Legacy Optical Recordings
Then and Now
by Ralph Sargent

Some Background

The success or failure of an analog optical recording system is totally dependent on the linearity of the medium upon which it is recorded. Of course a number of other factors will come into play when judging these systems, but all things considered, linearity must come first. Linearity is achieved in optical audio records by distortion and sensitometric analysis and adjustment of the interaction of the various electronic and photographic steps involved.

Historically speaking, there have been three principal ways of recording optical tracks for motion pictures and each is inextricably linked to sensitometry for freedom from distortion and efficient utilization of the chosen method. Two of these three methods are distinctly different ways of producing Variable Density recordings and the third is Variable Area recording. This article will discuss each of these methods and then present experiential techniques for producing modern prints for restoration applications and archival storage.

The Case Variable Density recording system, "Toe Recording"

If you take a trip to Auburn, New York and visit the Case Research Laboratory you'll have a chance to examine a highly modified Bell & Howell 2709 with an Aeolight recording tube, positioned to expose a variable density optical soundtrack on film as it traveled on the camera's main film sprocket having already passed the picture exposure gate. Both the modifications to the camera and the Aeolight were the brainchildren of Theodore "Ted" Case.

Aeolight was a recording device which depended upon a modulated, ionized gas to produce various intensities of light which were an analog of the original sound. However, an Aeolight was not a very powerful instrument. It could not produce enough light to place optical recordings on the portion of the characteristic curve which eventually became "home plate" for variable density recordings made with Electrical Research Products, Incorporated's (ERPI's) Western Electric recording equipment. Case and his assistant, Earl Sponable, later Chief Engineer for 20th Century-Fox, envisioned single system recording to be their method of choice. ("Single System" utilizes a camera which records both picture and sound simultaneously on the same film.) Case's camera became the parent of the newsreel camera of the '20s through to the '60s and it truly was the great-granddaddy of the Auricon camera, the 16mm workhorse of early television news.

You might think that Case and Sponable's development is an odd way to begin an article on sound recording systems, but figure this: millions of feet of film were shot and recorded with "Toe Recordings." Remember the early Fox sound feature, "Sunrise"? Well, that's a toe recording made with an Aeolight. Remember all those Fox Movietone Newsreels from the '20s and early '30s? Aeolight! The fact is that Fox-Case single system style cameras went around the world to record events that otherwise would be silent clips today. For studio production however, single system cameras quickly gave way to double system equipment, with separated but interlocked recorders and cameras.

Exactly what is a "Toe Recording?"

Pay attention to the classic H&D curve above. Its purpose is to tell us how a piece of film-in this case an idealized black and white print stock-reacts to various levels of exposure to light.

The right vertical side of the chart, from bottom to top, represents ascending density. The bottom line, looking left to right, shows increasing amounts of exposure. The curved lowest part of the graph is referred to as the "toe," the straight mid-section is called the "straight-line" and the top curved portion is called the "shoulder."

(So you have a touchstone for what this means, keep in mind that a normal print "gamma" for black and white is 2.40. This number is derived by extending a line down from the "straight-line" portion of the curve (the dashed line) to a point where it intersects the exposure or horizontal line at the bottom of the chart. By measuring the tangent of the angle of the intercept, we derive a numerical value for "gamma," or as people unfamiliar with "film-speak" might say, "contrast." See the curve below.)

(Today most of the calculations involved in deriving gamma are done in a computer, which spits out a graph of the characteristic curve and gives you a gamma upon which you make decisions related to how the film will be developed. Back in the `20s all of these calculations were carried out by hand and judgments about exposure and electrical setup of a recording chain were also carried out by listening and trial and error.)

One thing Case discovered really early on was that-as I previously mentioned-his Aeolight couldn't put out much in the way of exposure with then current films. To get as much photographic "speed" as possible, he used positive film and then played a trick on it. This was by any other measure a method to capitalize on gross underexposure. Here's what he did:

If you examine the H & D curve again very carefully in the toe region, you'll see that there is a portion of it which-for practical purposes-is straight. If the developing is particularly vigorous, as would be the case with Kodak's D-16 or similar developing formula, the toe portion of the characteristic curve could be exaggerated and brought to its own gamma of 1.0; or to put it another way, it could be completely linear. This meant that the negative could be played directly and not sound terrible, and if the print made from such a track was carefully exposed so that it too fell on the print's toe-region straight-line, it could be reproduced with the maximum volume possible for its day with the least amount of intrinsic distortion. I'm using the word "intrinsic" because when toe recordings were in vogue, intermodulation distortion analysis hadn't yet been invented as a control method for variable density recording and probably wouldn't have amounted to much in the analysis of toe recordings anyway.

