Is New Testament Scholarship Unusually Skeptical?

Summary

A common complaint runs like this.  The early church was near-unanimous about who wrote the Gospels.  No rival names survive.  Every manuscript that carries a title carries the same four.  Yet New Testament scholars treat those attributions as doubtful, when historians of Greece and Rome accept far worse-attested claims without a murmur.  Something other than evidence must be doing the work.

The complaint is testable, so this article tests it.  Five findings, four of them measurements.

1.  In classics, the age of an attribution predicts almost nothing.  Fifteen ancient works were tabulated below, each with the date of composition, the earliest surviving statement of who wrote it, and the field's current verdict.

• Xenophon's Anabasis is credited to Xenophon on one witness writing about 470 years later, who contradicts what the book itself says. Accepted without controversy.
• Procopius' Secret History is credited to Procopius on one Byzantine encyclopedia entry written about 425 years later. Accepted.
• Tacitus' Annals — the backbone of early Roman imperial history — survives in books 1–6 in a single manuscript copied around AD 850, roughly 735 years after the fact. Nobody doubts it.
• The Historia Augusta, by contrast, names its own six authors and dedicates its lives to named emperors. Classicists reject all six names as fictitious and hold that one man wrote it a century later than it claims.

The correlation between how late an attribution is and whether the field accepts it is close to nothing.  What decides is a different question:  does anything inside the text contradict the attribution?  Where nothing does, a very late single witness carries.  Where something does — the Historia Augusta uses office titles that did not exist until the fourth century — even an explicit self-naming falls.

2.  Apply that rule to the Gospels and you get roughly the verdicts New Testament scholarship reaches.  This series has already reached them independently:  Mark's attribution probably holds, because nothing internal contradicts it;  the apostle Matthew's does not, because something internal does — Matthew reproduces twelve separate runs of sixteen or more identical Greek words from Mark, including the account of his own call.¹  That is the Historia Augusta move, and no one calls classicists hyper-skeptical for making it.

3.  There is one real asymmetry, and it is worth naming precisely.  New Testament scholarship routinely treats a Gospel's formal anonymity as evidence against its traditional author.  Classics does not do this.  Caesar never says he wrote the Gallic War;  Tacitus never names himself in the Annals;  Xenophon actively attributes the Anabasis to somebody else.  All three are formally anonymous and no inference is drawn from it.  That is the one place where the charge of a double standard sticks, and it is narrower than the complaint usually claims.

4.  Four documented cases show conclusions outliving the arguments that produced them.  This is the strongest evidence that something is wrong with how the field maintains its results, and every case is verifiable.

• The birkat ha-minim. A rabbinic prayer, dated to a supposed synod at Jamnia around AD 85, was for two generations the fixed point that dated John's Gospel. Specialists in rabbinics have dismantled every link in the chain. The date it produced has not moved.²
• The genre reversal. In 1915 the Gospels were called popular biographies. From the 1920s Bultmann and K. L. Schmidt established that they were folk literature with no biographical interest, and that held for about seventy years. In 1992 Richard Burridge reversed the field. No new evidence arrived in either direction.³
• The criteria of authenticity. Morna Hooker published the decisive objection in 1970–72. The criteria remained the field's standard method for four more decades, through the Jesus Seminar and the Third Quest, until a 2012 volume announced their "demise." Forty years between a published refutation and the field acting on it.⁴
• The date of John. The nineteenth-century Tübingen school put it around AD 160–170. The field moved to around AD 90–100 after a scrap of papyrus, P52, was published in 1935 and dated to about AD 125. That date is now known to have been asserted without evidence — traced link by link by Brent Nongbri in 2005 — and P52's real palaeographic range may run to about AD 200.⁵

5.  The failures have a shape, and it can be measured.  Where New Testament criticism has been embarrassed by later evidence, the embarrassments cluster in arguments from the absence of corroboration.  A worked example:  Luke calls the magistrates of Thessalonica πολιτάρχαι, a title absent from all surviving Greek literature.  In 1898 Ernest DeWitt Burton assembled every known inscription containing it and found seventeen.  The Packard Humanities Institute's epigraphic database, queried for this article, returns fifty-six attestations spread over fifty-one texts, and fifty-three of the fifty-six — 94.6% — come from the one province where Luke uses the word.⁶

The base grew threefold in 128 years.  That is what an argument from silence is worth in a field whose evidence keeps arriving.

