Freedmen’s Bureau Records: Reading a Family Past 1870
The federal government spent decades counting enslaved people without writing their names down. That single administrative choice is why so much African American family research halts at the 1870 census, and it is why two bodies of Reconstruction paperwork — the Freedmen’s Bureau files in Record Group 105 and the Freedman’s Bank depositor books in Record Group 101 — carry a weight nothing else does. They name people, their parents, their brothers and sisters, and in places the person who had owned them. This page is about what is genuinely in them, what is not, and how unevenly they survive.
The short version
The wall at 1870 is a records artifact, not a mystery. The 1860 census had two forms: one that asked for the name of every person, and a second, for enslaved inhabitants, whose only name column was the owner’s. The Freedmen’s Bureau files are the largest federal body that names the same people as people. Labor contracts, ration rolls, marriage registers, hospital books, complaint dockets and school lists, filed office by office between 1865 and 1872 and kept together as Record Group 105. The Freedman’s Bank signature books are a separate record group entirely — Record Group 101, the Comptroller of the Currency — and they are the richer of the two per page, because a clerk opening an account wrote down parents, siblings, children, spouse and sometimes the former owner. Coverage is wildly uneven and the Bureau kept no personal-name index of its own. Whether your family appears depends less on how they lived than on which officer ran which office, and how well that officer did the paperwork.
Watch: the two record groups, and why they get confused
A short walkthrough of what sits in Record Group 105, what sits in Record Group 101, and why the difference changes where you write.
Watch the walkthrough
Why the paper stops at 1870
It is worth being blunt about this, because the polite version of it obscures what actually happened.
Anyone researching an African American family long enough arrives at the same place: the trail is workable through the twentieth century, workable through 1900 and 1880, workable through 1870 — and then it goes flat. The usual shorthand for this is a “brick wall,” which makes it sound like an accident of survival, as though a courthouse burned. It was not an accident. It was the design of the form.
The 1860 census used two separate schedules. Schedule 1, Free Inhabitants, asked the enumerator to record the name of every person in the household, along with age, sex, occupation, the value of real and personal estate, and place of birth. Schedule 2, Slave Inhabitants, asked for something different. Its columns, as the National Archives sets them out on its own 1860 census reference page, were: name of slave owner, number of slaves, age, sex, color, fugitive from the state, number manumitted, a disability column, and number of slave houses. There is no column for the name of the person being counted. The only human being named on that form is the one claiming ownership.
That is the wall, stated exactly. Two forms went out in the same year under the same act, and one of them recorded a person and the other recorded a quantity. The consequence is arithmetic: for a family enslaved until 1865, the federal population schedules contain no name before 1870, because no federal enumerator was ever asked to write one down. The same page notes that slave schedules exist for sixteen states and the District of Columbia, so the gap is not regional. It is the whole of the enslaving South — Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Texas and Virginia — plus the border states of Delaware, Kentucky, Maryland and Missouri, plus Utah.
Understanding it as a design decision changes what you look for next. If the missing names were lost, you would search for surviving copies. They were never taken, so you search instead for the other federal record-keeping that did name people — and between 1865 and 1874 there are two bodies of it, created by two agencies, on a scale that has no equivalent before or since. Everything below is about those two. If you are further forward in time and stuck on a twentieth-century generation instead, that is a different discipline, and the companion piece on applying locate technique to a stalled family tree covers it.
Two acts of Congress, one afternoon
March 3, 1865. Volume 13 of the Statutes at Large. Two chapters apart, and the difference between them decides where you write your letter.
Chapter XC of that session is titled “An Act to establish a Bureau for the Relief of Freedmen and Refugees.” Its first line establishes, in the War Department, “a bureau of refugees, freedmen, and abandoned lands,” committed to “the supervision and management of all abandoned lands, and the control of all subjects relating to refugees and freedmen from rebel states.” You can read the page itself at 13 Stat. 507. Note what the body of the act calls the thing: the title says relief bureau, the operative text says refugees, freedmen and abandoned lands, and it is the operative text that became the name on the boxes.