Be that as it may, toe recordings were certainly the mainstay for Fox features until 1932 with shorts and newsreels continuing as Aeolight recordings for some time afterwards. You should be aware of some of the audio specifications of toe recordings so that as I discuss the two other alternatives, you'll understand why the Aeolight and its toe recordings did not survive as the method of choice at this major studio indefinitely.

Because the light produced by the Aeolight was a glow-discharge similar to a neon light, like a neon light it required a polarizing potential which caused the ionization of the internally contained helium gas. It was this polarizing potential too which was modulated with the audio signal to produce the recording on film.

In the days before anyone had conceived of noise reduction methods, the unmodulated value for the polarizing potential was set at 50% of the available exposure range. This meant that the loudest typical cycle of sound could dip down to just above the point where the gas would fail to remain ionized and, on the positive excursion, the linearity of the record would be just below the point of onset of severe distortion. Exceed these two extremes and the recordings would not be acceptable. This led to a limited dynamic range for toe recording of approximately 30 dB. Even if the Aeolight could get brighter, the characteristic curve would prevent it from being useful! Compare this to the average range of a recording made by either Western Electric or RCA equipment of about 54 dB! There is quite a difference. Further, since both the negative and a print made from it were operating at the extreme bottom of the characteristic curve, there was no room to make use of any sort of noise reduction system that depended on exaggerating the negative's minimum density. You can't make something that's already transparent any more so!!!

Look at the soundtrack on the above sample of a Toe Recording. By definition these recordings were "thin" or lower in print density than those made by competing systems and therefore were subject to higher than normal wear-related noise as prints were repeatedly projected and rewound, projected and rewound, etc. On the good side, Aeolight recordings had relatively flat frequency response for the day, roughly 50 to 6000 Hz with no recording preemphasis. Slippage in the positive printing process probably brought the high-frequency response down 6 dB or so at 5 kHz and even more at 6 kHz. Even so, the audience of the day probably heard sound that was on the tinny side, given the extremely bumpy speaker response curves common at that time. But all in all, what the audience heard was not that much different from those made by other contemporary recording systems. But this situation did not last for long….

The Western Electric Variable Density Recording System

A more robust variable density recording system was that developed by Edward Christian Wente of Bell Telephone Laboratories in 1922. The equipment was sold under the name "Western Electric" and marketed and licensed to the motion picture community by Electrical Research Products, Incorporated, otherwise referred to as "ERPI." Unlike Case recordings, which depended on the Aeolight to be an audio-modulated variable light source, Western Electric recorders used a light valve which modulated a fixed external light source, and that light source could be as bright as necessary to produce whatever level of exposure might be required of it. In other words, Western Electric's modulator did not serve double-duty. This separation of functions made the system highly adaptable as track requirements, film stocks and processing constituents changed over the years. What made the Western Electric system succeed where Case failed? The answer lies in where the Western Electric recordings lie on the sensitometric curve!

If we go back to the classic H&D curve, we'll see that in fact there are three straight line portions. The toe and shoulder portions each have a short straight line while the classic, long straight-line has the greatest amount of real estate for placing a variable density audio recording and getting the most out of it; this is precisely where the Western Electric system put its recording.

Compared to the Case system, the Western Electric system had the potential to record a far wider dynamic range. Early experiments proved that this value was somewhere close to 60 dB; compare that to Case's 30 dB and it's not much of a horserace. But from a practical day-to-day standpoint the early utilization of the W-E system was limited to a range of 40 dB or so, and this was because of two reasons: first, strictly mathematical calculations of product gamma throughout the photographic process could not dynamically model the total distortion characteristics of the process; and second, noise reduction techniques had yet to be invented or applied to it. Techniques for coping with both of these early weak spots were rectified in the early and mid '30s. Even so, the W-E system WAS better than the Case system from the get-go and had the potential to get MUCH better. On the other hand, the Case system had nowhere to go.

From a purely business standpoint, the Case system was owned by Fox and had to be licensed by them. The Western Electric system on the other hand was available to anyone who had the bucks to buy the system and pay the license fees to ERPI to use it. Now, if you owned another studio, whose recording equipment would you buy: one supplied and controlled by a competing studio or one supplied by an "independent" non-competing source?

