A 20-M.eV BETATRON

By W. BOSLEY, M.Sc, J. D. CRAGGS, M.Sc., Ph.D., F.Inst.P.,D. H. McEWAN, B.Sc, Associate Member, and J. F. SMEE, Associate Member.

{The paper was first received \4th October, 1947, and in revised form 30th April, 1948.)

SUMMARYThe construction of a 20-M.eV betatron is described. Tests made

to measure the magnetic field and to locate the orbit position, etc.,are fully discussed, and details are given of some new measurementson the out-of-phase fields at the instant of electron injection.

(1) INTRODUCTION*Recent rapid developments and increasing interest in nuclear

physics have stimulated the great advances which have beenmade in experimental techniques, especially during the past fiveyears. The greatest importance is attached in particular to thestudy of problems involving the use of high-energy atomicparticles, and the equivalent voltage of the available nuclearprojectiles (protons, deuterons, alpha particles, etc.) has increasedcontinuously since the pioneer work of Cockcroft and Waltonin 1930 with protons of an energy equivalent to a few hundredkilovolts.

The acceleration of electrons to high energies has proved moredifficult than has been the case with heavier particles, since therelativistic variation of mass, which sets the limit to the energyobtainable with such accelerators as the cyclotron, becomesappreciable at lower energies for the lighter particles. In theperiod 1940-42, the induction acceleration of electrons to some2 M.eV and then 20 M.eV> was achieved, largely owing to thework of Walton,2 Wideroe,3 Steenbeck,* Tuck,s and finallyKerst,^ who first achieved practical results. His general designof accelerator (the betatron) developed in conjunction withCharlton and Westendorp at the General Electric Company'sresearch laboratories at Schenectady, reached an energy of100 M.eV.7

The historical development is described elsewhere;5'8-9 thepresent paper deals with the construction of a 20-M.eV betatronbuilt to the Kerst/G.E.C. design.

(2) THEORY OF OPERATION OF THE BETATRONThe theory of operation of the betatron (Kerst and Serber10),

is summarized below.The electrons in a betatron are accelerated in circular orbits

inside a toroidal vacuum chamber which is placed between thepoles of an electromagnet excited by alternating current. It isfound (be. cit.) that the electrons follow an orbit of constantradius r0 (see below) irrespective of the value of the magneticfield (increasing with time during the acceleration process) if

1(1)

where Ho and <f>0 are the field at r0 and the total flux within acircle of radius r0, respectively. The subscript / indicates thevalue of a quantity at the time of injection of the electrons intothe field.

(U.D.C. 621.385.83)Written contributions on papers published without being read at meetings are

invited for consideration with a view to publication.The authors are with the.Metropolitan-Vickers Electrical Co., Ltd.• <{>o< does not —> 0 in the case of flux-biased betatrons.22

Generally HOi and <j>Oi -> 0 so that

<J>0 = 27TrlH0 (2)

or, since Bo = ^o/77"^ = t^ie m e a n flu* density within the orbit,

H0 = ±B0 (3)

It is necessary, to ensure focusing of the beam, particularlyduring the early stages of acceleration, that the radial variationof the magnetic field should be given by

Hoc —, where 0 < n < 1r

(4)

It can be shown10 that under these conditions the electronsperform damped oscillations in both the axial and radial direc-tions about a path which tends asymptotically towards theequilibrium orbit at r0. It may also be shown that as thevelocity of the electrons during their acceleration tends towardsthe velocity of light, the force of repulsion between them tendsto zero, so that focusing forces are most important at the timeof injection.

(3) DESCRIPTION OF THE APPARATUSThe present 20-M.eV betatron (built closely to Kerst's design)

is shown in Fig. 1, the overall dimensions of the magnet and itsplinth being 5 x 3 x 6 ft high. The magnet weighs 3£ tons andis enclosed by walls of concrete loaded with barium sulphate for

._ ^ ^ 9

Fig. lA.—Front view of 20-M.eV betatron.[ 352 ]

BOSLEY, CRAGGS, McEWAN AND SMEE: A 20-M.eV BETATRON 353

Fig. IB.—Rear view of 20-M.eV betatron.

X-ray protection, these weighing a further 30 tons. Electronsare "accelerated in an orbit of 19 cm radius by a peak field of3 600 oersteds at full excitation, so producing X-rays of peakenergy 20 M.eV when caused to bombard a target.