Note also the duration. The act created the Bureau “to continue during the present war of rebellion, and for one year thereafter.” Congress built it to expire. It survived because Congress kept extending it — by acts of July 16, 1866 and July 6, 1868 — and then wound it down deliberately, ordering in July 1868 that the Bureau “be withdrawn from the several States within which said bureau has acted,” which took effect in early 1869. The National Archives microfilm pamphlet for the Bureau’s headquarters marriage records traces the statutory chain, including the final act of June 10, 1872 that abolished the Bureau effective June 30 of that year. For a researcher, those dates are a filter: an office that closed in 1868 cannot have filed anything in 1870, and a family that moved in 1871 will not appear in the state they left.
And two chapters later, a bank
Chapter XCII of the same session, on the same day, signed by the same President, is “An Act to incorporate the Freedman’s Savings and Trust Company.” Its Section 5 states the whole purpose in one sentence: the corporation’s business “shall be to receive on deposit such sums of money as may be from time to time offered therefor, by, or on behalf of, persons heretofore held in slavery in the United States, or their descendants, and investing the same in the stocks, bonds, treasury notes, or other securities of the United States.”
Two things in that text matter to a researcher. First, the statute confines the bank’s investments to federal government paper — which is why the 1870 amendment loosening that restriction is the hinge on which the bank’s collapse turns. Second, read the list of incorporators the act names in Section 1. Every one of them is identified as being of New York, Massachusetts, Rhode Island, Pennsylvania or Ohio. Not one Southern name, and not one of the depositors’ own community, appears among the men Congress trusted with the money. That is not a moral aside; it explains the institution’s later behavior and it explains why the records survive in the Comptroller of the Currency’s files rather than anywhere near the freedpeople who filled them.
The distinction almost every guide blurs
The Bureau’s records are Record Group 105. The bank’s records are Record Group 101. Different agency, different custody, different finding aids, different reading room. The National Archives says so plainly in its own published account of the bank, noting that “because of the bank’s close association with the Freedmen’s Bureau, researchers often confuse these records with those of the bureau, which is a separate body of records.” Commercial genealogy sites routinely bundle them under one heading, which is fine as marketing and misleading as method. When you write to ask for something, you have to know which of the two you are asking about, because the answer comes from a different desk.
What the Bureau’s clerks actually wrote down
Five groupings, and only two of them are worth your time if you are looking for a person rather than a policy.
The National Archives organizes Record Group 105 into five bodies: headquarters records, the state records of Assistant Commissioners, field office records, a separate series of marriage records collected at Washington headquarters, and the records of the Freedmen’s Branch of the Adjutant General’s Office, which carried on the claims work from 1872 to 1878.
Headquarters records are the least useful for family research, and the Archives says as much: they are “primarily official and statistical, but may contain some information on individuals at the local level.” Circulars, special orders issued by Commissioner Oliver Otis Howard, annual reports to the President, school schedules, rental accounts. Read them to understand what an officer was instructed to do; do not expect to find a grandmother in them.
The field offices are where the names are
The field office records are the opposite. These were kept by the men in daily contact with freedpeople — subassistant commissioners, agents, superintendents — and the Archives describes them as “rich with names and personal information of individuals,” listing marriage certificates, schooling information, labor contracts, hospital records, complaints, relief rolls, land applications, requests for legal aid and protection, and trial summaries.
The Archives’ published account of the microfilming program, in its magazine Prologue, is more specific still about what individual series hold. Ration records — registers, lists and applications from families who received relief — give “not only the names of people who received relief, but also their places of residence, the names of former owners, the reasons for their condition.” That phrase, the names of former owners, is the one that matters most, because it is the bridge. A ration roll that names both the freedperson and the person who had held them turns a 1866 document into a pointer at antebellum property records, probate inventories and deeds where the same names recur.
