Ancient Citation and Authorship Conventions

"Far be it from us to speak thus of a follower of Christ, who made it his care to formulate dogmas rather than to hunt for words and syllables." -- Jerome on Matthew 27:9, AD 395¹

"For we, brethren, 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." -- Serapion of Antioch, about AD 200²

Summary

Skeptics often say the New Testament misquotes the Old Testament, credits quotations to the wrong prophet, and contains letters written by people pretending to be apostles.  Defenders often answer that "ancient conventions" were looser than ours.  Both sides usually run five different questions together.  This article separates them and tests each against what ancient writers actually did and said:

1. Wording. Did a quotation have to reproduce its source word for word?
2. Source names. When a writer names a source, how exact did the name have to be?
3. Interpretive use. When a writer says a text was "fulfilled", what is he claiming about that text?
4. Historical authorship. Who actually wrote a given book or letter?
5. Ethics. If someone wrote under another person's name, did ancient readers count it as honest?

The main findings, each argued below:

1. Wording was free, and that is documented, not assumed. Jewish texts from before Jesus quote scripture with blended verses, rearranged wording and text forms that differ from the standard Hebrew Bible. One Dead Sea Scroll strings five passages together with no source names, and its first "two" passages turn out to be a single passage in a text form then in circulation. Another names "the book of Ezekiel, the prophet" over wording, partly restored by its editors, that matches no single Ezekiel verse. Jerome in AD 395 stated the principle outright: the apostles cared for meaning, "rather than to hunt for words and syllables." A wording difference alone therefore does not establish an error. Whether a quotation preserves or changes its source's meaning remains a separate question.
2. Source names are a different matter, and three cases remain hard. Mark 1:2 credits to "Isaiah the prophet" a quotation whose first line comes from Exodus 23:20 and Malachi 3:1. Matthew 27:9 credits to "Jeremiah" words mostly from Zechariah 11. Mark 2:26 says David ate the holy bread "when Abiathar was high priest", where 1 Samuel 21 names Ahimelech. A possible pre-Christian parallel for naming a single prophet over a blended allusion is (2 Chronicles 36:21). It does not establish a rule, and no ancient writer examined here states one. Ancient readers noticed all three cases: copyists changed Mark 1:2 to "in the prophets", some dropped "Jeremiah", some deleted the Abiathar clause, and Jerome left two of the three unsolved. Each case has defensible explanations and each keeps a residue.
3. "Fulfilled" covers more than prediction. The Dead Sea Scroll commentary on Habakkuk says God told the prophet what would happen "but he did not let him know the consummation of the era", and that the community's Teacher was given "all the mysteries of the words of his servants, the prophets." Matthew's use of Hosea 11:1 ("out of Egypt I called my son") reads a past event about Israel as a pattern completed in Jesus. Paul calls his reading of Hagar and Sarah an allegory in so many words, and Philo, his Jewish contemporary, allegorised the same two women. These are interpretive claims by recognised methods of the time. A familiar method can still be misapplied. Its use alone neither verifies the application nor establishes that the writer misunderstood the source.
4. An anonymous book is not a forgery. The four Gospels, Acts, Hebrews and 1 John never name their authors. Even Bart Ehrman, the best-known writer arguing that some New Testament books are forgeries, classes them as "false attributions" at worst, not forgeries. Hebrews shows how such attributions grew: ancient readers credited it to Paul, to Paul translated by Luke, to Barnabas, to Clement or to Luke, and Origen judged its style unlike Paul's and concluded "who wrote the epistle, in truth, God knows."
5. Ancient readers condemned false authorship when they detected it. Galen approved a reader who rejected a book sold under Galen's name after reading two lines. Serapion rejected writings "falsely ascribed" to the apostles. Tertullian reports that the presbyter who composed the Acts of Paul out of "love of Paul" was removed from office. The Muratorian list rejects two letters "forged in Paul's name." No ancient source examined here treats a letter falsely bearing an apostle's name as an accepted literary convention. The nearest thing to tolerance comes from philosophical schools (Iamblichus, about AD 300, praises Pythagoreans who credited their work to Pythagoras) and from a fifth-century Christian who defended a pseudonym and still felt he had to explain why it was not a lie.
6. Secretaries and authorised agents were real, and their reach is limited. Tertius says he wrote Romans (16:22). Paul adds a greeting "with my own hand" in five letters, two of them disputed. Cicero asked his friend Atticus to "compose" letters "in my name" and to "say whatever you think right." Jerome explained the different styles of 1 and 2 Peter by Peter's use of different "interpreters." But the phrase "I wrote through Silvanus" (1 Peter 5:12) is the standard way of naming a letter's carrier, as an evangelical specialist argues against his own side's use of it. No example verified for this article demonstrates a secretary independently composing an extended argumentative letter on another person's behalf.
7. The disputed letters must be argued one by one. The Pastoral Epistles have real and concentrated peculiarities: five Greek words and phrases that appear 32 times in their 3,485 words and never in Paul's seven undisputed letters. But they also have early users (Polycarp of Smyrna echoes 1 Timothy 6 in the letter with which he forwarded Ignatius's letters), and Titus fits Paul's rate of rare words once length is controlled. 2 Peter has sparse and disputed early attestation, and ancient readers doubted it on grounds of style. Among the selected letter pairs compared here, the 1-2 Thessalonians pair has the second-highest density of shared five-word sequences, after Colossians-Ephesians; that fits imitation or one author writing twice to one church, and the count cannot tell which.
8. Short-text stylometry needs validation for the actual task. Eder's broad tests often needed thousands of words for stable attribution, but he also discusses successful shorter samples. A separate Latin verification study uses 1,000-word slices. The disputed letters range from 659 to 2,416 words. This article's exploratory common-word test separates Paul poorly from other New Testament letter writers, although it separates him better from Acts and Revelation. Its limited performance does not establish that every method must fail.

Verdict.  Documented practices of paraphrase, blending and authorised composition answer some objections to the New Testament.  They do not settle a disputed source name or author.  The ancient texts examined weigh against treating deceptive apostolic attribution as harmless, while the first audience's understanding of a proposed literary fiction remains a separate historical question.  Each disputed passage and letter needs its own evidence;  the stylistic results reviewed here do not supply a general verdict.

Part 1: Five questions that must be kept apart

A single verse can raise all five questions at once, and a verdict on one does not carry over to another.  Mark 1:2 is an example.  Its wording differs from every known Greek and Hebrew text of its sources (question 1).  It names Isaiah for a line that is not in Isaiah (question 2).  It applies to John the Baptist words first spoken about a messenger before the Lord in Exodus and Malachi (question 3).  The Gospel itself names no author (question 4).  And nobody claims Mark falsely posed as someone else (question 5).

The last two columns matter most.  An appeal to "ancient convention" is evidence only when an ancient writer, Jewish, Christian or pagan, can be shown doing the same thing or describing it.  A proposed convention inferred only from the same disputed attribution cannot independently justify that attribution.  Other contemporary examples, inside or outside the Bible, can help;  their similarity and independence need to be assessed.

Part 2: Wording was free, and the evidence is older than the New Testament

Which text was being quoted

In the first century there was no single text of the Jewish scriptures.  Greek translations, Hebrew text forms close to the later standard (Masoretic) text, and Hebrew forms close to the Samaritan Pentateuch all circulated, and the Dead Sea Scrolls contain all three kinds.  New Testament writers sometimes follow the Greek against the Hebrew and sometimes the Hebrew against the Greek.  The counts, and why they are smaller than often claimed, are set out in Which Bible Text Gives the Right Date for Creation?.  Three examples illustrate the textual and translational questions:

1. Mark 1:3 closely follows the Greek wording of Isaiah 40:3, "the voice of one crying in the wilderness." The Hebrew poetic parallelism favours reading "in the wilderness" with the command to prepare a way. Punctuation and syntactic interpretation matter here; this is not by itself evidence of different underlying Hebrew consonants.³
2. Matthew 2:15 follows the Hebrew of Hosea 11:1, "out of Egypt I called my son", where the Greek has "his children."⁴
3. Matthew 27:9-10 has the money thrown to "the potter", as the Hebrew of Zechariah 11:13 does, where the Greek has "the furnace."⁵

A difference from a modern English Old Testament may therefore reflect another text form, translation or paraphrase.  It need not be a misquotation, though an individual difference still needs investigation.

