When Were the Gospels Written?

External witnesses set latest dates, while most precise Gospel dates remain disputed inferences. Markan priority links Matthew and Luke to Mark; John has separate dating arguments. The ending of Acts supports an early date only as part of a wider case, and a late Acts does not by itself date Luke equally late.

By AI (original draft model unrecorded, research dossiers by Claude Opus 5; quotation pass and corrections by Claude Opus 5.5, thinking level unavailable; additions of 7 October 2026 on Thessalonians, Matthew and the patristic dates by Claude Opus 5.5, thinking level unavailable, from sources read by Claude Sonnet 5.5; focused date-inference revisions by GPT-6.1 Sol, xhigh thinking; strengthening review by Claude Opus 5.5, thinking level unavailable)

Summary

Most books will tell you Mark was written around AD 70, Matthew and Luke in the 80s, and John in the 90s.  Those numbers are more confident than the evidence behind them.

The single most important thing to understand about Gospel dating is that almost none of it is direct.  There is no colophon, no dated manuscript, no ancient writer saying "Mark published in the ninth year of Nero."  Most proposed dates depend on several inferences.  Some are linked by literary dependence; others, especially for John, have separate grounds.

What is actually fixed

Evidence from outside the Gospels sets ceilings:  dates past which a Gospel cannot possibly have been written.  These hold even on the most grudging reading available, granting nothing soft.

Gospel hard latest date what forces it
Luke c. AD 144 Marcion produced an edited Luke at Rome. Two hostile witnesses in different provinces identify his book as Luke's.1
Matthew c. AD 155 Justin Martyr at Rome quotes Matthew's distinctive rewriting of Micah 5:2, and knows the magi and the star.2
John c. AD 155 Justin reproduces Nicodemus' objection about re-entering the womb, a narrative scene from John 3:4 and not a floating saying.3
Mark c. AD 170 Tatian's Diatessaron is the harmony "through four."4
all four as one book c. AD 250 P45, a four-Gospel-plus-Acts codex. Its first editor put it in the first half of the third century.5

There is no reading of the external evidence on which any Gospel is a late-second-century composition.  The latest date found in this research is Markus Vinzent's, who in 2014 argued for "a re-dating of the Gospels to the years between 138 and 144 AD."6  His book rests on an older idea, that Luke is an enlargement of Marcion's Gospel.  Dieter Roth traced that idea through the German debate of the 1840s and 1850s and found that "in previous eras, far from being a consensus position, it convinced only a few."  Roth wrote in 2008, six years before Vinzent, and left open whether the idea "can become broadly persuasive in the twenty-first century."7

And a point against our own side:  the papyri are the worst card, not the best.  Popular apologetics leads with "P52 dates to AD 125."  Brent Nongbri's study of the handwriting concludes that the possible dates for P52 "must include dates in the later second and early third centuries", so as a ceiling it is weaker than Justin.8  This article uses dates a skeptic would accept, because an argument that survives on those is worth more than one that needs its own dating.

What is not fixed

External evidence sets no floors at all.  Every witness above is a latest-possible date.  The earliest-possible date has to come from inside the texts, and there the evidence thins out badly.

If Mark was written in the 60s, the first datable trace of it outside the other Gospels is a century later.  That century of silence is real, and no arrangement of the Apostolic Fathers fills it.  It is a silence about Mark's own words, though, not about whether Mark existed:  if Matthew and Luke used Mark, as the relative order used below assumes, Mark is older than Luke's ceiling of about 144.

The five things the evidence actually supports

1.  The AD 70 argument runs both ways, and each direction is weaker than its advocates claim.

Luke 21:20 replaces Mark's cryptic "abomination of desolation" with "when you see Jerusalem surrounded by armies."  That is the main reason Luke is dated after 70.  But every distinctive word in Luke's siege description is standard Septuagint vocabulary, checked here word by word against the whole Greek Old Testament.  Isaiah 29:3 contains Luke's two signature words, "encircle" and "rampart," in one verse addressed to Jerusalem.9  The observation is C. C. Torrey's, from 1916:  Luke 21:20–24 "cannot be cited as throwing light on the date of the Gospel."10

And Luke's word for the siege-work is χάραξ, a palisade.  Josephus, who was at the siege, never uses that noun in the Jewish War.  He uses "wall" and its derivatives 381 times and "earthwork" 64 times, because what Titus built was earthworks and then a wall.11  Luke's wording is Isaiah's, not the eyewitness's.  He is glossing Mark's obscure phrase out of Isaiah.

But this only removes a reason for a late date.  It supplies no reason for an early one, because a writer in 85 who wanted to clarify Mark for Gentile readers would reach for the same Septuagint vocabulary.  That asymmetry is usually hidden by both sides.

2.  The strongest argument for an early date is what the Gospels do not say.

The ancient accounts of Jerusalem's fall dwell on three things:  the fire that destroyed the sanctuary, the famine inside the walls, and the civil war among the defenders.  Josephus gives all three, the famine as a register of horror and the burning as its climax.  Tacitus, whose surviving text breaks off before the city falls, has the three rival factions and the grain stores they burned.  Cassius Dio has the fire in the temple and the shortage of food.12

Mark 13, Matthew 24 and Luke 21 mention none of them.  No fire, no famine, no faction fighting, no Titus, no legions, no Roman standards sacrificed in the temple court.

That is an observation about the text, not about anyone's presuppositions, and it makes the "prophecy written after the event" reading pay a price it is rarely charged.  It does not prove an early date:  a writer in 85, addressing a Gentile church in Greece, was under no obligation to itemise.

And part of the discourse is demonstrably older than the war.  Paul's first letter to Thessalonica, written about AD 50, already has its "coming," its trumpet, its thief in the night and its birth pangs.13  That dates a tradition and not any Gospel, but it means the second-coming half of the discourse was not composed after the temple fell.

One refinement makes the silence slightly harder to dismiss, and it comes from Luke's own habit.  The usual reply is that Acts' narrative stops around AD 62, so the fall of Jerusalem simply lies outside the story.  But Luke does not confine himself to his story.  At Acts 11:28 a minor prophet named Agabus predicts a famine, and Luke steps outside the narrative frame to tell the reader it came true:  ἥτις ἐγένετο ἐπὶ Κλαυδίου, "which happened in the time of Claudius."14  He breaks frame for a famine.  He never does it for the destruction of the temple, which is the largest prediction in his own Gospel.

Weigh that carefully rather than triumphantly.  It is still an argument from silence, and one authorial aside does not establish a habit.  What it does is remove the easiest answer:  Luke's timeframe cannot be the whole explanation, because Luke was willing to reach past his timeframe when it suited him.

3.  Acts stops in the middle of a trial, and the silences around it are the evidence.

The last quarter of Acts is a legal narrative:  arrest at 21:27, hearings before the Sanhedrin, Felix, Festus and Agrippa, an appeal to Caesar, and transport to Rome for trial.  Eight chapters build toward a verdict.  The book ends with Paul under house arrest for "two whole years," preaching "unhindered," and never reports the outcome.

Set beside that, four things the book does not contain.  Acts records the deaths of Stephen and of James son of Zebedee, but not the death of James the Lord's brother in AD 62, which Josephus does record.15  Not Peter's death.  Not Paul's.  Not Nero's persecution, and not the Jewish War.

And one thing it does contain:  ἀκωλύτως, "unhindered," is the last word of the book.  The adverb occurs nowhere else in the New Testament and nowhere in the Septuagint, checked across all sixty-four files of the tagged text, while its root verb "hinder" is a Lukan habit:  twelve of its sixteen Gospel-and-Acts occurrences fall in Luke–Acts.16  Acts ends by cancelling its own recurring word for obstruction.  Whatever the ending is, it is not careless.

Adolf von Harnack, who had no confessional stake and had earlier dated Luke–Acts to 78–93, changed his mind on exactly this and put the difficulty in its sharpest form:  "Throughout eight whole chapters St Luke keeps his readers intensely interested in the progress of the trial of St Paul, simply that he may in the end completely disappoint them—they learn nothing of the final result of the trial!"17

But the ending will not carry a date by itself, and the honest reason is in the text, not in anybody's opinion:  the last sentence reports a condition, not a verdict.  It does not say Paul was acquitted, released, condemned or executed.  A writer in 62 and a writer in 110 could both compose it.  That is also the judgement of the argument's most thorough modern defender, who declines to rest the case on it.  Colin Hemer wrote that "The argument for early date drawn from the abrupt ending of Acts is an inadequate basis", useful only as confirmation for someone already convinced on wider grounds.18

Harnack is also routinely overstated in apologetic books.  He never states a year for Acts, and he refused to date Matthew before 70:  "In its present shape, however, it should be assigned to the years immediately succeeding that catastrophe."17  Anyone citing Harnack for a pre-70 Matthew is citing him against himself.

4.  Luke precedes Acts, but the interval is unknown.  Their common authorship is supported by the language comparison here.19  An early Acts would require an earlier Luke.  A late Acts, however, could follow a Gospel written years before.  If Acts used Josephus (published 93/94), that would date Acts after 93/94; it would date Luke after 93/94 only with further evidence that Luke itself used Josephus or that the two books were composed close together.

5.  Some standard dates are linked, but they do not all depend on Mark.

One common line of reasoning runs as follows:

  1. Mark ≈ 65–75, anchored on the Neronian persecution at one end and Mark 13 at the other.
  2. Matthew and Luke used Mark, so both are later.
  3. Allow time for Mark to circulate; the often-assumed ten to twenty years is not a measured minimum.
  4. Therefore Matthew ≈ 80–90 and Luke ≈ 80–90.
  5. Luke precedes Acts, with an unknown interval. John is usually dated last on separate grounds.

Markan priority supplies relative order, not a calendar date or a fixed interval.  Remove it and this particular route to the dates of Matthew and Luke disappears.  Their own passages, including Matthew 22:7 and Luke 21:20, still need to be assessed.  Luke preceding Acts and the arguments about John do not depend on Markan priority.

This is not a skeptical characterisation of the field.  It is how the standard conservative textbook describes its own method.  D. A. Carson and Douglas Moo say their decision on Mark's author "will rest almost entirely on external evidence", and that "the main reason for singling out Luke as the author of Luke-Acts comes from the external evidence".  On dates they write that "The date of Luke's gospel is closely intertwined with the dates of Mark and Acts", and that Matthew's date "is marginally bound up with the question of authorship."20  These statements show how literary order, external testimony and internal evidence interact.  They do not make all Gospel dates consequences of one assumption about Mark.

There is also a loop to name.  Mark Goodacre uses the AD 70 evidence to support Markan priority.  Elsewhere the same relative dating is derived from Markan priority.21  The two moves would be circular if the same disputed dating premise were their only support in both directions.  Independent arguments for literary order and for absolute dates can instead be combined; the dependence has to be stated.

The honest verdict

Gospel Working estimates discussed here, not hard bounds What the estimate depends on
Mark AD 40s to early 70s A decision on whether Mark 13 requires the war to have started. James Crossley's case for a date between the mid to late 30s and the mid 40s is the earliest anyone defends and has few followers.22
Luke Early 60s, or 80s-90s The Gospel's own indicators, literary sources and its priority to Acts. An early Acts narrows its latest date; a late Acts alone does not establish a late earliest date for Luke.
Matthew 60s to 90s Which indicators carry more weight. Matthew 22:7's king who "burned their city" points after 70; the temple tax (17:24–27), "immediately" (24:29) and the Sadducees point before it, and the unrevised Danielic phrase (24:15) is at most neutral. The article finds neither set decisive.
John 80s–90s Tradition, theological development and the disputed synagogue-expulsion setting, independent of Markan priority.

These working estimates place the Gospels within living memory of the events they describe.  That conclusion is a historical judgment, not a bound forced by the external witnesses alone.  Those witnesses allow later dates, and second-century proposals also require assessment.  A narrower date should be presented with the assumptions and evidence that produce it.

What external evidence can and cannot do

The ceilings, and how they were set

The rule used here:  the latest date is fixed by the earliest witness that establishes a written Gospel we possess, on a date that is itself independently fixed.  Where a witness's own date is a range, the late end is used, because the point is a ceiling a skeptic cannot argue past.

Marcion, about AD 144, produced an edited Gospel at Rome.  Two hostile witnesses in different provinces, Irenaeus in Gaul and Tertullian in Carthage, identify his book as Luke's.  Even Matthias Klinghardt, who puts Marcion's Gospel before canonical Luke, treats it as an older "proto-Lukan" gospel that Marcion used, not one he wrote.1

Justin Martyr at Rome, about AD 150–160, is the pivot for Matthew and John, and the evidence is better than the usual summaries suggest.

  • Justin attributes to "Micah" a text that is Matthew 2:6 word for word, twenty-two Greek words in a row, stopping two words before Matthew's sentence ends. The Septuagint of Micah 5:1 reads differently.2 Justin is not quoting the prophet; he is reproducing Matthew's rewriting of the prophet. That is proof of the book, not of a shared tradition.
  • Justin reproduces Nicodemus' objection about entering the womb a second time: a scene, not a saying.3
  • And he says the memoirs of the apostles are read aloud in Sunday worship. They are books, in churches, by 155.

One earlier witness would lower Matthew's ceiling if its own date were secure.  Ignatius of Antioch tells the church at Smyrna that Jesus was "baptised by John that all righteousness might be fulfilled by him."23  That is Matthew's phrase.  Only Matthew explains why Jesus was baptised, "to fulfil all righteousness" (3:15), and nowhere else in the New Testament is "all righteousness" fulfilled.  The wording is not identical, and Ignatius does not say he is quoting, so this is strong evidence for Matthew's wording and not proof of a written copy.  Kirsopp Lake, introducing his edition of the letters, reported a general tendency to think "that Ignatius was really a martyr in Rome in the time of Trajan," who reigned 98 to 117.23  On that date Matthew's distinctive wording was known in Asia before 117.  Timothy Barnes argued in 2008 that the letters "were probably written in the 140s."24  Even that date is no later than Justin.  The table keeps Justin only because the date of Ignatius's letters is itself disputed.

Tatian's Diatessaron, about AD 170, is the ceiling for Mark, because a harmony "through four" requires four.

What had to be struck from our own side

Two claims commonly made in apologetic books do not survive checking, and this article does not use them.

1 Clement does not quote a written Gospel.  Its sayings at 13.2 match no Gospel we have, and at 46.8 it twice reads "one of my elect" where every Gospel reads "one of these little ones."  What makes this decisive rather than arguable is that Clement demonstrably can name a written apostolic document when he has one, "Take up the epistle of the blessed Paul the apostle", and never does so for a Gospel.25

Ignatius' most Gospel-like line defeated Eusebius, who could not say where it came from, and Jerome assigned it to the Gospel he had "recently translated", the one he elsewhere calls the Gospel according to the Hebrews.26

The century of silence

State the problem plainly, because it is the real one.

