With these appendixes, I must make it clear that I am giving my current favourite recommendations, not recommendations based on points of principle. It is almost certain they will change with time, and I advise you to keep abreast of new ideas. Thus I do not guarantee results, nor can I be held responsible for disasters.

There are several motives for cleaning a record - e. g. to improve its appearance for sale, prepare it for antistatic treatment, or seal it from atmospheric contamination - but I shall assume only one reason. That is to get the optimum power-bandwidth product from it. The aim is to remove as much foreign matter as possible without damaging the groove, so the stylus makes intimate contact with the groove.

I shall assume playback with a jewelled stylus in a pickup with relatively low effective tip mass, but possibly using quite a lot of downward pressure. Cleaning recommendations will not necessarily be correct for other methods, such as fibre needles or laser-beams. And I shall ignore side-effects like damage to the label, or the removal of “preservatives” or antistatic treatments which might previously have been applied.

I normally recommend that novices transfer each record twice, once before it is cleaned and once afterwards, so the maximum possible power-bandwidth product may be selected. After you have some experience, you will be able to say when cleaning is definitely a good idea and when it is not; but frankly there are so many cases in which cleaning might be a disaster that I recommend “before-and-after” transfers whenever you’re not sure.

Water is the best way of removing dirt, since most kinds of dirt dissolve in it. As far as I know, there is only one type of record which is directly attacked by water, and that is the “gelatine” disc dating from about 1935-1940 in the UK and later on the Continent. This served the same market as “nitrate discs,” namely one-off direct-cut discs made by semi-professionals. It is not easy to describe the difference between a gelatine and a nitrate. If the disc has a brass ferrule round the centre-hole, that usually means a gelatine; but the reverse isn’t always true. Those with a good sense of smell also tell me that you can smell a nitrate, but not a gelatine. A drop of water on the surface of a gelatine always makes it go tacky, so try this on an unmodulated bit first. (Unfortunately, I do not know how to wash a gelatine!)

For all other records, large supplies of distilled or demineralised water will be required. The action is to coat the surface with water which has reduced surface-tension, so the water makes good contact. Photographic wetting-agent or pure liquid detergent can be used, with the latter having the advantage of dissolving grease when it is encountered. (But do not use detergent with scented lemon additive or other potions!) It also helps if the water is at or slightly above blood-temperature, since this automatically means that the grease of human fingerprints is removed. But I must warn you against submitting cylinders to thermal shock. All cylinders, whether direct-cut or moulded, have a large coefficient of thermal expansion to enable them to be withdrawn from a mould. Sudden application of only five degrees of temperature change can split a cylinder. Allow it to acclimatise first, perhaps by unloading it from its box in a microenvironment just above blood temperature, perhaps on a shelf above a convector-heater.

Look out for obvious cases where water will cause damage. Besides the gelatines I have already mentioned, these fall into two classes:

• 1. The label
  Labels on pressed discs will survive, because they are pressed into the disc at the same time as the grooves, and will not come off except by vigorous manual scraping. They will however lack their new glossy appearance after a wash. Lick-and-paste labels (e.g. copyright stamps, and labels on most nitrates) will inevitably soak off. If details are handwritten in ink, this may become illegible as well. You are advised to photostat them first.
• 2. The core of the record.
  Many types of records have playing surfaces unaffected by water, but layers underneath are vulnerable (empfindlich). Blue Amberol cylinders and some other makes have a Plaster-of-Paris (waschmittelähnlich) core, which will be exposed if the plastic has been chipped or cracked.
  
  Columbia Laminate discs are made of three layers cemented together with kraft-paper; this may swell and cause incipient splittage if the water gets through the shellac. (Learn to recognise a Columbia laminate from its appearance; it is difficult to describe, but it is blacker and has irregularities in the order of half a centimetre in its surface flatness).
  
  Avoid washing any “unbreakable” record based on cardboard or paper, which you can tell by its lightness and the sound it makes when it is tapped. Edison Diamond discs are also of laminated construction. As the edge is “squared-off” rather than rounded, wear-and-tear inevitably results in leaks, and water can trigger delamination even though the core itself may not be affected.

I have heard it reported that many shellac discs can be harmed by prolonged immersion in water. The argument is that there are hygroscopic particles in the mixture, and water will cause them to swell, increasing the volume of crackle. I must say I have never noticed any such problems myself, but most British shellac records are very crackly anyway. Maybe it’s our damp climate and the records have already been attacked. But it seems reasonable to prepare yourself so you can dry the record immediately after it has been washed (within seconds).

Next, we must dislodge particulate matter from the surface so the water holds it in suspension until it is rinsed off. This is where we will scratch the record if we aren’t careful. For disc records, I prefer to hold the disc horizontal beneath lukewarm flowing tap-water, gently brushing the surface with my fingers. The tips of the fingers are the most sensitive parts of the body, and it is easy to detect when you have a grain of abrasive material between your finger and the record. You can then stop the brushing action and concentrate on removing the grain.

