Show the Working

What the name data leaves out

The name index holds 181,336 names taken from eight birth registers. All eight are incomplete, and no two of them are incomplete in the same way — every one has decided for itself what a small number gives away, and they have reached six different answers. One of those differences will mislead you if nobody points at it. Every figure on this page is counted from the same files the index itself serves.

What is actually in it

Eight registers, and only eight.

Every one of those totals is of the counts that register published, so all but two are a floor: the names held back under the rules below are not in them. Austria and Greenland are the exceptions, as their own lines say — Statistik Austria suppress no name and no number, and Statistics Greenland's smallest published figure is 1, so those two totals are totals.

25,217 spellings are in both the American and the English and Welsh registers. 80,749 are in the American one and not the English one; 14,685 the other way about. Scotland's 5,954 names overlap heavily with both — only 134 of them appear in neither of the other two, and those are the ones you would hope for: Aonghas, Brighde, Beinn, Atholl. France, which is the first register here that is not English-speaking, brings 20,953 spellings the others do not have between them. Northern Ireland brings 10,331 and Ireland 524. Norway brings 324 spellings the others do not have and Sweden 2,497. Belgium, which overlaps France and the Netherlands both, brings 146. Austria brings 7,531 and Switzerland 160, Poland 932 and Canada 618. Argentina brings 4,597 and Finland 9,485. As the next sections show, those differences say as much about the sixteen files as they do about the sixteen countries.

Two of them count something different from the rest

Thirteen of the sixteen count births registered: an event, recorded once, at the time. Argentina, Greenland and Finland count people who are alive, and all three do it differently from each other.

RENAPER's figures are of those born alive and identified as their records stood in September 2024, so an Argentine figure for 2012 is how many people born that year were still on the register twelve years later — not how many babies were registered then. Anyone who died, or never received a document, is not in it.

Over twelve years that is a much smaller distortion than it would be over a century, and it is not nothing. It is here because this site is first of all a list of as many names as possible; it is labelled because a figure that answers a different question should not sit silently in a row of figures that answer the same one.

Argentina also publishes no sex. RENAPER release two files: one names 169 names and says which sex, the other names 14,917 and does not. The second is the one used, and rather than split its counts between girls and boys on no evidence, the Argentine rows say the sex was not recorded.

Finland goes further than Argentina in every direction. DVV publish the names in force of living Finnish citizens, resident in Finland or abroad. Argentina at least arranges its people by year of birth; this has no time dimension whatever, so there is no Finnish figure for any year and no Finnish peak. Somebody who has died is not counted; somebody who changed their name is counted under the new one; a Finnish citizen in Sydney is counted and a resident of Helsinki who is not a citizen is not. It is a census of who holds which name on one day, and it is here because it is also the broadest list on this site after the American and French ones: 23,554 spellings, every given name rather than only first ones.

Sixteen registers, thirteen silences, and two with none

Every register but two has had to decide what a small number gives away, and no two of them have decided the same thing. Four remove the name, at four different thresholds. One removes the year. One removes nothing much. One removes only the number. One changed its mind halfway through. One applies its threshold five hundred times over. One suppresses no figure at all and still leaves a tenth of its babies unnamed. And two hold nothing back at all: Austria, which withholds only which district a birth was in, and Greenland, whose smallest published figure is 1.

The two rules for England and for Scotland look alike and are not: two or fewer in total across thirty years is a far higher bar to clear than two or fewer in a year. So a name present in one column and missing from the next may have done nothing but fall on opposite sides of two different rules, and reading that as a fact about naming would be wrong six different ways.

Northern Ireland names every name

It is the most complete register here, and it is complete in a way none of the others are. NISRA publish the name of every baby registered in Northern Ireland since 1997. What they withhold, where the count is under three, is the number — and they distinguish that from a year the name was not used at all.

26,776 of their 29,788 records therefore have no published count at all. Those names are in this index and they show “count withheld”, never a nought. A nought would say nobody had the name; NISRA are saying somebody did, once or twice, in each of the years listed.

