The Biblical Canon

"There was no noticeable change of attitude before and after the Festal Letter of Athanasius, no flourish of trumpets at Hippo or Carthage." -- John Wenham, 1972¹

"I agree also that it is the work of a holy and inspired man. But I cannot readily admit that he was the apostle" -- Dionysius of Alexandria on Revelation, c. 250, in Eusebius²

Summary

The Bible did not come from one decision.  It came from a long process of use and judgement, carried out by churches and their leaders, and at the end of it different churches drew the edges in different places.  The popular story says the collection originated with an emperor or a council, usually Nicaea in 325.  The evidence of earlier use contradicts that story.  But the opposite story, that no one ever chose anything and the books simply rose to the top, is false too.  Church leaders were judging books from the second century on, and later councils did fix lists for the churches under them.  What the record shows is a large core that was in use early, a ring of disputed books that took centuries to settle, and a closing of the edge that came region by region.

1. "Canon" means several different things, and the date depends on which one is meant.  A book can be used, quoted as Scripture, read aloud in worship, copied with other books, named on a list, accepted almost everywhere, or fixed by an authority that forbids adding or removing anything.  These happened at different times.  The Greek word kanōn was not applied to the list of books until after 350.³  Albert Sundberg and Geoffrey Hahneman define canon as a closed list and so date it to the fourth century.  Michael Kruger uses three definitions together, including one in which a book is canonical because God gave it, whether or not anyone has yet received it.⁴⁴  The definitions do not change the evidence, but they do change what counts as an answer, so this article asks its historical questions separately.

2. The New Testament core was in use as Scripture by about 200, in more than one region, before any surviving council list.  About 155 Justin Martyr describes the apostles' "memoirs", which "are called Gospels", read aloud at Sunday worship in Rome beside the prophets, and his pupil Tatian combined four Gospels into one account he called the Diatessaron.⁵⁵⁶  About 180 Irenaeus of Lyon argued that the Gospels cannot "be either more or fewer in number than they are", and he quotes as Paul's the disputed letters too, including Ephesians, Colossians, 2 Thessalonians, the letters to Timothy and Titus.⁷⁸  About 207 Tertullian of Carthage named the same four Gospels and complained that Marcion had rejected the letters to Timothy and Titus.⁹¹⁰  Hurtado's 2006 survey of manuscripts dated before about 325 counted 36 copies of the four Gospels against about 8 to 10 of all other gospels, nearly all from Egypt.¹¹¹¹  This early use is good evidence against a fourth-century invention of the core.  It is not evidence that no one selected books before then:  Irenaeus was defending four Gospels against rivals, and Serapion of Antioch withdrew the Gospel of Peter after reading it.¹²

3. Seven New Testament books were disputed for a long time, and some books outside the canon were read as Scripture by some churches.  Eusebius, writing about 325, called James, Jude, 2 Peter, 2 John and 3 John "disputed" but "known to most", and he gave two places for Revelation because opinion was divided.¹³  Hebrews was read as Scripture in the East, but in 414 Jerome reported that Latin usage still did not receive it among the canonical books, while the Greek churches did not receive Revelation.³  On the other side, the Shepherd of Hermas, the Apocalypse of Peter and 1 Clement were read in some churches.¹⁴³  The disputes were argued with stated reasons:  doubtful authorship, different style, late date, limited use, and in Jude's case a quotation from the book of Enoch.¹⁵

4. The first surviving list of exactly the 27 is from 367, and agreement came region by region.  Athanasius of Alexandria listed exactly the 27 books in his Festal Letter of 367.¹⁴  Cyril of Jerusalem, about 350, had already left Revelation out, and so did Gregory of Nazianzus after Athanasius;  Amphilochius of Iconium reported that most called Revelation spurious.¹⁶³³  The Syriac church at Antioch still used a 22-book New Testament around 400.¹⁷  A surviving list of the 27 dates the writing down, not the first time anyone held that collection.

5. The core was in use before the surviving council lists.  Councils later closed lists for their own churches.  None of Nicaea's twenty canons concerns the books of Scripture, and Constantine's order for fifty Bibles names no books.¹⁸¹⁹  Jerome reported, around 400, that Nicaea had "counted" Judith among the Scriptures, a weak and unexplained report.²⁰  The first council lists that survive are African:  Hippo in 393 and Carthage in 397, repeated in 419.  They bound the African churches' reading, asked churches overseas to confirm them, and said they were passing on what "we have received from our fathers".²¹  Florence professed the full list as "true and necessary doctrine" in 1442, and Trent defined it with an anathema in 1546.²²²³  Eastern councils ratified collections that disagree with each other;  no council recognised as ecumenical by all Eastern Orthodox churches fixed one total.

6. The Old Testament is where the churches differ most.  Josephus, about 95, counted 22 books closed at the time of Artaxerxes, and 4 Ezra, about 100, counted 24 public books.  Josephus names none of his 22, but the usual reconstruction matches the Jewish 24 and the Protestant 39, counted differently.²⁴²⁴²⁵  There was no "Council of Jamnia".  Rabbis argued about Ecclesiastes, the Song of Songs and Esther into the third century, though some of those arguments were about ritual rules rather than inspiration.²⁶²⁷  The Catholic Old Testament has 46 books, the Orthodox a longer list with no single fixed total, and the Ethiopian church a wider list again.²⁸²⁹³⁰  The early church never agreed on this:  Jerome followed the Hebrew canon, and Augustine and the African councils followed the wider one.³¹²⁰

7. The positive case for the canon's authority is strongest at the core and needs more premises at the edges.  The argument Jonathan McLatchie summarises from John Wenham runs:  establish Jesus's authority historically;  Jesus treated the Jewish Scriptures as God's word;  he commissioned apostles as witnesses;  so writings that carry apostolic authority share that status.  The first step is argued elsewhere on this site.  Granting it, the argument is strongest for the books Jesus's contemporaries counted, for the four Gospels and for Paul's undisputed letters.  For Hebrews, 2 Peter, 2 and 3 John and Revelation it needs an extra link in each case, such as apostolic approval of an anonymous book, that the surviving evidence can make likely but not show directly.  Wenham himself concedes that "the criterion of prophetic or apostolic authorship ... does not cover all cases" and rests the boundary on providence working through the church's reception.¹  Catholic and Orthodox Christians rest it on the church's authority, and Kruger on qualities he finds in the books themselves.³²³³

Verdict.  The claim that the whole collection originated with an emperor or a single council that burned the rivals does not fit the evidence.  The four Gospels and most of Paul's letters were functioning as Scripture in several regions by about 200, chosen and defended by bishops and teachers within a network of churches.  Ehrman's scholarly skeptical account, that a proto-orthodox party built the canon in its fights with rivals, fits much of this history.  Doctrine, age, authorship and use all appear in the recorded decisions;  the records do not establish which motive controlled the whole process.  The claim that the whole canon was obvious from the start does not fit either.  Seven New Testament books took centuries to gain general acceptance, and the Old Testament's edges still divide the churches.  That these books, and only these, are inspired is a further claim.  History can support the Christian case well for the core, but identifying the exact boundary requires further premises.

Part 1: What "canon" means, and what is being tested

Arguments about the canon often go wrong at the first step, because the two sides are using one word for different things.  A book can stand in at least seven different relations to a church, and the evidence for each is different.

The word itself is late.  Kanōn means a measuring rod or rule.  In his 1987 history of the New Testament canon, Bruce Metzger found that "it was not until the second half of the fourth century that κανών and its derivatives κανονικός and κανονίζειν were applied to the Scriptures."³  The first known use is by Athanasius, soon after 350, who describes the Shepherd of Hermas as not belonging to the canon.

From this point the scholarship divides over definitions.  Albert Sundberg argued in 1968 that "Scripture" and "canon" must be kept apart.  On his view a church can have Scripture without a canon, and a canon exists only when a list is closed.  Michael Kruger, a Reformed scholar who disagrees with him, summarises the position fairly:  Sundberg "argued that we cannot speak of the idea of canon until at least the fourth century or later."⁴  Geoffrey Hahneman put it plainly:  "Once a distinction is made between scripture and canon, the idea of a New Testament canon does not appear applicable until the fourth century."⁴  A second, "functional" definition counts a book as canonical as soon as a church treats it as Scripture.  On that definition the core of the New Testament is second-century.⁴  Kruger adds a third, "ontological" definition:  "Books do not become canonical — they are canonical because they are the books God has given as a permanent guide for his church."⁴  He uses the three together, and grants what the first one stresses:  "it took several centuries for the edges of the canon to solidify."⁴

Each definition has a cost, and none settles the history by itself.  The exclusive definition does not require the modern 27.  Marcion's collection of one Gospel and ten letters was closed, and the Muratorian writer's graded list says which books are received, which may be read privately and which are rejected.  These are claims to bounded collections of different extents.  The first closed Christian list, the first surviving list of exactly the 27, universal closure and universal agreement are therefore four different questions.  They need not have one date, and universal agreement has not occurred.  Kruger points out that if a closed canon means uniform practice across all churches, "on those terms we still do not have a canon even today."⁴  The functional definition counts the Shepherd as canonical where it was treated as Scripture;  public reading alone does not show that status.  Lee McDonald objects to a definition that makes later-excluded books canonical.⁴  The ontological definition is openly theological:  it says what Christians claim the canon is.  A further premise about God's preservation and guidance is needed to connect that definition to the surviving collection.

The disagreement between these scholars is not only about words.  Sundberg, Hahneman and McDonald hold, as a historical claim, that the early church made no conscious decisions about a closed list until the fourth century.  Kruger holds, as a historical claim, that the idea of a canon grew from within the earliest churches.³⁶⁴  Those claims can be tested against the evidence below, whatever word is used for the result.

This article does not need to settle the word.  It asks four separate questions, and keeps them separate:

1. When did each book come to be used and treated as Scripture, and where? This is a question of evidence about practice.
2. Which books were disputed, by whom, and for what stated reasons?
3. What did councils and lists actually do? Did they choose the books, record a practice that already existed, or close an edge that practice had left open?
4. Can the authority of these books be argued from history? This is a different kind of question. Historical work can show who wrote a book, when it was received and why. The claim that a book is inspired by God goes beyond that, and needs at least one premise that is not purely historical. Part 7 tests how far the historical case reaches and where the extra premise comes in.

Part 2: The New Testament core was in use early, and leaders were already judging books

The earliest Christian writings that survive are letters, and letters were copied and passed on.  Paul asked that his letter to Colossae be read at Laodicea too (Colossians 4:16).  The letter called 2 Peter speaks of all Paul's letters and sets them beside the other Scriptures (2 Peter 3:15-16).  That verse proves an early collection only if 2 Peter is early, and its date is one of the disputes below.  Firmer evidence comes from outside the New Testament.  1 Clement, a letter from the church at Rome to Corinth that Kirsopp Lake dated to the 90s, tells the Corinthians to "take up the epistle of the blessed Paul the Apostle".³⁴³⁴  Polycarp, bishop of Smyrna in the first half of the second century, goes a step further.  In the covering letter he sent to Philippi with copies of Ignatius's letters, he writes:  "as it is said in these Scriptures, 'Be ye angry and sin not,' and 'Let not the sun go down upon your wrath.'"³⁴  The first line is also in Psalm 4, but the second is found only in Ephesians 4:26, so Polycarp is quoting a letter carrying Paul's name as Scripture.  Two cautions limit the point.  This chapter survives only in a Latin translation, and in the same breath Polycarp says that he is less well versed in the Scriptures than his readers, so he may have taken both lines for the Psalms.

