The Patronymic: The Field 52 Countries Don't Have

We audited the patronymic across 64 locales. The field carries information in 12 of them. In the other 52 the only honest value is an empty string.

That ratio is not the interesting part. The interesting part is that almost every wrong answer we found in the literature, in form designs and in other generators comes from the same mistake: three unrelated phenomena all get filed under "patronymic," and once they are merged nobody can tell which countries actually have the thing.

Here they are, separated.

Rows 1 and 2 are the 12. Row 3 is where most of the 52 live, and it is the single largest source of confusion in the whole subject: a surname that means "son of Jens" is not a patronymic, any more than a street called Mill Lane is a mill.

The 12 that have it

We wrote out the derivation rule for each locale and ran it over every male given name that locale's corpus holds — the full pool of possible fathers, both bearer sexes, every row inspected. Here is what came back.

5,716 distinct forms across the 12. Zero across the other 52.

Notice the column that splits the table in half. In five locales the form depends on whether the card holder is a son or a daughter. In seven it does not — and that is not sloppiness, it is grammar. A Greek πατρώνυμο is the father's name in the genitive, and the father is male whatever his children are; ΙΩΑΝΝΗΣ του ΓΕΩΡΓΙΟΥ and ΜΑΡΙΑ του ΓΕΩΡΓΙΟΥ carry the identical ΓΕΩΡΓΙΟΥ. The Slavic-type patronymic is the odd one out, not the norm.

Five where the bearer's sex changes the string

Ukrainian is the cleanest: 597 of the 603 names in the corpus take -ович / -івна, and exactly six take the old -ич suffix — Ілля → Ілліч, Лука → Лукич, Фома → Фомич, Хома → Хомич, Кузьма → Кузьмич, Косма → Космич. Six exceptions in 603 names, and every one of them is listed by name in the 2019 orthography.

Russian looks similar and is not. Running the rule over 316 names gives a five-way split:

The -ьевич / -иевич boundary inside names ending -ий is the trap. Георгий cannot take a soft sign — Russian orthography does not permit гь — so it goes Георгиевич. But the widely repeated shortcut "two consonants before the ending means -иевич" is simply wrong: Лаврентий → Лаврентьевич and Иннокентий → Иннокентьевич keep the soft sign, because a sonorant plus consonant behaves differently from an obstruent cluster. Eight names out of 316 sit on the wrong side of a bad rule.

Bulgarian is not a suffix at all in the Slavic-patronymic sense — it is the surname machinery applied one generation up. бащино име takes -ов or -ев plus a feminine -а, exactly as a Bulgarian surname does; the two are formed by the same rule from names one generation apart. Our run splits 360 names into 261 -ов and 99 -ев, with no residue.

Kazakh carries two parallel systems in one country, and the citizen picks. Native -ұлы / -қызы are not affixes at all — they are the words ұl "son" and qyz "daughter" with a third-person possessive, which is why they never vowel-harmonise: Серікұлы, Әсемқызы, Жақсылыққызы with the consonant cluster intact. Running the rule over the Kazakh corpus routes 33 of 285 names (11.6%) to the inherited Soviet -ович / -овна instead, because those 33 are Slavic given names and Иванұлы does not occur in real documents.

Moldovan is the one that is being switched off while we watch. The law makes it optional and — by its own text — for persons of non-Moldovan ethnicity; the modernised identity card introduced in 2025 stops printing it while keeping it in the population register. Our rule always returns a non-empty Romanian-orthography Slavic form (Alexandrovici / Alexandrovna), because deciding whether to show the field is a different decision from knowing what it would say.

Collisions are worth one line, because they are real and small: 10 Ukrainian pairs, 7 Moldovan, 2 Russian and 2 Bulgarian collapse two father names onto one patronymic. Антон and Антін both give Антонович. Гавриил and Гаврила both give Гаврилович. That is a property of the language, not a bug — the patronymic is lossy.