And the finding that cuts hardest against the complaint itself.  The most explicit enforcement of a conclusion in this whole field did not happen in a secular department.  On 17 December 1983 the Evangelical Theological Society voted 119 to 36 to go on record rejecting any position that a biblical writer "materially altered and embellished historical tradition," and then voted 116 to 41 to demand Robert Gundry's resignation over his commentary on Matthew.⁷  A conclusion about the composition of a Gospel was settled by a show of hands and made a condition of membership.

Implications

The honest verdict is not the satisfying one.  New Testament scholarship is not applying a standard that classics would refuse.  On the central question — how much weight a late attribution carries — it applies the same rule, and this series reaches the same answers.  What it does badly is what many humanities fields do badly:  it keeps numbers after losing the arguments that generated them, and it reasons from silence in a field where silence is a sampling artifact.

That is a smaller charge than "inbred groupthink."  It is also a more useful one, because it points at specific claims that can be checked rather than at a discipline's motives, which cannot.

How to Test a Charge Like This

"The field is biased" is not, by itself, a claim that can be checked.  Every discipline has fashions;  every scholar has commitments;  pointing at either proves nothing about whether a particular conclusion is right.  Worse, the charge is symmetrical.  A skeptic can say conservatives believe the traditional authors because they were raised to.  A conservative can say critics doubt them because doubt is what gets published.  Both are true of somebody and neither settles anything.

So the charge has to be turned into questions with answers.

Question 1.  Does another field, working on the same kind of evidence, reach different conclusions?  This is the comparison Eric's question asks for and it is the one that can actually be run.  Classical scholarship has hundreds of anonymous or disputed texts.  Tabulate what it does with them and see whether Gospel-quality attributions get accepted or rejected outside the New Testament.

Question 2.  Does the field hold conclusions after the arguments for them fail?  This is the sharpest available test of groupthink, because it does not require knowing anyone's motives.  If an argument is publicly refuted and the number it produced stays put, something other than evidence is holding the number in place.  It only needs cases where the refutation is documented and dated.

Question 3.  When the field made a claim the external world could check, how did it do?  Archaeology and epigraphy keep arriving.  Every arrival is a scored test.

Question 4.  Is the consensus as independent as it is described?  A "broad consensus including skeptics" is worth something only if the people counted reached their views independently.

Question 5.  What does the same test show about the other side?  A test that can only produce one answer is not a test.

The rest of this article runs these five.

Test 1: What Other Fields Do With the Same Kind of Evidence

The standard the Gospels do not have

Start with what a really well-attested ancient attribution looks like, so the comparison is not rigged.

Caesar's Gallic War was published around 51 BC.  Caesar never says in it that he wrote it;  he refers to himself in the third person throughout.  But in 46 BC — within about five years — Cicero puts this in Brutus' mouth in the dialogue Brutus:

"atque etiam commentarios quosdam scripsit rerum suarum." "And he also wrote certain commentaries on his own deeds."⁸

A named contemporary, in a datable work, five years after publication, attributing the book to a man he knew personally.  Nothing in the Gospel tradition is remotely like this.  The earliest Gospel attribution is Papias, preserved second-hand by Eusebius, writing perhaps thirty to sixty years after Mark.⁹

Say this plainly, because apologetic treatments usually do not:  by the best classical standard, the Gospels' external attribution is mediocre.  The question is whether the best classical standard is the one classics actually uses.  It is not.

What classics accepts on far less

Xenophon's Anabasis.  This is the crown case, and it is worth walking through slowly.

The Anabasis narrates the march of the Ten Thousand, in which Xenophon himself was a commander.  In his later Hellenica, Xenophon summarises those events and then says:

"Θεμιστογένει τῷ Συρακοσίῳ γέγραπται." "It has been written by Themistogenes the Syracusan."¹⁰

So the author of the Anabasis assigns his own book to another man.  There is no earlier testimony.  The first surviving statement that Xenophon wrote it comes from Plutarch, around AD 100 — roughly 470 years later:

"Xenophon, to be sure, became his own history by writing of his generalship and his successes and recording that it was Themistogenes the Syracusan who had compiled an account of them, his purpose being to win greater credence for his narrative by referring to himself in the third person, thus favouring another with the glory of the authorship."¹¹

One witness.  Four and a half centuries after the fact.  Contradicting the text's own attribution.  Classicists accept it, and the question is not live.