Labor contracts are the other high-yield series: thousands of agreements between freedpeople and planters, witnessed by Bureau officials, giving “the names of contracting parties, the obligations and responsibilities of each party, the period of service and wages, and the type of work to be performed.” A labor contract puts a named adult on a named piece of ground in a named year, with a named employer. For locating purposes that is a fixed point, and fixed points are what a search is built out of.
Beyond those two, the same account lists complaint and claimant registers, contracts of indenture, freedmen’s court papers, hospital registers, home colony registers, school and land reports. The indenture records deserve a flag of their own: apprenticeship was used in several states to bind Black children to former owners against their parents’ wishes, and the Bureau’s apprenticeship files are therefore both a genealogical source and a record of a family being separated. Read them knowing what they are.
The marriage series, and what it can prove
Marriage records were pulled together separately at Washington headquarters and cover 1861 to 1869 — earlier than the Bureau itself, because Union Army chaplains had begun solemnizing marriages in contraband camps during the war. The earliest item in the series is an October 1861 report by the Reverend L. C. Lockwood at Camp Hamilton, Virginia, listing thirty-two couples he had married that September.
What a Bureau marriage record can contain is unusual. Certificates from Louisiana, Mississippi and Tennessee run to the names and ages of both parties, their place of residence, the names and complexion of each of their parents, the number of years each had previously lived with another man or woman, how that relationship was separated, the number of children by that previous connection, and the witnesses. Read as a form, it is a marriage license. Read as a document, it is the state asking two people to itemize, on paper, the families that slavery took apart — and then keeping the answer. Those “previous connection” fields regularly name a first spouse sold away and children by that spouse, which is a whole additional line of descent that appears nowhere else.
A bank ledger that recorded whole families
The Freedman’s Bank signature books are, page for page, the densest family document the federal government produced in the nineteenth century.
The bank collected genealogy for an ordinary commercial reason: if a depositor died, it needed to know who could claim the balance. So the clerk opening an account asked for the family. The surviving registers from twenty-nine branches are reproduced on National Archives Microfilm Publication M816, fifty-five volumes in all, and the Archives’ published guide to the bank’s records lists what a full entry carries: name of the depositor, account number, age, complexion, date of application, place of birth, place raised, occupation, spouse, children, names of parents, brothers and sisters, remarks, and signature. Some of the earlier volumes add the names of former owners or mistresses and the plantations where depositors had lived. Some entries have a death certificate attached.
Read that field list against the 1860 slave schedule and the contrast is the entire argument of this page. One form records a tally under an owner’s name. The other, nine years later, records a woman’s mother, her father, her brothers, her sisters, her children, where she was born, where she was raised, and her signature in her own hand.
What one entry looks like in practice
The Archives’ own account works through a New Bern, North Carolina depositor named Dilla Warren, who opened account number 1333 in November 1869. Her record gives her birth in Chowan County, her age, her trade — she made her living sewing, knitting, washing and ironing — and then her family. Her husband, Pompey Nixon, had been sold away seventeen years before the Civil War. Eleven of her fifteen children are listed, with the ones who died and the ones who were sold marked as such. Her father had been sold thirty years before the war; her mother and two of her brothers seventeen years before it; one sister thirty-five years before Warren walked into that branch. She named a nephew to receive whatever was in the account when she died.
That is one page of one ledger in one branch. It contains four generations, two sales, a marriage, and an instruction about inheritance. No census before 1870 would have recorded any of it, and the reason we can read it now is that a bank needed to know who its depositor’s next of kin were.
The soldier’s route out of the register
Many of the earliest deposits came from Black soldiers receiving back pay and bounty money, and the signature forms had space for regiment and company. The Archives traces one: Jacob Reiley, twenty-two, Company B of the Thirty-third United States Colored Troops, who opened an account at the Savannah branch in April 1866. His signature record names his stepfather, his mother, a brother and three sisters, and places his natural father in the Barnwell District of South Carolina without naming him. His later pension file supplies the missing name. That is the pattern worth learning: a bank register hands you a regiment, and the regiment hands you a service record and a pension file, and the pension file is often where the relationship a family could not prove finally gets proved. If your line runs through a soldier, the same chain works forward as well as back, and the piece on tracing living descendants of a Civil War ancestor picks it up at the modern end.