Set beside France that is the whole range in two registers. Insee keep the count and remove the year. NISRA keep the year and remove the count. Both are answering the same question — what does a small number give away — and they have answered it in opposite directions.

One register moved its own bar, and the figures jumped

This is the clearest thing on this page, and it is worth reading before any of the numbers above.

Sweden published a name only where at least ten babies of that sex were given it, until 2003. From 2004 the bar is two. Nothing else about the register changed.

The number of Swedish names published goes from 790 in 2003 to 3,152 in 2004. Four times as many names in twelve months. Sweden did not become four times more inventive; the bar moved.

Anybody using that series to show naming becoming more varied over time would be showing the rule change and calling it a finding. It is the same trap as the peak years further down, and the same answer: the shape of the data is often the shape of the rule that made it.

France hides the year, not the name

This is the one genuinely different silence in the four, and it is worth understanding because it is the opposite trade from everybody else's. America, England and Scotland all decide that a small number is identifying, and remove the name. Insee keep the name and the count and remove the year.

926,369 French births are in that state — named, counted, and attached to no year at all. 2,316 name-and-sex records have no dated year whatever: the count is published and not one of its years is. Those names are in the index, they are findable, and where you meet one it says the years are not recorded rather than showing you a blank.

It also means the French column cannot answer a question about when. “Out of use” will never return one of those names, because nothing is known about when it was used — and a filter that treated an absent year as a year would have called every one of them long dead.

The other French rule costs more. A name is not named at all until it has been given twenty times, and everything below that is pooled into one record covering 1,799,189 births. That is the number of French people this data will not name.

The single-letter rule is the cleanest example. England and Wales record 9 single-letter names — A, C, D, J, K, L, M, S, T — and Scotland records 3, A, C, T. The American file records 0, because its own rule excludes them.

One register folds accents, and only one

NRS say so themselves: Chloé, Chlöe and Chloe “are all counted as the same name”. NISRA say the same thing in their own words — “Names containing accents have been recorded without those accents” — and nobody else does. France is where it shows: 4,159 French spellings carry an accent, and each one is counted separately. Neither the American nor the English and Welsh file contains a single accented name.

Searching this index ignores accents whichever register you are after, so typing Aaron finds Aarôn. What is shown is never folded: each spelling keeps its own row and its own counts, because they are separate entries in the registers and adding them together would invent a total nobody published.

And one register keeps spaces in a name

99 Irish names contain a space or a full stop — Anna Maria, Amelia Rose, A.J. France says in terms that a space separates two forenames and publishes only the first; America, England, Scotland and Northern Ireland have almost none. So a two-word name is a thing the Irish register can express and most of the others cannot, which is worth knowing before reading an absence as a fact about Irish naming.

This matters more than it sounds, because it is the one place where a column that looks comparable is measuring something else. A Scottish count for an accented name covers every way of writing it; the two counts beside it each cover one spelling. Nothing on screen can show you that, which is why it is written here.

Punctuation the American register does not keep

5,938 English and Welsh spellings contain a hyphen or an apostrophe: 5,825 hyphenated, 113 with an apostrophe. Scotland records 310 such names of its own. The number of American spellings containing any character outside A to Z is 0, and not one of those 5,938 names appears in the American file as written.

That is a fact about the file, not about American babies. Whether the punctuation was stripped before publication or never recorded, the data cannot say — so every double-barrelled name in this index reaches it through a British register, and its American figures, if it has any, are filed under the unpunctuated spelling instead.

A name here means a spelling

Neither register merges spellings, and this index does not either. ONS give the reason themselves: Grouping names with similar pronunciation would change the rankings. Exact names are given so users can group if they wish.

