Method

We publish only what a primary document prints: a census table, a peer-reviewed paper or a curated database. Every number links to its source. No estimates, recalculations or "approximate" aggregator figures.

Every figure is checked twice. A researcher first reads it in the official document and stores a quote. A separate script then downloads the document again and searches for the exact number. If the number is not found (for example, the source prints only rounded thousands), it stays unpublished until confirmed by hand.

Rosstat's original file Tom5_tab1_VPN-2020.xlsx, including the sheets for 85 regions. A people's regional breakdown is shown only when the regional sheets add up exactly to the national total.

Language classification: Glottolog 5.3. People-to-language links: Wikidata ("language used" and "native language"). Wikidata descriptions are not used because they contain errors.

Allen Ancient DNA Resource v66.p1 (CC0): locations, dates and publications of ancient genomes. Records without a passing quality assessment (PASS, PROVISIONAL_PASS, MERGE_PASS) are excluded.

Every migration statement was checked against the paper; wording was corrected to match the paper and unconfirmed claims were removed. Routes on the globe are illustrative.

No "national character", appearance ratings or other unverifiable generalisations. Genetics describes the history of populations, not the national identity of any individual.