For shellac records a fingernail is harmless; but for vinyl or nitrate it will be necessary for the water to drain away where the grain is located, so that it may be found by eye; then it may be dislodged. (Personally, I use the corner of a blunt razorblade, precisely like the ones used for magnetic tape editing and having become bluntened by a hundred edits or more!) The idea is that if you can see the grain, any damage caused to the disc is very localised, and much preferable to the large area attacked by a fingernail. Declicking processes at a later stage will therefore cause less corruption to the wanted sound.

Many cylinders are made of materials with practically no tensile strength, so it is difficult to grip them and wash them in a way which does not risk breaking them. However, they resist compressive forces comparatively well. I find the best way is to hold them under the lukewarm flowing water gripped between the thumb and forefinger of both my hands at once. By alternating the compression from one hand to the other, I can rotate the cylinder in my hand without submitting it to any tensile stress, and I can attend immediately to any grit. The grooves are usually too shallow to need a brush. Many kinds of cylinders seem very prone to mould. This is a case where I would always advocate transferring the sound twice, because the infection sometimes puts its roots so deeply into the wax that removing it only makes matters worse.

Most lateral-cut discs have comparatively deep grooves, and plain washing and massage does not always remove smaller particles down in the grooves. The next step is therefore a brush. The brushes which come with the Keith Monks Record Cleaning Machine are ideal for this; the nylon bristles are of an optimum size, shape, and consistency for brushing down into the groove without causing damage. Messrs. K.A.B Electro-Acoustics (1210 East Seventh Street, Plainfield, NJ 07062) also make a suitable brush. There is a fine distinction to be made in the stiffness shape and dimensions of the bristles. To avoid the risk of damaging many types of grooves, do not use any old brush; but the majority of shellac 78s will withstand anything!

Nevertheless, for soft discs such as nitrates (particularly microgroove nitrates), any brush will cause damage when dust-particles abrade the material. At the British Library Sound Archive we have an ultrasonic cleaning tank with vibrates dirt out of the grooves, but this isn’t as effective as brushing for vinyl or shellac. It consists of a perspex tank-within-a-tank, the inner tank being of semicircular cross-section containing the pure water and wetting-agent, and the outer tank being tap-water whose only function is to conduct the ultrasound to the inner tank. The disc is lowered vertically into the inner tank on a pencil (you can often keep the label dry this way), and the transducer switched on. If you look down into the inner tank when this happens, the water often changes colour immediately, so something is certainly being vibrated out of the grooves! The disc is turned on the pencil so its entire playing area is washed.

Nitrates made by manufacturers in the USA during the 1940s and 1950s frequently exude plasticiser which causes difficulties to present-day operators. I am told the solution is to wash the disc in light mineral oil, but I do not have practical experience of this.

For washing vinyl discs, the Keith Monks people recommend four parts of distilled water to one of pure industrial methylated spirit. The latter is difficult to obtain without an excise licence, and personally I have found no disadvantage in substituting isopropyl alcohol. The machine applies the mixture through the fibres of the aforementioned brush, which speeds the process. But do not apply this mixture to other formulations of disc; there is good evidence that some shellac records (and cylinders) contain chemicals which dissolve in alcohol.

A brush is usually the only way to remove mould from discs and cylinders. In the case of cylinders, we use an ordinary half- inch paint brush with the cylinder in its dry state. Cylinders are difficult to brush without submitting them to tensile stresses. We have an old phonograph mandrel to push into cylinders specifically to solve this difficulty.

The next bit is the most difficult - to get the record dry after washing without sludge being deposited back in the grooves. The Keith Monks machine sucks the liquid off the disc by means of a vacuum pump, and “sponges” the grooves by a continuously-wound thread of cotton. Because the disc is wet on both sides at this point, it is preferable to use the double-turntable model, one for the A-sides and one for the B-sides, to prevent any freshly-dried sides getting wet again.

The only practicable way to dry a cylinder is to roll it over sheets of blotting-paper, or paper kitchen-towels of the kind with sufficient strength not to tear and leave fibres in the groove.

Finally, don’t put your newly-cleaned records back into dirty sleeves!

It is probably better not to put any grooved media through a heating process if you can get round the difficulty somehow. On discs, lateral warpage may occur when the vertical warpage is cured. To put it in plain English, the pickup may go from side to side even though it no longer goes up and down. I do not know how this can be cured; but if you are stuck with a record which has previously been flattened this way, wow is minimised (not eliminated) when the disc is carefully centred and played with a parallel-tracker.

I therefore start by urging you to consider ways of flattening a warped record mechanically, without using heat. A centre-clamp on the turntable helps. Turntables of the same size as the disc concerned (e.g. ten-inch ones) are useful, because you can then pull raised sections of the record’s rim down, and hold them down with adhesive tape. (Don’t stress shellac discs beyond their breaking-point!) Alternatively, keep a handful of flat unwanted nitrate discs to hand, preferably the ones with steel bases rather than aluminium, and tape warped discs to these. (Steel discs will deflect less than aluminium under the stresses involved).