One caution about the bank’s collapse, because it shapes what survives. The bank closed in June 1874. Of the depositors eligible to receive liquidation dividends, the Archives records that 61,131 were entitled and only 29,996 ever sent in their passbooks; some 31,000 never claimed anything. The claim files generated by that process — letters, passbooks, notarized heirship affidavits, questionnaires filled out by depositors who had lost their books — were never microfilmed and sit at the National Archives facility at College Park. They are, for some families, the only surviving document of the relationship, and almost nobody looks at them.
What each record set names, and where it runs out
Six sources, ranked by how much of a family one document can reconstruct.
| Record | Who it names | Where it fails you |
|---|---|---|
| Freedman’s Bank signature register (RG 101, M816) | Depositor, spouse, children, parents, siblings, birthplace, where raised, sometimes the former ownerDensest per page | Only twenty-nine branches survive; the registers carry no index of their own |
| Bureau marriage certificates and registers (RG 105) | Both parties, their parents, prior partners, children of a previous connection, witnesses | Wildly uneven by state; several states produced almost nothing at all |
| Bureau ration and relief rolls (RG 105) | Recipient, household, place of residence, and the name of a former owner | Captures only families destitute enough to apply for federal relief |
| Bureau labor contracts (RG 105) | The freedperson, the employer, the wage, the term, the work | Adults under contract only; children, the elderly and the self-employed fall outside |
| Freedmen’s Branch claim files (Adjutant General, 1872–78) | Black soldiers and sailors and their heirs, with relationships stated for bounty and pension purposes | Only ten states are represented, and only where a claim was actually filed |
| Federal population census before 1870 | For an enslaved family, nobody. Schedule 2 of 1860 named the owner and counted everyone else. This is the gap the rows above exist to fill. | |
Note the shape of that table. The two most informative records were both created for administrative convenience rather than for posterity — a bank protecting itself against disputed inheritance, and an agency documenting who it had married so the marriage would hold up in a state court. Neither was built to preserve a family. Both did, and the reason a surname suddenly connects to an aunt three counties over is usually a clerk being thorough on a Tuesday in 1867. Turning that connection into a living person today is records-based locate work of an entirely different kind, and it starts where the historical trail stops.
The coverage is wildly uneven, and the Archives says so
This is the part that gets flattened into one line elsewhere. It deserves the numbers.
The Archives’ own description of the microfilming program puts it in a single sentence: the Bureau “was inconsistent in the kinds of records it created for each state.” The marriage series shows exactly how inconsistent, because the same order went to every Assistant Commissioner and the results differ by orders of magnitude.
Commissioner Howard’s Circular Number 5, issued May 30, 1865, authorized Assistant Commissioners to designate officers to keep a record of marriages. What happened next depended entirely on the individual officer. The Archives’ pamphlet for the headquarters marriage series records the outcome state by state:
Where the officers did the work
Mississippi. Assistant Commissioner Samuel Thomas issued his own circular on July 3, 1865, instructing officers how to keep marriage registers. The registers for Davis Bend, Vicksburg and Natchez alone document the registration of more than 4,600 freedpeople from Mississippi and northern Louisiana. Field offices at Brookhaven, Columbus, Davis Bend, Goodman, Grenada, Jackson and Pass Christian registered marriages as well. Louisiana and Tennessee. Several hundred certificates each in the headquarters files, plus subdistrict registers at Donaldsonville, Mansfield, Shreveport, Lebanon, Memphis and Trenton. Arkansas. Assistant Commissioner John Sprague ordered monthly reports in June 1865 and got them, with registers and certificates for ten named subdistricts including Arkadelphia, Fort Smith, Little Rock and Pine Bluff. Kentucky. Two documents only in the headquarters files, but subdistrict registers at Augusta, Bowling Green, Columbus, Cynthiana, Owensboro, Paducah and Winchester. Virginia. Colonel Orlando Brown ordered registration in March 1866, and his reports give names, ages, residence, parents and occupation, with subdistrict registers at Goochland, Lexington, Louisa Courthouse and Lovington.