(3.1) The Excitation Circuits and Frequency-Tripling TransformerThe magnet, which is shown in Fig. 2, and which was built

of 0-014 in silicon-steel laminations, was excited by two coils

63'

j r

i1 a

' f i •>h !

: ^lliMH|kii&Hp

i S e

L

11(11 | | | | ~ m I

iMHh1 °!

! :

Fig. 2.—20-M.eV betatron.(a) Upper magnet yoke. (h) Expander coil,(i) Lower magnet yoke. (i) Vacuum chamber.(c) Magnet side-piece. 0 ) Primary winding.(d) Upper pole-piece. (k) Secondary winding.(e) Lower pole-piece. (/) Lower coil.(/) Centre-piece. (m) Upper coil.(g) Electron gun. (n) Clamping bolt.

(p) Pumping tube.

placed round the pole-pieces and tuned to resonance at theoperating frequency of 150 c/s by a bank of condensers. Thepower required to maintain the oscillation in this circuit wasfed into two other coils connected to a specially designed fre-quency-tripling transformer, the circuit of which is shown inFig. 3.

700V150-c/soutput 9 L

440 V3-phase'O

Fig. 3.—Circuit of frequency-tripling transformer.Pi, P2, P3. Primary windings.Si, S2, S3. Secondary windings.TSi, TS2, TS3. Tapping switches.Li, L2. Auto-transformers.

The condenser bank consisted of 104 units of 1-2/AF each,connected so as to give an overall capacitance of 7 • 8 [x¥ rated at106 amp, 14 000 volts, 150 c/s. The "secondary" coils, acrosswhich the condensers were connected, were wound with strandedcopper cable, each strand being individually insulated to minimizeeddy-current heating. There were a total of 162 turns, woundwith two cables in parallel, as shown in Fig. 4. The primary

HV fuse

24-26 Capacitorsin parallel il

81 turns: Upper?coils <

14000 V106A ( .!50c/s <-A

81 turns;Lcwer«ceils*

700 V30 A

150 c/s

HV fuse

Fig. 4.—Magnet connections.

coils each consisted of six turns wound round the secondariesand connected, in parallel, to the output of the frequency-tripling transformer.* The particular transformer used had anoutput of 40 amp at 700 volts, 150 c/s, and took an 87-kVA,25-kW load from the 50-c/s 440-volt 3-phase mains.

(3.2) The Vacuum SystemThe toroidal vacuum chamber, which fitted between the

magnet pole-pieces and in which the electrons were accelerated,was built up in the manner described by Kerst.1 The chambermay be seen in position in Fig. 5 with the upper part of themagnet raised. In Kerst's 20-M.eV betatron a high-resistancelayer of silver had been deposited on the inner walls of thevacuum chamber and earthed to prevent the accumulation ofstray charges, but it had been stated that the technique ofdepositing such a film uniformly was difficult, and because of

• The theory of which has been described by Brailsford.11

354 BOSLEY, CRAGGS, McEWAN AND S.MEE: A 20-M.eV BETATRON

Fig. 5.—Vacuum chamber in position on lower pole-piece.

the comparative ease with which a coating of graphite could beprepared this was originally used in the present apparatus.

The electron gun was of the simple 3-electrode type1 with amolybdenum anode 0 01 in thick. The anode was normallyearthed, and high-voltage pulses were applied to the filamentand focusing electrode. Ionization gauges were mounted onthe pumping connection near the oil-pump and on the vacuumchamber itself. A large vacuum tap was provided so that thepumping set might be isolated in the case of a leak occurring inthe chamber. After several days' pumping (required to de-gasthe graphite, which could not be vacuum-baked) a pressure ofthe order of 1 x 10-5 mm Hg could usually be obtained.

Initially the pressure was found to rise to about 10~4 mm Hgwhen the electron beam was circulating in the chamber, owingto gas liberated from the graphite by electron bombardment.During the continued operation of the betatron this rise inpressure decreased, but it could not be completely avoidedexcept by the use of a silver-coated vacuum-chamber.

