Correct check, wrong edition
In a fictional case, two coherent ISBN-13s describe two book formats. A shared title does not establish removable duplicates.
In-depth practical guides
Separate format, check digit, assignment and matching when linking authors, publications and versions.
Updated :

AI-generated illustration
An identifier needs its namespace and referent. ORCID identifies a person, ISSN a continuing resource, ISBN a publication edition and format; a DOI can identify different objects. An arithmetic check catches some copying errors, not identity.
Keep leading zeros, punctuation and the observed URL. Separate original and normalised values; avoid spreadsheet numeric conversion.
Check the namespace first. Do not merge local article identifiers, DOIs, PubMed records and author identifiers.
ORCID uses MOD 11-2, ISSN modulus 11, ISBN-13 weighted modulus 10. Check the result without automatically repairing the last character.
Open the reference registry or record and compare title, author, edition and version. Link distinct objects rather than removing different identifiers.
Reading workshop
This fictional exercise separates namespace, received string and described object. DEMO identifiers are not assigned by any real registry.
Fictional data and records for learning: none of these results comes from a live service query.
One catalogue provides linked print and digital editions. Another catalogue reuses the same local string. The decision must retain missing evidence rather than invent equivalence.
If the table exceeds the screen, scroll horizontally; arrow keys work when the region is focused.
| Fictional record | Declared object | What is established | Decision |
|---|---|---|---|
| DEMO-A:000042 | Print edition of the DEMO manual | String retained with leading zeros | Keep this record. |
| DEMO-B:000042 | Record from catalogue B | A different namespace | Correspondence with A is not established. |
| DEMO-A:42 | Another received record | No rule links 42 and 000042 | Do not normalize to a number. |
| DEMO-A:000043 | Digital edition of the DEMO manual | Documented relation to the print edition | Link formats without merging records. |
Open each proposition to compare its diagnosis. More checks may still be needed.
No: equal strings across namespaces are insufficient. Request a documented mapping.
No: no producer rule justifies removing zeros. Retain both received values.
Yes in this fictional case because the relation is supplied. This does not establish a single object to deduplicate.
Origin: DEMO-A and DEMO-B catalogues Received: A:000042; B:000042; A:42; A:000043 Supplied relation: A:000043 = digital format linked to A:000042 A/B correspondence: not established Numeric normalization: not applied
Keep original strings in the record. A relation between editions differs from record identity.
In a fictional case, two coherent ISBN-13s describe two book formats. A shared title does not establish removable duplicates.
An illustrative list links one surname to two ORCIDs. Syntax cannot decide identity; read documented publications and affiliations.
Guided reading of an excerpt
These fictional identifiers are local to the named catalogues. None is presented as a DOI, ORCID or identifier assigned by a real registry.
Fictional exercise: identifiers, values and dates illustrate the check. This excerpt is not a live collected response.
catalogue-A and catalogue-B may assign the same string to different objects.
Keep 000042 as text. Numeric conversion would destroy part of the received form.
Compare object, title, version and the producer’s identification rule before linking records.
catalogue-A | notice | "000042"
catalogue-B | notice | "000042"
catalogue-A | notice | "42"No. That erases namespace and original form and creates an unjustified merge. Retain the received value and separately document any validated correspondence.
Official references : W3C — Data on the Web Best Practices
Worked exercise
Fictional example: the values and documents teach the method and do not describe a real organisation.
The citation gives DOC-A but uses a table from DOC-B. Both identifiers resolve; compare the actual objects retrieved.
If the table exceeds the screen, scroll horizontally; arrow keys work when the region is focused.
| Identifier | Title | Appendix with the table used |
|---|---|---|
| DOC-A | Same title | Absent |
| DOC-B | Same title | Present |
Successful resolution and matching titles are insufficient. The citation combines different objects. Cite DOC-B if that is what was consulted, or remove the table absent from DOC-A. Preserve editions rather than merging records.
Document a published version relation if available; it does not make contents interchangeable.
Keep ID, edition, file, table location and known relation.
Documentation reference : W3C — PROV Model Primer, derivation and revision · Consulted on
A documented match separating correct transcription from verified identity.
No. It checks arithmetic. Use the appropriate registry or record to verify assignment and correspondence.