Notice the second payload in that sentence, because it bears directly on a standard argument about the Gospels.  Plutarch states as a matter of course that third-person self-reference was a recognised ancient device for gaining credence.  When a New Testament scholar argues that John's Gospel cannot be by an eyewitness because it speaks of the beloved disciple in the third person, Plutarch is the counter-example, and he is not a Christian apologist.

Procopius' Secret History.  Procopius wrote around AD 550.  For a thousand years the only evidence that his scandalous Anecdota existed at all was one entry in the Suda, a Byzantine encyclopedia compiled after about 975:

"ἔγραψε καὶ ἕτερον βιβλίον, τὰ καλούμενα Ἀνέκδοτα, τῶν αὐτῶν πράξεων." "He wrote also another book, the so-called Anecdota, on the same events."¹²

The manuscript surfaced in the Vatican in 1623.  The attribution rests on that single notice, roughly 425 years after composition.  Accepted.  In fairness, the internal case is strong — the work cross-refers to Procopius' Wars — but that is exactly the point being established:  it is the internal evidence, not the external witness, that is doing the work.

Tacitus' Annals.  This is the comparison worth remembering, because the Annals is not a curiosity.  It is the single most important narrative source for the early Roman Empire, and much of what any educated person believes about Tiberius, Claudius and Nero comes from it.

• Tacitus does not name himself in it.
• Books 1–6 survive in one manuscript, written around AD 850, probably at Fulda — roughly 735 years after composition.¹³
• The earliest external attribution of any Tacitean history is Tertullian, around AD 197, who names "Cornelius Tacitus" and cites the fifth book of the Histories.¹⁴ For the Annals specifically the field leans on Jerome, around AD 406, referring to Tacitus' thirty volumes of imperial lives — nearly three centuries after the fact — and on the book-numbering colophons of the medieval manuscripts.¹⁵

Set that beside the Gospels.  Mark's earliest attribution is thirty to sixty years after composition;  the titles are attested in manuscripts from around AD 200;  the attribution is unanimous across every surviving branch of the tradition in three languages.  On external attestation alone, Mark is better attested than the Annals.  No classicist writes "Pseudo-Tacitus."

What classics rejects despite far more

The traffic runs both ways, and this half matters more, because it is where the charge of unique skepticism actually breaks.

The Historia Augusta.  A collection of imperial biographies that does everything the Gospels do not.  It names its authors — Aelius Spartianus, Julius Capitolinus, Vulcacius Gallicanus, Aelius Lampridius, Trebellius Pollio, Flavius Vopiscus — and dedicates individual lives to Diocletian and Constantine, placing itself firmly in the early fourth century.

In 1889 Hermann Dessau argued that all of this is a literary ploy:  one author, writing under Theodosius (379–395), invented the six names and the dedications.¹⁶  Ronald Syme, the thesis's most influential defender, called the author "a rogue grammaticus."¹⁷  The field went along.

What forced it was internal evidence.  The lives contain "mentions of office and titles that only came into being in the middle of the fourth century, decades after the supposed date of its composition," along with fabricated documents and cited authors who appear nowhere else and were probably invented.¹⁸

Then the stylometry, and its history is a caution rather than a triumph.  Ian Marriott ran the first computational analysis in 1979 and got single authorship.  Emily Tse, Fiona Tweedie and Bernard Frischer ran their own in 1998 and got multiple authorship.  Jeffrey Stover and Mike Kestemont ran a third in 2016 and concluded that "the authorial structure presented by the manuscripts … is not compatible with the features of the text," with two authorial layers inside a single hand.¹⁸  Marriott's study, they note, was "marred by methodological errors, particularly the use of sentence length as a criterion of authorship, which … should definitely not be used for ancient texts, where the punctuation is due to the modern editor."¹⁸

Three computational studies, three answers, thirty-seven years.  Any reader inclined to trust a stylometric verdict about a Gospel — including the ones in this series — should read that sentence twice.

[Longinus], On the Sublime.  The treatise survives in one tenth-century manuscript, Parisinus graecus 2036, whose heading calls it "by Dionysius Longinus" and whose table of contents calls it "by Dionysius or Longinus."  Classicists read the names as the conjectures of a medieval scribe torn between Dionysius of Halicarnassus and Cassius Longinus, and most reject both.¹⁹  The work is published today as by Pseudo-Longinus.

