nearly a straight line exoopt near the top part of the curve .
To determine the damping of the oscillations a discharge was passed in one direction and the deflection noted .
The detector was removed , magnetized and replm The direction of the discharge was reversed and the arm of the moved unUl ELECTRICAL WAVES AND SOME OF ITS APPLICATIONS .
the deflection was the same as before .
When this is the case the ratio of the maximum values of the first and second half oscillation is given by the ratios of the arcs traversed by the oharge .
In this way the rate of decay of oscillations in ordinary discharge vcuits was examined .
With short air breaks and copper wires for connection , it was found the damping was hardly appreciable .
As the of the spark gap was increased , the absorption of energy in the air-break caused the oscillations to damp pidly .
If the } ) wires of the discharge circuit were replaced by iron , there was in all cases a very rapid decay of the oscillations , whatever length of the airbreak .
If an iron cylinder were placed in a solenoid , the absorption of ener-gy by the cylinder caused a rapid damping , while a copper cylinder of the same diameter had no appreciable effect .
EXPERIMENTS on Damping of Oscillations .
Discharge circuit rectangular , 184 centims. 1oy 90 centims. ; Self inductance of circuit , ; Capacity , electrostatic units ; Frequemcy , millions per second .
In the third column the apparent resistance , to the absorptio ] ] of 6 MR. E. RUTHERFORD ON A MAGNIOTIC DRTEWOR OF energy in the conducting wires and spark gap is tabulated .
The the resistance of the wires of the discharge circuit ohm , that $he of the resistance is due to the great absorption of energy in the air brQak The above curve represents the relation between the length of the spark and the apparent resistance that the spark offers to the discharge .
The ohmic resistance of the air break is probably very variable , depending on the intensity of the charge at any instant , but the absorption of energy is quite definite and may be expressed in terms of the non-inductive resistance , which , when placed in the circuit , would absorb the same amount of energy .
It will be observed that the damping of the oscillations ases rapidly with the length of the spark , and that the resistance of the air break increases very rapidly with its length .
It was also found that the damping depended on the capacity when the inductance and spark length were kept constant .
The and ako the resistance of the spark were found to increase with increase of capacity .
For example with an air break centim. the damping and resistance are given below .
When the qapacity of the circuit was small the damping was found to be very small .
If iron wires were put into the place of the copper wires in the discharge circuit , the damping was found to be great for all capacities investigated .
When capacity of the circuit was only 130 electrostatic units , and inductance 2400 , no appreciable damping was found for an air break5 centim. : When the copper wire was replaced by an iron one of the same dimensions , the second h-.a !
foscillation was only6 of the amplitude of the first , ELECTRICAL WAVIIS AND SOME OF ITS APPLICATIONS .
Resistance of for High Frequency The rapid decay of the oscillations when iron wires form , the discharge circuit has been already noted .
Tbis has been observed by RIDGE ( ' Phil. Mag December , , who found , by photographing the spark , that there was evidence of much fewer oscillations when iron wires were used instead of copper .
For very rapid oscillations the resistance is given by ( sce Lord RAYLEIGH , " " On the Self-Induction and Resistance of Straight Conductors ' Phil. Mag 1886 ) , where is the resistance of the wire for steady currents , the le1lgth of the wire , the permeability , and , where is the number of oscillations per second .
Since the expression involves , we should expect the resistance to be much greater for iron wires than for wires of the same conductivity , but non-magnetic .
To determine the resistance of iron wires a very simple method was used .
The fall of deflection , due to the detector needle , arranged as in ( fig. 2 ) was noted .
The iron wire was then removed and a copper one of the same diameter substituted .
Since the inductance of the circuit was practically unchanged , if the damping in the two circuits are equal , the resistances should be the same .
A short piece of highresistance ] atinoid wire was introduced into the circuit of the copper wire , and the length adjusted until the deflection was the in the first case .
When this is , the resistance of the platinoid wire , together with the of the copper wire , is equal to the resistance of the iron wires for the frequency employed .
The resistance of the copper wires was calculated for the frequency but was , in general , smail compared with the resistance introduced .
resistance of the platinoid wire was also calculated , but was found to be practically the same as for steady curreuts .
The of wire placed in the current was 265 centims. ; spark length was25 centim. In the above Table , is the resistance for steady currents , the resistance frequency millims. second .
The last column gives the ratio for the different wires .
In the case of the wire centim. in diameter , bhe resistance is times the resistance for steady , while in the case of the wire of 025 rentim .
diameter , it is only times .
18 MR. E. RUTHERFORD ON A MAGNETIC DETBCPOR Qff If the value of , the permeability of the specimens of soft iron , be calculated from the formula it will be found that the value vary , with the diameter of the wire .
The above curve ( Curve 4 ) shows the relation between the diameter of the wire and the permeability for the discharge .
It be seen that varies approximately as the radius of the wire .
The very small value of the permeability for fine wires is to be expected when we consider the very large currents that pass through the wire , and the consequent Iarge value of the magnetic force that acts at the surface of the iron .
The maximum current of discharge , assuming the damping to be small , is given by , where is capacity and the potential to which condenser is charged .
