The first sentence of the Bible has a hinge in its first word. Read one way it stands alone — “In the beginning, God created the heavens and the earth,” a first instant, a making from nothing. Read another way it does not stand alone at all: “When God began to create…,” a clock-setting clause, and when it stops ticking the dark, the deep, and the waters are already there. One reading is the most familiar sentence in English. The other is what the grammar, left to itself, quietly prefers.
A reader who knows no Hebrew has still, almost certainly, met this sentence — it may be the single most-translated line of text in human history. And almost everyone has met it in one grammar only: In the beginning, God created. A full stop you can hear. A universe starting from a clean page. But the Hebrew does not obviously say that, and the people who fixed the Hebrew text — adding the vowels and the chant-marks a thousand years after it was written — left a set of fingerprints that point the other way.
The whole question lives in the first word, בְּרֵאשִׁית (bereshit). Everything below takes it apart: first the word itself and the one vowel that tips it; then the same word everywhere else in the Bible, where it never once means what English makes it mean here; then ten public-domain translators, every word byte-checked, laid side by side; and last the ladder down which the familiar reading actually reached us — through the Greek and the Latin, very early, and against the grain of the grammar.
בְּרֵאשִׁית is two pieces: the prefix בְּ (be-, “in”) and the noun רֵאשִׁית (reshit, “beginning, first, chief part” — from rosh, a head). The hinge is the vowel under that first letter. The text points it be-: a bare in, with no definite article. Had the writers of the vowels meant the absolute “in the beginning,” Hebrew has a way to say exactly that — the article is swallowed into the prefix and the vowel turns to ba-. They didn’t write it. Tap the word; then toggle the vowel the article would have forced.
The contrast is real and it is the heart of the matter: the Masoretes (7th–10th c. CE) wrote be-, not ba-. An anarthrous be-reshit reads most naturally as bound — “in the beginning of…”. The honest caveat: a temporal phrase can drop the article and still be read absolute (“in beginning”), so the missing “the” doesn’t settle it alone. It is the first of three witnesses, not a verdict.
Here is the second witness, and it is the strong one. The word רֵאשִׁית appears across the Hebrew Bible, and almost every time it is construct — glued to the noun behind it, meaning “the beginning of X,” “the first of X.” Below are four of those places. The killer is the first: בְּרֵאשִׁית opens the book of Jeremiah too — the same word, pointed the same way — and there nobody has ever read it absolute. It plainly means “in the beginning of the reign of Jehoiakim.” The English under each is the King James Version’s own — the very translators who, at Genesis 1:1, wrote the absolute “In the beginning.”
Rashi’s Hebrew is quoted verbatim (Sefaria; an 11th-century text, public domain); the English is this page’s crib, marked as such. He reaches for the same parallel verses, eight centuries before the modern grammars did — and names Jeremiah, Genesis 10, and Deuteronomy 18, the very ones shown above.
The Masoretes left two fingerprints, and they point opposite ways.
So is it settled? No — and the most honest part of the story is why. The same medieval scribes who pointed the vowel be- (which leans construct) also added the cantillation, the chant-marks — and the mark they put on בְּרֵאשִׁית is a disjunctive (a tipḥa), the kind that separates a word from the one after it. Read by the vowels, they bound the word forward; read by the accents, they cut it loose to stand alone. The medieval grammarian Ibn Ezra used exactly that accent to argue against Rashi, for the absolute reading. The tradition that fixed the text was itself of two minds — which is the truest thing anyone can say about this verse, and the reason it is still live.
Now watch the consequence. Below are Genesis 1:1–2 in ten public-domain English translations across six centuries — Wycliffe’s Middle English of 1395 to the public-domain modern WEB — each transcribed from a primary digitization and byte-checked. Pick one of the three decisions along the top to see how every hand takes it. Nine of the ten read bereshit absolute; one, Young’s relentlessly literal 1898, reads it construct and stands alone — the way one Bohn-prose hand stood alone in this venue’s Catullus.
The three places the Hebrew forces a choice: the grammar (bereshit — absolute “In the beginning,” or construct “In the beginning of”) · the chaos (tohu va-vohu — the earth’s first state, which the Greek read as “invisible and unformed”) · the spirit (ruaḥ elohim — “the Spirit of God,” or a “wind of God,” or “a mighty wind”).
Each line verbatim from the digitization named in the apparatus,
checked by research/genesis-1-1/verify.mjs. The badge marks how each hand
reads bereshit: absolute
or dependent. Three modern construct readings
still in copyright — the 1985 JPS, Robert Alter (2004), the NRSV’s footnote — are
named here, not quoted.
The chaos. Toggle it and watch the earth’s first state wander: without form, and void (King James), waste and void (the Revised tradition), unformed and void (the 1917 Jewish version), Tyndale’s blunt voyde and emptie, Wycliffe’s idel and voide — and, most strikingly, the Greek’s unsightly and unfurnished, where the Septuagint quietly turned a primeval murk into something closer to a room not yet decorated. (Robert Alter’s copyright welter and waste, named not quoted, is the one modern hand that hears the Hebrew’s rhyme, tohu va-vohu.)
