Ancestor Work

Family Tree Research When the Records Stop

The six walls American family research actually hits, which sources go around each one, what DNA will and will not do, and what to do when the record is gone for good.

Stack of worn leather folios with a brass magnifying glass and reading spectacles beside a candle
Quick answer

American family research stops at six recurring walls, and naming yours correctly is most of the work. Before 1850 the federal census lists only the head of household, so the way past it is land, probate, tax and church records rather than more census. The 1890 census was almost entirely destroyed by fire in 1921, leaving state censuses and city directories to fill the gap. For Black American families 1870 is the first census naming formerly enslaved people, and research below it runs on Freedmen's Bureau and Freedman's Bank records and on the probate and estate papers of the enslaving family. Burned counties usually have duplicates in state land offices, federal courts, newspapers and church books. Names were not changed at Ellis Island; search phonetically and follow the neighbors. DNA testing is strong at confirming a specific hypothesis within about six generations and weak at deep ancestry and ethnicity percentages. This is genealogical guidance, not legal advice.

Evidence
Verified

Federal censuses before 1850 name only the head of household; other household members appear as tally marks by age and sex.

Verified

The 1890 federal census was almost entirely destroyed following a 1921 fire in the Commerce Department building, with under one percent of schedules surviving.

Verified

1870 is the first federal census listing formerly enslaved people by name; earlier slave schedules record them by age and sex under the enslaver's name only.

Verified

Freedmen's Bureau and Freedman's Bank records, and the probate and estate papers of enslaving families, are the established route to pre-1870 Black genealogy.

Verified

Immigrant surnames were not altered by officials at Ellis Island, who worked from manifests written at the departure port; changes happened later and usually by the family's own choice.

Verified

Autosomal DNA is reliable for relationships within roughly six generations; ethnicity estimates are comparisons against reference panels and are revised over time.

Lived practice

Lighting a candle to open and close a research session, and documenting a negative result as a finding, are common working habits among researchers.

Traditional

Setting a place for ancestors whose names are unrecoverable is a widely kept form in ancestor practice.

Not proven

A spiritual practice can recover factual information about ancestors that the records do not hold.

VERIFIED = verifiable in scientific or official sources · TRADITIONAL = historically or culturally recorded · LIVED PRACTICE = widely practiced, experiential knowledge · NOT PROVEN = spiritual interpretation, not scientifically established.

Everyone who traces a family tree hits the same wall. It is usually four or five generations back, and it usually arrives without warning: the census stops listing them, the church has no record, and the name changes spelling three times and then vanishes.

This is the practical guide to that wall. Where American records actually stop, which stops are technical and which are the result of deliberate erasure, what to do at each one, and what to do when the answer is that the record is gone for good.

It follows on from ancestor work and how to begin it, which covers the practice side. This one is entirely about paper.

The walls, and which one you are at

Naming your wall correctly is most of the work, because each has a different way through.

Wall What it looks like First move
The 1850 census line Before 1850, the federal census names only the head of household. Everyone else is a tally mark by age and sex. Stop using the census as your spine. Switch to land, probate and tax records.
The 1890 gap Almost the entire 1890 federal census burned in 1921. Twenty years of nothing between 1880 and 1900. State censuses, city directories, and the 1890 veterans schedule, which partly survived.
The 1870 wall For Black American families, 1870 is the first federal census listing formerly enslaved people by name. A different research method entirely — see below. This is not the same kind of wall.
The ocean The trail is solid to the port of arrival and then stops. Find the exact town of origin. Nothing on the other side works without it.
The burned county Courthouse fire. Deeds, wills and marriages for a whole county gone. Records duplicated elsewhere: state land offices, federal courts, newspapers, church books.
The closed adoption The paper trail ends at an amended birth certificate. State law, which has changed a great deal recently, and DNA.

The 1870 wall

This one is different in kind from the others and deserves saying plainly: it is not a records gap. It is the documentary shape of slavery, in which people were recorded as property rather than as persons.

The work below 1870 is real and is done routinely, but it runs on different sources.