Marcion and the four Gospels

Marcion is the first person known to have published a fixed, closed collection of Christian books.  Metzger dates his break with the Roman church to a hearing in July 144.³  No copy of Marcion's Bible survives, so everything about it comes from his opponents.  Irenaeus, about 180, says Marcion "mutilates the Gospel which is according to Luke" and "dismembered the Epistles of Paul".³⁷  Epiphanius, writing around 375, two centuries later, lists Marcion's letters:  "1. Galatians. 2. Corinthians. 3. Second Corinthians. 4. Romans. 5. Thessalonians. 6. Second Thessalonians. 7. Ephesians. 8. Colossians. 9. Philemon. 10. Philippians."³⁸  In a later chapter Epiphanius says Marcion had no version "of First and Second Timothy, Titus, and Hebrews".³⁸  Tertullian, about 207, says Marcion "rejected the two epistles to Timothy and the one to Titus".¹⁰  These are hostile reports.  They agree on what Marcion lacked, but "rejected" is Tertullian's reading;  whether Marcion knew the Pastorals and threw them out, or never had them, the reports cannot settle.

Adolf von Harnack argued that the church's New Testament was its "creative act" in reaction to Marcion.⁴  Bart Ehrman puts it more cautiously:  some scholars think Marcion's decision spurred the proto-orthodox to make a canon of their own.³⁹  Marcion chose from writings already in circulation:  a Gospel his opponents identify as Luke and letters already called Paul's.  A church reacting to him could keep the books it shared with him and reject his cuts.  1 Clement's earlier use of Paul establishes that Paul's letters were valued before Marcion;  it does not establish an earlier closed collection.  Irenaeus's defence of four Gospels addresses several rival groups, including Marcion's, so reaction to Marcion could have been one influence among others.  These facts leave room for Marcion to have accelerated the formation of a bounded collection.  They do not establish him as its sole cause, and they do not refute Harnack's thesis in its full form, which is reported here through Kruger rather than assessed from Harnack's own work.

Between Marcion and Irenaeus, two writers in Rome show Gospels already doing the work of Scripture.  Justin Martyr, writing his First Apology about 155, says that "the apostles, in the memoirs composed by them, which are called Gospels", handed down the words of the Lord's Supper.  He then describes Sunday worship:  "the memoirs of the apostles or the writings of the prophets are read, as long as time permits".⁵⁵  Read aloud in worship in the same slot as the Old Testament prophets, the Gospels were functioning as Scripture in Rome a generation before Irenaeus.  Justin does not say how many Gospels there were, and he names no evangelist in these passages, so his "memoirs" could have included books outside the four.  His pupil Tatian, who broke with the church after Justin's martyrdom, "formed a certain combination and collection of the Gospels, I know not how, to which he gave the title Diatessaron", which is Greek for "through four".⁶  Eusebius admits he had not seen it, and Tatian's original is lost.  But a parchment leaf of a Greek Gospel harmony was dug out of a siege embankment at Dura-Europos on the Euphrates, an embankment that "was constructed after 254 and before 256-257 A.D."  Its editor, Carl Kraeling, describes it as "a 'harmony' in which Matthew, Mark, Luke and John are systematically combined".  He judged it a piece of Tatian's work, while naming two other possible sources.⁴⁰⁴⁰  Whoever made it, the Dura leaf shows a book built from exactly our four Gospels in use in eastern Syria before 256.  Tatian's title, if Eusebius reports it rightly, shows four Gospels treated as a set by Justin's own pupil, a writer active in the 160s and 170s.

By about 180 Irenaeus of Lyon defended exactly four Gospels:

"It is not possible that the Gospels can be either more or fewer in number than they are. For, since there are four zones of the world in which we live, and four principal winds ... it is fitting that she should have four pillars"⁷

His reason sounds strange today, and a skeptic is right that the four winds prove nothing.  What the passage shows is that Irenaeus was arguing for a fixed number against people who used more or fewer.  It does not show by itself whether the four were already settled where he lived or whether he was promoting them.  Two other things in the chapter bear on that.  He says each group he opposed used one of the same four:  the Ebionites Matthew, Marcion a cut-down Luke, those who separated Jesus from Christ Mark, and the Valentinians John.  "So firm is the ground upon which these Gospels rest, that the very heretics themselves bear witness to them".⁷  And he rejects the Valentinians' own Gospel of Truth because it "agrees in nothing with the Gospels of the Apostles".⁷  The first claim is a polemicist's report about his opponents, and it is not a census of every gospel they read:  the same groups also had books of their own, as his rejection of the Gospel of Truth shows.  So Irenaeus is good evidence that four Gospels were the norm in his network by about 180, and that he was drawing a boundary against real alternatives.

What was copied

Manuscripts show what Christians actually copied, which is a different kind of evidence from what bishops said.  Larry Hurtado set out in 2006 "to take account of all Christian Greek manuscripts dated to the second or third century, including some dated as third/fourth century (i.e., ca. 275-325 CE)".¹¹  He found "Matthew (12), Mark (1), Luke (7), John (16)".¹¹  For gospels and acts outside the canon he found "Gospel of Thomas (3), Protevangelium of James (1), Gospel of Mary (2), 'Egerton' Gospel (1), Gospel of Peter (2?), 'Fayum' Gospel (1), Acts of Paul (3)".¹¹  So the four canonical Gospels have 36 copies and all other gospels together have 8 to 10, depending on two doubtful fragments of Peter.⁴¹

What these numbers can and cannot show needs stating before they are used.  They are a 2006 count, using the dates editors had assigned from handwriting, and those dates have wide margins;  it is not a current census.  Almost all the papyri come from Egypt, where the climate preserves them, so they mainly show one region.  A codex holding several Gospels counts once for each.  Survival is not a random sample of what was read.  And copying does not equal canonical status:  Mark has only one copy, and Hurtado warns that Mark's single copy should make us cautious about reading too much into the low counts for other texts.¹¹  The Shepherd of Hermas, which was never in the final canon, has up to eight copies from before 300.¹¹  The narrow conclusion is about surviving witnesses:  this sample preserves the four Gospels several times as often as the other gospels combined.  It suggests more extensive use of the four in Egypt, subject to the survival problem.  Because the sample includes dates as late as about 325, its entire ratio cannot be assigned to the period before Constantine favoured Christianity.  Hurtado adds that those who read Thomas did not put it into the early collections of Gospels.¹¹

A bishop rejects a gospel

Around 200, Serapion, bishop of Antioch, first allowed the congregation at Rhossus to read the Gospel of Peter without having read it himself.  Told later that its users held heretical views, he obtained a copy from the Docetists who used it and found "many things in accordance with the true doctrine of the Saviour, but some things added".¹²  His rule is an early explicit statement of a test:  "we receive both Peter and the other apostles as Christ; but we reject intelligently the writings falsely ascribed to them, knowing that such were not handed down to us."¹²  Two tests appear in that sentence.  The first is apostolic origin.  The second is what had been "handed down", meaning use in the churches from the beginning.  The case also shows a bishop exercising authority over what a congregation read, a century before Constantine, and reversing a permission once he had read the book.  The order of events supports Ehrman's point that doctrine could drive a verdict on authorship:  Serapion judged the book not Peter's after finding its teaching wrong.³⁹

The core, region by region

Wenham, writing to defend the canon, put it this way:  "By AD 170 (at the latest) not only was the concept of New Testament Scripture firmly established, but the main contents of the new Canon were un- disputed : four Gospels, Acts, thirteen letters of Paul, 1 Peter, 1 John."¹  That is his conclusion.  The named witnesses support a narrower and more exact statement.

So by about 200 to 210, the four Gospels as a set are attested in Gaul and Carthage, and probably in Rome and Alexandria.  The Pauline collection, with the Pastorals, is attested in Gaul and Carthage.  1 Peter and 1 John were widely used:  Eusebius, who had the books, reports that Papias of Hierapolis drew on both, and that Polycarp's surviving letter uses 1 Peter.⁴³⁶  The Muratorian fragment as it survives does not name 1 Peter.  Eastern Syria is thinly documented apart from the Dura harmony, and the details of Egyptian practice are less well documented too.  This is strong evidence that the core was not invented in the fourth century.  It is evidence about what these churches and their leaders received, and the leaders were already ruling some books out.

The authorship of six of the letters carrying Paul's name is argued separately in Did Paul Write the Disputed Letters?, which leaves Ephesians and the Pastorals unresolved.  Their reception by about 180 to 210 in Gaul and Carthage is well attested.  Receiving a letter as Paul's does not prove that Paul wrote it.

Part 3: The Muratorian fragment

The date of the Muratorian fragment is disputed.  If it was written around 200, it is the oldest surviving list of New Testament books.  If it was written around 375, it is one fourth-century list among several.

What the manuscript is

Ludovico Antonio Muratori published it in 1740 from a manuscript in the Ambrosian Library in Milan.  The manuscript came from the monastery of Bobbio.  Metzger calls it an eighth-century manuscript, and Hahneman dates it to the seventh or eighth.³⁴⁴  Metzger also notes it is "a commonplace book of some monk", a miscellany of fourth- and fifth-century theology into which the list was copied.³  The Latin is very poor.  The text starts in the middle of a sentence about Mark and stops in the middle of a sentence about heretics, so its beginning and end are lost.  Many scholars, though not all, think the Latin translates a Greek original, and Metzger dates the Latin translation to after the start of the fifth century.³  If Chromatius of Aquileia used it, as discussed below, the Latin already existed before his death in 407.  Three dates have to be kept apart:  the copy (seventh or eighth century), the Latin translation (fifth century or later), and the original composition, which is the one in dispute.

What it says

The surviving text begins in the middle of a sentence, then says "The third book of the Gospel is that according to Luke", and later names John as the fourth.  So Luke and John are named, two Gospels came before them, and that those two were Matthew and Mark is a near-certain inference rather than a surviving word.⁴⁵  The fragment lists Acts, thirteen letters of Paul, Jude, letters of John, the Wisdom of Solomon, and the Revelations of John and of Peter.  It grades books into more than two classes:

1. Accepted. Paul's letters to seven churches, and Philemon, Titus and two to Timothy, which "are held sacred in the esteem of the Church catholic for the regulation of ecclesiastical discipline."³
2. Accepted, but disputed for public reading. "We receive only the apocalypses of John and Peter, though some of us are not willing that the latter be read in church." The Latin of the second clause is "quam quidam ex nostris legi In eclesia nolunt".⁴⁶³
3. To be read, but not publicly as Scripture. Of the Shepherd of Hermas: "it ought indeed to be read; but it cannot be read publicly to the people in church either among the prophets, whose number is complete, or among the apostles, for it is after [their] time."³
4. Rejected. A letter to the Laodiceans and another to the Alexandrians, "forged in Paul's name to [further] the heresy of Marcion", and the writings of named heretics, "for it is not fitting that gall be mixed with honey."³

The letters of John need care.  In the section on John's Gospel the writer says John "mentions these particular points also in his Epistles", and quotes the opening of 1 John.  Later he counts "the epistle of Jude and two of the above-mentioned (or, bearing the name of) John".⁴⁵  Samuel Tregelles reads the damaged Latin "superscrictio Iohannis" as "superscripti Iohannis", "of the John mentioned above".⁴⁶  The writer therefore used 1 John and counted two letters of John, but whether those two are 1 and 2 John, or 2 and 3 John on top of the 1 John already quoted, the text does not say.  Ehrman reads it the same way:  "he accepts two of the three that we have, but does not indicate which two".³⁹

Two other features stand out.  The list includes the Wisdom of Solomon, between Jude and the apocalypses.  That is its place in this list;  calling it a "New Testament" book imposes a later division the writer does not use.  And the fragment gives a reason for keeping the Shepherd out of public reading:  it is too recent to stand with the prophets or apostles.  That is an argument from date and apostolic origin, not from content.

