This paper was first presented at the Colloquium “Valoriser les archives des ethnologues Usages contemporains des collections”, BNF, Paris, 4-6 October 2018. Parts of it featured in a talk also in October 2018 at All Souls College, Oxford for the Open Science Seminar convened by Lisa Lodwick and Jasmine Nirody. Another, shorter, version was presented remotely in November 2018 at a AAA Roundtable “Approaches to Expanding the Use of Anthropological Archives” convened by Diana Marsh. A final version was presented at L’école d’été UMC – Marseille “Understanding Mediterranean Collections”, 17 July 2019. I am very grateful for all the invitations and for the feedback from these presentations.
1An image haunts this paper, that of the fire on the night of 2-3 September 2018 that destroyed Brazil’s 200-year-old National Museum and, with it, the ethnographic archives of Brazil (Figure 1).
fig. 1 – Fire at Museu Nacional, Rio de Janeiro, 2-3 September 2018
Photo Felipe Milanez, licence CC BY-SA 4.0; source: https://commons.wikimedia.org/wiki/File:Fire_at_Museu_Nacional_05.jpg
- 1 That this is not a lone case was sadly demonstrated in April 2021 in South Africa when the Universi (...)
An obvious lesson to draw from this example is, as the proverb goes, “not to put all one’s eggs in one basket”. Despite such cautions, older analogue materials may make this unavoidable: economies of scale are such that a single centralised archive may be possible while many distributed archives may not be practicable because not affordable. However, in some cases, and almost always with digital material, mitigation is possible by making copies. Another lesson that the tragic case of Brazil’s National Museum1 suggests is that centralising is sometimes dangerous, and so perhaps should be resisted: the hope is that dispersing the files will keep some safe, but at the possible price of inaccessibility (and lack of findability). I will leave this terrifying case as a framing context for the paper, but will return to it at various points during the discussion.
2In this paper I want to explore two contradictions that frame discussions about archiving ethnographic material. The first is about centralisation and economies of scale. The example of Brazil suggests centralised archives are unsafe (if a centralised archive burns, all the records are lost), but they do offer consistency of treatment, cataloguing, indexing and curation, as well as some fairly obvious economies of scale. They also provide centres of expertise for smaller archives. However, in addition to increasing vulnerability to catastrophe, the decision to concentrate resources into large, centralised archives also goes against the ethos of a decentralised, distributed world in which a thousand flowers are left to bloom. The second contradiction is in ethics, where it may be argued both that archiving is unethical and that not archiving is unethical (so archiving is an ethical obligation).
3As we shall see, such tensions push in many different directions. I will now discuss these points at greater length, concentrating more on the ethical issues than on those concerning centralisation. I provide examples from my own ongoing research practice.
- 2 See De Largy Healy’s paper, this volume.
- 3 See Armin Geertz (1994) for a case of Hopi Indians in the USA burning traditional altars.
4Against centralising, one could invoke some community-based archives. These empower neglected actors who have created research materials to hold them in their own local archives, or in some cases, old archival material is returned to source communities so they can curate it.2 There are some wonderful initiatives, but from the outside we must ask what happens if and when the enthusiasm is lost? If a community’s priorities change, then what will happen to the archives? A group may convert en masse to a form of religion that disparages previous traditions to the point of wishing their destruction.3 There is a delicate and highly political set of issues here. What responsibilities have people to, for instance, their grandchildren, who may not think as they do? How should the opinions of local custodians, who may wish to destroy material, be balanced with strangers who tend towards preservation? There are no simple solutions and no solutions which are not riven with colonial legacies.
5The bottom-line conclusion from Brazil is that if ethnographers had kept their materials in filing cabinets in their personal offices at universities, or in boxes in their attics, then more would have survived. The real answer is to duplicate, to make copies, although we must recognise that there are problems (chiefly of cost and effort) in making copies of analogue photographs and films. Putting a copy in an archive does not mean another copy cannot go elsewhere. Having said as much, we must recognise that this has cost implications which archivists will find hard to manage.
