Family tree charts: what the standard shapes assume and leave out

Before a chart records anything about a family, it has already decided what a family is. Two parents per person. One line of descent that matters. A shape that fans out cleanly in one direction and terminates neatly in the other.

Those are conventions, not findings, and they were built for particular purposes. They still govern almost every family diagram published, which means they quietly govern the research that fills them in: people look for the material the form has slots for, and stop looking when the slots are full.

What to take away

What the standard shapes assume

The two common layouts answer different questions and carry different assumptions, and confusing them produces charts that look complete while omitting most of a family. The ancestor-facing form is the pedigree chart, which fills up as it goes back and says nothing about anyone's siblings.

An ancestor chart runs backwards from one person and doubles each generation. It assumes exactly two parents per person, every generation, with no room for anything else.

Adoption has no place unless silently substituted for birth parentage. An unknown parent is only an empty box, read as a research failure. There is no way to show someone raised by a person outside the diagram.

A descendant chart runs forward from one person, and it makes a claim before it shows anything: that this individual is where the family starts. That choice is almost never explained, and it determines which branches are visible. A family looks different depending on which ancestor you decided to begin with, and the difference is invisible on the page.

Positional identifiers in a chart

Many charts identify people by position rather than by name: a number that says where in the structure someone sits. It is compact, and it is a trap, because the number is meaningful only if the structure is right.

Change one link, and every number above it silently points to a different person. Two people with the same number in two chart versions are not the same person. Citing "the individual at this position" cites a slot, not a human.

Positional identifiers also cannot represent someone who appears twice in one chart, as when two branches converge.

The alternative is dull and reliable: give every person a stable identifier that means nothing structural, and let position be a property of the current drawing. Then a restructure moves people around without changing who anyone is, which is the property you actually need when a correction lands.

Here is how that plays out. A chart shows a woman with one husband and two children, numbered 7, 14 and 15 by position. A marriage record then turns up showing an earlier husband, and a child from that marriage.

Adding the second husband and the child shifts the first husband and the two children down a place. Anyone who cited person 14 from the earlier version now points at somebody else. With a stable identifier, the two children keep their numbers, the new people get new ones, and every earlier note still says what it said.

The form shapes the research

This is the effect worth watching in your own work. A chart with empty boxes reads as an instruction to fill them, so effort flows toward the questions the diagram poses and away from the ones it cannot express.

Two consequences follow. Relationships with no slot go uninvestigated: informal fostering, guardianship, households that raised a child without any legal act. And empty boxes attract plausible occupants, because a blank in a symmetrical structure looks like a gap rather than like an absence of evidence.

The counter is to keep research notes ahead of the diagram and let them record what the chart cannot hold.

A line saying a person was raised in a household by people not in the chart is a finding. Discarding it because there is nowhere to draw it lets a drawing convention edit the evidence.

Reading a chart somebody else made

An inherited chart is a source, and like any source it was made by a particular person for a particular reason.

Ask what it was for. Charts have been drawn to establish a claim to something, to demonstrate a connection to a notable person, to support an application, and to satisfy curiosity, and the first three create pressure in a known direction. A chart built to prove a link will show that link confidently and treat every alternative as settled.

Then ask what it cites. Most published charts cite nothing, and a chart without citations is a set of assertions arranged attractively.

Where it does cite, follow the citations, not the shape, since the shape is the part most likely copied. Agreement among several charts is not corroboration when they descend from one another, the same chain problem described in the guide to reading a family's written record.

Finally, note what the chart is not saying. Missing branches are usually choices. A person absent from a diagram was frequently omitted rather than unknown, and the omission may have been deliberate, contested, or simply a matter of what would fit on the page.

A chart can be checked against records that were not made to support it. In the United States that means federal census schedules, open through 1950 and released 72 years after each count. The 1960 returns open in 2032.

County courthouses hold probate, land and marriage files. State vital records offices hold birth and death certificates, and their start dates vary. Many states began civil registration in the late 1800s, and town clerks kept earlier records in some places.

The National Archives catalog holds federal military, pension and immigration files. The Social Security Applications and Claims Index covers claims filed from roughly 1936 to 2007. Church registers and cemetery indexes locate a person but prove little on their own.

Check a chart in this order: match each person to an independent record, mark which links rest on a document and which on an inference, then rebuild only the sourced links.

For software, FamilySearch and Ancestry hold indexes and images. Gramps is a free desktop program, and GEDCOM is the format most programs use to exchange a tree. Any program works if it keeps a person identifier separate from a chart position.

Where the older conventions meet the living

The reference account of the family tree shows how much of the convention was settled long before the households it is now asked to describe existed.

A chart drawn on the standard shape will, on reaching the present day, present modern households in a vocabulary that cannot describe them. It flattens step, adoptive, and social parentage into one kind of line. The distinctions being lost are set out in the guide to three senses of parentage.

Name forms create the parallel problem going backwards. Spellings, transliterations, and naming customs change across generations and borders. A name match is a reason to look, not a finding. The guide to name data explains how names are recorded and counted.

The same lag applies to naming trends. Every naming story rests on a dataset with a definition and a coverage boundary.

The rule for the living end of any chart is the one that governs the whole site. The subject is the person with the public role, and everyone else appears only to the extent they have made the relationship public themselves.

Common questions

Is a chart with unlabeled links ever acceptable?

Only where every link genuinely rests on the same kind of evidence, which is almost never true. Otherwise the uniform line is a claim of uniform confidence, and it is the weakest link in the diagram that determines what the whole is worth.

How should a chart show an adoption?

Adoptive parentage is distinct from birth parentage where the people involved have made it public. Substituting it silently is a false statement about the record. Leaving it out is a false statement about the family.

Where it is not public, the parentage belongs in the chart because it is parentage. The adoption is not yours to disclose.

What about a person who appears in two places in the same tree?

Give them one identifier and show them twice, marked as the same person. Duplicating them as two entries produces two research trails that will diverge, and merging them later is exactly the operation that is hardest to undo.

Should a published chart show generations of living people at all?

Only the relationships those people have acknowledged, with nothing else attached. A structurally complete chart of a living family is a compiled identity record regardless of the intent behind it.