Its silences need to be weighed one by one.  The surviving text never mentions Hebrews, James, or either letter of Peter.  An omission is not a rejection unless the author says so, but a silence is not worthless either.  The text is damaged and was copied carelessly, so a missing book may be a gap in the copy.  Theodor Zahn proposed a reconstruction of the Greek original that makes room for at least 1 Peter, and even Hahneman allows for "omissions in the Fragment."⁴⁷  But the passage where 1 Peter would most naturally stand, after Jude and John and before the apocalypses, survives.  So the silence on 1 Peter is explained by damage only if a line dropped out in copying, which is a conjecture.  The fairest reading is that the writer probably did not list 1 Peter, Hebrews or James in this document, and that this does not show what he thought of them.

The date

The traditional date rests on one sentence.  The writer says that Hermas wrote the Shepherd "very recently, in our times, in the city of Rome, while bishop Pius, his brother, was occupying the [episcopal] chair".³  Pius was bishop of Rome about 140 to 155.³  If "in our times" means the writer's own lifetime, he wrote in the late second century.  Metzger concluded that the remark "points to a date in the latter part of the second century and certainly not later than the year 200", and that Rome is meant because the writer calls it simply "the city".³

Albert Sundberg argued in 1965 and 1973 for a fourth-century date and an Eastern origin.  Geoffrey Hahneman developed the case in a 1992 Oxford monograph and restated it in 2002.⁴⁴⁴⁸  His arguments, from his own chapter, are these.  First, the early date hangs on a three-word phrase, and "It is at least questionable to date such a cornerstone in the development of the Chris- tian canon on the basis of a three-word Latin phrase in a document first published as a striking example of scribal barbarism."  Second, the phrase can be read another way:  Sundberg takes "most recently" to mean the latest of the books listed, and "our time" to mean the post-apostolic age as against the apostles' age.⁴⁸  Third, the fragment is wrong about Hermas anyway.  Hermas never mentions a brother, speaks of the Roman church's leaders always in the plural, and "the internal evidence for dating the Shepherd all suggests a date around the turn of the first century, 40-50 years before the beginning of Pius's episcopacy".⁴⁸  In his monograph he adds that the fragment's mention of Miltiades depends on a copying error in Eusebius, and that its contents resemble lists of the later fourth century.⁴⁷⁴⁷

The defenders of the early date answer point by point.  Eckhard Schnabel's 2014 survey lists eight objections.  Charles Hill noted that a writer could be wrong about Hermas's date in the second century as easily as in the fourth.⁴⁷  Tertullian, about 210 to 220, already said that the Shepherd was judged "by every council of the churches ... among the apocryphal and false" writings, so church rulings against Hermas before about 220 are attested, and a second-century list that sets Hermas aside is not an anomaly.⁴²  Zahn and Lagrange argued that Victorinus of Pettau, about 270, borrowed the fragment's comment that John and Paul wrote to seven churches but spoke to all.  If so, the list existed before 300.⁴⁷  That parallel is suggestive but not decisive, because the idea of the seven churches standing for the whole church was common.

A third position has grown since then.  Clare Rothschild argued in 2016 that the fragment is "a fictitious attempt to provide a venerable second-century precedent for a later position on canon", which betrays itself "through anachronisms, including use of the Vulgate".⁴⁹  This answers Hill's question directly:  on her view a later author did adopt a second-century persona, and the persona's words would be expected to sound early.  So reading "in our times" literally does not refute her.  What would test her case is the specific anachronisms she alleges, above all the claimed use of Jerome's Vulgate, which would put the Latin, or its source, after about 400.  Her full argument, in a 2018 article and a 2022 commentary, could not be read for this article.  The first published reply, by Christophe Guignard in 2019, was read.  Guignard quotes her 2018 article as allowing settings for the forgery from the fourth century to "the eighth or ninth century when the codex in which it appears was copied".⁵⁰  His main answer is a fixed latest date.  Chromatius, bishop of Aquileia, who died in 407, reworked a passage of the fragment in the prologue to his commentary on Matthew without naming his source, so "the date of 400 represents a chronological bolt for any hypothesis."  His test case is a phrase about Luke.  The fragment calls Luke "iuris studiosus", which Guignard renders "expert in law", and Chromatius calls him "eruditissimus legis", "very learned in the [Mosaic] Law".  Guignard asks:  "how could we imagine that an author who had found eruditissimus legis in his source would rephrase it with the obscure iuris studiosus?"⁵⁰  He adds that Chromatius used the Latin translation we have.  If that is right, the Latin text existed by about 400.  Every setting for the forgery after that date falls away, and a trace of Jerome's Vulgate in the Latin would date the translation, not a Greek original behind it.  His second argument is from language.  A Latin text translated from Greek that is also "distinctively Western-minded" was most likely written when Greek was still the main language of the Western churches outside North Africa, "namely in the 2nd century or, at the latest, in the first half of the 3rd."⁵⁰  Two limits remain.  Guignard does not re-argue the Vulgate case in this reply;  he refers to his 2015 article on the fragment's original language, which has not been read for this article.⁵⁰  And the Chromatius date does not touch a fourth-century composition, which is what Sundberg and Hahneman proposed.  The language argument does bear on them, but only if the fragment is Western, and both of them placed it in the East.

Assessment.  On the arguments read here, the early date is somewhat better supported.  Its main evidence is the text's own words, plus Tertullian's report that councils had already ruled on Hermas, which removes the charge that a second-century list would be an anomaly.  Hahneman's strongest point is that the fragment's claim about Hermas's date is probably wrong, but that cuts against the writer's reliability in either century.  The late date's weakest point is the reading of "very recently, in our times", which strains the Latin.  Rothschild's forgery theory escapes that objection.  Guignard's reply, if Chromatius's borrowing holds, confines any forgery to before about 400, and his case for a Greek original written in the West points to the late second or early third century.  Her specific anachronisms, the Vulgate claim among them, are still not examined here from her own text.  So the date remains open, with a slightly firmer lean to about 200 than the arguments read before Guignard gave.  A late date now needs a Greek original composed in the fourth century, most plausibly outside the West, which is the form of the case Hahneman made.  Whatever the date, the second-century core does not depend on the fragment.  What depends on it is narrower:  whether a Western church around 200 already used a long, graded list that included the Apocalypse of Peter and Wisdom and drew sharp lines against forgeries.

Part 4: The disputed books, and the reasons given

Eusebius's three classes

Around 325 Eusebius of Caesarea, a historian with a large library, sorted the books into classes.  His words are often summarised badly, so they are given here in Kirsopp Lake's translation:

"In the first place should be put the holy tetrad of the Gospels. To them follows the writing of the Acts of the Apostles. After this should be reckoned the Epistles of Paul. Following them the Epistle of John called the first, and in the same way should be recognized the Epistle of Peter. In addition to these should be put, if it seem desirable, the Revelation of John, the arguments concerning which we will expound at the proper time. These belong to the Recognized Books. Of the Disputed Books which are nevertheless known to most are the Epistle called of James, that of Jude, the second Epistle of Peter, and the so-called second and third Epistles of John which may be the work of the evangelist or of some other with the same name. Among the books which are not genuine must be reckoned the Acts of Paul, the work entitled the Shepherd, the Apocalypse of Peter, and in addition to them the letter called of Barnabas and the so-called Teachings of the Apostles. And in addition, as I said, the Revelation of John, if this view prevail."¹³

Three points are often missed.  First, Eusebius says that even his "not genuine" books "would all belong to the disputed books".¹³  His "not genuine" class is a lower grade of disputed book, not a class of heresies.  A fourth class, the gospels of Peter, Thomas and Matthias and the Acts of Andrew and John, is "to be reckoned not even among spurious books but shunned as altogether wicked and impious", because no orthodox writer had quoted them and their style and teaching differed from the apostles'.¹³  Second, "the Epistles of Paul" in his first class means fourteen, Hebrews included, though he notes that "some dispute the Epistle to the Hebrews, saying that it was rejected by the church of Rome as not being by Paul".¹³  Third, Revelation appears in two classes at once, each time with a condition:  among the recognized books "if it seem desirable", and among the not genuine "if this view prevail".  Eusebius is reporting two opinions that both had defenders, and he promises to give the arguments later.  He leans against it elsewhere, but here he records a split rather than a settled verdict.

So "disputed" in Eusebius means three things that should not be run together:  doubt about who wrote a book, doubt about whether the churches had always used it, and a judgement that it was "not genuine", meaning not apostolic or not part of the received collection.

The seven books, one by one

The table separates two questions the sources themselves separate:  whether a book was read as Scripture, and who wrote it.  Jerome put the point bluntly about Hebrews:  "it makes no difference whose it is, since it is from a churchman, and is celebrated in the daily readings of the Churches."³  He received both Hebrews and Revelation, he says, because he followed "the authority of ancient writers" rather than the custom of his own day.  The Jerome entries on 2 and 3 John and 2 Peter are reported from his Lives of Illustrious Men and a letter, as summarised by Metzger.³

The stated reasons in the table are mostly about authorship, style, age and how widely a book had been used.  That is a fact about the reasons the writers gave.  It does not measure how often doctrine, rivalry or local loyalty moved the many churches whose reasons were never recorded, and a stated reason is not always the whole cause.  Revelation is the case where a doctrinal motive is on record.  Dionysius reports that some rejected it because its thousand-year kingdom suited the wrong people, and Wenham names "the growing unpopularity of chiliasm" as one reason it lost ground in the East.²¹  But the same Dionysius kept the book.  He wrote:  "I could not venture to reject the book, as many brethren hold it in high esteem."  He granted that it was "the work of a holy and inspired man" named John, and denied only that this John was the apostle.²  His stated grounds were his own judgement of its holiness and inspiration and the esteem of the brethren, and he argued about its author from style.  In his hands, reception and authorship came apart.

The order of the evidence also differs from book to book.  For 2 Peter and James, Origen is the first writer who clearly names the book, and he already hedges.¹²³  For Revelation and Hebrews the order runs the other way.  The oldest witnesses to Revelation accept it as John's.  Justin, at Ephesus, the first city the book addresses, calls its author "one of the apostles of Christ", and Irenaeus appeals to the testimony of men who "saw John face to face".⁵³⁵⁴  The doubts on record are reported by Dionysius about 250, who says "some before us" had rejected it.²  Hebrews was used in Rome in the 90s and doubted there later, because it was not thought to be Paul's.³⁴¹³  Early reception followed by later doubt is not proof that the early view was right.  Justin was himself a believer in the thousand-year kingdom, which gave him a reason to value the book, and an early date for the tradition does not answer Dionysius's argument from the difference in style between Revelation and John's Gospel.  But it does mean that for these two books the dispute was a later reversal, not a hesitation present from the start.

Books that were read but did not stay

The other side of the edge matters just as much.  Dionysius of Corinth wrote about 170 that his church had read a letter from Rome on the Lord's day, "as also from the former epistle, which was written to us through Clement".⁶  1 Clement was read in church at Corinth for decades but did not become one of the 27.  The Muratorian list accepted the Apocalypse of Peter.  Clement of Alexandria commented on Barnabas and the Apocalypse of Peter alongside the Catholic letters.¹²  Origen in his later years called the Shepherd of Hermas "a work which seems to me very useful, and, as I believe, divinely inspired".³  Codex Sinaiticus copies Barnabas and the Shepherd after Revelation, and Codex Alexandrinus copies 1 and 2 Clement after Revelation.⁵⁵⁵⁶

These books show that many early Christians drew their line in a different place from modern Bibles.  Some drew it sharply:  the Muratorian writer grades books with care, and Tertullian reports that church councils had already judged the Shepherd apocryphal.⁴²  What the sources record is a set of reasons, not the full causes in every region.  The Muratorian writer kept the Shepherd from public reading because it was recent.  Barnabas and the Didache stood in Eusebius's "not genuine" class.¹³  1 Clement appears in none of the fourth-century lists in Part 5 except the Apostolic Canons, which count two letters of Clement in their New Testament;  it is outside the 27, not outside every Christian list.³  Athanasius kept the Shepherd and the Didache as reading for new converts, "not indeed included in the Canon".¹⁴  Rufinus called such books "Ecclesiastical", to be "read in the Churches, but not appealed to for the confirmation of doctrine."⁵⁷

Origen's "trumpets": what the homily does and does not show

A quotation that circulates in apologetics has Origen, about 240 to 250, naming every book of the New Testament.  One English version reads:  "John also sounds the trumpet through his epistles [and Revelation]".  It is offered as a complete list of the 27 a century before Athanasius.  Its transmission and the unstated number of John's letters make the real evidence more modest.