Quotation at Qumran

The Dead Sea Scrolls show Jews of the period quoting scripture in the same ways the New Testament does.  The two scrolls most used below are dated before Jesus:  by handwriting the Community Rule was copied about 100-75 BC and the Habakkuk commentary about 30-1 BC, and their radiocarbon ranges (104-43 BC for the commentary, 159 BC to AD 20 for the Rule, at one standard deviation) also fall before or around the turn of the era.⁶

1. A chain of passages with no source names. The scroll called 4QTestimonia (4Q175) quotes Deuteronomy 5:28-29, Deuteronomy 18:18-19, Numbers 24:15-17 and Deuteronomy 33:8-11, then a passage that also appears in a non-biblical "Psalms of Joshua" (4Q379), and names no book.⁷ Its opening looks like two Deuteronomy passages spliced under the heading "And the Lord spoke to Moses." But Emanuel Tov, the editor-in-chief of the scrolls' official publication, shows that the Samaritan Pentateuch and another Qumran manuscript carry Deuteronomy 18:18-22 inside the Sinai story of Exodus 20, and that "the sequence of the anthology of verses in 4QTest follows" that text form.⁸ The sequence therefore need not have been assembled by this scroll's compiler: it could follow a harmonised scriptural text already in circulation. That supports an alternative source form without establishing word-for-word identity with a lost exemplar.
2. A named book over blended wording. 4QFlorilegium (4Q174) introduces quotations with "as it is written in the book of Isaiah, the prophet" and "as is written in the book of Daniel, the prophet." It cites "the book of Ezekiel, the prophet" for words the edition restores as "[they should] no[t defile themselves any more with all] their [i]d[o]ls."⁷ No single verse of Ezekiel reads so. Ezekiel 37:23 has "they shall not defile themselves any more with their idols"; Ezekiel 14:11 has "nor defile themselves any more with all their transgressions."³ The named book is consistent with this wording. A variant, paraphrase or blend could account for the small difference; damage and restoration prevent a confident reconstruction of the quotation's exact form.
3. A prophecy applied to the writer's own movement. The Community Rule (1QS VIII 13-16) tells members to withdraw "to the desert in order to open there His path. As it is written (Isa 40:3): «In the desert, prepare the way of ****, straighten in the steppe a roadway for our God». This is the study of the law."⁷ The Rule follows the Hebrew line division, which puts "in the desert" with the preparing, and applies Isaiah's words to its own community. Mark applies the same verse to John the Baptist, a century or more later.

Mark 1:2-3, word by word

Mark opens:  "As it is written in Isaiah the prophet:  'Behold, I send my messenger before your face, who will prepare your way;  the voice of one crying in the wilderness:  Prepare the way of the Lord, make his paths straight.'"  The table sets each clause beside its closest source.³

Two further facts matter.  Matthew 11:10 and Luke 7:27 quote the same Exodus-Malachi blend, in nearly the same Greek, in a saying of Jesus about John, introduced only by "it is written."³  So the blend occurs as a unit in all three Synoptics.  It may have circulated as a proof text about John, although shared written use could also account for the agreement.  And Matthew 3:3 and Luke 3:4 quote Isaiah 40:3 alone under Isaiah's name and put the Exodus-Malachi line elsewhere.

Adams and Ehorn's 2016 introduction defines a composite citation as passages fused and presented as one, and separates it from an unmarked allusion or an openly signalled list of sources.  Their survey treats deliberate combination, memory and earlier extract collections as possibilities to investigate, rather than assuming one cause for every example.⁹  This supports studying the practice beyond the Bible;  it does not independently establish a rule for choosing the more important prophet's name.

An ancient reader states the principle

Jerome, the most learned biblical scholar of the fourth century, discussed these passages in a letter of AD 395 defending his own translations, which critics had attacked for not being word for word.  He said his purpose was "not to convict the evangelists of falsification — a charge worthy only of impious men like Celsus, Porphyry, and Julian — but to bring home to my critics their own want of knowledge."¹  Of Matthew 27:9 he wrote that the passage "is not found in Jeremiah at all but in Zechariah, in quite different words and an altogether different order", and that the apostle "made it his care to formulate dogmas rather than to hunt for words and syllables."¹

That is an ancient statement of a convention about wording, made by a scholar who knew Hebrew, applied to the apostles, and offered as the reason his own free translations were legitimate.  Together with the Qumran evidence, it settles question 1:  a New Testament quotation that keeps the sense of its source while changing, combining or condensing the words follows the practice of its time.  It does not settle question 2.  Jerome, in the same letter, could not explain the names.

Part 3: Source names, three hard cases

Is there a rule for naming one prophet over a blend?

A common defence of Mark 1:2 and Matthew 27:9 says that when ancient writers combined passages they named only the more important or better-known prophet.  For that to count as evidence, someone must show a writer before or near the New Testament doing it.  The evidence found:

1. One pre-Christian case. Robert Gundry, in the standard technical study of Matthew's quotations (1967), points to 2 Chronicles 36:21: the exile lasted "to fulfil the word of the Lord by the mouth of Jeremiah, until the land had enjoyed its sabbaths ... to fulfil seventy years."⁵ The checked Hebrew bears him out. The seventy years are Jeremiah's (25:11-12; 29:10), but the words "until the land had enjoyed its sabbaths, all the days of its desolation" are Leviticus 26:34-35, which Jeremiah does not contain.³ Jeremiah's name heads a sentence partly worded from Leviticus. But the sentence can also name Jeremiah only for the number. This is evidence relevant to a proposed convention, not proof of its frequency or a rule about the more important prophet.
2. A later rabbinic practice. Gundry adds that "it was a rabbinical practice to quote various persons under one name if a similarity existed between the characters or actions of the persons."⁵ His authority is a nineteenth-century guide to Talmudic usage. Rabbinic texts were written down from about AD 200 onward, so they cannot show first-century practice without more evidence.
3. Copyists, not authors. Bruce Metzger, the leading textual critic of his generation, records that scribes inserted "the name of the best known prophet" where Matthew named none, and that two witnesses even changed "Jeremiah" in Matthew 27:9 to "Isaiah", "perhaps because, as the most prominent of the prophets, his name is met with most frequently."¹⁰ That shows later copyists attaching famous names. It is not evidence about the evangelists.
4. What the comparable texts actually do. The Qumran texts above either name no source (4Q175) or name a compatible book over partly restored wording (4Q174). Matthew himself credits his blend of Isaiah 62:11 and Zechariah 9:9 simply to "the prophet" (21:4-5).³

In the sources checked here, the "more important prophet" rule has one possible pre-Christian instance and no explicit ancient statement.  It is a hypothesis, not a documented convention.

Mark 1:2: "in Isaiah the prophet"

The earliest manuscripts of all three early text types read "in Isaiah the prophet";  the majority of manuscripts, most of them later, read "in the prophets", as the King James Version does.  Metzger:  "It is easy to see, therefore, why copyists would have altered the words 'in Isaiah the prophet' ... to the more comprehensive introductory formula 'in the prophets.'"¹⁰  Jerome put the problem plainly in 395:  the first part is "not written in Isaiah at all, but in Malachi the last of the twelve prophets.  Let ignorant presumption solve this nice question if it can, and I will ask pardon for being in the wrong."¹

The explanations and what supports each:

1. Isaiah names the theme. The Exodus-Malachi line was already a fixed proof text about John (Matthew 11:10 and Luke 7:27 have it). Mark's real announcement is Isaiah's voice in the wilderness, which the next verses fulfil with John "in the wilderness" (1:4). On this reading Mark names the prophet whose text frames his opening and treats the attached line as part of the same testimony. Support: the shared Synoptic wording, which does not require assuming a lost document called Q, and 2 Chronicles 36:21 as one parallel for a headline name. Weakness: the line Mark quotes first is the one not from Isaiah.
2. A slip. Mark, or a source he copied, misremembered. Support: Matthew and Luke both separate the two quotations. Weakness: it supposes Mark did not know where a proof text in common use came from.
3. A testimony collection. Mark took both lines from a written collection of proof texts headed by Isaiah. Support: 4Q175 shows such collections existed. Weakness: no such collection of this text survives.

