The "Napoléon" Comet: The Expert Appraisal of a Collection

Joël Daire - Georges Mourier - 24 juin 2015

The Napoleon Comet

"If you deconstruct the flight of my dove, it will fall while you explain it to me."

This phrase, attributed to Leonardo da Vinci and quoted by Gance in his correspondence, perfectly illustrates the dangers inherent in any detailed study of a work of art, and, a fortiori, when the analysis concerns a film of the stature of Napoleon. We should be wary of films buried under all sorts of honours and turned into monuments under the facile epithet of « works of genius ». This conveniently frees us from having to make any effort to understand the conditions by which the director achieved such transcendence. Legends often make us forget the humble creative craftsman. However, such an expert appraisal cannot be carried out without a secret fear of interfering with this legend and looking like Gance’s inquisitor. Would it be possible to maintain the legend, and the fascination that the work exerts everywhere? In this era of obsessive demystification, instead of a pure crystal, would we once again discover a complicated machine, full of tricks and processes, a « mechanism », thus destroying a bit of our dream ? In the worst case, would we let ourselves be trapped by vanity, starting from the principle that everything that has already been said and written about this film is highly suspect ? From this point of view, it would not be possible to ignore or systematically question the restoration work that has already been done on the film, notably that of Marie Epstein, Henri Langlois, Bambi Ballard, and Kevin Brownlow. On the contrary, our idea was to fully understand how previous restorers had worked using the means available to them, in particular Kevin Brownlow, whose career as a historian and film restorer was deservedly rewarded with an Oscar in 2010. Since we have been in correspondence with him for a number of years, we informed him of this enterprise from the beginning, as it was a delicate undertaking from every angle.

The « Napoleon Comet » was back.

Since we considered ourselves as continuing along the lines of our predecessors, we attempted to examine this film, which has never ceased to be both an object of fascination and an enigma, in the light of today’s methods and tools.

Affiche Napoléon

Reasons for the Expert Appraisal

First of all, we can legitimately ask why, after five restorations, it was necessary to clarify the situation concerning Napoleon and to try to trace its foundations. There were three main reasons :

  • the extreme complexity of the film’s history, its release, conservation, and restorations ;
  • the discovery of and access to non-film archives ;
  • the arrival of new digital technologies.

Short History of the Film’s Release

  • April 7, 1927 : Gala presentation at the Paris Opera in a 5,200-metre version (3 hrs. 47 min.), tinted and toned with triptychs (using three screens). We will call this the Opera Version.
  • May 8/9 and 11/12, 1927 : Paris trade screenings for distributors and the press at the Apollo Cinema, with a length of 13,261 m. (9 hrs. 40 min.), without triptych ; the length was reduced to 12,961 m. (9 hrs. 27 min.) for release. We will call this the Apollo Version.
  • November 24, 1927 : Shipping to Metro-Goldwyn-Mayer of a positive copy with triptychs (using three screens), then of a negative for international distribution of the film, with a length of 9600 m. (7 hrs.). This was improperly called the « definitive version ». It was reduced to 2,438 m. (1 hr. 46 min.) on January 7, 1928, by MGM.
  • May 1934 to January 1935 : Abel Gance decided to add sound to his silent movie. He filmed new scenes, post-synchronized the silent parts, and re-edited it. This new film, entitled Napoleon Bonaparte (3,850 m., 2 hrs. 20 min.), was released on May 11, 1935, at the Paramount Cinema, Paris. During the production of this version, the re-sized negatives from 1927 were lost.
  • December 1968 : Abel Gance added a new soundtrack to his film, just as he did in 1934, with a new post-synchronization. This gave rise to a third film, entitled Bonaparte and the Revolution (7,544 m., running 4 hrs. 35 min./24 fps), which was released on September 9, 1971, at the Kinopanorama Cinema, Paris.

