1 Framing the Digital Witness

In early 2020, the University of Udine signed with the Instituto Luce (Luce) a collaboration agreement aimed at digitally restoring a supposedly lost expedition film: La Spedizione Franchetti in Dancalia (l.t.: Franchetti expedition in the Danakil, 1929).¹ Luce and the University of Udine brought to light a large number of film fragments that included 35mm original negatives, DupPos Lavenders, Positives, and a 9.5mm reduction print, but there was no trace left of an edited version of the 35mm film.²

No international version was found nor documented, and no reliable, detailed reference about any complete edition survived.³

The Pandemic has forced the project to shift remote and forbid working on the original film materials. Therefore, we have dedicated a preliminary inspection to edge-to-edge digital scanned copies of every element from an assemblage of all the fragments into four reels. The Pandemic has produced a real-time simulation of a condition in which accessibility to the analogue witnesses of the film artefact is prevented. It has inspired philologists working on this film reconstruction to further consider the functions of the digital witness and the tradition of the film object and text.

In particular, this article will not present the final outcomes of the film restoration process. It rather discusses the preliminary questions solicited by the attempt to reconstruct a critical edition of this film: an experiment in phylogenetic and philology in a digital environment. Inside this “simulation” and for our philological purposes (an in-depth inspection and critical analysis of all the elements), we have requested a specific kind of digital transfer. Not yet a DPX scanning for digital intermediate (DI), nor a mere proxy access copy with the sequence of the frames, rather an integral, full-width edge-to-edge scan of every element. This article proposes that, under certain circumstances, a digital copy can serve as the authentic material and textual witness of a film, and that it can be used to excavate for philological and critical reconstruction. In doing so, we will also stress the current limits of such an approach and question how and when the recurse to analogue film originals is still inescapable.

2 Authenticating the Digital Witness

The digital intermediate usually introduces in the film restoration process a consistent amount of digital copies: raw digital tests, failed digital files, ProRes access copies, other than DPX files for post-production. In the critical scrutiny of what Lachmann’s ecdotic calls the tradition of a textual artefact, any physical inscription that transmits and conveys a given text is considered a witness. It can be included in the collection (recensio) and comparison (collatio) of its copies, whose innovations (i.e. artificial errors as well as material damages) determine correlations and genealogical relationships. Nevertheless, to what extent we can consider a digital copy a faithful witness of a historical film?⁴

Rudolf Gschwind, in a pioneering account, defined the kind of digital copy we have requested as “digital fac-simile”, or “a digital representation of the complete film strip as opposed to the sequence of frames provided by conventional film scanners (which ignore the rest of the film strip area)” (Gschwind 2002: 174). This article will explore the theoretical and operational implications of the digital fac-simile meant as a proper “digital witness” in the process of a film reconstruction. Indeed, the literature consistently emphasizes the DI process in film restoration – or the “film-to digital-to film” workflow (Fossati 2019: 63) – in questioning the muted approach to the authenticity of digital copies of films (Fossati 2019: 67). Conversely, minimum attention has been paid to particular kinds of digital copies such as the “digital fac-simile” or “a digital representation of the complete film strip” as an authentic critical witness of the film and an eligible artefact for film critical and philological inspection.

Although the issue of authenticity related to the digital copy of film has been widely debated in the film archiving sector, we will address this specific question from an ecdotic and philological perspective. We then stress a move from the film restorer to the film philologist, questioning the inner nature of a film witness.

According to Frank Kessler, each copy entering an archive is a document of its history (Kessler 1995: 30). Nevertheless, from the film philologist’s perspective, the authenticity and indexicality of the digital witness require specific evidence.

How specific this approach can be demands a technical reconsideration of such a term as “witness”. Bearing on a definition introduced by Cesare Segre in the domain of linguistics and philological semiotics, the witness is a “diasystem” where the linguistic and ideological system of one text's author intersects one of the copyist, or who (or what) produced a copy of that text (Chiesa 2007: 44). Therefore, one who inspects such an artefact must isolate and distinguish each original/copy manufacturing system, questioning the history, mentality, and culture that produced such an object. Leaving apart the strictly linguistic origins of the witness notion, this classical multilayered comprehension of the nature of a copy helps us reframing the debate about authenticity and indexicality in the philology of digital film copies.

