THE DESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE*By M. E. HAINE, B.Sc, Associate Member.f

(The paper was first received 28th October, 1946, and in revised form 21th January, 1947. // was read before the MEASUREMENTS SECTIONlist March, and the SOUTH MIDLAND RADIO GROUP 29th May, 1947.)

SUMMARYAn introduction to the electron microscope is given. Starting from

the limitations to the resolving power of the optical microscope, thediscovery of de Broglie concerning the wave aspect of movingparticles and the derivation of the electron lens by Busch are described.The limiting action of the spherical aberration of the electron lens isdiscussed and the theoretical resolving limit set by this and diffractioneffects is evaluated as 10 Angstrom units (A).

A very brief historical outline of early work is given.The functions and requirements of the various parts of a practical

instrument are discussed and a description is given of the new instru-ment designed by the author and his collaborators. The descriptionincludes details of the mechanical construction of the tube and itsaccessories, as well as general notes on the vacuum system, circuits andoperation of the instrument. Finally, a few examples of resultsobtained on the instrument are given.

(1) INTRODUCTIONThe resolving power of the microscope or the ability of the

instrument to resolve fine detail may be quantitatively definedby a dimension known as the resolution limit, this being theleast distance which can separate two minute or point particleswhile they still remain separate in the image. The definition isquite apart from contrast and intensity considerations, thesebeing assumed adequate for visibility.

The resolution limit (Aj) of the best optical microscopes isdetermined by diffraction effects. Abbe derived the followingexpression for this limit for one condition of illumination:

0-61Afj. sin a (1)

where A is the wavelength of the light in vacuo, [x. is the refractiveindex of the medium in the object space, and a is the semi-angleof the cone of rays leaving an object point and entering theobjective lens. Subsequent work showed this expression to holdto a close degree of approximation for other conditions ofillumination.

Since in practice sin a cannot exceed a value just less thanunity, and as suitable immersion mediums of refractive indexgreater than 1-7 are not available, minimum resolution limitsare of the order of half the wavelength of the light used. Forvisible light the limit is therefore about 2 500 A.J

It is of interest to note that this limit was reached and fairlywell understood by the middle of the nineteenth century. Forover fifty years the only worthwhile improvement resulted fromthe use of ultra-violet light with fused quartz lenses, makingpossible a resolution limit of the order of 1 500 A. Further re-duction of wavelength into the X-ray spectrum is limited by thelack of suitable materials to refract the radiation, though someadvance in this direction may be possible with the use of re-flecting surfaces.

After this long hiatus in the development of the microscope agreat stride forward started with the work of Busch on electronlenses. In 1926 he showed mathematically! that an axially sym-metric magnetic or electric field, as previously used to concentrate

* Measurements Section paper.t Metropolitan-Vickers Electrical Co., Ltd. X 1 A - 10-8 cm.

the electron beam in the Braun tube, acted on the beam in amanner closely analogous to the simple optical lens on a beamof light. Previous to Busch's discovery the "wave nature" ofthe electron in motion had been suggested by de Broglie in 1924and verified by Davisson and Germed in the United States andG. P. Thomsons in England. The equivalent wavelength (Ae) ofthe electron was shown to be a function of its velocity:

= . l(^i\ A (2)where V is the voltage through which the electrons have beenaccelerated from rest ("the volt velocity").

This expression showed the electron wavelength, even whenaccelerated through a few tens of kilovolts, to be 100 000 timessmaller than that of blue-green light, suggesting the possibilityof obtaining greatly improved resolutions in a microscope usingan electron beam instead of light.

The properties of practical field distributions in electron lensesand their relation to the trajectories of electrons traversing themhave been studied in great detail, mainly by mathematicalmethods due to a number of workers such as Busch, Scherzer,Glaser and others. Scherzer4* has shown that aberration-freeelectron lenses are possible only when they are very weak. Ingeneral the strong electron lens suffers from all the aberrationscommon to the uncorrected optical lens, and in the magneticlens three additional aberrations are effective.

Busch showed that for paraxial electrons % the focal length / o fa magnetic lens was given by

I = ^\HZ (3)where elm is the ratio of electronic charge to mass, V is the voltvelocity of the electrons, Hz is the value of the magnetic fieldalong the axis of the lens, and the integration is carried out fromone side of the lens to the other.