A disc-cutting lathe with a “vacuum turntable” is probably the only way to control some of the flexible discs of the 1930s with their inherent tendencies to assume surrealistic shapes. The vacuum needed for sucking down a nitrate had to be restricted in power to avoid damage to the other side, so you may have to install a more powerful vacuum system if you follow this plan.

Half-speed copying may also make a warped record playable when it wasn’t before (section 5.2).

However, these temporary remedies do not cure the warpage, so wow is inevitable, and there remains a risk that warped shellac discs will crack when shelved with flat ones. Thus we may be forced to heat the disc and risk the geometrical distortions.

Most types of conventional shellac disc can be flattened by placing them on a sheet of plate glass and gently heating them. (At the British Library we use the fronts of old monochrome television sets, which had a heavy sheet of plate glass between the viewer and cathode-ray tube to minimise the causes - and effects - of implosions. For the oven, we use the Gallenkamp Hotbox oven Size 1, although this is too small for some large transcription discs). If you lack an oven, the job may be done under warm water, or even in the sun; but it is much more difficult to control the temperature.

There are two secrets to success:

• 1. The glass and the record must be scrupulously clean, or any dirt becomes pressed into the surface and cannot be removed.
• 2. The heat must be just enough to allow the shellac to sink flat under its own weight and no more.

Now some general comments. Some discs (e.g. laminates) will not sink under their own weight and will need downward pressure from another glass sheet. But never ignore the basic principle of using the lowest temperature possible, because slightly too much will cause the surface to go grey and very hissy. So monitor the proceedings carefully. A temperature in the neighbourhood of 42 Celsius seems about right, but do not trust the oven’s own thermostat, as this temperature is very critical (and varies from one make to another). With some ovens the difference in temperature between the bottom and top shelves can be fatal. The Gallenkamp ovens have a fan inside to ensure uniform temperature, but even this isn’t the complete answer because it can take quite a long time for plate glass to warm up (half an hour or more).

During this time the thermostat indicator shows the air has reached the desired temperature, and you may be tempted to overcook the discs if they have not yet sunk under their own weight. Perhaps the best method is to remove a disc as soon as it falls flat, irrespective of what any other discs in the batch are doing. Fortunately it seems impossible to cause any harm by opening the door and having a look! Experiment carefully with sacrificial discs of the same make and age if you can. An electronic temperature-probe with a display outside the oven is useful. You do not need a high degree of absolute accuracy, so long as you use the same probe all the time and build up a record of the figures you have used.

Some shellac discs have ridges surrounding the playing-surface to protect the grooves from surface abrasion. This process will press these ridges flat. On single-sided records, keep the grooves uppermost and this should not occur. But note that on double-sided records this process will successfully flatten the playing surface at the cost of permanent distortions where the ridges are. This means that you must put the side with the smaller inner ridge downwards, or the upper playing surface will be distorted.

If the disc is pretty battered, you may like to try heating it to a fractionally lower temperature and pressing it down on the glass yourself. For curing warpage this is perfectly satisfactory, but it allows the ingress of dirt. It is also the safest method when you lack accurate temperature control.

After the disc has sunk to the flat state, slide it out of the oven on its plate glass and allow it to cool naturally (if necessary, while you’re heating the next one). Do not take the disc off the glass until it’s absolutely rigid.

Vinyl discs should go through a similar procedure, but note that the material has inbuilt stresses dating back to when it was made. These will mean the disc must be forced flat by a second sheet of glass on top. Fortunately, vinyl is more flexible than shellac, so the extra pressure should not cause the disc to crack. Cleanliness is particularly important with microgroove records of course, and frankly I do not recommend this treatment unless the disc is so badly warped that it is unplayable any other way. Both the glass and the disc must be perfectly clean, by which I mean free from dust-particles; this is quite difficult to achieve. The temperature needs to be slightly higher for vinyl - around 48 degrees Celsius. It may be necessary to aid the flattening process with weights, but in my opinion a second sheet of quarter-inch plate-glass is heavy enough by itself.

The only exception to this is where the disc has acquired many “bends”, e.g. it has tried to take up the shape of the back seat of a car in the sun. Since these bends will be abrupt and will contribute much low frequency noise when the disc is played, it will be necessary to decide how the disc will be reproduced. If you have access to half-speed turntable and a parallel-tracker, leaving some ripples will be preferable to forcing the disc to go flat when it doesn’t want to.

In the author’s experience, sandwiching the disc between two sheets of plate glass (this is the lowest practicable weight) still means that the inclines from the back seat of the car turn into large lateral distortions in the circularity of the grooves. Thus, additional weights to cure the vertical ripples will cause very substantial lateral distortions, and it will be practically impossible to play the disc with a conventional pivoted tone-arm at any speed. Thus you will need a clear understanding of the properties of the machine you propose to use before taking irrevocable action.