Thus, counting only the interventions by Gance, we now have three different films (1927, 1935, 1971), and for the 1927 film (at least) three different negatives, not counting the new montages for certain foreign versions and the 17.5mm (Pathé Rural) and 9.5mm (Pathé Baby) versions.

We can add two further revisions by Gance :

  • Circa 1947 : Gance filmed several pictorial illustration shots and began a modified version of Napoleon Bonaparte, without continuing.
  • February 23, 1955 : Revival at Studio 28, Paris, of Napoleon Bonaparte (1935) in a revised version with a final triptych (using three screens).

Next came the restorations :

  • 1953-1959 : starting in 1953, Marie Epstein and Henri Langlois established a 19-reel version of the 1927 Napoleon. They would work to perfect this version, known as the « Museum Copy », until 1959.
  • 1969-1982 : Kevin Brownlow made a first restoration (6,630 m., 4 hrs. 50 min.), shown successively at different stages, notably in London from October 1970 (at the National Film Theatre) to November 1980 (at The Empire, Leicester Square) with BFI assistance. The actual premiere was held at the Telluride Film Festival in Colorado on September 1, 1979.
  • 1983 : Kevin Brownlow made a second restoration (7,155 m., 5 hrs. 13 min.), again with BFI assistance, this time also with that of the Cinémathèque Française, which gave him access to its collections. This restoration was presented at Le Havre on November 13/14, 1982, and in Paris at the Palais des Congrès on July 23/24/25, 1983.
  • 1991-1992 : Bambi Ballard worked on a restoration for the Cinémathèque Française, based on Kevin Brownlow’s of 1983 and using new elements she had discovered. This resulted in a version totalling 7,500 m. (5 hrs. 28 min.). This restoration was presented under the Grande Arche de la Défense in Paris on July 29/30/31, 1992.
  • 2000 : Kevin Brownlow finished his third and final restoration (reintroducing tinting and toning) using the new elements discovered by Bambi Ballard, leading to a nearly identical projection time. Presented at the Royal Festival Hall in London in June 2000, it runs approximately 5 hrs. 30 min., with a length of 7,542 m. To create this restoration, all of the known elements of Napoleon at the Cinémathèque Française (including the 1991 work print) were sent to Kevin Brownlow.

In all, 22 versions of the film have currently been identified, without including the versions created in the Central European countries and Russia (these are under study) and numerous distributor mutilations.

Discovery and Availability of Non-Film Archives

1) Cinémathèque Française collections

Starting in 2002, the Cinémathèque Française (formerly the Bibliothèque du Film) began a project of archiving the immense Abel Gance collection, which had been pending since 1966 because of personnel and logistics issues. From 2002 to 2010 Delphine Warin, who was responsible for processing the collection, patiently worked her way through these incoherent archives, whose secret had been lost by Gance himself while he was still alive. This collection, which currently totals 664 archive folders in 141 boxes, is made up of different sources dating from 1958 to 2009, as well as acquisitions and deposits made by Abel Gance himself or by the Archives Françaises du Film du CNC. None of these documents, which are now available, had been consulted for previous restorations. Invaluable film set photos and administrative and legal archives from the Cinémathèque are also included in the collection. Starting in 2007, when the Cinémathèque and the Bibliothèque du Film merged, the two film and non-film branches, which had previously been separated, were thus able to work together in total synergy, using a modern and dynamic restoration approach.

2) French National Library (BnF) collections

When Abel Gance died on November 10, 1981, Claude Lafaye was able to save the personal archives of the man to whom he had been devoted for many years. He kept them in his narrow office with overflowing cabinets at the CNC, and, with his passion for transmitting knowledge, made them available to all research workers. Thus, they became the basis for Roger Icart’s excellent biography, Abel Gance ou le Prométhée foudroyé, which is the current reference on the subject. Bambi Ballard had also used the archives, primarily to elucidate the legal imbroglio that bogged down the film, create the 1988 restoration of the sound version of Napoleon Bonaparte (1935), and re-edit the 1927 script, whose publication offers a thorough photographic representation of the 1992 restoration, enriched with very pertinent comments on her choices. But because an overall inventory was lacking, many of the secrets contained in these archives remained to be discovered. In 1995, Claude Lafaye donated the archives that Gance had entrusted to him to the Bibliothèque nationale de France (Bnf; French National Library). They were then inventoried by Emmanuelle Toulet, who made them available to the Département des Arts du Spectacle (Department of Performing Arts) when she left in 1999. These documents had not been consulted since 1991.