[Cicero], Rhetorica ad Herennium.  Transmitted as Cicero's from late antiquity through more than a hundred medieval manuscripts.  Renaissance scholars pulled the attribution apart on internal grounds, and the author is now unknown.²⁰

The closest structural twin

One case matches the Gospel situation almost exactly, and it deserves its own heading.

Three books in the Caesarian corpus — the Alexandrian, African and Spanish wars — are anonymous continuations of Caesar's commentaries, written around 45 BC.  The earliest surviving statement about who wrote them is Suetonius, around AD 121:

"Nam Alexandrini Africique et Hispaniensis incertus auctor est:  alii Oppium putant, alii Hirtium, qui etiam Gallici belli novissimum imperfectumque librum suppleverit." "For of the Alexandrian, African and Spanish wars the author is uncertain:  some think Oppius, others Hirtius, who also completed the last and unfinished book of the Gallic War."²¹

An anonymous text.  A named attribution surfacing roughly 165 years later.  The witness himself hedging between two candidates.  That is a very close structural match to a formally anonymous Gospel with a title attested in manuscripts a century or more after composition.

And here is what the field does with it.  It splits them.  The Alexandrian War is usually given to Hirtius;  the African and Spanish wars are usually left to unknown officers of Caesar's army, because their Latin is markedly cruder.²²  The same external evidence covers all three, and the internal evidence decides them individually.

That is exactly what this series did with the four Gospels — the same external chain, four different verdicts, decided book by book on what is inside each one.

The table

Every row was checked against a primary text or a specialist study.  "Gap" is the interval between composition and the earliest surviving statement of authorship.

What the table shows

The gap column predicts nothing.  Attributions accepted by classicists sit at 5, 90, 290, 425 and 470 years.  Attributions rejected sit at 0, 165, 400 and 900.  Sort the table by gap and the verdicts scatter.

One column does predict:  whether internal evidence contradicts the attribution.  Every rejected case has a specific internal problem.  The Historia Augusta uses fourth-century office titles.  The African War's Latin is not the Alexandrian War's.  The ad Herennium does not read like Cicero.  The Longinus manuscript cannot decide between two names.  Every accepted case has none — including the ones with the most absurd gaps.

So the operative rule in classical scholarship is:  a late external attribution stands unless something inside the text contradicts it.

Now apply that rule to the four Gospels, and read the last four rows of the table.  Mark stands, because nothing internal contradicts a companion of Peter writing it.  Matthew falls, because something internal does:  a Gospel that reproduces Mark's Greek by the paragraph, including Mark's account of the call of the man the book is named for, is very hard to attribute to an eyewitness apostle.²³  Luke stands, and stands strongest.  John half-stands.

Those are the verdicts the classical rule delivers, and they are close to the verdicts the New Testament guild delivers.  On this test the charge of unique skepticism fails.

What the comparison cannot settle

Three limits, stated before anyone else states them.

The sample was chosen, not drawn.  Fifteen works picked to illustrate a range is not a random sample of classical scholarship.  A proper version would define a population — say, every anonymous prose work in the Oxford Classical Dictionary — and score all of it.  That would be a real study and this is not it.  What this table can support is a sufficiency claim:  classics both accepts and rejects on Gospel-quality evidence, so no single standard is being withheld from the New Testament.

The stakes really are different, and that is not automatically illegitimate.  Nobody's life changes if the Anabasis turns out to be by Themistogenes.  When a claim carries more weight, demanding more of it is ordinary practice, not prejudice.  The apologetic complaint tends to treat any extra scrutiny as bias.  It is not.

Classics has its own diseases and this article is not competent to audit them.  The Historia Augusta's three contradictory stylometric verdicts are a hint that the neighbour's house is not tidy either.

Test 2: The One Place the Charge Sticks

Run the comparison the other way and one asymmetry survives it.

New Testament scholarship regularly treats a Gospel's formal anonymity as evidence against the traditional attribution.  The argument is that a genuine eyewitness would have named himself, so anonymity points away from eyewitness authorship.²⁴

Classics does not reason this way, and the table shows why it cannot.

• Caesar never claims the Gallic War. Third person throughout.
• Tacitus never names himself in the Annals.
• Xenophon actively attributes the Anabasis to a fictitious Syracusan — and Plutarch says he did it to be believed more, not less.
• The Historia Augusta, the one work in the table that does name its authors, is the one work in the table whose names are fictitious.