Matthew 27:9-10: "Jeremiah the prophet"

Matthew writes:  "Then was fulfilled what was spoken by Jeremiah the prophet, saying, 'And they took the thirty pieces of silver, the price of him who had been priced, whom some of the sons of Israel priced, and they gave them for the potter's field, as the Lord directed me.'"  The thirty pieces of silver, the pricing and the potter are Zechariah 11:12-13.  A field bought with silver is Jeremiah 32.  A potter's vessel, the Valley of Hinnom, innocent blood and a place of burial are Jeremiah 19.³  No verse of Jeremiah supplies Matthew's words.  Metzger calls the reading "Jeremiah" "firmly established";  a few witnesses substitute Zechariah, some omit the name, and two read Isaiah.¹⁰

1. A composite named for its hidden half. Gundry argues that Matthew "sees two separate prophecies, one typical and one explicit, fulfilled in one event, and makes the ascription to Jer because the manifestness of the quotation from Zech and the lack of verbal resemblance to Jer would cause the Jer-side of the prophecies to be lost."⁵ He limits the Jeremiah link to chapter 19 and argues against linking chapters 18 and 32: "In Jer 18 a potter is mentioned, but no field, in Jer 32 a field, but no potter."⁵ Augustine had already proposed a composite reading about AD 400, using the field bought in Jeremiah 32.¹¹ Support: the proposed links include the potter, innocent blood, priests and elders, burial and the renaming of a place. Matthew does not explicitly name the Valley of Hinnom, so that location should not be counted as a stated verbal agreement. Weakness: it requires readers to recognise an allusion with no shared wording.
2. A slip of memory. Augustine considered this first: "the word Jeremiah occurred to his mind, in accordance with a familiar experience, instead of Zechariah." He then argued that providence let it stand, because "all the holy prophets, speaking in one spirit, continued in perfect unison."¹¹ Krister Stendahl called it simply a mistake (as reported by Gundry).⁵
3. "Jeremiah" as the name of the prophetic scroll. A Talmudic list (Baba Bathra 14b) puts Jeremiah first among the latter prophets.⁵ The list is later than Matthew, and Matthew elsewhere names Isaiah and Jeremiah as individual authors.
4. Textual or lost sources. Augustine weighed and rejected the copies without the name, because "a majority of codices contain the name of Jeremiah" and the older Greek copies had it.¹¹ Jerome thought of a Hebrew apocryphon of Jeremiah (as reported by Gundry); none survives.⁵

Mark 2:26: "when Abiathar was high priest"

Jesus asks, "Have you never read what David did ... how he entered the house of God when Abiathar was high priest, and ate the bread of the presence?"  In 1 Samuel 21:1-6 the priest is Ahimelech;  his son Abiathar escapes Saul's massacre a chapter later (22:20).³  Matthew (12:4) and Luke (6:4) have no Abiathar clause.  Metzger:  "In order to avoid the historical difficulty, D W al omit" the phrase, and other copyists added "the" so that it could mean the time of Abiathar, "who was afterward the high priest."¹⁰  Jerome again stated the problem without solving it.¹  Daniel Wallace, an evangelical textual critic, surveyed the solutions in 2006.¹²

1. "In the passage about Abiathar." The same Greek construction, ἐπί with a genitive, means "in the passage about" at Mark 12:26, "have you not read in the book of Moses, at the bush."³ John Wenham (1950) and William Lane revived this reading. Lane himself listed the objections, which Wallace quotes: the phrase is far from "have you not read", Abiathar is not central to that section, he first appears in chapter 22, and "numerous instances in Tannaitic documents indicate that a section was usually designated by a term which occurs early, not late, in the section."¹²
2. "In the days of Abiathar the high priest." Abiathar was alive and later became David's high priest. Wallace found no ancient text in which this construction without "the" means the general time of someone who was not yet in office, while admitting "I have not searched very diligently."¹²
3. Similar names in the tradition. 2 Samuel 8:17 names Ahimelech son of Abiathar; its parallel in 1 Chronicles 18:16 reads Abimelech son of Abiathar. These need not reverse the same father and son: a later descendant could bear a repeated family name.³ But no surviving Greek manuscript of 1 Samuel 21 reads Abiathar.¹²
4. An error by Mark, his source, or Jesus. Wallace calls this "probably the most popular among critical scholars."¹² It is the comment Bart Ehrman's Princeton teacher wrote on his student paper defending reading 1: "Maybe Mark just made a mistake."¹³

Wallace's own conclusion:  the textual solution "is clearly out", and "views 2-5 all have a certain plausibility";  he prefers the reinterpretation.¹²

What the three cases show

Two conclusions follow.  First, the freedom shown in Part 2 covers wording and arrangement, and no ancient text examined here extends it to names.  An appeal to "convention" does not dissolve these three cases;  each needs its own argument.  Second, the skeptic's case is weaker than "the Bible misquotes."  These three are cases ancient readers themselves discussed, among the New Testament's many formal quotations, and each has an explanation with some real support.  Whether a particular attribution is mistaken remains a historical and textual question, with implications for a doctrine of scripture.  The comparisons here do not conclusively choose between the explanations.  It is not evidence that the New Testament writers did not know their Bibles, since in two of the same passages they follow the Hebrew against the Greek on points of detail.

Part 4: "Fulfilled": prediction, pattern, reapplication and allegory

New Testament writers apply scripture in several ways.  Fulfilment language can concern prediction or a recurring pattern;  pesher and declared allegory supply related comparisons for understanding the wider interpretive practice.  These categories are not all introduced by the word "fulfilled", and their historical comparanda are not equally well established.

The Qumran commentaries say what they are doing

The Habakkuk commentary from Qumran quotes the prophet a line at a time and follows each line with "its interpretation concerns" (pishro).  It states its premise:  "God told Habakkuk to write what was going to happen to the last generation, but he did not let him know the consummation of the era."  Habakkuk 2:2 ("so that the one who reads it may run") then "concerns the Teacher of Righteousness, to whom God has made known all the mysteries of the words of his servants, the prophets."⁷  The interpretations identify the prophet's terms with the community's world:  the Chaldeans of Habakkuk 1:6 are "the Kittim", and in Habakkuk 2:17 "Lebanon is the Council of the Community and the animals are the simple folk of Judah."⁷

This is reapplication with an explicit theory:  the prophet wrote more than he understood, and the inspired interpreter now reveals it.  1 Peter states a similar idea about the prophets, who searched "what person or time the Spirit of Christ in them was indicating", and to whom "it was revealed that they were serving not themselves but you" (1:10-12).³  The likeness is in the premise.  The Qumran commentaries go line by line and treat the whole book as a code for their sect;  the New Testament writers mostly apply single texts to single events.

Hosea 11:1 in Matthew 2:15

Hosea 11:1-2 reads:  "When Israel was a child, I loved him, and out of Egypt I called my son.  The more they were called, the more they went away;  they kept sacrificing to the Baals."  In its context the verse looks back to the exodus.  Matthew applies it to the child Jesus brought back from Egypt.  Two evangelical scholars give opposite accounts of what Matthew is doing.