(3.3) The Injector and Expander CircuitsThe circuit used to generate the high-voltage high-frequency

pulses required to fire the electrons into the vacuum chamberjust after the time of zero flux was similar to that described byKerst.1 An oscillogram of the injector voltage is shown inFig. 6(a), the peak voltage in this case being 12-5 kV. The

Fig. 6.̂ -Oscillograms of: (a) Injector voltage with 3-̂ sec timingoscillation; (6) Expander current, with time scale in microsec.

calibration wave in this Figure has a period of 3 microsec, sothat the pulse is about 8 microsec long. Voltages of up to35 kV have been obtained from this circuit.

After being accelerated for one quarter-cycle of the magneticfield, the electrons were made to spiral outwards from theirorbit towards the anode of the electron gun, which acted as anX-ray target. This was achieved by a heavy current pulseobtained from the expander circuit and passed through a coilsituated near the orbit region so as to produce a field only overthis region. The field from this coil was arranged to oppose

the main magnet field, so that the flux relation of eqn. (2) wasmade inapplicable and thus the electrons moved outwards.Current pulses of up to 850 amp were obtained from this circuit,such a pulse being shown in Fig. 6(6), the duration beingapproximately 80 microsec.

(4) INTTIAL ADJUSTMENTSBefore the magnet was excited it was necessary to determine

the resonant frequency of the tuned secondary and to adjust itto the operating frequency of 150 e/s. This was first doneapproximately by connecting a beat-frequency oscillator acrossthe primary coils. When the magnet was excited the resonancecurves shown in Fig 7 were plotted, in which (a) shows primary

~4>o 40-8 500 502Mams frequency, c/s

Number of capacitors in circuit

Fig. 7.—Variation of primary current with (a) Mains frequency;(6) Secondary tuning capacitance.

current against frequency and (b) primary current againstsecondary tuning capacitance, the first curve being obtained bymaking use of the frequency variations of the supply mains.These variations had limits of ± 4%, corresponding to a requiredchange in tuning capacitance of ± 8% It was arranged thata suitable number of the condensers were connected in circuitby short straps which could readily be removed. Since it wasdesirable to limit the current drawn from the frequency-triplingtransformer at any time to 50 amp, it may be seen from Fig 7(b)that the tuning was critical to one condenser unit.

When the magnet was energized a loud note was emitted.An acoustical analysis of this note showed it to have an intensityof approximately 106 phons at a distance of one metre in frontof the magnet, consisting of a 97-phon 300-c/s fundamental witha 93-phon 600-c/s and a 79-phon 900-c/s harmonic.

Oscillograms were taken of some of the waveforms in themagnet circuit. These are shown in Fig. 8, in which (a) showsthe primary voltage against a linear time-base, the upper tracebeing a 50-c/s calibration wave. Fig S(b) shows the secondaryvoltage, (c) the primary current and (d) the rate of change ofmagnetic flux. All were sinusoidal except the primary currentwhich, due to the non-linearity of the hysteresis curve of themagnet steel, had an appreciable third-harmonic content.Figs. S(e) and 8(/) show respectively the primary current againstprimary voltage and secondary current against secondaryvoltage, both taken with the secondary tuned to resonance atthe tripler output frequency, the primary current then being aminimum.

BOSLEY, CRAGGS, McEWAN AND SMEE: A 20-M.eV BETATRON 355

(c)

(e)

Fig. 8.—Magnet oscillograms.(a) Primary voltage with 50 c/s calibration wave.

Secondary voltage.Primary current.Rate of change of flux,

(e) Primary current against primary voltage.(/) Secondary current against rate of change of flux.

\a) rrin(b) Sec<(c) Prir(of) Rat

Table 1 shows a typical set of readings of the various currentsand voltages in the circuit, together with the mains frequencyand corresponding number of tuning condensers.

o 18 20 22 24Radius r , cm

Fig. 9.—Variation of magnetic field with radius.

transformer, sufficient to give an output of approximately22-23 M.eV.

(5.2) Determination of Orbit RadiusA rough estimate of the radius of the equilibrium orbit was

obtained from measurements of total flux inside the centre-pieceand of the radial variation of the magnetic field, since it waspossible to determine graphically the radius at which the relation<f>0 = 27rrlHQ was satisfied. This method was not accurate,

Table 1

Mains frequency,c/s

50-2

Operatingfrequency,

c/s

150-6

No. of tuning icondensers i

i95 ;

amp

106

Tripler input

volts

445

Magnet primurj

amp

34

volts

500

Magnetsecondarycurrent,

amp

96

(5) MEASUREMENTS OF THE MAGNETIC FIELDThe preliminary adjustments having been made, it was neces-

sary to determine the radial variation of the magnetic field, thecircumferential uniformity of the field and the radius of theequilibrium orbit.