Two where it is a case, not a suffix

Greek takes the genitive, and the genitive moves the stress. Άγγελος → Αγγέλου, Νικόλαος → Νικολάου, Αθανάσιος → Αθανασίου, Βασίλειος → Βασιλείου: any name stressed on the third syllable from the end must move its accent to the penult in the genitive singular. String concatenation cannot do this; you have to syllabify, and you have to know that αι ει οι ου αυ ευ ηυ υι are single syllables while a diaeresis breaks them apart (Παΐσιος → Παϊσίου).

Sixteen names in the Greek corpus come back unchanged, and they are all the same kind of name — biblical indeclinables: Αβραάμ, Αδάμ, Βενιαμίν, Γαβριήλ, Δαβίδ, Δανιήλ, Εμμανουήλ, Ευφραίμ, Ισαάκ, Ιωακείμ, Ιωσήφ, Μιχαήλ, Ναθαναήλ, Ραφαήλ, Σεραφείμ, Συμεών. A validator that flags "output equals input" as a failure would raise sixteen false alarms here.

There is a second Greek subtlety worth naming because it decides what your output looks like: the box on a form labelled Όνομα πατέρα takes the nominative (ΓΕΩΡΓΙΟΣ), while the πατρώνυμο as a component of the name — in registry extracts, electoral rolls, notarial and tax documents — appears in the genitive after του. Same information, two grammatical forms, depending on which document you are looking at.

Armenian does the same thing with a smaller rulebook: genitive in -ի, with an epenthetic յ after -ա/-ո/-օ (Հրաչյա → Հրաչյայի, Կամո → Կամոյի). One genuine wrinkle: names in -ի split by origin — Russian borrowings stay in the -ի declension (Յուրի → Յուրիի) while native Armenian ones move to the -ու declension (Աղասի → Աղասու, a form attested in Abovian). Armenian stress is always final and unwritten, so nothing has to move.

Five where it is a verbatim copy

Jordan, Saudi Arabia, Iran, Israel and Turkey between them contribute 1,559 father names to the table above, and in all 1,559 cases the output equals the input. Character for character, identical for a son and for a daughter. There is no morphology.

This is the category where the temptation to add something is strongest, and where adding something does the most damage.

Do not add بن. Ceremonial, tribal and royal Arabic usage makes it extremely visible — محمد بن سلمان بن عبدالعزيز آل سعود — and it is exactly for that reason that inserting it into ordinary civil-registry data is wrong. Saudi and Jordanian identity documents write the chain by simple apposition: فهد عبدالعزيز سعود القحطاني. Add the particle and every generated record starts to read like a royal genealogy. Nor is there a feminine بنت in the register: a daughter's entry looks like a son's.

Do not add בן for the same reason, plus a second one specific to Hebrew. בן is itself a given name (Ben), and it is a component of compound given names (בן ציון, בן עמי) and surnames (בן גוריון). Prefix it mechanically and you get בן בן ציון. The Israeli population registry prints שם האב — bilingually labelled שם האב / اسم الأب on the identity card — and the box contains the father's name, unadorned. The בן/בת construction belongs to the religious and ceremonial register: aliyah to the Torah, the ketubah, a gravestone, a prayer for the sick.

Do not add -oğlu in Turkey, which is the sharpest case of all, because that suffix was legislated out of exactly this use. The 1934 Surname Law replaced Ottoman patronymics with fixed heritable surnames; Mustafaoğlu Mehmet has been impossible since. The surnames in -oğlu that survive — Süleymanoğlu, Sakaoğlu — are frozen family names, 78 entries in our 1,337-name Turkish surname list, 2.9% of its weight, and they have nothing to do with the living Baba adı field, which still sits printed on the chipped identity card alongside Ana adı.

Persian has the same shape: نام پدر is a separate line on the shenasnameh and the national card, filled with the father's name exactly as it appears in his certificate, no izafe and no suffix. The patronymic suffixes ‑زاده, ‑پور, ‑نژاد, ‑یان are surnames, hereditary since the registration laws of the 1918–1935 period, and they are structurally the Icelandic case in reverse: the derivation happened once and stopped.

