Returning to a historical image

What remains in the frame, and what a second visit makes visible

A photograph fixes a fraction of a second. It also fixes a place to stand. When the photographer leaves, the place goes on changing: a road moves, a house falls, a tree grows, a village loses its inhabitants or acquires new ones. Rephotography begins by returning with a camera and asking what the first photograph can still tell us.

The procedure sounds simple: locate the view, recover the photographer’s position as closely as possible, and make another image. In practice, the old photograph is both the map and the puzzle. Its background may contain the ridge that identifies a valley, the line of a roof that reveals the height of the camera, or a street corner that no longer exists. Even a failed attempt to repeat the view can yield evidence. The obstruction may itself be historical: a new building, a road, a forest, a restricted border or a changed riverbank.

The comparison has at least three forms of value.

Historical and spatial. Repeated views provide a readable record of changes in buildings, vegetation, mobility, public space and land use. They can also expose errors in a photograph’s inherited caption. A mountain silhouette is not interchangeable with another; when the background of a Manakia panorama does not match its supposed city, the image must be located again.

Anthropological. Rephotography is not merely a measurement of visible change. The act of searching brings the researcher into contact with people who recognise a house, dispute a place name, remember a vanished use of space or know why a route was abandoned. Their accounts belong beside the two photographs, although they remain testimonies to be interpreted rather than automatic proof of a date.

Visual. A historical view and a contemporary view do not cancel each other. They create an interval the eye is invited to explore. The spectator notices continuity and disappearance simultaneously; more importantly, the comparison reveals that both photographers selected a viewpoint and made decisions about what could enter the frame.

A method with a history

Repeat photography was used by geologists and landscape scientists before it acquired an established place in photographic art and the humanities. In 1977–1979, Mark Klett, Ellen Manchester and JoAnn Verburg organised the Rephotographic Survey Project in the American West. Returning to nineteenth-century survey images, they reconstructed locations and framing while recognising the choices made by both the historical and the contemporary photographer. See Mark Klett’s account of the project and Second View: The Rephotographic Survey Project (University of New Mexico Press, 1984).

The method was subsequently discussed explicitly in visual anthropology. Trudi Smith, in “Repeat Photography as a Method in Visual Anthropology” (Visual Anthropology 20, 2007, 179–200), approached the return to an earlier image as situated fieldwork, attentive to encounters, place and interpretation. Publisher’s record.

There are Balkan precedents. The Dutch photographer and traveller Casper Molenaar has retraced places pictured by the German photographer Kurt Hielscher in his 1926 album of Yugoslavia, setting historical and contemporary views beside one another. His project, In the Footsteps of Kurt Hielscher, is a particularly close visual comparison.

What distinguishes Beyond Yesterday

Beyond Yesterday is built around a different historical archive and question. The starting images were made by Ianaki and Milton Manakia, Aromanian photographers who worked in a multilingual Ottoman and post-Ottoman Balkans. The project follows not one national landscape, but routes across the present borders of Greece, North Macedonia, Albania and Bulgaria. It brings together photographic prints, glass negatives, archival documents, studio portraits and, where relevant, the brothers’ moving images.

The contemporary photograph is not the final answer. It is a means of testing a location and of beginning another inquiry: who lived here; what road connected this village to another; when did people leave; why was a name translated or forgotten; how did the original image travel into an archive? The research has included repeated journeys over more than a decade, consultation of the National Museum of the Romanian Peasant in Bucharest and the State Archives in Bitola, and field conversations along the routes traced by the brothers.

The project’s distinctiveness lies in joining transnational photographic rephotography to item-level archival investigation, historical toponymy, community memory and a travelling exhibition. Often the most interesting thing is not the photograph itself but the story: the stories that surfaced along the way, from people met in villages, from archive files, from wrong turns and unplanned encounters.

The evidence of this approach can be seen in the project’s photographic map and Journal: the reconstruction of a London photograph dated through a theatre programme, a panorama moved from Plovdiv to Sliven because the mountains did not match, and the reunion of images of Laista preserved in different countries.

What a pair of photographs cannot prove

Two pictures can demonstrate that visible elements differ. On their own, they cannot always explain the cause. An empty square does not prove that a community has disappeared; a ruined building does not establish when it was abandoned; a mountain that seems closer might result from a different lens or crop. For each comparison, Beyond Yesterday distinguishes observable change, historically documented explanation, local testimony and the researcher’s interpretation.

It also preserves the possibility of failure. The correct historical camera position may be inaccessible or destroyed. In such cases, a near-view is labelled as a visual approximation, not presented as an exact optical match. Precision is an account of the limits of the evidence, not a promise that time can be reversed.

Field methodology: how the second photograph is made

From a historical caption to a reproducible modern viewpoint

The practical task is to find, with the greatest accuracy the terrain allows, the location from which a photograph was taken around 1900. An original caption is a starting hypothesis, not the end of the investigation. Some captions were assigned long after the negative was exposed; others use obsolete place names or preserve a previous misidentification. The stages below are guidelines drawn from the project’s own practice, not a fixed protocol; each case may call for adaptations.

