How the books came to count as Scripture — the lists, the arguments, the books that almost made it, and the edges that never quite closed
This study traces how the biblical canon formed, from the settling of the Hebrew Scriptures through the four centuries in which the church received, disputed, and listed the New Testament books. It works from the primary record — the early lists, the councils, the surviving manuscripts, and the books that were argued over — and ends with the question the history cannot answer.
Every Bible on a shelf makes a silent claim before a word of it is read: these books, and not others. The claim raises a question that deserves a plain answer. Who drew the line? When, and on what grounds? And why does the line sit in different places for a Protestant, a Catholic, an Orthodox, and an Ethiopian believer, all of whom call their book the Bible?
The word itself is a piece of the answer. "Canon" is rule, measuring rod (κανών | kanōn | ) — a straight reed used for measuring, and from there a standard, and from there a list of what meets the standard. Paul uses the word for a rule of conduct (Gal 6:16). Applied to books, it carries both senses at once: the books that measure the faith, and the list that measures the books.17
This study is written for a site with a stated confession, and the writer believes the books in question are the word of God. That belief is exactly why the history deserves to be told straight: a canon whose formation cannot survive honest description would not be worth confessing. What follows is the process as the record preserves it — slower, messier, and more argued-over than most church introductions suggest, and better documented than most skeptical summaries allow.
The Hebrew Bible describes some of its own contents being recognized in real time. A "book of the law" found in the temple under Josiah is read to the king, who tears his clothes and binds the nation to it (2 Kgs 22–23). Ezra reads "the book of the law of Moses" to the restored community "from early morning until midday," and the people weep (Neh 8:1–9). Whatever else canon means, this is its raw material: a community treating a written text as binding on itself.
By the second century BC the shape is visible. The grandson of Ben Sira, translating his grandfather's book into Greek around 132 BC, refers three times in his short prologue to "the Law and the Prophets and the other books of our ancestors" — a three-part collection, named as a known thing, with the third shelf still loosely titled.1 That matches the structure the Hebrew Bible carries to this day: Torah (Law), Nevi'im (Prophets), Ketuvim (Writings) — the Tanakh, from the three initials. The Law settled first, the Prophets second, the Writings last; the prologue's vague "other books" is the third shelf photographed while still being arranged.
Jesus, in Luke's Gospel, names the same three-part shape:
"Then he said to them, 'These are my words that I spoke to you while I was still with you — that everything written about me in the law of Moses, the prophets, and the psalms must be fulfilled.'"
Luke 24:44 · NRSVue
"The psalms" here is most naturally the Writings, named for their largest book. By the end of the first century AD, two Jewish witnesses count the collection: Josephus defends "twenty-two books" that Jews hold sacred, claiming no one has dared add, remove, or alter them2; the apocalypse called 4 Ezra, written about the same time, counts twenty-four public books alongside seventy reserved ones.3 The two numbers are most simply read as the same list differently grouped (Ruth folded into Judges, Lamentations into Jeremiah, and similar combinations), and both match the books in the Hebrew Bible now — which are the same books, differently divided, as the Protestant Old Testament's thirty-nine.
The Dead Sea Scrolls put physical evidence under the claim: the Qumran library preserves copies of every Hebrew Bible book except Esther. It also preserves Tobit, Jubilees, and 1 Enoch in quantity — a reminder that a first-century Jewish bookshelf was broader than the collection later counted sacred, and that owning a book and canonizing it are different acts.
Settled shape is not the same as closed list. The Mishnah records rabbis still debating, into the second century AD, whether Ecclesiastes and the Song of Songs "defile the hands" — the rabbinic category for sacred books — and Esther's status drew discussion later still.4 On the Christian side, the earliest surviving Old Testament list from a church writer — Melito of Sardis, around AD 170 — omits Esther altogether.5 These were arguments about books already on the shelf rather than a gate deciding what got in, but they are arguments, preserved in the record, at dates later than a tidy account would like. No decree closed the Hebrew canon; use and consensus closed it, and the last edges stayed warm for centuries.
