How to Find Primary Sources for a History Paper
A historical primary source is original evidence from the time you are studying — a letter, newspaper, photograph, census schedule, oral history, or building — not a later textbook that retells it. The Library of Congress and the American Historical Association both treat that line as real and as question-dependent. This page is the where do I actually find the sources for a history paper? guide. For the type contrast across disciplines, use Primary vs secondary sources. For the archive visit itself, use How to find and use archives.
By CitationEasy Editorial Team23 min read

On this page
Quick Answer: How to Find Historical Primary Sources
Start with a research question, not a database. Then search from the cheap, indexed layer down to the unique box on a shelf. History papers usually cite what you find in Chicago Notes-Bibliography.
A working search order
- Name the question. "How did newspaper editors in 1918 describe the influenza?" is searchable. "World War I home front" is a topic, not a hunt.
- Search aggregators first. Digital Public Library of America, the Library of Congress digital collections, the National Archives Catalog, and ArchiveGrid.
- Then newspapers. Chronicling America digitizes selected U.S. pages from 1756 through 1963. It is a sample, not every American newspaper.
- Then a specialist collection. Founders Online for the founding generation; Born in Slavery for WPA narratives; DocsTeach for classroom-ready NARA scans.
- Then a finding aid. If the item is unpublished, ArchiveGrid or the repository site should lead to a collection guide with box and folder numbers.
- Cite so a stranger can retrace it. Use a persistent URL for a digital edition. Do not invent a box number you never saw.
What Counts as a Historical Primary Source?
The Library of Congress Ask a Librarian page (last updated 17 August 2020) gives the working definition students meet first:
Library of Congress, Ask a Librarian
A primary source is a first-hand record of an event or topic created by a participant in or a witness to that event or topic.
Although primary sources can include first-hand accounts that were documented later, such as memoirs or oral histories, primary sources created or written closest to the time of the actual event are considered to be the most useful sources for research purposes.
Archivists and rare-book librarians use a slightly wider formula. The Guidelines for Primary Source Literacy, approved by ACRL in February 2018 and by the Society of American Archivists Council in June 2018, open with this sentence:
Guidelines for Primary Source Literacy (ACRL/RBMS and SAA, 2018)
Primary sources are materials in a variety of formats that serve as original evidence documenting a time period, an event, a work, people, or ideas.
That is why a 1918 newspaper, a 1937 WPA interview, a 1963 photograph, a census microfilm, a city ordinance, a quilt, and a born-digital email can all be primary sources. Format does not decide. The question does.
The AHA rule: the question draws the line
The American Historical Association's Statement on Standards of Professional Conduct (wholly revised 2005; latest amendments January 2023; AHA published date 7 January 2023) is the profession's ethics document, not a library handout. It is also the cleanest official statement of the rule students get wrong:
American Historical Association, Statement on Standards of Professional Conduct
By "documents," historians typically mean all forms of evidence—not just written texts, but artifacts, images, video, statistics, oral recollections, the built and natural environment, and many other things—that have survived as records of former times. By "secondary literature," we typically mean all subsequent interpretations of those former times based on the evidence contained in primary documents. This distinction between primary and secondary sources is among the most fundamental that historians make. Drawing the boundary between them is a good deal more complicated than it might seem, since determining whether a document is primary or secondary largely depends on the questions one asks of it.
Worked cases of that rule:
- A 1918 issue of a city newspaper is a primary source for wartime advertising, local politics, or how editors described the influenza. The same paper is a secondary source if it is retelling the Civil War fifty years later.
- Eric Foner's Reconstruction (1988; updated editions later) is secondary for Reconstruction. It would be primary if you were studying how late-twentieth-century historians rewrote that field.
- A 2024 textbook chapter on the New Deal is tertiary or secondary for the 1930s. The letters distressed farmers sent to Franklin D. Roosevelt — the examples the National Archives uses in History in the Raw — are primary.
The Library of Congress pairs the same idea with named objects. These are LOC's own examples, not invented citations:
| Primary in LOC's pairing | Secondary in LOC's pairing |
|---|---|
| An original painting by Mary Cassatt | A book about the artist Mary Cassatt |
| President Lincoln's Gettysburg Address | An article about Abraham Lincoln |
| A photograph of Harry Houdini | A website about Houdini's magic tricks |
| An original Gershwin musical score | A recent recording of Gershwin songs |
Why Historians Still Go After the Raw Record
The National Archives' teaching essay History in the Raw (last reviewed 15 August 2016) states the pedagogical case in one sentence:
National Archives, History in the Raw
Documents--diaries, letters, drawings, and memoirs--created by those who participated in or witnessed the events of the past tell us something that even the best-written article or book cannot convey.