1. It survives only in Latin. The Homilies on Joshua exist only in a Latin translation by Rufinus, made about 400.³ Rufinus did change some of Origen's works. In his preface to On First Principles he says he left out or reworded statements on the Trinity that he judged corrupt, and he says he expanded the homilies on Genesis, Exodus and Leviticus. But of this work he wrote: "The works upon Jesus Nave and the book of Judges ... I translated simply as I found them, with no great labour."⁵⁸ Metzger thought it "not impossible that Rufinus altered Origen's words so as to reflect a later, fourth- century opinion concerning the extent of the canon", while leaning against it.³ The translator's own statement makes a wholesale update less likely, though it does not guarantee every phrase.
2. The critical text does not name Revelation. W. A. Baehrens's 1921 critical edition prints: "Addit nihilominus adhuc et Iohannes tuba canere per epistolas suas et Lucas Apostolorum gesta describens", meaning "John too adds the sound of the trumpet through his epistles, and Luke describing the acts of the apostles".⁵⁹ His apparatus shows that "et apocalypsin", "and the Apocalypse", is added by the manuscripts he groups as C, a family found mostly in France, by one manuscript of another family, g, an eleventh-century copy now in Munich, and by Delarue's printed edition of 1732.⁵⁹⁵⁹⁵⁹ Baehrens judged it an addition, and Metzger agrees: "The words et Apocalypsin are lacking in most manuscripts, and are probably a scribal expansion of the text."³ The brackets in the popular English mark exactly this addition.
3. The numbers are not all stated. The homily does not say how many letters John wrote, and the family Baehrens marks G gives Peter three letters instead of two.⁵⁹⁵⁹
4. Origen's Greek scholarship adds doubts. In his Greek commentary on John, quoted by Eusebius, Origen says Peter left "one acknowledged epistle; perhaps also a second, but this is doubtful", and John left one letter, "perhaps also a second and third; but not all consider them genuine".¹²¹²

So the homily, as far as the critical text can be trusted, is evidence that Origen, preaching to a congregation in the 240s, could name James, Jude and two letters of Peter among the apostles' writings without comment, along with fourteen letters of Paul.  That fits Metzger's suggestion that Origen spoke loosely in a sermon and precisely in his scholarship.³  It is not a third-century Greek list of the 27, it does not include Revelation in its best text, and it does not tell us that every church Origen knew used the same books.

What codices show, and what they do not

The great codices of the fourth and fifth centuries are sometimes treated as canon lists.  They need caution.  A codex records what one workshop copied for one customer.  Codex Vaticanus has lost its last quires and breaks off in Hebrews 9:14, so we do not know whether it once held the Pastorals or Revelation.⁶⁰  Sinaiticus includes Barnabas and the Shepherd.  Alexandrinus includes 1 and 2 Clement, and its ancient table of contents ends with the Psalms of Solomon, now lost.⁵⁶  Inclusion in a volume shows that a book was valued.  It does not show that the copyist ranked it equal with the Gospels.

The earlier papyrus codices answer a narrower question:  which books were bound together.  The Chester Beatty papyrus P45, from the third century, held the four Gospels and Acts in one volume.  Its first editor, Frederic Kenyon, dated it to "the third century, and the first half of it rather than the second", while A. S. Hunt preferred the second half.⁶¹  The proof that the five books formed one book is physical:  "the identity of the hand throughout" and "the numeration of the pages", two of which survive.⁶¹  Kenyon concluded that it "proves that at any rate in the third century it was possible to have the four Gospels in a single volume".⁶¹  P45 is an Egyptian copy and shows one owner's choice, not a rule.  But the four it binds together are the four that Irenaeus defended and that the Dura harmony combined, and Hurtado observes that readers of the other gospels did not put them into "the early Gospel collections".¹¹

The earliest Pauline codex, P46, raises a different problem.  It holds Romans, Hebrews, 1 and 2 Corinthians, Ephesians, Galatians, Philippians, Colossians and 1 Thessalonians, with the end lost.³⁵  Frederic Kenyon calculated that the Pastorals could not have fitted in the missing leaves:  "The space required is therefore nearly twice as much as is available".⁶²  Brent Nongbri has since shown that page numbering in such single-quire books can undercount the leaves, so "We should, however, also allow for the possibility that we have underestimated the size of the quire".⁶³  P46 is weak evidence either way about the Pastorals.  It is good evidence that the Egyptian Christians who made and used it in the third century counted Hebrews among Paul's letters.

Part 5: Lists and councils: what they actually did

The early lists side by side

The table gives each surviving list or survey, the route by which it reaches us, and how its New Testament differs from the 27.  A list's date is the date someone wrote down a practice.  It is not the date the practice began.

The sources for the rows are given in Parts 3 and 4 and below.^12161457333  The table shows two things.  No surviving list before Athanasius gives exactly the 27, and several later ones still do not.  That is a fact about what survives, not proof that no one held the 27 earlier.  And every row that is a full inventory includes the core of Part 2, where the core is the four Gospels, Acts and Paul's letters;  the Muratorian fragment as it survives does not name 1 Peter, and Origen's excerpts are not an inventory.  The variation in full lists is at the edge.

Nicaea and Constantine

The claim that the Council of Nicaea chose the books is widespread.  Its most famous modern form is in Dan Brown's 2003 novel The Da Vinci Code, where a character says:  "The Bible, as we know it today, was collated by the pagan Roman emperor Constantine the Great", and that "Constantine com- missioned and financed a new Bible, which omitted those gospels that spoke of Christ's human traits".⁶⁴⁶⁴  Brown wrote a novel, but many readers took these lines as history.

The evidence against the claim is direct:

1. The canons of Nicaea say nothing about it. The council issued twenty canons. The word "Scripture" appears once, in canon 17, in a rule against clergy lending at interest.¹⁹
2. Constantine's Bible order names no books. Some time after 330, when Constantinople was founded, Constantine wrote to Eusebius asking him "to order fifty copies of the sacred Scriptures, the provision and use of which you know to be most needful for the instruction of the Church".¹⁸ The emperor paid for copies and left their contents to the bishop's knowledge. No record says which books Eusebius put in them. His Ecclesiastical History, written before this, shows he was still unsure about Revelation. So the copies were probably chosen by a bishop, which is an inference from the letter, not an observed selection.
3. Critical scholars reject the claim too. Bart Ehrman, an agnostic and a leading popular critic of traditional Christianity, wrote: "The historical reality is that the emperor Constantine had noth- ing to do with the formation of the canon of scripture: he did not choose which books to include or exclude, and he did not order the destruction of the Gospels that were left out of the canon (there were no imperial book burnings)."⁶⁵ Ehrman's own account puts the choosing earlier, among second- and third-century Christians, as Part 7 discusses.³⁹
4. The altar story is late. Tim O'Neill, an atheist who writes on the history of Christianity, traces the popular version to a quip by Voltaire, who cited a "Supplement" to the council's acts in which the books were piled on an altar and the false ones fell to the floor. O'Neill connects it to the Synodicon Vetus, an anonymous Byzantine work of the ninth century, in which the inspired books are found on top and the spurious ones beneath, and notes that the story "is only found in this one work and is not referenced in any earlier material on the Council of Nicaea".⁶⁶⁶⁶ That genealogy is O'Neill's; the Synodicon itself was not read for this article.

One ancient text does link Nicaea to a book.  Jerome, translating Judith around 400, wrote that he agreed to do it "because this book is found by the Nicene Council to have been counted among the number of the Sacred Scriptures".²⁰  The Latin "legitur computasse" means "is read to have counted", that is, is reported to have counted.  Nothing in the surviving canons supports it.  Jerome may have known of a citation of Judith in the debates, or he may have been repeating a report.  It is weak, unexplained evidence that someone at or about Nicaea treated Judith as Scripture.  The silence of the twenty canons tests one thing well:  the council issued no ruling on the books that its own canons preserved.  It cannot show that no bishop mentioned a book.  Nothing supports the claim that the council voted on the canon.

Book burning

The popular story adds that the losers' books were burned.  Book burning by emperors did happen, but not as the story tells it.  In 303 Diocletian, a pagan emperor, ordered "that the churches be leveled to the ground and the Scriptures be destroyed by fire".⁶⁷  The church's own Scriptures were the target.  Lee McDonald argues that this persecution, which forced Christians to decide which books they would die rather than surrender, was one of two pressures that "provide the social context that led to the clos- ing of the biblical canon."³⁶  Constantine too ordered books burned.  Socrates, writing about 439, copies a circular in Constantine's name ordering that "if any treatise composed by Arius should be discovered, let it be consigned to the flames", with death for anyone caught hiding one.⁶⁸  That order concerns the writings of a living theologian condemned at Nicaea, not gospels, and the report does not show how far it was carried out.  Ehrman's parenthesis is therefore too broad if read as a denial of all imperial book burning.  The sources examined here do not report an order by Constantine or a council to destroy rejected gospels.  That is a limit of this evidence, not proof that every rejected gospel circulated freely.

The African councils

The first surviving conciliar list is African.  The Synod of Hippo in 393 issued a list of the Scriptures that the Third Council of Carthage in 397 repeated.¹⁷  The African code of 419 preserves it as canon 24:

"Item, that besides the Canonical Scriptures nothing be read in church under the name of divine Scripture. But the Canonical Scriptures are as follows:"²¹

The list gives the 27 New Testament books, and an Old Testament that includes Tobit, Judith and two books of Maccabees.  Three features show what the council thought it was doing.  First, the rule is about what may be read in church, the same question Laodicea's canon 59 asks.  Second, the 397 Latin lists "thirteen Epistles of the Apostle Paul, one Epistle of the same [writer] to the Hebrews", still keeping Hebrews apart.  B. F. Westcott notes that the Hippo version reads "fourteen Epistles of Paul" instead.¹⁷  Third, the canon closes:  "Let this be sent to our brother and fellow bishop, Boniface, and to the other bishops of those parts, that they may confirm this canon, for these are the things which we have received from our fathers to be read in church."²¹  Boniface became bishop of Rome only in 418.  Westcott explains the anachronism as a note added when the African canons were collected into one code in 419.¹⁷  In either form, the bishops presented their list as what they had inherited, and asked churches overseas to confirm it.  That is their claim about continuity, and it is probably true of the core.  It does not mean the decree changed nothing.  It turned varied usage into a fixed list for the African churches, it bound what could be read in them, and its two forms of the Hebrews line show that even the wording was still being settled between 393 and 397.  The forms of 393, 397 and 419 are known here from Westcott and the English of the 419 code, not from a modern critical edition.

Laodicea, Trullo and the problem of ratifying everything

A list attached to the canons of the Council of Laodicea, held sometime between 343 and 381,⁶⁹ is often cited as the first conciliar canon.  Its canon 59 is genuine:  "No psalms composed by private individuals nor any uncanonical books may be read in the church, but only the Canonical Books of the Old and New Testaments."⁶⁹  The list in canon 60 is doubtful.  The editors of the standard English translation mark it "[N. B.— This Canon is of most questionable genuineness.]", and Westcott concluded that it "was not I believe an integral part of the original conciliar text."¹⁷⁶⁹  It lacks Revelation.