1. Walter Kaiser (1985) argues that the use is contextual. Matthew's quotation "apparently is an independent and accurate translation of the Hebrew text, for the LXX reads on Hosea 11:1, 'Out of Egypt I summoned his children (plural!).'"⁴ The title "my son" was first used of Israel in Exodus 4:22-23, where "the whole community was pointedly designated by a singular noun, thereby allowing for the many to be represented in the One and the One to stand for the many."⁴ He concludes that "both Israel and Jesus are intimately related to God in the title 'My Son'" and are linked "in a corporate solidarity."⁴
2. Peter Enns (2005) argues that it is not. "Hosea 11 is not looking to the future at all but simply alluding to the past." Matthew's motive "is not a desire to do grammatical-historical exegesis": "in writing to a Jewish audience, such a handling of the Old Testament would not have seemed strange but very familiar."¹⁴ The New Testament shares with other Second Temple texts "the interpretive methods used and the interpretive traditions they adopt", and its writers "began with the event from which all else is now to be understood."¹⁴

Both agree on the facts that matter here.  Hosea's sentence describes the past, and Matthew is not claiming Hosea predicted Jesus' childhood in the way a forecaster predicts.  The disagreement is whether "my son" carries a pattern Hosea's own theology supports (Kaiser) or whether Matthew reads Hosea backward from the resurrection (Enns).  Two observations narrow it.  Matthew's argument works only with the singular "son", which the Hebrew has and the Greek lacks, so he chose his text deliberately.  And Enns's claim about how non-Christian Jewish readers would have reacted is a reasonable inference from Qumran practice, but no first-century Jewish response to Matthew 2:15 survives to test it.  The larger question of which Old Testament texts make predictions belongs to How to Test a Fulfilled Prophecy and Messianic Prophecy:  An Overview.

Galatians 4 and Philo

Paul writes that Abraham had two sons, one by the slave woman and one by the free woman, and then:  "these things are allegorized (ἅτινά ἐστιν ἀλληγορούμενα);  for these women are two covenants, one from Mount Sinai, bearing children for slavery;  she is Hagar ... but the Jerusalem above is free, and she is our mother."³  Paul names his method and keeps the history:  "it is written that Abraham had two sons."

Philo of Alexandria, a Jewish contemporary of Paul, read the same two women allegorically to a different end.  "We are not as yet capable of becoming the fathers of offspring of virtue, unless we first of all have a connection with her handmaiden;  and the handmaiden of wisdom is the encyclical knowledge", that is, grammar, geometry, astronomy, rhetoric and music, "of which Hagar, the handmaid of Sarah, is an emblem."¹⁵  An announced allegory is not being presented simply as the literal sense of the story.  Whether its proposed correspondence is justified still needs an argument.  A reader may reject allegory as a method, but cannot fault Paul for passing it off as plain sense.

What this settles

Question 3 has an answer.  New Testament uses include prophecy, pattern, reapplication and declared allegory.  The parallels checked here establish Jewish practices of reapplication and allegory;  this article has not independently checked a non-Christian typological example corresponding to Matthew 2:15.  Galatians 4 explicitly names allegory rather than using a fulfilment formula.  A skeptic who reads every "fulfilled" as a claim that the Old Testament author consciously forecast the event will find many "misuses";  most disappear once the kind of claim is identified.  That does not make every application valid.  A pattern can be imagined, and a reapplication proves nothing to someone who does not share the interpreter's premise.  The evidential force of a prophecy is a separate question from whether its New Testament use followed the conventions of its day.

Part 5: Names on books: anonymous, attributed and pseudonymous

Three different things are often all called "pseudepigraphy":

1. An anonymous book, whose author never names himself. The four Gospels, Acts, Hebrews and 1 John are anonymous in this sense.
2. A later attribution, when readers or copyists attach a name to an anonymous book. If the name is wrong, the author did not lie; later readers erred.
3. A pseudonymous book, whose author claims to be someone he is not, usually in the opening of a letter or in the first person.

Bart Ehrman, whose Forged (2011) argues that several New Testament letters are pseudonymous, draws the same line.  "None of the Gospels tells us the name of its author.  Only later did Christians call them Matthew, Mark, Luke, and John."¹⁶  In writings wrongly attributed later, "the author is not trying to deceive anyone.  He or she remained anonymous.  It is only later readers who claimed that the author was someone else."  His definition of forgery is "a writing that claims to be written by someone (a known figure) who did not in fact write it."¹⁶  Of the anonymous New Testament books he writes:  "since the real authors made no claims for themselves, the books are not forgeries.  They are simply false attributions", and a false attribution may be "just a well-intentioned mistake."¹⁶  Whether the traditional names of the Gospels are right is argued in Who Wrote the Gospels?;  whether classicists treat anonymity the same way is in Is New Testament Scholarship Unusually Skeptical?.

Hebrews: how attributions grow, and how ancient readers judged them

Hebrews never names its author, and its writer says the message of salvation "was confirmed to us by those who heard" the Lord (2:3), where Paul insists he received his gospel from no human being (Galatians 1:12).³  Ancient readers proposed Paul, Luke, Barnabas and Clement, and some rejected Paul:

1. Paul, translated by Luke. Clement of Alexandria, about AD 200, said Hebrews "is the work of Paul, and that it was written to the Hebrews in the Hebrew language; but that Luke translated it carefully and published it for the Greeks, and hence the same style of expression is found in this epistle and in the Acts."²
2. Barnabas, according to Tertullian as Jerome reports him, while Jerome says the letter "is not believed to be his", Paul's, "on account of the difference of style and language" (propter styli sermonisque dissonantiam).¹⁷
3. A disciple writing down Paul's thought. Origen, in the third century, judged that "the verbal style of the epistle entitled 'To the Hebrews,' is not rude like the language of the apostle", and that "the thoughts are those of the apostle, but the diction and phraseology are those of some one who remembered the apostolic teachings, and wrote down at his leisure what had been said by his teacher." He reported that some named Clement of Rome and others Luke, and concluded: "But who wrote the epistle, in truth, God knows."²
4. Not Paul. Eusebius records that "some have rejected the Epistle to the Hebrews, saying that it is disputed by the church of Rome, on the ground that it was not written by Paul."¹⁸

No one in this list calls Hebrews a forgery, because it claims no author.  Each judges authorship by evidence, and three of them use style.

Style as a test of authorship is ancient

Modern critics did not invent the argument from style.  Ancient readers used it, sometimes to reject a name and sometimes to explain a difference while keeping the name.

1. Galen, the second-century physician, tells how in the booksellers' street at Rome a buyer was about to purchase a book titled "Galen the physician." A man of letters read two lines and threw it down, saying (in Galen's Greek) that "this diction is not Galen's, and this book is falsely titled."¹⁹²⁰ Galen approves: the man had been trained, as Greek boys once were, by grammarians and rhetoricians.
2. Dionysius of Alexandria, a third-century bishop, compared Revelation with John's Gospel and 1 John: "I judge from the character of both, and the forms of expression, and the entire execution of the book, that it is not his." He describes Revelation as sharing "without even a syllable in common with them", and its author uses "barbarous idioms, and, in some places, solecisms." Yet he did not reject the name on the book: "that he was called John, and that this book is the work of one John, I do not deny."²¹ He concluded it was a different John.
3. Jerome reports that the second letter of Peter "is considered by many not to be by him", "on account of its difference from the first in style."²²
4. Eusebius rejected the gospels and acts that heretics put forward under apostles' names partly because "the character of the style is at variance with apostolic usage", adding doctrine and lack of use by church writers as further tests.¹⁸

So the argument from style is not a modern prejudice against the New Testament.  It is also not decisive by itself.  Dionysius kept "John", Origen kept the Pauline thought, and Jerome, as Part 6 shows, explained the difference between Peter's letters by a change of secretary.

Part 6: Secretaries, agents and carriers

What the New Testament itself says

1. Romans 16:22: "I Tertius, who wrote the letter, greet you in the Lord." A scribe wrote Romans. The verse says nothing about how much he shaped it.
2. Galatians 6:11: "See with what large letters I have written to you with my own hand." The change of hand implies that someone else wrote the rest, or most of it.
3. 1 Corinthians 16:21 and Colossians 4:18: "The greeting is by my own hand, Paul's." Philemon 19: "I, Paul, write this with my own hand: I will repay."
4. 2 Thessalonians 3:17: "The greeting is by my own hand, Paul's, which is a sign in every letter; this is how I write."³

These verses establish that Paul dictated, that he signed in his own hand, and that a signature could authenticate a letter.  They do not establish how much freedom his scribes had.