Where they did not
Alabama. One marriage license. Not one series, not one register — a single unbound license in the Commissioner’s files, for a couple named Abslum Susk and Martha Culbert, dated June 20, 1865. The Archives states there is no evidence in Bureau records that officers in Alabama registered or issued marriage licenses and certificates in the state at all. South Carolina. Assistant Commissioner Rufus Saxton issued formal marriage rules in August 1865 declaring that “the sacred institution of Marriage lies at the very foundation of all civil society,” and the surviving output is one certificate. Georgia and Texas. No evidence in Bureau files that officers registered marriages or issued licenses at all. Maryland. No evidence that the Assistant Commissioner even acted on Howard’s order.
Sit with that for a moment, because it is the single most important practical fact on this page. An Alabama family and a Mississippi family, in the same year, under the same federal agency, under the same circular, have completely different odds of appearing in the marriage records — and the difference is not about them. It is about which officer took the instruction seriously. Absence of a record here is administrative, not demographic. It says nothing about whether a marriage happened, whether a family existed, or whether they were where you think they were.
The practical consequence is a rule: never conclude anything from a negative result in a single state’s Bureau files. Check the neighboring state, check the earlier military records, and check the bank. The Archives makes the same point about the Freedman’s Bank from the other direction — if there was no branch in the city where your family lived, look at the branches in neighboring states and cities, because people traveled to deposit.
What these records cannot do for you
Four honest limits, each of which will otherwise cost you weeks.
They are not fully digitized, whatever you have read. The claim that “the entirety of the Freedmen’s Bureau records are online” circulates widely and needs care. The National Archives publishes its own measure through its Record Group Explorer, which for Record Group 105 reported, when we checked it in July 2026, 167,477 textual scans online against an estimated 3,059,650 total textual pages — 5.47 percent. The Archives is transparent about the method: the numerator is a file count from its own Catalog, and the denominator is an estimate produced by multiplying the volume of records in cubic feet by an average of 2,500 pages per cubic foot. So the figure measures scans in the National Archives Catalog specifically, not every copy that exists anywhere; a great deal of the microfilmed Bureau material is viewable through partner sites the Archives itself points to. Both things are true at once, and the honest summary is that the microfilmed portion is broadly reachable while the record group as a whole is nowhere near completely scanned into federal custody online.
There is no master name index. This is the limitation nobody advertises. The Archives noted, when the preservation program began, that a persistent problem with the Bureau’s records has been “the lack of personal name indexes among the bureau’s files that would allow easy access to the records when searching for freedmen and their families,” and that without them a researcher can spend countless hours in records that may contain nothing about their ancestor. The same is true of the bank: the signature registers are not indexed. A separate microfilm publication indexes the deposit ledgers, and the Archives warns that it is unreliable — more than one index exists for some offices, some are not in strict alphabetical order, some omit depositors who do appear in the signature books, many account numbers are missing, and some numbers in the index disagree with the signature cards. Crowdsourced indexes built by outside institutions have improved this enormously, but an index built from handwriting is a finding aid, not a census, and a null result in one is not a null result in the records.
They will not, on their own, get you before 1865. A ration roll naming a former owner is a pointer, not a proof. It tells you which antebellum household to examine, and then the work moves into county deed books, probate inventories, estate divisions and tax lists — records kept by the enslaving family, in which your ancestor appears as property. That transition is the hardest part of this research and it is a different skill from anything on this page.
They are handwritten, damaged, and in nineteenth-century hands. When the preservation program started, the Archives described records that were torn and in places crumbling, and its conservation laboratory built custom housings for more than four thousand bound volumes. The scale it was working at is worth knowing before you order anything: in the same Prologue account, the Archives puts the field office records of Record Group 105 held by its Old Military and Civil Branch at more than one thousand linear feet, and describes the Freedmen’s Bureau Records Preservation Act of 2000 as authorizing three million dollars to preserve them. The shelf measurement is the Archives’ own; the statute itself directs the microfilming and the university pilot project and names no quantity of records at all. Spelling of names is phonetic and inconsistent. Given names appear in forms a family never used. Plan for variants before you conclude an absence.