In 692 the Council in Trullo, in its second canon, ratified "the eighty-five canons ... handed down to us in the name of the holy and glorious Apostles", together with the canons of Nicaea, Laodicea, Carthage and other councils and the canonical letters of Athanasius, Gregory of Nazianzus, Amphilochius and others.  Those sources disagree about Revelation, the minor Catholic letters and the letters of Clement.  Metzger's verdict is blunt:  "Such an extraordinary situation can be accounted for only on the supposition that the members of the Council had not even read the texts thus sanctioned."³  The canon's own text tells against that.  It says the Apostolic Canons bid the church "to receive the Constitutions of the Holy Apostles [written] by Clement", which is an item in Apostolic Canon 85's list of New Testament books, and it rejects them:  "We therefore reject these Constitutions".⁷⁰  So the council read at least that list closely enough to strike one item from it.  Its stated purpose was discipline, the canons were confirmed "for the cure of souls and the healing of disorders", and it ratified bodies of church law without reconciling the book lists inside them.  An Eastern council that confirms several lists and fixes no total has endorsed collections of rules, not closed a canon in the strict sense.

From Florence to Westminster

Wenham dates something like Western agreement on the New Testament to Hippo in 393.¹  In 1442 the Council of Florence, in a decree of union with the Coptic church, issued "with the approval of this sacred ecumenical council of Florence, the following true and necessary doctrine".  The decree professes that one God is the author of both Testaments, "since the saints of both testaments spoke under the inspiration of the same Spirit", and says the church "accepts and venerates their books", listing the 27 and an Old Testament with Tobit, Judith, Wisdom, Ecclesiasticus, Baruch and two books of Maccabees.²²²²  The anathema that follows is aimed at the Manichees, who taught two gods, not at anyone who doubted a book.

The Council of Trent, on 8 April 1546, gave the list again "lest a doubt may arise in any one's mind, which are the books that are received by this Synod", and attached a penalty:  "if any one receive not, as sacred and canonical, the said books entire with all their parts ... let him be anathema."²³²³  Trent's list matches Florence's.  What Trent added was the word "canonical", an explicit definition against a live dispute, and a penalty for denying it.  So "the first binding canon came in 1546" is too simple.  A Catholic theological account makes the distinction more precise.  George Reid's 1908 Catholic Encyclopedia entry calls Florence's decrees infallible statements according to the common teaching of theologians, but distinguishes their teaching about inspiration from Trent's formal declaration of canonicity.⁷¹  The Lutheran-Roman Catholic Commission on Unity's 2006 study explains that Trent's participants appealed to Florence as an earlier decision they were bound to respect.⁷²  These accounts give Florence real doctrinal force;  they do not treat the list as a nonbinding suggestion.  Florence professed the books as inspired, and Trent explicitly defined them as sacred and canonical under anathema.

Trent was answering the Reformation.  In his 1522 New Testament, Martin Luther moved Hebrews, James, Jude and Revelation to the end and left them unnumbered, and he called James "an epistle of straw" compared with the best books, a sentence later editions dropped.¹  Luther's doubts about the New Testament did not last among Protestants;  Beckwith notes that they "had no permanent influence, even among Lutherans."⁷³  His view of the Old Testament did.  The Thirty-Nine Articles of 1563 kept the extra Old Testament books for the church to "read for example of life and instruction of manners; but yet doth it not apply them to establish any doctrine".⁷⁴  The Westminster Confession of 1646 was stricter:  "The books commonly called Apocrypha, not being of divine inspiration, are no part of the canon of the Scripture".⁷⁵

The Orthodox answer came at the Synod of Jerusalem in 1672.  It named Wisdom, Judith, Tobit, the History of the Dragon, Susanna, the Maccabees and Sirach, and said:  "For we judge these also to be with the other genuine Books of Divine Scripture genuine parts of Scripture."⁷⁶  That synod was local.  No council that all Orthodox churches recognise as ecumenical has fixed the total number of books.

What councils did

The pattern from 393 to 1672 has two sides.  A council writes down the books its churches already read, often in response to a dispute, and claims to be passing on what it received.  Roger Beckwith, an evangelical scholar of the Old Testament canon, puts it this way:  "The role of councils, therefore, was not so much to decide the canon as to confirm decisions about the canon already reached in other ways."⁷³  Wenham agrees:  "There was no noticeable change of attitude before and after the Festal Letter of Athanasius, no flourish of trumpets at Hippo or Carthage."¹  The early use of the core supports that judgement.  But at the edge, a council's list did more than record.  It settled, for its own churches, books that had been doubted, such as Hebrews in Africa and the extra Old Testament books, and it bound clergy and readers who might otherwise have chosen differently.  In the exclusive sense of the word, regional councils did help make canons, each for its own churches, and Trent formally defined the Roman Catholic list.  Michael Kruger writes that "If the closing of the canon refers to a formal, official act of the early church, then we are hard pressed to find such an act before the Council of Trent in the sixteenth century."⁴  That formulation needs a qualification:  the African councils and Florence are earlier formal acts, though none secured agreement among all churches.  Whether a council's act makes a book Scripture is a different question.  Catholic teaching itself says the church receives books already inspired by God, rather than conferring inspiration by approving them.⁷²

Part 6: The Old Testament

The Jewish Scriptures in Jesus's time

One of the earliest references outside the Bible to a collection of Jewish Scriptures in three parts is in the prologue to Sirach, written by the author's grandson in Egypt after 132 BC.  He speaks three times of "the Law and the Proph- ets and the others that followed them", "the Law and the Prophets and the other ancestral books", and "the Law itself and the Prophets and the rest of the books".⁷⁷  The first two parts have names.  The third does not, and the prologue does not say which books it holds.  Luke has Jesus speak of the law of Moses, the prophets and the psalms (Luke 24:44), which is also a three-part description.

Josephus, a priest writing in Rome around 95, gives the first count:

"Our books, those which are justly accredited, are but two and twenty, and contain the record of all time. Of these, five are the books of Moses ... the prophets subsequent to Moses wrote the history of the events of their own times in thirteen books. The remain- ing four books contain hymns to God and precepts for the conduct of human life. From Artaxerxes to our own time the complete history has been written, but has not been deemed worthy of equal credit with the earlier records, because of the failure of the exact succession of the prophets."²⁴

Josephus does not name his 22 books.  His editor, H. St J. Thackeray, suggests that the thirteen are "Probably" Joshua, Judges with Ruth, Samuel, Kings, Chronicles, Ezra with Nehemiah, Esther, Job, Isaiah, Jeremiah with Lamentations, Ezekiel, the Twelve and Daniel, and that the four are Psalms, Song of Songs, Proverbs and Ecclesiastes.²⁴  That reconstruction gives exactly the books of the Protestant Old Testament.  It is a reconstruction, not a list Josephus wrote.  The alternatives proposed mostly move the same books between his groups, for example Job among the thirteen rather than the four.  A set of 22 that swapped in a book such as Sirach or Wisdom would clash with his own rule, since those books were known to be later than Artaxerxes.  The weak point is that his count needs some books combined, and a different combination could drop one book, such as Esther or the Song, and add another.  So the reconstruction is probable, not proven.  His stated criterion is clear:  books written after Artaxerxes, the Persian king of Ezra's and Nehemiah's day, do not have the same standing, because the exact succession of prophets had failed.  That is Josephus's belief about when the books were written, not a demonstrated date for each of them.

Two more facts bear on how far back his count reaches.  He presents the collection as long closed:  "although such long ages have now passed, no one has ventured either to add, or to remove, or to alter a syllable".²⁴  And he learned the Scriptures in Jerusalem before the war of 66-70.  He was born "in the year in which Gaius Caesar became Emperor", AD 37, to a father who was "among the most notable men in Jerusalem", and he says that at about fourteen "the chief priests and the leading men of the city used constantly to come to me for precise information on some particular in our ordinances".²⁴  So his 22 books are the view of a Jerusalem priest educated a generation after Jesus, before the rabbinic discussions that followed the Temple's fall.  The boast about the age of the collection is an apologist's, and the rabbinic debates below show that a few books were still argued over.  It does show that by about 95 a Jewish priest could treat a fixed number of books as old and settled, and expect his readers to accept it.

The apocalypse called 4 Ezra, written around 100, gives the other Jewish count.  Ezra dictates books for forty days, and God says:  "The twenty-four books that thou hast written publish, that the worthy and unworthy may read (therein): but the seventy last thou shalt keep, to deliver them to the wise among thy people."²⁵  The total is 94 in the critical text.  The King James Apocrypha prints "two hundred and four" and the Clementine Vulgate "nine hundred and four", which shows how easily numbers change in copying.⁷⁸⁷⁹  The editor G. H. Box notes that "The number 24 is the ordinary reckoning of the O.T. books (5+8+11)", while 22 combines Ruth with Judges and Lamentations with Jeremiah.²⁵  The passage attests 24 as the number of public books about 100.  It does not say these were the only inspired books.  The same Ezra dictates seventy more, to be kept for the wise, "For in them is the spring of understanding, the fountain of wisdom, and the stream of knowledge."²⁵  The partition is between public and hidden books, and the author of 4 Ezra, whose own book is outside the 24, plainly valued the hidden ones.

A teaching in the Babylonian Talmud, attributed to the earlier rabbis, gives the order.  The Prophets are "Joshua and Judges, Samuel and Kings, Jeremiah and Ezekiel, and Isaiah and the Twelve Prophets".  The Writings are "Ruth and the book of Psalms, and Job and Proverbs; Ecclesiastes, Song of Songs, and Lamentations; Daniel and the Scroll of Esther; and Ezra and Chronicles."²⁷  Five books of Moses, eight Prophets and eleven Writings make 24.  Counted as Protestants count them, splitting Samuel, Kings, Chronicles and Ezra-Nehemiah into two each and the Twelve into twelve, they make 39.⁸⁰

There was no Council of Jamnia

Many textbooks used to say that rabbis meeting at Jamnia, or Yavneh, around 90 closed the Jewish canon.  The idea goes back to Heinrich Graetz in 1871.⁷³  Jack Lewis's review of the evidence, published in 2002, found that the only early text is a passage in the Mishnah, edited around 200:  "M. Yadayim 3:5 remains the only text dating within a hundred years of Jamnia which speaks of discussion of biblical books."²⁶  That passage reports:

"All the Holy Scriptures defile the hands. The Song of Songs and Kohelet (Ecclesiastes) defile the hands. Rabbi Judah says: the Song of Songs defiles the hands, but there is a dispute about Kohelet. Rabbi Yose says: Kohelet does not defile the hands, but there is a dispute about the Song of Songs."²⁷

"Defiles the hands" was the rabbis' way of marking a book as holy.  Rabbi Shimon ben Azzai then cites a ruling "on the day when they appointed Rabbi Elazar ben Azariah head of the academy".²⁷  So a session of the academy discussed two books.  It did not issue a list, and it was not a council.  Roger Beckwith adds that the rabbis who still questioned Ecclesiastes and the Song lived in the later second century, and those who questioned Esther in the third, so "either this evidence shows that the canon was still open long after AD 90 (incredibly long afterwards!), or it does not show that the canon was open at all."⁷³  In the Babylonian Talmud, Shmuel, who died about 254, is quoted as saying, "The book of Esther does not render the hands ritually impure."²⁷  The Talmud itself asks whether this means Esther was not inspired, and answers no:  Shmuel elsewhere said Esther was given by the Holy Spirit, and the answer is that "It was stated with the Divine Spirit that it is to be read in public; however, it was not stated that it is to be written."  On the same page Rabbi Shimon ben Menasya says Ecclesiastes does not defile the hands "because it is the wisdom of Solomon, and not divinely inspired", and is answered that it was.²⁷  So some rabbinic doubts were about inspiration, and some were about ritual status, and the rabbis themselves kept the two apart.  Lewis reports that even Sundberg, who dated the closing of the Jewish canon late, "recognizes that the 'Council of Jamnia' hypothesis is dead."²⁶

