Moria
For every one of the 31,102 verses, Moria has already worked out which other texts are related to it, by five methods that do not depend on one another — and it shows you, verse by verse, what each link is actually made of.
A note on language. Moria was written in Dutch and built on the Statenvertaling; opened from these pages it speaks the language of the page — English here, Dutch or Spanish there — and starts in a reading text to match. The analyses themselves were computed on the Dutch text and the King James Version, so switching the reading translation changes what you read, never the scores.
What is in it
The text, in four translations and two original languages. The Statenvertaling in the extended edition that carries Strong’s numbers on every Dutch word and the original marginal notes; the King James Version, the Berean Standard Bible and the Reina-Valera 1909; the Hebrew of the Westminster Leningrad Codex and the Greek of MorphGNT/SBLGNT, both fully parsed — 444,000 words with lemma, part of speech and morphology.
The Septuagint, aligned. Swete’s edition of the Greek Old Testament, set against the ordinary verse numbering. That alignment is not a hand-written table: it is derived from the text itself, because proper names survive translation — Ἰσραήλ next to Yisraél — and a monotone alignment per book follows from name agreement and verse length. On a set of 34 known checkpoints it lands correctly in 33.
A map, a synopsis and a concordance. Biblical places with coordinates from Tyndale House’s TIPNR over a simplified Natural Earth coastline; the four gospels side by side, each verse against its nearest counterpart in the other three; searchable morphology, so all qal imperatives of שׁמע is a query rather than a project.
A dossier you own. Verses collected while reading, each with a note, saved as a small JSON file next to your work — deliberately a file and not browser storage, so nothing disappears when a browser tidies up. It exports to Word and Markdown.
The five methods, and what each one can and cannot show
Most study Bibles hand you a column of cross-references and no account of where they came from. Moria does the opposite: it records five kinds of link for every verse and shows, for each, what that link rests on. The five are deliberately different from one another: where several of them fall on the same pair of texts independently, that agreement means something. Where only one does, the kind of evidence matters, and it is named.
| Method | What it rests on | What it finds |
|---|---|---|
| Cross-references | The openbible.info dataset (CC-BY): about 700,000 directed pairs with vote scores, drawn largely from the Treasury of Scripture Knowledge. A tradition of interpretation, not a measurement of the text — and the only method that shows no word-level evidence, because there is none to show. | The classical references, weighted by how many readers have affirmed them. |
| Original-language overlap | Lemmas and Strong's numbers from the Westminster Leningrad Codex and MorphGNT/SBLGNT; tf-idf over single words and word pairs. Cannot reach between Hebrew and Greek; that is what the Septuagint bridge is for. | Parallel passages and verbatim borrowing inside one language — 2 Kings 18 against Isaiah 36. |
| Quotation via the Septuagint | Every New Testament verse compared against the Greek Old Testament (Swete) on word forms, word pairs and rare word chains. Silent where the New Testament follows a different Greek text, or the Hebrew. | Old Testament quotations in the New at the level of the original — with no translation in between. |
| Quotation via the translation | Rare four-word content phrases shared between verses, weighted by how rare they are. Runs through a translation, so its wording partly decides what turns up. | Quotations and allusions across the canon, including where the Greek diverges. |
| Thematic kinship | Latent semantic analysis in 300 dimensions over two translations at once, mixed with direct word overlap. Statistical, not exegetical. Judge every result yourself. | Texts about the same thing that do not share the same words. |
The point of separating them is that they fail differently. A statistical method drifts toward vocabulary; a traditional reference list carries the assumptions of its compilers; a quotation search finds only what was quoted. Read together, and with their evidence visible, they let you see which kind of claim you are looking at. How each method works in detail →
What makes it different
Every link comes with its evidence, recomputed live between the two verses using exactly the same word segmentation and stop-word lists as the analysis itself — otherwise the highlighting would show something other than what caused the score.
For a quotation through the Septuagint, that evidence is the shared Greek chain: at Psalm 110:1 and Hebrews 1:13, Κάθου … δεξιῶν … ἕως … ἐχθρούς … ὑποπόδιον … ποδῶν. For original-language overlap it is the shared lemmas, ordered by rarity so the distinguishing word comes first and not the article. For thematic kinship it is the shared content words, marked in the text.
And for the classical cross-references there is nothing — which is the point. Those are judgements handed down by readers, not measurements of the text. They are not worth less for that, but they rest on something different, and that difference ought to be visible.
No server, no account, no tracking. Moria is one HTML file. Everything — all four translations, both original languages, the Septuagint, the precomputed analyses, the map — is inside it. Once it has loaded, you can disconnect. Save the file and it will still work in ten years.