Primary sources force students to realize that any account of an event, no matter how impartially presented it appears to be, is essentially subjective.
NARA's essay is also honest about the cost. The same documents that humanize the past carry the creator's social position, the archive's collecting habits, and your own biases into the reading room. A WPA interview recorded in 1937 is a primary source for memory of slavery and a primary source for 1930s interview practice. It is a weak source if you treat it as a verbatim 1850s transcript.
The AHA statement adds a digital-era duty the old paper hunt did not have. After the primary-versus-secondary paragraph, it warns that a growing share of recent records is born digital, and that "the ease with which digitally preserved records are reproduced and accessed" makes accurate citation more vital, not less.
What This Page Is Not
Three sibling guides already cover nearby jobs. Keep them distinct:
- Primary vs secondary sources is the type lesson for every discipline, including lab data and literary works. This page is the history-specific hunt.
- How to find and use archives is the visit: registration, pencils, handling rules, reproduction policies. This page tells you which collection to request.
- How to cite archives in Chicago is the note and bibliography form. This page stops at enough citation to keep you from losing the item.
Where to Search: Digital Collections That Still Exist
Several destinations older syllabi still print have moved. Use the current homes below. Every URL here is one you can open without a proxy; subscription databases are labelled as such.
Aggregators (start here when you do not yet know the repository)
- [Digital Public Library of America](https://dp.la/). One search across contributing U.S. libraries, archives, and museums. Use it to discover which institution holds the scan, then click through to that institution's record.
- [Europeana](https://www.europeana.eu/). The European Union's cultural-heritage aggregator. Same habit: treat it as a switchboard, not the citation.
- [ArchiveGrid](https://researchworks.oclc.org/archivegrid/). OCLC's index of archival collection descriptions. This is how you find an unpublished collection that has a finding aid but no full-text scan.
- [HathiTrust](https://www.hathitrust.org/) and [Internet Archive](https://archive.org/). Full-text books, pamphlets, and government reports. Public-domain scans are the ones you can actually read off campus.
- [World Digital Library](https://www.loc.gov/collections/world-digital-library/). The UNESCO/Library of Congress project now lives on loc.gov. The old wdl.org site was archived after the 2020 charter ended; do not cite a dead wdl.org URL as if it were still the collection.
Government and official records
- [National Archives Catalog](https://www.archives.gov/research/catalog). Federal records, including digitized series. DocsTeach is NARA's classroom subset of the same holdings.
- [Library of Congress digital collections](https://www.loc.gov/collections/). The old American Memory brand was folded into this site. Search the current collections index rather than an American Memory URL from a 2008 syllabus.
- [GovInfo](https://www.govinfo.gov/). Official U.S. government publications. FDsys, the name older guides still print, was retired.
- [Founders Online](https://founders.archives.gov/). Annotated papers of Washington, Franklin, Jefferson, the Adamses, Hamilton, and Madison. A real letter, opened for this page: John Adams to Thomas Jefferson, 13 July 1813.
- [CIA FOIA Electronic Reading Room](https://www.cia.gov/readingroom/). Includes the CREST 25-year program archive. Declassified files are still government records; they are not automatically public-domain text you can republish without reading the stamp.
- [The National Archives (UK)](https://www.nationalarchives.gov.uk/) and [Library and Archives Canada](https://www.bac-lac.gc.ca/). Use these when the question is British or Canadian, not as a substitute for NARA.
Newspapers
Chronicling America is the National Digital Newspaper Program: selected historic U.S. pages, permanently hosted by the Library of Congress and funded with the National Endowment for the Humanities. The live site searches digitized pages from 1756 through 1963 and a U.S. Newspaper Directory of titles from 1690 to the present. It is not a complete run of every American title. State partners choose which papers to digitize. If your town paper is missing, that is a selection decision, not proof the paper never existed.
Library subscriptions still matter for the gaps: ProQuest Historical Newspapers, Gale's nineteenth-century U.S. collections, and the British Newspaper Archive. Cite the paper, date, page, and the database or Chronicling America persistent URL — not "online newspaper." See How to cite a newspaper in Chicago for CMOS 18 §§14.89–14.98.