In addition to this it was shown that the magnetic lens has thepeculiar property of rotating the image through an angle (ftgiven by:

The derivation of the lens aberrations5 by consideration ofthe extra-axial electron trajectories shows that the sphericalaberration predominates in limiting the resolution of the electronmicroscope. This aberration results in a beam of electrons fromone image point being focused in a "disc of confusion" ofdiameter A* given by:

A, = Q/a3 (5)where Cs is a constant dependent mainly on the field geometryand termed the spherical aberration constant, / is the focallength of the lens for paraxial electrons as given above, and a hasthe same significance as in equation (1).

[447]X Electrons traversing the lens near the axis.

448 HAENE: THE DESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE

From this expression it is seen that the effect of the sphericalaberration can be reduced indefinitely by reducing a. On theother hand it has been shown earlier in the paper that the reso-lution limit due to diffraction as given by equation (1) becomesgreater as a is reduced. It will be noted here that we are as-suming the latter expression to be valid for electrons. That thisis so can be seen from considerations of the uncertainty principleas shown by Heisenberg6 in 1927.

It is clear from the foregoing that when both the sphericalaberration and the diffraction error are present, a minimum valueof the resolution limit will occur at some optimum value of a.To obtain a rigid value for this optimum it is necessary to treatche combination of the two errors on a wave-mechanical basis,though a close approximation may be obtained by assumingthem to sum up as error functions.7 Thus the total error A, isgiven by: ,

(6)By substituting (1) and (5) in this expression and differentiatingwith respect to a to find a minimum value, Gabor8 derives thefollowing expressions for the optimum aperture angle (

HAENE: THE DESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE 449

Electron gun

Condenser lens

Objective lens

Intermediate screenProjector lens

Fluorescent screen

Fig. 2.Schematic ray diagram of electron microscope.(3.1) General Arrangement of the Main Tube

The general arrangement of the main tube assembly is shownin Fig. 3. The tube is mounted on a cast brass pedestal (1) whichcontains the pumping system except the rotary pump which ismounted separately to minimize vibration. The tube itself con-sists of a cast brass viewing chamber, (2) the intermediatesection (3) and object section (4) of drawn brass tube, and theelectron gun (5). The individual sections will now be describedin detail.

(3.2) The Illuminating SystemThe illuminating system consists of the electron gun and the

condenser lens assembly. In combination these fulfil thefunctions of producing a beam of electrons of the requiredvelocity, intensity and convergence angle.

The electron velocity is a direct function of the accelerationvoltage. The choice of voltage is reached by consideration ofits effect on the resolving power and contrast in the image.Cosslett14 and others have shown theoretically that resolutionwould be expected to improve as the accelerating voltage in-creases, though the rate of improvement would be slow above60-100 kV. Optimum contrast conditions depend on the typeof object in use. For a very thin object of low atomic number,such as a virus particle, voltages above 30 kV may give poorcontrast, while a few thick biological objects may require severalhundred kilovolts to show up internal detail effectively. For ageneral-purpose instrument it was decided that a voltage variablefrom 25 to 50 kV would be most suitable and economic.

Fig. 4 shows the arrangement of the electron gun. The fila-

ment (1) consists of a 0 005in-diameter tungsten-wire hairpinwith the sharply bent tip centrally positioned by a simple settingjig (Fig. 5) just behind a 0 020 in diameter aperture in the cathodeshield (2). The cathode shield, shaped to maintain a narrowlydivergent beam, is at cathode potential. The anode (3) is acopper block drilled centrally with a 0-012 in diameter aperturethrough which the central portion of the beam passes. Theanode supports a shield (4) which prevents secondary electronsfrom the anode building up a charge on the insulating glasscylinder (5) and also absorbs X-rays.

The cathode and anode assemblies are spigoted into the endflanges (6) which are fitted with demountable rubber vacuumseals. The flanges are sealed on to the ends of the insulatingglass cylinder (5) by means of Apiezon wax, the wax seals beingmade with the electrodes held in alignment by an assembly jig.

