Scotland’s Heritage Records: National Resources, Public Access and AI

Scotland’s national heritage resources provide the essential foundation for discovering, studying and celebrating its buildings and places. Generations of survey, photography, cataloguing and designation have made evidence available far beyond the archive reading room. Local websites, walking guides and AI-assisted tools can create additional routes into that knowledge, while keeping its sources and custodians visible.

Read alongside Building Conservation in Scotland and Sources & Method.

Before the database: survey, drawings and inventories

The national record has grown from the work of RCAHMS, founded in 1908. County inventories, field notes, measured drawings, photographs and research accumulated into a body of evidence covering archaeology, buildings, industry and maritime heritage. Its historical depth comes from sustained observation and the preservation of the resulting material.^{1, 3}

RCAHMS’s early chronological remit ended at 1707; it was extended in 1948. The Scottish National Buildings Record transferred to the commission in 1966, helping form the National Monuments Record of Scotland. Recording threatened buildings was an important part of this work: a drawing or photograph might preserve evidence that later alteration or demolition removed.²

A catalogue entry and a survey drawing do different work. The entry helps identify a place and find related material; the drawing may reveal its construction, section or successive phases. Digitisation has made these connections easier to follow, while the underlying sources retain their individual dates, authors and purposes.

Canmore and the connected national record

Computerised cataloguing allowed names, locations, classifications and archive references to be searched and connected. Canmore became the familiar online route into the National Record of the Historic Environment, linking place records with photographs, drawings, bibliographic references and survey activity. HES now makes those place and archive records accessible through Trove.^{1, 3, 4}

The national record has always drawn on more than one institutional collection. It includes contributions from other organisations, communities and individuals, alongside the survey work of HES and its predecessors. A record can bring together material created decades apart, by people asking different questions about the same place.³

This cumulative structure explains why a short site description may lead to an extensive archive. Some holdings have digital images; others are represented by catalogue information and require further consultation or digitisation. The presence of a catalogue record does not guarantee an immediately viewable scan.

Scran was conceived in 1995 as an educational resource. Its founding partners included national museums, RCAHMS and the Scottish Museums Council, with curriculum involvement. It joined RCAHMS’s remit in 2008 and became part of HES in 2015. The accessioning of its collection into the HES archive began in 2024.⁴

ScotlandsPlaces brought together material from HES, the National Records of Scotland and the National Library of Scotland. Following its retirement, HES directs users to the successor services of those bodies. A building’s documentary history may therefore require several resources: heritage survey records, historic maps, estate papers, photographs, valuation material and local archives.⁴

Trove: a shared point of access

HES launched trove.scot on 13 February 2025. The launch brought together designation information, the archaeology and built-heritage catalogue, material from properties in care and Scran collections. The launch announcement described approximately three million records; that figure belongs to the announcement, rather than a fixed measure of the service.⁵

Canmore and Scran were among the services retired in June 2025. HES explained the change in terms of supporting ageing websites securely and sustainably. Canmore links were intended to redirect to the corresponding Trove records; migration and replacement of features formed an ongoing programme.⁶

Trove’s roadmap describes continuing development and invites feedback. Functions such as contribution tools, licensing, export and map facilities should be checked against the live service. A feature described as planned or being explored may differ from one already released.¹⁰

PastMap: seeing several records together

PastMap brings information from HES, local-authority Historic Environment Records and other sources onto a common map. Layers can be compared against modern mapping, historic mapping and aerial photography. Links lead back to the organisations responsible for the underlying information.⁷

This is particularly useful when a building belongs to a larger landscape of roads, earlier settlement, industrial remains or archaeological sites. Overlapping symbols can represent different records associated with one place, such as an archive location, an archaeological entry and a statutory designation.

PastMap’s current welcome page directs users to the Historic Environment Portal for live designation information and says its map search should not be used to confirm current designations. A convenient map layer and the current designation record have different purposes.⁸

Listings, scheduling and the record’s legal purpose

HES lists buildings for their special architectural or historic interest. The statutory address is the legal part of a listed-building record; the descriptive account and statement of interest provide supporting information. Records can change through review, amendment and new evidence.⁹

A place can be represented in the national archive without being listed, and a listed ensemble can include several buildings or structures. A search result, a photograph of one entrance and a listing boundary should therefore be read together. The listing date, construction date and date of a photograph describe different events.

The national and local records also inform planning. NPF4 Policy 7 calls for assessments based on an understanding of cultural significance and for proposals to be informed by Historic Environment Records. The relevant authority and current designation material remain essential when an enquiry concerns proposed works.¹⁸

Historic maps and the wider evidence network

The National Library of Scotland’s georeferenced mapping allows older surveys to be compared with modern geography. It also provides map tools, reusable viewers and specified API services. These resources can help locate vanished streets, follow an estate’s growth or establish that a building existed by a particular survey date.¹¹

Map evidence has its own limits. Survey and publication dates can differ; small structures may be omitted, and a building’s outline may conceal several uses or phases. A name on one map may identify a property, settlement or business whose extent changed over time.

