Uploaded 16 November 1998
Film history is like a massive graveyard in which lost films are buried, never again to be recovered or seen. As in life, the dead outnumber the living by a long shot. Although we have only just celebrated the first centenary of cinema, the statistics of mortality are frightening. Of all the films produced during the silent era, i.e. between 1895 and 1930, approximately 90% have been lost. In other words, only ten percent of all films from that era are still in existence. Of all films produced during the nitrate sound film era, i.e. between 1930 and 1955, only about 50% survive in any form. Independent, avant-garde, and documentary filmmakers are notoriously unconcerned about their past work, once a film has played its commercial run. They are justifiably more interested in finding the funding for their next project. Meanwhile, the negatives are lost, the remaining distribution copies are routinely destroyed or worn out through continual use. Films disappear from view and consciousness, unless some interested party manages to put a print away for safe-keeping.
Even the films that have survived are not necessarily safe. Many films only exist in mutilated form or in foreign archives or in the hands of commercial interests unconcerned with anything but the profit motive. Films made in certain wide screen formats or with specific now obsolete sound systems or most colour films from the last three decades are in grave danger. Unprotected nitrate films continue to decompose in the vaults, due to chemical instability and adverse climate conditions. Acetate films are subject to vinegar syndrome, i.e. they, too, have now been found to decompose, as a result of insufficient temperature and humidity controls. Colour films not based on imbibition colour systems (where the negative is actually black & white) are subject to fading, so that eventually only a monochromatic magenta remains.
We now live in the age of video and seemingly instant accessibility. Most non-professionals believe that every film ever made has already been transferred to video and is now just waiting for a push of the VCR button. It needs to be emphasized that the project of film preservation and restoration is far from completed. The great majority of films presently housed in the film archives are not archivally secured. It is estimated that in the United States alone nearly 35 million meters of nitrate film remain unprotected. When we begin to scratch the surface of the acetate era, when we begin to think about color preservation, when we consider the many “obsolete” wide-screen formats of the 1950s, then the work of film archiving is not even close to completion. The preservation of only the first 100 years of cinema will take at least until the middle of the next century, maybe longer.
Given these huge gaps in the accessibility of historically important film titles, it stands to reason that the writing of film history is also a function of film preservation, and vice versa. On the one hand, film preservation priorities are established in the film archives, according to the classical texts of film history. On the other hand, film history can only be revised, if film archives make certain film titles available. In other words, a symbiotic relationship exists between film preservation and film history.
I would like to illustrate this point with a film classic, that has been seemingly available for years and is a standard offering in many film history courses. Dozens of film historians have written about the film, yet it is my contention that none of the historians writing about the film has ever seen The Joyless Street. Until now. Anglo-American film critics have relied on the version distributed by Kino International, a bastardized version of The Joyless Street, which in no way resembles the original. The Kino print is based on a French version that survives at the Cinémathèque Française which is missing more than four reels from its original length of 3,700 meters.
II. The Joyless Street – a damaged masterpiece
Die freudlose Gasse(The Joyless Street), directed Georg Wilhelm Pabst from a script by Willy Haas, based on a novel by Hugo Bettauer, is not only one of the most important films of the Weimar Republic, it is also one of the most spectacular censorship cases of the era. While the film made its director famous, the state institutions of control guaranteed that no one would ever see the film in its original form. The film was considered too much of a provocation. Its story from the inflationary period in Vienna in the years immediately after World War I offered enough dynamite for several muckraking novels: nouveau riches currency and stock market speculators who wallow in Babylonian luxury, homeless and unemployed Lumpenproletariat living in barns, women who sell their souls for a bit of fresh meat at the butcher’s, arrogant but impoverished former bureaucrats unaware of their social slide, young social climbers willing to prostitute themselves with high society women, sexual orgies and bordellos, murder out of jealousy, murder out of despair, and, finally, a revolution in the streets.