If Mark was written in the 60s, the first secure external trace of it is Tatian, around 170.  A hundred years.  Papias is earlier and better than usually granted, since Jerome's Latin of Eusebius' Chronicle enters him under AD 100,27 but Papias describes a book by Mark without quoting it, so he cannot show that the Mark he means is the Mark we have.  One earlier trace exists and is disputed.  About AD 160 Justin says it is written "in his memoirs" that Jesus renamed the sons of Zebedee Boanerges, a detail found only at Mark 3:17.  Georges Archambault, who edited the Greek text read here, took "his memoirs" to mean memoirs of Peter, and judged that the passage refers to the Gospel of Peter and not to Mark.28

On the order of the Gospels this article uses, the gap is narrower than the table makes it look.  If Matthew and Luke used Mark, then they are themselves the earliest traces of Mark, since they reproduce much of its wording, and Mark is older than Marcion's edited Luke of about 144.  The table prints 170 for Mark only because it grants nothing a minority view disputes:  on the view that Mark came last and drew on Matthew and Luke, Tatian is the first trace.  What stays silent for a century is Mark outside the other Gospels.

Nothing in this article fills that gap.  What the gap does not do is push the composition late:  the ceilings above are ceilings, and a text can be much older than its earliest surviving trace.

The AD 70 argument

The case for a late date, stated fairly

Two passages point after the war, and a historical case places Mark near it.

Luke's rewriting.  Mark 13:14 has "the abomination of desolation standing where it ought not."  Luke 21:20 has "Jerusalem surrounded by armies."  The obvious reading is that Luke knew what had happened and said so.

Matthew 22:7.  In Matthew's version of the wedding-feast parable, the king "sent his troops, destroyed those murderers, and burned their city."  Nothing in the parable requires it.  Harnack, arguing for early dates everywhere else, called this verse "of special weight" for a post-70 Matthew.17

A case for Mark near AD 70 does not have to begin by excluding prophecy.  Joel Marcus reads Mark against the Jewish War: conflict over the temple, Jew-Gentile tensions and competing messianic claims.  He identifies the person implied by the masculine participle in Mark 13:14 with Eleazar's occupation of the temple in 67-68.  His note also allows a Caligula-era source behind the passage.  A source's date and Mark's final composition date need not be the same.29

Marcus prefers a date shortly after 70 because of the destruction prediction's precision, while allowing that people could have anticipated it beforehand.  This article disputes that the prediction is precise enough to settle the date; it cannot dismiss his cumulative case as one assumption about miracles.29

One point often listed with these cuts the other way.  Matthew 24:15 keeps Mark's Danielic phrase, names Daniel as its source and places the abomination "in a holy place," where Luke rewrote it as armies around Jerusalem.  If Luke's rewriting reflects the event, Matthew's retention suggests he did not feel the same pressure.  That makes it an objection to a late Matthew, not evidence for one.  W. D. Davies and Dale Allison, who date Matthew after 70, list the wider form of it among the objections to their view:  "why does Mt 24 fail to recite the characteristic features of that war—such as the burning of the temple?"  Their answer is "Matthew's conservative nature":  he did not find detailed pictures of the siege in his source, Mark, so he did not add them.30  That answer makes the verse neutral.  It does not make it a post-70 indicator.

The case against, and one original observation

Luke's siege vocabulary is Septuagintal, word for word.  Every distinctive term in Luke 19:43–44 and 21:20–24 was traced through the entire Greek Old Testament for this article.9

Luke's word occurrences in the Septuagint notable
χάραξ (rampart) 15 Isaiah 29:3, Ezekiel 4:2, Jeremiah 40:4
κυκλόω (encircle) 93 Isaiah 29:3, 1 Maccabees ×8
ἐδαφίζω (dash to the ground) 6 Hosea 10:14: "they dashed the mother to the ground with her children"
ἐρήμωσις (desolation) 23 Daniel 9:27, 11:31 and 12:11, where Mark's phrase "abomination of desolation" comes from
παρεμβάλλω (encamp) 198 1 Maccabees ×41, among them 6:51, an army encamped against the sanctuary with siege engines, and 13:43, a town encamped against and encircled
πατέω (trample) 19, and 46 more as καταπατέω Isaiah ×7, Lamentations 1:15; as καταπατέω, Zechariah 12:3 (Jerusalem "a stone trampled by all the nations") and 1 Maccabees 3:45

The counts are of dictionary forms in the tagged text.  Three of them correct an earlier draft of this table, which counted spellings and so missed some forms of two words and mixed "walk about" in with "trample."9

Isaiah 29:3 contains both of Luke's signature words, in one verse, addressed to Jerusalem:

"And I will encircle you as David did, and I will cast a rampart around you, and I will set towers around you."

And Luke's word is not the eyewitness's word for AD 70.  He writes χάραξ, a palisade.  Titus raised earthworks and then built a continuous wall of circumvallation.  Josephus, who was present, uses "wall" and its derivatives 381 times across the Jewish War and "earthwork" 64 times.  The noun χάραξ he uses not once.  Its cousin χαράκωμα, "palisade," appears six times in the seven books, and only one of the six is at Titus's siege, where hurdles are stretched over the palisades to shelter the men building the earthworks.11  A man remembering the siege does not reach past the words everybody used for the word Isaiah used.

The weakness of this point should be stated.  In the Septuagint χάραξ stands for the siege mound an attacker throws up against a city.  At Isaiah 37:33 and Ezekiel 4:2 it translates the Hebrew word for exactly that, sollelah.  A siege mound is what Titus built.31  So Luke's word does not contradict the event.  The argument is only about vocabulary:  Luke's term is the Greek Bible's, and the man who was there used other terms.

Luke is also editing Mark's sentence, not composing a fresh one.  Set them side by side and the frame is identical:  "When you see … then let those in Judaea flee to the mountains."  Luke swapped only what sits between.  And he kept Mark's Danielic noun "desolation."  He is glossing an obscure phrase for Gentile readers, which is what he does elsewhere with Aramaic and with Jewish customs.

Mark 13:2 is a building idiom, not a demolition report.  The phrase "stone upon stone" occurs exactly once in the entire Septuagint, and it is Haggai 2:15, of the construction of the second temple, "before stone was laid upon stone in the temple of the Lord."32  Mark 13:2 is Haggai run backwards.

Warning

An apologetic own-goal to avoid.  "Not one stone was left on another, exactly as predicted" is an overstatement, and every reader has seen a photograph of the Western Wall.  Three distinctions keep the claim truthful.  Jesus points at "these great buildings":  the structures, not Herod's retaining wall.  Josephus records that the sanctuary and city were razed so completely as to leave visitors no ground for believing the place had been inhabited.  And Josephus names the exceptions in the same breath:  three towers, and the stretch of the city wall on the west side, left deliberately as a camp for the garrison.  That is not the Western Wall of the photographs, which is part of the Temple Mount's retaining wall and which Josephus does not mention here.33  Even the eyewitness's "levelled completely" is hyperbole with stated exceptions.  Mark 13:2 is the same register of speech.

Paul had the discourse twenty years before the war

The return-of-Christ half of the Olivet Discourse was in circulation by about AD 50, because Paul uses it.  1 Thessalonians was written about AD 50 or 51, and its authorship is not in dispute:  "no contemporary scholars of repute seem to doubt the authentic Pauline character of the letter."34  Its teaching on the Lord's return, and that of 2 Thessalonians, shares a run of words with the discourse.  The counts below are of dictionary forms in a tagged Greek New Testament.13

Word In the discourse In the letters to Thessalonica Rest of the New Testament
παρουσία (coming) Matthew 24:3, 27, 37, 39; no other Gospel has the word 1 Thess 2:19, 3:13, 4:15, 5:23; 2 Thess 2:1, 8, 9 13
σάλπιγξ (trumpet) Matthew 24:31; no other Gospel has the word 1 Thess 4:16 9
ἀπάντησις (meeting) Matthew 25:6 1 Thess 4:17 1, Acts 28:15
ὠδίν (birth pang) Matthew 24:8, Mark 13:8 1 Thess 5:3 1, Acts 2:24
κλέπτης (thief), of the Lord's coming Matthew 24:43, Luke 12:39 1 Thess 5:2, 4 2 Peter 3:10, Revelation 3:3 and 16:15
θροέομαι (be alarmed) Matthew 24:6, Mark 13:7 2 Thess 2:2 0
ἐπισυνάγω, ἐπισυναγωγή (gather, gathering) Matthew 24:31, Mark 13:27 2 Thess 2:1 the noun once, Hebrews 10:25; the verb 6 times
ἀνομία (lawlessness) Matthew 24:12; no other Gospel has the word 2 Thess 2:3, 7 9
αἰφνίδιος (sudden) Luke 21:34 1 Thess 5:3 0
ἐφίστημι (come upon) Luke 21:34 1 Thess 5:3 19, of them 17 in Luke and Acts
ἐκφεύγω (escape) Luke 21:36 1 Thess 5:3 6

Paul presents this as teaching his readers already have.  What he says about the dead in Christ he says "by a word of the Lord" (1 Thessalonians 4:15).  On the timing he writes, "you yourselves know accurately that the day of the Lord comes like a thief in the night" (5:2).  He had founded the church only months before.

David Wenham worked through the parallels in 1981.  He discounted every one that two Jewish writers could have reached separately from older apocalyptic writing,35 and concluded of the rest that "Paul is seen to be our earliest witness to the Synoptic Apocalypse and his evidence must be brought in when the different Synoptic accounts are compared."35

Does this show that Paul had read Matthew?  Bernard Orchard argued in 1938 that it does.  His best observation is that the parallels sit together in one block of Matthew, 23:31 to 25:46, while Luke's are scattered over several chapters.36  He concluded that Paul depended on that block, and that "this passage is something absolutely primordial and must be dated somewhere between 40 and 50 A.D."36  In 1987, writing with Harold Riley, he put the whole Gospel "before 46/47."37  If that were right, Matthew would be the oldest book in the New Testament.

The evidence stops short of that, for three reasons.

First, the closest wording of all is Luke's and not Matthew's.  "Sudden," "come upon" and "escape" stand together at 1 Thessalonians 5:3 and at Luke 21:34–36, a passage with no parallel in Matthew or Mark, and "sudden" occurs nowhere else in the New Testament.  Orchard saw this and answered that Luke borrowed the phrase from Paul's letter.36  Wenham argued the reverse, that Luke's wording is the older form.35  A letter that overlaps the Matthew-only and the Luke-only parts of the discourse is more simply explained by a source older than both Gospels.

Second, Orchard's own argument allowed for that.  The shared wording, he wrote, requires "either indebtedness of both to a common source, or else the direct indebtedness of one to the other."36

Third, the scholars most inclined to agree with him did not.  J. A. T. Robinson, who dated every book of the New Testament before 70, said Orchard's conclusion "seems to me to be pushing the evidence much too far."38  Allan McNicol, who holds with Orchard that Matthew was the first Gospel written, concluded that "sufficient literary evidence constituting a number of parallels does not exist to cause this investigator to hypothecate that there is copying between Paul and the Synoptics."39

What Robinson kept is the part that bears on dating:  "the evidence of Thessalonians at any rate shows that this way of thinking was rife in the year 50.  The marks of it in Matthew 24 cannot therefore be used to require any later date for that gospel."38

Three limits should be stated.

  • 2 Thessalonians is a disputed letter. By 1990 "the balance of expressed opinion has tilted decisively against the authenticity of 2 Thessalonians."34 Three rows of the table rest on it alone. The rows from 1 Thessalonians stand without it: the coming, the trumpet, the meeting, the thief, the birth pangs, and Luke's three words.
  • Paul has the return, not the ruin. Neither letter says the temple will fall. The nearest either comes is the "man of lawlessness" who takes his seat "in the temple of God" (2 Thessalonians 2:4), which assumes a temple still standing. So Paul does not show that Jesus predicted the destruction. He shows that the discourse was not built from nothing after it.
  • Older than Paul is not the same as from Jesus. Wenham says so himself: "Even if the common Pauline/Synoptic traditions are pre-Pauline, this is not the same as saying that they were thought to derive from Jesus."35 Paul's "by a word of the Lord" shows that he, at least, thought they did.

The methodological question

How much of the post-70 consensus rests on ruling out predictive prophecy in advance?

Some of it, and it should be said without either exaggerating or excusing.  But two things keep this from being the whole story, and both cut against a lazy apologetic use of it.

First, predicting the temple's destruction required no miracle.  Rome had destroyed cities before.  And Jesus was not the only Jew who predicted this one:  Josephus reports a peasant named Jesus son of Ananias crying "woe to Jerusalem" through the streets from about AD 62 until he was killed during the siege.40  A prediction that a rebellious province's capital would be sacked is not a supernatural claim.

Second, the more specific the prediction, the stronger the case for late composition, and the Gospels are not specific.  That is precisely why the absence of fire, famine and faction matters.  The Olivet Discourse reads like the stock apocalyptic vocabulary of the Hebrew prophets, because that is what it is made of.

The end of Acts

The text

Acts 28:30–31 ends:  Paul "remained two whole years in his own rented lodging … proclaiming the kingdom of God and teaching the things concerning the Lord Jesus Christ with all boldness, unhindered."

Four things in the Greek carry weight, and one of them is genuinely striking.

  • "Two whole years" is a defined, completed period. The verb is aorist. The natural reading is that the two years are over when the writer writes. That is not the same as saying he is standing at the end of them: a historian in 110 would also use the aorist.
  • ἀκωλύτως, "unhindered," is the last word of the book. It occurs nowhere else in the New Testament and nowhere in the Septuagint, checked across all sixty-four files of the tagged text.16 Its root verb, however, is a Lukan favourite: twelve of the sixteen Gospel-and-Acts occurrences of "hinder" are in Luke–Acts. Acts ends by cancelling its own recurring word for obstruction. Whatever the ending is, it is not careless.
  • But it reports a condition, not a verdict. It does not say Paul was acquitted, released, condemned, or executed. That gap is where the whole debate lives.

The argument, and its real force

The last quarter of Acts is a legal narrative:  arrest, hearings before the Sanhedrin, Felix, Festus and Agrippa, the appeal to Caesar, the voyage to Rome for trial.  Eight chapters of build-up to a verdict that never comes.

Add the silences.  Acts records the deaths of Stephen and of James son of Zebedee.  It does not record the death of James the Lord's brother in AD 62, which Josephus does record.  It does not record Peter's death, Paul's death, Nero's persecution, or the Jewish War.

Three of these silences have a particular shape.  Colin Hemer, who lists them, also states their limits.