One practical consequence of the copy-verbatim group: because the value is a plain copy, these five fields must be rendered as their own labelled box, not spliced into a full-name string. A Turk's name is Ad + Soyad; a Persian's is نام + نام خانوادگی. The Jordanian and Saudi chains are the exception inside the exception — there the father's name genuinely does sit inside the name string, between the personal name and the family name.

The fossils: why "-son means patronymic" is false

Now the 52. Most of them once had a living patronymic; the reason they no longer have the field is that the patronymic won. It stopped being recalculated, hardened into a surname, and the surname slot swallowed it.

Here is the census, measured on our surname corpora. The "share of file weight" column is the share of the dataset's total weight carried by entries with that ending — roughly, how much of the population walks around with a fossilised patronymic in the surname box.

Denmark is the extreme case and deserves the emphasis. Three names — Nielsen, Jensen, Hansen — carry 21.0% of our Danish surname file's weight, and the top ten carry 46.3%. Every one of the ten is a frozen patronymic. Denmark froze first and hardest in Scandinavia, and its surname distribution has never recovered.

Two entries in that table need an honest caveat rather than a footnote. Armenian -յան covers 444 of our 448 Armenian surnames (99.94% of weight) and Russian -ов/-ев covers 864 of 1,335 (81.87%), but both suffixes attach to an ancestor's given name or to a trade, a place or a nickname. They are possessive, not specifically patronymic, so we have left them out of the table above: you cannot read "descendant of a man called X" off the suffix alone.

And one number in our own data is a property of our list rather than of the country: our Serbian surname corpus is 958 entries, of which 958 end in -ić. Serbian really is dominated by -ić, but not to four significant figures. Treat 100% as a selection artefact of the list we hold, and compare it with the Croatian 69.78% — the same suffix, the same history, a differently built list.

The diagnostic that actually works

If you want a mechanical test for "living patronymic vs. fossil," ignore the suffix and look at whether the form tracks the bearer's sex.

Across our 64 locales, the male and female surname files are byte-for-byte identical in 51 of them. They differ in 13: bg_BG, cs_CZ, el_GR, is_IS, ka_GE, kk_KZ, lv_LV, pl_PL, ru_RU, sk_SK, uk_UA, bn_BD and en_IN.

Ten of those thirteen differ because the surname itself carries grammatical gender — Czech Novák/Nováková, Polish Kowalski/Kowalska, Greek nominative/genitive, Latvian declension classes. Two more (bn_BD, en_IN) simply hold lists of different lengths.

Exactly one differs because of patronymy. That one is Iceland.

Iceland: where the patronymic won and became the document

Our Icelandic surname list holds 343 entries in each file. 284 of them are patronymics; 59 are inherited family names. The male file reads Jónsson, Sigurðsson, Guðmundsson…; the female file reads Jónsdóttir, Sigurðardóttir, Guðmundsdóttir…; and the weight vectors are identical row for row — not approximately, exactly, in all 343 positions. The 59 inherited names are the same 59 strings in both files.

Two details from that comparison are worth pulling out.

Exactly one pair is not a suffix swap. Of the 284 patronymic pairs, 283 share a stem; one does not. Sigurðsson / Sigurðardóttir — because Sigurður has two competing genitives, Sigurðs and Sigurðar, and usage settled the son-form on one and the daughter-form on the other. If you built the female list by string surgery on the male list you would get 283 right and 1 wrong, and the one you got wrong would be the second most common name in the country.

Fifteen of the 59 inherited names end in -sen: Petersen, Hansen, Thoroddsen, Stephensen, Ottesen, Nielsen, Jensen, Knudsen, Finsen, Olsen, Thorarensen, Sívertsen, Berndsen, Siemsen, Claessen. Read that next to the Danish row of the census table. The suffix that is the fossilised patronymic in Denmark is, in Iceland, the marker of a name that is not a patronymic — a Danicised, inherited family name from the era when studying in Copenhagen meant coming home with a surname. Same string, opposite meaning, 500 km apart.

