A museum that places its collection online usually does so in the language of generosity. Objects long confined to storerooms become visible to anyone with a connection, images are released for reuse without charge, and the institution can say with some justice that it has shared what it holds in trust.

Among the objects swept into these programmes are things made for religious use and acquired, in many cases, during colonial rule: ancestral figures, masks, reliquaries and ritual regalia. The people from whose communities they came frequently regard them as living or powerful, and surround them with restrictions. Such objects may be seen only by the initiated, by one sex, by members of a lineage, in a certain season, or after proper preparation.

The title asks whose heritage is at stake when a restricted object becomes an unrestricted file. The position taken here is that sacred objects carry obligations that travel with their images, that the communities who recognise those obligations have the first claim to decide the terms of online access, and that museums should treat that claim as a condition of digitisation and not as a courtesy to be extended afterwards.

Open-access policy rests on a view of culture as information. On that view an image of an object is a description of it, descriptions of public collections belong to the public, and any barrier to their circulation needs special justification. The FAIR principles, which ask that data be findable, accessible, interoperable and reusable, express the same ideal for research data (Wilkinson et al., 2016).

Christen (2012) has shown how poorly this fits many Indigenous systems of knowledge, in which access is graded by kinship, age, gender and ritual standing, and in which the freedom of information is not a value at all. Where knowledge is held under such rules, a default of openness is not neutral. It replaces one regime of access with another, and it does so on the authority of the institution that happens to hold the object.

The assumption that a copy is only a copy is equally fragile. Many communities hold that a likeness participates in the power of what it depicts. Isaac (2011) documents the Zuni position that replicas of religious objects carry the knowledge embodied in the originals and are subject to the same control. If that is so, a high-resolution scan released for download does not merely represent a restricted thing. It reproduces it without restriction.

Law offers the museum a straightforward account. The institution owns the object or holds it in trust, the object is usually too old for copyright, and any rights in the photograph belong to the photographer or the museum. The community of origin appears nowhere in this chain. Intellectual property law protects identifiable authors for limited terms and has no ready place for collectively held, inherited and sacred knowledge (Brown, 2003). In this way the legal order makes the community's interest invisible. Table 1 lists instruments that take a different view.

The broadest of them is the United Nations declaration of 2007, which recognises the right of Indigenous peoples to maintain, protect and have access in privacy to their religious and cultural sites, to the use and control of their ceremonial objects, and to maintain and control their cultural heritage and traditional knowledge (United Nations, 2007). In the United States, federal law provides for the return of sacred objects and objects of cultural patrimony to affiliated tribes (United States Congress, 1990). The museum profession's own code asks that material of sacred significance be housed and displayed in a manner consistent with the beliefs of the communities from which it came (ICOM, 2017). Each recognises a relationship that ownership does not extinguish.

Three reasons support placing the decision with the community of origin. The first is that the harm of wrongful exposure is defined by religious norms that only the community can state. A curator cannot know from a catalogue entry whether an object may be seen by women, by children or by outsiders. The second is historical. Much sacred material entered museums through sale under duress, confiscation by missionaries and officials, or removal from graves and shrines. An institution that acquired an object in that way has a weak moral basis for deciding alone how it is shown to the world.

The third reason is that the alternative has been tried. Museums have long presented themselves as contact zones in which collections become occasions for relationship between institutions and the peoples represented in them (Clifford, 1997). Boast (2011) argued that such collaboration remains unequal for as long as the museum keeps final control over the terms. Digitisation sharpens the point, because a decision to publish is practically irreversible. Once files have been copied they cannot be recalled, and a consultation held after release has nothing left to decide.

Digital projects led by communities show what a different arrangement looks like. The Mukurtu platform grew from work with the Warumungu community in central Australia and allows each item to carry protocols that determine which users may see it (Christen, 2011). Traditional Knowledge Labels let a community attach its own statements about proper use to records held elsewhere (Anderson and Christen, 2013). Tlingit clan leaders worked with the Smithsonian Institution on three-dimensional scanning of clan objects for purposes that the clans themselves defined (Hollinger et al., 2013). In each instance digitisation went ahead. What changed was who set the terms.