Contemporary film critics, as well as film historians have always recognized The Joyless Street as a seminal film, a film on the border between German Expressionism and Neue Sachlichkeit (new realism). Two quotes from contemporary reviews may illustrate the point: the Berliner Tageblatt wrote: “In terms of its acting, this is a beautiful film with great direction and wonderful actors.”  In fact, the film starred Asta Nielsen, Werner Kraus, Hertha von Walter, and a young Greta Garbo who would go on to a stellar career in Hollywood. In the Deutsche Allgemeine Zeitung, Otto Friedrich wrote:
He (Pabst) achieves his goal in great style, does not tread on well-worn paths, presents new ideas which are uniquely his own, and with this achievement gives us a new hope. He captures the atmosphere of an idea, keeps redrawing it on the surface in ever more intensive images, in contrasts of action that make an ever stronger impression. He illustrates the down-trodden souls of the era and the manic lust for pleasure of its people who are trying to do nothing more than escape for a few minutes from the terrible truth of the present. 
In this context it is interesting to note that both Lotte Eisner and Siegfried Kraucauer characterized the film as unsuccessful, claiming it to be too melodramatic, too stereotypical, too over-laden with symbolism. Eisner indeed criticized the film’s blending of expressionistic symbolism with a realism characteristic of the later Pabst. Yet one must remember that both Kracauer and Eisner only had access to mutilated versions of The Joyless Street, or reviews of mutilated versions.  In point of fact, the film was chopped up, mutilated and censored, like no other film in the Weimar Republic. Only a few short years after the film’s premiere, Paul Rotha described its fate in The Film Till Now:
The film was shot in 38 days, since they worked sixteen hours a day. When it was finished it had a length of 3700 meters, almost as long as Ben Hur or The Big Parade. France immediately trimmed 700 meters, including every shot of the street. The Viennese cut without reason every scene in which Werner Kraus was seen as the butcher. In Russia the Lieutenant is transformed into a doctor and the butcher the murderer (instead of the girl). After a year, the German censors went back to work. In England the film was banned altogether and shown only once at a London Film Society screening. 
Butcher (Werner Kraus) at his window, eyeing legs of Else.
Else (Hertha von Walter), homeless and hungry.
Butcher thinking about meat for sex.
Butcher in his meat locker, after closing shop for lack of meat.
Else after the “rape” in the butcher shop.
Else, Marie (Asta Nielsen), Butcher: The payoff.
Not only was the film cut for political and moral reasons in every country where it was publicly shown, it was also reedited in order to close the huge gaps created by censorship cuts. Thus, in almost every country totally different narratives of The Joyless Street emerged. According to Mark Sorkin, Pabst’s editor, he and Pabst had made the first cuts the night before the premiere, because the theatre owner insisted the film be shortened.
While the first version of The Joyless Street still had a length of 3738 meters (only four meters were initially cut by the censorship board on 25 May 1925), the film was back in court on 29 March 1926, because the police had issued a decree calling for a total ban on the film, due to its “lewd” and ” seditious” tendencies. Now only 3477 meters remained of the film. None of the later surviving versions ever came close to this length. Neither a German version, nor the original German censorship cards are known to exist.
As a result, film historians have only seen mutilated versions that in no way represent the original, leading them to false conclusions, based on evidence that only existed in such false versions. For example, Patrice Petro’s close analysis of the film in Joyless Streets: Women and Melodramatic Representation in Weimar Germany  does consider two different versions of the film, but neither one was even close to genuine, so that her conclusions are off the mark. She also fails to consider the provenance of the two versions, just as most film historians have failed in their analyses to take into account the extreme losses suffered by film history. Unlike literary critics studying sources and versions in print, film historians often assume the film they are seeing is the text. However, no two film prints are alike, and the final editor of a film is the projectionist who has removed his favorite scenes for his personal collection.
In 1989, Enno Patalas of the Munich Filmmuseum attempted a first restoration of The Joyless Street. That reconstruction was based on three different versions: a print from the Cinémathèque Française with French intertitles (2,309 m); a version with English intertitles from the National Film and TV Archive at the British Film Institute, London (2,168 m); and a print with Russian intertitles from Gosfilmofond, Moscow (2,950m), which, however, included at least forty meters of material from other films. All three copies differed from each other not only in their length and intertitles, but also in terms of their editing of scenes and shots. During the course of this first reconstruction, which took as its guide an original script with notations by Mark Sorkin, several seemingly unsolvable problems arose:
1. The completely different editing structures in the three versions, as well as the absence of censorship cards, which would have allowed for the intertitles to be reconstructed,  made an exact reconstruction of the original version impossible. While every surviving shot could be identified, numerous scenes were still missing, so that huge gaps in narrative logic remained. Furthermore, Patalas designed new titles from dialogue in the script.