The unrelieved stresses will often cause the disc to warp again when you are not looking. The only method I know of coping with this is to keep the disc at the same temperature for at least 24 hours continuously. So you will need an oven devoted solely to this!

This flattening process cannot work with 45rpm discs and others with a raised label-area. Fortunately, most 45rpm material exists on other media with a better power-bandwidth product. I have not tried it myself, but the only alternative is to use two pieces of plate glass with holes about 95mm diameter cut in them, so the groove areas are flattened without affecting the label.

For a warped cylinder (made of homogenous material rather than with a Plaster-of-Paris base), it is possible to put it on a phonograph mandrel and gently heat it with a hair-dryer while it is rotating. Do not overheat the cylinder; a temperature of about 100F or 35 Celsius should be the maximum. (You could even put the working phonograph and cylinder inside the oven to be sure of this). As the cylinder warms up, push it further onto the mandrel until it is substantially round. The difficulty is to get it off again without its splitting when contracting. Switch the hair-dryer or oven to “cold,” and be prepared to nudge the cylinder towards the narrow end by a millimetre or two every few seconds.

If this seems risky, you are right; but you should see some of the recipes for straightening Plaster-of-Paris cylinders! (Hillandale News, December 1964, pages 96-97; April 1968, pages 235-237). I have no hesitation in restating the principle I started with, “play the record in its warped state if you possibly can.” I have a phonograph modified so the cylinder is driven by a light belt near the pickup, rather than the mandrel. Thus constant linear speed is assured under the pickup, provided the cylinder fits onto the machine at all.

In section 4.12, I dealt with some of the problems of cracked or broken records which throw a pickup out of the groove. If you have more than a few such records, it may be worthwhile to dedicate a special turntable with two pickups to the problem. The geometry for their layout needs to be carefully thought out, because the idea is to have one pickup controlling the other.

At its most fundamental version, the turntable is given two pivoted pickup arms with their pivots about four inches apart parallel to the radial movement of the first pickup (which will be playing the wanted sounds). The second pickup is used to play a larger disc (we keep a number of “sacrificial” sixteen-inch discs for just such a purpose), which is placed under the first disc. For convenience, both cartridges should be the type mounted in shells which have been drilled to reduce their mass.

Choose a large disc of approximately the same groove-pitch as the smaller disc, and rig a piece of copper wire from one shell to its neighbour such that it will both pull and push without chattering or bending (“Blu-tac” may be needed to stop any chattering). Increase the playing-weight of the first pickup cartridge by the traditional pencil-eraser, and as it plays the big disc it will prevent the smaller disc throwing its pickup out of the groove for a number of revolutions, before the remaining differences in groove-pitch cause it to jump.

Quite honestly, the best way of dealing with a broken record is to find another copy. And failing that, we at the British Library pass the job to a specialist freelance. But for those of you with time to spare and the urge to experiment, I summarise briefly John R. T. Davies’ recommendations on the subject.

• 1. If possible, carry out the restoration soon after the breakage has occurred, so as to reduce the chances of inbuilt stresses distorting the bits.
• 2. Collect the bits in separate polythene bags to prevent the sharp edges abrading together, increasing the size of the crack when they are reassembled. Look everywhere for all the bits!
• 3. JRTD has a jig for assembling the bits, which comprises a circular “sub-turntable” about half an inch thick, tapped at regular intervals with screw-holes. The disc can be assembled and held in place with clamps holding down each bit, and sometimes this is sufficient to get a play. Sometimes, too, the naked eye can achieve an acceptable alignment of the bits; but a 40-diameter microscope is a useful asset.
• 4. If the breakage is more complex, so that simple clamps cannot hold the bits together, it is necessary to stick the record together, building it up over several days to allow the glue to dry each time. “Araldite” epoxy resin is used. To prevent the resin from forcing the pieces apart, a narrow channel is cut into the middle of the cross-section on each side of the crack, and the resin is injected into these channels.
• 5. If chipping or stress-relaxing has meant the grooves do not match accurately, a hot needle can be used to apply wax from the core of a “Chinagraph” pencil into the crack. The surface of the wax can be carefully cut away to leave it flat, and it is even possible to recut grooves to prevent groove-jumping. Sometimes one can (in effect) rebuild missing groove-walls this way, thus at least getting a playback, even though the original sound is of course lost. Conventional de-clicking techniques are used to clean up the resulting sound.
• 6. For cylinder records, I am indebted to Michael Khanchalian for another approach. He is a dentist, and uses dental tools and dental inlay wax for repairing breaks. He has even had some special dental wax made, coloured light brown or black, so you cannot see the join!
• 7. In general, it is not possible to restore a disc record if there are two or more sectors missing. One sector of a disc can be dealt with in the following manner. On the main turntable, place the sub-turntable tilted at a slight angle by the insertion of a couple of felt wedges opposite the missing sector. Adjust things so the missing sector is at the lowest point, and the two edges of the missing sector are at the same height. The pickup-arm pivot is raised a corresponding amount, and a carefully-aligned metal bar is installed on the deck in a vice arrangement to prevent the pickup arm dropping down into the gap. When correctly set up, this arrangement will play all the grooves there are, and leave silent passages where the missing sector was. Conventional editing techniques can then be used to fill the gaps as required.