Not automatically: characters may belong to the suffix. Compare the source and resolver; retain the original before editing.
Keep the form received first. A transformation can facilitate research if its rules are known, but it must be documented and not remove a significant character from the system concerned.
TOOLS / REGISTERS
Open the reference registry or record and compare title, author, edition and version. Link distinct objects rather than removing different identifiers.
DOCUMENTED REFERENCE
The Gallica address ark:/12148/bpt6k83037p/f143 has distinct components. ARK is a persistent identifier system. The domain provides access while the assigning authority and assigned name identify an object within its namespace.
| Reference | Value or form | What it identifies |
|---|---|---|
| ark:/ | ARK | ARK label in the form used by this Gallica example. |
| 12148 | NAAN | Name Assigning Authority Number, or NAAN; neither a date nor a page number. |
| bpt6k83037p | Name | Assigned document name in that namespace; the string alone does not reveal title or year. |
| f143 | Qualifier | Reference to a particular view in this Gallica path; a whole-document citation and a passage citation differ in precision. |
Keep full ARK, actual destination, title, edition or version and consultation date. Persistence relies on identifier management; it does not certify immutable content. An ARK is not a file’s cryptographic hash.
No. Identifier stability and content stability are separate questions. Record version and consulted passage. For file comparison, also preserve a hash and observation date rather than treating the identifier as proof of binary identity.
Documentation consulted on .
DOCUMENTED REFERENCE
The two letters on an intellectual property document do not always identify a country. WIPO Standard ST.3 also covers organisations. These entries were checked against the August 2026 revision and must be read in their field context.
| Reference | Value or form | What it identifies |
|---|---|---|
| ST.3 | 2026-08 | This is the revision date consulted, not the file date. Older archives may use historical codes documented in the annexes. |
| FR / GB | FR / GB | FR refers to France and GB to the United Kingdom. In a priority field, these letters describe the recorded priority, not the applicant’s nationality. |
| EP | EPO | EP identifies the European Patent Office. Do not expand this prefix as a country code or infer identical protection throughout Europe. |
| EM | EUIPO | EM identifies EUIPO for EU marks and designs and relevant designations in international systems. It is not interchangeable with EP or EU. |
| WO / IB | WIPO | WO is used notably for PCT international publication; IB identifies the International Bureau in the capacities specified by the standard. Publication and receiving office are different information. |
Keep the code, field name, complete number and documentary system together. A prefix alone establishes neither grant, current ownership nor the countries where a right remains in force.
No. The code identifies the international publication context. Follow a PCT file through its national or regional phases and their events in the competent registers.
Documentation consulted on .
DOCUMENTED REFERENCE
An ISBN identifies a publishing product: a particular title, edition and format, not every manifestation of a work. The local check follows International ISBN Agency arithmetic to detect transcription errors. It does not query allocation records or identify a book.
| Reference | Value or form | What it identifies |
|---|---|---|
| ISBN-13 | 13 | New ISBNs have thirteen digits since 1 January 2007. Historical ten-character ISBNs use different arithmetic. |
| Prefix | 978 / 979 | The agency currently lists prefixes 978 and 979. An EAN checksum with another prefix is insufficient for ISBN-13. |
| Elements | 5 | Five elements: prefix, registration group, registrant, publication and check digit. Variable lengths rule out fixed-position publisher extraction. |
| Weights | 1, 3, 1, 3… | Weights alternate 1 and 3 from the left across the first twelve digits. |
| Check digit | (10 − (sum mod 10)) mod 10 | The check digit makes the complete weighted sum divisible by ten. A divisible partial sum requires zero, not ten. |
| Example | 978-92-95055-12-4 | The seventh international manual bears this ISBN, also cited in its official Slovene translation. It is the local documentary example. |
| Scope | Title · edition · format | Different languages, editions or product forms may need distinct ISBNs. Different ISBNs do not establish distinct intellectual works. |
Enter all digits, with or without spaces and hyphens. X is allowed only at the end of an ISSN. Arabic and Persian digits are accepted. No network request.
Documented example: ISBN of the seventh international manual. Weighted sum of first twelve digits: 126; check digit: 4.
Removing spaces and hyphens is only for arithmetic. ISBN group boundaries and allocation are not checked. ISBN-10 and ORCID use different rules and are not handled here. A checksum does not detect every error, including some digit transpositions.