2. The script only approximated the completed film: a number of scenes described in the script were not shot (according to Mark Sorkin’s hand-written notes in the margins); other scenes which were supposedly not produced, actually showed up in surviving prints; other scenes were constructed differently or appeared at a different point in the film.
3. The varying quality of the three prints – all of them positive prints from dupe negatives from nitrate originals – made it difficult to achieve a unified look. Since the original nitrate prints were not viewed, it was also impossible to tell where splices had been made, evidence which would have indicated where scenes, shots or titles had been removed. Also, the tinting and toning – two of the nitrate prints were tinted – could not be reconstructed from the b & w dupes acquired by Patalas.
Other problems only became apparent, once work on a second reconstuction commenced.
III. Close analysis at an editing table
In 1995, the Munich Filmmuseum applied to the “Project Lumiere” of the European Union’s Media I Project for funds to attempt a new reconstruction of The Joyless Street. In particular, the rediscovery of the Cinémathèque Française’s original, tinted nitrate print made such a new project feasible and desirable. Furthermore, the planned complete retrospective of all of Pabst’s films at the Berlin Film Festival in February 1997 offered another incentive for once again attempting a more complete reconstruction. However, at the time the decision was made, the reconstruction team, consisting of Jan-Christopher Horak, Gerhard Ullmann and Klaus Volkmer could hardly have imagined how complicated the reconstruction would become or that totally new sources would turn up in the course of their work.
In the Spring of 1996 the Munich Filmmuseum requested from its FIAF partners all surviving nitrate prints of the film, i.e. a survey of all surviving materials and their origin was made. Of the 21 copies from Austria, Germany, the United States, Canada, Great Britain, Italy, Hungary, etc. five projection prints were identified as Ur-nitrate prints:
1) a print from the estate of film collector/distributor, Raymond Rohauer, (now deposited at the Library of Congress), which had served as pre-print material for the British Film Institute’s safety dupe negative (London).
2) the Cinémathèque Française tinted print (Paris).
3) a print from the Cineteca di Milano with French intertitles which had apparently been cut together from two different versions, and only partially identical to the French print (Milan).
4) the Russian print from Gosfilmofond which had two different typographies in the intertitles, leading to the conclusion that the original version had later been reworked (Moscow).
5) a 59-minuted version from George Eastman House, Rochester, which James Card had bought from a Cleveland junk dealer in the early 1960s. This version had English intertitles and had apparently been released after Greta Garbo’s rise to fame in America in an attempt to cash in on her star attraction (Rochester).
After requesting and receiving these five nitrate originals, an exact shot by shot protocol was made of all five prints, as well as correcting and updating the shot protocol of Munich’s first reconstruction. We soon ascertained that these prints originated from two different negatives which had been cut differently and also used different shots and angles. The London and Paris prints, and half of the Milan print came from one negative, and the Rochester and Moscow prints and the other half of the Milan print from a second negative. Thus, a first theory that there had been a negative for the domestic German version and a second negative for foreign release was shot to pieces, since we were dealing exclusively with foreign prints. According to Mark Sorkin’s notes, he and Pabst had simply cut an “A” and a “B” negative.
We realized very quickly that we would not be able to reconstitute either an A or a B negative, since none of the prints were anywhere near complete, but rather we could only complete a synthetic version from the two negatives. Furthermore, we knew that within a single scene, we would have to choose between shots from negative A or negative B, since it was impossible to match shots within a scene from the different negatives, without risking losing the film’s elegance. At the same time, it was clear from seeing the original nitrates that the image quality of the London, Paris, and Rochester version was vastly superior to all surviving dupes and that at the very least we could improve the surface image in our restoration. We also noted that each version had different scenes missing, i.e. all the versions ended with a happy end between Greta and Lieutenant Davy, except the Russian one, which ended with the burning down of the bordello. The Rochester version, on the other hand, while extremely short, included a few key shots and scenes, not found in any other version. For example, we discovered a freeze frame of three naked women in a tableaux vivant on a nightclub stage. My first impulse was to assume that the shot had been added later by James Card, not an impossibility, if one knew Card’s predilections or our discovery that the Moscow version included footage from at least two other films.  However, the fact that the shot was nitrate and no splice could be found forced us to look at the shot more closely. After examining the nightclub scenes in other prints with a lupe, we noticed the silhouette of a naked woman at the edge of one shot where Greta flees the nightclub. That silhouette perfectly matched our freeze frame image.