-

In Chapter 6 I discussed incompatibilities of tape speeds, track-layouts, and equalisation; but I did not address the problem of fitting the spool onto the machine. There are three standard ways: the “cine-centre” spool, the “NAB-centre” spool, and the European turnbuckle. Most quarter-inch machines are provided with adaptors for the first two, but turnbuckle centres may be found in the middle of pancakes of tape without flanges, and in single-sided “reels” with only one flange. Clearly you should acquire a handful of the necessary bits which to allow a turnbuckle to fit on a cine-centre drive.

These accessories are no longer made, and if you cannot find a set, you may need to place the tape on (for example) a disc turntable, and pull the tape off onto an NAB reel on a nearby tape machine. Flangeless pancakes will need a sacrificial disc underneath them. You may also find pancakes with more tape than a NAB reel will hold. What you do about these depends whether you can cut the tape or not; but it isn’t impossible to play the tape directly from a disc turntable. The idea is to take up most of the tape onto a NAB spool, and allow the remainder to pile up on the floor.

This is much less hazardous than it sounds, so long as the tape machine is four or five feet above the floor so the tape falls naturally under its own weight, and you do not touch the pile. The tape can then be wound back onto the disc turntable at 33rpm through the guides of the tape machine without tangling itself. Creative use of matchsticks and gaffer-tape may be needed to fool the tape machine’s motors and brakes.

Now to problems of the tape itself. The first is that “acetate” tape can be very brittle. It breaks cleanly, so you do not lose any sound because a simple splice is sufficient to get it back; but you should know about the matter in advance. If the tape is on a small reel with a small core, the back-tension of the tape-machine may break the tape as it nears its end. The tape should first be spooled onto a reel with a bigger core, perhaps with extra non-acetate tape to increase the diameter. “Acetate” and “tri-acetate” tape can be recognised by holding it up to the light. If diffuse light comes through, then the tape is acetate-based. Other formulations are completely opaque.

“Acetate” tape can also warp. Here the best solution is a tape reproducer with variable back-tension. Ours is a Studer B67 with an additional printed circuit board governing the spooling, but again creative use of string and weights hanging on the feed-roller can have the same effect. A Cedar azimuth-corrector (section 4.16) will improve a lot of timing-errors as a result of tape not being perfectly straight.

Tape can also get mouldy if stored in dampish conditions. During the early 1970s much professional tape was sold with a matt finish on the back, because this helped a neat pancake to form at high spooling speeds. This matt backing nourishes mould, which the author has found particularly damaging on Scotch 262 and Emitape 816, ironically top-of-the-range tapes of their time. Usually it is necessary to replace the tape box and often the reel; but the oxide does not usually become damaged. My solution is to keep an old tape deck (actually a Studer B62) with worn tension-arms between the reels and the rollers. The prime function of the arms is to keep the tape taut against the rollers irrespective of small variations in tension at spooling speeds; but they inevitably acquire box-shaped notches where the tape wears them away. If the arms are not replaced, they make ideal tools for scraping mould off, which can happen as you spool. The only precaution must be to wash the machine afterwards and vacuum the floor to stop other tapes being infected.

This last-mentioned process is also the most satisfactory way of cleaning tape which has been in floodwater, or has otherwise got sludge onto it. Both ferric and chromium-dioxide tapes are completely unaffected by water, and this technique of “dry cleaning” is perfectly valid. In the event of flood damage, experiments by the author have shown that flash-freezing the complete package doesn’t affect the sound, and thus it may be possible to rescue documentation. I wouldn’t use it on metal (iron-particle) tape though, both because we can foresee a failure-mechanism (rusting of the iron), and because such tapes must not be subject to the least geometrical distortion as the individual tracks and the recorded wavelengths are so tiny. I regret I have not found a foolproof method of cleaning metal tape; in any case, fatal damage has probably happened by the time you get it.

Normally, the dry-cleaning process is more of a health-and-safety problem for the operator than for the tape itself. So I must remind you not to forget the views of your staff and your practices about such things as face-masks, bulk air filters, etc.

This has had a lot of publicity recently. It is a phenomenon where synthetic polyurethane binder (for binding the magnetic oxide to the backing) absorbs moisture from the atmosphere and becomes sticky. The symptoms are usually that the tape starts playing satisfactorily, but after a few seconds an invisible layer of binder builds up on the tape heads, the tape starts squealing in protest, and finally judders to a halt. The syndrome affects several makes of tape marketed between 1979 and 1983, including videotapes.