Two opposite errors are possible here.  One is to say Jamnia closed the canon.  The other is to say the Jewish canon was so fixed that no Jew questioned any book.  Beckwith himself concedes that if a closed canon means "that no individual ever again questions the right of any of them to its place in the Bible, the canon of neither Testament has ever been closed, either among Jews or among Christians."⁷³  The evidence supports a middle position.  By the late first century, Jewish writers spoke of a fixed number of books, 22 or 24.  Debate about a few of those books went on among the rabbis for two centuries, and not all of it was about whether they were Scripture.  Books written later, such as Ben Sira, were placed outside:  "The scrolls of Ben Sira and all scrolls that are written from that point forward -- they do not defile the hands."²⁷

What was read at Qumran

The scrolls from the Judaean Desert show what Jews there were copying.  The table below comes from Eugene Ulrich's 2010 Index of Biblical Manuscripts and Editors, which principally lists Hebrew manuscripts from the eleven Qumran caves.  It also includes a few fragments without a known cave designation.  It counts only books of the traditional Hebrew Bible, excludes translations, and depends on editors' decisions about which fragments belong to one scroll.  Ulrich's separate Index of Biblical Passages has a wider scope, including Greek and Aramaic translations and other Judaean Desert sites;  the table does not use that wider index.⁸¹⁸¹

Ulrich's index covers only books of the traditional Hebrew Bible.  The caves also held many copies of books outside it.  J. T. Milik identified eleven manuscripts of the Enoch literature in Cave 4 alone.⁸²  Those eleven are copies of several Enoch books, some with one book each and some combined, not eleven copies of the later 1 Enoch, so they are not counted on the same basis as Ulrich's figures for single biblical books.  Even so, the Enoch writings were copied about as often as Daniel or the Twelve, while no copy of Esther appears in Ulrich's index.  This is a real problem for anyone who reads copying as canonicity.  The finds preserve, mostly, the library of one group in one region.  They show what that group copied, not what all Jews held to be Scripture.  The absence of Esther is not a measured rejection:  it may reflect that group's views, a short book's lower chance of survival, or chance.

Enoch, Jude, and quotation

The letter of Jude quotes a prophecy of "Enoch, the seventh from Adam" (Jude 14-15), which matches 1 Enoch 1:9.  Some argue that Jude thereby made Enoch Scripture.  Others argue that Enoch must be ancient because Jude quotes it.  Neither follows.

A quotation does not make a book canonical.  Paul quotes the poet Menander in 1 Corinthians 15:33, and Luke reports him quoting Aratus in a speech at Athens (Acts 17:28), without either treating the poets as Scripture.  Jude's case is harder, because he says Enoch "prophesied".  That is a claim about one saying, not about a whole book, and early Christians were divided over what it meant.  Jerome reports that Jude was "rejected by many" because of the quotation, and that "by age and use it has gained authority".¹⁵  The Ethiopian Orthodox Church, by contrast, does count 1 Enoch as Scripture.²⁸

A further argument is sometimes made in apologetics:  that 1 Enoch survives only in Aramaic and Greek, not in any older language, so it cannot go back to the real Enoch.  That inference does not work, because the language of a copy says nothing about the language of its source, and it is not needed for the canon question.  The facts are these.  The complete book survives only in Ge'ez, the classical Ethiopic language.  The oldest copies are Aramaic fragments from Qumran.  Milik dated the first copy of the Book of Watchers, which contains chapter 1, to "the first half of the second century B.C." and showed that "from the first half of the second century B.C. onwards the Book of Watchers had essentially the same form as that in which it is known through the Greek and Ethiopic versions."⁸²⁸²  The date of the oldest copy is a latest date for that section, not the date it was written.  The book is a collection:  R. H. Charles judged that chapters 1 to 5 "do not come from the same author or period as vi—xxxvi", and no fragment of the Parables, chapters 37 to 71, has been identified at Qumran.²⁵⁸²  Missing fragments do not prove the Parables were absent from every copy once there, but they leave that section without early attestation.  For the canon question the point is narrower.  Jude's quotation can show at most that he took one saying as true prophecy.  It cannot show who wrote the whole collection or that Jude counted all of it as Scripture.

The Christian Old Testament divided early

Many of the first Christians read the Jewish Scriptures in Greek, and Greek Bibles carried more books.  But the Greek Bibles were not one fixed collection either.  Codex Vaticanus has no books of Maccabees.  Sinaiticus has 1 and 4 Maccabees, Tobit, Judith, Wisdom and Sirach.  Alexandrinus has 1 Esdras, four books of Maccabees and Psalm 151, and its table of contents lists the Psalms of Solomon.⁶⁰⁵⁵⁵⁶  There was no single "Septuagint canon".

When early Christian writers made lists, many of them followed the Jewish count:

1. Melito of Sardis, about 170, went "East" to "the place where these things were preached and done" to learn "accurately the books of the Old Testament". His list has no Esther. The standard translation renders one item "the Proverbs of Solomon, Wisdom also", but the Greek, Solomōnos Paroimiai hē kai Sophia, more naturally means Proverbs, also called Wisdom: one book, not two. Eusebius himself says that Irenaeus "and the whole company of the ancients, called the Proverbs of Solomon All-virtuous Wisdom."⁶⁶⁸³
2. Origen, about 230, says "the canonical books, as the Hebrews have handed them down, are twenty-two", puts the Letter of Jeremiah with Jeremiah, and says "besides these there are the Maccabees".¹² His list as preserved names 21 units, not 22, because the Twelve Prophets are missing, apparently by a copying slip.⁸⁴
3. Athanasius, 367, counts 22 books but includes Baruch and the Letter with Jeremiah and leaves Esther out. He puts Wisdom, Sirach, Esther, Judith and Tobit among books "appointed by the Fathers to be read by those who newly join us".¹⁴¹⁴
4. Cyril of Jerusalem, about 350, counts "the twenty-two books of the Old Testament" with Baruch and the Letter.¹⁶
5. Jerome, about 400, translated from the Hebrew. He wrote that the church reads Judith, Tobit and Maccabees without receiving them among the canonical Scriptures, and that Wisdom and Sirach may be read "for the strengthening of the people, (but) not for confirming the authority of ecclesiastical dogmas."²⁰
6. Rufinus, about 400, called the same books "not Canonical but Ecclesiastical".⁵⁷

Other writers followed the wider Greek use.  Augustine, in the 390s, listed "forty-four books" of the Old Testament, including Tobit, Judith, two Maccabees, Wisdom and Ecclesiasticus.³¹  The African councils did the same.²¹  Augustine also gave his rule for disputed books:  prefer "those that are received by all the catholic churches to those which some do not receive", and among the rest, "such as have the sanction of the greater number and those of greater authority".³¹  That rule describes reception, and on the Old Testament the churches of his day did not agree.

The modern lists compared

The Catholic count of 46 is the Catechism's.²⁹  The Ethiopian narrower list counts the 39 as 40 by splitting Proverbs, then adds "Enoch, Jubilees, Wisdom, 1 Esdras, Ezra Apocalypse, Judith, Tobit, Ecclesiasticus, Baruch, 'the rest of Jeremiah', book of Susanna, 'the rest of Daniel', 1 and 2 Maccabees" for 54, and with the 27 of the New Testament reaches 81.  A broader Ethiopian list reaches 81 differently, with 46 Old Testament books and a New Testament of 35 that includes church-order books.²⁸²⁸  R. W. Cowley, who studied the Ethiopian canon, notes that "the canon lists found in Ethiopian MSS and printed books present many variations".²⁸

The Orthodox Church in America says these books "are considered by the Orthodox to be fully part of the Old Testament because they are part of the longer canon that was accepted from the beginning by the early Church."³⁰  That is the church's own account of their authority, and it is one jurisdiction's statement.  As a claim about uniform early acceptance it needs qualifying:  the lists of Melito, Origen, Athanasius, Cyril and Jerome differ from the OCA's longer collection.  Some include books beyond the later Hebrew list, such as Baruch in Athanasius and Cyril;  several other books stand outside their counts.  The Catholic and Orthodox positions should not be merged either.  Rome has a defined list;  the Orthodox churches agree on the core of the longer canon but have no conciliar total.

The Old Testament differences all concern books outside the Hebrew Bible.  Every modern tradition accepts the books of the Jewish 24, though Melito and Athanasius left Esther off their lists.  The New Testament is not quite uniform either:  Cowley's broader Ethiopian list counts 35 New Testament books by adding church-order works to the 27, and the Syriac Peshitta tradition long kept 22.¹⁷²⁸  Cowley's figures are his 1974 report of Ethiopian lists, which he says vary;  they are not a current official count.²⁸  The main disagreement is over whether a second group of Jewish books from the last centuries BC, read by many early Christians, is Scripture or useful reading.

Part 7: Why these books? The case for their authority, tested

The history so far explains how the churches came to hold these books.  It does not yet say whether they were right.  This part examines four accounts of why these books, and only these, have authority:  an evidentialist argument from Jesus, the Catholic and Orthodox appeal to the church, Kruger's self-authenticating model, and the naturalistic accounts of Ehrman and McDonald.  Each Christian account needs at least one premise that history alone cannot supply.  The question is which premises each needs, and how well the historical record fits.  The verdicts below are about these accounts, not every possible one.

The evidentialist argument from Jesus

The Christian apologist Jonathan McLatchie, who calls himself both a Protestant and an evidentialist, has summarised his own route to the canon and recommends John Wenham's Christ and the Bible for the details.⁸⁵  In outline, the argument has four steps:

1. Establish who Jesus is from historical evidence, chiefly the resurrection.
2. Jesus treated the Hebrew Scriptures as having God's authority. So a person who accepts Jesus's authority should accept those Scriptures.
3. Jesus commissioned apostles as his witnesses and promised them the Spirit's guidance (John 14:26; 15:27; 16:12-15; Acts 1:8, 21-22).
4. So books that come from the apostles carry that authority. The early church's tests of apostolicity, which McLatchie defines as connection with an apostle, and of universal acceptance as genuine were the right tests. A book's place in the canon then depends on how confident we can be that it is apostolic in that sense.

This argument is not circular in the way it is sometimes said to be.  It uses the Gospels and Acts as historical sources, which is how any historian uses ancient texts, not as inspired Scripture.  Its conclusion that they are Scripture comes at the end, not the start.  McLatchie is himself a sharp critic of circular arguments for the canon.  He calls Michael Kruger's self-authenticating model "epistemologically muddled" and says it begs "the question in favor of the New Testament canon".⁸⁶

Step 1 is argued elsewhere on this site, in articles such as Hallucination Theories of the Resurrection.  The question here is what follows if it is granted.

Step 2 is strong for the core of the Old Testament and depends on Jewish evidence at its edges.  The Gospels show Jesus citing Moses, the Prophets and the Psalms as authoritative.  They never show him listing books.  Wenham argues that Jesus's phrase "from the blood of Abel to the blood of Zechariah" spans "the whole sweep of biblical history from the first book of the Hebrew Canon to the last", since Zechariah's murder is told in Chronicles, which ends the Hebrew Bible.¹  The Talmud's list does put Chronicles last.²⁷  But that list was written down later, and Matthew calls Zechariah "the son of Barachiah" while Chronicles names his father Jehoiada (2 Chronicles 24:20), so the identification is debated.  The firmer evidence is Josephus, a contemporary Jewish priest, who speaks of a fixed set of 22 books closed at Artaxerxes and says that "it is an instinct with every Jew, from the day of his birth, to regard them as the decrees of God".²⁴²⁴  If Jesus shared the Scriptures of the Judaean Jews of his day, those were probably the books in Josephus's count, and the usual reconstruction of that count is the Hebrew Bible.²⁴  Two inferences sit in that sentence.  One is that Josephus's late-first-century count held fifty or sixty years earlier, in Galilee and Judaea.  The other is that the reconstruction is right at its edge, for books such as Esther and the Song that the rabbis still debated.  Both are reasonable;  neither is shown.  For the extra books the argument is weaker still.  Jesus is never shown discussing them, and the New Testament has possible echoes of Wisdom and Sirach but no quotation introduced as Scripture.  So the argument supports the Jewish and Protestant core well, and the exclusion of the extra books only as far as Josephus's testimony can be carried back to Jesus.