Put those four facts together and the inference collapses in both directions.  Naming yourself is compatible with fraud.  Not naming yourself is compatible with being the eyewitness commander of the army you are describing.  Whatever formal anonymity is evidence of, it is not evidence about eyewitness status, and the field that handles the most anonymous prose works has quietly known this for a long time.

This is a narrow finding.  It does not show that the Gospels were written by their traditional authors.  It shows that one specific argument routinely used against them would not be made in the discipline next door.

Test 3: Does the Consensus Track the Evidence?

This is the test that produces real evidence of dysfunction, and it does not require guessing at anyone's motives.  If an argument is publicly demolished and the conclusion it supported does not move, then the conclusion is not resting on the argument.

Four cases.  All four are documented, dated, and reported by scholars inside the field.

Case 1: a dating argument collapsed and its dates did not move

For most of the twentieth century the standard reason for dating John's Gospel to around AD 90 was the birkat ha-minim — the "benediction against heretics," a Jewish prayer thought to have been promulgated by a rabbinic synod at Jamnia around AD 85 to expel Christians from synagogues.  John's Gospel uses a rare word, ἀποσυνάγωγος, "put out of the synagogue," three times.  J. Louis Martyn's History and Theology in the Fourth Gospel (1968) tied the two together and dated the Gospel from the prayer:  "a Takkanah, published by the Jamnia academy about A.D. 85."²⁵  One specialist's judgement is that Martyn's book above all gave the reading "an almost canonical status."²⁶

Every link has since failed.  Reuven Kimelman showed in 1981 that the benediction was not aimed at Gentile Christians and marked no watershed.  Ruth Langer's 2012 monograph is the standard account:

"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.  For our purposes, these arguments are sufficient to establish that the New Testament sources themselves present an unreliable witness to the birkat haminim."²⁷

The date is still standard.  Open almost any introduction and John is placed around AD 90–100.  The argument that put it there is gone.

Two honesty notes, because this case is easy to abuse.

First, removing a bad late-dating argument widens the window;  it does not move the book earlier.  Anyone citing the birkat ha-minim collapse as evidence for an early John is overreaching.

Second, the theory is not universally abandoned.  William Horbury defended it in 1982;  Joel Marcus has defended a nuanced version;  Martyn himself, according to his editor, "sees no reason to retreat."²⁸  What is documented is a collapse of the specialist consensus in rabbinics that the argument depended on, not unanimity against it.

Case 2: the same evidence, two opposite consensuses

In 1915 Clyde Votaw called the Gospels popular biographies.  From the 1920s K. L. Schmidt and Rudolf Bultmann established the opposite — that they were folk literature, sui generis, with no biographical interest — and that held for roughly seventy years.  In 1992 Richard Burridge tested the Gospels against Greco-Roman bioi on formal features and reversed the field.  One of his measures:  Jesus is the grammatical subject of 24.4% of Mark's verbs, against 18.7% for Xenophon's Agesilaus and 18% for Tacitus' Agricola.²⁹

Nothing new arrived.  Burridge's comparanda were on the shelf in 1920.  His method needed no technology Bultmann lacked.  The field simply held one confident position for seventy years and then held the opposite.

A discipline that can do that has shown that its confidence is not a reliable guide to its evidence.  This is a stronger finding than any anecdote about gatekeeping, because it is a property of the field's output rather than of anyone's behaviour.

Case 3: forty years between a refutation and its acceptance

The "criteria of authenticity" — dissimilarity, multiple attestation, embarrassment, coherence — were for decades the machinery by which historical Jesus scholars decided which sayings were genuine.  Morna Hooker published the decisive objections in 1970–72:  the criteria cannot deliver what they are designed to deliver, because the criterion of dissimilarity in particular can only find a Jesus unlike both Judaism and the church, which is to say a Jesus no historian should expect to exist.

The criteria remained standard practice for another forty years.  They powered the Jesus Seminar's votes through the 1980s and 1990s and most of the Third Quest.  In 2012 Chris Keith and Anthony Le Donne published Jesus, Criteria, and the Demise of Authenticity, whose contributors argue either that the criteria should be abandoned or that they need fundamental recasting, and whose lead essay's position is that Hooker had been right all along.³⁰

And here is the part that matters:  the historical reconstructions the criteria produced are still in circulation.  Retiring a method does not automatically retire the results it generated;  somebody has to go back and redo them, and largely nobody has.