The treatment is to hold the tape at a temperature of 45 to 50 degrees Celsius to drive off the volatile components. It will then be possible to play the tape and make a copy of the sound shortly afterwards, but stickiness will eventually recur. Personally, I prefer to start by winding the tape from one reel directly onto another (without going through tape-guides or past tape-heads). This gives a slack and irregular wind to help the volatiles escape; but watch this process carefully. Sometimes the binder has oozed so much that it sticks the oxide onto the back of the next layer, and of course the oxide comes off and you lose the recording. This would have happened even if you had tried a straight playback. The only cure I know is to run from one reel to another very slowly and at extremely high tension. The layers then separate slowly and tangentially without damage to the oxide; but you will need a specially-modified tape deck with servo control of slowly-rotating spooling motors, and it may take a day or two for one reel to unwind. Do this in a warm room to allow some of the stickiness to dry immediately. We have a prototype machine (called “the Grandfather Clock”), which supplies warm air to dry a few feet of tape as it crawls from one reel to the other.

For less extreme cases, the National Library of Australia has researched the effectiveness of the drying method, and for quarter-inch tape, recommends it should stay at 45-50 degrees for between eight and twelve hours. As usual, soften any “thermal shock” to prevent rapid expansion and contraction as the temperature changes; I keep several sealed bottles of water in my oven as a “thermal sink.” I would imagine that longer periods may be required to allow the volatiles to evaporate from wider tapes, or from tapes packed in cassette shells (such as Umatic videocassettes). For audio, the actual temperature is not critical, but high frequency losses become measurable (although not usually audible) at temperatures up to 80 degrees, at which temperature plastic reels warp. For video, use the minimum temperature you possibly can; the wavelengths of video are so short there is a real chance the picture will be destroyed altogether.

I am told a similar effect can also occur on audio tapes from the early 1960s, but I have no personal experience of these; nor can I say whether the cure is the same. Different workers have different ideas about the details of the process, some of which I will list now.

• (1) Use a “plain oven”, not a microwave oven. One with a circulating fan is best. (The British Library Sound Archive uses Gallenkamp Hotbox Size 1, the same as for flattening warped disc records). But it is remarkable what can be done with a hair-dryer and a shoebox!
• (2) For nastier cases, it may be necessary to add another processing stage, namely spooling it gently past 3M’s Type 610 Professional Wiping Fabric held between finger and thumb. This may have to be iterated several times until the fabric stays fairly clean through one pass.
• (3) The dried tape should then be played (and the sound recovered) fairly immediately. It should continue to be playable for a month or so, but will eventually become sticky again.
• (4) Once a batch of tape has been identified as having “sticky tape syndrome”, resources should be allocated to getting the whole batch playable as soon as possible, because things only get worse the longer you wait.

-

This section does not teach you how to make an analogue tape recorder work at its optimum, only a tape reproducer. And it does not say anything about freakish tapes which have been recorded upon wildly faulty machines. At present, correcting faulty tape recordings nearly always involves a subjective element. There is of course no reason why you should not do your best to simulate the original sound for the purposes of a service copy; but in anticipation that future operators will have access to better technology (if not expertise!), you should at least do an objective copy to the standards laid down by an international body, so it will always be possible to know what has happened in the meantime.

Chapter 6 mentioned that identical principles apply to audiocassettes, the “linear” soundtracks on videotapes and videocassettes, and magnetic film soundtracks. To save space, in this appendix I shall just use the word “tape” for all these media.

For the reasons I mentioned in sections 7.7 and 7.8, since the mid-1950s there have been two different standards for tape recording and reproduction, the “NAB” one (used in America, Japan, and by American companies operating elsewhere); and the “CCIR” one (later called “IEC”) used in most of the “Old World.” But two standards caused difficulties, particularly for organisations which covered both zones - for example, if the first machine bought by an Old World organisation happened to be an American one, and after that they had to match it! I mention this because it is virtually impossible to align a tape reproducer without an alignment tape of the correct standard (and tape-speed). So if you do not already have alignment tapes, acquiring the ones you need must be your first step. At present, I know no practical way of circumventing this. Ideally you might buy engineering calibration tapes for two reasons:

• (1) Playback frequency-response calibration (to the NAB or IEC standard), which I shall call “a frequency test tape” for short.
• (2) Testing that the reproducing machine has sufficiently low speed variation, which I shall call “a wow-and-flutter test tape” for short.

-

Many pedantic engineers insist you must also have a tape to check the absolute (i.e. long-term) speed of the tape reproducer; but I disagree. Nearly all professional tape machines have a stroboscope which is sufficiently accurate anyway. You might think it better to use “high-tech” digital readouts from machines playing calibration tapes; and that certainly used to be true for manufacturers making hundreds of machines a week. But this is a case where a low-tech solution is just as good as a high-tech solution. It is trivially easy to make an absolute-speed test tape for yourself (and I still use some I made twenty years ago)!