  • James. James the Lord's brother is not a minor figure in Acts. He gives the deciding speech at the Jerusalem council (15:13–21), and Paul reports to him on his last visit (21:18). Josephus says the high priest Ananus, a Sadducee, convened a council while one Roman governor was dead and the next was still on the road, had James stoned, and was removed by King Agrippa after three months in office, when the new governor wrote to him in anger.15 That is the kind of story Acts tells again and again: temple authorities move against a Christian, and Roman officials find no case, as Gallio does at 18:14–16 and Festus and Agrippa do at 26:31–32. Hemer's own verdict on it: "This point, which depends for its force on a prejudgment of Luke's purpose and setting, should not be pressed."18
  • Nero. Acts presents Paul's appeal to Caesar (25:11) as his protection. The Caesar was Nero, who in 64, by Tacitus's account, blamed the fire of Rome on the Christians and put a great number of them to death with elaborate cruelty.41 Hemer: "The chosen appeal to Roman justice is harder to place when the character of the later Nero was known, or after the conflicts which followed."18
  • The voyage. Acts 27 gives a long chapter to the sea voyage to Rome: the winds, the harbours of Crete, the crew's manoeuvres and the shore of Malta. Hemer argued that "these inconsequential details are hard to explain except as vivid experiences recalled at no great distance in time."18 David Seccombe restated the point in 2020. The voyage is one of the longest sections of the book and "Verse for verse, it is the least theological," which he reads as the work of an author writing before "the extraordinary disruptions of the following decade."42

The skeptic's reply limits all three.  A writer who wanted Roman readers to see the faith as harmless had his own reason, at any date, to leave out a Christian leader's execution and an emperor's persecution.  And vivid detail can come from a travel diary used later.  Seccombe himself records that F. C. Baur, who dated Acts to the middle of the second century, "felt the author must have had in his hands a genuine diary of the voyage."42  These points make an early date fit the book more easily.  They do not prove it.

Harnack's conclusion, reached against his own prior position:

"the concluding verses of the Acts of the Apostles, taken in conjunction with the absence of any reference in the book to the result of the trial of St Paul and to his martyrdom, make it in the highest degree probable that the work was written at a time when St Paul's trial in Rome had not yet come to an end."17

The answers, stated at full strength

  • Acts 20:25 and 20:38 have Paul telling the Ephesian elders they will not see his face again. Read as foreshadowing, the author knows Paul is dead. The early-date reply turns the verse around. Paul's own letters expect release and more travel (Philippians 1:25, Philemon 22), and 1 Timothy 1:3 and 2 Timothy 4:20 put him back at Ephesus and at Miletus, where the farewell speech was given. If he did return, a writer who knew his later life left standing a prediction that events had falsified. Seccombe argues that the verses "may be evidence that Luke did not yet know of Paul's release (if he returned to Ephesus)."42 The reply rests on letters whose authorship most critics deny.
  • The structural reading. Acts 1:8 sets the programme: "to the ends of the earth." The book ends when the gospel reaches Rome, because that is where its programme ends, not where its information stops.
  • A deliberate open ending. Ancient works do end in ways modern readers find abrupt. Daniel Marguerat finds the device, which he calls narrative suspension, in Homer and Herodotus, and concludes: "Everything leads one to think that the author of Acts made use of this model."43 His examples are four and five centuries older than Acts, which is the weak point of the comparison.
  • The Josephus argument, which is the strongest late-dating card. Acts 5:36–37 names Theudas and then Judas the Galilean in that order, the same pairing and the same order as Josephus, Antiquities 20.97–102, published in 93/94.15 And Acts places Theudas before Judas, who revolted decades earlier. This is a genuine problem and it is already a named unresolved residue elsewhere in this archive. If Acts used Josephus, Acts is after 93/94. Luke precedes Acts, so that result alone does not put Luke after 93/94.

Where this lands, and it is less than apologists claim.  Colin Hemer is the most thorough modern defender of an early Acts, and he will not rest the case on the ending:  "The argument for early date drawn from the abrupt ending of Acts is an inadequate basis, though it offers a neat reconstruction and may be a valuable confirmatory consideration for one who is convinced of the early date upon a different and broader base."  He adds:  "It is thus difficult to explain the ending exactly, regardless of the view one takes about the early or late date of the book."18  Hemer still dates Acts to 62 and still counts the ending toward that date, but as one item in a cumulative case.

An article that leads with "Acts must be early because it stops in 62" is leading with the weakest form of a case whose own champion will not rest on it.  What the ending supports is narrower and defensible:  the author had nothing to say about anything after the two years, which is odd on a late date and unremarkable on an early one.  That is a probability, not a proof.

The answers to it are real, and every one of them is an explanation of a silence rather than positive evidence of a late date.  The Josephus argument is the exception: it supplies a possible source and therefore a possible earliest date for Acts.  It is one reason an early date is not established by the ending alone.  It neither fixes the interval between Luke and Acts nor exhausts the arguments for later dates.

The individual Gospels

Mark.  The traditional Roman setting under Nero puts it in the 60s.  James Crossley has argued for a "date for Mark between the mid to late thirties and mid-forties", on the ground that Mark assumes a law-observant setting which Matthew and Luke have to spell out for their readers.22  He is not a Christian apologist, which makes the argument evidentially interesting; he has also attracted few followers, which is the honest other half.  The two reviews the Review of Biblical Literature ran in January 2006 show both halves.  David du Toit wrote that Crossley "clearly shows that the consensus date around 70 C.E. is based on arguments of doubtful validity," and also that his evidence "is not conclusive and does not necessarily demand an early date for Mark's Gospel."  John Painter was not persuaded, and added:  "flimsy evidence is flimsy for all comers."44  The external ceiling is 170; the internal floor is the events themselves.

Matthew.  The post-70 indicator is 22:7's burned city, the strongest internal argument for a late date in any Gospel, and it points to the 70s or 80s.  Harnack, who wanted early dates, conceded this one.  The retention of Mark's Danielic phrase at 24:15, sometimes paired with it, points the other way or nowhere, as shown in the AD 70 part above.

But Matthew also contains the best internal indicator pointing the other way, and the two sit inside the same book.  Matthew 17:24–27 is the only passage in any Gospel to name the δίδραχμον, the two-drachma temple tax, and Jesus tells Peter to pay it "so that we may not give offence."  After the war that tax did not cease.  It was redirected.  Josephus, Bellum Judaicum 7.218, on Vespasian's settlement:

φόρον δὲ τοῖς ὁπουδηποτοῦν οὖσιν Ἰουδαίοις ἐπέβαλεν δύο δραχμὰς ἕκαστον κελεύσας ἀνὰ πᾶν ἔτος εἰς τὸ Καπετώλιον φέρειν, ὥσπερ πρότερον εἰς τὸν ἐν Ἱεροσολύμοις νεὼν συνετέλουν. "He laid a tribute on the Jews wherever they lived, ordering each to bring two drachmas every year to the Capitol, just as they had formerly contributed to the temple in Jerusalem."45

So for a reader after AD 70, a passage in which Jesus endorses paying the two drachmas endorses paying for the temple of Jupiter Capitolinus.  Matthew keeps it.  He also keeps the instruction to be reconciled with your brother before offering your gift at the altar (5:23–24), and the argument about swearing by the gold of the temple (23:16–22), all of them teachings a congregation could no longer act on.

Three smaller details in Matthew point the same way.

  • The Sabbath. Mark 13:18 has "pray that it may not happen in winter." Matthew 24:20 has "pray that your flight may not be in winter or on a Sabbath" (μηδὲ σαββάτῳ). Luke drops the sentence.
  • "Immediately." Mark 13:24 puts the darkened sun "in those days, after that tribulation." Matthew 24:29 begins "Immediately (εὐθέως) after the tribulation of those days." A writer who knew that years had passed since the siege had reason to loosen Mark's link, and Matthew tightens it. Hagner names this as the detail that changed his own mind about the date.46 In his words, the evangelist "aggravates the problem considerably by his insertion of εὐθέως, 'immediately,' in 24:29, which leaves the clear impression that he expected the parousia of the Son of Man to occur in close succession to the fall of Jerusalem."47
  • The Sadducees. Matthew names them seven times, Mark once and Luke once.48 Robinson: "Since this was a group whose power disappeared with the destruction of the temple, preoccupation with them argues strongly for an earlier period."38

Each has an answer.  A Sabbath clause suits a law-observant readership at any date.  A writer in the 80s could still expect the end soon.  And five of Matthew's seven references are the fixed pair "Pharisees and Sadducees," which reads more like a label for Jesus's opponents than like news of a living party.48  The scholars who press these points grant that they prove nothing.  R. T. France:  "It may be readily conceded that none of these arguments for a pre-70 date for Matthew is conclusive."49  Hagner:  "The evidence briefly canvassed here cannot, of course, demonstrate an early date for Matthew.  But a cumulative argument of some strength emerges from it."46  W. D. Davies and Dale Allison, after reviewing the arguments on both sides, conclude that "Matthew was almost certainly written between A.D. 70 and A.D. 100, in all probability between A.D. 80 and 95."30

How much this is worth, stated carefully.  It is not a floor.  An evangelist may preserve a saying because Jesus said it, not because his readers can use it, and faithful preservation of obsolete material is exactly what a careful tradent does.  What it is worth is this:  Matthew contains indicators pointing in both directions, and any dating of Matthew has to spend one of them.  Whoever leans on 22:7 for a post-70 Matthew owes an account of why the same author kept the temple tax;  whoever leans on the temple tax for a pre-70 Matthew owes an account of the burned city.  This article does not think either side has paid.  Donald Hagner reached the same verdict from inside the conservative camp:  "the dogmatism of critical orthodoxy concerning a post-70 date is unwarranted.  But this is no reason to give rein to a similar dogmatism concerning a pre-70 date."47

Luke and Acts.  The ending of Acts and possible use of Josephus are important, but they are not an exhaustive choice between an early and a late date.  Luke's preface also places it after earlier narratives, without measuring the interval.  And some late-date proposals appeal to the use of Paul's letters and to second-century disputes.  Joseph Tyson, for example, dates Acts to about 120-125 because he reads its portrait of Peter and Paul as answering the Marcionite movement, not merely because of the Josephus parallels.50  This article does not establish that interpretation.  It does rule out presenting the end of Acts as the only relevant variable.

One further datum could connect the date of Luke with another New Testament letter.  1 Timothy 5:18 reads:

λέγει γὰρ ἡ γραφή· Βοῦν ἀλοῶντα οὐ φιμώσεις, καί· Ἄξιος ὁ ἐργάτης τοῦ μισθοῦ αὐτοῦ. "For the Scripture says, 'You shall not muzzle an ox while it treads out the grain,' and, 'The labourer is worthy of his wages.'"

The first quotation is Deuteronomy 25:4.  The second is the saying of Luke 10:7, all six words in the same forms and the same order, lacking only Luke's connecting word γάρ, "for."  And it is not Matthew, whose version of the same saying reads τῆς τροφῆς, "his food," rather than τοῦ μισθοῦ, "his wages."  The match establishes a shared form of the saying.  Whether it establishes a quotation from written Luke as Scripture is a further question.

Two things keep this from settling anything, and both are visible in the text rather than in anyone's opinion.  First, it dates Luke only as early as it dates 1 Timothy, and Pauline authorship of 1 Timothy is exactly what most critics deny.  Second, the same pairing appears in a letter nobody disputes, and there it runs the other way.  1 Corinthians 9:9 quotes the same verse of Deuteronomy, and 9:14 gives the same principle about paying the worker as "the Lord commanded":  a dominical saying, in Paul's own words, not a quotation of a book.

A third question concerns the scope of "Scripture says": does it introduce both sayings or only Deuteronomy?  Philip Towner thinks extending it to the saying of Jesus is unlikely.51  Even if it covers both, the authority given to a saying of Jesus does not by itself show that it was copied from a written Gospel.  The verse therefore dates Luke only if written dependence is established; the close wording alone does not establish it.52  The alternatives are assessed more fully in Did Paul Write the Disputed Letters?, Part 7.

John.  Traditionally last, written at Ephesus in old age.  The scholarly dating rested for a generation on J. Louis Martyn's reading of the "put out of the synagogue" passages as reflecting a formal expulsion of Christians around AD 85–90.  Martyn tied those passages to a synagogue prayer against heretics that he said was "published by the Jamnia academy about A.D. 85."  That anchor has given way.  Reuven Kimelman showed in 1981 that "There is no unambiguous evidence that Jews cursed Christians during the statutory prayers", and Ruth Langer's 2012 history of the prayer records that from the mid-1990s "some prominent New Testament scholars who had previously advocated Martyn's theory backed away from it".  It is not dead.  Langer also reports Joel Marcus defending "the basics of Martyn's reconstruction" in 2009.53  What has replaced it is mostly an argument from theological development, which is weak.

John 5:2's present tense, "there is in Jerusalem … a pool", is sometimes used for an early date.  It is a real oddity, because John elsewhere puts his topography in the past tense.  But it is not strong, because the pool outlived the war.  Eusebius, early in the fourth century, says its twin pools were still being pointed out, and the Bordeaux Pilgrim described them in 333.  A writer in 95 could say "is" and be telling the truth.54

Important

The largest honest concession in this article.  Nothing in the per-Gospel evidence produces a floor worth the name.  Every internal indicator either turns out to be conventional language any writer could use at any date (Matthew 22:7; "to this day"; "let the reader understand"), or survived the event it was supposed to date (the pool of Bethesda was still being shown to pilgrims in the 330s), or has lost many of the specialists who once held it (the synagogue-expulsion argument).  The ceilings are solid.  The floors are not there.

Two verses commonly used for an early date have to be given up.  Matthew 27:8's "to this day" and 28:15's "to this very day" are Septuagintal narrator's formulas, and the Septuagint uses them of short spans.  Joshua 22:3 has Matthew's wording, "to this very day," of the few years of the conquest.  Deuteronomy 3:14 has the sister formula, "to this day," of a name given within Moses' own lifetime.  The two verses prove nothing about a late date and nothing about an early one.55

And one datum points late, which this article owns rather than buries.  John 21:23 corrects a rumour that the beloved disciple would not die.  That fits a community facing the death of a long-lived disciple, and so can support a late date for chapter 21.  But the verse gives neither his age nor a year, and a rumour of exemption from death could arise earlier.  It supplies no firm AD 80 floor, and the date of a possible concluding chapter cannot simply be assigned to every earlier part.