What ancient letter-writers show

The fullest evidence comes from the letters of Cicero, the Roman statesman, in the first century BC.

1. Dictation at two speeds. Cicero tells his friend Atticus that he did not dictate one letter "even to Tiro, who usually takes down whole periods at a breath, but syllable by syllable to Spintharus."²³
2. Letters composed by a friend in Cicero's name. From exile in 58 BC: "If there are any people to whom you think that letters ought to be delivered in my name, pray compose them and see them delivered." In 48 BC: "if there are any persons whom you think ought to have a letter from me, compose one yourself", and, when ill, "Pray write to Basilus and to others to whom you think it proper ... in my name, and say whatever you think right."²³
3. And the recipients were not told. "Pray send letters in my name to any to whom you think it right to do so. You know my intimates. If they remark on the absence of my signet or handwriting, pray tell them that I have avoided using either owing to the military pickets."²³

Cicero's letters prove that a letter could go out under a man's name, with his authority, composed by someone else, and that the sender might not disclose it.  They also show the limits.  The composer was a close friend, not a hired secretary.  The requests are occasional correspondence;  the resulting letters do not survive in an identified form, and their lengths are not stated.  And the seal and handwriting were the usual signs of authenticity, which is exactly what 2 Thessalonians 3:17 appeals to.

Two ancient Christian secretary theories

1. Clement of Alexandria explained the style of Hebrews by translation: Paul wrote in Hebrew and "Luke translated it carefully", "hence the same style of expression is found in this epistle and in the Acts."²
2. Jerome explained the style of 2 Peter by a change of helper. Paul "had Titus as interpreter, as blessed Peter had Mark, whose Gospel was composed with Peter narrating and Mark writing. And the two letters that circulate as Peter's differ from each other in style and character and in the arrangement of words; from which we understand that he used different interpreters as circumstances required."²⁴

Ancient readers, then, faced the same style evidence modern readers do and sometimes answered it with a secretary.  That shows the explanation was available and plausible to them.  It does not show it was true in any case.

"I wrote through Silvanus"

1 Peter ends:  "Through Silvanus, the faithful brother as I regard him, I have written to you briefly" (5:12).  Defenders of Peter's authorship often read this as naming Silvanus as the secretary who polished the Greek.  E. Randolph Richards, an evangelical specialist in ancient letter-writing, argues against that reading:  "It will be argued here that this formula identified only the letter-carrier and was not used to identify the secretary."²⁵  Ignatius of Antioch, writing early in the second century, uses the same formula for Burrhus, an escort sent with him by the churches:  "I write to you through Burrhus, who was sent along with me by the Ephesians and Smyrnaeans."²⁶  Richards still holds that "Silvanus certainly could have been the secretary" and that secretarial help "should be assumed for all NT letters.  However, 1 Pet 5:12 does not address the role of the secretary."²⁵

Ehrman's objection, and what it leaves

Ehrman writes:  "I don't know of a single piece of evidence or a single analogy to suggest that Peter or Paul used a secretary who significantly—or insignificantly, for that matter—added to the contents of the letter."  His reason is that "secretaries did not produce the contents of letters (at least letter-essays of this sort)."¹⁶  He grants that "even the upper classes would sometimes instruct a secretary to spin off a quick stereotyped letter for them", "as is evidenced on several occasions by Cicero", but holds that this "is completely different from composing a long, detailed, finely argued, carefully reasoned, and nuanced letter like 1 Peter or Ephesians."¹⁶

The evidence supports half of this.  No such extended secretary-composed letter is demonstrated by the examples verified here, and Cicero's agent was a friend.  But Ehrman's word "stereotyped" is his own description;  "say whatever you think right" is a grant of content, and the letters' length is unknown.  And Clement and Jerome show that ancient Christian readers themselves proposed translators and secretaries to account for style.  Those later explanations are evidence of what ancient readers found plausible, rather than independent documentation of the original composition.  The fair conclusion is that a secretary or co-writer can explain differences of style and wording in an authentic letter, as Jerome thought, but it must be supported separately for the letter at issue, and the further a letter's content departs from the named author's, the less a secretary can explain.  Kenneth Neumann, whose computer study is discussed in Part 9, reached the same place from the other side:  letters that do not match Paul's style might come from "Paul's amanuenses or co-senders writing predominantly under or with the apostle's guidance, or they may be later Pauline students."²⁷

Part 7: What ancient readers called forgery

When a false author was detected

1. Galen, second century: a book sold under his name was rejected as "falsely titled" on style, and Galen wrote a whole treatise listing his genuine works so that readers could tell them from others. He complains that his lecture notes, given "to friends or students without a title", were read by others as their own, and that people altered his books by "removing some things, adding others, changing others."²⁰
2. Serapion, bishop of Antioch about AD 200, had allowed a congregation to read a "Gospel of Peter" before reading it himself. "As I had not read the Gospel which they put forward under the name of Peter, I said, If this is the only thing which occasions dispute among you, let it be read." After reading it he found "many things in accordance with the true doctrine of the Saviour, but some things added to that doctrine", and wrote: "we reject intelligently the writings falsely ascribed to them, knowing that such were not handed down to us."² His general rule names false ascription; his decision on this book followed from its content.
3. Tertullian, about AD 200, against those who cited the Acts of Paul to let women baptize: "let men know that in Asia the presbyter who compiled that document, thinking to add of his own to Paul's reputation, was found out, and though he professed he had done it for love of Paul, was deposed from his position." The Latin names the work as "Acta Pauli" in the oldest manuscript, though the printed edition behind the older English translation lacks the word "Acta."²⁸ Ehrman rightly notes that this is not a case of a false name: "the Acts of Paul is not a book that claims to be written by Paul; it claims to be a true account about Paul."¹⁶ It shows that a pious fiction about an apostle, written out of devotion, cost its author his office.
4. The Muratorian fragment, a Latin list of New Testament books, rejects letters "to the Laodiceans" and "to the Alexandrians" that were "forged in Paul's name to [further] the heresy of Marcion ... for it is not fitting that gall be mixed with honey."²⁹ The Latin is "Pauli nomine fincte."³⁰ Forgery and heresy are named together.
5. Earlier critics of Revelation, reported by Dionysius, rejected it by "maintaining that the title is fraudulent" and said that Cerinthus, "desiring reputable authority for his fiction, prefixed the name."²¹
6. Eusebius, about AD 325, calls the apocryphal gospels and acts "the fictions of heretics" and rules that they "are not to be placed even among the rejected writings."¹⁸

Even 2 Thessalonians, whose own authorship is disputed, warns readers not to be shaken "by a letter as if from us" (2:2) and gives Paul's handwriting as the sign of a genuine letter (3:17).³  Whoever wrote it, the warning assumes that a letter falsely claiming Paul was possible and was wrong.  Ehrman notes the irony, since "Second Thessalonians ... is itself widely thought among scholars not to be by Paul, even though it claims to be written by him."¹⁶

Where tolerance appears

1. Philosophical schools. Iamblichus, about AD 300, praises the Pythagoreans: "This also is a beautiful circumstance, that they referred every thing to Pythagoras, and called it by his name, and that they did not ascribe to themselves the glory of their own inventions, except very rarely." The same book complains that Pythagorean teaching had been "obscured by false and spurious writings."³¹ Iamblichus wrote more than six centuries after the Pythagoreans he describes, and two centuries after Paul.
2. A fifth-century Christian pseudonym. About AD 440, books "to the church" appeared under the name of Timothy.³² Bishop Salonius asked why, warning that unless the name were explained the books might be counted among the apocrypha. Salvian of Marseilles, probably their author, defended the name in a letter that never names the author.³³ His defence is revealing. The books are obviously recent, and "those are free of the suspicion of the apocryphal which are recognised not to have been the apostle Timothy's." The author hid his own name because readers "consider not so much what they read as whose it is." And he chose "Timothy", which means "honour of God", as Luke chose "Theophilus", "love of God", because, in the printed text, "he feared a lie in this exchange of names, thinking that not even in the service of a holy work should the stain of falsehood be admitted." The words "he feared" rest on an editor's correction of the manuscript.³³ The one ancient Christian defence of an apostolic pseudonym examined here argued that the name was not really a claim of authorship, and treated a straightforward false name as a lie.