  • Freedmen's Bureau records (1865–1872) — labor contracts, marriage registers, ration rolls, letters. Digitized and searchable, and often the single richest source for the immediate post-emancipation years.
  • Freedman's Bank records (1865–1874) — the registers frequently list parents, siblings, spouse, former enslaver and plantation. Unusually detailed for the period.
  • The 1850 and 1860 slave schedules — enslaved people are listed by age and sex under the enslaver's name, without their own names. Genealogically thin on their own, but they establish who held whom and where.
  • Probate and estate records of the enslaving family. This is the core method: people were inventoried, willed and divided as property, and those documents name them. Researchers call this reverse genealogy — you trace the white family's paper in order to find the Black family inside it.
  • Bills of sale, plantation ledgers, county deed books. Often in state archives or university special collections.

This is slow, and it is emotionally heavy in a way that ordinary genealogy is not. It is also frequently successful. If it is your line of research, look for the community around it before starting alone — this is well-organized territory with experienced people in it.

Plain hand-poured soy candle in a clear glass vessel burning on a dark wooden surface

From the temple

Timeless Candle — White Moss & Cotton

$25.00 · hand-poured organic soy, ~50 hours

Something a lot of people who do this work end up doing without being told to: light one candle at the start of a session, work until it feels finished, put it out. It gives the research a beginning and an end, which matters more than it sounds when the material is heavy and there is no natural stopping point. Any candle does this. If you want one that lasts a whole winter of Sundays, this is the one we use.

See the candle →

What to do when the census runs out

The single most common mistake is treating the census as the backbone of a tree. It is a snapshot every ten years, taken by a stranger, often from a neighbor, with spelling at the enumerator's discretion. Before 1850 it barely names anyone. Once you accept that, better sources open up.

  • Land records. The most underused source in American genealogy. Deeds are continuous, they were kept carefully because money was involved, and they routinely name heirs, spouses and neighbors. A deed that says "and his wife Sarah, daughter of the late John Weeks" does in one line what ten census pages cannot.
  • Probate and wills. Estate files name children, sometimes grandchildren, and often the ones who had already moved away.
  • Tax lists. Annual rather than decennial. They catch a man appearing in a county, and the year he stops appearing.
  • Church records. Baptism, marriage and burial registers, frequently older than any civil registration in that place.
  • Newspapers. Obituaries, marriage notices, legal notices to heirs. Digitized newspaper archives have transformed this in the last decade.
  • Military and pension files. Civil War widow's pension applications are extraordinary — to prove a marriage, women submitted family Bible pages, affidavits from neighbors, and detailed statements about who was where and when.

Names, and why yours keeps changing

A persistent myth worth killing: names were not changed at Ellis Island. Officials there worked from manifests written at the port of departure and did not create new documents. Name changes happened later, gradually, and usually by the family's own choice — anglicizing, simplifying, or just accepting whatever a clerk had written.

Practical consequences when searching:

  • Search by sound, not spelling. Soundex and similar phonetic matching exist for exactly this. Most large databases support it.
  • Spelling was not standardized until surprisingly recently, and the person themselves may have been unable to write. One family can appear as Schneider, Snyder, Snider and Schnyder within a single generation.
  • Watch for translated given names — Wilhelm to William, Giovanni to John, Anna to Annie to Ann.
  • Search the neighbors. Families migrated in groups. If you have lost a surname, find the three families who lived next to them and follow those instead. The one you want is usually in the same wagon.

DNA, and what it will and will not do

Autosomal DNA testing has genuinely broken walls that paper could not, particularly for adoptees and for lines lost before 1870. It has also created a large amount of confidently wrong family tree information.

What it does well: confirms or breaks a hypothesized relationship, identifies close relatives you did not know about, and works backward through clusters of shared matches to a common ancestor. What it does poorly: anything beyond roughly six or seven generations, where the inherited segments become too small to be reliable. And ethnicity estimates are the least reliable part of the product — they are comparisons against reference panels that get revised, which is why your percentages change when the company updates them.