Matthew 22:7, weighed honestly.  "Sent his troops and burned their city" is a genuine Old Testament formula.  The Septuagint uses Matthew's verb eighteen times of a town or its buildings being burned, including Jeremiah on the first burning of Jerusalem.56  But that cuts both ways.  A writer after AD 70 would reach for exactly that formula, because it was his Bible's language for the same event happening to the same city.  W. D. Davies and Dale Allison make the sharper reply:  "even if Mt 22.7 comes from the OT or reproduces a common prophetic topos, we must inquire into the reason for its placement in the parable of the great supper, and for this the events of 70 lie near to hand."  Their conclusion is that the verse, "while it does not demand a date after A.D. 70, does strongly imply one."30  The topos weakens the verse as a proof;  it does not remove it as the best post-70 indicator in the corpus.

Objections and responses

"The Gospels must be late, because they contain a prophecy that came true." That reasoning assumes its conclusion.  It is also unnecessary here:  the destruction of a rebellious province's capital was predictable, and at least one other Jew predicted this one.  The stronger form of the objection is about specificity, and on specificity the Gospels are notably poor:  no fire, no famine, no factions, none of the details every real account contains.

"P52 proves John was circulating by AD 125." It does not, and this article does not use it.  Palaeographic dating of a scrap this small carries a range of a century or more, and Nongbri's window for P52 runs into the early third century.8  The ceiling for John is Justin, not the papyrus.

"The church fathers tell us when the Gospels were written." They tell us relative order and a few circumstances, and they contradict each other on the circumstances.  Irenaeus says Mark wrote after the "departure" of Peter and Paul, which is usually read as their deaths.  Clement of Alexandria says Peter was alive and heard of it.57

For Matthew there is one ancient time marker.  Irenaeus, about AD 180, says Matthew issued his Gospel "while Peter and Paul were preaching at Rome, and laying the foundations of the Church."58  R. T. France reads that as a date:  "This would place it in the early sixties at latest."49  Paul is named, and Paul reached Rome about AD 60, so as a date it cannot mean much earlier.  John Chapman argued in 1905 that it is not a date at all.  The clause "cannot possibly be pressed to mean 'during the time that Peter and Paul were preaching'";  its point is the contrast of "at Rome" with "among the Hebrews."59  Harnack and Robinson took Chapman's side.38  Eusebius adds one circumstance and no year:  Matthew wrote "when he was about to go to other peoples."60

Only one ancient text puts a year on a Gospel, and the line is a later addition.  Jerome's Chronicle enters Peter's move to Rome at the second year of Claudius, AD 42, and "Mark the evangelist, Peter's interpreter, preaches Christ in Egypt and Alexandria" at the third, AD 43.  The old printed text adds, among the entries for AD 39 to 44, "Matthew in Judaea is the first to write a gospel."  The Merton manuscript has a blank space there, and the English edition made from the modern critical text prints the line only in a footnote.27  The Armenian version of Eusebius's own Chronicle has the Peter and Mark entries and no entry for Matthew.61

"Papias and Irenaeus converge on AD 42 for Matthew." This is David Pallmann's argument, and it joins two chains.62  First, Eusebius says Matthew wrote when he was about to leave for other peoples, a passage Charles Hill traces to Papias,63 and three early texts say the apostles stayed twelve years in Jerusalem.  Twelve years from a crucifixion in AD 30 is 42.  Second, Irenaeus says Matthew wrote while the church at Rome was being founded, and Eusebius and Jerome put Peter's arrival there in 42.

Parts of this are sound.  The twelve-year tradition is real and early.  Apollonius, about AD 200, knew "a tradition that the Saviour commanded his apostles not to depart from Jerusalem for twelve years,"60 and the Preaching of Peter and the Acts of Peter have the same number.6465  A first visit by Peter to Rome in 42 has been defended by George Edmundson and John Wenham, and Robinson thought Edmundson "may be right."6638  And Orchard and Riley put Matthew in the same decade on other grounds.

But neither chain reaches Matthew.

  • No ancient writer connects Matthew to the twelve years. Eusebius's sentence has no year and does not mention Jerusalem. None of the three twelve-year texts mentions Matthew, and none says where the twelve years begin. "After the crucifixion" is Robinson's wording, not theirs.38 The join is made by the modern reader.
  • The Acts of Peter works against the date. In its story Peter sails for Rome when his twelve years are up, and he arrives after Paul has already been in Rome and left for Spain.64
  • Irenaeus names Paul. Paul was not in Rome in 42. His letter to the Romans, about AD 57, says he has "often intended" to come and been prevented (1:13) and has wanted to "for many years" (15:23). Read as a date, Irenaeus's clause means the 60s, which is how France takes it. Read with Chapman, it is no date. On neither reading is it 42.
  • The year 42 for Peter is the product of a sum. Edmundson, who defends the date, shows where it comes from: "the date of this martyrdom was according to the received tradition the fourteenth year of Nero or 67 A.D.; if then we deduct twenty-five years, we arrive at 42 A.D."67 The twenty-five years are a figure from the lists of Roman bishops. Another ancient list, the Liberian Catalogue, has the same twenty-five years and runs them from AD 30 to 55.67
  • Hill's Papias fragment is a proposal. Richard Bauckham grants that "Hill has made a good case for his proposal that Papias was the source," and still doubts the claim's main point, that the passage reports what Papias said about John's Gospel: "Attractive as this claim is, I doubt if it can be sustained."68
  • Both fathers are speaking of a Gospel in Hebrew. Irenaeus says "in their own dialect" and Eusebius "in his native tongue." Whatever date this tradition carries belongs to that book. Whether it is the Greek Matthew read today is a separate question.

So the gain is smaller than the conclusion.  Peter may have been in Rome in the 40s, and the tradition that the apostles left Jerusalem about then is old.  Neither fixes a year for Matthew.

"Silence proves nothing." Correct as a general rule, and this article says so.  What silence can do is shift a burden.  A book that spends eight chapters on a trial and never reports the verdict is not merely silent; it is silent in a place where a reader has been taught to expect speech.

"Acts used Josephus, so the early-dating case is finished." The Theudas passage is a real problem and this article does not minimise it.  But the argument requires the parallels to be literary dependence rather than shared common knowledge of recent revolts.  Steve Mason, the Josephus specialist who argues the case for dependence, writes that "we cannot prove beyond doubt that Luke knew the writings of Josephus", and says only that he finds dependence "easier to believe" than coincidence.69  Colin Hemer reads the same contradictions the other way:  "If either used the other, he misused him."18

The numbers are the concrete form of Hemer's point.  Where Acts and Josephus describe the same rebel, Acts never has Josephus's figure.  Josephus says Theudas persuaded "the greater part of the crowd";  Acts 5:36 says "about four hundred" men.15  For the Egyptian prophet, Josephus's Jewish War gives "about thirty thousand" followers;  Acts 21:38 has the tribune speak of "the four thousand men of the Assassins."45  And Acts reports that Judas the Galilean died, which Josephus does not.  Mason answers this himself, and his answer is fair:  "numbers are notoriously fluid in ancient texts (compare Josephus' own differences with himself on the size of the Egyptian's following!)"69  Josephus's own Antiquities gives no total for the Egyptian, only 400 killed and 200 captured, and a reader who thinks Luke worked from memory may notice that this 400 is the number Acts gives Theudas.  So the numbers do not refute dependence.  They do show that a reader who wants Luke to have used Josephus must also have him ignore or misremember every figure in the passages he is supposed to have used.

"Conservative scholars date the Gospels early because of what they believe." Some do.  Harnack did not, and Crossley does not.  The reverse is equally true:  a naturalistic premise, applied without argument, is doing work in some of the standard late dates.  The right response to both is to state which arguments survive without the premise, which is what the ceilings section above does.

What is not known

  • No Gospel has an externally fixed earliest date. All the external evidence is ceilings.
  • Whether the ending of Acts reflects the author's date or his design. A century of argument has not settled it.
  • Whether Acts used Josephus. A consequential open question for the date of Acts. Its implications for Luke depend on the interval between the books and on whether the Gospel itself used the same source.
  • The date of Q, if Q existed. Undetermined by anything.
  • Whether Papias' "Mark" is our Mark. He describes without quoting.
  • Whether Paul's source for the discourse was written. The letters to Thessalonica show the material existed by about AD 50. They do not show what form it was in.

Sources

  1. 1.

    Irenaeus.  Adversus Haereses.  n.d.  Section 1.27.2:  In the Ante-Nicene Fathers translation, Irenaeus says of Marcion:  “mutilates the Gospel which is according to Luke.”  Section Tertullian, Adversus Marcionem 4.2.4:  Tertullian says, in Latin, which means of the writers we have that Marcion seems to have picked Luke to cut up:  “ex iis commentatoribus quos habemus Lucam videtur Marcion elegisse quem caederet”.  Section Tertullian 1.19.2:  The date comes from Tertullian's arithmetic at this place, about 115 and a half years from the fifteenth year of Tiberius to Marcion, which is about AD 144.  Section Klinghardt 2008, pages 1–27:  Matthias Klinghardt's article in Novum Testamentum 50 (2008) is titled.  “The Marcionite Gospel and the Synoptic Problem:  A New Suggestion”.  Section Klinghardt 2008, page 26:  and speaks at this page of.  “the 'proto-Lukan' gospel which was used by Marcion.”  Section Vinzent, as its publisher describes it:  The publisher describes the book as giving Marcion a.  “crucial role in the Gospel making”.  A work by Irenaeus, one of two hostile witnesses in different provinces who identify Marcion's edited Gospel as Luke's.  The article cites it together with Tertullian, the other witness, as two separate hostile witnesses who give the same identification. ↩ ↩

  2. 2.

    Justin Martyr.  First Apology.  n.d.  Section 34.1:  Justin attributes to Micah a text that reproduces Matthew 2:6 for twenty-two consecutive Greek words, from Καὶ σὺ Βηθλεέμ to τὸν λαόν μου, and stops before Matthew's last two words, τὸν Ἰσραήλ.  The Septuagint of Micah 5:1 (5:2 in English Bibles) reads differently, with οἶκος τοῦ Εφραθα and ὀλιγοστὸς … ἐν χιλιάσιν Ιουδα.  Justin is reproducing Matthew's rewriting of the prophet, not the prophet.  This is proof of the book rather than of a shared tradition.  Section Dialogue with Trypho 106.4:  The magi and the star are at this place, in the Ante-Nicene Fathers translation:  “when a star rose in heaven at the time of His birth, as is recorded in the memoirs of His apostles, the Magi from Arabia, recognising the sign by this, came and worshipped Him.”  A work by Justin Martyr, in which he attributes to the prophet Micah a text that is Matthew's own wording.  The article uses it because Justin quotes Matthew's distinctive rewriting of Micah 5:2 and knows the magi and the star, which shows that Matthew's Gospel was in use by about AD 155. ↩ ↩

  3. 3.

    Justin Martyr.  First Apology.  n.d.  Section 61.4–5:  Justin reproduces Nicodemus' objection about entering the womb a second time (John 3:4).  Section 67.3:  The memoirs are read aloud in Sunday worship at this place.  A work by Justin Martyr.  The article uses it to show that Justin reproduces a narrative scene from John 3:4, Nicodemus' objection about re-entering the womb, and not a floating saying. ↩ ↩

  4. 4.

    Eusebius.  Historia Ecclesiastica.  n.d.  Section 4.29.6:  On Tatian, Eusebius says:  “Τὸ διὰ τεσσάρων τοῦτο προσωνόμασεν”.  “he named it the Through Four.”  Eusebius adds.  “οὐκ οἶδ᾽ ὅπως”.  “I do not know how”.  so he had not seen the book.  Independently supported by P45 and by the Dura fragment 0212.  Section Kraeling 1935, page 6:  In Carl H. Kraeling's A Greek Fragment of Tatian's Diatessaron from Dura (Studies and Documents 3, London: Christophers, 1935), the Dura fragment was dug out of a siege embankment that, in its first editor's words,  “was constructed after 254 and before 256-257 A.D.”  the second date being the fall of the city.  Whether the fragment is Tatian's own harmony is disputed.  The church history in which Eusebius reports the name Tatian gave his harmony of the four Gospels.  The article uses the name to show that a harmony of all four Gospels existed by about AD 170, which is its date for Mark. ↩

  5. 5.

    Kenyon, Frederic G.  The Chester Beatty Biblical Papyri, Fasciculus II: The Gospels and Acts, Text.  London: Emery Walker.  1933.  The first edition of P45, the papyrus of the four Gospels and Acts.  The article uses it because its first editor dated P45 to the first half of the third century.

    1. 5a.

      Page x:  “The date which I should assign to it is the third century, and the first half of it rather than the second.”  Kenyon adds that H. I. Bell and W. Schubart reached the same estimate, and that A. S. Hunt.  “is inclined to place it in the second rather than the first half of the century.”  The table uses 250, the late end of Kenyon's own range, by the rule this article applies to every witness.  On Hunt's reading the figure would be later in the same century. ↩

  6. 6.

    Vinzent, Markus.  “Marcion and the Dating of the Synoptic Gospels.”  Studia Patristica Supplements 2.  Leuven: Peeters.  2014.  A book by Markus Vinzent arguing for an early re-dating of the Gospels, to the years between 138 and 144 AD.  The article cites it for the latest date its research found, and notes that the book rests on the old idea that Luke is an enlargement of Marcion's Gospel.

    1. 6a.

      Section Publisher's summary:  The summary says the book.  “argues for a re-dating of the Gospels to the years between 138 and 144 AD.” ↩

  7. 7.

    Roth, Dieter T.  “Marcion's Gospel and Luke: The History of Research in Current Debate.”  Journal of Biblical Literature 127(3):513–527.  2008.  A history of the mid-nineteenth-century German debate over whether Luke was an enlarged version of Marcion's Gospel.  Roth argues that the consensus later writers claimed for that view never existed.  The article uses it to show that the idea Vinzent's book rests on has never been a consensus position, and Roth wrote six years before Vinzent.

    1. 7a.

      Page 527:  "This position" is the view that Luke is an enlarged version of Marcion's Gospel, as debated in Germany in the 1840s and 1850s.  “Whether this position can become broadly persuasive in the twenty-first century remains to be seen;  however, it appears that in previous eras, far from being a consensus position, it convinced only a few.” ↩

  8. 8.

    Nongbri, Brent.  “The Use and Abuse of P52: Papyrological Pitfalls in the Dating of the Fourth Gospel.”  Harvard Theological Review 98(1):23–48.  2005.  A journal paper by Brent Nongbri of Yale University, who argues that scholars use the John papyrus P52 in inappropriate ways.  He finds that the handwriting date for P52 is a window that reaches into the early third century.  The article uses it because the AD 125 date that popular apologetics leads with is weaker than claimed, so Justin, not the papyrus, sets the ceiling for John.

    1. 8a.