The condenser lens system and aligning mechanism are alsoshown in Fig. 4. The condenser lens (8) is an iron-shroudedsolenoid with an air-gap (9). The lens is mounted outside thevacuum, the fringe field from the air-gap extending into thevacuum through the walls of the brass tube (10) and acting onthe beam therein. A further brass tube (11) supports the beam-defining aperture (12) in the centre of the condenser tube. Theaperture limits the maximum value of the convergence angle ofthe beam to 0-004 radian when the condenser lens is focused.Reduction of this angle is effected by defocusing the lens. Thelens slides on the flat upper surface of the block (13), and iscentred by the adjusting screw (14) and a similar one mountedperpendicular to it. Return motion is obtained by springs (15).Two perpendicular pairs of adjusting screws (16) are providedto move the gun assembly over the spherical surface (17), tiltingit about the centre of the sphere which lies in the plane of theobject. This adjustment allows the direction of the illuminatingbeam to be aligned to the axis of the objective lens withoutaffecting the position of the beam at the object.* Similarly twofurther perpendicular pairs of screws (18) move the whole gunlaterally to align the illumination to the objective field of view.Vacuum connection between the electron gun assembly and themain tube is made by a flexible metallic bellows (19).

The entire electron gun, together with the terminating bushingfor the 50-kV supply cable from the d.c. power unit, is mountedwithin an earthed protecting cover. One half of the cover canbe swung back on hinges to gain access to the interior. Asimple automatic earthing device operates as the hinged halfopens.

(3.3) The Object StageThe object stage of the electron microscope holds the specimen

rigidly at the correct height and position relative to the objectivelens, and can be moved laterally to explore different fields in theobject.

The supporting membrane on which the object is mounted isdiscussed in more detail in Section 8. It is sufficient to say nowthat it usually consists of a very thin organic film supportedacross a small aperture or a series of apertures in the form of ametal grid.

The object support is held in a cap on the end of a flangedtubular capsule (Fig. 6) which has an accurately ground cone onits outer surface. This cone fits with great precision into aconical hole (1) in the stage plate (Fig. 7). The plate has threeball-bearings let into its underside, and these run on the groundupper surface of the objective lens. Three phosphor-bronze leafsprings (2, etc.) hold the balls in contact with the lens surface.The two perpendicular mechanical stage traversing drives eachconsist of a push rod (3), mounted to the plate at one end andengaging in a V-block (4) at the other. The V-block is attachedto a further push rod (5) leading out of the vacuum by a flexible

* The method of alignment is discussed in Section 6.

450 HAINE: THEfDESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE

/

Fig. 3.Cross-sectional elevation of type E.M.2 electron microscope.

HAESE: THE DESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE 451

Fig. 4.Cross-section" elevation of electron gun and condenser assembly.

metallic bellows (6). The outer end of this rod contacts a ball-ended adjusting screw in a bell crank (7) which bears at rightangles on the main driving lead screw (8). The lead screw has60 threads to the inch and is turned through a flexible couplingby the operating shaft (9) which terminates in a knurled knob (10)at a position convenient for the operator to grasp when viewingthe final image. Return motion of the stage is obtained bysprings which are suflBciently strong to ensure the stage followingthe lead-screw motion with less than 0-00001 in backlash. Twocompression springs, located in the flexible metal bellows, forcethe outer pair of push rods against the bell cranks, and a tensionspring, mounted to the stage plate half-way between two pushrods, pulls the stage plate against the V-blocks.

The advantage of this system over others tried, is that, whilethe necessary fine control is obtained, the object holder can stillbe rapidly moved over its full diameter.

(3.4) The Object Air-LockWhen removing the object from the

vacuum, an air-lock obviates the necessity ofletting the whole tube down to atmosphericpressure. The design of the air-lock followsin general principle that of Hillier,5 thoughit has the advantage of being designed as acomplete sub-assembly unit on a demount-able side-plate. Through the port exposedby the removal of this assembly, access isobtained to the mechanical stage and objec-tive lens. This feature has proved of parti-cular value in experimental work on theinstrument.

A cross-sectional diagram of the object-air-lock assembly is shown in Fig. 8. It con-sists essentially of the air-lock tube (1) joinedto the flange plate (2), the insertion block (3),the lifting lever (4), the trap (5), and thetrap-closing mechanism. The following de-scription of the operating cycle also explainsthe construction of the air-lock arrangement.In the running position, as illustrated in theFigure, the specimen holder rests in itsconical seating in the mechanical stage plate.Rotation of the air-lock control knob turnsa lead screw on the control shaft (7) andpushes a block (8), vacuum-sealed by metalbellows, away from the front flange. Thismotion, through a simple lever system,closes the air-lock trap (5) towards the endof the air-lock tube. Two pins, projectingfrom the door and pushing through spring-loaded push rods, tilt the lifting lever (4)about its pivot, thus lifting the stirrup (9).The stirrup engages with the flange on theobject holder and lifts it into the air-locktube so that the door can close over it. Avacuum seal is made by a rubber gasket.The door clamp (10) is now released, lettingair into the lock through the rubber-seatedleak valve (11), which is normally heldclosed by the clamp. As the pressure in thelock rises, the sealed-off metal bellows (12)compresses, moving a catch (13) over the

1 inch 1 inch

Fig. 5.Filament-setting jig. Fig. 6.The object holder.