Let's stay with the early days of the Western Electric system for just a few moments more. Wente's theory of laboratory processes for variable density of this kind states that if the negative and positive gammas together produce a product gamma of 1.0 or thereabouts, the result will be a "distortion free" recording and reproduction of the original sound.

But how would you arrive at what was a true or real system-wide gamma of 1.0? Early literature indicates values used for these calculations to be a negative gamma of 0.6 and a positive gamma of 1.75. Multiplying both together yields a product gamma of 1.05. The unmodulated negative track exposure of 50% valve opening would produce a density of 0.6 above base and residual fog. The positive track density for the same unmodulated negative spot would be 0.5. Allowable error was 20%.

But simply knowing each film's gammas and densities does not bring into the mix all of the principal factors which affect distortion in the variable density process. Look at this:

At the very least the recording modulator has its own optical gamma or "flare factor," the negative certainly has its own gamma, as does the printer used to make the print, the print itself and the projector's sound reproducing optics. Each of these must be taken into mathematical consideration if you stand a prayer of getting anywhere close to a truly representative product gamma for the whole chain of events!

But, hold on! One other thing must be added to the mathematical soup literally and figuratively. Belying the underpinnings of Wente's theory of a simple product gamma of unity for negative and positive was this: the law of reciprocity states that exposure = intensity x time. However, for recording slit heights of .0001 inches or less, the exposure times during recording of the track negative become less than 1/20,000 of a second at 90 feet per minute of film travel. In this case the recording film no longer follows the law of reciprocity and simple math will not give us a reliable prediction of what's going to happen with negative densities.

How do you get around this mathematical "gotcha?" The answer was found in these three main areas: one, the introduction of sound recording films especially constructed for variable density recording; two, the introduction of intermodulation distortion analysis equipment to the setting and control of the total recording, laboratory and reproduction process and the removal of the uncertainties due to the effects of reciprocity law failure; and three, the introduction and complete adoption of noise reduction means.

Points You Should Know

Intermodulation distortion is the primary distortion characteristic associated with incorrectly manufactured variable density positive prints. Mathematically, it is an extremely complex distortion based on the nonlinear transmission of two or more frequencies, resulting in the generation of both sum and difference components between each of the original frequencies. Since common audio in a motion picture is rarely as simple as two pure tones, the complexity of the resulting undesirable components can quickly overwhelm a simple math description of the offending signals. However, the invention of intermodulation distortion analysis equipment literally took the guesswork out of the process of setting correct exposure and developing conditions for both negative and positive track.

One other thing happened with the changes I've just mentioned. Up to this point (the early '30s) a considerable amount of negative and positive developing had been controlled by eye without regard to sensitometry.

The recording of sound tracks forced the developing and printing processes of laboratory work into the age of objective science. Some might lament the loss of artistic control in this regard, but no one can argue with the resulting dependability and repeatability of the results.

Later Developments

As the era of, "Gee whiz, the screen talks…" gave way to the era of, "We gotta make this sound better," experience and technical analysis honed the craft in many areas. Newer, simpler modulators were designed that had higher flux levels which could be driven with more modern, low distortion amplifiers. Simpler, more direct optical and mechanical designs let the useful upper frequency range be extended from 5 kHz to 8 or 9 kHz offering a more lifelike, realistic recording. This simplification also allowed a wide variety of ribbon configurations and applications. Noise reducing circuits increased the apparent signal-to-noise ratio by 6 to 10 dB, and some studios, notably led by MGM and Columbia, added frosting on the cake with the introduction of "squeeze track" which varied the width of the recorded track to increase or decrease its playback volume at the discretion of the mixer.

A squeeze track negative and a positive print from it

Finally, in the mid-forties, push-pull recording was widely adopted as the preferred method for pre-release tracks. Push-pull made it possible to significantly reduce residual distortion in these recordings, making for quieter, cleaner mixes and producing a significant improvement in the quality of sound heard in theaters.

All in all, careful control of the process from beginning to end produced variable density optical recordings which came very close to those produced by magnetic means, which were beginning to be introduced in 1947 - but this is a big topic for another time. Let's finish the story of optical with a discussion of Variable Area.

Variable Area and the RCA Recording System

Developed in the early and mid '20s by Charles Hoxie of General Electric, variable area recording under the name of "Pallophotophone" languished from GE's disinterest for some time before David Sarnoff of RCA came knocking. Sarnoff was motivated by his distaste for AT&T's domination via ERPI and Western Electric of the motion picture theatrical sound recording field with its variable density recording system; he, and therefore RCA, wanted into the game.