So Iceland is why the taxonomy at the top of this article has three rows and not two. The Icelandic patronymic is fully alive: it is recomputed every generation, it agrees with the bearer's sex, it is derived from the father's given name in the genitive. It ticks every box for row 1. And the correct value for a separate patronymic field in Iceland is still empty — because the patronymic is already on the document, in the surname box. Þjóðskrá keeps eiginnafn, millinafn, kenninafn, and kenninafn is the patronymic. Add a fourth field and you print Jón Jónsson Jónsson.

The rule that comes out of this: "does the country have a living patronymic" and "does the card need a patronymic field" are two different questions, and Iceland answers yes to the first and no to the second.

Six impostors that sit in the same slot

The remaining confusion is not about suffixes at all. It is that many naming systems have something between the given name and the surname, and that something gets read as a patronymic because that is the reader's nearest available category.

Philippines — the middle name is the mother's maiden surname. A legitimate child is [given] + [mother's maiden surname] + [father's surname]. The middle component is taken from the mother's surname, unchanged; the father's given name does not appear anywhere in it. The clinching detail is what happens when the schema breaks: a child born outside marriage takes the mother's surname and has no middle name at all. A patronymic does not behave that way. (The Jr./III suffixes are namesake markers; Art. 375 of the Civil Code makes the middle name obligatory precisely so that a grandson cannot also be "Junior.")

Vietnam — Văn and Thị encode sex, not paternity. Nguyễn Văn An is a man and Nguyễn Thị An is a woman; the contrast is sex and nothing else, and the same two morphemes serve every family in the country. Our own corpus reflects this from a completely independent direction: Văn occurs as a name element in 5 of 294 male given-name entries and 0 of 248 female ones, and Thị occurs in neither list — consistent with the well-documented collapse of Thị/Văn among younger cohorts in favour of semantic middle names like Bảo, Gia, Ngọc.

China and Taiwan — the generation syllable cannot encode a father, for a reason of physics. 字辈 / 班次 is a syllable shared by brothers and cousins of one generation. The sequence is fixed by a "generation poem" composed by the clan elders and recorded in the 族谱, often covering dozens of generations in advance. The syllable a child receives was therefore chosen before the child's father was born, in many cases before his grandfather was born. It marks a rung on the ladder, not a parent. Taiwan preserves the practice noticeably better than the mainland, where it faded after the one-child policy — but the logic is identical in both.

Korea — same structure, different name. 항렬자 / 돌림자 is assigned by the clan book 족보 for an entire generation, often cycling through the five elements, and even the position of the syllable (first or second) alternates by generation. Cousins with different fathers share it. The modern habit of giving siblings a common syllable (Minji / Mina / Minyoung) is not even 항렬자 any more — it is parental taste.

Uganda — the surname is drawn from the clan's list and is deliberately not the father's name. Among the Baganda, each of the 50-plus clans owns a list of male and female names, and a child's "surname" is chosen from the clan's list. The normal outcome is that the child's surname differs from the father's — the exact inverse of patronymy. The clan name is sex-marked (Nsubuga male, Nansubuga female, with the Na-/Nna- prefix); in our 466-entry Ugandan surname list, 38 entries carry that prefix, though our male and female files are byte-identical, so we do not currently model the pairing. The Runyoro/Rutooro empaako — Akiiki, Amooti, Atwooki and about a dozen others — is a shared praise name given by custom, again not derived from a father.

Tamil Nadu — real patronymy, wrong slot. In Tamil Nadu and parts of Kerala and Karnataka the father's name precedes the personal name, usually abbreviated to an initial: R. Anirudh, where R is Ramachandran. That is patronymy: the reference is to the individual father, and it updates every generation. But there is no derivation — the father's name is taken as it stands, identically for a daughter (R. Priya) — and when a passport is issued the initial is expanded and placed in the SURNAME field. The phenomenon is real and it belongs to the surname generator, not to a patronymic field; producing both would print the father's name twice.