Case 4: the field moved on a papyrus, and the papyrus was misdated

The Tübingen school of the mid-nineteenth century dated John's Gospel to around AD 160–170 on the ground that its theology was too developed for the first century.  That was a confident consensus of a leading school.³¹

It was overturned by an object.  P52, a credit-card-sized scrap of John 18 in the John Rylands Library, was published in 1935 and dated to about AD 125.  A second-century date for the manuscript makes a mid-second-century date for the composition impossible, and the field moved.

Two lessons run in opposite directions and both belong here.

The first is against the critics:  a whole school's confident date was wrong by sixty years or more, and nothing internal to the discipline caught it.  A scrap of papyrus did.

The second is against the apologists, and this series has already adopted it.  Brent Nongbri traced the "AD 125" figure link by link in 2005 and found it had no evidentiary basis:  Roberts's cautious "first half of the second century" became Aland's "Anfang 2. Jahrhundert," became "um 125," became a later limit in the standard handbook.  Nongbri's verdict:  "This so-called 'consensus' in 'recent opinion,' as it rests on assertions with no evidence, is highly dubious."³²  Roberts himself had written that "any exact dating of book hands is, of course, out of the question."³³  The honest palaeographic range for P52 may extend to around AD 200.

So the field's most famous self-correction was itself driven by an overconfident number.  The same pathology produced the error and the correction of the error.

What four cases can and cannot show

They cannot show that any current New Testament conclusion is wrong.  A field with a bad record of retiring conclusions can still be right about a given one.

What they show is narrower and still worth having:  when a New Testament date or reconstruction is cited as "the consensus," that alone is worth close to nothing, because the field has demonstrably kept dates whose arguments died, held opposite positions on identical evidence, used a method for forty years after its refutation, and moved on a number that had been asserted rather than measured.

That is not a claim about anyone's honesty.  It is a claim about a documented failure rate, and it licenses exactly one thing:  going and looking at the evidence yourself.

Test 4: The Failures Have a Shape

The four cases above are literary.  This test asks a different question:  when New Testament criticism made a claim that the physical world could check, how did it do?

The politarch numbers

Acts 17:6 and 17:8 call the magistrates of Thessalonica πολιτάρχαι.  The title appears nowhere in surviving Greek literature.

In July 1898 Ernest DeWitt Burton published the definitive catalogue in The American Journal of Theology.  He opens:

"It is a fact well known to New Testament scholars that the word πολιτάρχης found in Acts 17:6, 8 does not occur elsewhere in extant Greek literature, and has its only other vouchers in inscriptions."³⁴

His conclusion, after thirty-four pages assembling everything he could reach across Europe:

"There are in all seventeen known inscriptions which attest the existence of the office of politarch in ancient cities, to which two more may be added if we accept somewhat probable restorations of them."³⁵

The Packard Humanities Institute's Searchable Greek Inscriptions, queried for this article on 24 August 2026, returns 56 matches in 51 texts.  By region:  Northern Greece 53, Central Greece 1, Egypt and Cyrenaica 2.³⁶

Two figures, then.  The evidence base for one word roughly tripled in 128 years.  And fifty-three of the fifty-six attestations — 94.6% — come from Macedonia, the one province where Luke uses the title.

That is the mechanism behind a whole class of failures.  An argument from surviving text is an argument from a sample you are holding.  An argument from the silence of surviving sources is an argument from a sample whose size you do not know.  In 1898 a scholar collecting every attestation he could reach found seventeen.

What the usual telling gets wrong

The politarch story is normally told like this:  critics said Luke invented the word, then inscriptions vindicated him.

No named critic has been found who said so.  Burton, writing the standard catalogue, reports no such charge;  he treats the word as an ordinary lexicographical problem and cites Grimm, Thayer, Boeckh and Tischendorf.  Searching for a nineteenth-century scholar who alleged a Lukan error over πολιτάρχης turns up apologetic sources that say "critics" and name nobody.

This matters more than it looks.  A recurring feature of popular apologetics is an unnamed antagonist:  "scholars used to say."  Once the antagonist has no name, the claim cannot be checked, and an argument that cannot be checked is not evidence.  The honest version of the politarch case is weaker and still useful:  a title unattested in surviving literature turned out to be precisely right for precisely one province, and the evidence that showed it arrived after the reference works of the period were written.