At its peak, quite literally ERPI did have most of the studios under contract. Even as early as the late '20s only a few studios were available to a potential competitor. Along with General Electric, RCA's then parent company, and Westinghouse, Sarnoff struck out on a path to build his own theater chain and the studio facility to feed it, using of course the now renamed RCA Photophone equipment.

On the theatrical side RCA bought the 100 or so theaters of the dying Keith-Albee-Orpheum vaudeville circuit and began equipping them with motion picture projectors and RCA sound equipment. On the studio side, RCA bought up Joe Kennedy's Film Booking Office of America studios (nicknamed FBO) and joined it with several other smaller film-producing operations, renaming the conglomeration RKO, for Radio-Keith Orpheum or Radio Pictures. The physical location of the new studio remains today where it always was: on the corner of Melrose and Gower, now a part of the Paramount lot in Hollywood.

Examples of unilaterial Variable Area Recordings

Variable Area recording was the antithesis of everything Variable Density. It used (and still does use) high contrast, high density recording materials upon which a theoretically perfect photographic record of a moving light-stop edge or edges is exposed. To be more specific, in the earliest unilateral non-noise reducing systems, with no audio applied, the resulting track was half black and half clear. As audio was applied, the dark edge would oscillate in accordance with the waveform of the sound and the excursions of the waveform would increase or decrease depending on the strength of the applied signal. 100% modulation meant a signal amplitude just at the point of touching the extremes of the physical track width, with abrupt distortion beginning just beyond 100% modulation. From a practical standpoint, "maximum or 100% modulation" was defined as 1 dB into "clash" or the point where excessive distortion began. This extremely loud signal was rarely encountered in early release tracks even though automatic equipment such as "brick-wall limiters" did not exist at that time. The danger of severe distortion from over-modulation in production tracks was seriously discouraged by management!

There is one fly in the ointment of theory vs. reality: namely, no photographic material is capable of an immediate and instantaneous transition from total transmission to total density. If you were to take a microdensitometer and scan across such a transition, you would always find a slope between one state and the other. In the case of variable area recording, this slope is called image spread. Image spread and its effects on variable area recording may be more easily understood by examining the following illustration of three perfect sine waves as traced by the galvanometer.

All three thin-line traces are "perfect" in shape when laid down by the recorder; but when you examine the photographic record produced by under-exposure, correct exposure and over-exposure, shown here as the diagonal hash marks, you'll see that, in the case of the underexposed, silver image on the left there are gaps between its image and the original exposure. Likewise, the over-exposed silver image on the right shows density exceeding what should be-in other words, the image has spread.

Of course I'm writing of a two-part system of recording: the original sound track negative and the print from it. To produce a final recording which has no distortion, the image spread of one film must be counteracted by the image spread of the other. But how do you select what the exposure and developing conditions of the track negative should be when the gamma of the positive print is fixed by the requirements of picture content and track density is set by technicians who like to keep track specifications constant from one show to the next?

In the early years of Variable Area recording the only answer to the preceding question was to resort to listening and trial and error - a system which once again could only approximate what the correct negative exposure might be, not deliver a finite yet simply arrived-at and accurate answer.

The answer came in 1936 or thereabouts, with a system of electronic cross-modulation distortion analysis suggested by G. L. Dimmick and brought to fruition by J. O. Baker and D. H. Robinson, all of RCA. This type of distortion measurement system made setting both the exposure and development of a given negative track emulsion easy and predictable and took into account the entire chain of events from recorder to film to reproducer. It is still the dominant means of accomplishing this end today.

More Points You Should Know

Unlike peak distortion produced by exceeding 100% modulation of the system, cross-modulation distortion exists in negatives as a natural product throughout the recorded signal, whether loud or soft. In the 1950s, efforts were undertaken by RCA to build a cross-modulation compensator to reduce the inherent distortion in direct positive, one-step recordings which of course were functionally negatives. The technique involved the intentional pre-distortion of the recorded signal in a manner such that it would counteract the anticipated image spread occurring in processed film. Such a system was instituted in a number of recording chains for 16mm use, but was not widely adopted by the 35mm studio industry. However, the same conceptual approach may be applied after a conventional negative has been made to allow its direct playback with a considerable reduction in distortion. Modern digital techniques have made this approach more useful.