      Page 46:  “What I have done is to show that any serious consideration of the window of possible dates for P52 must include dates in the later second and early third centuries.”  Pages 30–31:  At note 22, Nongbri traces the.  “AD 125”.  figure through a chain of restatements, C. H. Roberts's.  “first half of the second century”.  hardening into.  “about 125”.  and then into a later limit.  Page 31:  On this page he calls the resulting consensus.  “highly dubious”.  Section Roberts 1935, page 13:  Roberts himself had written, in his 1935 edition (An Unpublished Fragment of the Fourth Gospel in the John Rylands Library, Manchester: Manchester University Press):  “Any exact dating of book hands is, of course, out of the question”. ↩ ↩

  9. 9.

    Every distinctive word of Luke 19:43–44 and 21:20–24 traced through the complete CATSS morphological Septuagint (library key catss-lxxmorph-full, 64 files, about 685,000 tagged words).  Working notes:  research/luke-siege-vocabulary.md.  The Isaiah 29:3 quotation is read from that text, not from a translation:  καὶ κυκλώσω ὡς Δαυιδ ἐπὶ σὲ καὶ βαλῶ περὶ σὲ χάρακα καὶ θήσω περὶ σὲ πύργους.The counts in the table are of the dictionary form in the tagged text, recounted on 1 October 2026.  The 64 files carry Judges, Daniel, Tobit, Susanna and Bel in two recensions each, so a verse in those books can count twice:  the 23 for ἐρήμωσις includes six in the Old Greek of Daniel and five in Theodotion's.  An earlier draft printed 13, 58 and 124 for ἐρήμωσις, παρεμβάλλω and πατέω.  The first two came from matching the opening letters of the spelling, which misses ἐρημώσεως and the past tense παρενέβαλεν.  The third lumped every compound of πατέω together:  περιπατέω "walk about" 41, καταπατέω "trample down" 46, συμπατέω 10, ἐμπεριπατέω 8, and the simple verb 19.  The helper bin/lxx-lookup.py still does this when a compound is not named in the same query.  Mark's phrase is closest at Daniel 12:11 in the Old Greek, τὸ βδέλυγμα τῆς ἐρημώσεως;  Daniel 9:27 has βδέλυγμα τῶν ἐρημώσεων in both recensions, and 1 Maccabees 1:54 has βδέλυγμα ἐρημώσεως.  Isaiah 37:33 also joins Luke's two words, οὐδὲ μὴ κυκλώσῃ ἐπ᾽ αὐτὴν χάρακα. ↩ ↩ ↩

  10. 10.

    Torrey, Charles Cutler.  “The Composition and Date of Acts.”  Harvard Theological Studies I.  Cambridge, Mass.: Harvard University Press.  1916.  Page 70:  “It appears, then, that every particle of Luke's prediction not provided by Mark was furnished by familiar and oft-quoted Old Testament passages.  It is therefore obviously not permissible to call Luke 21, 20–24 a vaticinium ex eventu, and it cannot be cited as throwing light on the date of the Gospel.”  C. H. Dodd is usually credited with this and should not be, at least not for the dating conclusion.  Dodd's 1947 Journal of Roman Studies article opens by granting the opposite: "That the Third Gospel was in fact produced after the Fall of Jerusalem is on other grounds fairly certain" (JRS 37, page 47, the article's first page; reprinted in More New Testament Studies, 1968, page 69).  His thesis is that Luke used a pre-70 oracle, not that Luke is pre-70.  The article therefore credits Torrey with the dating verdict and Dodd with the vocabulary.  Bo Reicke argued the same in 1972: "Luke's apocalyptic scenery had nothing to do with the attack of Titus and the legions" ("Synoptic Prophecies on the Destruction of Jerusalem," in D. E. Aune, ed., Studies in New Testament and Early Christian Literature, Leiden: Brill, 1972, page 128).  A scholarly book on when Acts was written, the first volume of the Harvard Theological Studies series.  The article uses it for Torrey's 1916 verdict that Luke 21:20–24 cannot be cited to date Luke's Gospel. ↩

  11. 11.

    Counted across all seven books of Josephus, Bellum Judaicum, in Niese's Greek text (library key niese-1895), recounted on 1 October 2026.  τεῖχος and its derivatives 381, of which 299 are the noun itself;  χῶμα, earthwork, 64;  words in περιτειχ-, to wall around, 8;  χαράκωμα, palisade, 6, at BJ 1.253, 1.381, 1.383, 3.163, 3.169 and 5.269;  χάραξ 0.An earlier draft printed 414, 66 and "χάραξ 7".  Those were counts of letter strings.  The string τειχ also occurs in 33 past-tense forms of the verbs κατέχω, ἀντέχω and μετέχω, such as κατεῖχεν, "he held";  the string χωμα in two forms of ἀγχώμαλος, "nearly equal";  and the seven hits for χαρακ are the six of χαράκωμα and one χαρακτῆρας, "features," at 2.106.  The one palisade at Titus's siege is 5.269, γέρρα τῶν χαρακωμάτων ὑπερτεταμένα, "hurdles stretched over the palisades," sheltering the men who had been ordered at 5.262 to raise earthworks, ἐγείρειν χώματα.  Josephus' phrase for the later decision is περιτειχίζειν ὅλην τὴν πόλιν, "to wall around the whole city," at 5.499.  The wall, "one short of forty stadia" with thirteen forts, is at 5.508, inside the account of its building at 5.502–511. ↩ ↩

  12. 12.

    Josephus.  Bellum Judaicum.  n.d.  Section 6.193–213:  The famine, including the cannibalism of Mary of Bethezuba at 6.201.  Section 6.249–266:  The burning of the sanctuary.  Section 6.316:  At BJ 6.316 Josephus also records the legions carrying their standards into the temple court and sacrificing to them, while the sanctuary and everything round it were burning, when flight to the mountains would have been months too late to be useful advice.  Section Tacitus, Histories 5.12:  Tacitus, in Latin, which means three leaders and as many armies, with battles, treachery and arson among them, and a great store of grain burned:  “Tres duces, totidem exercitus … proelia dolus incendia inter ipsos, et magna vis frumenti ambusta”.  Tacitus does not give all three: the Histories break off in book 5 before the temple burns, and at 5.11 the famine is still only something the Romans decline to wait for.  Section Cassius Dio, epitome of book 65, 5.4:  In Earnest Cary's Loeb translation, volume 8, pages 267 and 269, the defenders.  “short of food”.  Section Cassius Dio, epitome of book 65, 6.3:  In the same translation, volume 8, pages 267 and 269,  “they were not conquered until a part of the temple was set on fire”.  Dio does not mention the factions.  Josephus's account of the war that ended with the fall of Jerusalem.  The article uses it for the three things the ancient accounts of Jerusalem's fall dwell on, the fire, the famine and the civil war among the defenders, all of which Josephus gives. ↩

  13. 13.

    Counted for this article by dictionary form in the MorphGNT tagging of the SBL Greek New Testament, all 27 books.  Totals for the whole New Testament:  παρουσία 24, σάλπιγξ 11, ἀπάντησις 3, ὠδίν 4, θροέομαι 3, ἐπισυναγωγή 2, ἐπισυνάγω 8, ἀνομία 15, αἰφνίδιος 2, ἐφίστημι 21, ἐκφεύγω 8.  The last column of the table is each total less the verses in the two middle columns.  The thief row lists only the places where a thief stands for the Lord's coming;  the noun κλέπτης occurs 16 times in all.  The three Lukan words of 1 Thessalonians 5:3 are αἰφνίδιος and ἐφίσταται, matching ἐπιστῇ and αἰφνίδιος at Luke 21:34, and ἐκφύγωσιν, matching ἐκφυγεῖν at Luke 21:36. ↩ ↩

  14. 14.

    “Acts 11:28.”  SBLGNT.  n.d.  The Greek of Acts 11:28, where the fulfilment notice that a famine foretold by Agabus happened in the time of Claudius is a relative clause outside the narrative voice.  The article uses it for Luke's habit of stepping outside his story, which it sets against his silence about the fall of Jerusalem.  A fuller test would need every authorial aside in Luke–Acts catalogued, which has not been done, so the point is a refinement of the argument from silence, not an independent argument. ↩

  15. 15.

    Josephus Flavius.  “Antiquitates Judaicae.”  Perseus Digital Library.  Edited by B. Niese.  c. 94 (edition of 1892).  Josephus's history of the Jews, cited here for two passages.  The article uses it for his account of the death of James the Lord's brother in AD 62, which Acts does not record, and for his list of Theudas and Judas the Galilean, which resembles the list in Acts 5:36–37.

    1. 15a.

      Section 20.200:  The high priest Ananus, in the gap between the death of Festus and the arrival of Albinus, brought before a council a man described as.  “τὸν ἀδελφὸν Ἰησοῦ τοῦ λεγομένου Χριστοῦ, Ἰάκωβος ὄνομα αὐτῷ”.  which in English is.  “the brother of Jesus who was called Christ, James by name”.  and handed him over to be stoned. ↩

    2. 15b.

      Section 20.97–98:  On Theudas under the procurator Fadus, and.  Section 20.102:  on the crucifixion of the sons of Judas the Galilean under Tiberius Alexander, where Judas's own revolt in the time of Quirinius is recalled.  Section 20.267:  The date of publication is Josephus' own, given at this place as.  “τρισκαιδεκάτου μὲν ἔτους τῆς Δομετιανοῦ Καίσαρος ἀρχῆς”.  the thirteenth year of Domitian, which is AD 93/94. ↩

    3. 15c.

      Section 20.199–200:  The younger Ananus, a Sadducee, chose a moment when Festus had died and Albinus was still on the road, convened a council of judges, and handed James and some others over to be stoned.  “Ἀλβῖνον δʼ ἔτι κατὰ τὴν ὁδὸν ὑπάρχειν, καθίζει συνέδριον κριτῶν”.  Section 20.203:  Albinus wrote to Ananus in anger, and King Agrippa took the high priesthood from him after he had held it three months.  “γράφει μετʼ ὀργῆς τῷ Ἀνάνῳ”.  “ἄρξαντα μῆνας τρεῖς”. ↩

    4. 15d.

      Section 20.97:  Theudas, in Greek, which means he persuaded the greater part of the crowd.  “πείθει τὸν πλεῖστον ὄχλον”.  Section 20.171:  The Egyptian's followers under Felix, in Greek, which means he killed four hundred of them and took two hundred alive.  The Antiquities gives no total for the Egyptian's following.  “τετρακοσίους μὲν αὐτῶν ἀνεῖλεν, διακοσίους δὲ ζῶντας ἔλαβεν”. ↩

  16. 16.

    ἀκωλύτως occurs once in the New Testament (Acts 28:31) and zero times in the Septuagint, both counted from the tagged corpora on disk.  Of the sixteen Gospel-and-Acts occurrences of κωλύω, twelve are in Luke–Acts:  Matthew 1, Mark 3, Luke 6, John 0, Acts 6.The related adjective ἀκώλυτος does occur once in the Septuagint, at Wisdom 7:22 (verse 23 in the CATSS numbering), of the spirit of Wisdom.  The claim is about the adverb. ↩ ↩

  17. 17.

    Harnack, Adolf.  “The Date of the Acts and of the Synoptic Gospels.”  New Testament Studies IV.  London: Williams & Norgate.  Translated by J. R. Wilkinson.  1911.  Harnack's 1911 study of the dates of Acts and the Synoptic Gospels, in English translation of his German work Neue Untersuchungen zur Apostelgeschichte, Leipzig 1911.  The article uses it for two of his conclusions: that Acts was written while Paul's trial in Rome had not yet ended, and that Matthew in its present shape was written in the years after the destruction of Jerusalem.

    1. 17a.

      Page 95:  “Throughout eight whole chapters St Luke keeps his readers intensely interested in the progress of the trial of St Paul, simply that he may in the end completely disappoint them—they learn nothing of the final result of the trial!”  Section Robinson 1976, page 90:  Harnack had earlier dated Luke–Acts.  “between 78 and 93.”  That figure is John A. T. Robinson's report of Harnack's 1897 Chronologie (Redating the New Testament, London: SCM, 1976, page 90). ↩

    2. 17b.

      Page 99:  He never states a year for Acts.  “the concluding verses of the Acts of the Apostles, taken in conjunction with the absence of any reference in the book to the result of the trial of St Paul and to his martyrdom, make it in the highest degree probable that the work was written at a time when St Paul's trial in Rome had not yet come to an end.”  Page 93:  And at this page he refuses the apologetic inference outright:  “the decayed beams of a building are not made stronger and better by the proof that they are older than was at first thought!”  Section Robinson 1976, page 91 note 26:  John A. T. Robinson, arguing on Harnack's side, says so bluntly:  “Despite the English title of his second book, he never actually dates Acts”. ↩

    3. 17c.

      Page 134:  “The book must be placed in close proximity with the destruction of Jerusalem.  In its present shape, however, it should be assigned to the years immediately succeeding that catastrophe.  Here St Matt. xxii. 7 (a verse wanting in St Luke) is of special weight.  And yet composition before the catastrophe cannot be excluded with absolute certainty.”  Harnack is quoting his own Chronologie of 1897 and says, on page 133, that he has.  Page 133:  “nothing new to add”.  to it.  He put Mark.  “during the sixth decade of the first century at the latest”.  and called that date.  “certain”.  (page 133).  On Matthew he did not move.  Anyone citing Harnack for a pre-70 Matthew is citing him against himself. ↩ ↩

  18. 18.

    Hemer, Colin J.  “The Book of Acts in the Setting of Hellenistic History.”  Wissenschaftliche Untersuchungen zum Neuen Testament 49.  Tübingen: Mohr Siebeck.  Edited by Conrad H. Gempf.  1989.  A scholarly study of Acts set against the history of the Hellenistic world.  The article uses it because Colin Hemer is the most thorough modern defender of an early date for Acts, and it quotes him on the ending and on the Josephus parallels.

    1. 18a.

      Page 365:  “The argument for early date drawn from the abrupt ending of Acts is an inadequate basis, though it offers a neat reconstruction and may be a valuable confirmatory consideration for one who is convinced of the early date upon a different and broader base.”  Page 387:  “It is thus difficult to explain the ending exactly, regardless of the view one takes about the early or late date of the book.”  Hemer's own date for Acts is AD 62.  He places the writing inside Paul's two years: "Luke-Acts, essentially as a unity, was in process of composition during the διετία, which provides a natural slot for it" (page 405).  His summary on page 414 does count the ending:  the fifteen features "and the 'immediacy' of the later chapters in Acts taken together with the peculiar ending to the book make such a date highly probable."  The fifteen are listed at pages 376–382.  On this article's reading only three of them, numbers 4, 11 and 13, bear on the date of composition rather than the date of the setting.  Number 13 depends on 2 Timothy, as Hemer says at page 381.  Number 11, period detail, is the kind of evidence Robinson had already allowed may come from a writer "using the language of his sources or consciously archaizing" (Redating the New Testament, page 91). ↩ ↩

    2. 18b.