In one of the first and most compelling arguments on historicity in the digital era of film, Philip Rosen stressed how the core question in debating the indexicality and authenticity of the digital copy relies upon a similar dual system of intertwined authorities. Here a profilmic event and a prodigital event require equal critical consideration. In other terms, Rosen stressed the fact that the “authority of the referent” can be excavated into the digital both ways: tracking down the “profilmic ‘here and now’ (or ‘there and then’)”⁵ (Rosen 2001: 306), and the prodigital event attesting to the encoding of “light intensities as numbers on a magnetized substrate” (Rosen 2001: 308). More particularly, Rosen suggested moving from indexicality stricto sensu to the “capacity of digital to imitate such preexisting compositional forms of imagery” in terms of digital mimicry.⁶

Digital mimicry is precisely the quality of digital copying and the argument Rosen raised against the ideological opposition between indexicality and the digital, a kind of idealization of the digital that would definitively prevent any assumption of credibility of a digital copy as a witness.

Conceptualizing a digital witness means then – as Rosen put it – “de-idealizing the digital” and stressing the hybrid temporalities and materialities overlapping into the digital copy: mimicry convey impure temporal and material histories, attesting the intrinsic hybridity status of the digital, its capacity to indicate and refresh the authority of its referent(s) – the material determinations of the profilmic and the prodigital – and, more importantly, it repristinates a specific criterion of credibility that proves how digital imagining “is not separable from prior histories of mediated representations” (Rosen 2001: 314). Digital mimicry would attest to the credibility of the digital fac-simile as a witness in film philology. Matthew Kirschenbaum went even further in sustaining this aspect when, relying on Nelson Goodman semiotic category, stated that “digital computers support and sustain an ideal allographic environment” (Kirschenbaum 2008: 135), namely, a state in which a given object fulfils its ontology in reproduction through what he called an illusion of immateriality. This has allowed us to trust in the “sameness” of the composition and representation of the Spedizione Franchetti film as text. What is at stake here, to paraphrase Kirschenbaum, is the formal materiality of the digital witness, meaning the formal symbolic manifestations elicited by a certain software. The illusion of immateriality Kirschenbaum is stressing, in fact, is given by the fact that “technology must produce perfect outputs from imperfect inputs, nipping small errors in the bud” (Kirschenbaum 2008: 133). While continually restoring the signals to their optimal state, the digital technology perpetuates a conservative environment where transmission of the film as text is guarantee and credible: a completing argument to Rosen’s digital mimicry.⁷

For what is at stake here, it has to be said that Rosen insisted on three levels of preexistent authority that the digital witness carries: the actual pro-filmic event itself, the manifestations of film as text and material object – what Barbara Flueckiger described as “recording process in combination with subsequent development, editing, optical works, colour grading, and printing” (Flueckiger 2012) – and the digital transfer (and subsequent operation in post-production). The individualization of the physics of the media apparatus and its operationality is thus crucial in defining the diasystem (S1=analogue indexing + S2=digital mimicry). According to that, excavations into the digital copy elicit forms of reverse engineering aimed at reconstructing specific technological operations that occurred in what Flueckiger defines as the first two histories of the film object⁸ (Flueckiger 2012): during the filming process (i.e., camera magazine’s film loading procedures, camera exposing operations etc.), in the printing phase (i.e., Light printing protocols), and an initial stage of editing (cuts and editing style). Conversely, the same attention must be paid to the technical procedures in the digital environment during the film transfer to digital.

That stated, no proper technical standard has been defined yet by the literature to produce a digital copy eligible as a philological witness belonging to the film tradition. Differently said, on the one hand, the authenticity argument about the digital mimicry could serve any digital copy of the film. Nevertheless, at what conditions a digital copy deserves to be formally included in the film copies scrutiny as a new witness (and a proper material object taking part in the film tradition)?