The New Testament books were written across the second half of the first century, and the churches used them from the start — read in worship, copied, exchanged. Two passages inside the New Testament show the collection already becoming conscious of itself. Paul instructs one congregation to swap letters with another: "when this letter has been read among you, have it read also in the church of the Laodiceans, and see that you read also the letter from Laodicea" (Col 4:16) — letters written for one city already circulating as church property at large, and one of them, the Laodicean letter, now lost. And 2 Peter, likely the latest book in the collection, speaks of Paul's letters as a known corpus that can be misread the way Scripture is misread:
"So also our beloved brother Paul wrote to you according to the wisdom given him, speaking of this as he does in all his letters. There are some things in them hard to understand, which the ignorant and unstable twist to their own destruction, as they do the other scriptures."
2 Peter 3:15–16 · NRSVue
"As they do the other scriptures": inside the New Testament itself, Paul's letters are already being shelved with the sacred books. What did not yet exist was a boundary. The first person known to have drawn one was not a council or a bishop of the consensus — it was Marcion of Sinope, around AD 144. Marcion taught that the God of the Hebrew Scriptures was a lesser deity, not the Father of Jesus Christ, and he built a Bible to match: one edited Gospel (a shortened Luke) and ten letters of Paul, with the Old Testament rejected entire. The church at Rome expelled him; his movement outlived him by centuries.
Marcion's role in the story is debated in an instructive way. An older scholarly view made him the inventor of the very idea of a Christian canon, with the church's list a defensive copy of his; most scholarship now reads him as an accelerant rather than an originator — the four Gospels and the Pauline corpus were already functioning collections before him.6 Either way, the effect is not in dispute. A rival list forces the question a working consensus can defer: which books, exactly, and why these?
The churches never adopted a formula. What the record shows instead is a set of working tests, visible in how actual cases were argued: apostolicity — does the book come from an apostle or an apostle's circle; antiquity — does it date from the apostolic age; catholicity — is it received and read across the churches, not in one region or faction; and conformity to the rule of faith — does it square with the teaching the churches already carried. The tests are best seen in operation.
Around AD 200, Serapion, bishop of Antioch, found the church at Rhossus reading a "Gospel of Peter" in worship. He first allowed it, unread; then he read it, found teaching in it that cut against the reality of Christ's body, and revoked the permission — remarking, as Eusebius preserves the letter, that the churches receive the apostles, but reject writings that falsely circulate under their names.7 The case shows the tests interlocking: the book failed the rule of faith, and its apostolic name could not save it, because the name was judged part of the forgery.
The Shepherd of Hermas — a book of visions and moral teaching from second-century Rome — was among the most read and copied Christian texts of its era. The Muratorian Fragment, the earliest surviving canon list (traditionally dated to the late second century, though some scholars argue for the fourth8), draws the line with unusual clarity: the Shepherd may be read privately, but not in public worship among the prophets or apostles, because it was written "very recently, in our times."9 Not heretical; loved; too late. Antiquity was a real test, and it excluded a popular book.
Hebrews names no author. The East received it early as Pauline; the West doubted it for two centuries precisely because the attribution would not hold still. Origen, the third century's greatest textual scholar, reviewed the options — the thoughts are Paul's, the style is someone else's — and ended with the sentence that became the classic verdict: who wrote the epistle, God knows.10 Hebrews entered the canon anyway, carried by its antiquity, its content, and eventually the West's deference to the East's usage. The case matters because it shows the tests were weighed rather than mechanically applied: a book with an unanswerable authorship question was received on the strength of everything else.
Notice what all four tests have in common: they are backward-looking. None of them asks "is this book inspiring?" or "does it teach what we wish were true?" They ask whether the book comes from the apostolic deposit — the right people, the right period, the whole church's usage, the received teaching. The churches understood themselves to be identifying an inheritance, not curating an anthology. Whether that self-understanding was accurate is a separate question, taken up at the end; that it was the self-understanding is what the cases show.