Where a Freedmen’s Bureau search goes wrong
Six failures we see repeatedly, and what each one is actually telling you.
Searching the wrong record group
Half the queries we see for “Freedmen’s Bureau bank records” are actually about Record Group 101, which is held by the Comptroller of the Currency and described in a different finding aid. If you are asking about a signature book, you are asking about the bank, not the Bureau.
Reading a state’s silence as an answer
Nothing for Alabama, nothing for Georgia, nothing for Texas in the marriage series. That is a statement about Bureau officers, not about families. Move to the adjoining state, to the labor contracts, and to the bank before you conclude anything.
Trusting a single index hit as identity
Given names repeat heavily within a county and a plantation. One matching name with a plausible age is a candidate, not an ancestor. Two independent records agreeing on a relationship is the minimum before you build anything on top of it.
Looking only where they ended up
The Bureau provided transportation to freedpeople trying to reunite with family or relocate. A family enumerated in Ohio in 1880 may have a Bureau file in Virginia, a bank account in the District of Columbia, and a marriage registered in neither. Follow the movement, not the destination.
Stopping at the microfilm
The bank’s liquidation correspondence, the passbooks mailed in for dividends and the heirship affidavits sworn by descendants were never filmed. They sit unfilmed at College Park with mixed arrangement schemes, and for some families they are the only surviving proof of the relationship.
Treating the historical find as the finish
An 1869 register entry naming four siblings is a discovery, not a contact. Every one of those lines has a hundred and fifty years of descendants, and reaching a living one is a separate piece of work with its own consent questions attached.
How a Reconstruction-era file gets worked
Four stages. The order is not optional, because each one supplies the input the next one needs.
Fix the place, not just the name
Bureau records are arranged by state and then by office, so a search without a locality is not a search. We establish where the family was between 1865 and 1872 from later records first, and we treat that county as the query.
Work the bank before the Bureau
If there was a branch within reach, the signature register is the highest-yield single page in either record group. It also hands over a regiment, a birthplace and an owner’s name, all of which redirect the rest of the search.
Read the office, then the entry
Before drawing a conclusion from what an office holds, we read its own descriptive pamphlet: who commanded it, when it opened and closed, what it was ordered to record. An absence in a office that shut in 1866 means something different from an absence in one that ran to 1870.
Corroborate before anyone is contacted
Two independent documents agreeing on a relationship before it goes in a report, and a separate verification that any living descendant we name is the right individual at a current address. Historical certainty and present-day identity are two different burdens of proof.
What this firm is, and what it declines
Research that reaches back into slavery and forward into living households needs its limits stated up front, not buried.
We are a skip-tracing and public-records research firm. Nobody on this team holds a private investigator’s license and we are not licensed private investigators. We read records, we file written requests with archives and agencies under our own name, and we say who we are and who instructed us. We do not surveil, follow or photograph anyone, and we do not describe ourselves as investigators to any archivist, clerk or family member. Every file opens with a stated, permissible purpose written down before research begins — an estate that needs its heirs identified, a family reunion, a descendant community documenting its own history, a court filing that requires proof of a diligent search. Wanting to know where a stranger lives is not on that list, and we will say so rather than take the money.
We do not pretext. No calling an archive or a church pretending to be a relative, no letting a wrong assumption stand because it is convenient, no impersonating a claimant or an official to get a file opened. On this subject the temptation is real, because descendant communities are cautious about strangers asking after their families and they are right to be. The honest approach is also the effective one: an archivist who knows exactly who is asking and why can usually help, and one who has been misled tells the next person in the building. We also do not go after private financial account contents — balances, statements, account numbers — which are behind a financial institution’s door for good reason and are usually obtained by the pretexting we just refused. Note that the historical Freedman’s Bank records are a different thing entirely: they are archived federal records of an institution that closed in 1874, open to any researcher.