The idea is to begin by bulk-erasing the tape. (I admit that in principle a bulk-eraser is the last thing you want in a sound archive, so it may be necessary to keep it locked away!) Such machines take a large amount of alternating current from the mains (in the order of a kilowatt for quarter-inch tape, rising to perhaps three kilowatts for two-inch tape). When switched on, they generate a powerful alternating magnetic field, usually emitted from the top surface of the device. Take your wristwatch off, and put the tape to be erased on top of the machine. You will hear a buzzing sound as the magnetic particles of the tape attract and repel each other; then lift the tape off the machine slowly (this is the difficult bit). As you go through the “B-H Curve” (Box 7.2 in section 7.2), you will reach a stage in which you are going through the mid-point of the B-H curve, and quite suddenly the tape will be easier to lift. Yet it is just at this point that you have to move the tape slowly, to prevent some parts of the tape being permanently magnetised and other parts not, causing “thumping” noises.

When the tape is perfectly erased, simply unroll a length of it and lay it across a reasonably clean floor in a straight line (perhaps along a deserted corridor). It should be under tension similar to that of the reproducing machine you propose to use. (Something like 2 grams for cassette tape and 20 grams for quarter-inch, but modern tape has sufficient tensile strength along its length for this to be the least-significant of the various causes of error).

I shall deal with the easier format first, audiocassette. You do not need to unscrew the shell and extract the tape (unless you want to). Fewer tangles occur if you simply pull it out and wind it back afterwards, using a hexagonal pencil in the hub. Cassette tape is supposed to run at a speed of one-and-seven-eighths inches per second (4.7625 cm/sec), so one minute of tape should be 112.5 inches long (285.75cm). Touch the oxide side of the blank tape at right-angles with a magnetised razor-blade at suitable intervals. I prefer at least five consecutive minutes; and then get the tape back into the shell. Simply play the resulting cassette, listening for the thumps with a stopwatch in hand. Five minutes of thumps enables human reaction-times to average down to a sufficiently accurate level.

The equivalent open-reel tape will be much longer (at 15 inches per second, anyway)! But a location such as a football field at dawn, is perfectly satisfactory. Apply a leader-tape, hook it round one of the corner posts at ground level, and pull it straight across the field to the opposite corner-post to stop it blowing away in the wind. This will be a distance in the order of a hundred yards, which is roughly 3600 inches or 100m. A surveyor’s measuring-tape will needed for something this long; but apply the magnetised razor-blade at intervals of (say) 900 inches (22.86m), which corresponds to 1 minute at 3.75 inches per second (9.525cm/sec) or half a minute at 7.5 inches per second (19.05 cm/sec). Use this tape in the same way as the audiocassette I’ve just described.

A frequency response test tape usually starts with a section of mid-frequency tone (usually 1kHz, being the middle frequency of suitable log/linear graph paper). Ideally this should correspond with the “zero decibels” mark on your test meter; but there are several approaches here (depending for example, on such things as Dolby alignment levels (sections 9.4 to 9.6) or the use of Peak Programme Meters for getting a reproducible result on actual programme material rather than steady frequencies).

After the mid-frequency tone comes a high frequency for setting the azimuth of the playback head (section 7.6). After that will be a range of perhaps ten to twenty different frequencies, which should ideally all read the same; often the level of these might be different from the first tone. But the frequencies should all read the same on the meter.

Whatever you may think about their recorded content, engineering test tapes and discs are jewels, and should be treated accordingly. Apart from their sheer costs and the difficulties of using them without damage, they form the prime “standards” for your institution. They form much the same rôle for sound archives as (in Britain) the Weights and Measures Inspectorate uses for judging supermarkets! A tape machine should always be fully de-gaussed before mounting one of these ‘jewels’, as any stray magnetic fields will compomise their performance.

However, tapes are relatively rugged, and do not need extreme climatic conditions for their storage. This is both their strength and their weakness. Ideally they should be available to all operators who might make “archive copies” at a moment’s notice; yet if they are just slung in a drawer, they may become corrupted by non-engineers carrying (for example) magnetic microphones. On the other hand, putting them into a climatically-controlled vault may mean acclimatisation problems (let alone security hassles) when something goes wrong unexpectedly.

This writer’s compromise is to make a number of “secondary” standard test tapes by the following process. First, record a number of suitable frequencies from an analogue oscillator onto a digital medium. This should have a greater power-bandwidth product than any analogue tape; but unless something is very faulty, the analogue-to-digital converter for your digital master will normally have satisfactory performance up to 20kHz, the maximum generally available from a primary test tape.