Writing around 325, the church historian Eusebius took stock of where the argument stood. His ledger has four bins, and the honesty of the middle two is the most valuable thing in it11:
| Category | Contents (abridged) |
|---|---|
| Acknowledged (homologoumena) | The four Gospels, Acts, the letters of Paul, 1 John, 1 Peter — and Revelation, "if it really seem proper," a hedge Eusebius could not resist repeating. |
| Disputed (antilegomena) | James, Jude, 2 Peter, 2 and 3 John — "spoken against," yet "known to most." |
| Spurious (notha) | The Shepherd of Hermas, the Epistle of Barnabas, the Didache, the Apocalypse of Peter — books the churches used but did not receive as Scripture. |
| Heretical forgeries | Gospels under the names of Peter, Thomas, and Matthias; acts under the names of Andrew and John — "to be cast aside as absurd and impious." |
Forty-two years later, the first list that matches a modern New Testament page for page. In his Easter letter of AD 367, Athanasius of Alexandria named twenty-seven books — the twenty-seven — and wrote of them: "in these alone is proclaimed the doctrine of godliness. Let no man add to these, neither let him take ought from these."12 Synods at Hippo (393) and Carthage (397) ratified the same list for the West. No ecumenical council of the early church ever ruled on the canon; the lists that survive are bishops and synods writing down what their churches had come to hold.
Two details in Athanasius's letter keep it honest history rather than a triumph scene. His Old Testament list omits Esther — the book's old instability visible even here, in the very letter that fixes the New Testament.12 And after the canon he appends a second shelf: books "not included in the canon, but appointed to be read" to newcomers — Wisdom, Sirach, Esther, Judith, Tobit, the Didache, the Shepherd. Esther, missing from his canon, lands one shelf down in the same letter. The church that drew the line kept the books just outside it in the room.
Lists are one witness; bindings are another. The two oldest complete Christian Bibles that survive were both copied within living memory of Athanasius's letter, and neither matches it exactly. Codex Sinaiticus (fourth century) ends its New Testament with the Epistle of Barnabas and the Shepherd of Hermas. Codex Alexandrinus (fifth century) appends 1 and 2 Clement. Whoever commissioned those volumes bound the disputed shelf into the book — which is roughly what a library does with books it values and has not quite decided about. The canon on paper and the canon in bindings converged over the following century; the gap between them is the settling process caught in the act.
Ask "how many books are in the Bible?" and the answer depends on where the question is asked. The differences are old, documented, and almost entirely about the Old Testament's Greek shelf.
The Septuagint — the Greek Old Testament the early church carried into the world — included books with no place in the Hebrew canon: Tobit, Judith, Wisdom, Sirach, Baruch, 1 and 2 Maccabees, with expansions of Daniel and Esther. The churches read them for centuries. Jerome, translating the Vulgate around 400, pressed the distinction: as the church reads Judith, Tobit, and the books of Maccabees "but does not admit them among the canonical Scriptures," so it may read these books "for the edification of the people, not to give authority to doctrines of the Church."13 The Reformation took Jerome's side; Luther's 1522 German Bible grouped these books under the heading "Apocrypha: books not held equal to Holy Scripture, yet useful and good to read." Rome answered at the Council of Trent in 1546 by defining them as canonical — the "deuterocanon," the second-listed. The Orthodox churches receive a wider Greek shelf still, and the Ethiopian Orthodox Tewahedo Church holds the widest canon in Christendom, with 1 Enoch and Jubilees inside it — which means the book Jude quotes as prophecy (Jude 14–15, citing 1 Enoch 1:9) is Scripture in Addis Ababa and background reading nearly everywhere else.