- 4 The LOCKSS Program, based at Stanford Libraries, provides services and open-source technologies for (...)
6Digital solutions for managing systems of copies are now well-established. If we are thinking about the future, one in which records are digital, then there is a robust and well-developed answer: LOCKSS (Lots of Copies Keep Stuff Safe).4 This has two meanings. It is a technical, software solution to distribute copies of digital journals to avoid catastrophic loss from fire or from e-journals either running out of money or ceasing publication. And it is also a general tenet, a philosophy of how to manage data that, certainly with digital material, is quite easy to manage:
Spread copies widely (with cross references to make it clear what is happening so other researchers are not mislead into thinking there are two different documents rather than one with two copies).
7It might be that rather than trying to establish a few BIG archives, we should develop distributed systems where there are centralised catalogues – harvesting metadata from many repositories – which can point people at the source material wherever it may be. This is the ambition of <linked data> and the semantic web as applied to interconnected archives. I note that strictly this approach is consistent across both analogue and digital archives (catalogues are neutral as to whether they are pointing at paper or digital items). Indeed, there are many examples of these sorts of things in different subject domains.5 So at first sight this seems practical. But we should note that so-called portals (which provide single unifying points of access to distributed systems) are not without expense, and they can be fragile: if metadata standards or hardware platforms with associated URLs change, then the data harvesting (or search dissemination) may break (the technical term for this is “linkrot”), so what works one year may not work in ten years’ time even though the material may still be housed in the same place. There is a problem maintaining the systems even in the short term. Once again it all comes down to money.
- 6 Small Computer System Interface is a set of standards for physically connecting and transferring da (...)
8One can put paper records in a box and they are usually still readable in fifty years’ time (if acid paper has not disintegrated or biro ink has not faded). As we know, the same approach does not work for digital. Benign neglect does not work well in the digital domain: discs may not be readable, the wires and connectors may no longer be used. How many people now use, for example, SCSI6 interfaces to connect hard drives? Without the cables and computers with the appropriate ports, the drives are unreadable. Curation has to be active. This is problematic for “Dark” or “Dim” archives (described below). It is all but impossible for truly hidden ones.
9The only consolation I can offer is that anthropologists and archivists are not alone in all this. Solutions to these sorts of problems affect governments, commercial companies and all academic disciplines. By comparison to physics or astronomy, anthropological digital data are relatively small in scale, so we may be able to benefit from adopting solutions developed elsewhere, in other disciplines with larger and more pressing data management needs. This is where centralisation can have benefits: a single solution can be rolled out for all records… Although, as we have to keep reminding ourselves, this is at the risk of fire (as in the Brazilian case with which we started), flood and other catastrophes.
- 7 Kirsten Bell (2018) looks at ethics more generally, considering conflicting relationships and stanc (...)
10Turning now to the contradictions and conundrums in the ethical position of archiving,7 these are even more irresolvable than the contradictions about centralisation. Consider the possibly conflicting, certainly different, responsibilities that anthropological researchers have towards:
- Informants
- Colleagues
- Informants at a later date and their descendants
- Future colleagues (including older versions of ourselves)
- Research funders or institutions
In the two examples discussed below (The Northern Irish Troubles and the research of Alice Goffman), some other parties are involved (legal actors, police forces and journalists).
11These responsibilities and obligations point us in different directions; they mutually conflict. Respect for the privacy of individuals suggests anonymization, closure or not archiving, whereas respect for the descendants of those individuals in the distant future suggests openness and archiving for the long term.