Step 3 depends on how reliable John's Gospel is.  The promise that the Spirit would "teach you all things and bring to your remembrance" is in John's farewell discourse.  Critics who doubt that John records Jesus's words will doubt this step.  The argument can lean instead on the apostles' role as witnesses in Acts and in Paul's letters, which is broadly attested.  Even Wenham admits the promises "do not of course refer specific- ally or exclusively, to the inspiration of a New Testament Canon, but they provide in principle all that is required for the formation of such a Canon".¹  So the step from the apostles were authorised witnesses to what they wrote is Scripture is an inference, and it needs stating as one.

Step 4 works for the core and needs a further link for each edge book.  Paul calls himself an apostle in seven letters that almost all scholars accept as his.  The Gospels of Matthew and John carry apostles' names, and Mark and Luke are tied by early tradition to Peter and Paul.  The evidence is weighed in Who Wrote the Gospels?.  Four things must be kept apart here:  that an apostle wrote a book, that an apostle approved it, that it came from the apostles' circle, and that we know the author's name.  An anonymous book need not be unconnected to the apostles.  For each edge book the argument needs a specific extra link.

1. Hebrews is anonymous. The early church did not know who wrote it. Origen said "God knows", and the Roman church doubted it because it was not thought to be Paul's.¹²¹³ If apostolic authorship is the test, Hebrews fails it. If the test is connection with an apostle, Hebrews needs the link that its author belonged to Paul's circle, which its mention of Timothy as a companion (Hebrews 13:23) makes likely, and that the apostles approved it, which no surviving source records. Its earliest known reader is the Roman letter 1 Clement, which Lake dates to between 75 and 110 and most probably the 90s, and which repeats Hebrews' opening chapter.³⁴³⁴ Wenham's comment on its history is candid: "It looks as though Hebrews was accepted at an early date in Rome, although known not to be Pauline; it was later doubted because its apostolic authorship was in doubt; later still it was accepted again because it was believed to be Pauline!"¹
2. 2 Peter needs the link that Peter wrote it, perhaps with a different secretary, as Jerome suggested. Its early attestation is thin: no quotation in Origen's surviving Greek, and Origen called it "doubtful".¹²³ Jerome reports doubts on the grounds of style.¹⁵ 2 and 3 John are short, rarely quoted and also doubted, so the comparison among these small books is close; 2 Peter is the one whose authorship was most openly questioned.
3. Revelation's author calls himself John but not an apostle. It needs the link that this John was the apostle, or that a prophet called John had apostolic standing. The oldest testimony supplies that link: Justin, about 155-160, calls him "one of the apostles of Christ".⁵³ That statement is a century older than Dionysius's doubt, but it is one sentence, and Justin does not say where he learned it. Dionysius of Alexandria denied on grounds of style that he was the apostle, while still treating the book as inspired.²
4. James and Jude were by brothers of Jesus, if the traditional attribution holds. Paul appears to count James among the apostles (Galatians 1:19), but whether James wrote the letter was doubted in the early church.¹⁵
5. The disputed letters of Paul are received as Paul's, but Did Paul Write the Disputed Letters? leaves Ephesians and the Pastorals unresolved.

McLatchie's test of connection with an apostle is close to Wenham's own definition, and Wenham is frank about how far it can be checked.  He writes:  "The criterion of prophetic or apostolic authorship (taken in its simple and natural sense) does not cover all cases."¹  He widens apostolicity to "apostolic authorship or approval (perhaps only tacit approval) by one or more of the apostles".¹  He gives both sides of the evidence in one sentence:  "It is indeed extremely likely that Mark and Luke and Acts and Hebrews (to mention no others) had the warmest apostolic approval, but that there was direct historical proof of that fact a hundred years later is quite uncertain."¹  His reasons for "extremely likely" are that the apostles' circle was "a readily recognizable entity" and that books from it were assumed to be true to the apostles' teaching.  A lack of direct proof is not a failure of inference:  historians reason to approval from association, early use and the absence of protest, and that reasoning can be strong.  But it is a probability for each book, and it differs from book to book.  His final ground is providence:  "Grounds of canon- icity are to be found in an interplay of subjective and objective factors over-ruled by Divine Providence."¹

Closure.  Paul wrote a letter to Corinth before 1 Corinthians, on the usual reading of 1 Corinthians 5:9, and Colossians 4:16 mentions a letter "from Laodicea" without naming its author;  it may be a lost letter of Paul or a known one under another name.  It is sometimes argued that the apostolic test cannot explain closure, because a lost apostolic letter would pass it.  That is not the test Wenham uses.  He says a book "had in practice to pass two tests before gaining canonical recog- nition. It had to pass the test of continuous and widespread approval amongst Christians, otherwise its candidature would have automatically disappeared", and adds that "It is possible that many writ- ings of the apostolic circle never survived."¹  Kruger likewise says that lost apostolic books need not be called canon.⁴  On these accounts apostolic origin alone does not establish membership in the church's permanent collection.  Reception and preservation also matter.  The historical fact that some books survived cannot by itself show that God intended exactly those books to survive as Scripture;  that connection depends on providence.  A newly found, demonstrably authentic apostolic letter would sharpen the question of why continuous reception is required.  The passages cited here do not answer that hypothetical case in detail.

So the evidentialist argument supports a strong conclusion about a large core:  the books of the Jewish Scriptures in Josephus's count, the four Gospels, Acts, Paul's undisputed letters, 1 Peter and 1 John.  For several other books it supports a weaker conclusion, that they come from the apostolic circle and were received early, with differing degrees of confidence.  Getting from there to exactly 27 books requires either trust in providence, which is a theological premise, or trust in the church's judgement, which is the next account.

The church's authority

Roman Catholic teaching rests the canon on tradition.  Vatican II's Dei Verbum says:  "Through the same tradition the Church's full canon of the sacred books is known".³²  The Catechism adds:  "It was by the apostolic Tradition that the Church discerned which writings are to be included in the list of the sacred books."²⁹  Catholic teaching distinguishes recognising an inspired book from making it inspired by approval.⁷²  The Orthodox Church in America gives a related reason for its longer Old Testament:  the books are "part of the longer canon that was accepted from the beginning by the early Church."³⁰  The two are not the same model.  Rome has a defined list and a teaching office that defined it;  the Orthodox churches appeal to the church's received use without a single conciliar total.

This account explains the boundary.  The church received certain books, and its judgement fixes the edge where history alone cannot.  It fits the evidence that reception, and not only a theory of authorship, decided some hard cases.  Dionysius kept Revelation partly because "many brethren hold it in high esteem", though he also judged it the work of an inspired man;  and Jerome received Hebrews and Revelation by following "the authority of ancient writers".²³  Those are examples of reception counting.  They do not by themselves show that the church's reception is protected from error, which is the further claim.

It has two costs.  The first is that premise, that the church's collective judgement is reliable or protected from error, and it needs its own argument.  The second is historical.  The church did not speak with one voice for centuries.  On the Old Testament, Jerome and Athanasius disagreed with Augustine and the African councils, and Eastern and Western churches still differ.  The appeal to "the church" has to say which church, at which date, and why that body's judgement and not another's is the one protected.  Catholic theology answers by locating the authority in councils with the bishop of Rome, which explains why Florence and Trent matter to it;  Orthodox theology answers through the consensus of the churches.  Those answers are theological, and history can test only whether the bodies said what is claimed.

The self-authenticating model

Michael Kruger's model is often described by its critics, so it should be stated in his words first.  He writes that believers know the Bible is God's word "from the attributes and characteristics present in the Bible itself", and that "the Bible bears evidence within itself of its own divine origins. This is what we mean when we say that the Bible is self-authenticating."³³  In Canon Revisited he names three marks of a canonical book:  divine qualities, apostolic origins and reception by the church as a whole, with the Holy Spirit enabling believers to recognise the first.⁸⁶  He presents this as evidence available in the books and in history, not as a private feeling, and he lists the charge of subjectivism as an objection he answers.³³  Separately, in The Question of Canon, he defends a historical "intrinsic model", which he says "does not require a commitment to any particular theological perspective":  the idea of a canon grew from inside the earliest churches rather than being imposed later.⁴

Jonathan McLatchie, an evidentialist who shares Kruger's conclusions, objects to the epistemic model.  He calls it "epistemologically muddled" and says it begs "the question in favor of the New Testament canon".⁸⁶  His specific objections are that "many believers (myself included) do not have any conscious awareness of an internal witness to Christianity's truth", and that "people of other religious faiths (such as the Mormons) claim similar internal experience."⁸⁶  Those objections bear on the role of the Spirit's witness.  They bear less on Kruger's other two marks, apostolic origins and corporate reception, which are open to historical argument and overlap with the evidentialist case.  The divine qualities themselves can be assessed by an outsider only as claims about the books' content.  Whether they are there is a judgement a skeptic is entitled to dispute;  a skeptic's refusal, though, is not a test of the claim.  The historical intrinsic model is testable, and the evidence of Part 2 supports it for the core:  Irenaeus, Tertullian and Serapion were treating apostolic writings as authoritative and ruling others out long before the fourth century.

The naturalistic accounts

The scholarly naturalistic accounts do not say that Constantine chose the Bible.  Four of them, in their authors' words, are these.

1. Harnack: the canon as a reaction to Marcion. The canon was the church's "creative act" in response to Marcion.⁴
2. Ehrman: the canon as a proto-orthodox victory. The proto-orthodox, one Christian party among several, "collected other early writings into a sacred canon of Scripture to advance their" views, and "The victory of proto-orthodox Christianity in its quest for dominance left a number of indelible marks on the history of Western civilization. Of these, none has proved more significant than the formation of the New Testament as a canon of Scripture."³⁹³⁹ Ehrman names four criteria the proto-orthodox used: a book had to be ancient, apostolic, accepted across the churches and orthodox. Orthodoxy was the most important, and "To some extent, in fact, the other criteria were handmaidens to this one."³⁹
3. McDonald: closure as a fourth-century response to pressure. "Currently it appears that at no time before the fourth century did the church consciously make decisions about what literature was “canonical” and what was not." Diocletian's persecution and Constantine's push for unity "provide the social context that led to the clos- ing of the biblical canon."³⁶³⁶
4. Sundberg and Hahneman: no canon until the fourth century, on the exclusive definition, with the Muratorian fragment moved to that century.⁴⁸⁴

None of these needs state power, doctrine as the only test, or equal survival of every gospel.  They have to be tested on what they say.