(5.1) Measurement of the Radial Variation of the Magnetic FieldThe radial variation of the field was determined by means of a

small search-coil supported by a Bakelite board so that itscentre lay in the median plane of the gap, the board beingdrilled with holes at radii of from 6 in to 11 in in £-in steps.Three turns of wire were wound round one side-limb of themagnet, and the voltage generated in them was measured foreach reading of the search-coil voltage so that the latter couldbe corrected to a standard value of total magnetic flux. Theradial variation was measured at six angular positions in thegap, and the readings at a given radius were found to be constantround the gap within the accuracy of measurement (0-5%), thevalue of n (see Section 2) being 0-75 ± 0 02 over the regionoccupied by the vacuum chamber, as shown in Fig. 9.

From a knowledge of the effective area of the search-coil, theorbital field value could be calculated and the output energy ofthe electrons estimated. The field was found to have the requiredvalue of 3 600 oersteds (peak) for normal excitation (a secondarycurrent of 106 amp) within the accuracy of measurement,corresponding to 20 M.eV output energy. It is possible, how-ever, to obtain a higher voltage from the frequency-tripling

however, because the leakage flux from the centre-piece couldnot be measured. The second method used was originally dueto Kerst6 and made use of the fact that at the equilibrium orbitthe tangential electric field produced by the flux is a minimum.A set of 13 coils was used, all the coils being wound with thesame length of wire about a common axis at radii such that eachhad one more turn than the adjacent coil of larger radius. Thecoils fitted inside the magnet air-gap, and, since the minimumwhich occurred when the coil voltage was plotted against radiuswas very shallow, adjacent coils were connected in opposition.A voltmeter then showed the magnitude of the resultant voltage,and a cathode-ray oscillograph showed its polarity by comparisonwith a standard 150-c/s wave picked up by a one-turn coil woundround a side-limb of the magnet.

A curve of coil-voltage against radius is shown for a particularcase in Fig. 10, the minimum in this case occurring at a radiusof 21 cm, beyond which there is a maximum due to the falling-offof the magnetic field. This method was found to be less accuratethan the one finally used, because the radii of the coils wereaccurate only to 0-5%, which corresponded to an appreciablechange in the difference voltage near the orbit. '

The method of orbit-radius determination finally used was amodification of a null method originally due to Westendorp andCharlton,? who used a single-turn coil wound at the radius ofthe required orbit and a small search-coil whose effective areawas equal to twice the area of the single turn and which wasplaced at the circumference of this turn. When the orbit radius

356 BOSLEY, CRAGGS, McEWAN AND SMEE: A 20-M.eV BETATRON

Fig.

"15 17 19 21 23 25 27Radius , c m

10.—Variation of pick-up coil voltage with radius.

had the required value the voltages induced in the two coilswere equal, since Bo = 2H0, so that, with the two coils connectedin opposition and to an oscillograph, the magnet air-gap couldbe varied until the orbit was at the required radius as shown bya null reading on the oscillograph.

In the present case it was required to determine the orbitradius rather than merely to have a null method of determiningwhen it had the required value, and so single-turn coils werewound at radii of 17, 18, 18-5, 19, 19-5, 20 and 21 cm, and,in addition to determining the difference between the voltageinduced in the 19-cm single turn and that in the search-coil placedon a 19-cm radius, the actual voltage in each single-turn coilwas measured, the orbit radius then being readily calculated froma knowledge of the coil voltage-gradient.

The separation between the upper portion of the magnet yokeand the two side-pieces was varied to adjust the value of theequilibrium orbit radius, since raising the top portion increasedthe air-gap at the centre of the pole-pieces by a greater percentagethan that at the orbit. This caused a greater reduction of centralflux than in the orbital field and decreased the orbit radius. Itwas found that raising the upper yoke by 3^ in reduced theorbit radius by approximately 1 cm.