      Pages 372–373:  “If either used the other, he misused him.  They are surely independent, and follow independent, indeed conflicting, sources.” ↩

    3. 18c.

      Page 377:  The third of his general indications of date.  “The chosen appeal to Roman justice is harder to place when the character of the later Nero was known, or after the conflicts which followed.”  The fifth begins:  “There is no hint of the death of James at the hands of the Sanhedrin in c. 62”.  Page 378:  “though it has been argued that the mention of this case would suit so perfectly Luke's apologetic aim to a Roman audience as to merit inclusion.”  “This point, which depends for its force on a prejudgment of Luke's purpose and setting, should not be pressed.” ↩ ↩

    4. 18d.

      Page 389:  On the voyage narrative of Acts 27–28.  “It is our contention that these inconsequential details are hard to explain except as vivid experiences recalled at no great distance in time.” ↩

  19. 19.

    Burrows's Delta over the Greek Gospels and Acts, computed with bin/nt-stylometry.py cluster and rerun on 1 October 2026:  162 blocks of 500 words, the 100 commonest dictionary forms.  The mean distance from Acts to Luke is 0.947, against 1.032 to Matthew, 1.036 to Mark and 1.120 to John.  That closeness cannot be explained by copying, because Acts shares no text with Luke.The limit of the test should be stated.  Acts and Luke are each closer to themselves, 0.863 and 0.875, than to each other, and Luke is slightly closer to Mark (0.933) and Matthew (0.941), whose text it shares, than to Acts.  The method cannot separate the three Synoptic Gospels.  Full result and that failed control:  Who Wrote the Gospels? and ../gospel-authorship/research/stylometry-luke-acts-and-we.md. ↩

  20. 20.

    Carson, D. A., and Douglas J. Moo.  An Introduction to the New Testament.  Grand Rapids: Zondervan.  2nd edition.  2005.  The standard conservative introduction to the New Testament.  The article quotes it because a textbook describing its own method is better evidence for the shape of the field than any outside characterisation of it.

    1. 20a.

      Page 175:  On Mark's author:  “Our decision, then, will rest almost entirely on external evidence.”  Page 205:  “Yet the main reason for singling out Luke as the author of Luke-Acts comes from the external evidence.”  Page 207:  “The date of Luke's gospel is closely intertwined with the dates of Mark and Acts.”  They add that the ending of Acts.  “may point to the date at which Luke published Acts (about A.D. 62).”  Page 155:  The first of five arguments for a Matthew before 70:  “The question of date is marginally bound up with the question of authorship.” ↩

  21. 21.

    Goodacre, Mark.  The Synoptic Problem: A Way Through the Maze.  Sheffield: Sheffield Academic Press.  2001.  Page 68:  In the chapter on Markan priority.  “If it were clear, for example, that the best evidence placed Mark's Gospel earlier than Matthew's or Luke's, we would have a useful additional reason for thinking that his Gospel was the first to be written.”  Compare the Carson and Moo sentence on the date of Luke, quoted in the article.  Circularity would require the disputed dating premise to be the only support in both directions.  The dependency chain is set out in full in The Synoptic Problem.  A guide to the Synoptic Problem, the puzzle of why Matthew, Mark and Luke share so much of their wording, written for students and general readers.  Goodacre argues that Mark was written first and that there was no lost sayings source called Q.  The article uses it for his use of the AD 70 evidence to support Markan priority, the view that Mark came first, when elsewhere relative dating is derived from Markan priority.  The article asks whether the two arguments have independent support; their interaction alone does not establish a circle. ↩

  22. 22.

    Crossley, James G.  “The Date of Mark's Gospel: Insight from the Law in Earliest Christianity.”  JSNTSup 266.  London: T&T Clark International.  2004.  A book arguing for a date for Mark's Gospel between the mid to late thirties and the mid-forties, on the premise that Mark assumes a law-observant readership.  Crossley is the non-believer in How Did Christianity Begin? A Believer and Non-Believer Examine the Evidence, written with Michael Bird in 2008.  The article uses it as the earliest date anyone defends, a case that has few followers.

    1. 22a.

      Page 208:  “A date for Mark between the mid to late thirties and mid-forties has important ramifications not only for our understanding of Mark's gospel but for Christianity in general before the Pauline letters.”  Page 172:  The argument in one clause:  “Mark assumes what Matthew and Luke could not.”  A third review, Lois Fuller's in the Journal of Greco-Roman Christianity and Judaism 2, pages R1–R5, presses a point Crossley does not answer: a Gospel written for readers who need no explanation of the law is a strange Gospel to fill with explanations of Aramaic words and Jewish customs.  No commentary or introduction found in this research has adopted Crossley's date. ↩ ↩

  23. 23.

    Lake, Kirsopp.  The Apostolic Fathers, vol. 1 (Loeb Classical Library 24).  London:  Heinemann;  New York:  Putnam.  1919.  The Greek text with a facing English translation of 1 and 2 Clement, the letters of Ignatius, Polycarp's letter and the Didache, each with a short introduction on its date.  The article uses Ignatius's letter to the Smyrnaeans, which echoes a phrase found only in Matthew, and Lake's report of the usual date of the letters.

    1. 23a.

      Page 252:  Ignatius, To the Smyrnaeans 1.1, in Greek.  “βεβαπτισμένον ὑπὸ Ἰωάννου, ἵνα πληρωθῇ πᾶσα δικαιοσύνη ὑπ᾽ αὐτοῦ”.  Page 253:  Lake's English, which prints the last six words inside quotation marks as a quotation of Matthew 3:15.  “baptised by John that”.  “all righteousness might be fulfilled by him”. ↩

    2. 23b.

      Page 166:  On Eusebius's date for the martyrdom, the tenth year of Trajan, which Lake gives as AD 108.  “Modern critics are by no means unanimous as to the correctness of this date”.  “that Ignatius was really a martyr in Rome in the time of Trajan”. ↩

  24. 24.

    Barnes, Timothy D.  “The Date of Ignatius.”  The Expository Times 120(3):119–130.  Dec 2008.  A journal article arguing that Ignatius knew the writings of the Gnostic teacher Ptolemaeus, so that Eusebius's date under Trajan cannot stand.  The article uses it as the main case for a later Ignatius, which is why his letters stay out of the ceilings table.

    1. 24a.

      Section abstract:  “it is suggested that the letters were probably written in the 140s.” ↩

  25. 25.

    Clement.  1 Clement.  n.d.  Section 13.2:  Its sayings here match no surviving Gospel.  Section 46.8:  It twice reads.  “ἕνα τῶν ἐκλεκτῶν μου”.  where Matthew 18:6 and Mark 9:42 read ἕνα τῶν μικρῶν τούτων and Luke 17:2 has the same words in another order.  Section 47.1:  Against this, at 47.1 it says:  “Ἀναλάβετε τὴν ἐπιστολὴν τοῦ μακαρίου Παύλου τοῦ ἀποστόλου”.  “Take up the epistle of the blessed Paul the apostle.”  Clement can name a written apostolic document when he has one.  He never does so for a Gospel.  1 Clement must be struck from lists of early Gospel attestation.  Section Polycarp, Philippians 2.3:  Polycarp has one of 1 Clement's clauses word for word,  “ἐλεᾶτε, ἵνα ἐλεηθῆτε”.  “show mercy, that you may be shown mercy”.  which is in no Gospel, and has it beside a clause that is Matthew 7:1 exactly and another that uses Luke 6:38's verb.  So the two writers share a form of the saying that is not a Gospel's, though Polycarp also knows Gospel wording.  A letter from Clement, which the article checks for quotations of the Gospels.  The article uses it to show that 1 Clement does not quote a written Gospel, so it cannot count as early evidence for the Gospels. ↩

  26. 26.

    Ignatius.  Smyrnaeans.  n.d.  Section 3.2:  “Λάβετε, ψηλαφήσατέ με καὶ ἴδετε, ὅτι οὐκ εἰμὶ δαιμόνιον ἀσώματον”.  “Take, handle me and see that I am not a bodiless demon.”  Section HE 3.36.11:  Eusebius could not identify its source, and says:  “οὐκ οἶδ᾽ ὁπόθεν ῥητοῖς συγκέχρηται”.  “he has used words from I know not where”.  In Lake's Loeb edition, volume 1, page 282.  Section Jerome, De viris illustribus 16:  Jerome says that Ignatius took it, in Latin:  “de Evangelio quod nuper a me translatum est”.  from the Gospel he had recently translated.  Jerome wrongly places the line in the letter to Polycarp.  Chapter 2:  At chapter 2 of the same work he names that Gospel as the one.  “quod appellatur secundum Hebraeos”.  A letter by Ignatius with a line that sounds like a Gospel saying.  The article uses it because Eusebius could not say where the line came from, and Jerome assigned it to a Gospel he had recently translated. ↩

  27. 27.

    Jerome.  Chronicle.  Edited by Roger Pearse et al.  c. 380 (edition of 2005).  Jerome's Latin version and continuation of Eusebius's Chronici Canones.  The article uses it because it is the only ancient text that puts a year on a Gospel, and the Matthew line in it is a later addition.

    1. 27a.

      Section Entry at the third year of Trajan, AD 100, in the 219th Olympiad:  “Bishop Irenaeus writes that John the Apostle survived all the way to the time of Trajan:  after whom his notable disciples were Papias, Bishop of Hieropolis, Polycarp of Smyrna, and Ignatius of Antioch.”  The translation prints the apostle's name in capitals.  The later date of about 130 for Papias rests on Philip of Side alone. ↩

    2. 27b.

      Section the years AD 39 to 44:  “Peter the Apostle, by nation a Galilean, first high priest of the Christians, after he had been the first to found a church at Antioch, proceeded to Rome, where as bishop of the same city he remains, preaching the gospel for 25 years.”  “Mark the evangelist, Peter's interpreter, preaches Christ in Egypt and Alexandria.”  “In the Merton manuscript there is a space following.  In the Patrologia Latina, there follows this entry:  Matthew in Judaea is the first to write a gospel.”  The third quotation is the translators' footnote.  The Peter entry stands at the second year of Claudius and the Mark entry at the third. ↩

  28. 28.

    Justin Martyr.  “Dialogue with Trypho.”  Justin: Dialogue avec Tryphon 2.  Paris: Picard.  Edited by Georges Archambault.  n.d. (edition of 1909).  Section Dialogue with Trypho 106.3:  “γεγράφθαι ἐν τοῖς ἀπομνημονεύμασιν αὐτοῦ … υἱοὺς Ζεβεδαίου ὄντας, μετωνομακέναι ὀνόματι τοῦ Βοανεργές, ὅ ἐστιν υἱοὶ βροντῆς.”  Page 152:  On page 152 of volume 2, Archambault's note says the pronoun.  “ne peut se rapporter qu'au Christ ou à Pierre”.  can refer only to Christ or to Peter, and he chooses Peter.  He grants that readers.  “thought at once of the second Gospel”.  whose traditional author was Peter's disciple, and concludes from the newly found fragments of a Gospel of Peter that.  “il semble bien qu'il faille voir dans le présent passage la mention de cet Évangile de Pierre.”  The date of about 160 for the Dialogue is the usual one and is not established here.  A work by Justin Martyr.  The article uses it for the one earlier trace of Mark's Gospel that it finds, which is disputed: Justin says that a detail found only at Mark 3:17 is written in his memoirs. ↩

  29. 29.

    Marcus, Joel.  “The Jewish War and the Sitz im Leben of Mark.”  Journal of Biblical Literature 111(3):441-462.  1992.  A study connecting Mark's temple language and social conflicts with the Jewish War.

    1. 29a.

      Page 448.  Page 454, note 59:  Marcus proposes Eleazar's occupation in 67-68 for Mark's setting while allowing a Caligula-era reference in an earlier source. ↩

    2. 29b.

      Page 460, note 90:  “Contemporary critics are about evenly split”.  His report concerns scholarship in 1992; he prefers shortly after 70. ↩

  30. 30.

    Davies, W. D., and Dale C. Allison.  “A Critical and Exegetical Commentary on the Gospel according to Saint Matthew.”  International Critical Commentary 1.  Edinburgh: T&T Clark.  1988.  A scholarly commentary on Matthew's Gospel, volume 1 of the International Critical Commentary series.  The article uses it for the authors' weighing of the argument that Matthew 22:7 was written after the events of AD 70, and for their dating of Matthew to between A.D. 70 and A.D. 100.

    1. 30a.

      Page 131:  After listing the objections to reading Matthew 22:7 as written after the event:  “In our judgement, all of these questions can be readily answered;  that is, none of the objections to the standard interpretation is decisive.”  Page 132:  “In other words, even if Mt 22.7 comes from the OT or reproduces a common prophetic topos, we must inquire into the reason for its placement in the parable of the great supper, and for this the events of 70 lie near to hand.”  “In conclusion, when all has been said, Mt 22.7, while it does not demand a date after A.D. 70, does strongly imply one.”  The argument that a later writer would reach for the Bible's own formula is the article's own.  The quotations are Davies and Allison's reply to it. ↩

    2. 30b.

      Page 131:  Among the objections they list to reading Matthew 22:7 as written after the event.  “if Matthew were writing after the horrors of the Jewish wars, why does Mt 24 fail to recite the characteristic features of that war—such as the burning of the temple?”  Page 132:  Their reply.  “(i) We must keep in mind Matthew's conservative nature.”  “So if he did not find in his sources detailed depictions of Jerusalem under siege (cf. Mark 13), we should not really expect to meet with such in his gospel.” ↩

    3. 30c.

      Page 138:  “Matthew was almost certainly written between A.D. 70 and A.D. 100, in all probability between A.D. 80 and 95.” ↩

  31. 31.

    Isaiah 37:33 in the Septuagint reads οὐδὲ μὴ κυκλώσῃ ἐπ᾽ αὐτὴν χάρακα, where the Hebrew has וְלֹא־יִשְׁפֹּךְ עָלֶיהָ סֹלְלָה, "nor cast up a siege mound against it."  Ezekiel 4:2 has περιβαλεῖς ἐπ᾽ αὐτὴν χάρακα for וְשָׁפַכְתָּ עָלֶיהָ סֹלְלָה.Greek from the CATSS text, Hebrew from the Westminster Leningrad Codex as tagged by the Open Scriptures Hebrew Bible project (library key openscriptures-2026b), both read on 1 October 2026. ↩

  32. 32.

    The sequence λίθος … ἐπί … λίθος was searched across all 64 CATSS Septuagint files.  It occurs once:  Haggai 2:15, πρὸ τοῦ θεῖναι λίθον ἐπὶ λίθον ἐν τῷ ναῷ κυρίου — of building the second temple. ↩

  33. 33.