Luther's New Testament carried its own quieter edge. His 1522 edition moved Hebrews, James, Jude, and Revelation to the back, unnumbered in the table of contents, and his preface famously called James "really an epistle of straw" next to John and Romans — a line dropped from editions after 1537, and a judgment no Lutheran church body adopted.14 The episode is worth keeping in view for what it shows: the antilegomena were still feelable as a category a millennium after Eusebius named it. The seams did not vanish; the churches decided, repeatedly, to live with them.
"If anyone adds to them, God will add to that person the plagues described in this book" (Rev 22:18–19) is sometimes quoted as though it sealed the Bible shut. In context it is a curse protecting the integrity of that book — the Revelation of John — using a formula ancient scribes attached to individual works (compare Deut 4:2, which does the same for the law of Moses and was never taken to forbid the prophets that followed). Revelation sits last in the modern binding; it was not written last, and its curse guards its own text, not the table of contents.
The facts above are largely common ground. What they mean is not, and the disagreement sorts into two live accounts.
On one account, the canon is the church's product. Books became Scripture by being received: the community conferred the status, and a different history — a Marcion who won, a West that kept doubting Hebrews — would have produced a different Bible. Scholars in this line distinguish "scripture" (a book functioning authoritatively) from "canon" (a closed list), and date the canon proper late, to the fourth century's lists and synods.15 Rome's version of this account gives it doctrinal form: the church, teaching with authority, defined the canon — which is why Trent could rule on it.
On the other account, the church's role was recognition, not conferral. The books carried their authority intrinsically — from their origin in the apostolic witness, and behind that from their inspiration — and the community's reception was the effect of that authority rather than its source. Advocates point to the pattern in the record itself: the lists ratify prior usage; Athanasius's own language is of handing down what was received; no council claims to be granting status.16 On this account the canon was functionally present as soon as the books existed and were heard, and the fourth-century lists are the paperwork, not the event.
The accounts agree on more than their partisans advertise: both grant that reception took centuries, that real books were really disputed, and that the process ran on human argument. The genuine fork is behind the history: whether there existed a fact of the matter — a right list — that the church could be correct or mistaken about. That is not a question the surviving evidence can settle, because every piece of evidence is an act of reception, and reception is what both accounts predict. The reader should know that this study's author stands, by confession, in the second account; the history in the sections above has been told so that a reader in the first account can check every step of it.
The history establishes this much: the canon was recognized far more than it was decreed. No council invented it; the lists wrote down what long usage had already sorted; the disputed books were argued about openly, for generations, and the record kept the arguments. The edges — the Greek Old Testament shelf, the wider Eastern canons — were drawn in different places by different communions and remain where they were drawn.
And the history cannot establish this: that the books received are the right books. That the process happened as described is documentation; that the process was providential is confession. The line between those two is exactly where this site's method stops the sentence, and the honest reader is owed the stop.
The recognition story is the winners' story. Every document in this study's evidence base — Eusebius's ledger, Athanasius's letter, Serapion's ruling — was preserved by the party that prevailed. Communities that read the Gospel of Peter or kept Enoch on the top shelf mostly did not get to keep archives. When the record shows "the churches" converging on twenty-seven books, it is showing the convergence of the network that won; a fuller record might show the consensus as narrower, later, and more enforced than it looks. The response that the surviving usage evidence is broad and geographically dispersed has weight, but it is drawn from the same curated survivals it defends.
The criteria may be rationalizations. Apostolicity, antiquity, catholicity, and the rule of faith read less like a gate the books passed through and more like a description, after the fact, of the books that were already loved. Hebrews shows the tests bending for a book the church wanted (an unknown author waved through); the Shepherd shows them tightening against a book that was arguably more widely read than 2 Peter. And catholicity is circular on its face: "received by the whole church" means received by the network doing the receiving. A defender can answer that criteria which bend under judgment are how all canons of judgment work — precedent, evidence law, textual criticism — but the objection stands: the tests did not operate as an algorithm, and presenting them as one flatters the process.