Not a consumer reporting agency
People Locator Skip Tracing is not a consumer reporting agency, and nothing we produce is a consumer report. Do not use anything traced through this page to decide on employment or a promotion, on housing or a tenancy, on credit, insurance, a professional license or a benefit. Those are decisions the Fair Credit Reporting Act governs, and they belong to a screening firm operating with the consent, disclosure, adverse-action and dispute process that statute requires. We do not run that process. Asked for a locate that turns out to be a covered decision in disguise, we decline the assignment.
Where we stop, and why this subject in particular
We decline any file where the person at the end of it appears to have stepped away on purpose. Genealogy is one of the easiest covers there is for reaching someone who does not want to be reached — a family-history framing sounds harmless, arrives with a plausible reason for wanting an address, and is very hard for a third party to check. So a request that reaches a living relative gets more scrutiny at intake, not less. If the circumstances suggest a protective order, a stalking pattern, a custody dispute being run outside a court, or a relative who has already asked plainly for no contact, we stop and we do not restart under a rewritten instruction. Nobody here is owed an ancestor’s descendant’s address. People live where they live, and some of them have very good reasons for the distance.
Two variations on this are specific enough to name. One is the descendant-community approach: a stranger who has found a shared surname in an 1867 ration roll and wants an introduction to a family that has never heard of them. We will carry a letter; we will not hand over an address so a doorstep visit can be a surprise. The other is the inheritance angle — heirs-property and unclaimed-estate matters cluster hard around exactly these records, and they attract people who are not what they say they are. If somebody has contacted you about an estate, ask them in writing for the decedent’s name, the case number and the court, and check the terms of anything before you sign it; a legitimate researcher will answer all three without hesitating. Where an estate genuinely needs its beneficiaries proved, that is the discipline set out in the guide to confirming a claimed beneficiary is the right person.
Everything on this page is general information about federal records and is not legal advice. Who inherits, what a court will accept as proof of descent, and whether a particular document establishes a legal relationship are questions for an attorney in the relevant jurisdiction. We find people and show our sources; the legal conclusions are somebody else’s job.
Who asks us to work a Reconstruction-era file
Six recurring instructing parties. They want very different things out of the same volumes.
Family historians
Stuck at 1870 and needing to know which of the two record groups to work first.
Heirs-property counsel
Clearing title on land where the chain of descent runs through unprobated deaths.
Genetic-genealogy clients
Holding a documented match and needing the paper trail that explains it.
Descendant associations
Documenting a whole community’s families from one plantation, church or settlement.
Museums and archives
Attaching living descendants to an object, a photograph or a name in a collection.
Reunion organizers
Rebuilding a scattered line before a gathering, and needing current people not old names.
The genetic-genealogy clients are the group whose expectations most often need adjusting, in both directions. A test result can prove a relationship that no surviving document names — which is exactly what these families were denied — but it cannot tell you which of four brothers in an 1867 labor contract is the one, and only the paper can. It also cannot make a match reply, which is a separate problem with its own etiquette; the note on what to do when a genetic match stays silent deals with that, and the short answer is that silence is an answer. Working sideways through a documented sibling line is frequently the faster route, and the piece on tracing the descendants of an ancestor’s brother or sister sets out how.
Every name arrives with the volume it came out of
Nothing goes in a report without the record behind it: the publication number, the roll, the office, the year, the account number. You should be able to order the same page and see the same thing, and if you cannot, we have not finished. Where the answer is that the record does not exist, we write that down as a finding with its own sourcing rather than dressing it up as a search still in progress — and where we cannot locate the person, you are not billed for a locate that did not happen. A family’s history has been handled carelessly enough already.
Questions about Bureau and Freedman’s Bank records
Why does African American research hit a wall at 1870?