The consequence is worth seeing. Therese, Theresé, Therèse, Therése, Thérese, Thérèse, Thérése are 7 separate entries with 7 separate counts, and adding them together is a decision, not a correction. The same applies across the sexes: 11,854 American names, 2,656 English and Welsh ones, 307 Scottish ones, 2,453 French ones, 1,238 Northern Irish ones, 252 Irish ones, 5 Norwegian ones, 469 Swedish ones and 50 Belgian ones, 871 Austrian ones and 108 Swiss ones, 82 Polish ones and 1,557 Canadian ones, 84 Greenlandic ones, and 1,796 Finnish ones appear on both the girls' and the boys' list, kept apart rather than added up. Argentina is not in that sentence because RENAPER publish no sex at all, so no Argentine name can be on two lists.

The two records cover different years, and this one matters

This is the gap most likely to mislead, because nothing on screen looks wrong. America begins in 1880, France in 1900, Norway in 1945, Ireland in 1964, Scotland in 1974, England and Wales in 1996, Northern Ireland in 1997 and Sweden runs 1998 to 2022, Austria in 1984, Switzerland in 2000, Poland in 2000, Canada in 1991, Greenland in 2002, Argentina in 2012, and Belgium 2016 to 2025. When a name's figures sit side by side in the same row, they cover spans of 146, 124, 81, 62, 52, 42, 34, 30, 29, 26, 26, 25, 24, 13 and 10 years, and an English or Welsh peak can only ever fall between 1996 and 2025.

Belgium is the extreme of this and worth saying on its own. Its 10 years against America's 146 is not a smaller version of the same measurement. A name absent from the Belgian column has had ten years to appear in, not a century — and because Belgium publishes no years at all, there is no Belgian peak to compare with anybody's, and no way to tell a name arriving from a name leaving.

Finland is not on that list at all, and that is the point. Belgium has a window of 10 years and declines to say where in it a birth falls. Finland has no window: DVV publish who is alive and holding a name on one day, with no year column and no birth in it. So a Finnish figure cannot be short or long, early or late — it is answering a different question, and every Finnish row on this site says so.

Norway's label is worse than short, it is wrong: the table says 1880 and holds nothing before 1945. Its own note explains why — figures earlier than 1945 are not published — and the span used on this site is counted from the figures rather than copied from the title.

Ethel is the case to hold on to. In America she peaked in 1918, with 278,922 registrations since 1880. In the English and Welsh column her highest year is 2024. That is not a revival — it is the only kind of answer a register starting in 1996 can give. Whether Ethel was more common in Britain in 1918 is a question this data cannot reach, and the Scottish record, which starts in 1974, cannot reach it either.

It is not one name. Of the 26,942 name-and-sex records that appear in both registers, 12,171 peak in 2015 or later on the British side, and 851 peak before 1950 in America and after 2015 in Britain. Every one of those 851 looks like a comeback in the table. The data cannot tell you which of them is one.

How a name is said, where anyone has written it down

17,539 of the 181,336 names have a recorded pronunciation, from the CMU Pronouncing Dictionary. 163,797 do not. A name without one is not a name nobody can say — it is a name that is not in that dictionary, which is a list of English words and common American names rather than a register of every name given to a child.

Meanings are not here yet

There are none on the site at present, and a made-up etymology is worse than none at all. Where meanings do arrive they will come from named scholarship old enough to be out of copyright, they will be attributed, and where two authorities disagree the page will say so rather than picking one.

Every name is searchable; not every name has a page

There are 24,204 name pages, one for each group of spellings, and between them they cover 27,037 of the spellings in the index. A name earns a page by having been given to a thousand or more babies, or by having a recorded pronunciation. The remaining 154,299 are not missing: they are all in the index, all searchable, and every one of them has a link. Typing a name puts it in the address bar — /baby-names#ethel — and that link works whether the name has a page or not.

Nearly every country records the name. Almost none counts it

This is the most useful thing on the page, and it took a while to see. When a country is missing from this site it is almost never because nobody wrote the name down. It is because nobody was ever asked to add it up.