(6) THE INJECTION PROCESS: DETECTION AND MEASURE-MENT OF OUT-OF-PHASE MAGNETIC FIELDS

(6.1) Effect of Circumferential Variations of Orbital MagneticField

Since the injection voltage and orbital magnetic field necessaryfor the electrons to move on the required orbit in the field are

H2 er2

connected by the relation V, = J31 ", then for any given2mc2

electron r0ccl/H0l, so that if the orbital magnetic field variescircumferentially the radius of the equilbrium orbit will alter,and, as can readily be shown, the electrons will tend to oscillateabout their new orbit and may bombard the walls of the vacuumchamber.

A high gas pressure (perhaps > 10-s mm Hg) or the existenceof large out-of-phase magnetic fields will probably cause muchloss of beam current during the early part of the accelerationperiod, i.e. until the loss of energy by scattering becomes lessthan the increase of energy by acceleration between collisions.It appears that in all the electron accelerators known to theauthors a considerable background of very soft (~ 1 M.eV)radiation is emitted, and it is suggested that much of this arisesat about the time of injection.

(6.2) Detection of Out-of-Phase Magnetic FieldsThe search-coil measurements of the orbital magnetic field had

failed to show any circumferential variation in this field. It wasrealized, however, that comparatively small localized magneticfields caused by excessive eddy-currents in parts of the pole-piecesmight produce appreciable deflections of the electron beam atthe time of injection, even though these fields were too small tobe detected by such measurements. At the time of injection,the'main orbital magnetic field is small (about 25 oersteds for20 kV injection) and is increasing rapidly, as shown in curve Aof Fig. 11. The field caused by an eddy-current, however,

Out-of-phasefield • 5 oersteds

Curve A.Curve B.

10 20Electrical minutes of arc

Fig. 11.—Effect of out-of-phase fields.Main field (3 600 oersteds peak).Out-of-phase field (5 oersteds peak).

30

Curve C. Resultant field with time of zero field shifted by 5 min of arc.

being 90° out-of-phase with the main field, has its peak valueand remains sensibly constant during the period of injection(curve B, Fig. 11). If the out-of-phase fields are distributeduniformly round the pole-pieces then no disturbance of the beamwill result, except in so far as they produce a phase differencebetween the orbital field and the central flux, resulting in agradual expansion or contraction of the orbit during acceleration.If the out-of-phase fields vary round the orbit, however, theresultant of the main field and the out-of-phase fields (curve C,Fig. 1,1) will also vary and serious deflections of the beam mayresult.

It may be seen from Fig. 11 that one effect of a change inout-of-phase field is to alter the time at which zero flux occurs,and the method devised to detect such changes depended uponthis. "Peaking strips" were built which gave sharp voltagepulses as the flux through them passed through zero, and byplacing two such strips in the magnet air-gap, one at a referencepoint and the other at some other point on the orbit, and arrangingfor their pulses to oppose one another, the size of the resultantpulse could be used to give a measurement of the difference inthe time of zero flux and so of the difference in out-of-phasefields at the two points.

The peaking strips originally used consisted of five pieces of0-0025-in thick Permalloy "C" metal, an alloy with a highpermeability and low saturation flux density. Each of the fivepieces was 1 cm wide by 6 cm long, the pieces being enamelledand wound with 25 turns of thin copper wire, and then bound toa piece of -fam Bakelite board. Two such strips were made,and when placed in the field of the magnet each was saturated

BOSLEY, CRAGGS, McEWAN AND SMEE: A 20-M.eV BETATRON 357

except for a short time-interval including the time of zero field,during which interval a sharp voltage pulse was induced in thewinding round the strip. The pulses were observed separatelyon a high-speed oscillograph to check their similarity and werefound to be trapezoidal and about 7microsec long, having aheight of 20 volts. The two windings were then connected inopposition, and the resultant pulse was examined when the stripswere as close together as possible and at the same radius. Theresultant was found to be small and to be inverted when thepositions of the strips were interchanged, indicating zero phasedifference when the positions coincided, and confirming thatthere were no spurious shifts in the apparatus.

Comparisons were next made between a reference point takenat the front of the magnet gap and points at 90° and 180° to thisposition. Differences in the time of zero flux of up to 2 microsecwere detected, indicating out-of-phase fields of up to 6 oerstedspeak value.