Many variations of the appearance of the track itself have occurred since the original simple unilateral configuration. The above chart shows a variety of pre-1948 track configurations. Unilateral transformed to dulateral to bilateral, then dual-bilateral to multi-bilateral with as many as 13 or more individual area tracks packed into the same space as unilateral. What was the point of all of this? Most of the changes you see here were made with the view of sidestepping non-uniform illumination and poor azimuth in early or poorly maintained reproducers.

As in the case of Variable Density recording, Variable Area recording benefitted greatly from the application of noise reduction techniques produced by the addition of side-shutters, which blocked out the otherwise wide open areas of track unused during silent or quiet effects or speech.

Finally, as in Variable Density, push-pull recording offered a whole new dimension of lower distortion and noise with broader, more easily attained recording conditions for the technicians producing these recordings in the late 1940s.

How do you print Legacy Optical Track Negatives today?

There are no easy answers to this question. There is no book, no roadmap, no great guru to give a definitive answer and there are no test signals on the track leaders which can be measured to help in this matter, but there are a few good guideposts to observe, and here they are:

Know the history of the films that were used

Become familiar with the history of the system with which you are working. A lot can be gleaned by learning the characteristics of and developing suggestions given by data sheets of the various films which were used at the time a given film was made.

Learn the science of the recording systems

Immerse yourself in the science of the recording systems used to produce a given type of track.

Know what can be expected from a certain era of track

Know the time spans within which a particular type of track was made and what the realistic expectations are for a given era's product.

Learn to recognize what type of track each recording system makes and each system's typical distortion characteristics

Understand and learn to recognize what dominant type of distortion is typical for a given system and know what you can do to influence and minimize that type of distortion. To wit:

Toe Recordings

For Toe recording: Be aware that not only was Toe Recording used for early sound newsreels and some features, but that it was also used for a great many direct positive kinescope recordings of television shows from the mid-1940s through to the late-1950s. This also holds true for Kodachrome Commercial prints with sulfide tracks during the same period. On the other hand, Western Electric style (NOT Toe Recording style) Variable Density single-system recording was commonly used for many negative kinescope recordings intended for conventional printing to positive films. Adjust your thinking accordingly.

Western Electric (Westrex) Variable Density Recordings

For Western Electric (Westrex) Variable Density Recordings: Learn what intermodulation distortion sounds like. In the simplest sense, it's this: "Where low frequencies chop holes in high frequencies." This usually means run a series of exposures of a relatively loud music section (usually main titles) and listen carefully how bass notes influence high strings. If you hear holes in the string sound which correspond to what's happening in the bass line, you've got intermodulation distortion. Select the exposure which has the least of this effect.

RCA Variable Area Recordings

For RCA Variable Area Recordings: Learn what crossmodulation distortion sounds like. Typically excessively sibilant dialogue will be an obvious tipoff. Rough sounding speech surrounded by a rattling effect or grating characteristic that is not natural is what you're listening for. Run a series of test prints of a sequence of dialogue containing lots of "Ss" at various exposure levels and select the exposure with a minimum of this distortion.

Other Manufacturers' Equipment... Time Marches on!

Bear in mind that manufacturers other than RCA or Western Electric made optical recording equipment, for example: J. A. Maurer, W. A. Palmer and Bach-Auricon. Also, remember that Western Electric (Westrex) equipment, patents and designs were folded into Photophone in the 1980s and that the bulk of Variable Area recorders in use today either were Variable Density machines by Western Electric to begin with or are modern clones of the Western Electric design, set up to produce Variable Area tracks using light valves, not galvanometers.

Brush up on you knowledge of photographic processing

Know your way around photographic processing and recognize what you can and can't do by careful manipulation of gamma and exposure.

Examine the original carefully!

Carefully examine your incoming track negative and look for the following tips: Is the negative notched, indicating that it should be timed or graded to match levels or scene-to-scene changes? Is the negative composed of pieces of both Variable Area and Variable Density recordings? Should these sequences receive different exposures? Many labs today are not equipped to time or grade sound tracks and/or don't care.

Relax, this is not rocket science....

Finally, I recommend that, if you're faced with material with which you are not familiar, take a deep breath, do your homework, approach the work systematically and you will not be disappointed with the results. Remember that the work you do in fact includes not just the protection of the images we cherish but the sounds that accompany them. The future will thank you for them.

Back to Top | Home Page