iczn

International Code of Zoological Nomenclature

Working knowledge of the Code — 4th edition, adopted by the International Union of Biological Sciences, in force since 1 January 2000, as amended.

What this skill is. An operational synthesis of all 90 Articles, the Glossary and the Appendices, written rule by rule with the citation attached, plus the data-modelling consequences that follow from them. The Code's text is copyrighted (© International Trust for Zoological Nomenclature 1999) and is not reproduced here; the rules are restated in our own words and every one of them points at the Article it comes from.

Its standing. None. Art. 87 gives equal force to the English and French texts of the Code and to no other version, so a synthesis — this one included — settles nothing in a dispute. Cite the Article, and when the exact wording decides the case, fetch it:

python scripts/fetch_article.py 23.9        # the letter of the law, from the source

Answer in the user's language. The rules are stated here in English because the Code's vocabulary is English and Latin, and translating available name or nomen oblitum into a database field is how those fields end up wrong. Keep the technical terms as they are; write the surrounding explanation in whatever language the user used. references/glossary.md carries a pt-BR column for recognition.

Answering a nomenclatural question

Almost every real question is one of six. Find it here, then open the file named — do not try to answer from memory, because the traps in this domain are all in the sub-articles.

1. "Is this name available?"

Availability is the threshold question: an unavailable name is not a name at all, and nothing else about it matters. It is a permanent property of the name, fixed at publication and untouched by anyone's later taxonomic opinion (Art. 10).

→ `references/04-availability.md` — opens with the checklist, in the order a practitioner applies it. Arts. 10–20. Then, depending on where the checklist stops:

• was the work published at all? → references/03-publication.md (Arts. 7–9)
• electronic-only publication? → `references/zoobank.md`. Without ZooBank

registration under Art. 8.5 the name does not exist. This is the most common modern failure and it is silent.

• published after 1999? → Art. 16 requires explicit indication that the name is new,

and for a species-group name, explicit type fixation naming the depository (Art. 16.4).

• is the entity even within the Code's scope? → Art. 1.3 excludes hypothetical concepts,

teratological specimens, hybrids as such, and infrasubspecific entities: references/01-nomenclature.md.

2. "Which of these two names is valid?"

→ `references/06-validity-priority.md` — Art. 23 Priority, Art. 24 simultaneous publication and the First Reviser.

The trap: Art. 23.9 reversal of precedence has two conditions and they are conjunctive. The senior name must not have been used as valid after 1899 (23.9.1.1) and the junior name must have been used as valid in at least 25 works by at least 10 authors within the preceding 50 years spanning at least 10 years (23.9.1.2). Almost everyone recalls the first and drops the second. If both hold, the author must publish the act citing both names (23.9.2), and the junior name becomes the nomen protectum while the senior becomes the nomen oblitum — the naming is counter-intuitive, so check which is which every time. If the conditions do not hold, the case goes to the Commission (23.9.3); it is not the author's to decide.

3. "Is this the right spelling, ending, or author citation?"

→ `references/07-formation-spelling.md` — Arts. 25–34.

• gender agreement (Art. 31.2) applies only to adjectival epithets, never to nouns in

apposition or genitives, and the spelling alone does not tell you which it is

• justified emendation / unjustified emendation / incorrect subsequent spelling

(Art. 33) are three different things with three different authorship consequences → `references/11-authorship.md` — Arts. 50–51, including Art. 51.3: parentheses around the author mean the species sits in a genus other than the original one. That is a statement about the current combination, not about the name.

4. "Are these homonyms?"

→ `references/12-homonymy.md` — Arts. 52–60. Species-group homonymy exists only within a genus (Art. 53.3), so it is not a property of a name string; Art. 54 lists names that never enter homonymy at all; Art. 58 makes certain spelling variants count as identical — but only when the names share derivation and meaning.

5. "Is this type designation valid?"

→ `references/13-typification.md` (Art. 61, the principle and the objective / subjective synonymy distinction that falls out of it) → `references/16-species-group-types.md` (Arts. 71–76: holotype, syntype, lectotype, neotype, type locality). Two live traps: a lectotype designation after 1999 must satisfy Art. 74.7 in full, and a neotype designation must satisfy every one of the seven conditions of Art. 75.3 — they are conjunctive, and one omission voids the act. → `references/15-genus-group-types.md` (Arts. 66–70: type species, and the precedence of the fixation mechanisms — original designation, then monotypy, then tautonymy, before any subsequent designation can be valid) → `references/14-family-group-types.md` (Arts. 62–65: type genus)

6. "How do I store this?"

→ `references/data-modeling.md` — the name / taxon-concept / nomenclatural-act separation, and the eight cases where naive schemas break. → `references/dwc-mapping.md` — the Darwin Core fields, and the nomenclaturalStatus vs taxonomicStatus split that most datasets get wrong.

Everything else

Two distinctions that decide most answers

Available vs valid. Available means the name passed Arts. 10–20 at publication. It is permanent, objective, and says nothing about whether anyone should use the name. Valid means it is the correct name to use for a taxon under someone's current taxonomic hypothesis — an opinion, revisable, and relative to a circumscription. Every available name has exactly one status on the first axis and a changing status on the second. A dataset with one status column has already lost this and cannot get it back.