In April 2009, acquisitions from Nelly Kaplan were added to this collection, and included :

  • the last typewritten version of the Napoleon script, dated December 3, 1925, which we compared to previous versions ;
  • a filming notebook written by Gance ;
  • a dossier about the recording of the soundtrack of Napoleon Bonaparte (1935) using elements of the 1927 version.

New Digital Technologies

The object of this article is not to discuss the role of digital methods in film restoration. The colloquium recently held at the Cinémathèque Française (October 13-14, 2011) has provided a wealth of material on this topic. Providing generation gains, the ability to repair lost images using « patches » or interpolations, and the correction of certain types of deterioration, the « digital revolution » has profoundly changed our way of selecting which source elements to use for a final restoration. In addition, the simple fact of being able to « dematerialize » the elements (i.e., making systematic digital video captures at an editing table) enables us to compare them by simultaneously watching several video files, without having to manipulate the original element. Although Napoleon had been restored five different times, the systematic analysis of film elements and paper archives, supported by new digital technologies, enabled putting the history of the film and its successive restorations into perspective.


Napoléon vu par Abel Gance

Origin of the Film Elements Conserved by the Cinémathèque Française

The history of the constitution of the film collection relating to Napoleon at the Cinémathèque Française, established according to the study of the paper archives, has enabled us to better understand the scope and limits of the currently existing restorations. We tend to forget that Abel Gance went through a very difficult period from 1943 to 1954, when none of his projects were completed. In December 1948, a debt owed to the Kodak-Pathé laboratories caught up with Gance. He then contacted Henri Langlois for an « arrangement ». In exchange for a payment of this debt by the Cinémathèque Française, they agreed that the Cinémathèque would acquire the material property rights to the 40 cans of Napoleon that Gance had left with the laboratory as a financial guarantee. In return, Gance pledged to provide the Cinémathèque with a complete copy of the 1927 film (8,600 m., 6 hrs. 15 min., without the triptychs). It took Langlois six months to find the money and overcome administrative and accounting difficulties (an association cannot pay a supplier on behalf of a third party), and the arrangement was finally concluded in June 1949. But when Langlois and Marie Epstein received the material from the labs, they were bitterly disappointed to find that neither the Opera Version nor the Apollo Version were complete. To make matters worse, « even attempting to reconstitute a single version with reels from both films, it is impossible to obtain a coherent version which includes all the essential moments of your work. »

By 1953, they were patiently able to piece together as best they could what remained of the film. But Marie Epstein and Henri Langlois did what all later restorers would do : they mixed what remained of the two original versions to produce a single continuity. Nothing was surprising about this, a priori, because silent films were usually made from two closely similar negatives. This reconstituted copy was presented with triptychs at the Venice Film Festival in 1953, then at São Paulo in 1954. It greatly contributed to the rediscovery of Gance, whom certain people even believed had already died. With a great deal of difficulty, Langlois spent several years (1953-1959) trying to preserve Gance’s collection. These duplicate internegatives (and the surviving positive copies) were, except for the elements discovered by Bambi Ballard in 1986, the main material for all the film restorations that followed. Twice, there was strong tension between Gance and Langlois : in 1955 when the reworked version Napoleon Bonaparte was shown at Studio 28, and in 1969 when Bonaparte and the Revolution was created. Langlois was constantly obliged to protect what he had been able to save of Napoleon from Gance’s « creative fury ».