From a national record to a citizen’s question

A resident may begin with an old photograph, a local building name or a question about the houses on a street. A teacher may want a short comparison of school buildings. A walking group may need an accessible route linking public viewpoints. A community organisation may be preparing an exhibition about a hall or former factory.

National resources make such work possible on a scale that once required repeated visits to specialist collections. Additional access platforms can organise material around a place, a theme, a route or a reader’s level of knowledge. Their value grows when they lead people onwards to the fuller evidence and acknowledge the institutions and contributors who assembled it.

Participation can also add evidence: a dated photograph, an identified architect’s drawing, a building’s local name or a correction to a location. Archaeology Scotland’s Adopt-a-Monument programme provides an established example of communities working with advice, training and practical stewardship.¹⁷

What AI could make easier

The following are potential uses of AI-assisted access tools, rather than a claim that every national service currently provides them. A system can help a reader turn an ordinary-language question into useful search terms, explain architectural vocabulary, assemble a first list of relevant records, or compare passages supplied by the user.

It can also assist with drafting captions, producing different reading levels, preparing translations and structuring material for a local trail. A person with limited technical experience may be able to create a small website or an illustrated guide, then spend more of their time checking photographs, speaking to contributors and investigating the buildings.

The most useful design keeps the route back to evidence short. A summary can sit beside the source title, record identifier, relevant date and a link to the full record. Readers can then move between an inviting introduction and a specialist description without losing the relationship between them.

AI slop: a problem of evidence and editorial quality

The expression ‘AI slop’ commonly describes abundant, plausible-looking output that adds little reliable information. In built-heritage writing it may take the form of invented architects, unsupported dates, generic praise, irrelevant photographs or repetitive descriptions that could apply to almost any town.

Generative systems can produce fabricated explanations and citations as well as factual errors. NIST’s risk profile identifies this problem as confabulation. Its work on machine-generated reports also stresses the need to assess whether citations actually support the claims attached to them.^{15, 16}

A fluent account can therefore make uncertain evidence look settled. Repeating a claim across several generated pages can amplify the appearance of agreement even when the pages descend from the same mistaken source. The practical response is to preserve provenance and check claims before publication.

The relevant standards are accuracy, usefulness, attribution and openness to correction. Those standards support the same shared endeavour across national services, professional research, local history and citizen publishing. AI can widen participation when the resulting work remains connected to that endeavour.

Checks that make wider access worthwhile

Element	A useful publishing practice
Identity	Check the name, address, town, record identifier and subject of each photograph.
Location	Distinguish an entrance, a representative point, the camera position and the extent of an ensemble.
Dates	Separate construction, alteration, survey, listing, photograph and publication dates.
Authorship	Distinguish architect, practice, builder, patron and later restoration team.
Evidence	Read the cited source; preserve qualifications and explain material disagreements.
Rights	Check each photograph, map and dataset’s applicable reuse terms and credit requirements.
Accountability	Give a contact or correction route and retain a dated record of substantive changes.

For example, a guide to a large housing development should identify which phase it describes. A photograph of neighbouring recent houses does not establish the architect or tenure of a named phase. A map pin for one listed row does not establish the extent of a whole estate. These are ordinary research checks made more pressing by the speed at which material can now be assembled.

Rights, reuse and keeping the sources visible

Digital availability and permission to republish are separate questions. HES distinguishes categories of information and different reuse arrangements in its statement of public task. The Portal provides specified spatial downloads under the Open Government Licence, with exceptions and an attribution statement. Trove has its own licensing terms for material offered for educational reuse.^{12, 13, 14}

A website should check the terms for the particular item and intended use. Credit the photographer or creator, the collection and the licence where applicable. A dataset’s open licence should not be assumed to cover every associated photograph or an Ordnance Survey background map.

Links and identifiers help keep the national resource visible even where a platform writes its own summaries. They also allow a correction or new designation to be traced back to its source. Quoted description, source-based summary, local testimony and editorial interpretation should remain distinguishable.

A small local project, worked through

A group preparing a housing walk could begin by choosing a few developments and finding their national and local records. It would then compare historic and current photographs, map the streets and consult the relevant council appraisal or housing archive. AI could help organise the notes and suggest a sequence of stops.

Before publishing, the group would check each stop against the original sources, confirm which buildings the photographs show, obtain the necessary rights and use public viewpoints. Draft captions would identify the image date and any buildings since demolished. Contributors’ recollections could be attributed with their agreement.

After publication, a source-linked correction log would allow residents and researchers to improve the guide. Material offered to a national collection would follow that collection’s contribution process. The local guide would provide a welcoming route into the evidence, while its links would support further investigation and continuing use of the national resources.

Continuing access and public investment

Search interfaces and AI tools can change quickly. The underlying work of surveying, describing, conserving, digitising and maintaining identifiers requires sustained expertise and investment. Wider access benefits from those foundations remaining dependable, well documented and available to future users.

The possibilities for citizen participation are substantial: more people can notice buildings, ask informed questions, share images responsibly and tell well-supported local stories. National resources supply the bedrock; additional platforms can help more people find and use what that bedrock makes possible.

Services, links and published feature descriptions checked 6 November 2026. The AI uses discussed above are editorial proposals; individual services’ available features should be checked on their live sites.