Three smaller ones, for completeness:

• Romania writes Popescu I. Ion on diplomas, court filings and cadastral records — the initial of the father's given name. It is one letter, it does not agree with the bearer's sex (Ion I. Popescu / Ioana I. Popescu), it is not part of the legal name under Art. 83 of the Civil Code, and when the father is unknown it is replaced by the mother's initial. Four reasons it is not a patronymic, any one of which would be enough.
• Hungary does have a mandatory identifying field that everyone fills in on every form — and it is anyja neve, the mother's birth name, recorded in full with no derivation. The -né suffix (Kovács Jánosné) is a married woman's name formed from her husband's full name. The -fi/-ffy surnames (Péterfi, Pálffy) are the fossil case.
• Czechia, Slovakia and Poland have -ová and -ówna, which look patronymic and are not: they derive from the surname, not from a given name. Jan Novák's daughter is Nováková, never "Janovna." Since 2022 any Czech woman may register the undeclined masculine form without giving a reason, which settles the question — it is a variant of a surname, not a separate field.

And one country that really does print a parents field, verbatim. The Polish dowód osobisty carries IMIONA RODZICÓW / PARENTS' GIVEN NAMES on the reverse. The values are the parents' names as they are — Jan, Anna — with no derivation of any kind. That is row 2 of the taxonomy, in a country that everyone files under row 3. If you need to fill that box, you copy a given name; you do not compute one. The same is true of the Serbian identity card, where "name of one parent" lives in the chip and not in print, and of the Indonesian KTP, where bin / binti — genuinely used, mostly in religious-legal documents and by particular communities — sits inside the single Nama Lengkap string because Indonesian law recognises no surname field at all.

The four questions

If you are deciding whether a locale needs a patronymic field, these four questions settle it, in order:

1. Is it derived from the father's given name? No → it is a surname (Uganda, Philippines, Nepal, Bangladesh, most of India's north), a generation marker (China, Taiwan, Korea), or a sex marker (Vietnam).
2. Is it recomputed each generation? No → it is a fossil, and it lives in the surname slot (-sen, -son, -ez, -ić, -owicz, -oğlu, -ძე, -sons, -fi).
3. Does the answer to 1 and 2 come out yes, but the value already occupies the surname box? Yes → Iceland. Field stays empty.
4. Is it part of the legal name, or a separate box on the form? Both are real, but they render differently: Ukrainian, Russian, Bulgarian, Kazakh, Moldovan, Greek, Armenian and the Arabic chains go inside the name; Persian, Hebrew and Turkish go in a labelled box beside it.

Question 2 is the one that eliminates the most locales, and question 3 is the one everybody forgets.

What we are not confident about

• Moldova is a moving target. The field exists in the population register, was printed on the older card, is optional by law and restricted by law to persons of non-Moldovan ethnicity, and stopped being printed on the 2025 card. Any single answer for ro_MD is wrong for some documents.
• Kazakhstan has a legal branch we do not model. A Kazakh who adopts -ұлы/-қызы as a surname is not given a separate patronymic at all. Our rule always returns one.
• Greek learned genitives are not covered. Ancient names in -ης have a parallel learned genitive (Περικλής → Περικλέους, Σωκράτης → Σωκράτους). We produce the demotic Περικλή, which is correct modern Greek, but which registry you are imitating decides the answer. Ιωάννης → Ιωάννου is in our exception table because the learned form dominates there overwhelmingly.
• Ugandan clan names are not sex-paired in our data, although the real system pairs them.
• The Serbian 100% -ić figure is ours, not Serbia's — see above.
• The -oğlu restriction. The 1934 Surname Law and its accompanying regulation ended patronymic surnames; the specific prohibition on new -oğlu endings comes from the regulation rather than the statute, and we have not verified the exact article wording. It does not affect anything above.

The one-line version

Twelve locales out of sixty-four have a patronymic field. Five of those twelve change the string depending on whether the card belongs to a son or a daughter; five copy the father's name verbatim; two decline it into the genitive and hand the same form to both sexes. Everywhere else the patronymic either never existed, or won so completely that it became the surname — and Iceland, the one country where it is both fully alive and printed on every document, is precisely the country where the separate field must be left blank.