All his life, Gance constantly revised his original work to make it conform to new aspects of his artistic personality. Although he appreciated the tributes made to him, his past work had value in his eyes only if it benefited his new projects. « There is always a precipice behind me, » he used to say. The formidable conflict between the curator (Langlois) and the creator (Gance) nonetheless involved a sincere mutual admiration. Two documents bear witness to this, one official, the other personal :

  • In February 1968, in the midst of the Langlois Affair, Gance demanded that the new president of the Cinémathèque stop using Gance’s deposits immediately and ordered that they be at his disposal within 24 hours if he requested their withdrawal.
  • A second document, found during this appraisal project in Gance’s personal notes kept by his daughter Clarisse, contains a handwritten message by Gance on the death of his « friend and adversary » : « C’est grâce à Langlois que Napoléon existe encore. (« It is thanks to Langlois that Napoleon still exists. »)

Appraisal Method for the Napoleon Film Elements

We carried out our detailed examination in close collaboration with Laure Marchaut of the Cinémathèque Française’s Service « Restauration et Sauvegardes » (Restoration and Conservation Service). Her skill and patience were key to the successful completion of this work, which was carried out over several years.

From a practical point of view, to appraise the film collection in a reliable way, we needed to start from the stock database and continue by type of support, following the steps in the making of the film (from the original negative and first positive to the preservation elements).

Databases and Source Files

  • The LISE database :
    This is the CNC’s computer database for managing film collections. It has been used since September 2004 by the Cinémathèque Française to manage its own collections, and has gradually incorporated data from preceding cataloguing records.
  • Paper records :
    This contains the information we transferred to LISE. Each element corresponds to a cataloguing record filled out by the research librarians in charge of the inventory and the processing of the assessed elements.
  • A series of paper records preceding this organization which have been conserved at the Cinémathèque Française.


Based on a relatively simple idea, it consists of a series of intersecting two procedures :

a) Follow the steps in the making of the film and the generation of new material :

  • First, inventory the nitrate negatives,
  • Next, inventory the nitrate (or diacetate) positives,
  • Then, the nitrate duplicate internegatives,
  • the fine-grain acetates,
  • the acetate duplicate internegatives,
  • the triacetate positives.

b) Cross-check the data, going from the most recent files to the oldest

  • LISE database,
  • Cinémathèque Française paper records,
  • Previous files.

This approach took into account all film formats : 35mm, 17.5mm, 16mm, and 9.5mm.

c) We carried out the following chain of operations :

1/ Make note of all items in the LISE database listed under the title Napoleon (1927), then under the title Napoleon Bonaparte (1935), and finally under Bonaparte and the Revolution (1971), support by support
2/ Study and inventory (screening and/or listening, noting of film information, scan of images and film stock information if necessary, DVD capture on the editing table)
3/ Tracing search through the records existing before LISE
4/ Writing the appraisal report
5/ Delivery of the report and debriefing with film collection managers
6/ Checking of the information with the elements by the research librarians and preparation of a catalogue record
7/ Return to us for checking the final file record
8/ Final changes in the LISE database.

Next, we began a second search for all homonyms or similar titles, first in LISE, and then in all previous file records. This data was then also submitted to the same process of detailed checks in the chain described above.

d) Each item (can or box) was given a standardized label containing :

  • A general description: the title (on the can or box when it arrived), its former numbers, its current number, a detailed description of its contents, and a scan of frames if necessary
  • A list of film information :
  • Edge number: this is the number found on the edge of the film that enables mounting on the negative editing tables in compliance with a working copy. There is a start number and an end number ( e.g., G 25283 to 25353)
  • Manufacturer’s number: if there is no other way to date the film (e.g., 322 15 2 2)
  • Manufacturer’s code: the brand and the symbols that sometimes enable dating a batch (e.g., black and white, EASTMAN KODAK + (1934) NITRATE FILM).