The modern positions, stated by their authors

1. Convention, no deception. Richard Bauckham, a leading conservative New Testament scholar, judges 2 Peter a "testament" written after Peter's death and argues that its Petrine authorship "must be a deliberately transparent fiction, a literary convention which the author expected his readers to recognize as such, just as modern readers are not likely to mistake a historical novel for factual history."³⁴ David Meade argued, as Donald Guthrie summarises him, "that literary attribution is primarily an assertion of authoritative tradition, and not of literary origins." Guthrie replied: "The real Achilles heel of Meade's case is when he is obliged to admit that there was an element of deception in the process."³⁵
2. Deception. Ehrman: "in virtually every instance in which an ancient author mentions forgery, he condemns it", and "forgery was not a transparent fiction."¹⁶ Terry Wilder, a conservative scholar, agrees on the ancient evidence and draws the opposite conclusion about the canon: the second-century church rejected pseudo-apostolic works, and "the first-century church probably also would have rejected pseudonymous letters written in their names."³²

The ancient evidence assembled here supports Ehrman and Wilder on the historical question.  The examples condemn deceptive attribution or fabricated apostolic material, while their circumstances differ.  Tertullian's narrative is not a letter claiming Paul as author;  Serapion and the Muratorian list also judge doctrine.  Iamblichus praises attribution within a philosophical school, and Salvian denies that his pseudonym claimed the apostle's authorship.  Together these texts weigh against a blanket claim that ancient readers accepted false apostolic letters.  Bauckham's case for 2 Peter rests on the letter's internal features, which are real (Part 8), but it needs first readers who recognised the fiction.  The earliest readers who discuss 2 Peter treat its authorship as a factual question:  Origen calls a second letter of Peter "doubtful", Eusebius lists it as disputed, and Jerome reports doubts on grounds of style.²  None says it was understood as a fiction.

The consequence is sharp.  A letter written to deceive readers about an apostolic author would conflict with the standards expressed in these sources.  Authorised composition on another's behalf is a different case.  A claim that a pseudonym was transparent needs evidence about the genre and its intended audience;  absence of an early explanation is a difficulty, not proof that every first reader was deceived.  That does not decide whether any letter is pseudonymous.  It means the question must be answered, letter by letter, from the evidence.

Part 8: The disputed letters, one at a time

General ancient practice cannot tell us who wrote a particular letter.  What can is the evidence about that letter:  who used it early, what it says about itself, and how its language compares with the named author's.  This part gives the main evidence each way for four representative cases.  It does not reargue every disputed letter;  Ephesians and Colossians, for example, raise similar questions and are left aside.

The Pastoral Epistles (1 and 2 Timothy, Titus)

1. Early use. Polycarp of Smyrna, in the letter with which he forwarded Ignatius's letters (13.2), writes, in a literal translation of his Greek: "the love of money is the beginning of all troubles. Knowing therefore that we brought nothing into the world, but neither have we anything to carry out."²⁶ 1 Timothy 6:10 and 6:7 have "the love of money is a root of all evils" and "we brought nothing into the world, and we cannot carry anything out", with eight Greek words shared in the same order. Both sayings were proverbial, so a skeptic can say Polycarp drew on common stock; the two together are evidence that he knew 1 Timothy. Polycarp gives the two sayings in the reverse order from 1 Timothy, not the same order.³
2. Marcion and the Muratorian list. Tertullian says that Marcion, the second-century teacher who made his own short canon of Paul, kept Philemon but "rejected the two epistles to Timothy and the one to Titus."³⁶ Whether Marcion rejected letters he knew or did not know them, Tertullian treats them as Paul's. The Muratorian list includes "one to Titus, and two to Timothy", "held sacred in the esteem of the Church catholic."²⁹
3. Vocabulary. P. N. Harrison's 1921 study found that "of the remaining 848" words of the Pastorals, "306 or over 36 per cent. are not to be found in any one of the ten Paulines", and that 175 words appear nowhere else in the New Testament.³⁷ A check for this article finds five words and phrases that occur 32 times in the Pastorals' 3,485 words and never in Paul's seven undisputed letters (23,999 words): εὐσέβεια, "godliness" (10 times); ἐπιφάνεια, "appearing" (5); σώφρων, "sensible" (4); ὑγιαίνω, "be sound", of teaching (8); and the formula "the saying is trustworthy" (5).³⁸ These illustrate a concentration of particular vocabulary, not a prespecified statistical test: the items were selected because they are characteristic of the Pastorals. Each rests on a handful of occurrences, and some reflect the letters' subjects: church order and false teaching.
4. Vocabulary controlled for length. Jermo van Nes (2018) tested whether the disputed letters use more once-only words than their length predicts, using Paul's seven undisputed letters as the baseline. He found that "Colossians, Ephesians, and 2 Thessalonians neatly fit the undisputed Pauline pattern", that "Titus among the Pastorals also fits this pattern, be it close to the maximal prediction value", and that "only the numbers of hapaxes in the Timothy correspondence are significantly higher."³⁹
5. Personal details. 2 Timothy 4:13 asks Timothy to bring "the cloak that I left at Troas with Carpus, and the books, above all the parchments."³ Defenders read such details as the marks of a real letter; critics read them as the touches a careful imitator adds. Neither reading can be tested from the verse alone.

Finding:  the Pastorals present a substantial internal challenge to Paul's authorship, and the vocabulary evidence is not an illusion of method.  It is also not decisive:  the letters are short, their subjects differ from Paul's other letters, a different secretary or co-writer could account for some of the language, and the length-controlled measure puts Titus inside Paul's range.

2 Thessalonians

1. The letter's own claims. It warns against "a letter as if from us" claiming "that the day of the Lord has come" (2:2), and closes with the handwritten greeting, "a sign in every letter" (3:17).³
2. Overlap with 1 Thessalonians. Leaving aside the formulaic openings and closings, the two letters share nine distinct five-word sequences, 12.4 per thousand words of 2 Thessalonians, and a ten-word run: "working night and day so as not to be a burden to any of you" (1 Thessalonians 2:9 = 2 Thessalonians 3:8). Among the eleven pairs selected for this comparison, only Colossians and Ephesians share more by this measure (14.8); 1 and 2 Corinthians share 2.1, Romans and Galatians 8.0.⁴⁰ Imitation would produce this; so would one author writing twice to one church within a short time, a possibility not unique to this pair: Paul also wrote repeatedly to Corinth. As a control, Jude and 2 Peter, whose parallel order and content make a literary relation very likely, share only two five-word sequences: a writer who borrows can rephrase almost completely. The count cannot choose between the explanations.
3. Teaching on the end. 1 Thessalonians 5:2 says "the day of the Lord comes like a thief in the night"; 2 Thessalonians 2:3 says it will not come "unless the rebellion comes first and the man of lawlessness is revealed."³ Read as two general descriptions, these pull against each other. Read as the letter presents itself, the second corrects a claim "that the day of the Lord has come" (2:2), which a reminder of preceding signs would answer.
4. Language. Van Nes finds that its rate of rare words fits Paul's.³⁹

Finding:  undecided on the evidence reviewed here.  The letter's warning against forged letters is evidence of what its author thought wrong, not of who he was.