452 HAINE: THE DESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE

Fig. 7.Cross-sectional elevation of object stage.

linch

Fig. 8.Cross-sectional elevation of object air-lock.end of the lever arm and holding the lever and object holder inthe raised position. The insertion block (3) is now removed fromthe tube and the object replaced. On returning the block andresetting the clamp the air-lock is pumped down to a roughvacuum through an auxiliary pipe line as described in Section 4.The air-lock trap is re-opened, dropping the specimen into theoperating position, the diffusion pump clearing away the residualair present in a few seconds after the trap opens.

(3.5) The Lens SystemIn Section 1 it was shown that under optimum conditions a

resolving power of the order of 10 A might be obtained in theelectron microscope. A more practical value lies about 50 A.The average eye, on the other hand, is able to resolve detail onlyas small as 0-1 mm in order of size, so that a magnification of0-1/(50 X 10-7) = 20 000 times is necessary to make the finestresolved detail in the instrument visible to the eye.

In practice it is desirable to limit the electron optical magnifica-

tion to about 10 000 times, as by so doing abrighter final image field is obtained for agiven beam current density at the object.Additional magnification can then be obtainedby photographic enlargement up to the limitset by the grain of the emulsion.

To obtain a magnification of 10 000 timesin a reasonable tube length a two-lens systemis necessary. Focal lengths of 0 6 cm for theobjective lens and 0 3 cm for the projectorwere found practicable, so that two stageseach 40 cm long were decided upon to give therequired magnification. Objective lenses ofshorter focal length were designed later,enabling higher magnifications to be usedwhen required.

The earlier lens design is illustrated in thesectional elevation of Fig. 8. A bobbin-wound energizing coil (14) is surrounded byan iron circuit except for a gap (15) in thecentral bore. The gap is bridged by a pair

of accurately fitting pole-pieces (16) which concentrates the mag-netic flux into a restricted region round the axis of the lens.A brass spacer is inserted between the pole-pieces, and this isdrilled with a central hole to allow for the passage of the elec-tron beam. The energizing coil in this design is sealed up in atoroidal brass box with rubber-sealed terminal bushings, thisbeing necessary to prevent the coil "gassing" in the high vacuum.

The improved design of lens which is used on later instrumentsis shown in Fig. 9. The coil is here wound directly on the iron

linch

Fig. 9.Cross-sectional elevation of improved lens.

circuit and fully impregnated with a solventless varnish of thetype discussed by W. Simpson. ^ After impregnation the out-side of the coil is "skimmed up" in a lathe and the outer cylin-drical portion of the iron circuit fitted. A further impregnationnow seals the coil completely within the iron so that only thevarnish in the joints and in the holes through which the leadsare brought out is exposed to the vacuum. The vapour pressureof the varnish is too low to affect the vacuum noticeably.

The design of the energizing coil and pole-pieces is based onthe necessity of obtaining a minimum focal length in the lenses

HAINE: THE DESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE 453Sufficient excitation is allowed almost to satu-rate the iron in the pole-piece tips, and thewinding space is determined by considerationof the power losses in the coil. The pole-pieces give an axially parallel field exceptnear the central bore where the lens actiontakes place. The shape of the pole-pieces isnot critical, but the axial symmetry is of vitalimportance and considerable care is takenduring construction to minimize asymmetries.Accurate grinding methods are resorted to forfinishing most surfaces, and careful checksare made for errors. Nevertheless, asym-metry of the objective lens field is the limitingfactor in performance in this and other instru-ments. Recent work by Hillier has shown that greatly improvedresults can be obtained by paying great attention to this questionof pole-piece symmetry.