Subject collections with identifiable items
- U.S. political and legal texts. Avalon Project (Yale Law Library) for treaties, constitutions, and diplomatic documents. These are transcriptions: check the bibliographic note before you treat a web page as the archival original.
- World history teaching packets. Fordham's Internet History Sourcebooks and BYU's EuroDocs are curated gateways. They are excellent for orientation and weak as your only footnote.
- Civil War communities. Valley of the Shadow (University of Virginia) publishes paired sources from Augusta County, Virginia, and Franklin County, Pennsylvania.
- Slavery and emancipation memory. Born in Slavery: Slave Narratives from the Federal Writers' Project, 1936 to 1938 at the Library of Congress. These are 1930s interviews about earlier lives. Cite the narrator, the interviewer, the date of the interview, and the LOC item — and say that the recording is a New Deal interview.
- Japanese American incarceration. Densho publishes encyclopedic context plus oral histories and photographs. Use the encyclopedia as secondary; use the interviews and documents as primary.
- Women's activism. Women and Social Movements (Alexander Street) is a licensed database your library may subscribe to. Do not list it as a free site.
How to Search When the Words Have Changed
Catalogers and OCR engines store the words on the page, not the words you would use in 2026. If you only search modern vocabulary, you will miss the folder that actually holds the letters.
Period language, not a slur you have to repeat in your prose
- Modern "World War I" or "First World War" → contemporary U.S. and British papers often say "the Great War" or "the European War."
- Modern "influenza pandemic of 1918" → many 1918 papers say "Spanish influenza," "the grip," "la grippe," or just "the epidemic." The Library of Congress Spanish Flu topic guide lists those terms on purpose.
- Modern "mental illness" → asylum and court records may say "insanity," "lunacy," or a diagnosis that no clinician would use now.
- Modern "enslaved people" → the catalog and the document will often use the period's dehumanizing legal terms. Search those terms in the database; do not recycle them as your own voice unless you are quoting.
Other search habits that actually move the needle:
- Try every form of a name: John Smith, J. Smith, Smith, John, a married name, a title ("Mrs. John Smith"), and a misspelling the clerk used.
- Add a place and a date limiter before you add a third keyword. "Strike 1919 Seattle" beats "labor unrest."
- Read the subject headings on one good hit, then search those headings. Library of Congress Subject Headings are a controlled language; keywords are not.
- When OCR is dirty (long s, faded microfilm), search a distinctive proper noun instead of a common noun.
- If a digital collection has no full text, search the finding aid for the folder title, not the document text you wish were there.
Worked Hunt: Newspapers on the 1918 Influenza
This is a hunt you can repeat. The Library of Congress research guide Influenza Epidemic of 1918 (Spanish Flu): Topics in Chronicling America (last updated 12 June 2025) tells you to search the digitized papers between 1918 and 1920 and to combine period words: Spanish influenza, Spanish flu, grip, la grippe, plague, epidemic, pandemic, and Pfeiffer's bacillus.
The same guide's selected-articles list points at identifiable pages, including "Epidemic of Grip and Pneumonia," The Ogden Standard (Ogden City, UT), 8 May 1918, 3:30 P.M. City Edition, page 4. That bibliographic line is LOC's, not an invented citation. Open the linked page from the guide; do not reconstruct a URL from the headline.
A page opened for this rewrite is the 16 October 1918 Evening Star (Washington, D.C.), page 6, whose editorial columns treat influenza as a wartime efficiency problem. The persistent Chronicling America URL is the citation, not "online newspaper":
Chicago note for a Chronicling America page (real item)
Evening Star (Washington, DC), 16 October 1918, 6, Chronicling America, https://chroniclingamerica.loc.gov/lccn/sn83045462/1918-10-16/ed-1/seq-6/.
That form follows CMOS 18 §§14.89–14.98: the dated newspaper is the source; the URL or database name locates the copy you read. History papers often keep newspaper stories in notes and omit them from the bibliography unless the paper itself is central to the argument. For the full newspaper pattern, including bylines and news sites, use How to cite a newspaper in Chicago.
Finding Aids and Physical Archives
A finding aid is the collection's map: creator, dates, extent, arrangement, restrictions, and a box/folder inventory. Read it before you email the archive or book a trip. The How to find and use archives guide covers the visit. The research failure that happens before the visit is requesting "everything on the Civil War" from a repository whose finding aid already lists the three folders that match your question.