Stage one: identify the object

Before travelling, record the original image’s holding institution, collection or fonds, item identifier, physical form, caption, attributed author and date or date range. Distinguish a negative from a print and a photographic card from a later published reproduction. If two archival copies exist, examine both when possible: a contact print may reveal more of the margins, while a negative may preserve details cropped from the print. Often I did not even know what the villages in the archive of the National Museum of the Romanian Peasant were called. Many photographs had no name and no location, so the first task was detective work: identifying the village. There is no real method here; it comes down to intuition and luck, but also to a visual training that can be practised, learning to recognise the shapes of mountains. One caution: mountains change too. Forests grow back and landslides reshape slopes, as happened at Pades.

Example, Avdela. The project’s photographic map displays a glass negative of the village held in Bitola, Fonds 0580, GPN 5701, beside a print held at MNȚR, Mk-1910. Another pair depicts Holy Trinity, with Bitola GPN 5676 and MNȚR Mk-1932. These identifiers permit a researcher to trace the historical objects independently of the contemporary rephotographs. See the project’s archival captions.

Stage two: test the geographical attribution

Examine ridge profiles, peaks, rivers, valley geometry, street intersections, church towers, masonry, rooflines, bridges and sightlines. They are not equally reliable. Mountains often survive as large-scale landmarks, while façades and roads may be transformed. Compare historic maps, current topographic maps, aerial imagery and photographs made by other people. Search old, official and Aromanian place-name variants in parallel.

Example, Plovdiv and Sliven. A historical panorama thought to represent Plovdiv was tested against its supposed setting. Its mountain background failed the comparison and directed the research to Sliven. The change of attribution is an outcome of visual and geographical analysis, not a change made for a better narrative. Journal case study.

Stage three: reconstruct the camera station

Once the settlement is secure, narrow the search to a defensible camera station, meaning the approximate position and height of the historical camera. Match the relative placement of fixed landmarks, not only the general resemblance of the view. A few metres of lateral movement may shift the alignment between a church tower and a distant summit; moving uphill changes the relative position of rooftops; altering focal length changes the apparent scale and field of view.

The practical order is: choose at least three control landmarks at different depths; adjust position and height until their relationships match; then select an appropriate field of view. A historical lens’s focal length should not be guessed from a cropped image. If the surviving image is already enlarged or trimmed, record that limitation.

Stage four: make and document the contemporary photograph

Record the present viewpoint with latitude and longitude, coordinate reference, approximate GPS accuracy, photographer’s height or camera-centre height, camera direction or azimuth when measured, camera model, lens or focal length, exposure time, date and local clock time with time zone. Retain the original image file and EXIF metadata, a field note and an unaltered reference photograph showing the broader surroundings.

Record also season, weather, accessibility, visible changes, obstacles and sources of uncertainty. Time of day matters when attempting to approximate light and shadows, but identical sunlight may be impossible because the skyline and built environment have changed. An image taken at a different time must not be presented as though it had reproduced the historical lighting.

Unknown fields stay unknown. Where a value was not recorded in the field, the record says “not recorded”; a measurement is never reconstructed after the event as an exact figure.

Stage five: distinguish optical alignment from interpretive comparison

Classify each pair:

• Rephotography, close alignment: the historical viewpoint is strongly supported by multiple landmarks and the second framing approaches the first.
• Rephotography, approximate alignment: the locality is secure, but barriers, demolition, vegetation, terrain or uncertain cropping prevent a closer match.
• Site identified, viewpoint unresolved: the scene can be located, but the exact historical camera station cannot be defended.
• Interpretive pairing: the present photograph explores the historical subject or community without claiming the same view.

The description should explain why a result belongs in one category. Never equate “visually persuasive” with “geometrically identical.” A short, honest note about the remaining uncertainty is a research finding in its own right.

Stage six: conduct the anthropological follow-up

Show the original image to people in the place where it was made. Ask open questions, distinguish direct recollection from inherited family memory, and seek permission before recording names or publishing personal accounts. A resident may identify a building that has vanished or a family photograph preserved privately; such testimony should be attributed with consent and compared with documentary evidence.

Example, Laista. Two Manakia images of Laista/Laka, one present in the Bitola holdings and another in Bucharest, enter a wider history through later photographs and local recognition. The project’s map gives the glass negative as Bitola GPN 1115 and the print as MNȚR Mk-1919, both dated 1898 on the site. The Journal recounts how apparently isolated objects were connected. Map and collection references and research narrative.

Stage seven: publish the evidence, not just the pair

Each public comparison is best accompanied by a compact record: historical photograph (author, title or descriptive title, date/date range, institution, identifier); contemporary photograph (author, date and place); match category; key landmarks; brief note on transformations; source references; and a link to the fuller journal entry where one exists.

For selected research cases, publish an optional alignment overlay and a map point with approximate location accuracy. We advise against publishing precise coordinates for sensitive private properties or vulnerable sites without consent and assessment, and against digitally adding or removing historical objects in either original photograph of the evidentiary pair. Any creative AI treatment is better kept in a separate, unmistakably labelled visual category.

For each pair the project keeps a field record: the custodian’s identifier for the historical object; the date and original file of the new photograph; the place, with its official and historical names; the position, direction and height of the camera, or “not recorded” where unknown; the camera and lens; the crop applied; and the source of the comparison.