2 Peter

1. Early attestation. Sparse and explicitly disputed. Origen: Peter "has left one acknowledged epistle; perhaps also a second, but this is doubtful." Eusebius: "we have learned that his extant second Epistle does not belong to the canon; yet, as it has appeared profitable to many, it has been used with the other Scriptures."²¹⁸ Jerome reports that "many" doubted it because of its style.²²
2. Strong claims to be Peter. The writer says his death is near, "as our Lord Jesus Christ made clear to me" (1:14), and that "we were eyewitnesses of his majesty ... we ourselves heard this voice borne from heaven, for we were with him on the holy mountain" (1:16-18).³ If Peter did not write it, the letter is pseudonymous in the strongest sense, not an anonymous work later attributed.
3. Signs of a later date, as critics read them. Scoffers ask where Christ's coming is, "for ever since the fathers fell asleep, all things continue as they were" (3:4). Paul's letters are known as a group, "all his letters", which the ignorant twist "as they do the other scriptures" (3:15-16).³ Bauckham reads the scoffers' "fathers" as "that generation" of first Christians, "now passed away."³⁴ Paul meant his letters to circulate: the undisputed 1 Thessalonians orders that it "be read to all the brothers" (5:27).³ A known group of letters placed beside "the other scriptures" is important evidence for reception. Its dating force depends on how quickly such a collection and authority could arise; the phrase supplies no fixed minimum interval. The argument from "the fathers" likewise depends on whom that term denotes.
4. Use of Jude. 2 Peter 2 runs parallel to Jude 4-18 in order and content. The exact wording shared is small (Part 2's yardstick finds a longest run of five words), so the case for literary dependence rests on the order and content as well as wording. Extensive paraphrase would still be use of a written source.⁴⁰
5. Two explanations. Jerome's change of secretary (Part 6) and Bauckham's testament genre (Part 7) answer different problems. Jerome keeps Peter as author and explains the style. Bauckham gives up Peter as author and argues there was no deception. The ancient evidence on reception tells against Bauckham's second step; nothing ancient decides between Jerome's view and pseudonymity.

Finding:  2 Peter presents serious difficulties for traditional authorship, and ancient Christians explicitly disputed it.  It is decided, if at all, by the specific evidence above, not by any convention.

Hebrews, the control

Hebrews claims no author, so the question of forgery does not arise.  Its style differs from Paul's, its writer received the gospel from those who heard the Lord (2:3), and ancient readers proposed four names and settled on none.  Origen's verdict, "who wrote the epistle, in truth, God knows", is still the honest one.²  Hebrews shows the difference between not knowing who wrote a book and showing that a book lies about its author.

Part 9: What stylometry can and cannot show

Stylometry measures habits of language that writers do not consciously control, usually the frequencies of common words such as "and", "but", "the" and prepositions, and asks whose known writings a disputed text most resembles.  It works well on long texts with a closed list of candidate authors.  Its limits are just as well documented.

Length

Maciej Eder, a specialist in computational stylistics, tested how much text the standard methods need by cutting known-author texts into samples of different sizes.  For English novels "samples shorter than 5000 words provide a poor 'guessing', because they can be immensely affected by random noise.  Below the size of 3000 words, the obtained results are simply disastrous."  For other corpora the point of reliable success lay "between 5000 and 10000 words per sample (and there was no significant difference between inflected and non-inflected languages)."  Latin prose did best, at "some 2500 words", yet his conclusion was that "using 2500-word samples will hardly provide a reliable result, to say nothing of shorter texts."⁴¹

Those corpus-specific figures are not universal minimum lengths.  Eder's fuller paper discusses a 209-word poem correctly attributed in an earlier study and explains that some writers are easier to identify than others;  stable performance across difficult cases requires more text.⁴²  The disputed letters are short:  Titus 659 words, 2 Thessalonians 820, 2 Timothy 1,235, Colossians 1,580, 1 Timothy 1,591, Ephesians 2,416.⁴³  A. Q. Morton, the first to put Paul's letters through a computer, used sentence length and excluded three letters for exactly this reason:  "II Thessalonians, Titus and Philemon are excluded from the analysis as they are not large enough to provide reliable samples in tests of this type."⁴⁴

Punctuation

Ancient Greek manuscripts have little or no punctuation;  the sentence divisions in a printed Greek New Testament are the modern editors'.  Justin Stover and Mike Kestemont, reviewing earlier work on a Latin text, fault one study for "the use of sentence length as a criterion of authorship, which is no longer considered an effective stylometric feature even for modern texts, and should definitely not be used for ancient texts, where the punctuation is due to the modern editor."⁴⁵  That undermines treating Morton's punctuation-based sentence counts as a direct measure of the original author.  A separate analysis would need boundaries and features that do not simply reproduce a modern editor's choices.

Successes and disagreements outside the Bible

1. Caesar's war commentaries. Suetonius says "the author of the Alexandrian, African, and Spanish Wars is unknown; some think it was Oppius, others Hirtius, who also supplied the final book of the Gallic War."⁴⁶ Kestemont, Stover and colleagues (2016) tested 1,000-word slices of Latin, which they call "a challengingly limited document size", against known authors, and concluded that "Hirtius wrote Book 8 of the Gallic Wars and the remainder of the Alexandrian War."⁴⁷ The study benchmarks its verification methods and argues for those attributions. It does not require assuming that every possible author is known. Its 1,000-word slices also show why Eder's thresholds should not be turned into a universal ban on shorter samples.
2. The Historia Augusta. This collection of Roman imperial biographies claims six authors. A 1979 study using sentence length found one author; a 1998 study using function words found several; Stover and Kestemont's 2016 study concluded that "the authorial structure presented by the manuscripts - six authors ... is not compatible with the features of the text."⁴⁵ Three studies, three methods, three answers. The lesson is not that stylometry fails, since the 2016 method was validated on known authors, but that a result depends on the method and its validation.

Studies of Paul

1. Kenneth Neumann (1990) used discriminant analysis on several measures with samples from known authors. The disputed letters he tested (Ephesians, Colossians, 2 Thessalonians) were "assigned to the Pauline group with a rounded-off probability of 1.0", yet lay far from Paul's centre, and he warned that "these results, despite the Pauline classification, may still indicate a non-Pauline authorship." He also reported "a disturbing result": a fresh control sample from Galatians, which nobody doubts, "is classified with Ignatius with a probability of 0.9910" by his best set of measures, and with Hebrews by the second best.²⁷ The Pastorals and 1 Peter "are classed with non-Pauline groups with a high degree of probability." His conclusion on method: "Only in cases where all the potential authors for disputed documents are known may discriminant analysis or other stylistic analytical methods decide authorship."²⁷ That is a limitation of forcing a choice among listed candidates. Modern authorship verification instead tests a particular candidate without assuming the true author must belong to a complete list, as the Caesar study explains.⁴⁷
2. Jermo van Nes (2018), controlling for length, found that only 1 and 2 Timothy have significantly more once-only words than Paul's undisputed letters predict (Part 8).³⁹

A test for this article

To test one implementation on short texts, this article ran an exploratory common-word comparison on the Greek New Testament.⁴³  The method is a Delta-style standardised distance based on the 50 to 150 commonest dictionary forms.  These include content words as well as function words, so subject matter can contribute.  Each of Paul's seven undisputed letters was held out in turn, and blocks of it were scored against a profile built from the other six.  Blocks from six letters nobody assigns to Paul, Hebrews, James, 1 and 2 Peter, 1 John and Jude, were scored against the same profiles.  Blocks from Acts and Revelation provide a contrasting task involving different authors and genres;  success on that task cannot by itself validate the implementation or its authorship inference.  The disputed letters were never used to fit anything.

AUC is the chance that a non-Pauline block scores as less Pauline than a genuine Paul block:  0.5 means no separation and 1.0 perfect separation.  The figures shown use the 50 commonest words;  100 and 150 give the same picture (AUC 0.35 to 0.57 against other letter writers, best two-group accuracy 73%).  The one-profile distance separates the selected epistles weakly, while the two-group classifier performs better in some settings.  These are descriptive results, not a statistical demonstration of chance-level performance or a limit on all stylometry.  The easier contrast with narrative and apocalyptic mixes author and genre.  One example shows the instability:  the single 1,200-word block of 2 Timothy falls at the 6th percentile of genuine-Paul distances with 50 words and at the 100th with 100 words.  Same text, same method, opposite impression.