The first step, then, was to make a dupe negative of the complete London nitrate, since we decided to use this as our base print, due to its superior image quality and the length of its shots. In addition, dupe negatives were made of all those scenes and sections from the other versions which were either missing in the London print or were possibly of better image quality. Work prints were then generated from all new negative materials.
Working from the London version, a rough cut was put together in four separate passes, during which scenes and shots from the Milan, Paris, Moscow, and Rochester copies were successively incorporated into the English print. In order to facilitate constant comparison between the five versions and check our own work, we generated VHS-video transfers for easier reference of every version, including a German version, produced in the former GDR, and the first Munich reconstruction. We also made sure that we never “mixed negatives” within a scene, as stated above.
During the ensuing fine cut, we realized that not only had numerous scenes been removed or altered, due to censorship cuts, but also that various narrative strands had been totally mixed up. It was therefore not just a matter of adding missing material, the most complex part of the job lay in discovering the film’s original editing structure. The editors responsible for the various foreign versions had indeed taken incredible liberties. Scenes were chopped to pieces, shots from the end of the film were put at the beginning, or vice versa, longer shots were cut up into shorter ones with other images slapped in-between. As a result, persons often appeared in two different scenes at the same time or shots from one scene were cut into another scene. For example, a close-up of a female piano player with a cigar in the corner of her mouth appeared in a nightclub at the end of the film in the London and Moscow versions, yet this woman was no where to be found in the long shots of the same scene – on the contrary, a male dance band was seen playing in the establishing shot. It was only during the third pass that we noticed this very same piano player – very small – in a corner of the bordello scene at the beginning of the film.
IV. Approaching Pabst’s realism
In order to create a fine cut out of this seeming chaos, without censorship cards or an original print, the restoration team formulated a working thesis, namely that G.W. Pabst had for the most part attempted a realistic narrative, i.e. spatial and temporal relationships functioned chronologically. Although Pabst had not yet perfected his technique of invisible editing on movement, the plot had to make narrative sense. Characters could not be in two places at once, their movement from one space to another had to fit into a realistic space/time continuum. An exact philological analysis of each shot was therefore necessary. Diagrams were drawn of individual rooms and spaces, based on spaces, movements and looks within the frame, allowing us to reconstruct sets in space, while the many individual sub-plots were painstakingly chronicled. Again and again the team discovered “mistakes”, e.g. Patrice Petro describes how the butcher’s large dog runs after his murderer, a series of shots found in at least two versions. Yet, close analysis of the shots revealed that the woman is running on a cobblestone street pavement, while the dog is on a wooden floor, and therefore must have been in a completely different scene which, as it turned out, was still almost completely missing, forcing us to throw out the shot completely (this was one of only three shots that could not be inserted into the completed film). On the other hand, the team noticed that transitions would suddenly work and the editing structure became much smoother, once shots had been reordered according to our method of close analysis. Pabst was indeed very close to a classical narrative style as our “matched” cuts (if they were from the same negative) proved.
In the course of our work, the team developed a second working thesis, the so-called “Potemkin-Principle”: If a shot had been placed in the wrong place, or, as was the case in the Moscow print, duplicate shots from elsewhere in the film were inserted,  one could assume with certainty that a scene or sequence of shots had been removed. For example, we discovered that the rape scene in the butcher shop (unseen in the meat locker) was rendered harmless by inserting a shot of Maria’s father walking down the street. Thus, in the censored version it appears as if Maria is scared, because she hears her father (which in fact makes no narrative sense because he can’t see her in the basement), rather than the rape of her girlfriend, occurring in the meat locker. Thanks to our hypothesis, the team also discovered the film’s complex flash-back structure.
