Genealogy

Cuneiform legal texts are dense with filiation: PN son of PN, over and over. That information is what turns a list of individuals into a society — but it lives at three different levels of interpretation, and DAPCA keeps them apart so that automatic processing never overwrites your judgement.

The distinction this page rests on is the one drawn in the introduction: a historical person is not a marked filiation. What a tablet states is evidence; what a family tree shows is an interpretation built on top of many such statements, and the step between the two is neither automatic nor guaranteed to succeed.

1. The three levels

The distinction between the first two is the important one. The evidence is tablet-local: it says that this token is the son of that token, and it remains true regardless of who those tokens turn out to be. The primitive level is historical: it says that this individual is the son of that individual, which is only meaningful once both tokens have been assigned to persons.

The third level is a pure consequence of the second. If two people share a father, they are siblings — there is nothing to record, and recording it by hand would just create something that can fall out of sync. DAPCA computes those relations and shows them as read-only.

2. Recording the evidence (tablet level)

Filiations are marked on the tablet itself, in the Mark genealogies mode. Two ways:

• Auto-detect Family Relations finds the PN DUMU PN pattern and proposes the links for the whole tablet at once. It is a pattern match, so review the result: it is right far more often than not, but not always.
• Add a Family Relation — select two personal-name tokens in the text and connect them by hand, with a secure flag and notes.

Existing links are listed underneath in the Family Relations table, where they can also be removed.

These links belong to the text. They are worth recording even for names you have not yet promoted to persons — in fact that is the normal order of work, and the next step depends on having them.

3. From evidence to individuals — Seed from texts

The Seed from texts button on the Genealogy tab of a person page walks every recorded filiation, looks up which individual each of the two occurrences belongs to, and creates the corresponding kinship edge between those individuals.

It reports what it did: how many edges were created, updated, left unchanged, and skipped.

If an occurrence is assigned to more than one candidate individual, the seeder still creates the edge but marks it not secure — an identification is ambiguous somewhere along the line.

Where the bottleneck actually is

The corpus currently holds some four thousand filiations recorded at the level of the text, and a handful of kinship edges between individuals. The disproportion is not a defect of the bridge: it measures how many occurrences have been promoted to persons.

A worked case makes the point. In one Emar text the filiation pil₂-su-DINGIR-da-gan DUMU DINGIR-IM-GAL is recorded as evidence, and the son is identified as a known individual; the father's occurrence, written logographically DINGIR-IM-GAL, is the perfectly familiar Baʿlu-kabar — who also exists as an individual in the register. The edge is skipped only because that occurrence was never assigned to him. Assign it in the tablet's Prosopography view, seed again, and the father edge appears, attested by one more text.

The work that unlocks a family tree is therefore not marking more filiations: it is identifying the occurrences of the fathers.

4. Adding a relation by hand

On the Genealogy tab, the relation form records edges the texts do not state plainly.

Direction matters. The relation is read as "this person is <type> of the other person". Adding son of on Aḫiu's page, pointing at Abdi-ili, means Aḫiu is the son of Abdi-ili. To record a child, add the relation from the child's page.

Available types: son of, daughter of, husband of, wife of, brother of, sister of.

Basis distinguishes what the texts state from what you conclude:

• attested — the relation is expressed in the documents.
• inferred — a historical deduction. A rationale is required: the form will refuse an inferred relation without one. That text is the argument; without it, a later reader cannot tell your reasoning from an error.

You can additionally record secure/uncertain, a bibliographic reference with page numbers, and free notes.

Relations can be removed from either endpoint's page.

5. Derived kinship

Below the asserted relations, the page lists what follows from them:

These are read-only. If one is wrong, the error is in a primitive relation underneath it — fix that, and the derived one corrects itself. If any supporting link is marked uncertain, the derived relation inherits the uncertainty.

6. Reading the family tree

The tree page — Family tree on the Genealogy tab of a person page, or the tree icon in the person list — draws the individual in their kin network, oldest generation at the top.

Cuneiform tablets are rarely dated absolutely, so the tree doubles as a relative chronology: beside each generation it lists the tablets involving that generation's members, giving you a sense of which documents are contemporary with which.

Text chips are coloured by how the person appears in the document:

Only active attestations date a generation, which is why the patronymic and papponymic roles are worth setting carefully when you attach an occurrence: a man named only as someone's father is not thereby attested as alive at that moment.

The three categories are toggles — switch them off to see only what you care about.

Bridge lanes (dashed, between two generation rows) mark tablets where members of two different generations are both active — a father and a son both witnessing the same contract, for instance. These are genuine synchronisms, the strongest chronological anchors the corpus offers, so they are pulled out of the ordinary strips and shown in the gap between the generations they connect. A tablet that merely mentions the father in a filiation formula is not a bridge: expected filiation is not a synchronism.

Each node has an info button opening a dossier — attestations grouped by role, seals owned, derived kinship. The node's name opens the person page; the derived kinship links move the tree's focus to that relative.

7. Who may do what

Note the asymmetry, which follows the layers: the evidence on the tablet is written by editorial accounts, the interpretation at person level by holders of prosopographic editing rights. To make the bridge productive you need both — the filiations must have been marked, and the occurrences assigned.