What to steal from a finding aid
- Scope and content. The one-paragraph claim about what the collection actually is.
- Biographical or historical note. Context for the creator — and a warning about what the creator never saw.
- Arrangement. Chronological, alphabetical, and "in original order" collections search differently.
- Box and folder list. This is the request slip. Write the collection title, box, and folder before you travel.
- Restrictions and language of materials. Classified files, donor seals, and untranslated scripts are not surprises you want at the desk.
Where the paper still lives:
- National and state archives for government records.
- University special collections for personal papers and organizational records that arrived as gifts.
- Historical societies for local newspapers, city directories, and family collections.
- Museum archives for exhibition files and object documentation.
- Corporate and religious archives for firms and congregations; access is often by appointment and sometimes by permission.
- Community archives built by the people the official repository never collected. These are often the only place a marginalized community's record survived.
Worked Examples You Can Recheck
A published letter with a stable URL
On 13 July 1813 John Adams wrote Jefferson from Quincy about the French Revolution, aristocracy, and Adams's own Defence of the Constitutions. The annotated text is on Founders Online, a National Archives / University of Virginia Press edition of the papers. That is a real letter. A Chicago note can point at the digital edition:
Chicago note for a Founders Online letter (real item)
John Adams to Thomas Jefferson, 13 July 1813, Founders Online, National Archives, https://founders.archives.gov/documents/Jefferson/03-06-02-0238.
Do not invent a box and folder for a letter you only read online. If you later see the manuscript at the Massachusetts Historical Society or the Library of Congress, add that repository's identifiers. The fabricated "Box 45, Folder 3" habit is how students produce citations no archive can fill.
The CMOS 18 collection example
The Chicago Manual of Style, 18th edition (University of Chicago Press, 2024; doi:10.7208/cmos18), keeps unpublished manuscripts in chapter 14. CMOS 18 §14.119 states the identification rule; §§14.119–14.129 cover unpublished collections. University CMOS 18 guides reprint the handbook's long-running Egmont example:
Chicago Manual of Style 18, §14.119
Full identification of most unpublished material usually requires giving the title and date of the item, series title (if applicable), name of the collection, and name of the depository. Except for placing the cited item first [in a note], there is no general agreement on the sequence of the remaining elements in the citation. . . . Whatever sequence is adopted, however, should be used consistently throughout the same work.
Chicago Manual of Style manuscript example
Note
James Oglethorpe to the Trustees, 13 January 1733, Phillipps Collection of Egmont Manuscripts, 14200:13, University of Georgia Library (hereafter cited as Egmont MSS).
Bibliography (the collection, not every letter)
Egmont Manuscripts. Phillipps Collection. University of Georgia Library.
Two CMOS habits students skip: the note cites the item; the bibliography usually cites the collection. Specific items go in the bibliography only when you used just one piece from that collection (CMOS 18 §14.120).
A New Deal interview about an earlier life
The Library of Congress collection Born in Slavery holds more than 2,300 first-person accounts collected by the Federal Writers' Project in 1936–1938 and assembled in 1941 as the seventeen-volume Slave Narratives microfilm. A named Virginia interview in that set is Fannie Berry of Petersburg, recorded by Susie Byrd, in Federal Writers' Project: Slave Narrative Project, Vol. 17, Virginia, Berry–Wilson. Cite the narrator, the interviewer, the interview date printed on the typescript, the volume, and the LOC resource. Date the source to that interview, not to the 1850s.
A later, separately recorded oral history is Fountain Hughes, interviewed by Hermond Norwood in Baltimore on 11 June 1949, in the American Folklife Center's Voices Remembering Slavery collection. That recording is not a Federal Writers' Project 1936–1938 typescript. Do not file it under Born in Slavery just because both are first-person memories of slavery.
A labelled template when you do not yet have the folder number
If the finding aid is incomplete, do not invent identifiers. Use a template and fill only what you have:
Template (not a real citation)
Creator to Recipient, Day Month Year, Series Title (if any), Collection Name, box X, folder Y, Repository, City, State, persistent URL if you used a scan.
How to Evaluate What You Found
Every primary source is someone's version of events. NARA's History in the Raw names the two facts students discover the hard way: the record reflects the participant's point of view, and you bring your own. Ask the five questions archivists teach, then add the two the 2018 Guidelines insist on.