Articles vs Recommendations. Articles are mandatory. Recommendations (Rec. 73F), Examples, and Appendices A and B are not (Art. 89.2) — they are good practice, and citing them as binding is a real error in manuscript review. Every reference file here marks which is which; keep that distinction in your answers.

Scripts

Standard library only, no installation.

# Route to the rule offline: the synthesis, its citation, and the URL
python scripts/explain.py                       # chapter map
python scripts/explain.py --article 23.9        # the rule and its sub-rules
python scripts/explain.py --term "nomen oblitum"
python scripts/explain.py --search lectotype    # across all reference files

# Get the Code's exact wording from the source (network required)
python scripts/fetch_article.py 74.7
python scripts/fetch_article.py glossary
python scripts/fetch_article.py 23 --url

# Check names against the mechanically checkable provisions
python scripts/validate_name.py "Panthera onca (Linnaeus, 1758)"
python scripts/validate_name.py --rank family "Felinae"
python scripts/validate_name.py --csv names.csv --name-column scientificName
python scripts/validate_name.py --csv names.csv --json > report.json
python scripts/validate_name.py --selftest      # assertions pinning rules to Articles

# Has the Code changed upstream since these references were written?
python scripts/sync.py --articles 8 9 10 21 78

Reading validator output

Three levels, and the distinction is the whole value of the tool:

• `violation` — decidable from the string itself: a numeral in an epithet

(Art. 11.2), a lower-case genus (Art. 28), -inae on a declared family (Art. 29.2), a date before 1758 (Art. 3), an exact epithet collision inside one genus (Art. 57.2). Safe to act on in bulk.

• `warning` — suspect, and the script cannot decide. Possible gender disagreement

(Art. 31.2) needs the Latin grammatical category of the epithet; an Art. 58 variant collision needs the two etymologies. A warning is a worklist, never an edit.

• `info` — worth knowing, no defect. Tautonymy (Art. 18) is legal; parenthesised

authorship (Art. 51.3) is reported so you notice it is combination-dependent.

validate_name.py deliberately refuses to guess where the Code requires judgement. If you need the judgement call made, make it yourself from the original description — and say that is what you did.

Working practice

Cite, always. ICZN Art. 74.7.3 with a link. A nomenclatural conclusion without an Article number cannot be checked by the person who has to defend it to a reviewer, and is therefore worthless however correct it happens to be.

Never invent an Article number. If you are unsure, explain.py --search or fetch_article.py will tell you in one call. A plausible-looking wrong citation is worse than no citation, because it will be trusted.

Say when the Code is silent. It does not cover taxonomic judgement — whether two populations are one species is not a nomenclatural question, and the Code says so (Preamble, Art. 1.1). Sending a user to the Code for an answer it does not contain wastes their time; name the boundary instead.

Distinguish the act from the conclusion. Reversal of precedence, lectotype designation, and nomen novum publication are acts: they take effect when published, by an author, on a date. Reaching the conclusion privately changes nothing. When a user says "so this name is now the valid one", check whether they mean they have published the act or that they believe the conditions are met.

Pending / contested

Not rules. Do not act on anything in this section. It exists so that a question about contested ground gets a clear "the current rule is X, and there is an active debate" rather than false confidence.

• A fifth edition has been under discussion within the Commission for years. Nothing

in it is in force. Until ratified under Art. 90, the 4th edition governs.

• Names based only on photographs or on DNA sequence, without a preserved specimen.

The Code does not currently require a preserved specimen as the name-bearing type — an illustration can serve in circumscribed cases (Arts. 72.5.6, 73.1.4) — and several such names have been published and heavily criticised. The debate is about whether the Code should require preserved material, and Declaration 45's addition of the Glossary term "specimen, preserved" and Recommendations to Art. 73 (2020) pushes toward preservation without mandating it. Current rule: Art. 73 as amended. Contested: whether photograph-only types should remain acceptable.

• Mandatory registration of all new names in ZooBank. Required now only for

electronic publication (Art. 8.5.3). Extension to print works has been proposed repeatedly and is not in force. See references/zoobank.md.

• Interpreting molecular data as a description or definition under Art. 13.1.1.

Declaration 47 added Examples to guide this; the boundary is still argued in practice.

When a user asks about any of these, give the rule in force, name the debate, and do not blur the two.

References

Chapter by chapter, mirroring the Code's own structure so each file can be checked against its source and updated when the Commission amends it.

scripts/code_index.json maps every Article to its canonical URL, its chapter, the reference file that synthesises it, and a hash of the page as retrieved — which is what lets sync.py notice an upstream amendment.

Related skills

[darwin-core](../darwin-core/) — the standard this skill's names travel in. Use both together whenever nomenclatural data has to be published as DwC-A or DwC-DP; the Darwin Core Conceptual Model's Taxon and Identification classes are where the name / concept separation described in references/data-modeling.md lands.

[skos-xl](../skos-xl/) — for the controlled vocabularies behind field values (nomenclaturalStatus, typeStatus, taxonomicStatus), and for the TDWG NameThing pattern representing names with their nomenclatural provenance as RDF.

[DataProvenance](../DataProvenance/) — nomenclatural acts are events with an agent, a date and a source, which is exactly what W3C PROV records. Use it when a database has to answer "who asserted this synonymy, when, and on what basis".