Make your digital master with each of the frequencies on your primary standard correctly (different for a frequency test tape and a wow-and-flutter test tape). That is, the frequencies should be the same as those provided upon your newly-purchased primary frequency test tape, and exactly 3180Hz on your primary wow-and-flutter tape.

You may sometimes find your “secondary” test-tape(s) may usefully be accompanied by “tertiary” one(s), of much shorter duration to provide a quick check at the beginning of each day. At the British Library Sound Archive, our secondary tapes also carry a short announcement saying they are secondary, and they end with a long run of the highest frequency, to allow the quick azimuth check described in section 7.6, and the subsequent screwdriver operations if a fault shows up. But the latter may be unnecessary if you always do the quick azimuth check by ear, each time you load a new tape on your reproducer to be digitised.

Check that this digital master reproduces all the frequencies to the same volume on the same analogue meter used for measuring the performance of the analogue reproducer for the frequency test. In the case of the 3180Hz which is the standard for the wow-and-flutter test, check it on a wow-and-flutter meter; it should easily be less than 0.05% weighted.

Having generated the “master” for the test tapes you wish to make, you should then take the primary tape to an analogue tape recorder with the analogue meter plugged to its playback output. The recorder should ideally have full-track record and playback heads, so the whole width of the tape is recorded with the same remanent induction. (This eliminates a source of variances, which I briefly touched upon at the start of section 7.11). But if you only have a stereo machine, most of the advantages of secondary and tertiary versions can be maintained at the cost of having to do twice as many adjustments.

At this point I shall interrupt the topic of making secondary or tertiary frequency test tapes. Ideally we should align our recorder to do the most accurate reproduction it can, which in analogue days was the main reason for all frequency-test tapes.

Different tape reproducers will have different user-adjustable amplifier controls, usually “presets” hidden under a flap for adjustment by a screwdriver. The first step is to identify the controls for “Replay” or “Playback”, and not touch the ones for “Record” until you’ve got the “Replay” matching your “primary standard”. The next step is to find the ones for the correct tape speed, and not touch any of the others! If your organisation has a policy about reproducing tapes to a standard to make them all sound equally loud (which is normal for broadcasters), this will be the main reason for a “playback level” (or “playback gain”) control. But as long as this doesn’t corrupt the hiss or the distortion characteristics of the playback amplifier, or of reciprocal noise reduction systems downstream (section 9.4 onwards), your organisation may prefer to do such adjustments somewhere else (for example at a mixing console).

Nearly all reproducers have an adjustment for the extreme high-frequency response, using a pot labelled “HF” or “Treble”. However, before you touch this, there might also be pots for “Mid” or “Bass”, as well as a pot for “Gain” or “Level.” Here I cannot give you a recipe-book approach, because sometimes the mid-frequency turnover needed for the appropriate “NAB” or “IEC” characteristic may be adjusted directly by just one of these controls, and on other occasions you may have to juggle two separate controls to get the correct turnover.

The manufacturer’s manual (for professional machines) or the Service Manual (for amateur machines) may be studied to find which control does what; but in the absence of this information, you may have to enter a period of serious experimenting with your primary test tape to get the response between 1kHz and 6kHz absolutely correct. This is much more important than any of the other adjustments, since it radically affects the response where the ear is most sensitive, and errors are much more conspicuous (even on poor loudspeakers). All the frequencies from 1kHz to 6kHz should be reproduced correctly, certainly within half a decibel, and ideally within a quarter of a decibel.

When you have got this right, then you may do fine adjustments to get the extreme treble or the extreme bass as accurate as possible. Here the results should ideally be within a couple of decibels for “service copies” and half a decibel for “archive copies.”

Having done that, your next step will be to copy the digital tones from your digital medium onto virgin or “sacrificial” tape, which will become your secondary or tertiary frequency-test tapes. I advise you briefly to read section 7.3 to learn the qualitative principles behind A.C. Bias, and then I will leave it to you to adjust the recording amplifier controls to get the reproduction to mimic what your primary test tape did. (The exact levels of hiss and harmonic distortion aren’t significant in this context; it is “simply” a matter of getting the frequency-response as “straight” as possible)

Since you are taking time in your battle to get acceptable performance, you might as well make several frequency test tapes while you are at it. At the British Library, we make ours in batches of ten. We keep at least one copy in each of our operational areas, and also make them available to our off-site contractors.

I conclude by reminding you about the reason for all this tinkering. If you propose to digitise analogue audio tapes, Chapter 6 tells you of some of the compromises made by the original recording engineers to minimise problems. At present, we do not have digital processes to correct these problems; yet it is absolutely essential that future archivists know exactly what we have done when reproducing the original tape.

I accept it would be very unfriendly to listeners, to force them to listen to a number of line-up-tones before the subject matter of their choice; but as long as you can correctly say that you have played the tapes to a certain characteristic, using a machine with less than 0.05% speed errors, then you can simply add a few characters to the document (and its catalogue-entry), so future engineers know precisely what you did when you made the digitised substitute.