Greta Rumford (Greta Garbo), pleading with would-be client.
James looking over at Greta and Rosenow in bordello.
Rosenow (Mario Cusmich) listening to Greta’s case.
Rosenow gives up on his date.
Rosenow complains to Greifer (Valeska Gert), the madame of the bordello.
Title: “I ordered a little girl friend from you, and you deliver me something like that!”.
Despite the fact that the Paris version included several scenes of Maria’s murder of the lawyer’s wife, Enno Patalas had removed the second of these scenes, because Maria’s confession at the end of the film explained all. It was however my opinion that the presentation of the scene three times, each time revealing a bit more, corresponded to silent film conventions: the first time, one sees Asta Nielsen looking into the next room, where she discovers her boyfriend with another woman; the second time, she relates how she has witnessed the murder to her customer, Canez; the third time, she confesses the murder in the police station. However, it was not until we discovered a duplicate shot where the first flashback should have been that my team accepted my opinion.
Another important discovery in relation to the film’s overall narrative structure was made when we attempted to incorporate a number of shots from two reels of “outtakes” that had been discovered in Berlin by film archivist Martin Koerber at the Stiftung Deutsche Kinemathek. The two reels included actual outtakes, shots from several unidentified films, and two sequences which were eventually identified as belonging to three different scenes from The Joyless Street. Of major consequence for our edit were several shots with Greta Garbo at a table with the speculator Rosenow and other shots with Davy’s friend, James, and an unidentified companion looking from their nightclub table at something. These shots were in no other version, but we were able to identify them in the script. The question was, however, where did they fit into the film? We only knew about the seduction scene with the butcher, which had been placed after Davy’s departure from the flat.
Only after repeated viewings of the crucial scene in the apartment did we notice that James actually recognizes Greta and then begins to push Davy out of the flat which meant that he must have assumed that she was a prostitute, because he had seen her in the bordello with Rosenow. We concluded that not only did the shots with Greta/Rosenow and those with James/companion belong together in a scene before Davy moves out, but that also the scene with Greta/the butcher had to come much earlier. For the first time there appeared a clear motivation, why James literally drags Davy from the flat, even as Greta tries to hold him back. More importantly, the whole dramaturgy of the Rumfort’s economic plight, forcing Greta again and again to find money for food, because the family funds keep disappearing, suddenly became clear. Greta indeed nearly prostitutes herself three times in the course of the film, until she is finally rescued by Davy, providing the second of three climaxes.
A second dramatic structure involving Marie continually listening at doorways and then looking in, thus setting a whole chain of unhappy circumstances in motion (as in Pandora’s Box, in which she starred two years earlier) also became visible as work progressed. Thus, the more we worked, the more we realized that Pabst had created editing patterns and rhythms which had never been evident in any of the censored versions.
V. The mystery of Joyless Street’s intertitles
After we completed the first fine cut, the restoration team turned to the intertitles. We had originally assumed that we would be able to use the intertitles made for the first Munich reconstruction of 1989. However, the massive changes brought about by our reediting of scenes forced us to reconsider our procedure. It was only at that point that I learned that the German intertitles had been formulated wholesale by Patalas from dialogue in the script. All along, I had had the feeling that there were both too many intertitles and that the titles were too long. Intertitles in silent films were usually added after the fact and kept to a minimum, while the dialogue in the script was understood to be help for the actors creating a character, rather than what they were actually saying. The first scene in the Hotel Carlton is exemplary in this respect: A number of persons approach the speculator Canez, in order to get him to invest in their schemes. Five intertitles had been cut into the scene, although the scene worked without any intertitles, and, as we later realized, based on evidence from a German censorship report from 1926, probably was shown without titles. We decided, therefore, to proceed completely differently. 
First, we electronically photographed and translated every single intertitle from all five foreign versions, as well as from the GDR version, and pasted them together in sequence. By comparing the intertitles side by side we were able to discover which titles were used in two or more versions. In a first pass we only added those titles which were documented in several versions, since we assumed that they had probably been translated from the original German. In a few spots we then had to use intertitles from the script, because there were no corresponding scenes in any of the surviving versions, e.g. the scene with Garbo/Rosenow.