Two things to consider before testing. First, results are permanent and unpredictable — misattributed parentage is discovered often enough that you should decide in advance how you would handle it. Second, your result is information about your relatives too, and they did not consent. Read the privacy terms, including what happens to the sample if the company is sold.

Adoption and sealed records

This area has changed substantially and much of the advice online is out of date. A growing number of states now allow adult adoptees access to their original birth certificate, some unconditionally and some with a contact-preference form filed by a birth parent. The rules are state-specific and have been moving steadily in one direction — check the current law for the state of birth rather than relying on what was true a decade ago. Alongside that, state reunion registries, court petitions for cause, and DNA matching are the standard routes.

When the record is genuinely gone

Sometimes it is. The courthouse burned, the parish register was pulped, the village was destroyed, the ship's manifest was never kept. No amount of persistence produces a document that does not exist.

Two things are worth doing at that point.

Write down the wall itself. Record what you searched, where, and what was not there. A negative result documented properly is real research — it stops you repeating it in two years, and it is the thing a future relative will be most grateful for. "The 1863 fire destroyed the county deed books" is an answer, not a failure.

Then let the practice do the part the paper cannot. Every tradition that keeps ancestors keeps unnamed ones. You do not need a name to set a place, and "all those I come from whose names I do not have" is a complete and honest address. The forms are in the Samhain ancestor altar and how to host a dumb supper.

Genealogy answers who. It does not answer why it matters that you are still here to ask, and it was never going to.

What we don't promise

  • We do not claim every line can be traced. Many cannot, for reasons that are historical rather than a failure of effort.
  • We do not offer legal advice on adoption records, which is state-specific and changes; check the current law where the birth occurred.
  • We do not claim any spiritual practice recovers factual information about ancestors. Research produces facts. Practice produces relationship. They are different things and conflating them makes both worse.

Frequently asked questions

Why does my family tree stop around 1850?

Because before 1850 the federal census names only the head of household — everyone else appears as a tally by age and sex. The fix is to stop relying on the census and move to land, probate, tax and church records, which name people individually and go back further.

What happened to the 1890 census?

It was almost entirely destroyed by a fire in the Commerce Department building in 1921 and its aftermath. Fewer than one percent of the schedules survive. State censuses, city directories and the partly surviving 1890 veterans schedule are the usual substitutes.

How do I research Black ancestors before 1870?

Through Freedmen's Bureau and Freedman's Bank records, the 1850 and 1860 slave schedules, and above all the probate, estate and deed records of the enslaving family, in which enslaved people were inventoried by name. It is a different method from ordinary genealogy and is routinely successful, though slow.

Were names really changed at Ellis Island?

No. Officials worked from manifests created at the departure port and did not issue new documents. Name changes happened afterward, usually gradually and usually by the family's own decision.

Will a DNA test break through my brick wall?

Sometimes. It is strongest at confirming or disproving a specific hypothesis and at finding relatives within roughly six generations. It is weakest at deep ancestry and at ethnicity percentages, which are estimates against reference panels and get revised.

Can I get my original birth certificate if I was adopted?

It depends on the state of birth, and the law has been changing steadily toward access. A growing number of states now release the original certificate to adult adoptees, some with a contact-preference form attached. Check the current statute for that state rather than older guidance.

What do I do if the courthouse burned?

Look for duplicates. Land grants often exist at the state land office, federal court records survive separately, newspapers printed legal notices, churches kept their own registers, and neighboring counties frequently hold records for families who crossed the line. A burned county slows research; it rarely ends it.

Sources and further reading

  • National Archives (NARA) research guides on census, military, and Freedmen's Bureau records.
  • NARA account of the 1921 fire and the fate of the 1890 census schedules.
  • Freedmen's Bureau and Freedman's Savings and Trust Company records, digitized and indexed.
  • State-by-state adoptee access statutes, which are amended frequently.
  • Genetic genealogy standards literature on autosomal segment reliability across generations.

Continue here: how to start ancestor work · the Samhain ancestor altar · how to host a dumb supper

Back to the Desire Diaries