Because of how the census forms were written. In 1860 there were two schedules: Schedule 1 recorded the name of every free person, while Schedule 2, for enslaved inhabitants, had columns for the name of the slave owner, the number of slaves, age, sex, color and a few other counts. There was no column for the name of the enslaved person. So for a family enslaved until 1865, the federal population schedules simply contain no name before 1870. It is not a gap in survival; the names were never collected.
Is the Freedman’s Bank part of the Freedmen’s Bureau?
No, and the confusion is common enough that the National Archives addresses it directly. The Bureau’s files are Record Group 105, held as records of the War Department agency. The bank’s records are Record Group 101, records of the Office of the Comptroller of the Currency, because the Comptroller was eventually put in charge of winding the bank up. Congress created both on March 3, 1865, two chapters apart in the same volume of the Statutes at Large, and their staff overlapped heavily, but they are two separate bodies of records with separate finding aids.
What is actually written on a Freedman’s Bank signature record?
A full entry gives the depositor’s name, account number, age, complexion, date of application, place of birth, place raised, occupation, spouse, children, the names of parents and of brothers and sisters, a remarks field, and the signature. Some earlier volumes also record the names of former owners or mistresses and the plantation where the depositor had lived, and some entries have a death certificate attached. Soldiers’ forms had space for regiment and company, which is the link across into military service and pension files.
Are the Freedmen’s Bureau records fully digitized?
Not in the sense that phrase implies. The National Archives publishes a figure for Record Group 105 through its Record Group Explorer, and in July 2026 it showed 167,477 textual scans online against an estimated 3,059,650 total pages, or 5.47 percent. That measures scans in the Archives’ own Catalog, and the total is an estimate derived from shelf volume rather than a page count. A large share of the microfilmed material is separately viewable through partner sites the Archives links to, so the practical availability is much better than 5.47 percent suggests, while the record group as a whole is nowhere near completely scanned.
There is nothing for my ancestor’s state. Does that mean they are not there?
Almost certainly not. The Archives states that the Bureau was inconsistent in the kinds of records it created for each state, and the marriage series shows how extreme that is: Mississippi’s registers at Davis Bend, Vicksburg and Natchez alone document more than 4,600 registrations, while Alabama produced a single marriage license and Georgia and Texas produced no evidence of registration at all. The difference is about which officer followed the order, not about which families existed. Check the adjoining state, the labor contracts, the ration rolls and the bank before drawing any conclusion.
Can these records identify the person who enslaved my ancestor?
Sometimes, and it is one of the most valuable things they do. Ration registers and applications routinely record the name of a former owner alongside the recipient’s name and residence, and some of the earlier Freedman’s Bank volumes record former owners or mistresses and the plantation. That name is a pointer rather than a proof: it tells you which antebellum household’s deeds, probate inventories and estate divisions to examine next, which is where research before 1865 actually happens.
Do I have to go to Washington to use them?
Not for the microfilmed material, which is what most researchers need and which the Archives directs people to online. You may need to for the rest. The Freedman’s Bank liquidation files in particular were never microfilmed, and the letters, mailed-in passbooks, heirship affidavits and questionnaires that make them up are available at the National Archives facility at College Park, Maryland. They have mixed arrangement schemes and no single index, and for some families they hold the only surviving statement of a relationship.
Can you use this research to find a living relative for me?
We can, with conditions. The historical work identifies lines of descent; locating a living person on one of those lines is separate work with its own rules. We require a stated lawful purpose, we identify ourselves and whoever instructed us in any approach, and we make it easy for the person to decline. Where the circumstances suggest someone has deliberately stepped away — a protective order, a stalking pattern, an explicit request for no contact — we decline the file rather than deliver the address. We come back within 24 hours with an honest read on what your particular records are likely to hold before anyone commits to anything.
The names are in those volumes. Let us read them with you
Send the surname, the county and roughly where the family was between 1865 and 1872, and we will tell you which record group is worth working first and what the odds honestly look like for that state — before you commit to anything. If the answer is that the records for your county barely exist, we would rather say that on day one. Tell us what you are trying to reconstruct and we will tell you where the paper is.
Open a family research file