The international standard for vital statistics is the United Nations’ Principles and Recommendations for a Vital Statistics System, and it draws the line in two separate places.

The name is in the record. Annex I lists the information a live birth record should hold for judicial and administrative purposes, and under Characteristics of the child the first item is Name, followed by sex and the assigned personal identification number. That is the birth certificate, and it exists in every country with working civil registration.

The name is in none of the statistics. The same document recommends twenty tabulations of live births — LB‑1 to LB‑20 — classifying births by date, place, sex, birth weight, gestation, the age and education and marital status of the mother, birth order, attendant at delivery and more. Not one of the twenty involves the child’s name.

So a statistical office that publishes no names is not hiding anything and is not behind. It is doing exactly what the standard asks. The thirteen registers on this site are the ones that went further than they had to, and that is the unusual thing — not the silence of everybody else.

It also explains why two of the best sources found so far are not here. New Zealand told the world in a press release that 18,666 different first names were given to 57,889 babies in 2023, and publishes the top hundred. Chile released a hundred and two years of registered names — 858,782 rows — to somebody who asked under its transparency law. In both cases the data was sitting there, counted, and simply not published, because nothing obliged anyone to publish it.

The rest of the world

Thirteen registers, six of them English-speaking in whole or in part, is still a narrow view of naming and this page is not going to pretend otherwise. Adding a country needs two things at once: figures published in a form a machine can read, and a licence that permits reuse including commercially. Where both hold, a register can be added. Where the licence does not, nothing is taken.

Spain counts the living, not the born

Spain publishes name figures, they are machine-readable, and the licence permits reuse. They are still not here, and the reason is worth more than the data would have been.

What the National Statistics Institute publishes is a count of people resident in Spain who bear a name — in their own words, las personas que se encuentran residiendo en el territorio español, drawn from the Padrón and the Censo de Población. Every other register on this site counts birth registrations: events, recorded once, at the time.

Those two things answer different questions. A census of the living forgets everyone who has died and counts everyone who has arrived, so it measures migration and mortality as much as naming — and a name popular a century ago is thinned by both. Put a Spanish figure in a row beside eight counts of births and it would read as the ninth of a kind. It is not one.

So Spain is left out on purpose. If this site ever adds a second kind of figure, it will be in its own table, with its own heading, and not in that row.

Germany: the record exists, and nobody holds all of it

This is the first country here that has actually been looked at, and it is not the simple case the paragraph above was written for. Germany does not publish nothing. Germany publishes a great deal, and no national body assembles it.

The Federal Statistical Office says so itself. Its release on the commonest names given in 2018 gives Paul and Marie — and credits them to somebody else: Quelle: Gesellschaft für Deutsche Sprache, a language society rather than the statistical office. Destatis count the births; a private body counts the names.

What is published instead is municipal. Every German dataset on the European open data portal belongs to a single city — Münster, Aachen, Kleve, Berlin, Munich — each keeping its own list. That is a real record of German naming, and it is scattered across town halls in a form nothing on this site can use as a country.

Said carefully, because the checking was not exhaustive: no national German name register was found, in the federal office's own words about its own figures or in the European portal. That is weaker than the Eurostat finding above, which was tested against a catalogue proved to contain what it was searched for. It is not proof that none exists.

France is the reason this is worth writing down at all. INSEE publish a national file and dozens of French communes publish their own, so a municipal record is not an alternative to a national one. Germany has the second without the first.

What this page will not do is publish a list of countries that supposedly publish nothing. Checking what every national statistics office puts out, and on what terms, is a piece of work nobody here has done yet — and a confident list assembled from memory is precisely the kind of thing this site exists to be the opposite of. Germany above is the first entry earned by checking; the rest of the list arrives the same way or not at all.

Why say all this

Because a missing name and a rare name look identical in a search box, and so do a revival and a short record. Every site built on these registers inherits the same gaps; the difference is whether it tells you. The index itself repeats the parts that matter where you will meet them.