The choice is often presented as open against closed. It is better understood along two separate lines: how widely the files circulate, and who has authority over that decision. Figure 1 crosses the two. The lower right quadrant, where sacred images are released because release is the default, is the situation this article criticises. The lower left is no improvement: a museum that locks material away on its own judgement still excludes the community from the decision, and may exclude community members from their own heritage.

The two upper quadrants are both legitimate outcomes. A community may decide that an object should be seen by everyone, perhaps with a statement of its meaning attached, or that it should be seen only by those entitled under its own rules. The CARE Principles describe this as authority to control and set it beside, not against, the technical aims of open data (Carroll et al., 2020). The important movement is upward in the figure, not leftward.

The most serious objection comes from the cosmopolitan defence of museums, which descends from what Merryman (1986) called cultural internationalism: the idea that the heritage of each people is part of the heritage of all. Cuno (2008) argues that cultural objects speak to humanity at large and that claims of exclusive national or group ownership impoverish everyone. The argument has force for works of art and for antiquities whose makers have no successors. It has less for objects in current religious use by a living community, and less still when what is asked is not the return of the object but restraint in broadcasting it. A reader in another continent loses little when a mask restricted to initiates is described in words and not shown.

A second objection asks who speaks for the community. Communities are divided by generation, by faction and by distance, religious authority may be disputed, and some members will want images published for teaching or for cultural revival. Brown (2003) warns against treating cultures as bounded entities with a single voice. The warning is sound, but it describes a difficulty of consultation and gives no reason to skip it. Museums already negotiate with estates, lenders and donors who disagree among themselves. Protocols written for archives advise working through recognised governing bodies and returning to them as circumstances change (First Archivist Circle, 2007).

A third objection is that restriction is censorship and will chill scholarship. Yet research involving living people is routinely limited by confidentiality, consent and privacy, and scholars do not regard those limits as censorship. Differentiated access, in which records exist but some images are withheld or shown only to authorised users, leaves the documentary record intact. A related worry concerns the proposal that digitised African collections be released freely as part of restitution (Sarr and Savoy, 2018). The better course is for decisions about digitisation and access to pass to the communities of origin along with the objects, since open release decided in Europe would repeat the original appropriation in a new medium.

Several practical consequences follow. Sacred and ceremonial material should be identified before a collection is photographed in bulk, and held back from publication until those with authority over it have been asked. Catalogue records should be able to carry community statements and access conditions, which requires collection software that supports more than a public and a private setting. Agreements should state who may alter the terms later, and a named member of staff should be responsible for keeping each relationship alive after the project that began it has ended. Those who catalogue and photograph collections need enough training to recognise restricted categories of material and to stop and ask. Where images have already been released, museums can still withdraw them from their own sites, add notices and stop promoting reuse, while being candid that copies persist elsewhere.

Digital copies are sometimes offered to communities in place of the objects themselves. Boast and Enote (2013) object that this so-called virtual repatriation is neither virtual nor repatriation, since nothing is returned and control stays where it was. Digital access negotiated with a community can support the return of objects and the renewal of knowledge, but it is no substitute for either. Funding bodies that require open licences as a condition of digitisation grants should allow exceptions for culturally restricted material, so that institutions are not forced to choose between their funders and their partners.

Open access is a good default for most of what museums hold. Sacred objects are the hard case because their meaning includes limits on who may look, and because the institutions that hold them often acquired them in ways that left their communities without a voice. Asking those communities first, and abiding by the answer, costs the public very little and corrects a real wrong. The heritage in question belongs most immediately to the people for whom the objects are still sacred, and digital policy should be written on that footing.

To be declared by the author(s) before publication.

To be declared by the author(s) before publication.