In a next step, we compared our intertitles with a German censorship report from 1926. This text quotes directly or indirectly numerous intertitles from the original version, but, more importantly it numbered the quoted titles sequentially (reel X, title XY). Inexplicably, this report had not been consulted during the first reconstruction for creating intertitles, but rather merely as a guide to plot construction. We thus began counting titles in each reel and soon noticed that some reels still had too many intertitles. We also realized that some scenes were still in the wrong place, and were able to make corrections. Now the reconstructed film has ten intertitles less than the first reconstruction (236 vs. 245), although a number of new scenes have been added. Seventy new titles were eventually generated, while the remaining titles were close enough to the titles in the foreign versions to be reused.
Thus, the new reconstruction has a length of almost 3,000 meters (only slightly longer than the Moscow print), but has been almost totally reedited in comparison to all previously known prints. While the literature describes versions centered around two major female figures, Marie (Nielsen) and Greta (Garbo), the new reconstruction has four major female characters: Else (Hertha von Walter), Marie, Greta, and Regina Rosenow (Agnes Esterhazy) who represent every class in post World War I Vienna: homeless “Lumpenproletariat”, working class, destitute middle class, and nouveau riche upper middle class. Around these four female figures Pabst constructs a true tableau vivant of more than eighteen characters, whose individual stories reveal the era’s social inequalities as surely as the naked women in the nightclub expose their bodies. The first reel is nothing more than one long narrative of exposition, introducing virtually all the characters and their circumstances, giving the film a previously unrecognized depth. Over the next eight reels these stories are continuously interweaved, making this film a true masterpiece.
VI. Giving The Joyless Street its colour
After the Munich Filmmuseum’s reconstruction of The Joyless Street was premiered at the Berlin Film Festival in February 1997, work continued on the film. First, the new dupe negative was cut together to match the work print. Then a colour plan had to devised to tint and tone any prints made from the new negative. This was not an easy procedure either, since the tints in the London and Paris versions were not identical (the Rochester material was tinted amber throughout), and many new scenes were in neither of those two versions, so that there was no evidence to guide our tinting for those scenes.
We therefore decided to use the Paris version as our primary guide, and only consult the London version, when the scene was missing in the French print. We decided to take this course of action, because the colours in the Paris version were simply richer, although we were also well aware that Pabst had insisted on a very limited palette of colour, mostly browns, greens, and yellow tints and tones. Once we had identified the colour in every existing colour scene, we tried to see whether there was any pattern to the tinting. We noticed that certain spaces were indeed consistently coloured (except in night scenes), and therefore continued to use the same colour in shots we only had in black and white. Only in one case, the fire scene at the end of the film, did we use a colour tint without direct documentation: in that case, we followed silent film conventions and tinted the scene red. However, imperfections in the Russian material strongly indicated that the scene had been tinted some colour.
We encountered one last problem while trying to put our tinting scheme into practice. It is of course no longer possible to tint and tone, as in the days of silent film when every positive print was put together from shots that had been individually dunked in a dye bath. Furthermore, we knew that the dyes in the original tints and tones had most probably faded, so that it was not really possible to recreate the original colours. Working together with the Cinemateca di Bologna which was responsible for all the lab work, we decided instead to use the “Desmet-Method” which involved generating a black and white negative, and then flashing the colour print material with coloured lights, before exposing the material to the negative.  The advantage of this method was that we could colour material that had been previously black and white, and our new dupe negative would not be subject to colour fading, as so many colour negatives made during the past forty years have been. After a whole battery of colour tests, we were able to achieve tints and tones that roughly approximated the colours in the surviving nitrates. The new colour print was premiered at the “Cinema di Retrovato” Film Festival in Bologna in June 1997. 
Everyone who has seen the new reconstruction of The Joyless Street agrees that it is a vast improvement over all previously existing versions. Yet it remains only a subjective attempt at a reconstruction. While it is true that the reconstruction team tried to base every one of it decisions on documentary evidence, it must also be admitted that without a surviving original German version, we will never know what the film really looked like. It is, for instance, possible that Pabst actually included mismatched shots in his edit, because he probably rightly assumed that no one would look at every shot that closely, and the precepts of classical narrative editing were only beginning to fall in place, at least in Europe. Given the incredible complexity of the narrative, involving at least six different subplots, and the fact that the film is still missing at least 700 meters, this version must be characterized as at attempt at a reconstruction. A rolling title at the beginning of the film in fact characterizes the film as such, and gives a brief description of the process described above. In is not our intention that future historians mistake the Munich edit for Pabst’s original version.