Corpus figures measured 2026-07-22 on the 64-locale name datasets. Counts described as "in the file" are entry counts and weight shares of those datasets; the weight column of a dataset is the relative frequency it assigns each entry. Derivation counts (father-name pools, distinct forms, collisions, suffix splits) were produced by running each locale's rule across every male given name in that locale's corpus for both bearer sexes.

Sources — the 12:

• Ukraine: Civil Code of Ukraine, Art. 28(1); Ukrainian Orthography 2019 (CMU res. 437 of 22.05.2019), § 41 "Formation of masculine and feminine patronymics"; Ministry of Justice order 52/5 on civil status registration.
• Russia: Federal Law 143-FZ "On Acts of Civil Status", Art. 18; Family Code Art. 58; passport form per MVD orders 1259 / 889. Formation rules: Petrovsky, Dictionary of Russian Personal Names (patronymics appendix); Superanskaya; Rosenthal.
• Bulgaria: Закон за гражданската регистрация, чл. 13 (SG 96/2004); Institute for Bulgarian Language, BAS, "За имената на бащите ни".
• Kazakhstan: Code on Marriage and Family, Art. 63; Presidential Decree 2923 of 02.04.1996 on the writing of Kazakh surnames and patronymics; Қазақ тілі емлесінің ережелері (Baitursynuly Institute of Linguistics).
• Moldova: Legea nr. 100/2001 privind actele de stare civilă, art. 5(5); Agenţia Servicii Publice, 2025 identity-card redesign.
• Greece: standard civil-status and notarial formula Όνομα Επώνυμο του <πατρώνυμο>; M. Triantafyllidis, Νεοελληνική Γραμματική, on the genitive of masculine nouns in -ος/-ης/-ας and the accent shift.
• Armenia: Law of the Republic of Armenia on Identification Cards (parliament.am, legislation ID 4348); U.S. Department of State Visa Reciprocity Schedule, Armenia.
• Jordan: قانون الأحوال المدنية رقم 9 لسنة 2001; Civil Status and Passports Department (cspd.gov.jo), family register and ID content.
• Saudi Arabia: Ministry of Interior, Civil Affairs Agency (moi.gov.sa) — بطاقة الهوية الوطنية and سجل الأسرة field lists.
• Iran: قانون ثبت احوال, art. 41; National Organization for Civil Registration (sabteahval.ir) — نام پدر on the shenasnameh and the smart national card.
• Israel: חוק מרשם האוכלוסין, תשכ"ה-1965, § 2(2); תקנות מרשם האוכלוסין (רישומים בתעודת זהות), תש"ן-1990.
• Turkey: Soyadı Kanunu No. 2525 (21.06.1934) and Soyadı Nizamnamesi (27.12.1934); Nüfus Hizmetleri Kanunu No. 5490 — baba adı / ana adı as nüfus kütüğü fields printed on the chipped T.C. Kimlik Kartı.

Sources — the fossils and the impostors:

• Nordic: Lög um mannanöfn nr. 45/1996 arts. 8–9 and lög nr. 80/2019 (Iceland, the -bur option; Þjóðskrá fields eiginnafn / millinafn / kenninafn); Navneloven lov nr. 524 af 24.06.2005 § 7 (Denmark); Lov om personnavn LOV-2002-06-07-19 § 4 (Norway); lag (2016:1013) om personnamn (Sweden); Etu- ja sukunimilaki 946/2017 and the 1920 surname act (Finland). All four mainland acts permit a parent-derived name as a surname and define no patronymic field.
• Western Europe: BW Boek 1 arts. 1:4–1:7 and RvIG Logisch Ontwerp BRP rubrics 01.02.20/30/40 (Netherlands); Belgisch BW art. 335 as amended 8 May 2014 (Belgium); the Napoleonic decree of 18 August 1811 and the Dutch royal decree of 8 November 1825; BGB §§ 1616–1618 and PAuswG § 5(2), plus the Frisian patronymic-as-Geburtsname option effective 01.05.2025 (Germany); ZGB arts. 160, 270–270b and the Zivilstandsverordnung art. 8 (Switzerland); Code civil art. 57 and Loi nº 2002-304 (France); D.P.R. 396/2000 arts. 34–35 and Corte costituzionale sent. 131/2022 (Italy).
• Iberia and Latin America: Ley 20/2011 art. 49 (Spain); Código do Registo Civil art. 103 (Portugal); Lei nº 6.015/1973 arts. 54–55 as amended by Lei nº 14.382/2022 (Brazil); Código Civil del Perú arts. 19–20 with Tribunal Constitucional Exp. 02970-2019-PHC/TC; Ley Orgánica de Registro Civil art. 93 (Venezuela); Código Civil y Comercial art. 64 (Argentina).
• Central and Eastern Europe: Prawo o aktach stanu cywilnego art. 49(2) and Dz.U. 2015 poz. 212, the IMIONA RODZICÓW rubric (Poland); zákon č. 301/2000 Sb. §§ 61, 69 with the 2022 amendment (Czechia); zákon NR SR č. 300/1993 Z. z. §§ 2, 4, 7 (Slovakia); Porodični zakon RS čl. 342 and Zakon o ličnoj karti (Serbia); Zakon o osobnom imenu NN 118/12 čl. 2 (Croatia); Zakon o ličnom imenu CG, Sl. list CG 47/08 (Montenegro); Zakon o osebnem imenu ZOI-1 3. člen (Slovenia); MK noteikumi Nr. 114 (Latvia); 2010. évi I. törvény az anyakönyvi eljárásról (Hungary); Codul civil art. 83, Legea nr. 119/1996 and Ordinul MENCS nr. 3844/2016 on the father's initial (Romania); Law of Georgia on Civil Status Acts, matsne.gov.ge doc. 1541247.
• Philippines: Civil Code (RA 386) arts. 364 and 375; In re Adoption of Stephanie Nathy Astorga Garcia, G.R. No. 148311 (2005); Alanis III v. CA, G.R. No. 216425 (2020); the 1849 Clavería decree and the Catálogo alfabético de apellidos.
• East and Southeast Asia: Luật Hộ tịch 2014, SBS Cultural Atlas "Vietnamese — Naming" and published research on gender markers in a ~1 million-name 2006 entrance-examination corpus (Vietnam); 中华人民共和国居民身份证法 (2003), 户口簿 and 民法典 § 1015 (China); 姓名條例, 戶籍法 and 民法 §§ 1000, 1059 (Taiwan); 가족관계의 등록 등에 관한 법률 (2008) with descriptions of 항렬자 / 돌림자 and 족보 (Korea); 戸籍法 and 民法 § 750 (Japan); UU No. 23/2006, UU 24/2013 and Permendagri No. 73/2022 with the single Nama Lengkap KTP-el field (Indonesia); descriptions of 字辈 / 班次 and the generation poem.
• South Asia and Africa: Passport Seva (MEA) on the optional surname field and UIDAI's single Name field, with descriptions of Tamil, Telugu (intiperu) and Malayali (tharavad) naming (India); FamilySearch "Nepal Naming Customs" and the thar/gotra system; Births and Deaths Registration Act 2004 (Bangladesh); M. B. Nsimbi, Baganda Traditional Personal Names (Uganda Journal, 1950) and Luganda Names, Clans and Totems (1980), J. Roscoe, The Baganda (1911), and Births and Deaths Registration Act Cap. 309 (Uganda); Births and Deaths Registration Act 51 of 1992 with material on izithakazelo / iziduko (South Africa).

A note on verification. Everything above that is stated as a count — 12 of 64, 5,716 forms, 284 patronymics and 59 family names in Iceland, 15 of those 59 ending -sen, 597 of 603 Ukrainian names in -ович, 33 of 285 Kazakh names routed to -ович, Văn in 5 of 294 male given-name entries and 0 of 248 female ones, 51 of 64 surname file pairs byte-identical — is a measurement on our own datasets, reproducible from them, and describes our data. Where our data and a national registry agree, that is a consistency check and not independent confirmation. Where they disagree, we have said so: the Serbian -ić share and the unpaired Ugandan clan names are both ours to fix, not the countries' to explain.