Preserve leading zeros and original text. This check does not use allocation ranges to validate group boundaries or publisher existence. Correct length and checksum do not establish assignment; check title, language, date and format in a catalogue.
No. It establishes limited mathematical consistency. Allocation, bibliographic description and availability require a separate catalogue or publisher source; author identity is not confirmed either.
Documentation consulted on .
DOCUMENTED REFERENCE
A UUID contains 128 bits. In the RFC 9562 variant, four bits indicate its version. The table reproduces published examples from Appendices A and B.1, not identifiers assigned here to companies or documents.
| Reference | Value or form | What it identifies |
|---|---|---|
| UUIDv1 | C232AB00-9414-11EC-B3C8-9F6BDECED846 | Time-based version with Gregorian timestamp, sequence and node; the node field does not prove a machine’s hardware address. |
| UUIDv3 | 5df41881-3aed-3515-88a7-2f4a814cf09e | Deterministic namespace-and-name computation using MD5, not an object signature. |
| UUIDv4 | 919108f7-52d1-4320-9bac-f847db4148a8 | Random or pseudorandom version with 122 bits outside version and variant. Quality depends on the generator. |
| UUIDv5 | 2ed6657d-e927-568b-95e1-2665a8aea6a2 | Namespace-and-name computation using SHA-1; stable results also require identical name bytes. |
| UUIDv6 | 1EC9414C-232A-6B00-B3C8-9F6BDECED846 | Rearranges version 1 timestamp fields for ordering. It is not version 7’s Unix timestamp. |
| UUIDv7 | 017F22E2-79B0-7CC3-98C4-DC0C0C07398F | The first 48 bits encode Unix milliseconds. This example contains 1645557742000, or 2022-02-22T19:22:22.000Z. |
| UUIDv8 | 2489E9AD-2EE2-8E00-8EC9-32D5F69181C0 | Custom layouts. Example B.1 does not define a universal time layout for every version 8 UUID. |
Local reading of hexadecimal 8-4-4-4-12 notation. Any timestamp is encoded by the producer, not proof of creation. Selector examples come from RFC 9562.
Field reading updated.
Correct syntax proves neither effective uniqueness, authenticity nor the associated object’s creation time. The reader decodes time only for version 7 with the RFC variant; Nil and Max are handled separately. It generates no identifiers or documents.
Documentation consulted on .
DOCUMENTED REFERENCE
An ORCID retains leading zeros and encodes neither country, institution nor discipline. Its last position checks transcription. The examples below are published by ORCID; they are not identities that a reader may claim.
| Reference | Value or form | What it identifies |
|---|---|---|
| body | 15 × 0–9 | The first fifteen digits feed MOD 11-2 in order. Hyphens separate four groups of four characters and do not enter the calculation. |
| MOD 11-2 | t ← (t + d) × 2 | Start at zero, then double the sum of the current total and each digit. The check value is (12 − (total modulo 11)) modulo 11. |
| X | 10 · 0000-0002-1694-233X | The value 10 is written as a final X. The official example ending in 233X tests this case without replacing X with a numeral. |
| URI | https://orcid.org/0000-0002-1825-0097 | The full HTTPS URI with hyphens retains registry context. Preserve zeros on import and avoid a numeric spreadsheet column. |
| scope | arithmetic ≠ registry ≠ authentication | A consistent calculation proves neither assignment by ORCID, control of the account nor the source of an affiliation. Consult the record, then distinguish manual entry from an authenticated connection. |
Short code or HTTPS URL on the exact orcid.org domain, without query or fragment. 16 characters, in four groups or without hyphens. Final x is accepted and converted to X; only ordinary outer spaces, tabs and line breaks are removed.
Format and arithmetic check are consistent. Assignment and identity not checked.
MOD 11-2 t = 1314 1314 mod 11 = 5 (12 − 5) mod 11 = 7
Local calculation: no registry request. The source’s published example remains readable without JavaScript.
This reader does not query the registry or verify allocation blocks. It does not create an ORCID or automatically correct a wrong check digit. People sharing a name must not be matched solely on the arithmetic consistency of a code.
Documentation consulted on .