    Josephus.  Bellum Judaicum.  n.d.  Section 7.1–4:  At 7.1 Titus orders the whole city and the sanctuary razed, leaving the towers Phasael, Hippicus and Mariamme and.  “τεῖχος δ᾽ ὅσον ἦν ἐξ ἑσπέρας τὴν πόλιν περιέχον”.  which in English reads.  “as much of the wall as enclosed the city on the west.”  At 7.2 the wall is kept as a camp for the garrison and the towers as a monument.  At 7.3 the rest was so levelled.  “ὡς μηδεπώποτ᾽ οἰκηθῆναι πίστιν ἂν ἔτι παρασχεῖν τοῖς προσελθοῦσι”.  so that visitors would have no ground for believing it had ever been lived in.  Josephus's account of the war that ended with the fall of Jerusalem.  The article uses it because Josephus names the exceptions to the destruction of the city and the sanctuary: three towers, and the stretch of the city wall on the west side, which was left as a camp for the garrison. ↩

  34. 34.

    Wanamaker, Charles A.  “The Epistles to the Thessalonians: A Commentary on the Greek Text.”  New International Greek Testament Commentary.  Eerdmans.  1990.  Wanamaker himself argues that Paul wrote both letters.

    1. 34a.

      Page 17:  “no contemporary scholars of repute seem to doubt the authentic Pauline character of the letter”. ↩

    2. 34b.

      Page 17:  “in the last two decades the balance of expressed opinion has tilted decisively against the authenticity of 2 Thessalonians”. ↩

  35. 35.

    Wenham, David.  “Paul and the Synoptic Apocalypse.”  Gospel Perspectives, volume 2: Studies of History and Tradition in the Four Gospels, 345–375.  JSOT Press.  Edited by R. T. France and David Wenham.  1981.  The fullest study of the parallels between Paul's letters and the Olivet Discourse.  Wenham finds a common tradition older than Paul and does not conclude that Paul used a written Gospel.

    1. 35a.

      Page 345:  “Parallels therefore that can reasonably be explained as due to two authors' independent acquaintance with pre-Christian Jewish apocalyptic tradition must be discounted.” ↩

    2. 35b.

      Page 365:  “We may indeed say that our conclusions, if valid, enlarge the Synoptic problem in a significant way, since Paul is seen to be our earliest witness to the Synoptic Apocalypse and his evidence must be brought in when the different Synoptic accounts are compared.” ↩

    3. 35c.

      Page 354:  “The differences point to Lukan originality”.  This is the heading of the section, pages 354 to 358, in which Wenham compares 1 Thessalonians 5:3 with Luke 21:34–36. ↩

    4. 35d.

      Page 364:  “Even if the common Pauline/Synoptic traditions are pre-Pauline, this is not the same as saying that they were thought to derive from Jesus.” ↩

  36. 36.

    Orchard, J. Bernard.  “Thessalonians and the Synoptic Gospels.”  Biblica 19:19–42.  1938.  The article that first argued Paul depended on Matthew 23 to 25.  Orchard was a Benedictine monk of Ealing Abbey and later a leading defender of the view that Matthew was the first Gospel written.

    1. 36a.

      Page 36:  “All the Matthean parallels are grouped together in the eschatological discourse and the verses immediately preceding it”.  The fourth of the four points Orchard draws from his table of parallels.  He gives the block as Matthew 23:31 to 25:46. ↩

    2. 36b.

      Page 39:  “it naturally follows that this passage is something absolutely primordial and must be dated somewhere between 40 and 50 A.D.” ↩

    3. 36c.

      Page 29:  “it seems to remain that this expression is original to St Paul and that St Luke has derived it either directly or indirectly from him”. ↩

    4. 36d.

      Page 37:  “it is necessary to admit that there exists between the two authors some kind of literary dependence - either indebtedness of both to a common source, or else the direct indebtedness of one to the other”. ↩

  37. 37.

    Orchard, Bernard, and Harold Riley.  The Order of the Synoptics: Why Three Synoptic Gospels?  Mercer University Press.  1987.  A defence of the order Matthew, Luke, Mark.  The argument from Thessalonians is restated on pages 118 to 120.

    1. 37a.

      Page 245:  “that is to say, before 46/47; so that its formation could have taken place in the first decade of the church's life”. ↩

  38. 38.

    Robinson, John A. T.  Redating the New Testament.  SCM Press.  1976.  Robinson was a liberal Anglican bishop and New Testament scholar who argued that every book of the New Testament was written before AD 70.  His final date for Matthew is before 62.

    1. 38a.

      Page 106, note 77:  “But this goes with his belief in the priority of Matthew in its present form and seems to me to be pushing the evidence much too far.” ↩

    2. 38b.

      Page 106:  “But the evidence of Thessalonians at any rate shows that this way of thinking was rife in the year 50.  The marks of it in Matthew 24 cannot therefore be used to require any later date for that gospel.” ↩

    3. 38c.

      Page 104:  “Since this was a group whose power disappeared with the destruction of the temple, preoccupation with them argues strongly for an earlier period.” ↩

    4. 38d.

      Page 110, note 99:  “The genitive absolute is not temporal; it does not imply that the gospel of St Matthew was written at that time; it simply contrasts the ministry of the two great Apostles with that of St Matthew”.  Robinson is quoting Harnack, who was following Chapman.  Of Irenaeus Robinson adds in his own voice that his chronology merely reflects the canonical order. ↩

    5. 38e.

      Page 113, note 121:  “that the apostles were to stay in Jerusalem for twelve years after the crucifixion”.  Page 113:  “There is ground therefore for thinking that Edmundson may be right in dating the death of James and the imprisonment of Peter in the spring of 42”.  Robinson's own estimate for Peter's first preaching in Rome, later in the chapter, is about AD 45. ↩ ↩

  39. 39.

    McNicol, Allan J.  “Jesus' Directions for the Future: A Source and Redaction-History Study of the Use of the Eschatological Traditions in Paul and in the Synoptic Accounts of Jesus' Last Eschatological Discourse.”  New Gospel Studies 9.  Mercer University Press.  1996.  McNicol belongs to the school that puts Matthew first and has written on this discourse jointly with Orchard.

    1. 39a.

      Page 20:  “However, sufficient literary evidence constituting a number of parallels does not exist to cause this investigator to hypothecate that there is copying between Paul and the Synoptics”. ↩

  40. 40.

    Josephus.  Bellum Judaicum.  n.d.  Section 6.300–309:  On Jesus son of Ananias, who cried woe against Jerusalem from about AD 62 until he was killed during the siege.  Josephus's account of the war that ended with the fall of Jerusalem.  The article uses it to show that predicting the city's fall was not unique to Jesus, because Josephus reports a man named Jesus son of Ananias who cried woe against Jerusalem. ↩

  41. 41.

    Tacitus.  “Annales, book 15.”  The Latin Library.  c. 116.  The Roman historian's account of the fire of Rome in 64 and of Nero's punishment of the Christians for it.  The article uses it to date the persecution that Acts, which ends with Paul awaiting Nero's court, never mentions.

    1. 41a.

      Book 15, chapter 44:  “ergo abolendo rumori Nero subdidit reos et quaesitissimis poenis adfecit, quos per flagitia invisos vulgus Chrestianos appellabat.”  In English, so to end the rumour Nero put forward as culprits, and punished with the most refined cruelties, people hated for their crimes, whom the crowd called Chrestians.  “multitudo ingens haud proinde in crimine incendii quam odio humani generis convicti sunt.”  In English, a huge multitude was convicted, less of arson than of hatred of the human race. ↩

  42. 42.

    Seccombe, David.  “Dating Luke-Acts: Further Arguments for an Early Date.”  Tyndale Bulletin 71(2):207–227.  2020.  A journal article that adds three arguments to the case for writing Luke and Acts in the early 60s:  the danger after 66 of Luke's language about Israel's national hope, a church whose Jewishness is taken for granted, and the space and detail given to the voyage to Rome.  The article uses the third argument, which restates Colin Hemer's, and Seccombe's reply to reading Acts 20:25 as foreshadowing Paul's death.

    1. 42a.

      Page 221, note 64:  “may be evidence that Luke did not yet know of Paul's release (if he returned to Ephesus)”.  Note 65:  “In Phil. 1:25 and Phlm. 22 Paul expects to be released.” ↩

    2. 42b.

      Page 224:  “Even F. C. Baur, who saw Acts as a mid-second-century creation, felt the author must have had in his hands a genuine diary of the voyage.”  Page 226:  “Verse for verse, it is the least theological.”  Page 227:  “the prominence given to the voyage to Rome and the vividness of detail in the account indicate an author at work prior to the extraordinary disruptions of the following decade.” ↩ ↩

  43. 43.

    Marguerat, Daniel.  “The First Christian Historian: Writing the Acts of the Apostles.”  SNTSMS 121.  Cambridge: Cambridge University Press.  Translated by K. McKinney, G. J. Laughery and R. Bauckham.  2002.  A scholarly study of how the Acts of the Apostles was written.  The article uses it for Marguerat's argument that Acts ends on purpose, in the open way that ancient works such as Homer and Herodotus sometimes end.

    1. 43a.

      Page 215:  “The attestation of narrative suspension in the ending, both in Homer and in that great master of historiography Herodotus, is impressive, when one realizes the considerable role played by these works in ancient culture.”  And, three sentences on:  “Everything leads one to think that the author of Acts made use of this model.”  Page 229:  His summary is that the device.  “allows the author of the Acts consciously to use silence and ambivalence in composing the end of his monumental work.”  Hemer had objected that the deliberate-ending view lacked a literary parallel "at the focus of the problem" (page 384); Marguerat's Herodotus is the answer offered. ↩

  44. 44.

    du Toit, David S., and John Painter.  “Reviews of Crossley, The Date of Mark's Gospel.”  Review of Biblical Literature.  Jan 2006.  Section du Toit review, last page:  “Crossley has written an interesting and provocative book.  He clearly shows that the consensus date around 70 C.E. is based on arguments of doubtful validity.”  The publisher prints this first sentence as a blurb.  Of Crossley's assumption that Torah observance lapsed everywhere at the same pace:  “This would imply that the evidence presented by Crossley is not conclusive and does not necessarily demand an early date for Mark's Gospel.”  Section Painter review, last page:  “Certainly Crossley has reminded us, as J. A. T. Robinson did before him, upon what flimsy evidence our views of the authorship and dating of the Gospels are based.  But flimsy evidence is flimsy for all comers.  It seems to me that the straws in the wind continue to point more in the direction of a date around 70 C.E. than elsewhere.”  Two reviews of Crossley's book in the same issue of the Review of Biblical Literature, one by David S. du Toit and one by John Painter.  The article uses them to show both halves: du Toit finds the consensus date around 70 C.E. based on arguments of doubtful validity, and Painter says the evidence still points more toward a date around 70 C.E. than elsewhere. ↩

  45. 45.

    Josephus, Flavius.  Bellum Judaicum.  Edited by Benedikt Niese.  1895.  Section 7.218.  Section Cassius Dio, epitome of book 65, 7.2:  Cassius Dio says the same, in Cary's Loeb translation, volume 8, page 271:  “should pay an annual tribute of two denarii to Jupiter Capitolinus”.  Josephus's history of the Jewish war with Rome, in the Greek text edited by Benedikt Niese.  Book 7, section 218 describes the tax that is called the fiscus Iudaicus, and the article quotes it for the yearly two-drachma tribute.  The article uses it as background to Matthew 17:24, the only place in the New Testament where the word for the two-drachma coin (δίδραχμον) occurs, and the stater of 17:27 is likewise unique to Matthew, the two coins that pay it. ↩

    1. 45a.

      Section 2.261:  The Egyptian false prophet, in Greek, which means he gathers about thirty thousand of those he had deceived.  At 2.263 most of them are killed or captured and the Egyptian escapes.  “περὶ τρισμυρίους μὲν ἀθροίζει τῶν ἠπατημένων”. ↩

  46. 46.

    Hagner, Donald A.  “Determining the Date of Matthew.”  Jesus, Matthew's Gospel and Early Christianity: Studies in Memory of Graham N. Stanton, 76–92.  T&T Clark.  Edited by Daniel M. Gurtner, Joel Willitts and Richard A. Burridge.  2011.  Hagner's fullest statement of his case for a Matthew written before AD 70.  He gives no year.

    1. 46a.

      Page 86:  “Some years ago I changed my mind about the date of Matthew.”  “I noticed Matthew's surprising addition of the word eutheos, 'immediately', at the beginning of 24.29.” ↩

    2. 46b.

      Page 91:  “The evidence briefly canvassed here cannot, of course, demonstrate an early date for Matthew.  But a cumulative argument of some strength emerges from it.” ↩

  47. 47.

    Hagner, Donald A.  “Matthew 1–13.”  Word Biblical Commentary 33A.  Word Books.  1993.  Hagner taught New Testament at Fuller Theological Seminary.  His section on the date of Matthew is pages lxxiii to lxxv of the introduction.

    1. 47a.

      Page lxxv:  “It needs to be re-emphasized that the dogmatism of critical orthodoxy concerning a post-70 date is unwarranted.  But this is no reason to give rein to a similar dogmatism concerning a pre-70 date.” ↩

    2. 47b.

      Page lxxiv:  “Indeed, the evangelist aggravates the problem considerably by his insertion of εὐθέως, “immediately,” in 24:29, which leaves the clear impression that he expected the parousia of the Son of Man to occur in close succession to the fall of Jerusalem.” ↩

  48. 48.

    Σαδδουκαῖος by dictionary form in the same tagged text:  Matthew 3:7, 16:1, 16:6, 16:11, 16:12, 22:23 and 22:34;  Mark 12:18;  Luke 20:27;  Acts five times.  At 3:7 and in the four verses of chapter 16 Matthew writes "Pharisees and Sadducees" as a pair, four of the five times in the form τῶν Φαρισαίων καὶ Σαδδουκαίων, with one article for both. ↩ ↩

  49. 49.

    France, R. T.  Matthew: Evangelist and Teacher.  Paternoster Press.  1989.  France was Principal of Wycliffe Hall, Oxford.  The date of Matthew is discussed on pages 82 to 91.

    1. 49a.

      Page 82:  “This would place it in the early sixties at latest”.  Page 83:  “Altogether, then, the patristic evidence seems unanimous that Matthew was written not later than the early sixties.” ↩

    2. 49b.

      Page 90:  “It may be readily conceded that none of these arguments for a pre-70 date for Matthew is conclusive.” ↩

  50. 50.