Where the critical position won, including against this series

A track record with only wins is a curated list.  Here is the other column, and several entries cost this project something.

• The Gospel titles are secondary. Conceded in this series. Even Simon Gathercole, who argues the titles are very early, dates them to the second century.
• The apostle Matthew did not write the Greek Gospel of Matthew. This series' own verdict, reached on internal evidence, and it is the critics' verdict.
• Markan priority, against the unanimous ancient tradition that Matthew came first. The Fathers were wrong; the moderns are right.
• Mark 16:9–20 and John 7:53–8:11 are not original. The manuscripts settle it.
• The Anti-Marcionite Prologues are late. Engelbert Gütwenger dated the Luke prologue to about AD 400 — in 1946, twenty-three years before the standard treatment said the same. That removes an early-attestation claim that apologetic literature still uses.
• P52 does not date to AD 125, as above. A critical correction of an apologetic overreach, adopted by this series.

The pattern in both columns is the same and it is not about ideology.  The critical position wins where the evidence survives in quantity — manuscripts, and the Greek text itself.  It loses where it argued from absence.

The limit of this test

Everything above is drawn from lists other people assembled, and those lists are curated by whichever side won.  A properly designed version would fix a set of falsifiable claims before the outcomes were known and score them afterwards.  Nobody has done that, this article has not done it, and until somebody does, the shape described here is a hypothesis with good supporting cases rather than a measurement.

Test 5: Is the Consensus as Independent as Advertised?

Sociology is not epistemology

The sharpest available criticism here comes from Lydia McGrew, an analytic philosopher who works on evidence and testimony, and it does not depend on her being right about anyone's motives.

Her target is the practice of licensing a historical claim by the percentage of scholars who hold it.  In the "minimal facts" argument for the resurrection, a proposition earns the label "historical bedrock" — a term that sounds like an assessment of the evidence — only if it also commands a consensus of somewhere between two-thirds and ninety-five per cent of scholars across the spectrum.  McGrew's objection:

"The problem lies in withholding terms of apparent epistemic approbation on the grounds of the absence of social approval of those propositions."³⁷

Whether a proposition is well-evidenced and whether scholars accept it are two different facts about the world.  Requiring both, and then using a word that sounds like the first, quietly converts a head-count into a warrant.  This objection is correct as a matter of logic, and it applies to every use of a consensus figure — including the ones in apologetics.

She adds a claim about the sociology which is harder to check:

"In the contemporary milieu, the very concept of 'heterogenous majority,' applied to people who are all part of the guild and seeking a place at the table during our own time, is often a misleading concept.  It is an illusion of diversity."³⁷

And elsewhere, more bluntly:  "the field of NT studies in particular is inbred, subject to perpetuating epistemic pathologies, and needs an outsider perspective."³⁸

What could be checked, and what could not

Checkable:  the appeal to credentials was made.  McGrew documents Michael Licona repeatedly discounting her criticism on the ground that "the most vocal is not a gospels expert or even a New Testament scholar," and declining to engage arguments not published in "respected, peer-reviewed journals in the relevant fields."³⁸  The podcasts and posts are public.  What this establishes is that the argument-from-credentials was used in a live methodological dispute — not that the field as a whole runs on it.

Not checkable here:  the composition of the guild.  Testing "illusion of diversity" properly would mean assembling the population of scholars whose agreement is being counted and measuring how independent their training and institutions actually are.  Nobody has published that dataset, this article did not build it, and the claim therefore stands as an allegation by an interested party rather than a finding.  It is worth someone's dissertation.

Partly checkable:  where the percentages come from.  The consensus figures most often quoted in resurrection apologetics derive from Gary Habermas' private bibliography of roughly 3,400 publications, from which he traces scholarly positions on some 140 sub-questions.  The bibliography itself has not been published in a form that allows an outsider to recount it.  A statistic whose underlying data cannot be inspected is not a measurement, whichever side quotes it.

The density problem, which nobody chose and everybody inherits

One structural fact about the field is measurable and probably explains more than any allegation of bias.

The Greek New Testament contains 137,554 word tokens, counted for this article from the morphologically tagged SBL text.³⁹  The Society of Biblical Literature describes itself as "a membership organization with over 7,000 scholars."⁴⁰

That is roughly twenty words of primary text per member of the professional society — and SBL is only part of the field, while the New Testament is only part of SBL's remit.