This version was created according to certain well-defined theses, concerning the nature of Pabst’s work. In this sense, it is both a film and a meta-film. Future film historians when dealing with this or any other film are going to have to be much more exact in their definition of the text. They can no longer assume that the print they are viewing and analyzing is anything more than one possible text, making their reading only one of many possible readings. Just as philologists have long assumed that they must reveal the provenance of the literary text under discussion, so too must film historians spend much more time, defining the exact characteristics of their particular texts. Not only are we dealing with release versions, rerelease versions, and “director’s” cuts, each individual print shows the ravages of time, making it a unique object. The work of the film archivist is to discover the best possibleur-text, using all the methodologies and documentary evidence at their disposal, while resisting the urge to improve both the technical and the aesthetic quality of the work, just because modern laboratory methods make it possible. At the end of the day, though, there must always be the realization that any reconstruction is as much a work of subjectivity as it is of documentary truth.
 This is a revised version of “Der Fall Die freudlose Gasse. eine Rekonstruktion im Münchner Filmmuseum”, in: Ursula von Keitz (ed.): Frühe Filme, späte Folgen. Restaurierung, Rekonstruktion und Neupräsentation historische Kinematographie (Marburg: Schüren, 1998) translated from the German by J.-C. Horak.h
 “Die freudlose Gasse“, in: Berliner Tageblatt, Nr. 237 (20 May 1925).
 Otto Friedrich, ” Die freudlose Gasse“, in: Deutsche Allgemeine Zeitung, No. 239 (23 May 1925).
See Lotte Eisner, The Haunted Screen (Berkeley, CA.: University of California Press, 1969), p. 256; Siegfried Kracauer, From Caligari to Hitler. a Psychological study of German Cinema (Princeton, N.J.: Princeton University Press, 1947), p. 67. While Kracauer probably only had access to reviews and his own memory of the film, Eisner viewed the surviving print at the Cinémathèque Française.
 Paul Rotha, The Film till Now: The Film Since Then (London: Spring Books, 1970), p. 37.
Patrice Petro, Joyless Streets: Women and Melodramatic Representation in Weimar Germany (Princeton, N.J.: Princeton University Press, 1989), p. 201ff.
 German censorship records were published on so-called censorship cards. Those cards listed all intertitles in numerical order that had been passed by the censors. Based on such records, it is often possible to reconstruct the narratives of German silent films, even when no intertitles survive. Unfortunately, no censorship cards for The Joyless Street have surfaced to date.
 Having worked for Card in the 1970s, I could imagine him fooling with the print. I was able to identify an extensive stock market scene as originating in Joe May’s 1923 four-part, German serial, Tragödie der Liebe. Another long night club scene from an unidentified film was also clearly inserted into the film.
 The Munich Filmmuseum staff had noticed during the reconstruction of Potemkin that the Russians were fond of duping shots from other parts of the film and inserting them, when footage had been removed. The concept also refers to Potemkin Villages.
 In the Berlin outtakes were two original German intertitles from the film with the original typography. Since the style was very close to the style chosen for the first Munich reconstruction, we decided to continue to use that typography, thus saving us the cost of redoing every single intertitle. The credits at the beginning of the film were also done in the same style and were put together using various sources, since again no direct documentation on the film credits survives.
 This method was developed by Karl Desmet at the Cinematheque Royale de Belgique (Brussels) and has become standard practice at many archives. Other archives still prefer to make a colour dupe negative of the original nitrate and work from that. The advantage of colour film is that it is more vibrant, even if it distorts the original colour tinting. Clearly, in the case of The Joyless Street, a colour negative was not an option, given the extreme homogeneity of the material.
 The reconstruction of The Joyless Street took three staff persons a total of one and a half years to complete, of which half a year was spent on research and documentation, half a year at the editing table, and half a year creating new titles and a colour print.