The spirit, and the wind. Here the old gallery is unanimous and the dissent is entirely modern. Every one of the ten public-domain hands writes the Spirit of God (Wycliffe, alone, writes “the Spirit of the Lord”) moving, hovering, fluttering, borne over the waters. But the Hebrew ruaḥ is also the plain word for wind and breath, and elohim can act as an intensifier — so the same three words can mean a mighty wind, with no spirit in them at all. The modern translations that read the grammar construct tend to read the wind here too: a wind sweeping a dark sea that already exists. Not every crux splits the old hands; this one splits the centuries.
The familiar “In the beginning.” is not a mistake — it is ancient, older than the vowel-points that lean against it. The decision was taken once, very early, by the Greek translators, and the whole West inherited it down this ladder.
One grammatical choice, and the theology of the first verse changes shape. Read absolute, verse 1 is the first event: God makes heaven and earth, and then verse 2 describes what he has just made. This is the reading that has carried the doctrine of creatio ex nihilo — creation out of nothing — and one English hand, Wycliffe in 1395, even wrote the doctrine straight into the line: “God made of nouyt heuene and erthe” — made of nought, words that stand in neither the Hebrew nor the Latin he was turning.
Read construct, verse 1 is a clock, not an act: when God began to create — and the main sentence is verse 2 or verse 3, where the tohu va-vohu, the darkness, the deep (tehom), and the waters are already present for the spirit-or-wind to move over. On this reading the verse does not describe a beginning from nothing at all; it describes God bringing order to a chaos that is simply, unaccountably, there.
The honest end of it: text-criticism cannot decide the doctrine, and does not try to. Creation from nothing is argued elsewhere too, on other verses and on later theology, and it does not stand or fall on this one grammar. What the grammar shows is narrower and stranger — that the most familiar sentence in the language rests on a choice its own scribes were divided about, and that the less-familiar reading is the one the bare words, left alone, drift toward. The venue’s habit is only ever to lay the readings side by side and let the disagreement be the evidence. Here the disagreement is as old as the Greek and as live as this morning’s study Bible, and it is hiding in a single missing vowel.
Everything quoted here is re-derived offline by
research/genesis-1-1/verify.mjs (run node it from a fresh
checkout; exit 0 = all pass). The texts are first fetched, never recalled, by
research/genesis-1-1/fetch-sources.mjs, which writes
sources.json from the primary digitizations below; the verifier then asserts
that this page reproduces the Hebrew of Genesis 1:1–3, the four construct parallels and
their KJV renderings, Rashi’s Hebrew, the Greek and the Latin, and all ten English
translations — each verbatim, byte-for-byte. Full provenance and the honest edges are in
research/genesis-1-1/SOURCES.md.
Hebrew — Sefaria, Miqra according to the Masorah (the Masoretic / Leningrad tradition), Genesis 1:1–3, Jeremiah 27:1, Genesis 10:10, Deuteronomy 18:4, Genesis 49:3.
Rashi on Genesis 1:1 §2 — Sefaria (Rashi, c. 1075–1105; public domain).
Septuagint Greek — getBible v2, “OT LXX Accented,” Genesis 1:1–2.
John 1:1 — getBible v2, Textus Receptus (εν αρχη ην ο λογος και ο λογος ην προς τον θεον και θεος ην ο λογος).
Latin — bible-api.com, Clementine Vulgate, Genesis 1:1–2.
Wycliffe (c.1395), Tyndale (Pentateuch, 1530), Douay (1609), Young’s Literal (1898), American Standard (1901) — getBible v2 digitizations.
King James (1611), World English Bible (public domain), Darby (1890, English) — bible-api.com.
JPS 1917 (public domain) — Sefaria, The Holy Scriptures: A New Translation (1917).
Brenton’s Septuagint in English (1851) — eBible.org, eng-Brenton.
The construct parallels’ English is the King James Version (bible-api.com), chosen because it is the same hand that reads Genesis 1:1 absolute.
The grammar of bereshit is genuinely unsettled; this page argues that the construct reading is grammatically the stronger, not that it is proven, and it shows the Masoretic accent that cuts the other way. The be-/ba- contrast and the בָּ form shown under the toggle are standard grammarians’ illustration, not a manuscript variant — Genesis 1:1 is pointed be- in every witness. Historic English spellings are reproduced as the named digitization records them (Tyndale’s ad for and; Wycliffe’s nouyt); the wording each rests on is the part the page reads. Per-word glosses, the English crib of Rashi, and the morphological labels are this page’s editorial apparatus and are marked as such; the Hebrew, Greek, Latin and the ten English renderings are quotation. All source texts were fetched directly from the digitizations named — none supplied from memory.