Each lens is mounted in the microscope tube on three verticaladjusting screws which may be used for initial levelling. Theobjective lens is fixed in position by a spigot and clamped downby three tie-bars. The projector lens can be moved laterally byfour adjusting screws, working through metallic flexible bellows.Electrical connections from the lenses are brought through thewall of the tube by rubber-insulated bushings.

A double Mumetal screening tube (see Fig. 3) extends over thefull distance between the two lenses and screens the beam fromstray magnetic fields; a maximum a.c. field of a few millionthsof a gauss only being permissible in the region between the lenses.A further, similar, shield-extends below the projector lens.

A fluorescent screen lightly coated with willemite is positionedcentrally on top of the projector lens for observation of the inter-mediate image through an auxiliary viewing port. A shieldedhole in the Mumetal tube is arranged to allow free vision of thisscreen. In the centre of the screen a 1-mm diameter hole allowsthe central part of the image to pass through to the projectorstage for further magnification and projection on the finalfluorescent screen mounted on the camera.

A port is provided to gain access to the projector lens and thecentral part of the tube. The lens may, if necessary, be removedfrom the column through this port without disturbing the restof the tube.

(3.6) The CameraTo make a photographic record of the final image on the elec-

tron microscope it is usual, as in the electron diffraction cameraand demountable high-speed cathode-ray oscillograph, to place thephotographic plate within the vacuum and to allow the electronbeam image to impinge directly upon it. This method has theadvantages over photography of the light image on the fluorescentscreen, that exposure times are about ten times shorter, and thatthe resolution in the photographic image is not limited by thegrain in the fluorescent screen. The former is of importancesince short exposure times call for less stringent requirements inmechanical and electrical stability and other factors affecting thesteadiness in position and focus of the final image. At the sametime the limitation to resolution of the photographic image im-posed by the fluorescent screen would in general be greater thanthat due to other causes, and would therefore limit the instrumentperformance.

The camera in the present instrument incorporates an air-lockand is mounted in the cast brass viewing chamber supporting themain column and containing three pairs of binocular viewingports with eye-shields and light-tight shutters. The connectionto the diffusing pump is made at the back of the casting.

The camera body (Fig. 10) consists of a long rectangular brassbox (1) open at one end and having a square hole in the upper

Fig. 10.The photographic camera removed from tube.

side symmetrically placed under the fluorescent screen (2) whichis hinged to the side of the box.

The camera box is sweated into a flange (3) which is sealedinto an opening in the viewing chamber by a rubber gasket, thefluorescent screen then being situated symmetrically below thecentre of the main tube. The end of the box projects throughthe back of the casting into a vacuum-sealed cover. A hingeddoor (4) on the open or front end of the camera box seals theopening with a rubber gasket. The door is fitted with a conicalseated air-leak (5) and two clamps (6) to secure it in the closedposition. Longitudinal grooves are milled inside the walls ofthe box, and in these grooves runs the cassette carrier plate witha rack mounted on its underside and positioning pieces on itsupper side to locate and hold the photographic plate cassette.The plate is driven along the box by the rack and an engagingpinion wheel which is driven through two pairs of skew gearsby a knob on the side of the camera. The drive passes throughthe main camera flange by a gland-type vacuum seal as describedin Section 4. An indicator coupled to this drive shows theposition of the plate within the box.

When the carrier plate approaches the front of the camera box,two lugs, projecting from the rear end of it, engage a simple levermechanism which, on further movement of the drive, raises acircular plate with a rubber gasket, sealing off the square photo-graphing port and isolating the camera box from the main tube.The air-leak in the camera door now allows the operator to letthe camera down to atmospheric pressure and replace thephotographic cassette. The door is then closed, the camerapumped out to a rough vacuum, and the trap opened by a turnof the carrier-plate drive control to allow the diffusion pump toclear away the remaining air.

The light-tight cassette holding the 10 in x 2 in photographicplate consists of a chemically-blackened, flat rectangular boxwith sliding lid. At one end holes drilled in the lid allow freepumping of the air from the cassette. Baffles prevent light,entering these holes, from reaching the photographic plate.When the cassette is inserted in the camera two projecting lugson the lid engage in two short slots in the side of the camera box,and a flat projecting spring underneath the box engages in a slotin the carrier plate, which thus moves the cassette along thecamera box, the lid being held back by its lugs. During thereturn motion the lid slides back into the closed position. Thisarrangement is similar to that used on the Siemens and the R.C. A.instruments.