Questions that belong on every source card
- Who made this? Name, office, race, gender, literacy, and what they stood to gain.
- What is it? A private diary, a published pamphlet, a form letter, a posed photograph, a later memoir.
- When? Creation date versus the date of the event. A 1937 interview about the 1850s is both dates.
- Where? Geography and institution. A War Department memo and a church minute book do not see the same town.
- Why? Petition, propaganda, love letter, required report, grant application.
- What is missing? The 2018 Guidelines tell users to look for silences, not just bias in the surviving page.
- How did it reach you? Digitization crops, editorial selection, a database that only scanned page one.
Guidelines for Primary Source Literacy, Core Ideas (Theoretical Concepts)
Collections in cultural heritage institutions reflect and reinforce societal power structures. Users must seek to understand resulting silences and absences by critically considering what sources were never created, what sources may no longer exist, and what sources are collected, as well as communities' abilities to engage in these activities.
Authenticity is a separate job from bias. Provenance (who held the file), anachronisms, and a second contemporary source will catch more forgeries than a hunch about tone. The AHA statement is blunt: historians do not fabricate evidence, and an undetected counterfeit poisons later work that trusted it.
Failure modes that waste a week
- Database versus original. A scan can hide the verso, the envelope, the docketing, or the color of the ink. If your argument depends on those, you still need the box.
- No date, no author. Describe the item ("unsigned memorandum, filed with the 1919 strike correspondence") instead of writing "Anonymous, n.d." and walking away. See Cite a source with no date and Cite a source with no author.
- The newspaper is not an eyewitness. Many nineteenth-century papers reprinted telegraph stories and other cities' reporting. Compare two papers before you call a dispatch local testimony.
- The memoir is not the diary. Closest-in-time still wins, which is why LOC privileges contemporaneous records over later first-person accounts.
- Digitization is a sample. Chronicling America, DPLA, and NARA Catalog hits are the digitized slice. ArchiveGrid exists because most collections are still paper.
- You cited a syllabus URL. American Memory, FDsys, and wdl.org have moved. Prefer the current loc.gov, archives.gov, or govinfo.gov record.
- You mixed a newspaper with an archive form. A dated Chronicling America page follows CMOS 18 §§14.89–14.98. An unpublished letter in a named collection follows §§14.119–14.129. Do not invent a record group for a newspaper, and do not drop a letter into italic newspaper punctuation.
Working with Common Source Types
Letters and diaries give daily texture and a single person's stakes. They are weakest when you treat one eloquent writer as the whole town. Ask whether the diary was kept private or rewritten for heirs.
Official records (court files, census schedules, legislative hearings, after-action reports) are systematic and official. They omit informal work, unofficial violence, and anyone the state did not count. Read for what the form refused to ask.
Newspapers catch public argument and local advertising that never reached a bound volume. They also sell papers. Separate news columns from editorials, and check a second paper before you build a chronology on one headline.
Photographs, maps, and posters are composed. Studio portraits, government posters, and cropped news photographs are arguments with a lens. Research the photographer or agency; do not treat the image as an unmediated window.
Oral histories recover voices that never entered the official file. Memory is dated to the interview, not only to the event. Note the interviewer, the date, the language, and whether a transcript was edited.
Citing Historical Primary Sources
History departments almost always want Chicago Notes-Bibliography, not APA author-date. The AHA statement calls bibliographies and annotations the scholarly apparatus that lets the next historian retrace your steps. Minimum elements:
- Creator (or a description if there is no name)
- Title or a plain-language description of the item
- Date of creation
- Collection, series, box, and folder — or a digital edition's persistent URL
- Repository and city
- The scan's collection name if you never saw the paper
Use How to cite archives in Chicago for notes versus bibliography, How to cite a manuscript in Chicago for unpublished texts, How to cite a newspaper in Chicago for dated papers, and the history citation guide for the rest of a history paper. CMOS 18 is the current edition; do not follow a handout that still says 17th edition as if it were the latest. Federal record groups have their own leaflet: NARA General Information Leaflet 17.
Copyright, Credit, and People in the File
U.S. Copyright Office Circular 15A (revised April 2026) is the duration circular. For published U.S. works that were already under statutory protection before 1978, it states the current cutoff in one sentence:
U.S. Copyright Office, Circular 15A, Duration of Copyright
Applying these standards, all works published in the United States before January 1, 1931, are in the public domain.