If necessary, film information was scanned, including information that did not correspond to any known classification but which hopefully could someday be deciphered :

  • The results of the tracing search through all the cataloguing records and archive files relating to this element (from its deposit up to its current archiving) :
  • the conclusion,
  • LISE database corrections requested for inaccuracies found during our expert appraisal.

The final file made by the research librarians was the last step in our appraisal, and now provides the reference point for new database entries.

Napoléon vu par Abel Gance

Extension of the Expert Appraisal and the Discovery of New Collections

The extreme complexity of the collection soon became evident. To recover its coherence, it was necessary to have as broad an approach as possible. The Cinémathèque Française thus proposed to the Archives Françaises du Film du CNC that the Cinémathèque’s expert appraisal be extended globally to include all elements conserved at the CNC, subject to approval by the depositors. Since the CNC had encountered the same issues for identifying their elements, they proposed in April 2008 to transmit to the Cinémathèque Française, after agreement by the depositors, an overall list of all known elements relating to Napoleon. It was thus that the collection, which at the beginning of our appraisal project was estimated to be 250 cans, was now over 450.

But in January 2009, as we were preparing to finish our report, we discovered that one bizarre number actually represented a set of 179 cans. Why had they remained dormant ? In 1974 Gance had made a deposit to the CNC that was pre-inventoried in 1979. At that time computers were not yet used for inventories, and a number was given to the entire batch instead of to every single can. On top of this, the legal problems in which Napoleon had become mired had resulted in a freeze on the processing of this collection. By cross-checking the work orders we found, we realized that this stock had been cut in half. The other half is currently conserved at the Cinémathèque de Toulouse, and includes 202 cans. Thus, only six days before the date we were scheduled to submit our final report, 381 boxes that hadn’t been opened since 1971 suddenly reappeared !

Finally, on July 24, 2009, after a long personal investigation, we found the negatives of Bonaparte and the Revolution that were « sleeping » at the L.T.C. film laboratories in the name of Claude Lelouch’s Les Films 13, in spite of Lelouch’s donation of his rights to the film in 1989 to the Cinémathèque Française. When we contacted Claude Lelouch the following day, he immediately transferred the ownership of the set of elements to the Cinémathèque Française.

The Cinémathèque thus decided to continue our grand inventory project, to take into account the scrutiny of the additional Napoleon collections conserved at both the CNC and the Cinémathèque de Toulouse. The 179 cans from the CNC were appraised from March to August 2009. The 202 cans from the Cinémathèque de Toulouse were appraised on site (with photographs of each can, fortunately still in its 1971 packaging) during July 2009. At the end of 2009, evaluation reports on these collections were given to the CNC and the Cinémathèque de Toulouse.

Our expert appraisal on a national level was thus finished.

We were finally able to submit our report, which included four volumes totaling over 1,000 pages, two operational logs (500 pages), and over 200 DVDs with video captures. Data from 80% of the elements in the collection were modified in the LISE database. We had appraised 930 cans and looked at over 100,000 metres of film between 2007 and 2009.

Results : From Expert Appraisal to Restoration Project

For the first time, we have an overall view of the chaotic itinerary that this film followed from its creation up to the present day.

The Mixture of Versions

From the first elements studied, we observed that three Gance versions (1927, 1935, and 1971) had often been mistaken for each other, a direct consequence of Gance’s revisions for over 40 years. Since Laure Marchaut and I had both worked with Bambi Ballard on her different projects, we were able to give each element its original title. What was more complex was that in 1935 Gance wanted to evoke moments from the Empire that he had not been able to film in 1927. To do this, he borrowed excerpts from other films on the same theme. Some of these films had been mistakenly archived in Gance’s Napoleon collection. We therefore created a new category of works in the database :

  • Films other than Napoleon, improperly classified as one of Gance’s three versions. For example, a Napoléon produced by Pathé in 1909, or L’Agonie des Aigles by Dominique Bernard-Deschamps and Julien Duvivier (1922).

  • The « making of » documentary Autour de Napoléon, which Jean Arroy filmed on the set in 1927.

