Methodology & data sources

Every number on this site comes from an official birth register or national statistics office. We clean each source, put them on one scale, and group the forms of each name across languages.

1. Collecting the data

We only use official open data: national statistics offices (INSEE, ONS, CSO, Statbel, BFS…), civil registries (SSA, RENAPER, provincial registrars) and regional open-data portals. Each source keeps its own threshold: names given fewer than 3 or 5 times in a year are not published, so very rare names are missing everywhere. Some countries only publish a top 50 or top 100; for those we show ranks, never shares. The table below lists every source, its years, publisher and licence.

2. Putting countries on one scale

Raw counts are not comparable between a country of 330 million and one of 5 million, so charts show births per 100,000 newborns (both sexes combined) in each country and year. The denominator is the sum of births in the published lists, which is close to, but slightly below, total births.

Argentina's register stores every given name as one string ("María Florencia"), so we count the first given name only. Where a register has no sex column, the sex of a name is taken from other registers in the same language.

3. Spellings, names and families

We work on three levels:

• Spelling: Élodie and Elodie, or Александр and Aleksandr, are the same name written differently. Accents, case and script are folded together and every spelling stays visible on the name's page.
• Name: a spelling plus a sex. Jean is a boy's name in France and mostly a girl's name in the US, so these are two different names.
• Family: the forms of one name across languages. Victoria (English, Spanish), Victoire (French), Vittoria (Italian) and Viktoriya (Russian) form one family, with one main page.

4. How names are grouped

Four kinds of evidence are used, strongest first. Each family page lists the evidence behind every link.

1. Checked by hand: a curated list of about 200 cross-language families (John = Jean = Juan = Giovanni = João = Ivan…). Two curated families are never merged automatically.
2. Verified against linguistic sources: when our rules are unsure (Marcel / Marcelo, Laure / Laura, Eva / Evie), the pair is checked against etymological references and decided as same, related or different.
3. Spelling variants: forms that sound the same once spelling conventions are normalised (ph/f, th/t, y/i, doubled letters), with the same first letter and the same sex: Sofia / Sophia, Matteo / Mateo.
4. Wikidata: the open "said to be the same as" links, used only to attach an isolated name to a family when both forms look alike. Wikidata alone links too much (Kyle ~ Caleb), so it is never allowed to merge two existing families.

Diminutives and derived names that live their own life (Jack, Sean, Ian for John; Isabelle for Elizabeth; Evie for Eva) keep their own page and are shown as related names. Masculine and feminine forms (Victor / Victoria) are linked too.

5. Statistics

• Name length: average number of letters, weighted by births.
• Diversity: share of births going to the ten most common names, number of names needed to cover half of births, and the effective number of names (1/Σp²).
• Sounds: share of births whose name ends in -a, -o, -n, an "ee" sound, rhymes with Aiden, contains a Y, K or Z, and so on.
• Convergence: cosine similarity between two countries' distributions of name families. Because families cross languages, a Victoire in Paris and a Victoria in London count as the same choice.
• Similar trajectories: names whose popularity curve over the last 60 years correlates most closely in the same country.

6. Limits

The current INSEE file rounds French counts to a multiple of 5, and the SSA omits names given fewer than 5 times. Registers differ in what they record (registration year or birth year, accents kept or stripped, hyphenated names split or not). Small populations produce noisy curves. Grouping names across languages is a judgement call; if you think a link is wrong, tell us and we will review the evidence.

Data sources