The tidy first-century Old Testament may be apologetic. The claim that the Hebrew canon was effectively settled by Jesus' day leans hard on Josephus, who was writing a defense of Judaism for a hostile audience and had every motive to overstate the fixity and unanimity of the collection. Qumran's broad library, Melito's missing Esther, and the Mishnah's live debates all suggest softer edges than "twenty-two books no one has dared alter." The three-tier shape by the second century BC is well evidenced; the closed list by the first century AD is an inference with an apologist as its chief witness.
The canon did not fall from the sky bound in leather. It was written over a millennium, sorted over centuries, argued about by name, listed late, and bound with its runners-up. A reader who finds that history unsettling is reacting to a story the church's own archivists chose to preserve — Eusebius did not have to record the disputes; Athanasius did not have to append the second shelf. They kept the seams in view, and a site devoted to reading texts as they are can do no less.
Where the author lands, said plainly and held as confession rather than conclusion: I receive these books as the word of God, and I take the long, argumentative, human process by which they were gathered to be the ordinary way providence works — through use, testimony, and time, rather than around them. Nothing in this study proves that reading, and the criticism section names real cracks in any attempt to prove it. What the history does supply is this: the collection I confess was not assembled carelessly, or quickly, or in the dark. The church knew which books were disputed, said so for centuries, and left the record where I could check it. That is not the same as a guarantee. It is the difference between an inheritance with a paper trail and an inheritance with a story — and it is the paper trail that lets the question stay open on the table, where a reader who does not share the confession can still handle every document in the case.
Sources
The Holy Bible, New Revised Standard Version, Updated Edition (NRSVue), 2021. Primary text for scripture quotations.
Primary source. All scripture quotations verified verbatim against NRSVue, 2026-07-16.
Eusebius of Caesarea, Ecclesiastical History (c. 325), esp. 3.25; 4.26; 6.12; 6.25.
Primary source (4th c., pro-Nicene court historian; preserves earlier documents — its selections reflect the winning party, as the Criticism section notes).
Athanasius, Festal Letter 39 (AD 367).
Primary source (Alexandrian bishop; anti-Arian).
The Muratorian Fragment; Prologue to Sirach; Josephus, Against Apion; 4 Ezra 14; Mishnah Yadayim.
Primary sources of mixed provenance; each confirmed at its note against the edition named, 2026-07-16.
Metzger, B. M. (1987). The Canon of the New Testament: Its Origin, Development, and Significance. Oxford University Press.
Position: Mainline Protestant; the standard critical survey. Cited for framework and the κανών appendix; no direct quotations.
Kruger, M. J. (2012). Canon Revisited: Establishing the Origins and Authority of the New Testament Books. Crossway.
Position: Conservative evangelical (Reformed); the self-authenticating model surveyed in §Two accounts. Position confirmed at survey level; no direct quotations.
McDonald, L. M. (2007). The Biblical Canon: Its Origin, Transmission, and Authority. Hendrickson.
Position: Historical-critical; the community-formation account surveyed in §Two accounts. Position confirmed at survey level; no direct quotations.
Blind gate panel (2026-07-16, this version, after the primary-source verification pass): three independent cold Claude Opus 4.8 runs — 20, 20, 20; Fidelity 12, 11, 11. Two different-lab blind reads the same day: ChatGPT (default model) core 20/20; GPT-5.5 (API) 17/20 core · 11/12 fidelity — the full spread, 17–20, is disclosed rather than smoothed. GPT-5.5's residual: overconfident compression on disputed historical inferences (the Josephus / 4 Ezra numbers mapping; early closure) — its flagged "almost certainly" was softened on the page after grading, and the Criticism section already presses the same point. The panel's residual is F1 (3/4): the scholar-position footnotes are confirmed at survey level rather than from primary page-pulls. All quotations verified verbatim against the editions named, 2026-07-16.
How We Got the Bible — the full pipeline (writing, transmission, canon, translation) surveyed in one sitting; this study is the canon chapter of that story slowed down and given its arguments back.