    Tyson, Joseph B.  Marcion and Luke-Acts: A Defining Struggle.  Columbia: University of South Carolina Press.  2006.  A case for dating Acts in the early second century, reading its portrayal of Peter and Paul against the Marcionite movement.

    1. 50a.

      Page 78:  “the challenge of Marcion and Marcionite Christianity forms a remarkably meaningful and probable context for Acts”.  Tyson concludes here that Acts was probably written about 120-125.  That date depends on his proposed context, not on Josephus alone. ↩

  51. 51.

    Towner, Philip H.  “The Letters to Timothy and Titus.”  New International Commentary on the New Testament.  Grand Rapids: Eerdmans.  2006.  A commentary that assesses the Pastorals as historical letters while distinguishing their composition from their authorship.

    1. 51a.

      Page 59:  “encompasses the Jesus material as Scripture is neither likely nor, from the standpoint of authority, necessary.”  On the assumption that the introductory formula applies to the Jesus saying as well as Deuteronomy; he also compares 1 Cor 9:9-14. ↩

  52. 52.

    “Greek text of 1 Timothy 5:18, Luke 10:7, Matthew 10:10, 1 Corinthians 9:9 and 9:14.”  SBLGNT.  n.d.  The Greek of these verses in the SBLGNT (Society of Biblical Literature Greek New Testament), where Luke 10:7 reads ἄξιος γὰρ ὁ ἐργάτης τοῦ μισθοῦ αὐτοῦ and 1 Timothy 5:18 has the same six words without γάρ.  In Luke γὰρ stands between ἄξιος and ὁ, so the six words are not consecutive there.  The verb difference in the Deuteronomy citation looked like a second development marker, but the SBLGNT apparatus records φιμώσεις as a variant reading at 1 Corinthians 9:9, where the text has κημώσεις, so the article deliberately does not use it.  The article uses it to distinguish shared wording from proof of dependence on written Luke; the scope of the Scripture formula is disputed. ↩

  53. 53.

    Martyn, J. Louis.  History and Theology in the Fourth Gospel.  New York: Harper & Row.  1st edition.  1968.  Page 36:  “the wording of the Benediction Against Heretics was a Takkanah, published by the Jamnia academy about A.D. 85.”  Reuven Kimelman, "Birkat Ha-Minim and the Lack of Evidence for an Anti-Christian Jewish Prayer in Late Antiquity," in E. P. Sanders and others, eds., Jewish and Christian Self-Definition, volume 2 (Philadelphia: Fortress, 1981), pages 226–244, at page 244, says: "There is no unambiguous evidence that Jews cursed Christians during the statutory prayers."  Ruth Langer, Cursing the Christians? A History of the Birkat HaMinim (New York: Oxford University Press, 2012), page 29, says: "Beginning in the mid-1990s, some prominent New Testament scholars who had previously advocated Martyn's theory backed away from it, lessening its influence in scholarly circles."  Joel Marcus is quoted through Langer, page 266 note 65: his "Birkat Ha-Minim Revisited," New Testament Studies 55 (2009), "explicitly defends 'the basics of Martyn's reconstruction while nuancing the extent of rabbinic control in the early Christian centuries' (523)."  A scholarly study of the Gospel of John.  The article uses it because Martyn dated a synagogue prayer against heretics to the Jamnia academy about A.D. 85, which gave the expulsion passages in John a late date, and the article calls that anchor broken. ↩

  54. 54.

    Eusebius.  Onomasticon.  Translated by C. Umhau Wolf.  n.d.  Section Onomasticon, Bezatha:  In C. Umhau Wolf's translation of 1971, Eusebius says of the pool:  “Once … had five porticos.  There are now pointed out twin pools”.  The Greek is Καὶ νῦν δείκνυται ἐν ταῖς αὐτόθι λίμναις διδύμοις.  Section Bordeaux Pilgrim, A.D. 333:  The Bordeaux Pilgrim says:  “Within the city are the twin pools, with 5 porticoes, called Betsaida.”  Wieland Willker's textual commentary also shows that no manuscript reads "was" for "is" at John 5:2.  John's past-tense descriptions of places are at 4:6, 11:18, 18:1 and 19:41.  A list of the places named in the Bible, each with a note, written by Eusebius in the early fourth century.  The article uses it for his report that the twin pools of John 5:2 were still being pointed out, which shows the pool outlived the war. ↩

  55. 55.

    Matthew 27:8 and 28:15, Joshua 22:3 and Deuteronomy 3:14.  n.d.  A comparison of the Greek phrase for "to this day" in Matthew 27:8 (ἕως τῆς σήμερον) and 28:15 (μέχρι τῆς σήμερον ἡμέρας) with Joshua 22:3 and Deuteronomy 3:14 in the Septuagint, the Greek Old Testament.  Joshua 22:3 uses ἕως τῆς σήμερον ἡμέρας of the time the eastern tribes had served through the conquest, which is the campaign, a period the book's own figures at 14:7–10 put at about seven years, while Deuteronomy's wording is different, since 3:14 uses ἕως τῆς ἡμέρας ταύτης of the villages Jair had just named.  The article uses it because both of Matthew's verses are commonly cited for an early date, and the same formula in the Septuagint is used of short spans. ↩

  56. 56.

    The verb is ἐμπίμπρημι, Matthew's ἐνέπρησεν.  It occurs 35 times in the CATSS Septuagint, and in 18 of them, spread over 15 verses, the thing burned is a town or its buildings:  Joshua 8:19, 11:11, 11:13 twice, 16:10;  Judges 1:8, 18:27, 20:48;  4 Kingdoms (2 Kings) 25:9 twice;  2 Chronicles 36:19 twice;  Jeremiah 52:13 twice;  Baruch 1:2;  1 Maccabees 1:31, 5:28, 5:35.Recounted on 1 October 2026.  Seven of the eighteen are the Babylonian burning of Jerusalem in 587 BC, among them Jeremiah 52:13, καὶ ἐνέπρησεν τὸν οἶκον κυρίου καὶ τὸν οἶκον τοῦ βασιλέως καὶ πάσας τὰς οἰκίας τῆς πόλεως.  Working notes, with Karl Heinrich Rengstorf's 1960 study of the formula:  research/per-gospel-internal-dating.md. ↩

  57. 57.

    Irenaeus.  Adversus Haereses.  n.d.  Section 3.1.1:  “After their departure, Mark, the disciple and interpreter of Peter, did also hand down to us in writing what had been preached by Peter”.  The Greek word for "departure" is ἔξοδον, which Eusebius keeps at HE 5.8.3, and it can mean a death or a leaving.  Section Eusebius, HE 6.14.6–7:  Clement of Alexandria, as Eusebius reports him, in the Nicene and Post-Nicene Fathers translation:  “When Peter learned of this, he neither directly forbade nor encouraged it”.  Irenaeus's Adversus Haereses, quoted in the Ante-Nicene Fathers translation.  The article uses it for his report that Mark wrote down Peter's preaching after the "departure" of Peter and Paul, a word usually read as their deaths, and sets it against Clement of Alexandria's account, in which Peter was alive and heard of it. ↩

  58. 58.

    Irenaeus.  “Against Heresies.”  Ante-Nicene Fathers, volume 1.  Translated by Alexander Roberts and William Rambaut.  c. 180 (edition of 1885).

    1. 58a.

      Book 3, chapter 1, section 1:  “Matthew also issued a written Gospel among the Hebrews in their own dialect, while Peter and Paul were preaching at Rome, and laying the foundations of the Church.”  The Greek, preserved by Eusebius at Ecclesiastical History 5.8.2, is τοῦ Πέτρου καὶ τοῦ Παύλου ἐν Ῥώμῃ εὐαγγελιζομένων καὶ θεμελιούντων τὴν ἐκκλησίαν. ↩

  59. 59.

    Chapman, John.  “St Irenaeus on the Dates of the Gospels.”  Journal of Theological Studies 6:563–569.  1905.  Chapman was a Benedictine patristic scholar who himself held that Matthew was the first Gospel.

    1. 59a.

      Page 566, note 1:  “the genitive absolute εὐαγγελιζομένων cannot possibly be pressed to mean 'during the time that Peter and Paul were preaching'.  The notion of contemporaneousness is almost as faint in the phrase as in the English 'While Peter and Paul preached at Rome'.  The chief point in the clause is the contraposition of 'at Rome' to 'among the Hebrews'.” ↩

  60. 60.

    Eusebius.  “Ecclesiastical History.”  Nicene and Post-Nicene Fathers, second series, volume 1.  Translated by Arthur Cushman McGiffert.  c. 325 (edition of 1890).

    1. 60a.

      Book 3, chapter 24, section 6:  “For Matthew, who had at first preached to the Hebrews, when he was about to go to other peoples, committed his Gospel to writing in his native tongue, and thus compensated those whom he was obliged to leave for the loss of his presence.”  The Greek is ὡς ἤμελλεν καὶ ἐφ᾽ ἑτέρους ἰέναι, as he was about to go to others also. ↩

    2. 60b.

      Book 5, chapter 18, section 14:  “He speaks, moreover, of a tradition that the Saviour commanded his apostles not to depart from Jerusalem for twelve years.”  Eusebius is reporting Apollonius, who wrote against the Montanists. ↩

  61. 61.

    Karst, Josef.  “Eusebius Werke, fünfter Band: Die Chronik aus dem Armenischen übersetzt.”  Die griechischen christlichen Schriftsteller 20.  Hinrichs.  1911.  A German translation of the Armenian version of Eusebius's Chronicle, which survives independently of Jerome's Latin.

    1. 61a.

      Section the entries under Gaius and Claudius:  “Petros der Apostel begibt sich, nachdem er zunächst die Antiochener Kirche gegründet, in die Römerstadt und predigt dort das Evangelium”.  “Markos der Evangelist, der Dolmetsch des Petros, zog nach Egiptos und Alexandria”.  The translation has no entry for Matthew writing a Gospel.  It gives Peter twenty years at Rome where Jerome gives twenty-five. ↩

  62. 62.

    Pallmann, David.  “How We Know Matthew's Gospel Was Written in AD 42.”  Faith Because of Reason (YouTube).  Oct 1, 2026.

    1. 62a.

      At 9:28:  “despite giving two distinct time markers, Papias and Irenaeus converge exactly on the precise year of AD 42”. ↩

  63. 63.

    Hill, Charles E.  “What Papias Said about John (and Luke): A 'New' Papian Fragment.”  Journal of Theological Studies 49(2):582–629.  1998.  Hill argues that Eusebius drew Ecclesiastical History 3.24.5–13, which includes the sentence on Matthew, from Papias without naming him. ↩

  64. 64.

    Anonymous.  “The Acts of Peter.”  The Apocryphal New Testament.  Clarendon Press.  Translated by Montague Rhodes James.  c. 190 (edition of 1924).  A legendary narrative of Peter's contest with Simon Magus at Rome.  James dates it not later than AD 200.

    1. 64a.

      Chapter 1:  “At the time when Paul was sojourning in Rome and confirming many in the faith”.  Chapter 5:  “For whereas the twelve years which the Lord Christ had enjoined upon him were fulfilled, he showed him a vision”.  Chapter 1 goes on to send Paul from Rome to Spain.  Chapter 4 says that Paul was not at Rome when Simon Magus arrived, and chapter 5 brings Peter from Jerusalem to deal with him. ↩ ↩

  65. 65.

    Clement of Alexandria.  “Stromata.”  Clemens Alexandrinus, volume 2 (Die griechischen christlichen Schriftsteller 15).  Hinrichs.  Edited by Otto Stählin.  c. 200 (edition of 1906).

    1. 65a.

      Book 6, chapter 5, section 43, page 453:  “μετὰ δὲ δώδεκα ἔτη ἐξέλθετε εἰς τὸν κόσμον· μή τις εἴπῃ· οὐκ ἠκούσαμεν”.  In English, after twelve years go out into the world, so that no one may say, We did not hear.  Clement is quoting the lost Preaching of Peter, where the Lord is speaking to the apostles. ↩

  66. 66.

    Wenham, John.  “Did Peter Go to Rome in AD 42?”  Tyndale Bulletin 23:94–102.  1972.  John Wenham argues from Acts 12:17 that Peter went to Rome in 42.  His subject is the date of Mark;  he does not discuss Matthew.

    1. 66a.

      Page 95:  “we put Peter's first visit to Rome in 42, the whole position is revolutionized”.  Page 97:  “The later lists of the bishops of Rome credit Peter with a twenty-five year episcopate”. ↩

  67. 67.

    Edmundson, George.  The Church in Rome in the First Century.  Longmans, Green.  1913.  The Bampton Lectures for 1913, and the fullest defence of a visit by Peter to Rome in AD 42.

    1. 67a.

      Page 50:  “Now the date of this martyrdom was according to the received tradition the fourteenth year of Nero or 67 A.D.; if then we deduct twenty-five years, we arrive at 42 A.D., which is precisely the date given for St. Peter's first visit to Rome by St. Jerome in his work 'De Viris Illustribus.'” ↩

    2. 67b.

      Page 71:  “Both the Eusebian Chronicle and the Liberian Catalogue give twenty-five years as the term of St. Peter's episcopacy, but they differ as to the dates of its beginning and its end.  We have already seen that the Eusebian date-limits are from 42 A.D. to 67 A.D.; the Liberian, however, are from 30 A.D. to 55 A.D.” ↩

  68. 68.

    Bauckham, Richard.  Jesus and the Eyewitnesses: The Gospels as Eyewitness Testimony.  Eerdmans.  2nd edition.  2017.

    1. 68a.

      Page 423:  “Attractive as this claim is, I doubt if it can be sustained”.  Page 433:  “It is clear that at least 3.24.6-7 comes from a single source, and it is likely that 3.24.11 is from the same source.  Hill has made a good case for his proposal that Papias was the source.” ↩

  69. 69.

    Mason, Steve.  Josephus and the New Testament.  Peabody, Mass.: Hendrickson.  1st edition.  1992.  A scholarly book on Josephus and the New Testament, by the Josephus specialist Steve Mason.  The article uses it for Mason's case that Luke depended on Josephus for the parallels in the Theudas passage of Acts, which the article calls a real problem.

    1. 69a.

      Page 224:  “In short, we cannot prove beyond doubt that Luke knew the writings of Josephus.”  Page 225:  “I find it easier to believe that Luke knew something of Josephus' work than that he independently arrived at these points of agreement.  Nevertheless, we await a thorough study of the matter.” ↩

    2. 69b.

      Page 210:  “Admittedly, Theudas' following of four hundred hardly captures Josephus' claim that he persuaded the majority of the masses: and Josephus does not mention Judas' death as Luke does.  But numbers are notoriously fluid in ancient texts (compare Josephus' own differences with himself on the size of the Egyptian's following!), and Luke's statement that Judas was “destroyed” is quite vague.” ↩