Which Version to choose as a reference ?

Since we were unable to screen the 2000 restoration, our access to the film « from the top down » was Bambi Ballard’s 1992 version. But what about the basis ? What should be our reference for the original version ?

Oral tradition and letters and criticism from the time have always praised the superiority of the Apollo Version, enriched with triptychs, the way the film was shown in a few large cities, after the trade screening for press and distributors. But no homogeneous film element was able to back up this assertion, because this « amputated, devastated, and dismembered » version had become an undecipherable mosaic mixed with other fragments.

This is when we discovered in the Cinémathèque archives a key document: the sequence inventory composed by Marie Epstein and annotated by Gance (with information transferred from a 1927 document) that was drawn up during their research from 1966 to 1969 to list all surviving elements just before the production of Bonaparte and the Revolution. It contained the continuity breakdown of the Apollo commercial copy in 36 reels, with the original length in metres of each sequence.

In addition, cross-checking with archived set photos enabled us to establish more precisely than ever the list of scenes that were written but not filmed, filmed but edited, or removed on release against Gance’s wishes. Studying them confirmed the accuracy of the document and removed all ambiguity. This therefore became our reference document.

We thus used two extremes: on the one hand, the 1992 Ballard restoration, on the other, the « Epstein sequence inventory » of the 1927 Apollo Version, in order to « corner » the elements as we progressed.

The Vertical Approach, support by support

We moved ahead using a « geological drilling » approach.

1) Reconstituting the Branches of the « Element Tree » :

The first step was to re-establish the links between elements on different supports. Which positive copy was at the origin of which duplicate ? Which duplicate was later used to make a fine-grain print ? And which element did successive restorers use ? From the beginning, the help of non-film archives was decisive, because we were finally able to bring together :

  • reports and correspondence between Gance and his editors (1925, 1927, and 1958),
  • printing instructions from 1927, and then from 1968 and 1971,
  • reports from witnesses who watched Gance fight against film mutilations (1928),
  • correspondence and production files for Napoleon Bonaparte (1935),
  • inside information on the  »Kodak Affair« ,
  • correspondence between Langlois and Gance from 1953 to 1977,
  • the list of preceding inventories,
  • correspondence between Marie Epstein and Abel Gance (1966/1970),
  • finally, all of Bambi Ballard’s notes and reports on screenings of elements from Napoleon, which gave a clear picture of the collection available in 1988.

All this information, which is a treasure that until now had never been used, would be used to cross-check the scientific examination of the films.

Let us start out by emphasizing two factors without which this work would have been impossible :

  • the importance of having kept in the cataloguing files a record of the item numbers used when these were deposited. Without these numbers, we would not have been able to link them to current elements and discover their origin.

  • the importance of the preservation material (internegatives or fine-grain prints), whose execution very frequently retained all the information written on the leaders and bands (edge code numbers, etc.). Sometimes two, three, or four edge numbers were found superimposed one over the other, which enabled us to identify the number of generations and which element (sometimes missing) this duplication chain had used.

All this information allowed us to « reconnect » each element with its source element in the best possible way, as well as to understand how they had been « broken down » in subsequent editions.

2) Support by Support, differences appear :

The 1927 negatives were lost during the production of the 1935 version. During this expert appraisal, only a very small amount of the original footage therefore was found. In spite of this, the appraisal enabled the recovery of 80% of the original negative of the « Marseillaise » scene at the Club des Cordeliers, and the « rushes » of the « Bal des Victimes ».

The most original elements would thus be duplicates of release positives or nitrate or diacetate work prints.

In addition to work elements from the time (title cards, durations, emplacements, interpositives for 17.5mm and 9.5mm, etc.), a significant part of the release copies were found, among them probably a few surviving tinted and toned reels of the original Apollo Version. In addition, since April 2008, with the assistance of Jean-Pierre Mattei, Vice-President of the Cinémathèque of Corsica, we have been able to examine the very beautiful print found by Bambi Ballard in 1987 at Ajaccio. When compared to the duplicates made by Langlois in 1953 and 1959, these elements allowed us to gradually discern the differences in artistic treatment between certain scenes.

