Documentary pages live on images: a harbour in 1918, a convoy on a map, a ship in a storm. Footage and photographs are also the most common way a small history page gets into trouble. People assume that anything old, or anything held by an archive, or anything government-made, is free for anyone to use. Each of those assumptions is partly true and dangerously incomplete.
This post sets out what "public domain" actually means for archival material, where the traps are, and how to credit and track what you use. It includes a small checker for a rights log. This post is general information, not legal advice. Copyright law differs by country and by the facts of each case, and I am not a lawyer. For anything high-stakes, ask a qualified copyright lawyer.
Public domain depends on where you are
"Public domain" isn't a universal label. It is a legal status in a particular country, and it differs between countries. The Duke Center for the Study of the Public Domain makes this point on its own Public Domain Day page: its explanations are about US law only, and terms in other countries are different. For an audience in the US, the UK and Canada, a work can be free to use in one and protected in another, so say clearly which country's rules you rely on, and when in doubt be more cautious.
Route one: US government works
Under section 105 of the US Copyright Act, copyright protection is not available for any work of the United States Government. In practice that covers material made by federal agencies and by officers and employees as part of their official duties, such as many military photographs and films. This is the most useful category for military and environmental history pages.
But three limits matter:
- It is a US rule. Not being protected by US copyright says nothing about protection elsewhere.
- Contractors are not employees. Material made for a government agency by an outside company or a donor isn't automatically a government work.
- Other people's material can be inside. A government film may include music or footage from a third party.
The archive trap
The National Archives' own page on permissions for special media is blunt about this. Works created by federal agencies or employees are treated as being in the public domain, it says, but not all records in its holdings are, "especially" its special media holdings. Some records may contain material restricted by copyright law, contract, deed restrictions or publicity rights. It also says it does not confirm copyright status, does not license its content, and that where a status is unknown it may supply the item with a warning that responsibility for later use lies with the user.
The Library of Congress takes a similar line in its collection-level statements. For one photograph collection, it tells researchers to watch for items published in the United States less than 95 years ago, says you are responsible for deciding whether your use is legal, and notes that you may also need permission from holders of publicity or privacy rights. Each collection has its own statement and its own credit line, so read the one that applies.
The lesson is simple: being in an archive doesn't make an item free. It makes the archive the place to read the rights statement.
Route two: age
Duke's Public Domain Day page records that on 1 January 2026, works published in the United States in 1930 entered the US public domain, as did sound recordings from 1925. So a US-published film or photograph from 1930 or earlier is now generally out of copyright in the US, subject to the usual checks. That is not the rule in every country, and unpublished works follow other rules. In the UK, the National Archives explains, a published work protected by Crown copyright lasts 50 years from the end of the year of publication, while an unpublished one lasts 125 years from the end of the year it was made or until 31 December 2039, as the National Archives puts it. That means an old British government film might be free of copyright or might not, depending on whether it was published, so don't assume.
Route three: licences and permission
Many archives and museums publish material under an open licence or ask for permission. In the UK, Crown copyright material should generally be available for re-use under the Open Government Licence. Whatever licence you rely on, write down its name and the page you found it on, and follow its conditions, which usually include a credit.
Credit, properly
The National Archives asks that audiovisual productions that use its special media records carry the line Courtesy: National Archives and Records Administration. The Library of Congress gives each collection its own credit line. Use the wording the source gives, put it where the audience can see it (in the video, and in the caption for a social post), and don't treat a credit as a substitute for permission. Credit tells people where something came from. It doesn't change who owns it.
A rights log you can check by machine
For every clip or image, I'd keep one row: what it is, the web address where you found it, the rights basis you relied on, the credit line, the date you checked, any restrictions, and notes. The script below reads such a log and flags rows that are incomplete: a missing credit line, an unknown rights basis, a source that is not a web address, a missing note where a note is needed, a date that is malformed, in the future or more than a year old. It warns on fair use every time, because that is a judgement a court makes after the fact.
rights_log.py
"""Check a rights log for archival footage and images: every item needs a source, a rights basis and a credit.
python rights_log.py rights.csv
Exit status 1 if any row has an ERROR. This checks that the paperwork is complete; it does NOT decide whether a
rights basis is correct. That judgement is yours (or a lawyer's).