12Delicate, highly sensitive material sometimes should not be archived, and perhaps should be destroyed. I note that future historians would disagree with that statement since this tends to be the material they particularly want to see. It is very hard to know what to do about this contradiction. Hidden archives may be the only solution. Moreover, there may be radically divergent opinions about what materials are delicate and sensitive, as well as who has the right to decide to destroy items, and/or control access to certain types of materials. The issue of deciding what gets accessioned, what enters the archive in the first place, is old-style “archival appraisal”, but now reimagined for a digital twenty-first century. Against those growing up after the invention of the Internet etc., I have to insist that this is not a new issue, but rather an extremely old problem in a new and somewhat different form. Two extreme or dramatic examples illustrate what I am talking about:
- An oral history project about the “troubles” in Northern Ireland was archived in Boston (USA). It sought to resolve the issue of informant confidentiality by promising former participants in the conflict who were interviewed that the recordings would not be released until after the interviewee’s death. But the Northern Irish police used legal subpoenas to break these promises. The researchers had made promises in good faith that in the end they were not able – were not allowed – to keep;8
- More recently there has been controversy about the evidential base of Alice Goffman’s book On the Run: Fugitive Life in an American City, 2014. She has stated that she destroyed all her fieldnotes so that she cannot be subpoenaed about the young black criminals she worked with.9 (It should be noted that the case became celebrated partly because she is the daughter of Erving Goffman).
I recognise these are extreme cases, and for the remainder of the paper I will not consider material of such sensitivity. Rather I will consider the bulk of far more mundane, less sensational, research material.
- 10 See Bell (2014) for a wider discussion of informed consent in anthropology.
- 1
13As has been mentioned, there are arguments that archiving is unethical. For older material, this includes the absence of consent for archiving. In any case, contentiously, archiving is unethical because of the impossibility of obtaining meaningful informed consent. It is impossible to state in advance what uses unknown others will make at some unspecified future date of the material to be archived. This makes impossible prior informed consent in the strict sense.10 And there are concerns about legal frameworks, since anthropological material often crosses continents. The quite recently agreed European Union-wide GDPR (General Data Protection Regulation) covers all material about identifiable living individuals (from anywhere) that is processed within the EU. This has not resolved the issues about archiving, although it actually has explicit provision for research data that makes legal the personal data handling (processing) of social scientific materials over the long term (and consent is not the only ground for doing this: the public-interest ground, i.e. science in its most general sense, is an explicitly acceptable ground for personal data handling).11
14From the list of different parties interested in anthropological research materials, I will now consider the position of funders and colleagues. The responsibilities of researchers to informants and their descendants are discussed in the examples below.
15Funders are included in the above list of parties with conflicting ethical positions. In the past, anthropologists tended not to consider their funders as having a significant (or any) role to play in the consideration of the future of the material that their funding enabled to be gathered, let alone the ethics of this. However, driven by discussion and developments in other disciplines, this is no longer the case: if public or charitable funds are given to a researcher to collect anthropological material, then although the researcher has first call on that material, it does not fully belong to them. (Oddly Mauss and The Gift [2016] has not featured in the discussion as much as one might expect, granted its central role in anthropology). Others may wish to check results or to reanalyse some aspect of the material in ways that the original researcher either may not have thought of, or is simply not interested in. Since a funding award implies a choice not to fund others, it seems unethical to resist any means to preserve and make available the material that the funding has enabled (or to deny that the funders may have a legitimate position on this, which the researcher, in accepting the funding, has implicitly agreed to respect).
- 12 During the 1970s and 1980s a Cameroonian photographer named Jacques Toussele took many thousands of (...)
- 13 This is one of the motivations of ODSAS (the Online Digital Sources and Annotation System). See htt (...)
16Colleagues and future colleagues have an obvious interest in research material. These may be not only colleagues from one’s own subject area, but those from entirely different disciplines. One example might be the use that fashion historians could make of the photo archives from African studio photographers such as Jacques Toussele.12 Some anthropologists feel embarrassed that their notes are so slight: they rely on memory for what they write, and feel that their sketchy notes or aides mémoires can be of little use to others. To this my response is that a little use is still something, and it may be important to know that one piece of published work relies more on memory than another. However, the most important future colleague that we have to consider is ourselves: we are not only the most privileged readers of the sketchiest of notes (which may jog important memories), but we need to consider that we may return to our notes and photographs twenty or thirty years after the fieldwork. There is a nice co-incidence of interests here: by working to help our future selves (so ensuring that the materials in question survive13 – which is not trivial with digital files – and will be comprehensible), we do work that can also benefit some others.
