Methodology
How the atlas keeps itself honest
The site's credibility comes from method, not from who runs it. This page sets out how evidence is labelled, how contested views are handled, and how anyone can check or challenge an entry.
Purpose
yhwh.cc documents the evidence for how the divine name was written, transcribed, used in other names and rendered in translation. It also documents the pronunciations proposed on that evidence. It is an evidence atlas, not an argument for any one reading, and not a theological resource.
Evidence tiers
Every record carries exactly one of five tiers. The tier answers a single question: what kind of evidence is this? It does not answer whether something is true.
Read directly on an ancient object or in a critical edition of one. Says what is written, not how it was pronounced.
Preserved through later copying, reading traditions, or an author's report of other people's usage.
An inference or synthesis published by named scholars. Attributed, never presented as a reading.
A proposed form or claim with no attested witness yet. Documented under the same rules as every other entry.
A reading, identification or manuscript status that is contested, unverified, or awaits direct collation against an edition or image.
Three rules
The corpus is built on one provenance principle, quoted from its metadata:
Consonantal spelling ≠ phonetic reading; name element ≠ full theonym; medieval pointing ≠ automatic ancient vowels.
- A consonantal spelling is not a phonetic reading. יהוה on a stele tells us the letters, not the vowels.
- A name element is not the full theonym. -yahu in a personal name shows how the element was used. It does not show how the full name was pronounced.
- Medieval pointing is not automatically ancient vowels. Masoretic vowel signs on the name encode a reading convention.
How statuses map to tiers
Each table records its own detailed status (for example secure_consonantal or collation_needed). A published, version-controlled vocabulary maps every status to a tier. The build fails if any record uses a status that is not in the vocabulary.
artifacts.evidence_status
attestations.status
forms.status
textual_witnesses.witness_status
pronunciation_candidates.classification
claims.epistemic_status
theophoric_names.witness_kind
renderings.verification
Neutrality
Two kinds of neutrality can conflict. Procedural neutrality means every hypothesis is documented with the same template, judged against the same evidence rules, and reviewed through the same process. Due weight means readers should not be misled about how scholarly opinion is actually distributed. Presenting a view held by a few as equal to a broad consensus is a form of bias, sometimes called false balance.
yhwh.cc applies both:
- Equal rules. All pronunciation candidates share one page template (evidence offered, limitations, what would test it) and appear alphabetically by default, in lists, in the matrix and in navigation.
- Attributed weight. The distribution of scholarly support is reported as sourced, dated claims. For example, “Yahweh is the dominant modern academic reconstruction” is claim C08, with its sources and caveat. The site does not express weight through ordering, size, colour or prominence.
- No site verdict. There is no score and no “most likely” label. The tier describes the evidence type, not the candidate's merit.
- No popularity contests. Votes, likes or reaction counts never rank hypotheses.
- Claims, not people. The same author can publish advocacy and peer-reviewed work. Each contribution is classified on its own terms.
Editorial policy
Every submission, from an editor or a first-time contributor, must include:
- A source. A primary witness (with edition, museum number or manuscript siglum) or a published scholarly reference. Links to images are welcome. Re-hosted images without rights are not.
- A claimed tier. What kind of evidence the submitter believes this is. Reviewers may change it, with a written reason.
- A stated limitation. What the evidence cannot show. A reconstruction submitted without a stated limitation is returned.
- A disclosure of any affiliation or interest (see below).
Review states
Data and hypotheses move through different states, on purpose:
- Data (a reading, a date, a locus, a citation): Submitted → Needs review → Verified reading. Verification means a reviewer checked it against the cited edition or image.
- Hypotheses (a pronunciation, an etymology, an identification): Submitted → Needs review → Documented. A hypothesis can be documented accurately. The site never marks it “verified”.
All review happens in public on GitHub. Every change to the corpus is a versioned commit, and every release carries a version number you can cite.
Conflicts of interest
Contributors disclose ministry, commercial, institutional or advocacy interests relevant to a submission. Disclosure does not disqualify anyone; it informs readers and reviewers. Editors are held to the same rule. An editor who has proposed or publicly advocated a hypothesis recuses from reviewing submissions about it, and another reviewer handles them. Editor-proposed hypotheses are classified exactly like anyone else's.
Disputes & corrections
Anyone can challenge any entry with a sourced correction. Disputes are resolved in writing, in the open, with the reasoning archived alongside the change. When a dispute cannot be resolved, both positions are recorded as attributed claims and the record is marked Disputed .
Conduct
- Discuss evidence and arguments, not people or their faith.
- No proselytising, polemics or devotional appeals in records or threads. Theology is welcome as a subject of study, not as a way to settle a reading.
- Moderation applies to conduct, never to which conclusion someone holds.
- Respect readers who do not pronounce or write the name.
Respect for practice
Many readers, particularly in Jewish tradition, do not pronounce the name, and some do not write it out in full. The display settings (⚙ in the header) include an option to veil vocalized forms: spelled-out pronunciations are blurred until tapped, while consonantal spellings and transcriptions in their original scripts stay visible. The site has no audio of the name.
Images & rights
The site links to museum and library images rather than re-hosting them. Script specimens are rendered in Unicode fonts and labelled as schematic: they show which letters are written, not what the object looks like. Linked publications and museum assets keep their own rights. The corpus annotations are published under CC BY 4.0.
Known limitations
- This is a starter corpus (v1.1.0). It is not an exhaustive critical edition, and the confidence values are not validated probabilities.
- Several attestations rely on catalogue summaries or secondary reports and are flagged for direct collation.
- Site coordinates are approximate editorial values, each with a precision note.
- The renderings table is an editorial seed; most rows still need a cited source.
- Theophoric name glosses are conventional interpretations; debated glosses are marked.