Ehrman's account fits much of the evidence.  Proto-orthodox leaders did select books, against real rivals, from the second century.  Irenaeus argued for four Gospels against groups that used others;  Serapion withdrew the Gospel of Peter;  the Muratorian writer refused "gall" mixed with honey.⁷¹²³  Doctrine could drive a judgement of authorship, as with Serapion.  Most surviving accounts of these decisions come from the tradition that prevailed.  Orthodox books being excluded does not refute Ehrman:  he himself names the Shepherd's recent date and the doubts about Hebrews' authorship.  His claim is that doctrine was the most important test and sometimes shaped the others, not that doctrinal agreement was sufficient for admission.³⁹

The harder question is whether doctrine explains the specific authorship disputes beneath their stated reasons.  Ehrman argues that Hebrews' severe view of repentance and Revelation's earthly thousand-year reign drove those debates.³⁹  There is evidence for a doctrinal motive in some opposition to Revelation:  Dionysius reports critics who attributed it to Cerinthus and his earthly kingdom.²  But Dionysius himself accepted it as inspired while arguing from style that its John was not the apostle.²  Thus authorship, inspiration and approval of teaching could come apart.  These cases require a book-by-book explanation.  They support doctrine's importance without proving that every authorship argument concealed a doctrinal motive.  They also do not establish the opposite, that the other criteria operated independently of doctrine across the whole process.  The surviving records show several criteria at work and cannot measure their relative causal weight.

McDonald's account fits the closing of the edge better than its start.  The surviving lists do cluster after 303:  Eusebius, Cyril, Athanasius, the councils.  It is plausible that persecution and imperial pressure for unity pushed churches to fix their lists.  But his claim that no conscious decisions about canonical literature were made before the fourth century is too strong.  He himself concedes that "Irenaeus spoke of a four-gospel canon, those four and no more".³⁶  Serapion made a decision about a gospel, and Tertullian reports that "every council of the churches" had judged the Shepherd apocryphal.⁴²¹²  Those are conscious decisions, by local and regional authorities, about which books could be read as Scripture.

Marcion's influence remains plausible.  As Part 2 explains, earlier use of Paul does not rule out a later reaction to Marcion in forming a closed collection.  Irenaeus also opposed several other rivals.  The sources examined here support an existing stock of valued writings followed by sharper boundaries;  they do not isolate how much Marcion caused that change.  Harnack's full argument has not been assessed directly here.

The copying evidence bears on one narrow question.  Hurtado's predominantly Egyptian sample preserves substantially more witnesses to the four Gospels than to the other gospels combined.⁴¹  If survival broadly reflects use, that supports wider use of the four in that region.  The sample's uneven survival and dates extending to about 325 prevent a precise comparison of copying rates before Constantine.  The direct second-century witnesses provide the firmer evidence against a fourth-century invention of the core.

The naturalistic accounts are right that the canon was formed by people, over time, with disagreement and with doctrinal tests.  The evidence does not fit a fourth-century invention of the core, which is not Ehrman's account.  His account explains much of the selection process;  the claim that doctrinal motives underlay particular authorship disputes needs testing against each writer's reasons and conduct.  Whether the human process was also guided by God requires a further argument connecting the historical evidence to God's purposes.

Objections and responses

1. "The Council of Nicaea voted on which books went into the Bible."  The twenty canons of Nicaea say nothing about it, Constantine's order for fifty Bibles names no books, and the altar story is first found in a ninth-century work.¹⁸¹⁹⁶⁶  Jerome's later report that Nicaea "counted" Judith among the Scriptures is the only ancient link, and it concerns one Old Testament book.²⁰

2. "The winners chose the books and burned the losers."  Proto-orthodox bishops and teachers did choose, from the second century on, and doctrine was one of their tests.  Imperial suppression of books is also documented:  Diocletian targeted the church's own Scriptures in 303, and Constantine ordered Arius's treatises burned.⁶⁷⁶⁸  Neither is an order to destroy gospels excluded from the canon, and the ancient texts examined here record no such order.  That limits what can be claimed about a canon-related burning;  it does not show that rejected books escaped every kind of suppression or neglect.

3. "There was no canon until the fourth century, so the early church had no New Testament."  On the exclusive definition the first part is true of the full list, though Marcion and perhaps the Muratorian writer had closed lists of their own.  The second part does not follow.  Justin describes Gospels read in Roman worship about 155, Irenaeus defended four Gospels and quoted most of Paul's letters as Scripture about 180, and Tertullian did the same in Carthage.  What came in the fourth century was the closing of the disputed edge, not the core.

4. "Councils had nothing to do with it;  they only rubber-stamped."  The core was widely read before the surviving conciliar lists.  Councils later fixed lists where usage varied, bound their own churches, and in Trent's case defined the list under anathema.  Saying that councils did nothing confuses the earlier use of books with the later establishment of binding boundaries.

5. "The Muratorian fragment proves a second-century canon."  Its date is disputed between about 200 and about 375.  On the arguments read here the early date is better supported.  Rothschild's forgery thesis has been tested only through Guignard's reply, which confines a forgery to before about 400 and argues for a Greek original written in the West.  The second-century core does not depend on the fragment.

6. "Origen listed all 27 books around 240."  The homily survives only in Rufinus's Latin, the critical text does not name Revelation, and Origen's own Greek calls 2 Peter doubtful and 2 and 3 John disputed.¹²⁵⁹  Rufinus said he translated these homilies as he found them, so the homily is still evidence that Origen named the other books in a sermon.⁵⁸

7. "The Septuagint was the Bible of Jesus and the apostles, so the Catholic or Orthodox Old Testament is the original."  There was no single Septuagint list.  The great Greek codices disagree about which extra books they carry, and Jewish writers of the first century, Josephus and 4 Ezra, counted 22 or 24 books.²⁴²⁵  Christian writers who studied the question, Melito, Origen, Athanasius, Cyril and Jerome, followed the Jewish count with small variations, while Augustine and the African councils did not.

8. "The Protestant Old Testament is a sixteenth-century novelty."  Its books match the usual reconstruction of the Jewish count attested by the first century, and Jerome held the same view around 400.²⁰  What is new in the Reformation is a binding confession that the other books are "no part of the canon", in answer to Trent's definition that they are.²³⁷⁵

9. "Jude quotes Enoch, so either Enoch is Scripture or Jude is wrong."  Jude quotes one saying and calls it prophecy.  That claims truth for the saying, not Scripture status for the whole collection.  The early church saw the difficulty:  Jerome reports Jude was "rejected by many" for this reason.¹⁵  The Ethiopian church draws the other conclusion and counts 1 Enoch.²⁸

10. "The argument from apostolicity is circular, because it uses the Gospels to prove the Gospels."  It uses them as historical sources first, and that is legitimate.  Its weakness is not circularity but reach.  It needs an extra link for each anonymous or doubted book, and it explains closure only by adding reception and providence.

11. "If the church took centuries to agree, the canon is arbitrary."  The New Testament disagreement concerned seven books that make up 13.5 percent of its Greek text, and Revelation alone is more than half of that.⁸⁷  That figure measures the size of those seven books, not how uncertain anything is.  The disputes were argued with stated reasons:  authorship, style, date and use.  A slow, reasoned process is not an arbitrary one, but it is not thereby a correct one either.  Whether it reached the right edge is the theological question of Part 7.

12. "The canon is obvious to believers;  Wenham concedes too much."  If it were obvious, Dionysius, Gregory of Nazianzus and the Syriac church would not have read different New Testaments, and Jerome would not have reported in 414 that the Latin and Greek churches each rejected a book the other read.²¹⁷³³  Wenham's concession is what the record shows.

Issues and unanswered questions

1. The Muratorian date. Rothschild's 2018 article and 2022 commentary were not obtained; her 2016 abstract was read, and Christophe Guignard's 2019 reply was read in full. Guignard answers her claim of Vulgate anachronisms in a 2015 article on the fragment's original language, which was not obtained, so that specific claim is still unexamined here. Later reviews of her commentary were found but could not be downloaded. Lemarié's identification of Chromatius's borrowing, on which Guignard's latest date rests, is reported through Guignard. Hahneman's case was read in his own 2002 chapter and the opening of his monograph. Sundberg's 1973 article and Ferguson's 1982 rebuttal were not obtained.
2. Origen's homily. Baehrens's sigla were read in his preface, but the manuscripts themselves were not. Rufinus's statement about his method is general, so it lowers but does not remove the chance of an update at this passage.
3. Florence and Catholic doctrine. Reid's Catholic account distinguishes Florence's binding teaching on inspiration from Trent's formal declaration of canonicity. The 2006 Lutheran-Catholic study explains Trent's appeal to Florence. These treatments clarify the distinction; the article does not survey every Catholic theological assessment of the decree.
4. The Orthodox Old Testament. One official Orthodox statement, from the Orthodox Church in America, was read. The Greek, Russian and other churches' current lists were not, and a 2023 OCA document that may say more was found but not read.
5. The Ethiopian canon. Cowley's 1974 article was read in a web transcription without page numbers. No current statement from the Ethiopian church was read.
6. The Catalogus Claromontanus. This list, perhaps from about 300, places Barnabas, the Shepherd, the Acts of Paul and the Apocalypse of Peter with the New Testament books, apparently marking some with dashes. The marks could not be checked against the manuscript, so the list is left out of the table.
7. The Decretum Gelasianum, a list often attributed to a Roman council of 382, is now generally dated to the sixth century. Its editor's dating was not checked in the original German for this article, so it is not used.
8. The African councils. The 393, 397 and 419 forms are known here from Westcott and the English of the 419 code, not from Munier's critical edition.
9. Josephus's 22. Thackeray's identification is a reconstruction. The case that Josephus meant exactly the books of the Hebrew Bible is probable but not proven. Josephus was educated in Jerusalem before 70, so his count probably reflects priestly opinion there a generation after Jesus, but carrying it back to Jesus's own day is still an inference.
10. The Qumran counts come from one edition's manuscript index, principally Hebrew fragments from Qumran, with some uncertain identifications and fragments lacking a cave designation. They are not a census of all Judaean Desert finds, all languages, or all books the community might have treated as Scripture.
11. The positive case. This article tests the structure of the argument from Jesus's authority. It does not assess the historical case for the resurrection, which the argument needs as its first step.
12. Kruger and McLatchie. Kruger's Canon Revisited was not available; his model is stated from his own 2015 summary and his 2013 book, with McLatchie's review for the three marks. McLatchie's own framework was read in a short social-media post saved in the research notes.

Verdict

The New Testament core was in use well before the first surviving conciliar lists.  The four Gospels and most of Paul's letters were functioning as Scripture in Gaul, Carthage and probably Rome and Alexandria by about 200, and Gospels were already read in Roman worship beside the prophets about 155.  No council list survives from that time, and the surviving evidence places these choices within churches rather than at an emperor's command.  The choosing was done by bishops and teachers within their own churches and networks, and they were already ruling books out:  Irenaeus against rival gospels, Serapion against the Gospel of Peter, councils against the Shepherd.  Hurtado's predominantly Egyptian sample, with dates extending to about 325, preserves the canonical Gospels about four times as often as the other gospels combined.  That is a ratio of surviving witnesses, not a measured copying rate.

The idea that the whole canon was settled from the start is also false.  Seven New Testament books were disputed for two centuries or more, for stated reasons:  doubtful authorship, different style, limited use.  Respected bishops left Revelation out as late as 380, the Syriac churches read a shorter New Testament for longer still, and Jerome in 414 found the Latin and Greek churches each declining a book the other read.  Other books, such as 1 Clement and the Shepherd, were read in churches and later dropped.  The first surviving list of exactly the 27 is from 367.

Councils recorded what their churches already read, and at the edges they closed the list for their own regions:  Africa from 393, the Roman church with Florence's profession in 1442 and Trent's definition in 1546.  The Eastern churches ratified collections without fixing a total.  The Old Testament is still divided:  the Jewish and Protestant 24 or 39 books, the Catholic 46, and longer Orthodox and Ethiopian lists.

The case for the books' authority is a further step.  The argument from Jesus's authority, if its first premise is granted, supports the Jewish Scriptures of Josephus's count and a large New Testament core strongly.  For each edge book it needs an extra link, such as apostolic approval of an anonymous letter, which the evidence can make more or less likely but cannot show directly.  Every Christian account of the exact boundary examined here adds a premise, whether providence working through reception, the church's authority or the qualities believers find in the books, that history can make reasonable but cannot prove.

Sources