(6.3) Plot of Out-of-Phase Fields round OrbitMore sensitive peaking strips were made, their lengths being so

chosen that they almost completely bridged the air-gap in themagnet so as to reduce their self-demagnetizing effect and sosharpen the voltage pulses. Each consisted of five strips ofPermalloy "C" 0-0025 in thick, wound with 150 turns of0 0076-in copper wire. One strip was fixed to the edge of aBakelite ring fitting round the magnet centre-piece and restingon three supports so that it could be rotated at the orbit radius.The other remained stationary at the reference point, the angularseparation between the two strips being read off from a scalefixed to the circumference of the Bakelite ring.

The shape of the pulses obtained from these "peakers" isshown in Fig. 12(a), the pulse being about 3-5 microsec wide at

Fig. 12.—Pulses from improved peaker strips.(a) Single pulse.(b) Resultant of two pulses in opposition with peakers placed close together.(c) Resultant of two pulses with difference in out-of-phase field of approximately

3 oersteds.

half-peak amplitude, and having a peak of 200 volts. Fig. 12(6)shows the resultant pulse obtained when the peakers were placedclose together, and (c) shows that observed when there was adifference of about 1 microsec in the times of zero flux at thepositions of the strips, corresponding to a difference in out-of-phase fields of 3 oersteds (peak). The difference in the timesof zero flux was determined by measuring the height of the firstpeak of the resultant pulse and using the oscillogram of a singlepulse [Fig. 12(a)] as a graph of pulse height against time. Acheck was made that the presence of the strips did not affect the

out-of-phase fields appreciably, by placing "dummy" strips inthe gap and repeating earlier measurements.

In this way a plot of the variations of the out-of-phase fieldsround the orbit was made, giving the curve shown in Fig. 13.The accuracy of these measurements was about ± 0 - 5 oersted.

5

Fig 13.—Variations of magnetic field round orbit at time of injection.

An attempt was made to correct these field variations by meansof small coils of copper wire attached to the face of one pole-piece and extending over the region of each of the longest varia-tions of Fig. 13, a suitable direct current being passed throughthe coil to produce a field equal and opposite to the peak valueof the difference in fields at the region of the coil and thereference point. This was not very successful, however, owingto the rapid changes in the out-of-phase fields which occurredover short distances round the orbit, and it was found that ifsufficient current was passed through the coil to reduce one peakappreciably, it usually made the adjacent trough much worse.The effect of the variations was eventually reduced by increasingthe injection voltage so that the electrons could be injected at alater time in the acceleration period when the main magnet fieldwas increased. An injection voltage of 30 kV (peak) was foundto be satisfactory.

By means of two sharp probes attached to a resistance-metera measurement was made of the resistance across each packet oflaminations in the pole-pieces, and it was found that in mostcases a conducting path existed, although it was not possible todetermine whether the two paths in parallel, necessary for aneddy-current field to be set up in the packet of laminations, werepresent. It was found, however, that those positions at which ahigh resistance was registered coincided with the positions ofminimum out-of-phase field variations in Fig. 13. This suggestedthat short-circuits in the pole-pieces rather than in the mainmagnet yoke were responsible for the eddy-current fields. Theadvisability of increasing the injection voltage was apparent, ofcourse, from Kerst's papers, but it was considered that thiscould best be done after measurements of the out-of-phase fieldvariations had been made, since the magnitudes of the latter wereinitially unknown. It is clear that the pole-pieces used in theseexperiments could be greatly improved, and work to this end isin hand.* These measurements have been described briefly bythe present authors elsewhere.12 (Since this paper was first

* The importance of these out-of-phase fields in disturbing the beam during theearliest part of the accelerating period would appear to be appreciable, since experi-ments with a single-piece (not fabricated, Section 3.2) vacuum chamber giving alower pressure did not, apparently, give less scattering. Further, a detailed butapproximate mathematical analysis of the effect of the out-of-phase fields on thebeam movement taken for several beam revolutions indicates that for 20 kV injectionmost of the beam will be disturbed so much as to strike the chamber walls and be lostsoon after injection. At 30 kV the effect is less serious but probably still appreciable,and experiments with 50-kV injection are in hand.