The fine-grain masters discovered in the material from the CNC and the Cinémathèque de Toulouse provided us with invaluable information on the steps in making the film.

Bringing together and comparing all the stock has enabled us to reconstruct the 1991-92 production chain. Each link in this chain opened the door to the preceding one : first, Brownlow’s 1983 restoration, and then Gance’s versions of 1970 and 1935.

This approach led us to speculate on Gance’s attitude between 1968 and 1970. In view of the cuts found, he had access at that time to the original tinted copies of several important sequences. He told this to neither Kevin Brownlow for his 1969 restoration, nor to Langlois or Marie Epstein, keeping them to be reused in Bonaparte and the Revolution. These « treasures » can no longer be found. All that remains are the editing cuts from the fine-grain copy and the duplicate positives and internegatives discovered among the stock of the CNC and the Cinémathèque de Toulouse.

« There is always a precipice behind me… »

For more than two years, we climbed up from the bottom of this cliff, step by step, hoping to reach the top and finally understand the film’s original structure. To do this, we needed to complete our vertical approach (by drilling) with a horizontal approach.

D. The Horizontal Approach

Simultaneously, using the Epstein sequence inventory for the Apollo Version, we gradually established a table as our work progressed. For each scene, this table indicates the box or can and collection where we found the images.

Each column mentions the number of the item where the images for each scene were found. The images are listed line by line. (This example shows only part of the columns, and does not take into account that the sequence breakdown starts again with each new format : 17.5mm, 16mm, 9.5mm.) This sequence breakdown provides a scene-by-scene list of the 36 reels in 35mm (without triptychs) that made up the original Apollo Version.

Using this comprehensive document, we were thus able to synthesize an overview of the « pulverization » that the original film underwent throughout its history and determine which elements contained the surviving parts. On the basis of this information, we undertook the last phase of the appraisal.

E. Overall Comparison, scene by scene

This phase took place in the third year of the appraisal project, during the summer of 2010. The 200 video-capture files made during the inventory were imported to our computer to carry out a simultaneous and methodical screening of all the elements, scene by scene.

This comprehensive, systematic, and simultaneous comparison, which until now had been impossible to do using a silver support, allowed us to clearly show :

  • which editing had been done, and by whom, when, and using which support (1935/1971 by Gance, 1953/1959 by Marie Epstein and Henri Langlois, 1983 by Kevin Brownlow, and 1992 by Bambi Ballard) ;
  • the existence of two negatives with different artistic treatments, and to which cut they had been mixed ;
  • that, for want of having at their disposal all the necessary information, the preceding restorations were essentially based on aesthetic considerations or the technical imperatives of linking shots together.

Finally, the comparative method and the reconstitution of the production chains enabled us to group shots together according to their original version (Opera or Apollo). Thus, we were able to observe very marked artistic differences between the two versions.

Napoléon vu par Abel Gance

Differences between the Opera and Apollo Versions

We are able to provide only a few examples here, because listing all the differences discovered would take up too much space in this already extensive article.

A. Example I : Scene known as « Les Ombres de la Convention » (« The Ghosts of the Convention »)

When Bonaparte comes to meditate in the deserted chambers of the Convention before leaving for Italy and the ghosts of the Revolution appear and speak to him, the Opera Version and the Apollo Version diverge.

In the Opera Version they appear superimposed on a static shot of the Assembly.

In the Apollo Version, the ghosts appear to Bonaparte via dolly shots while asking their question.

In the Opera Version, the lack of movement creates a confrontation. In the Apollo Version, the mobility of the ghosts literally « wraps » Bonaparte in the heritage of the Revolution.