In the above experiments the air-break was about inch , and the difference of potential about 10,000 voIts , and since 4000 , the maximum current was about 222 amperes .
The magnetic force at the surffice of t , he wire of radius , which conveys current , is given by C.G.S. units , if ELECTRICAL WAVES AND SOME OF ITS APPLICATIONS .
If we assume the value of to be about 14,000 , we see that the permeability for the extreme surface layer would be about 4 .
The value of the magnetizin force diminishes from tho surface inwards , so bhe mean permeability of the iron to the harge should be greater than the value at the face .
These considerations show that the permeability of iron to these discbarges is by no means constant , but depends on the diameter of the and the intensity of the discharge .
The resistance of iron wires was found to vary with the length of the spark .
Short air-breaks gave higher values of the resistance than long ones .
When the length of the spark was so adjusted that t , he maximum current was approximately constant for different periods , the resistance was found to vary as the square root of the frequency , as we should expect from theory .
Several specimens of pianoforte steel wire were examined to see whether the larger waste of energy due to hysteresis in steel affected the value of the resistance , but the increase of resistance was not so great as for soft iron wires of the same diameter , although the Ioss due to hysteresis in steel is much greater than in soft for slow cycles .
Absorption of Energy by Meted Cylinders .
This subject has been treated mathematically and experimentalLy by J. J. THOMSON ( ' Recent Researches , ' p. 321-326 ) .
It is there shown by observing the electrodeless discharge that a cylinder of iron placed in a solenoid absorbs considerably more energy than a copper one of the same dimensions .
The method adopted here admitted of quantitative as well as qualitative resuIts .
An ordinary Leyden .
was discharged through a solenoid of about thirty turns and 14 centims. long .
The metal cylinder was then placed in the solenoid and the damping of the oscillalions observed .
The cylinder was then removed and a noninductive resistanoe added until the damping was the same as when the metal cylinder was in the solenoid .
The absorption of energy in the cylinder was then equal to the absorption of energy in the added resistance , hose value was known .
the above we have taken no account.of the change of inductance of circuit due to the metal cylinder being placed in the solenoid .
The change is small , and could be made negligible by making the inductance of the solenoid small compared with the rest of the circuit .
( 1 .
) A test-tube was filled with finely laminated iron wire , 008 inch in diameter .
The test-tube was filled with paraffin 0il , to insure insulation from eddy rents .
The absorption of energy in this case corresponded to an added resistance of ohms to the circuit .
( 2 .
) A test-tube filled with steel and insulated as in ( 1 ) .
Increase of resistance , 9 ohms .
1 ) 2 MR. E. RUTHERFORD OAV A MAGNETIC DETECTOR OF ( 3 .
) A thin soft iron cylinder , centims. in diameter .
Increaso of resistanoe , ohms .
( 4 .
) Solid iron rod .
Increase of resistance , ohms .
A copper cylinder , a test-tube filled with a copper sulphate solution , and a platinum cylinder showed no appreciable absorption of energy .
( 6 .
) A carbon rod absorbed a large amount of energy .
Increase of resistance , ohms .
Absorption of Energy of various Conductors ; Absorption of Energy expressed in Terms of the increased Resistance of the Discharge Circuit .
The frequency of the oscillations in the above experiment two million per second .
If the experimental value ined for the increase of resistance due to the solid iron cylinder be compared with the theoretical value Recent Researches , , the value of the permeability will be found to be 172 , which accounts for the much greater absorption by an iron cylinder than a copper one .
From the peculiar deadeoed sound of the spark , it could always be told when much energy was being absorbed in the discharge circuit .
With copper wires for the discharge circuit , the spark was sharp and bright ; when iron wires substituted , the spark was weak ; when an iron cylinder was put in the place of a copper one , the spark was neither so bright , nor so sharp in sound .
Determination of the Period of a Discharge Circuit .
It is often a difficuIt matter to obtain even an approximation of the period of oscillation of a discharge circuit when the capacity of bhe condenser and the selfinductance of the circuit cannot be directly calculated .
The following simple method was found to work very accurately in practice , and could be used for a fairly wide range of frequencies .
Let ACB , ADB ( fig. 5 ) be two branches of a discharge circuit in paraIlel , and the resistance and of the branoh ACB , and resistance and inductance of branch ADB .
ELECTRICAL WAVES AN D SOME OF ITS APPLICATIONS .
Fig. 5 .
Let be the coefficient of mutual inductance between the two branches .
Let and currents in branches ACB , ADB respectively .
It is shown ( ' Recent Researches , ' p. 513 ) that for alapidly-alternating current of frequency , where , that , say ' A and are the maximum currents in the two branches ACB , ADB respectively , and If the circuits be so adjusted that and The value of the impedance is nearly independent of for rapid frequencies in ordinary copper wire circuits Suppose , and then If the value of the particular period was 2 ohms say , then a very small quantity .
We therefore see that , under ordinary circumstances , the resistances may be neglected in comparison with the inductanoes .