This test has limits.  It uses one classic method, not the more powerful methods of the Caesar study, and the non-Pauline set is small.  It does not show that all stylometry fails on Paul.  It shows why a proposed attribution needs validation appropriate to its text length, genre, reference material and candidate authors.  Several blocks come from the same letter, the reference profile changes between folds, and no uncertainty interval or independent threshold-validation set is supplied.  Blocks are not independent author samples.  The classifier's balanced accuracy is the mean of its correct-Paul and correct-non-Paul rates, not the overall percentage of blocks correct.

Objections and responses

1. "Calling loose quotation an 'ancient convention' is special pleading invented to protect the Bible." The convention is documented outside the New Testament, in Jewish texts copied before Jesus (the Qumran scrolls) and stated in so many words by Jerome, who used it to defend his own translations. The same standard is not extended here to source names, where no such documentation was found.
2. "Then the three misnamed quotations are simply errors." That is one of the explanations, and Augustine and Wallace both take it seriously. It is not the only one with evidence. Matthew's Jeremiah 19 parallels are specific, Mark 12:26 uses the same Greek construction to mean "in the passage about", and 2 Chronicles 36:21 shows a prophet's name over a partly borrowed sentence. The honest position is that each case is open, not that each is solved.
3. "Pesher and allegory are just ways of making a text say what you want." They can be. But the question here is whether the New Testament writers misunderstood or misrepresented their texts. A writer who says "this is an allegory" (Galatians 4:24), or who reads a past event as a pattern, is not claiming the source's author predicted his conclusion. Whether the method is persuasive is a separate question, discussed in the prophecy articles.
4. "Ancient readers accepted pseudonymity. The Pythagoreans wrote in their master's name." Iamblichus praises Pythagoreans for crediting their teaching to Pythagoras, about AD 300, and in the same book condemns "false and spurious writings." Christians who discussed the question, from Serapion to the Muratorian list, condemned false apostolic names. The one Christian defence of a pseudonym found, Salvian's, comes from the fifth century and argues that the name was not a claim of authorship.
5. "Serapion rejected the Gospel of Peter for its doctrine, not its name." True for the particular book: he let it be read until he found additions to true doctrine. But his rule, "we reject intelligently the writings falsely ascribed to them", names false ascription as a ground in itself, and Eusebius likewise joins style, doctrine and lack of use. Neither shows a name alone sufficed. Neither shows a false apostolic name was acceptable.
6. "The Acts of Paul case is not about a false name at all." Agreed; Ehrman makes the same point. The case shows how a devout fiction about an apostle was treated once found out, which bears on the "harmless convention" defence, but it is not an example of a forged letter.
7. "A secretary explains every difference in style, so style can never count against authorship." Ancient readers did use secretaries to explain style (Clement on Hebrews, Jerome on 2 Peter). But extensive freedom over vocabulary and arguments raises the question of co-authorship. The examples checked here do not document an extended argumentative letter composed by a secretary independently. A secretary can explain differences of wording more easily than differences of content.
8. "Ehrman says there is no evidence for secretary-composed letters, so the secretary theory is dead." Cicero repeatedly had a friend compose letters in his name, told him to "say whatever you think right", and prepared an excuse for recipients who noticed the missing seal or hand. That is evidence of authorised composition. Ehrman's narrower claim, that no long argued letter was composed this way, stands as far as the evidence found here goes.
9. "Stylometry has proven the Pastorals are not Paul's." The studies disagree, their lengths make validation especially important, Morton excluded the shortest as unreliable, Neumann's method misclassified a Galatians control, and this article's one-profile comparison separates the selected epistles weakly. The vocabulary of the Pastorals is unusual, and that evidence is real, but it is not a stylometric proof.
10. "Stylometry has vindicated Paul, since Neumann classed the disputed letters as Pauline." Neumann himself said the distances "may still indicate a non-Pauline authorship", and classed the Pastorals with non-Pauline groups. Neither side can claim the computers.
11. "If 2 Peter is pseudonymous, it is a transparent testament like many Jewish testaments, and no one was fooled." That may be true of the author's intention, and Bauckham argues it from the letter's genre and tense changes. But the readers who discuss 2 Peter in antiquity treat its authorship as a matter of fact, and none reports a shared understanding that it was a fiction. The convention has to be shown in its readers, not only inferred from its form.
12. "Hebrews was accepted as Paul's, so the church was fooled by false attributions." Some churches credited Hebrews to Paul and others denied it, openly and on stated grounds. The dispute shows uncertainty about an anonymous book's attribution; it does not by itself prove deception or prove that every attribution was correct.

Issues and unanswered questions

1. Independent comparanda for single names over composite quotations. Only 2 Chronicles 36:21 was found and checked. Adams and Ehorn's introduction and Adams's Homer chapter are now available, but the remaining chapters of Composite Citations in Antiquity were not read. A broader search could establish how often a single source name covered a combination, and whether the proposed rule about the more important prophet is supported.
2. The Hebrew and English translation of the Habakkuk commentary and 4QFlorilegium were read in the García Martínez and Tigchelaar edition only; the Ezekiel blend in 4Q174 partly rests on restored letters.
3. Galen's Greek was read from a damaged scan of Müller's 1891 Teubner text and checked against Kühn's facing Latin; a clean modern edition (Boudon-Millot, 2007) was not obtained.
4. Tertullian's text. The word "Acta" in On Baptism 17.5 stands in the twelfth-century Troyes manuscript and not in the 1545 printed edition; the article follows Ernest Evans's edition.
5. Salvian's "he feared" is an editor's emendation of the single manuscript's "Timothy"; the sense of the passage does not depend on it, but the quotation does.
6. The secretary question would be advanced most by new evidence from Greek papyrus letters showing how much freedom scribes had in long letters. E. Randolph Richards's 1991 monograph, The Secretary in the Letters of Paul, was not read for this article.
7. A stronger stylometric test. The test here used one classic method. Character n-grams, cosine Delta or a verification method like the Caesar study's, with known-author controls of letter length, would be a fair next step. Until such a test succeeds on known authors, no stylometric verdict on the disputed letters should be trusted.
8. The Pastorals' external evidence includes questions not examined here: whether the oldest Pauline manuscript (P46) had room for them, and whether Marcion knew them.
9. David Meade's book could not be read (the only archive copy is mis-scanned); his view is reported through Guthrie's review.

Verdict

For the skeptic.  "Ancient convention" is not a universal solvent.  No ancient evidence found here makes a wrong source name right, and three named quotations (Mark 1:2, Matthew 27:9, Mark 2:26) keep real residues that ancient readers already saw.  The sources checked condemn deceptive authorship and fabricated apostolic material;  they do not establish the response of every first audience to every pseudonymous genre, and the "transparent fiction" defence of 2 Peter lacks ancient readers who saw through the fiction.  The Pastorals' vocabulary is genuinely unlike Paul's undisputed letters, and 2 Peter's early authorship was explicitly disputed.

For the defender.  A charge based solely on differing words imposes an exact-wording requirement the ancient comparisons do not support.  Their freedom with wording, blending and text forms is documented in Jewish texts before Jesus and stated by Jerome.  "Fulfilled" covers pattern, reapplication and declared allegory as well as prediction.  Anonymous books are not forgeries.  Secretaries and authorised agents were real, and ancient Christians themselves used them to explain differences of style.  The stylistic tests reviewed here do not settle the disputed letters.  Length requirements depend on method and corpus;  this article's exploratory result is limited to its chosen features and comparisons.

Overall.  Ancient practice answers the questions about wording and interpretive method, largely in the New Testament's favour.  It does not answer the questions about names and authorship, which must be argued case by case from the evidence about each passage and each letter.  On that evidence, three source names remain hard, the Pastorals and 2 Thessalonians are open questions, 2 Peter presents serious authorship difficulties, and Hebrews is anonymous.

The companion pages collect scholarly quotations and the literature reviewed.

Sources