B. Example II : The Geography Lesson at Brienne

At the end of a geography lesson on island climates the teacher talks about Corsica, and then casually mentions St. Helena. In the Opera Version, young Bonaparte faces the blackboard with St. Helena drawn on it and becomes pensive. In the Apollo Version, in front of the drawing of St. Helena, he looks from one blackboard to another, then from Corsica to St. Helena, from the island where he was born to the island where he will die. This « short circuit » between the two limits of his life is what makes him pensive.

The character’s inner motivation is different and so is the artistic choice: it expresses one of Gance’s favorite psychological states, which he called « the memory of the future ». To these differences, we must add the wider framing in the Apollo Version, with wider and bigger shots. The Apollo Version thus acquires a different rhythm, more flexible and with greater internal homogeneity, which has not been seen since 1927.

In light of our study, certain legends were confirmed and some mysteries were elucidated.

C. « The Music of Light » : A Legend ?

Gance stated that he sometimes edited his images like a music orchestrator. For a long time, his often exalted prose made readers think that this was an exaggeration. However, during our appraisal we found the original negative of the « Marseillaise » scene, and the famous passage where the screen literally « sparkles »«  [an effect arising from the flashing images of Gance’s rapid editing]. A thorough study showed that the editing process is carried out in an intelligently rhythmical manner, and that at the scene’s climax each image was carefully chosen and numbered… in the same way that a composer fills in the notes on the staves of his score. ance’s concept of « the music of light » is thus a reality.

D. A Mystery Solved

A 37-second shot placed at the climax of the « Double Storm » sequence was a mystery. From a superimposed background, Bonaparte abruptly emerges, cool and unshaken, looking like a stoic eagle from the left to the right of the screen. Examining these elements, we were led to consider a cinematic effect that has now been lost: the meaning of the sequence of Napoleon mastering the unchained elements depends on a triple-screen framing, with complementary images on the left and right depicting the turbulent human tides of the Convention. Bonaparte seems to appear suddenly from the maelstrom, when in reality his sharp looks to the right and left  targeted  side screens that no longer exist.

The title card preceding this shot paved the way for this effect: Bonaparte was indicated as a « tamer of oceans ». We can give an idea of the desired effect with the following photomontage. The effect of this setting (a contrast between tumult and stability) echoes the film’s prologue at Brienne, where the young Bonaparte keeps absolutely still in the midst of the snowball fight, thus expressing his mastery of warfare.


Since 2009, several foreign film archives (all affiliated to FIAF) have been contacted concerning a worldwide inventory of Gance’s work which he himself carried out in 1977, as well as a second inventory personally conducted by the author with Gance’s daughter Clarisse in 1988. FIAF launched an appeal in late 2011. The results of this international research confirmed our conclusions and completed them.

In spite of what we can call the « Terror » (and this is not just a play on words) that the Napoleon collection long represented for the world of conservation, the Cinémathèque Française could not leave this cinematic monument to its mysterious fate. They realize it is necessary to start from the very beginning of the story of the building of this film in order to tame it, and to open it to the new techniques now possible in film investigation and restoration. Through the unveiling of new elements, what was first an inventory and then an appraisal has put forward a new possibility – the necessity even – of undertaking a reconstruction of the two original versions of the 1927 film, which would lead to a new restoration.

The work of reconstruction has now started. It leads us to hope that we can restore the cinematic monument represented by Napoleon as nearly as possible to its original versions, in particular the legendary Apollo Version. We would have the advantage of being able to go back to four generations of image quality and a montage now rediscovered thanks to three years of patient, painstaking exploration of films and archival material conserved in heritage institutions.

The Napoleon Comet could soon be back, ablaze with all the original clarity of its « music of light ».

Joël Daire est directeur du patrimoine à la Cinémathèque française.

Georges Mourier est réalisateur et chercheur. Il a tourné, entre autres, un documentaire consacré à Abel Gance, intitulé À l'ombre des grands chênes (2005). Il a été chargé par la Cinémathèque française de la reconstruction et de la restauration du Napoléon de Gance.