That sentence is still not a free pass. Unpublished letters and diaries follow different duration rules (often life of the author plus 70 years, with older unpublished works guaranteed protection through at least 2002, and through 2047 if they were published by then). Foreign works can have restored copyright. A scan on loc.gov can be public domain and still require the credit line the collection asks for. Fair use may cover a short quotation in a student paper; it does not automatically cover publishing a gallery of photographs.
Ethics is larger than copyright. Handle trauma, medical files, and recent personal records as if the people in them could read your paper. When the historical language is abusive, quote it only when the wording itself is the evidence, and say so. The 2018 Guidelines ask you to consider how your work affects creators, donors, and communities — including communities whose records were never collected.
Frequently Asked Questions
- How do I find primary sources for a history paper?
- Write the question first, then search DPLA, the Library of Congress collections, the National Archives Catalog, and ArchiveGrid. Use Chronicling America for selected U.S. newspapers from 1756 through 1963, then a specialist collection such as Founders Online or Born in Slavery. If the item is unpublished, read the finding aid and request the box. Do not stop at a textbook bibliography and call those books primary.
- Is a newspaper a primary source?
- Yes, when you are studying the period that published it — reporting, ads, obituaries, editorials. No, when the paper is summarizing an earlier war or reprinting someone else's history. The AHA statement is explicit: primary versus secondary depends on the question you ask of the document.
- Can I use digitized primary sources in a serious history paper?
- Yes. Founders Online, Chronicling America, DocsTeach, and LOC collections are standard in undergraduate and scholarly work. Name the digital collection and the persistent URL. Remember that digitization is selective: most archival boxes are still not online, and a scan can hide the back of the page.
- What is a finding aid?
- It is the collection guide: who created the papers, what years they cover, how they are boxed, and any restrictions. ArchiveGrid and repository websites publish them. The folder list is what you put on a request slip. It is not a catalog of every letter's first line.
- How do I cite an archival letter in Chicago style?
- CMOS 18 §14.119 puts the item in the note (creator, description, date, collection, identifier, repository) and usually the collection in the bibliography. The manual's own example is James Oglethorpe to the Trustees, 13 January 1733, in the Egmont manuscripts at the University of Georgia Library. For a digital edition you actually opened, cite that edition's URL instead of inventing a box number.
- How do I cite a historic newspaper I found in Chronicling America?
- Cite the paper, the full date, a page if you saw one, and the persistent Chronicling America URL. CMOS 18 §§14.89–14.98 treat the dated newspaper as the source. A page opened for this guide is the Evening Star (Washington, DC), 16 October 1918, page 6. Do not write "online newspaper" and walk away.
- Are oral histories primary sources?
- Yes. LOC counts later first-hand accounts as primary, and then warns that records closest to the event are usually more useful for reconstructing the event. A 1937 WPA narrative is primary for 1930s memory work and only indirectly primary for the 1850s. Always date the interview.
- What if two primary sources contradict each other?
- That is normal. NARA's History in the Raw essay treats disagreement as the point: each account is situated. Ask who each creator was writing for, what they could not have known, and which claim you can check against a third record. Do not silently drop the source that spoils a clean story. The AHA statement says historians should not omit evidence that runs counter to their interpretation.
- Do I need permission to quote a letter from an archive?
- Reading and citing a document for a class paper usually does not require a permission letter. Publishing a long extract or reproducing a photograph often does, because copyright and the repository's reproduction contract are separate from the reading-room rules. Unpublished material is the risky category. Ask the archive before you put images in a thesis that will be posted online.
- When do historical publications enter the U.S. public domain?
- Circular 15A, revised April 2026, says works published in the United States before 1 January 1931 are in the public domain. That is the 95-year term for pre-1978 published works, not a rule for unpublished letters. Sound recordings and some foreign works follow other schedules. Credit the archive even when the copyright has expired.
- What if almost nothing was written about my topic?
- Look past elite correspondence: census schedules, court dockets, newspapers, material culture, oral histories, and community archives. The 2018 Guidelines treat absence as evidence of collecting power, not proof the people did not exist. Ask a reference archivist; they know the uncatalogued accession that never hits DPLA.
Cite the item so the next historian can find it
Once you have the letter, newspaper, or scan, put it in Chicago Notes-Bibliography with the collection and a persistent URL. The archive guide walks through notes versus the bibliography entry; the generator will format a manual entry if you type the elements yourself.