If the branoh ACB of the divided circuit consist of a high resistanoe of sbortlengh , and consequently small inductance , and the other branch of an inductance , and the maximum of the currents in the two branches are equal , then since may be negIected in comparison with , and supposing the value of to be small compared with N. If the inductances are unchanged , is practically a oonstant for all periods , and is therefore proportional to R. In practice , one branch of the divided circuit consisted of a standard inductance , , and the other branch of an electrolytio resistance , R. The equality of currents in the two circuits was obtained by altering the value of until the effect on the detector needles was for both circuits .
Sinoe the effect on the needle was the same in both circuits , the maximum values of the current are the same , since each branch is traversed by an oscillation of the same Fig. 4 .
It was not found necessary to place the divided circuit in series with the discharge circuit , but the arrangement was more satisfactory when it was shunted off a portion , , of the discharge , circuit XPQS ( fig. 6 ) , whose period is to be determined .
The addition of the shunt circuit had no appreciable effect on the period of the oscillation for the equivalent inductance of the two branches , QBAP , was slightly less than that of , and the length of QP was generally not a tenth part of the whole discharge circuit .
ACB , ADB are the branches of the divided circuit .
In was placed a resistance consisting of zinc sulphate with zinc electrodes .
The amount of resistance in the current could be varied by altering the length of electrolyte through which the current passed .
In was placed a standard inductance consisting of six turns of insulated wire wound on a bobbin 10 centims. diameter .
The self-inductance of this could be accurately determined by tion , and was very ately the same for steady as for rapidly changing fields .
If the inductance of the resistance branch was small compared with , the value of is given by ELECTRICAL WAVES AND SOME OF I APPLICATIONS .
Even if the correction would only be 1 per cent. The resistance of the zinc sulphate solutiou was determined for steady currents , and it can be shown that the change of resistance due to the ation of the currenls on the surface is quite inappreciable for the periods investigated on account of the high specific resistance of the solution .
This can experimentally be shown as follows : a tube containing the solution to be tested is placed inside a solenoid of a few turns , and a detector needle placed in the .
solution .
After the passage of a discharge it will be found that the effect on the needle is the same as when the solution is removed , showing that there is no screening action on the needle due to the .
Since the law of of mag1letic force ffom the surface inwards is the eame as for the decrease of amplitude of a current through the conductor , it follows that the amplitude of the current at the centre of the solution was the same as at the surface , and that there was no alteration of the resistance of the electrolyte due to concentration of the current on the surface .
An air condenser of calculable capacity was discharged through a circuit whose inductance for rapid frequencies could be very approximately determined .
The value of obtained from theory was found to agree to within 3 per cent. of the experimentally-determined value , and , from the difficulty of accurately calculating the inductance , it is probable that the experimental determination is nearer the true value .
From the close agreement of theory and experiment , we have indirectly proved that the resistance of an electrolyte like zinc sulphate is the same for high frequencies as for low .
As an example of the determination of the period of oscillation , the value of , the standard inductance , was 6500 units .
When the current was the same in both circuits , the value of was 168 ohms .
'fherefore The frequency The value of the capacity and the induotance for rapid frequencies of the discharge circuit could also be determined .
If a Leyden jar of unknown capacity be replaced by an air condenser of known , the value of remaining unaltered , and the value of the resistance necessary for equality of currents in the two circuits determined as before , then if and be the new tlues of and and ' and ON A OF ELHiCTRICAL WAVER , Therefore Since and have both been is known in terms of the standard capacity Similarly , the value of , the self-inductance for rapid frequeucies , may ako be found .
If an additional standard inductance be introduced inbo the circuit , the value of the capacity remaining unaItered , and Therefore The value of for rapid frequencies is thus determined in terms of , a known inductance .
In the experiments on the determination of periods , a detector sisting of twenty or more fine insulated steel ires , about 1 centim. long , was used , with one or two of wire round it through which the oscillatory current passed .
This small tector coil was fixed before a magnetometer , and was so arranged that it could be switched either into the resistance or inductance branch of the divided circuit .
The inductance of the detector coil was too small to appreciably alter the distribution of the current in either circuit .
The equality of currents in the two circuits could thus be readily compared .
The " " longitudinal\ldquo ; detector ntay also be used for rough determinations ; but it is not so sensitive to slight changes of current as the noidal detector of fine wire .
It was found that the inductance , of a current when the was of iron was nearly the same as when replaced by copper of the same diameter .
It was difficuIt to determine the variation of inductance accurately in this case , in consequence of the oscillations being rapidly damped when iron wires were used .
Since the capacities of condensers for very rapid aIternations may be determined , it was interesting to observe whether the values of the specffic inductive capacity of glass was the same for slow as for very rapidly varying fields .
Some observers had found that glass had a much lower specific inductive capacity for rapid oscillations than for slow , while others again found values about the same in the two cases .
The value of the specffic inductive capacity found for plate.glass was about for periods of about 3 million per second .
This is considerably higher than the value gtined for plate-glass by J. J. THOMSON and BLONDLOT , who found values of respectively for periods of about 20 miUion per second .
The value of the specific inductive capacity of ebonite , tested by the same method , was found to be abom the same as for slow alternations .
These experiments were in the Cavendish ambridge .