"""
import csv
import sys
from datetime import date, datetime
COLUMNS = ["item", "source_url", "rights_basis", "credit_line", "checked_on", "restrictions", "notes"]
BASES = {
"us-gov-work": "made by a US federal officer or employee as part of official duties (17 U.S.C. 105)",
"pd-age": "public domain through age (you recorded the publication year and the rule you relied on)",
"licence": "an open licence (CC BY, OGL ...): the licence name goes in notes",
"permission": "written permission from the rights holder: where the email is filed goes in notes",
"fair-use": "you assessed fair use / fair dealing: the reasoning goes in notes (this is the riskiest basis)",
"own": "you created it",
}
NEEDS_NOTES = {"pd-age", "licence", "permission", "fair-use"}
STALE_DAYS = 365
def check(rows, today=None):
today = today or date.today()
out = []
for n, r in enumerate(rows, start=2): # row 1 is the header
name = (r.get("item") or "").strip() or f"row {n}"
problems = []
basis = (r.get("rights_basis") or "").strip()
if basis not in BASES:
problems.append(("ERROR", f"rights_basis must be one of: {', '.join(BASES)} (got {basis!r})"))
if not (r.get("source_url") or "").strip().startswith(("http://", "https://")) and basis != "own":
problems.append(("ERROR", "source_url missing or not a web address: you must be able to find it again"))
if not (r.get("credit_line") or "").strip():
problems.append(("ERROR", "credit_line is empty"))
if basis in NEEDS_NOTES and not (r.get("notes") or "").strip():
problems.append(("ERROR", f"{basis} needs a note explaining the basis"))
try:
d = datetime.strptime((r.get("checked_on") or "").strip(), "%Y-%m-%d").date()
if (today - d).days > STALE_DAYS:
problems.append(("WARN", f"rights last checked {d}: re-check, terms and sources change"))
if d > today:
problems.append(("ERROR", "checked_on is in the future"))
except ValueError:
problems.append(("ERROR", "checked_on must be a date like 2026-11-17"))
if basis == "fair-use":
problems.append(("WARN", "fair use is decided case by case by a court; consider a safer alternative"))
out.extend((name, sev, msg) for sev, msg in problems)
return out
def load(path):
with open(path, newline="") as f:
return list(csv.DictReader(f))
if __name__ == "__main__":
found = check(load(sys.argv[1]))
for name, sev, msg in found:
print(f"{sev}: {name}: {msg}")
print("OK" if not found else f"{len(found)} finding(s)")
sys.exit(1 if any(s == "ERROR" for _, s, _ in found) else 0)
A made-up example log, with one incomplete row on purpose:
rights_example.csv
item,source_url,rights_basis,credit_line,checked_on,restrictions,notes
Harbour photo (made-up example),https://example.org/item/1,us-gov-work,"Courtesy of Example Archive, Photographs Division",2026-11-17,none known,Created by a federal employee per the catalogue record
Convoy film clip (made-up example),https://example.org/item/2,pd-age,Example Library,2026-11-17,none known,Published 1921; US work more than 95 years old
Map scan (made-up example),https://example.org/item/3,licence,Example Museum / CC BY 4.0,2026-11-17,attribution,Licence stated on the item page
Newsreel clip (made-up example),https://example.org/item/4,permission,,2026-11-17,,
Running it flags only the last row: the credit line is empty and the permission has no note saying where the written permission is filed. The tests also check a bad basis and a bad URL, stale and future dates, and that your own work needs no URL. All four pass.