- 14 Rehfisch fieldwork photographs, Mambila, British Cameroon (now Nigeria), 1953-1954: http://era.anth (...)
17I should also note that making sense of other people’s field materials is not an all-or-nothing matter: it admits of degrees. Below I discuss the example of the work of Farnham Rehfisch, for whom, as another ethnographer working with Mambila people, I am a near-perfect other reader.14 But as already mentioned, other very different uses are perhaps possible, and these may not need the same background as I bring to Rehfisch’s notes.
- 15 See Dilger et al., 2018; Pels et al., 2018; Koning et al., 2019.
18I should note here a series of reports and papers sponsored by the European Association of Social Anthropologists which originated in a dispute about the probity of the papers of a long-retired researcher from Leiden. Their call is for the destruction or “return” of fieldnotes upon retirement. Nothing is said about archiving as an option, or about a researcher continuing to work on their notes after retirement. Indeed Mart Bax appears to have anticipated their call by destroying his notes – which impeded the enquiry into his research.15
19In the book I wrote with Roger Just, we tried to establish a median position between full-blown relativism and hard-nosed positivism (Zeitlyn and Just, 2014). We argue for some sort of wishy-washy realism for all the problems with that word. We talked about merology (the study of incompleteness or partiality) as a way of being rigorous but incomplete, about not having the last word. Anthropologists give knowingly incomplete accounts which are rigorous and correct (as far as these go). This is what I now prefer to call “sparse anthropology”, an anthropology in which theoretical exuberance is tamed or curbed. Among other advantages, such an anthropology is more comprehensible for those working in their second or third languages than those following the dictates of the current turn in high Theory (the capital letter is deliberate).
20In another part of that book we talk about anthropologists being characterised by a position of bad faith towards the people we are living with when we do fieldwork. In short, anthropologists are both the ultimate sceptics believing nothing they are told, and credulous fools believing everything. The trick is to turn this into a methodology.
- 16 As already mentioned, this is a motivation for ODSAS. See Hannoun, this volume.
- 17 I note that Diane Duclos has recently (2017) published a discussion of the impossibility and unethi (...)
21Anthropologists tend to have a similarly confused relationship to the field materials they collect. They cherish them assiduously, moving them from one filing cabinet, box or hard drive to another over decades and generations, holding on to them even long after retirement (if they are lucky enough to survive that long, and to have a job to retire from).16 Clearly they value them. At the same time they are often reluctant to grant access to others, in some cases even after they have died. Some suffer from a form of imposture syndrome: they are worried that if someone else sees their notes they will discover there is insufficient basis for the claims they have published. Other worries are about confidentiality, especially nowadays, which is why I need to say something about anonymization.17
22While the GDPR has been generating discussion over recent years,18 I have got increasingly concerned about this topic, especially as those concerned about ethics (e.g. the members of university and research funding ethics panels/committees) seem to be using the GDPR to justify obligatory and complete anonymization. So let me indulge in a small polemic about the differences between GDPR data protection, research ethics, and archives, and about why it is that although pseudonymization may have a place, anonymization may well be unethical in the long term. One of the things that has come out of the GDPR discussion is the importance of the distinction between anonymization and pseudonymization. An example of a visual anonymization technique is the blurring of faces in photos made accessible online, but preserving un-blurred originals in a long-term (dark/hidden/inaccessible) archive. As I recall, MédiHAL did this when it launched, but I cannot see this mentioned in 2021.19
- 20 I note that the man depicted is deceased, so these photographs are no longer personal data under th (...)