The digital witness we are considering for this article is a ProRes “digital fac-simile” of original negatives. This is assembled according to a general linear sequence, inferred by the archivist from the location of the reels into numerated cans. A general sequential numeric series localized on the side of perforations preliminarily guided the assemblage. However, not (yet) a real critical motivation led to the alignment of the fragments at this stage of the digital transfer. No edited versions survived to suggest a preliminary alignment historically justified, so this digital copy merely bears trace evidence of the appearance of the complete film strips as they were in the Luce archive’s storage. Furthermore, although an essential technical intervention of film repairing was made before digitalization to guarantee a safe passage through the scanner (a DObserver Cine7-D1 inspection table adapted for scanning), no digital post-production was applied then. Hence, this particular digital witness belongs to the mere S2=digital mimicry system,⁹ with the specific characteristics of an edge-to-edge digital fac-simile and the quality of an access copy. Yet, no DPX digital transfer was required at that stage. In fact, for critical inspection and excavations into the witness, an access copy quality would have to guarantee enough information and more suitable manageability for analysis and the comparison of fragments, for instance when working in the DaVinci Resolve software developed by Blackmagic Design.

3 The Prodigital and the Scanning Procedure

As we defined and processed it, the digital witness results from transforming a historical film as a tangible object into digital data. The full-width edge-to-edge copy is an inescapable characteristic that allowed to excavate a mimicked-trace-evidence of the pro-and-filmic event, thus providing a valuable fac-simile for the original film negatives, and a crucial support for a preliminary philological investigation aimed at film reconstruction. Nevertheless, the digital witness resulted from the film scanning is also introducing into the film’s tradition a new physical trace of the film with its proper qualities and innovations.

Flueckiger defined scanning as “a simple process of photographing; it is analogue in the first step, meaning that there is a proportional connection between incident light and the electric charge generated in the sensor. Only afterwards are these values mapped on a discrete scale by quantization and then assigned binary values by encoding the voltage into mathematical data” (Flueckiger 2012). This process is not neutral. It produces innovations into the film as a text and the film as a (new digital) material object incepting the film’s tradition. As Kirschenbaum highlighted, digital technology essentially “restores signals to near perfection at every stage” (Kirschenbaum 2008: 133) while continually producing innovations into the film witness as a digital material object. “The mathematics generates a bit sequence that serves as a redundant expression of the original data; if the two fail to match up during a read or write task, then an error is indicated, and the task is repeated. Users never see such errors, which are detected and corrected in the space of milliseconds. This contributes to the way in which the drive is perceived as an abstraction identified only by an arbitrary volume letter (‘C’) or an icon on the desktop” (Kirschenbaum 2008: 133).

If, on the one hand, the illusion of immateriality through signal and errors correction has allowed Kirschenbaum to talk of an ideal allographic environment for the film to be conveyed, on the other hand, these corrections are innovations in the forensic materiality of the digital witness. This can be excavated deep down into the range of small, localized glitches in the code and the surface perceived “at the junctions” between the analogue and digital states of film. The scanning procedure is primarily impacting such innovations. As Flueckiger put it, “the scanning is dependent on the properties of the light source(s), the camera properties, and the transport system applied to the film” (Flueckiger 2012). Any innovations produced in the scanned file are a significant intervention into the film's history and endangers its authenticity as a record (Flückiger 2012). If digital technology is producing innovations restoring signals, the scanner settings may also be responsible for producing its errors and damages. A first example concerns the creation of artefacts in the texture of even real digital glitches due to acquisition errors or to the scanning process speed. In the digital witness of Spedizione Franchetti, one frame has been wrongly digitized, and a portion presents inverted colours.

A second example concerns the identification of duplicated frames: crucial information here is the edge code reference in the peripheral area of the film, which allows clearing identification of duplicated passages.

These kinds of innovations are impacting simultaneously on both the formal and the forensic materiality of the film witness. Thus, these specific innovations are not easily distinguishable as textual errors (modifying the textual and formal materiality of the film) or physical damages (impacting on the material root of the code). Once again, the hybridity of the digital witness is attesting its specific nature.

These preliminary notes are stressing that the production of a digital witness for philological investigation is demanding a standard formatting procedure that is still missing in the ethics of film preservation.

