# Worldwide displacement and settlement audit — methodology

## Purpose

This audit asks, for every present-day country in the supplied list and for relevant non-sovereign territories:

1. Which historically identifiable populations based in, or substantially ancestral to populations now based in, that geography displaced, subordinated, absorbed, expelled, deported, or demographically replaced another population?
2. Which populations living in that geography were subjected to those processes?
3. Which processes are represented in the current *Taking their land* atlas, and whether that coverage is exact, partial, bundled, or absent?
4. Which absent or incomplete processes should be reviewed for inclusion?
5. Which related processes are important but fall outside the atlas’s settler-colonial or demographic-expansion threshold?

Modern states are geographic containers for the audit. They are not treated as timeless peoples or as collectively responsible actors. “Agent” means a historically named state, institution, army, settler stream, or population movement associated with the territory—not every present-day resident. “Victim” likewise names the affected population rather than implying that all residents experienced the event identically.

## Event unit

A **unique historical event** is one linked process with substantially continuous actors, affected land or population, geography, and chronology. A new regime, settler population, mechanism, or long interruption triggers a split review, but not an automatic split. A mechanism change may remain a dated phase when it implements the same causal land transition—for example, removal followed by successor settlement—or when one law or commission operates through several places. A proposed child becomes a separate event only when it has its own bounded coercive act or land transfer, a distinct affected population or tract, enough local evidence for an independent analytical profile, and an explicit supersession rule that prevents parent/child double-counting. Map nodes and phases are not themselves event counts. A process spanning several modern states remains one master event but creates several **country-event incidences**.

Examples:

- The Holocaust is one master event with victim incidences across many present-day countries and agent incidences for Nazi Germany and its collaborating regimes.
- The post-1945 expulsion of Germans is separated from Nazi occupation and Generalplan Ost because its actors, victims, chronology, and mechanism differ.
- The Spanish conquest and settlement of the Americas is divided into regional processes where different Indigenous populations, settler streams, institutions, and timelines make a single entry analytically useless.

Short duration does not disqualify an event. Scale, durable land or population consequences, and evidentiary clarity matter more than duration.

### Historical tempo and event-unit fairness

Calendar duration is not an event definition. A transition visible only at
century-scale archaeological resolution, an ancient campaign dated by season,
and a modern removal dated by day cannot be made comparable by forcing them
into equal time windows. Chronology triggers a split review only when it may
mark a change in actor, program, affected population, mechanism, geography, or
causal sequence. Events close in time may still be separate; a longer process
may remain one event when those elements are demonstrably continuous.

Archive density must not decide the count. A rich record does not turn every
named place or dated action into an event, and sparse evidence does not permit
distinct processes to be merged or silently omitted. Unresolved units are held
and unresolved traits are coded unknown. Parent and child rows may never carry
the same statistical weight: the parent must be inactive or its residual scope
must be explicitly disjoint. The rule is unchanged by actor or victim identity,
moral salience, familiarity, or whether a boundary makes the project's
statistical results look cleaner.

Beginning with the next promoted release, the project requires a release-wide
event-boundary audit, independent calibration across eras and source
resolutions, and grouped-parent/alternative-resolution sensitivity tests before
claiming that the one-event/one-row rule has been applied fairly. Statistical
effects cannot be used to admit, reject, split, or merge an event. The full
prospective gate is in
`analysis/second_pass/EVENT_UNIT_FAIRNESS_AND_RESOLUTION_PROTOCOL.md`.

## Temporal scope

The core audit runs from approximately 6500 BCE, matching the current atlas, to the present. Earlier hominin population turnovers and initial peopling of previously uninhabited land belong in the related-circumstances appendix unless they are necessary to interpret a later process. Prehistoric entries require convergent archaeological, linguistic, or ancient-DNA evidence; language spread alone is not treated as proof of population displacement.

## Inclusion families

Every candidate receives one mechanism family:

- **Settler colonization:** permanent incoming settlement tied to land transfer, sovereignty, Indigenous elimination, or durable demographic domination.
- **State resettlement/internal colonization:** a state moves favored populations into a frontier or minority homeland and transfers land, administration, or political weight.
- **Demographic expansion analogue:** a non-state or pre-state population movement produces substantial replacement, displacement, absorption, or stratification.
- **Expulsion/deportation/ethnic cleansing:** a population is removed without a qualifying replacement project, or replacement is secondary.
- **Population exchange/partition:** reciprocal or administratively paired removals, usually under interstate agreement or border formation.
- **Conquest with settlement:** military incorporation accompanied by durable incoming settlement, but without enough evidence for a stronger settler-colonial classification.
- **Conquest without substantial settlement:** domination or elite replacement; normally contextual rather than an atlas case.
- **Enslavement/labor removal:** demographic removal or coerced migration whose primary mechanism is labor extraction; normally contextual unless it clears land for settlers.
- **Development/conservation displacement:** dams, mines, plantations, military reservations, parks, or infrastructure remove residents; contextual unless coupled to replacement settlement.
- **Genocide-linked clearance:** mass killing and removal materially clear land or demographic space; classification depends on whether settlers or replacement populations follow.
- **Return/decolonizing displacement:** removal or flight of a formerly dominant settler population; retained for symmetry and historical understanding, not automatically equated with the original colonization.
- **Conflict displacement:** flight, expulsion, or internal displacement caused by war or state collapse where a replacement-settlement project is not established. These events answer the victimization audit and belong in context, not in the settler-colonial count.

## Atlas recommendation

- **Core:** clear settler project, durable incoming population, land/sovereignty transfer, and elimination or displacement of Indigenous residents.
- **Strong:** most core elements are present, but scale, metropole, continuity, or demographic dominance is less complete.
- **Partial:** real settlement and displacement, but commercial, military, elite, or mixed motives prevent a close fit.
- **Expansion analogue:** demographic replacement or displacement without the modern state/metropole structure.
- **Attempted:** planned or initiated replacement project that failed or was interrupted.
- **Excluded:** examined and found not to involve meaningful settlement or replacement.
- **Contextual:** important expulsion, conquest, forced migration, partition, enslavement, or development displacement that should inform the atlas but not necessarily become a stop.

## Comparison with the atlas

- **Exact:** the current case represents the same actors, victims, geography, chronology, and mechanism as one defensible event unit.
- **Incomplete:** the case identifies a real process but omits a major phase, population, geography, or mechanism, or uses a chronology too broad to answer the audit questions cleanly.
- **Wrongly bundled:** one atlas case combines several event units that must be split before they can be counted or mapped responsibly.
- **Wrongly classified:** the event exists, but the atlas assigns the wrong mechanism or strength—for example treating conquest, labor migration, or settlement on previously uninhabited land as demographic replacement.
- **Unsupported:** the asserted event, displacement, or replacement effect is not adequately supported by the cited evidence.
- **Adjacent:** a related atlas case exists but is not an adequate representation of this event.
- **Absent:** no current case captures it.
- **Context only:** intentionally outside the atlas’s main model but retained in the appendix.

“Exact” is not a synonym for “the title is broadly right.” It is reserved for cases that survive an event-unit test. A familiar multi-century regional story fails that test when governments, settler streams, affected populations, or mechanisms change materially.

## Evidence and confidence

- **High:** multiple independent scholarly syntheses, primary/official records, or convergent archaeological/genetic evidence agree on the population movement and material consequences.
- **Medium:** the event is well attested but the scale, causal mechanism, continuity, or displacement effect is materially disputed.
- **Low:** evidence is sparse, retrospective, politically instrumentalized, or unable to distinguish population movement from language/cultural diffusion. Low-confidence events remain visible but are not recommended as atlas additions without further research.

Each master event should have at least one directly relevant source. Recommended atlas additions should normally have two, preferably including a specialist scholarly source and an official, primary, archaeological, or demographic source where available.

## Counting rules

Two totals are maintained:

1. **Unique historical events** in the deduplicated master register.
2. **Country-event incidences**, because a single event can affect several present-day countries.

Country summaries separately count agent, victim, both, and container-only incidences. An event can count in more than one role for the same country when different populations or phases warrant it. Counts never imply collective guilt or homogeneous ancestry.

## Completeness rules

- Every supplied country must receive a row, including microstates and states for which no qualifying event is identified.
- Relevant territories and stateless or cross-border homelands are added rather than silently assigned to one sovereign state.
- Overseas colonial action creates an agent incidence for the metropole and victim/container incidences for affected territories.
- Indigenous and minority names are used where supported; obsolete or imposed names are retained only when necessary to identify a historical source and are explained.
- Absence of evidence is recorded as “no qualifying event identified,” not “no event occurred.”
- The audit records competing interpretations and avoids turning present-day borders into ancient ethnic boundaries.

## Quality-control checks

Before completion, the files must pass checks for:

- full country and territory coverage;
- unique stable event IDs;
- valid dates and chronological ordering;
- consistent actor/victim names;
- cross-border deduplication;
- atlas case IDs that actually exist;
- at least one source per event and two per recommended addition where feasible;
- consistency between master-event, country-incidence, and summary counts;
- explicit confidence and classification for every event;
- separation of historical findings from present-day political or legal claims.