DOCUMENTED REFERENCE
A ROR identifier is a random string with no information about the institution or its relationships. Its preferred form is a full URL. The documentation uses 02mhbdp94 as an example; this reader exposes the conversion before checking the digits.
| Reference | Value or form | What it identifies |
|---|---|---|
| pattern | 0 + 6 × base32 + 2 × 0–9 | The short form has nine characters: an initial zero, six Crockford base32 characters and two check digits. The initial zero remains in the URL. |
| alphabet | 0123456789abcdefghjkmnpqrstvwxyz | The alphabet includes 0–9 and the letters shown, excluding i, l, o and u. Do not map these excluded letters to similar-looking digits on import. |
| 2mhbdp | 88649142 | Decoding 2mhbdp in base32 for the official example gives the integer 88,649,142. That integer reveals no name, country or institutional ranking. |
| check | 98 − ((n × 100) mod 97) = 94 | The generator published by ROR calculates 98 − ((n × 100) modulo 97), as two digits. Here the remainder is 4 and the check pair is 94. |
| record | https://ror.org/02mhbdp94 | Code consistency comes before reading the record: check names, name types, status and documented relationships. Do not predict a subsidiary’s ROR by modifying its parent’s. |
Short code or HTTPS URL on the exact ror.org domain, without query or fragment. Nine characters; ASCII case converted to lowercase with no substitution for i, l, o, u. Only ordinary outer spaces, tabs and line breaks are removed.
Format and arithmetic check are consistent. Assignment and identity not checked.
base32(2mhbdp) = 88649142 8864914200 mod 97 = 4 98 − 4 = 94
Local calculation: no registry request. The source’s published example remains readable without JavaScript.
The tool runs locally and does not check presence in ROR. It rejects excluded letters and internal separators rather than silently repairing them. A normalised URL represents the entered code, not confirmation of existence or affiliation.
Documentation consulted on .
DOCUMENTED REFERENCE
A LEI may contain twenty characters yet fail the transcription check. The XBRL taxonomy also distinguishes an arithmetic error from a warning about certain endings. The example uses the LEI displayed by GLEIF for the foundation itself without inferring its current status.
| Reference | Value or form | What it identifies |
|---|---|---|
| pattern | 18 × A–Z / 0–9 + 2 × 0–9 | XBRL’s format check expects eighteen A–Z or 0–9 characters followed by two digits. This step does not query the registry. |
| A → Z | 10 → 35 | For arithmetic checking, replace A with 10, B with 11 through Z with 35, then concatenate digits without moving the LEI characters. |
| check | expanded(full LEI) mod 97 = 1 | The expanded decimal number for the complete code must leave remainder 1 modulo 97. Calculate digit by digit without storing this large integer in a floating-point number. |
| body + 00 | 98 − (expanded(body) × 100 mod 97) | For the eighteen-character body, append 00 after conversion, then calculate 98 minus the modulo-97 remainder. This is a calculated pair for comparison, not permission to change a published LEI. |
| warning | 00 / 01 / 99 | XBRL flags 00, 01 and 99 separately: some issued LEIs use these endings due to a historical ambiguity. If the remainder is 1, the tool warns and requests registry verification without declaring them nonexistent. |
20 characters without separators; ASCII case converted to uppercase. Last two characters must be digits. Only ordinary outer spaces, tabs and line breaks are removed.
Format and arithmetic check are consistent. Assignment and identity not checked.
A = 10 … Z = 35 5067001614116293252633363 mod 97 = 1 98 − (5067001614116293252633300 mod 97) = 63
Local calculation: no registry request. The source’s published example remains readable without JavaScript.
Even remainder 1 proves neither assignment, current registration validity nor entity activity. Check the GLEIF record and date the lookup. This reader does not check other allocation rules and does not require positions 5–6 to be 00.
Documentation consulted on .
DEEPER READING / PRACTICE
A file collects report numbers, DOI, opinion references and other identifiers. The control must find the right object and distinguish valid identifier and accurate correspondence.
Updated on .
Preserve the original form, identify the system used and examine the resolution result. Compare title, organization, date and version with the expected document.
A small mistake can lead to another real object. Normalizing an identifier without knowing its rules can also remove a significant character.
The register retains the original, form of search, document found and concordance. Unresolved identifiers remain visible for subsequent verification.