The following paragraph will excavate the profilmic and filmic events the digital fac-simile enabled to reconstruct, emphasizing the temporal complexity and the historical overlaps in it. Hence, we will question what historical information we could excavate from such a digital copy and how this approach can help copies scrutiny and the forensic film reconstruction of archival films. In particular, these excavations into the digital witness are about the nature of film as a text (i.e. the exact sequence of frames, or the origin of errors in the content of the image), as well as a material object (i.e. the historicity of damages, or defects in lightning).

The conclusive paragraph of this article will then address the specific and verified parameters that should determine the digital witness for philological proposes (differently meant from digital restoration purposes), and the issue of the unicity of the digital witness. It will finally discuss how institutions and film archives can contribute to each digital witness's “status of authenticity” (Elsaesser 1997: 207).

4 Excavating the (pro)filmic

During the recensio and collatio of the Spedizione Franchetti’s fragments from the Luce’s archive, we planned an inspection of edge-to-edge digital copies on QuickTime and DaVinci Resolve. Every occurrence has been systematized and transcribed onto an Excel sheet. However, the excavations of the digital witness through multimedia readers and post-production software have limitations. Every new digital witness brings a series of innovations and error corrections that risk collapsing into one undistinguished plain scanning process and the digital mastering of the original data. For example, to make a distinction of physical damages or lacunae we find in the original, from the digital innovations resulted from the attempt to “compensate” them (through the settings and formatting of the digital output) is challenging. Accordingly, the philological inquiry always demands peculiar hermeneutical attention. That stated, we questioned the digital fac-simile to excavate a genealogy of the film analogue copies and the history of the film production and circulation: in particular shooting, printing and editing, exhibition and hints of different film versions. Furthermore, to flesh out this philological/hermeneutical scrutiny, we provided evidences based on historical film-related materials, such as the travel diaries and expedition reports.

Shooting. The frame stabilization procedure on the digital witness first proved crucial traces of the shooting habits. In the historical journals’ accounts of the expedition film, director Mario Craveri explained that during the shooting, he used “one 120 meters movie camera with a single tripod; plus, two other portable cameras of 25 meters of film each.” [Translated by the author] (Craveri 1936). This appeared evident in the digital witness, when we noticed two different dynamics of frame instability while scrolling the digital fac-simile: these were corresponding to two different kind of matte shapes. Indeed, we realized that a vertical instability occurred in correspondence of sequences shot by the bigger and more stable camera with a blunt corner matte.

Instead, the two smaller cameras had pointed corners mattes, which we could recognize, isolate in few shots and relate to tilting instability caused by the crank manual use.

Still focusing on the pro-filmic event, the sequential order of the assembled fragments has been questioned through an integral transcription of the edge markings and numbers series, then systematized onto an Excell sheet. This strategy crucially revealed further aspects of the shooting habit. We noticed that the edge numbers scrolling order changed. Sometimes the edge codes proceeded in ascending order, and otherwise in decreasing one. This lets us infer about operational habits during the loading of the film in the camera magazines. Since the smaller cameras could load only a shorter film strip roller, the film needed to be prepared – blindly – putting the hands in a black sack (during the travels in the desert!) to safeguard the film from the light. This pragmatical habit suggests an “in reverse” film loading process. On the contrary, bigger camera’s magazine was usually loaded correctly, so that the ascending order is the one we can usually correlate to the shooting with the bigger camera – with few exceptions. In fact, the circular mattes rarely correspond not only to the ascending order but to decreasing one too, as the two extracts make clear.

Printing and editing. In the systematic scrutiny of the edge markers, we identified two sequences of handwritten numbers aside from both lines of the perforations. Through their organization onto the Excel sheet, we could compare and formulate an interpretation.