test_rights_log.py
import unittest
from datetime import date
from pathlib import Path
from rights_log import check, load
HERE = Path(__file__).parent
TODAY = date(2026, 11, 17)
class Tests(unittest.TestCase):
def test_example_file_flags_only_the_incomplete_row(self):
found = check(load(HERE / "rights_example.csv"), TODAY)
names = {n for n, _, _ in found}
self.assertEqual(names, {"Newsreel clip (made-up example)"})
msgs = [m for _, s, m in found if s == "ERROR"]
self.assertTrue(any("credit_line" in m for m in msgs))
self.assertTrue(any("permission needs a note" in m for m in msgs))
def test_bad_basis_and_url(self):
rows = [{"item": "x", "source_url": "somewhere", "rights_basis": "found-it-online", "credit_line": "c",
"checked_on": "2026-11-17", "notes": ""}]
msgs = " | ".join(m for _, _, m in check(rows, TODAY))
self.assertIn("rights_basis must be one of", msgs)
self.assertIn("source_url missing", msgs)
def test_stale_check_is_a_warning_and_future_date_an_error(self):
base = {"item": "x", "source_url": "https://e.org", "rights_basis": "us-gov-work", "credit_line": "c", "notes": ""}
stale = check([dict(base, checked_on="2025-01-01")], TODAY)
self.assertEqual([s for _, s, _ in stale], ["WARN"])
future = check([dict(base, checked_on="2027-01-01")], TODAY)
self.assertEqual([s for _, s, _ in future], ["ERROR"])
bad_date = check([dict(base, checked_on="17/11/2026")], TODAY)
self.assertEqual([s for _, s, _ in bad_date], ["ERROR"])
def test_own_work_needs_no_url_and_fair_use_always_warns(self):
own = {"item": "mine", "source_url": "", "rights_basis": "own", "credit_line": "Me", "checked_on": "2026-11-17"}
self.assertEqual(check([own], TODAY), [])
fu = dict(own, rights_basis="fair-use", source_url="https://e.org", notes="short clip for criticism")
self.assertEqual([s for _, s, _ in check([fu], TODAY)], ["WARN"])
if __name__ == "__main__":
unittest.main()
To repeat the limit: the checker only proves the paperwork is complete. It can't tell you whether "us-gov-work" really applies to a particular film.
Music, voices and people
Footage often has more than one layer of rights: the film, any music on the soundtrack, the narration and the people shown. A public-domain silent film may be fine while a modern recording of music on top of it isn't. On YouTube, the help page on three-minute Shorts notes that a Short over one minute with an active copyright claim of any type is blocked globally, so soundtrack choices matter. Use silent footage and your own narration where you can, and check the music separately.
What I haven't verified
- The rights status of any particular film, photograph or collection. Always read the item's own statement.
- Rules outside the US, apart from the UK Crown copyright durations quoted from the National Archives' page.
- How platforms such as Facebook treat claims on archival material.
- Any of this with a lawyer.
This post is general information, not legal advice. Copyright law differs by country and by the facts of each case, and I am not a lawyer. For anything high-stakes, ask a qualified copyright lawyer.
What I ran this against
- The rights-log checker was run with Python 3.13.5 (standard library only) on 17 November 2026. 4 tests pass. The example CSV in the post is made up and describes no real item.
- The checker verifies that your paperwork is complete (a source, a rights basis, a credit line, notes where needed, a date). It cannot judge whether your rights basis is correct, and nothing in this post was reviewed by a lawyer.
- I read the pages listed under sources on 17 November 2026. I did not request or license any footage, and I did not check any specific film or photograph's status. This post is general information, not legal advice. Copyright law differs by country and by the facts of each case, and I am not a lawyer. For anything high-stakes, ask a qualified copyright lawyer.
Sources checked
- National Archives (NARA), "Permissions / Rights for Use of Special Media Records at NARA" (works of US federal agencies and employees are not eligible for US copyright and are treated as public domain, but not all NARA holdings are; NARA does not confirm copyright status or license its content; the user is responsible; suggested audiovisual credit "Courtesy: National Archives and Records Administration"; page last reviewed 25 April 2023): archives.gov, read 17 November 2026.
- Library of Congress, "Rights and Access" page for the Detroit Publishing Company collection, as an example of its collection-level statements (watch for items published in the US less than 95 years ago; you are responsible for deciding whether your use is legal; collection-specific credit line; publicity and privacy rights may also apply): loc.gov, read 17 November 2026. Other collections have their own statements.
- 17 U.S. Code § 105, "Subject matter of copyright: United States Government works" (copyright protection is not available for any work of the United States Government; the section also has narrow provisions for certain Defense-related faculty works): law.cornell.edu, read 17 November 2026.
- Duke Center for the Study of the Public Domain, "Public Domain Day 2026" (on 1 January 2026, works published in 1930, and sound recordings from 1925, entered the US public domain; the site is about US law only and other countries' terms differ): web.law.duke.edu, read 17 November 2026.
- The National Archives (UK), "Crown copyright FAQs" (published Crown copyright works are protected for 50 years from the end of the year of publication; unpublished works for 125 years from the end of the year they were made or until 31 December 2039; Crown copyright material should generally be available for re-use under the Open Government Licence): nationalarchives.gov.uk, read 17 November 2026.