23However, this sort of approach has limits. A dramatic example of this is the case of someone whose silhouette is characteristic and will be easily identifiable whether or not the face is pixelated. This remains the case even if one uses an alternative to pixilation, such as software that exaggerates the edges and in effect turns a photograph into a line drawing. I give an example here from the Jacques Toussele archive of photographs from Cameroun (Figure 2). As I hope these two images make clear, in some cases a silhouette, even a line drawing, is disclosive.20
fig. 2 – The same image with the face blurred (left) and with an edge retention treatment (right)
Photo Jacques Toussele; EAP054/1/19/524
24To make concrete the importance of preserving copies of photographs without blurring or other forms of anonymization, let me give a clear example of why it would be unethical to destroy or fully anonymise data. Consider the following field photograph that I took in May 1986 (Figure 3).
fig. 3 – Field photograph
L-R: Barmi (alive 2018), Nde Donat (d. 1987?), Ndignoua Salomon, Suzana Thia (alive 2018), Ngon Luise (died), Blandine (died), Dissi (Mougna’s child, died), Jacqueline (alive 2018), the two children in the front: the boy is Kounaka Fidèle (Salomon’s junior brother, alive 2018), and the girl is Mbitti Josephine (alive 2018). The names were provided in 2018 by Serge Donat, son of Blandine.
Photo David Zeitlyn, DZ reference: 24_34.jpg 01/05/86
25Yanele Blandine, the girl in the white headscarf, is long dead. Her son, Serge Donat, had no photograph of his mother until recently, when I was able to send him a copy. Imagine his response if I had said either:
I took a photograph of your mother long ago but I cannot send you a copy because I destroyed it on completing my doctorate.
or
I took a photograph of your mother long ago but I cannot send you a copy because I do not have her permission to share it.
or
I took a photograph of your mother long ago but I can only send you a copy with her face blurred because I do not have her permission to share it.
I would suggest that all of these possible responses as suggested by the ethos of many ethics panels would be inhumanly rude and indeed would be deeply unethical.
- 21 Note that if they are truly anonymized and there is no key to reverse the process, then they are no (...)
26A researcher might well want to anonymise versions (I stress versions, not everything) of research material so it can be made accessible relatively soon after the data has been collected – in ways that parallel how the UK Census Office removes personal ID information from material from recent censuses. There is an important point to stress here: even though the publicly accessible version may be anonymized, if the key to the anonymization exists, then legally it remains personal data and subject to GDPR rules.21 I very much hope researchers retain the key, because the longue-durée view is that the full data should be retained and made available, but only in the long term. Perhaps just as happens with census data, this should default to being 100 years after the collection date.
27My favourite example of why this is important is to cite the book written by Michel Foucault in 1973 about events that occurred in 1835 (made into a film in 1976):
Moi, Pierre Rivière, ayant égorgé ma mère, ma sœur et mon frère…
I think similar books and films should be possible in 150 years’ time with the full names given rather than:
Moi, Homme03061835, ayant égorgé ma mère, ma sœur et mon frère…
So anonymization should be reversible. The jargon for this is pseudonymization.
28There is a further almost banal point to be made – unnecessary for most of this readership but important nonetheless. Archiving does not mean:
- giving free access to all comers
- putting all material online without regulation (even for digital archives).
29To emphasise this, it is sometimes useful to distinguish between Light, Dim and Dark archives.22 Light or Open archives are ones that are freely searchable and where access is possible to the materials found from searching the catalogue. However, even these may not be open (or freely accessible) on the internet. Registration may be required and other constraints on usage may apply, but as far as I am concerned this would still be a Light or Open archive. At the other extreme is a Dark Archive in which nothing is made accessible, at least perhaps for the duration of an extended embargo. Sometimes not only are the materials inaccessible, but also the metadata. Between these extremes are Dim Archives. These are archives that incorporate elements of both the Dark and Open Archive models. Access for some materials is restricted to organizational custodians, while access for others may be open to a broad user community. There are dark archives that still make their metadata available (an example is that of a doctoral thesis under an embargo where the catalogue may state that it is embargoed until 2025, but the catalogue may still include the abstract so readers can learn something about what they cannot access!). Other dark archives hide even the metadata, so in principle no one can discover that the material is there.23
30I now return to the more general issue of conflicting claims about what is ethical in archiving anthropological research material. There are arguments that not archiving is unethical: wasting money by not allowing others to benefit from the costs of undertaking the research in the first place; an absence of respect for the informants and their descendants; removing the possibility of important unintended research benefits; asserting in a very colonial fashion the right of the researcher over the research partners/subjects, who by being anonymized or deleted loose thereby their moral rights to be recognised.
