Open data and how to cite.
132,697 catalogue records are available as plain metadata, one row per record. Every row keeps the original institution, its identifier and link, the date as the source states it, and the rights statement, so a result can be traced back to the holder of the original.
What this dataset is, and is not
It is a discovery index of records from public institutions, normalised by automated adapters. It is not a copy of their holdings, not a complete list of what they hold, and not reviewed record by record by historians. Dates, countries and rights are kept as the source gave them; where PastTrace derives a value (for example a single sortable year) the column says so below.
Get the data
Each request returns 100 records and a cursor for the next page. Follow x-next-cursor (or next in the JSON form) until it is empty. The limit is 30 pages a minute per client.
# one archive, CSV, first page
curl "https://pasttrace.com/api/dataset?provider=<id>&format=csv"
# next page: pass the last id as ?after=
curl "https://pasttrace.com/api/dataset?provider=<id>&format=jsonl&after=<cursor>"
# one record in a citation format
curl "https://pasttrace.com/api/record/<record-id>?format=bibtex" # json | jsonld | csl | bibtex | risBy archive
- Library of Congress · Sanborn Maps50,600 records · JSON
- Library of Congress · Stereograph Cards39,478 records · JSON
- Early catalogue · Internet Archive · FedFlix, NASA Image and Video Library, Wikimedia Commons and others12,048 records · JSON
- Filmoteka Narodowa · National Film Archive (Poland)6,032 records · JSON
- Internet Archive · FedFlix & National Archives4,998 records · JSON
- Library of Congress · Sanborn Map Sheets4,044 records · JSON
- Europeana Photography · Israel collection2,956 records · JSON
- Library of Congress · Free to Use and Reuse2,608 records · JSON
- Europeana · Global historical collections2,269 records · JSON
- National Film Archive · Czechia1,802 records · JSON
- The Metropolitan Museum of Art1,463 records · JSON
- Japan Search1,211 records · JSON
- Early catalogue · Internet Archive · FOIA & Official Dossier, Internet Archive · Archival Motion Picture, Internet Archive · FOIA & Official Historical Collections and others1,112 records · JSON
- Early catalogue · Internet Archive · NASA Historical Photographs, Internet Archive · Universal Newsreels1,000 records · JSON
- Cinemateca Portuguesa · Museu do Cinema533 records · JSON
- Wikimedia Commons217 records · JSON
- Swedish Film Institute · Filmarkivet102 records · JSON
- Smithsonian · Open Access Historical Collections92 records · JSON
- Wellcome Collection82 records · JSON
- Internet Archive · Prelinger Archives25 records · JSON
- Europeana · Cinecittà Luce Photographs20 records · JSON
- Gallica · BnF5 records · JSON
CSV files start with a byte-order mark so spreadsheet programs read UTF-8 correctly. Text cells that begin with =, +, - or @ are prefixed with an apostrophe so a title can never run as a formula; numbers are untouched.
Metadata reuse terms — v1
Titles, descriptions, images and recordings belong to the institutions that hold them and stay under their stated terms: read rights_status and rights_text on every row, and the rights guide. Public access to a file is not permission to reuse it. The fields PastTrace prepares itself (the ones marked below as normalised or created by PastTrace) are offered under CC BY 4.0: reuse them freely, and cite the dataset.
These metadata-only terms also apply to the PastTrace-created or normalised fields in the Crime & investigations index and its case datasets. They do not license original documents, photographs, recordings or institution-supplied descriptions. Follow the source rights statement for each item.
How to cite
A record. Cite the original institution, using the source_url and source_id columns. Each record page and /api/record/<id> offers BibTeX, RIS and CSL-JSON that already name the institution as publisher and add the date you accessed it. The PastTrace page is a locator, not the authority.
The dataset. PastTrace. PastTrace archive catalogue metadata, version 2026.10.31. https://pasttrace.com/data. Accessed [date].
A permanent DOI for tagged dataset snapshots is planned through Zenodo and will be shown here once deposited; until then, cite the version and the access date.
Data dictionary
| Column | Type | Origin | Meaning |
|---|---|---|---|
id | string | Created by PastTrace | Permanent PastTrace record identifier. Stable across dataset versions; the record page is /record/<id>. |
title | string | Institution (verbatim) | Title as supplied by the originating institution. Never rewritten; any translation is a separate field. |
kind | string | Normalised by PastTrace | Media type: film, image, audio, map or document. |
year | integer | Normalised by PastTrace | Single indexed year (negative = BCE) used for sorting and filters. For ranges and approximate dates it is an index value, not a claim; read date_precision. |
date_text | string | Institution (verbatim) | The date exactly as the source states it, including "circa" and ranges. |
date_precision | string | Normalised by PastTrace | year, decade or range when the source date is approximate; empty when not stated. |
date_confidence | string | Normalised by PastTrace | How firmly the source supports the date, where the adapter could tell. |
country | string | Normalised by PastTrace | Country or region named in the source metadata. It describes a subject or association, not necessarily where the item was captured. "Unknown" is kept as unknown. |
place | string | Institution (verbatim) | Place text from the source, where given. |
source | string | Institution (verbatim) | Name of the archive or collection that holds the original. |
source_id | string | Institution (verbatim) | The institution's own identifier for the item. |
source_url | url | Institution (verbatim) | Link to the original record at the institution. Cite this, not the PastTrace page. |
creator | string | Institution (verbatim) | Creator as stated by the source. Placeholders such as "Unknown" are exported as empty. |
rights_status | string | Normalised by PastTrace | PUBLIC_DOMAIN, OPEN_LICENSE, REUSE_ALLOWED_WITH_CONDITIONS, RIGHTS_UNCLEAR, UNKNOWN or RESTRICTED. Public access to a file is not the same as permission to reuse it. |
rights_text | string | Institution (verbatim) | The institution's rights statement, unaltered. |
access_status | string | Normalised by PastTrace | PUBLIC, LOGIN_REQUIRED or UNAVAILABLE at the last check. |
original_language | string | Institution (verbatim) | Language code of the original description, where detected or stated. |
pasttrace_url | url | Created by PastTrace | Canonical PastTrace page for the record. |
indexed_at | date-time | Created by PastTrace | When PastTrace last normalised the record. |
Known limitations
- About three quarters of the records come from a few very large, repetitive sets (for example map sheets and stereograph cards). Counts describe the index, not the diversity of the sources.
- Country names describe the subject or association stated by the source, not where an item was captured. Unknown places stay unknown.
yearis one sortable value. For “circa”, decade and range dates it is an index convenience; readdate_textanddate_precision.- Meaning-based retrieval covers a part of the catalogue only; see the methodology and its evaluation.
- Rights statuses are the source’s, normalised. They are not legal advice. Report errors through corrections.