To make the exposure the fluorescent screen is raised onhinges by a driving spindle running, through a gland seal in themain camera flange, to a control knob. A third control operatesa shaft running through another gland seal in the main flangeand operates a masking device to enable the size of the photo-graphic exposure to be controlled. This masking device wasoriginally designed as a rotating brass masking plate with a series

454 HAINE: THE DESIGN AND CONSTRUCTION OF A NEW ELECTRON MICROSCOPE

of rectangular apertures which could be positioned in the beam.Later the camera was partly redesigned with a greatly simplifieddrive mechanism, and the mask was included as a split half ofthe fluorescent screen which could be held in the closed positionand thus allow an exposure of 2 in x 1 in to be made as analternative to the full 2 in x 2 in.

(4) THE VACUUM SYSTEMThe vacuum system follows the usual practice of continuously

pumped apparatus. An oil diffusion pump (Metrovac type O3B)is backed by a rotary pump (Metrovac type D.R.I). This com-bination has proved more than adequate to maintain the highdegree of vacuum necessary for the operation of the instrument.

Water supplyGeissler tube

I Pirani element

Diffusion pump

O5trap

Rotary pumpFig. 11.Schematic diagram of vacuum plumbing.

The vacuum pipework circuit is illustrated in Fig. 11. Threeinterlocked vacuum valves allow the rotary pump to be con-nected directly to the backing side of the diffusion pump or toeither of the air-locks when it requires pumping down to a"rough" vacuum. Rough pumping of the air-locks takes fromone to three minutes, and during this time the diffusion pumpfeeds into the reservoir formed by its rough-side pipe con-nections.

An entirely automatic system was developed and incorporatedin the first twelve instruments built. This arrangement hasfunctioned quite successfully but suffers inevitably from beingrather complicated in construction. Later instruments are beingfitted with a simpler manually-operated system which is onlysemi-automatic. The change is considered advantageous inview of the greater ease of assembly and maintenance.

The automatic system employs three solenoid-operated vacuumvalves electrically interlocked by a system of relays and interlockcontacts on the valve-operating mechanisms. The modifiedsystem uses three rubber-seated vacuum valves operated by camson a single control shaft. The cams are so arranged that onlyone valve can be opened at a time. Both systems employ anautomatic vacuum monitor which is similar to that described byMiller,*7 consisting of a Pirani element in a bridge circuit whichis arranged to trigger a thyratron at a preset vacuum. A P.O.-type relay in the anode of the thyratron controls the diffusionpump and the solenoid-operated vacuum valves in the firstsystem, while in the second system it controls the diffusion pumpand gives audible warning by a buzzer when it is necessary tooperate the manual valve control.

Vacuum joints throughout utilize rubber as a jointing medium.In the main tube most of the joints have a \ in diameter rubber-cord gasket which fits into a semicircular slot on one flange, thesecond flange being flat. The groove is arranged to allow suffi-cient compression of the rubber to give a seal when the two

flanges come together in metal-to-metal contact. This arrange-ment has the advantage of allowing accurate alignment of thevarious parts of the tube, as well as giving a greater overallrigidity.

To obtain linear movements inside the tube, flexible metalbellows are used as described in previous Sections. Rotarymovement is obtained by a simple modification of the Wilson

Fig. 12.Modified Wilson seal.

seal. This seal, illustrated in Fig. 12, consists of a gland packedwith a heavily-loaded synthetic-rubber washer. The outer edgeof the washer is clamped against the outside of the vacuum wallby a gland nut and washer, while the seal on the shaft is main-tained entirely by virtue of the hole punched in the rubber beingsmaller than the shaft and thus exerting a pinching pressure onit. This type of seal is very simple to manufacture and fit andhas proved quite reliable in operation. The frictional torquerequired to rotate the shaft is considerably less than in otherrotary vacuum seals previously used.

(5) ELECTRICAL CIRCUITS(S.I) Stability Requirements

The circuit design for the electron microscope is mainly de-termined by the high degree of stability required in the mainhigh-voltage and coil current supplies. It is one great advantageof the electrostatic electron microscope, using uni-potentiallenses, that this requirement is absent. However, the improvedresolution obtained on the magnetic type instrument wouldappear to justify the extra complication.

That a high degree of stabilization is necessary is readily seenon consideration of equation (3) giving the focal length of theobjective lens:

1