31I could go into these in much greater depth but I do not really want to get into those details, absorbing though they are. My purpose here is rather to point to the contradictions that confront us. These make the possibility of archiving a “wicked problem”, which is probably why many in the profession would rather avoid the issue altogether. For, indeed in the terms of my subtitle, we are Damned if we do, damned if we don’t. Or, as J.-F. Werner (2006) put it discussing photographic research: we are between the hammer and the anvil (nous sommes entre l’enclume et le marteau).
32To move to a more positive conclusion, I want to give some further examples from my own work which demonstrate (explain?) why I am so convinced that ethnographic archives are important and that – for all the problems – it is important to create and maintain them in all their disclosive detail.
33The field notes of Farnham Rehfisch are a good example. There have been two anthropologists who have done fieldwork with Mambila. I am the second. The first was Farnham Rehfisch, who studied in the Nigerian Mambila village of Warwar in 1953 (Figure 4). I was fortunate enough to have met him before he died, and he gave me his field notes. I was able to read them in the village where he wrote them on a visit to Nigeria in 1993, some forty years later. In many, many ways I am the perfect reader for his field notes. I can see echoes and resonances that I am sure very few other readers would notice. That does not mean they are useless to others, although they would be very challenging to someone without a Mambila background. But it is not all or nothing. There are plenty of median positions – possible readers who might not get as much out of the field notes as I may, but still can find them useful, those working on neighbouring groups, for example. The field photos of Farnham Rehfisch (depicting a wide range of activities and aspects of Mambila life, from weaving to ritual via agriculture and hair styles) are excellent, and could be used in many ways even by non-Mambila specialists. They have been made available online as part of the Experience Rich Anthropology project.24
I do not know who will replace me, but I look forward to somebody someday continuing to work with Mambila, and maybe discussing my field notes with the great grand-children of the people I have worked with.
34Another example going further back in time concerns a photograph taken on 11 January 1938 by a missionary named Paul Gebauer in Mbamga village (sometimes also written as Mbamnga), in what was then Cameroon but is now Nigeria. The photographs are now held in the archives of the Metropolitan Museum of Art in the USA (because they acquired from Paul Gebauer a large collection of art from Cameroon).25 Having visited the archive, I was able to take a copy of this photograph back to Nigeria (Figure 5), and there I was able to trace the son (Figure 6) and grandson of the man in the original photograph. They enabled me to record his name, so the documentation of the archive was improved and a man in Mbamga (the grandson) gained a photograph of his grandfather, whom he had never seen.
fig. 5 – Mbapu Dola died c. 1964
Photographed by Paul Gebauer 11 January 1938 in Mbamga Village. (ref 45_15); Arts of Africa, Oceania, and the Americas, The Metropolitan Museum of Art, New York, © The Metropolitan Museum of Art. Used with Permission
fig. 6 – Dola Mohammed (son of Mbapu Dola)
Photographed by David Zeitlyn 19 April 1993 in Mbamga Village (ref N93f3_21)
- 26 The digital files have been backed up by my University IT services, thus providing a minimal sort o (...)
These are the basic actions of archivally mediated interaction. One involves a formal archive (the Metropolitan Museum of Art as holder of the Paul Gebauer archives), the other my own research data, which is currently held and curated by me.26
*
* *
35As the fire at the Brazilian National Museum demonstrated, we cannot take for granted that the results of anthropological research will survive into the future. As well as concerns about physical survival, the archiving of anthropological materials is fraught with ethical challenges and beset with contradictions. In this paper I have given a few examples of ways of dealing with those contradictions, or perhaps it is better to say these examples show ways that can be found of living with the contradictions. Living with contradictions is what humans are used to doing. And anthropologists as well as archivists are (mostly) human.