358 BOSLEY, CRAGGS, McEWAN AND SMEE: A 20-M.eV BETATRON

written some similar measurements by American workers havebeen published.13)

(7) USES OF THE BETATRONBetatrons (and the new electron accelerators named syn-

chrotrons1^ is, 16, n, which have some important advantages overthe earlier betatron, especially for the higher energies) have sofar largely been used in three main fields, namely radiography ofmetals,^ preliminary deep-therapy work19 and nuclear physics.20The present installation, described above, was built mainlyto give preliminary experience in the operation of electronaccelerators, but also for nuclear physical investigations.

Several of the latter have already yielded preliminary results,i.e. the investigation of scattering and energy losses for fastelectrons and positrons, and the checking of certain aspects ofthe quantum theory of radiation.21 Results have also beenobtained in investigations of the photo-disintegration of thedeuteron; and the photo-disintegration of boron, beryllium andlithium are being studied.* Techniques for studying the photo-fission of uranium and thorium have also been developed.

(8) ACKNOWLEDGMENTSThe authors are indebted to many of their colleagues ana to

workers in other laboratories. They wish to thank Dr. T. E.Allibone, who initiated the work, and Mr. F. R. Perry, whoshortly afterwards became responsible for the direction of theelectron accelerator projects, for their support and encourage-ment. Mr. S. W. Redfearn has helped generously and invaluablywith the design of peakers and pulse-transformers. Mr. J. B.Hansell has been generally responsible for the design and manu-facture of the magnet, and designed the frequency-tripling trans-former for this work.

The authors are indebted to Prof. D. W. Kerst for helpfulcorrespondence, and to the staff of the Betatron Laboratory(under the direction of Dr. E. E. Charlton and Mr. W. F.Westendorp) of the General Electric Company of America.

Finally, the authors are grateful to Sir Arthur P. M. Fleming,C.B.E., D.Eng., Past-President, Director of Research andEducation, and to Mr. B. G. Churcher, M.Sc, Member,Manager of Research Department, Metropolitan-Vickers Elec-trical Co., Ltd., for permission to publish the paper.

(9) REFERENCES(1) KERST, D. W.: Review of Scientific Instruments, 1942, 13,

p. 387.(2) WALTON, E. T. S.: Proceedings of the Cambridge Philo-

sophical Society, 1929, 25, p. 469.(3) WIDEROE, R.: Archiv fur Electrotechnik, 1928, 21, p. 387.(4) STEENBECK, M.: U.S. Patent No. 2103303, .1937.(5) KERST, D. W.: Nature, 1946, 157, p. 90.(6; KERST, D. W.: Physical Review, 1941, 60, p. 47.(7) WESTENDORP, W. F., and CHARLTON, E. E.: Journal of

Applied Physics, 1945, 16, p. 581.(8) CRAGGS, J. D.: "Electronics, their Applications in Research

and Industry," edited by A. C. B. Lovell (Pilot Press,London, 1947).

(9) BOSLEY, W.: Journal of Scientific Instruments, 1946 23,p. 277.

(10) KERST, D. W., and SERBER, R.: Physical Review, 1941, 60,p. 53.

(11) BRAILSFORD, F.: Journal I.E.E., 1933, 73, p. 309.(12) BOSLEY, W., CRAGGS, J. D., MCEWAN, D. H M and SCAG,

D. T.: Nature, 1947, 159, p. 229.(13) ADAMS, G. D., and KERST, D. W.: Review of Scientific

Instruments, 1947, 18, p. 799.(14) VEKSLER, V.: Journal of Physics of the U.S.S.R., 1945, 9,

p. 153.(15) MCMILLAN, E. M.: Physical Review, 1945, 68, p. 143.(16) GOWARD, F. K., and BARNES, D. E.: Nature, 1946, 158,

p. 413.(17) OLIPHANT, M. L., GOODEN, J. S., and HIDE, G. S.: Pro-

ceedings of the Physical Society, 1947, 59, p. 666.(18) GIRARD, J. P., and ADAMS, G. D.: Transactions of the

'American I.E.E., 1946, 65, p. 241.(19) KOCH, H. W., KERST, D. W., and MORRISON, P.: Radiology

1943, 40, p. 120.(20) BALDWIN, G. C, and KOCH, H. W.: Physical Review, 1945,

67, p. 1.(21) BOSLEY, W., CRAGGS, J. D., and NASH, W. F.: Nature 1947,

160, p. 790.(22) WESTENDORP, W. F.: Journal of Applied Physics, 1945, 16,

p. 657.