Both appeared to be related to the notches: the shallow cuts along the edge of a film used to trigger the change of printing exposure in a film printing machine.¹⁰ The recurring numbers on the left side are written only at the beginning of one scene and are recursive: they usually go from 1 to 18 or at least 23 and then start again from 1, so that in each reel, we could count up to eight recursive numeric series; on this side, the handwritten number is located at the fourth frame after the splices, while the light notches are between the sixth and the seventh frame. On the right side, instead, a new numeric series is progressive along with the full extension of the four reels, from 1 to 517. Each number is usually marked both at the beginning and at the end of one scene. Here the splices precisely match with the notches and the numeration.

Moving from the archive documentation of the Società Geografica Italiana in Rome, we could make some hypothesis about this numeric series. Indeed, back to Italy, Franchetti himself reported about the expedition at a conference at Teatro Augusteo in Rome¹¹ on November 20^th, 1929, when he screened some scenes from the film. According to that, we firstly inferred that the recursive numeric series was standing for an episodic editing structure of a film spitted into six to eight modules/episodes, to be separately screened as educational and illustrative moving slides in a conference. A second interpretation came from the lab technicians of the Istituto Luce. They directly put in relation the notches’ shape with the numeric series, suggesting that a double print process could justify them. The position and the shape of the notches on the left side could be markers of a first printing procedure in shorter blocks. According to one printer's limit, the operator could probably manage only 90-100 meters long strips of footage at time, and this could motivate the recursive numeric series on the left side. Each short sequence of footage was then printed separately.

Instead, the numbers and the notches on the right side correspond to the editing line of the entire film. Sequential order is suggesting a first original editorial scheme. In a second printing passage, the film has been probably processed integrally. We could find a similar double notation system procedure in another colonial film of the same period, preserved at the Luce archive: Nelle oasi. Viaggio dei sovrani nella colonia d'oltre mare (l.t. In the oases. Journey of the sovereigns to the overseas colony 1929). This print brought the same notation system on the film borders. These double series of notches also suggested that printing followed two processes: a first one in shorter separated blocks and a second one (in a time to be questioned) with the final editing sequence. In fact, we found superimposed mattes produced by the reiteration of the printing process. Indeed, the first printer left a more internal and blurrier sign, as the second printer probably had a more oversized matte that appears here more external and sharper.

Unfortunately, to further confirm this hypothesis, we needed a direct observation of the original negatives. In fact, to date the handwritten numbers, we should know with precision weather both had been written with Indian ink during the initial printing stage, or they were more recent signs made with a marker. So, we asked that to the Istituto Luce’s technician who, through a magnifying glass and by using a swab soaked with alcohol, could confirm that both the markers were written with Indian ink in the same historical period.

Edition and Exhibition. Some historical and critical discrepancies left doubts about the editing scheme and following versions of the film. Indeed, we know that Spedizione Franchetti was screened on many occasions and cities in Italy between December 1929 and May 1930. Although an official historical report about the film mentions a division in six parts (Istituto Nazionale LUCE 1929a) (that would confirm the hypothesis of an editing structure in six modules, eventually attested by the recursive numeric series), this description does not match with the intertitles list published in the Catalogo generale di soggetti cinematografici (Istituto Nazionale LUCE 1937). A look through the numeric series systematized into the Excel sheet made patent some changes in the editing style. Indeed, the long and continuative numeric series starts only after the first recursive numeric series. In fact, the film's very opening holds numbers only on the left side of the film borders. This first “module”, with recursive numeric series on the left, shows Barone Franchetti greeting and shaking hands with the chiefs of the Ethiopian communities. Number 1 of the long and continuative numeric series (on the right side) starts with an animation sequence with the exploration itinerary on a map of Ethiopia. This opening animation sequence seems to be more justifiable as a film introduction. In contrast, the first module seems a later occasional insertion motivated by political and propaganda reasons stressing the utility of those colonial expeditions: as the numeric series suggests, the film's very beginning is the animation sequence. Crossing mismatching of scenes descriptions in the historical documentation and the systematization of the numeric series in the peripheral sides of the film unravel many suspicious editing changes.

Moreover, in a dubious historical time, some scenes have been lengthened through duplicates from a positive print. This intervention attempted to create a more complete and more extended version of the film, a kind of restoration, adding frames by duplicating them directly from a positive release print (which is currently missing from the archives), as these frames are missing from original negatives. It can be inferred from the digital witness, looking at the edge code. Here the original Agfa film (with original, first-generation negative frames) alternate a Gevaert Belgium (with duplicated frames that has a lower contrast and image’s quality besides an over exposition of them and even prints of original notches)

A hypothesis still under investigation is that a new version was prepared by Mario Craveri and Enrico Gras for the RAI (Radiotelevisione Italiana, the Italian national television) documentary of journalist Ettore Della Giovanna Raimondo Franchetti. Esploratore della Dancalia (l.t. Raimondo Franchetti. Explorer of Danakil, 1963), though no evidence of this re-edition was found in the RAI archives.

Further clues in the original negatives may lead to different versions of the film. We identified two different film insertions, although not undoubtedly identifiable.

We found a very short fragment of a sound film, Ferrania, which is probably a residual left of a subsequent reprint or addition, as Spedizione Franchetti is undoubtedly a silent film (La Greca 2017). The second is a different AGFA edge code, accompanied by a plain black point, which is not datable without approximation since this kind of code appears from 1925-1926, in new low-contrast, fine grain emulsions intended for duplicating or effects (Brown 2020).¹²

It is arduous to make an indisputable examination of the splice structure, shape, and thickness from the digital witness. It is a crucial aspect to focus on during an inspection of the physical witness. Moreover, in our specific case study, it would have helped to date the insertion of the duplicated elements.

We will need to discriminate which sign ought to be preserved and amended during the digital restoration phase, starting from the challenging distinction between analogue and digital damages and errors in the digital intermediate process. Thus, recurring to the original negatives is crucial.

However, we want to stress that, besides limits of information concerning the physical specificities of some material dimensions of the original, changes in the digitizing settings may create a different “witness” and produce textual and material innovations.

5 Conclusions

Under certain circumstances, a digital copy can be considered an eligible philological witness for film philology, phylogenetic and film reconstruction. Once authenticated, the digital witness takes part in the film's tradition as a new artefact and digital film material object.

Furthermore, the full-width edge-to-edge extension is a mandatory feature for recognizing and crediting a digital copy as a philological witness. Together with the digital copy's peculiar ontology, the full-width edge-to-edge allows an ideal allographic environment when the profilmic and filmic event can be excavated and reconstructed together with the individual digital innovations produced into the digital environment.

The standard resolution is a matter of debate. Depending on the depth of the inspection requested, a DPX may not be necessary, and a lighter, more manageable file is to be preferred. Nevertheless, it should be considered that scanning full-width under a specific resolution limit of 2K “does reduce the resolution of the image area to an extent which is visible when compared to using 2K for the image alone. This made the results less sharp in comparison” (Flueckiger et Al. 2018: 53).

A proper scanning speed is also a decisive matter of debate. This can prevent or limit the formation of glitches and the duplication of frames. That stated, glitches and digital innovations will inevitably take part in the film's tradition, and they should contribute to the individualization of that digital copy.

Finally, the digital witness must be unique. Giovanna Fossati reminds us that, within the film archiving working chain, assuming “a digital intermediate stored on a digital tape or a hard drive, as an ‘original’ is problematic”. Fossati also emphasized that the digital intermiediate is “unique, but it can be ‘copied’ without quality loss”, as it is not dealing with duplication, rather with data migration (Fossati 2019: 166). While this is typical evidence of how the “film as original” framework strongly influences the film archives, from the philologist's point of view, this raises some critical issues, not lastly an impact on the necessity – in the critical scrutiny of the copies – for each witness of the film tradition to be unique. Accordingly, data migration has to be made scarce. Differently said, it is necessary that the state of authentication and verification of a given digital witness has to be promoted and certified by the institution holding the originals. As part of a new generation of copies, any digital copy chosen as a witness should be treated as a digital non-fungible token. According to that, independently from the migrations of the same data, any digital witness produced should be accompanied by a certificate or a contract attesting to the scanner settings and the temporalization of the data production. This is an argument still under debate in the film archive sector (Hutchinson 2021). However, such a perspective on the utility of the digital witness and its historical values would demand such a way to rendering scarce a migrating ensemble of data.