Compare any other ancient corpus.  There is no body of classical literature attended by anything remotely like this density of professional interpreters.  And an academic career is advanced by novelty:  a doctorate needs an original contribution, a monograph needs a thesis, a journal needs a claim not already in print.

Put a very large number of career-motivated specialists on a very small text and hypotheses will proliferate faster than evidence can arrive to test them.  That is not a conspiracy and requires no bad faith.  It is arithmetic, and it predicts precisely the pattern documented in Test 3:  elaborate reconstructions built on thin foundations, held with confidence, abandoned slowly.

The Other Direction: the Most Explicit Enforcement Was Not Secular

Any test that can only return one answer is worthless, so run this one on the conservative side.

116 to 41

The Evangelical Theological Society met in Dallas on 17 December 1983.  Robert Gundry, professor of New Testament at Westmont College, had published a 652-page commentary on Matthew arguing that Matthew used midrash — that some of the material in the infancy narratives in particular was theological elaboration rather than reportage.

Norman Geisler organised the response.  He circulated a petition, gathered signatures from seminary faculty, and the evening before the vote distributed a document titled "Why We Must Vote Now on Gundry's Membership, and Why We Must Vote No on Gundry's Membership," hinting that a rival society would be formed if ETS did not act.

Two motions passed at the plenary business session.⁴¹

• George Knight III's motion "that the ETS go on record as rejecting any position that states that Matthew or any other biblical writer materially altered and embellished historical tradition or departed from the actuality of events." 119 to 36.
• Roger Nicole's motion to request Gundry's resignation. 116 to 41 — 73.9%.

ETS then had 1,698 members, each of whom signs the inerrancy statement annually.  Gundry resigned.

What this costs the complaint

Notice what happened there.  A scholarly society settled a historical question — how the Gospel of Matthew was composed — by a vote of the membership, in advance of the argument, and made the answer a condition of belonging.  Nothing in the secular guild is that explicit.  Soft consensus pressure, which is the charge against critical scholarship, is a weaker mechanism than a recorded ballot.

And 73.9% is a consensus figure of exactly the kind apologists quote from Habermas.  Any argument that treats a percentage of scholars as evidence has to accept this percentage too, or explain the difference on grounds other than which answer it produced.

What this project would fall under

This series concluded that the apostle Matthew did not write the Greek Gospel of Matthew.  That is a different claim from Gundry's — it concerns who held the pen, not whether the contents are historical — but it would fall under Knight's motion just as squarely, since it holds that the book's traditional attribution is mistaken.  A young-earth, inerrancy-affirming project reached it from the Greek text.

There is no way to say that and also claim the pressure only runs one way.

Verdict

On the question as asked — would another field of history be this skeptical of Gospel authorship, given the evidence and the early near-consensus? —  the answer is yes, and in places more so.

Classical scholarship rejects the Historia Augusta's six named authors, the Ciceronian ad Herennium, the Longinus of On the Sublime, and two of the three Caesarian continuations.  It does so on exactly the kind of internal evidence that leads New Testament scholars to doubt apostolic Matthew.  It also accepts the Anabasis on one witness 470 years late and the Annals on a ninth-century manuscript.  The rule it follows is consistent:  a late attribution stands unless something inside the text contradicts it.  New Testament scholarship follows the same rule, and this series' four verdicts are what that rule produces.

On the sharper charge — that the field is inbred and maintains conclusions by inertia — the evidence is real but narrower than the charge.

What is documented:  four dated cases of conclusions surviving the arguments that produced them, one of them a seventy-year consensus reversed without new evidence and one a method used for forty years after its refutation.  A characteristic failure mode — arguing from the silence of surviving sources — with a measurement showing what that silence is worth.  A structural incentive, twenty words of text per member of the professional society, that predicts exactly this pattern without requiring anyone to be dishonest.

What is not documented:  that the guild's diversity is an illusion.  That is an allegation, made by an interested party, and testing it would need a dataset nobody has built.

And one finding cuts squarely against the people most likely to enjoy this article.  The one place in this field where a conclusion about Gospel composition was fixed by an explicit vote and made a condition of professional membership was a conservative evangelical society, in 1983, by 116 to 41.

The practical upshot is the same either way, and it is the reason this series measures things rather than counting scholars.  A New Testament consensus, on its own, is worth close to nothing as evidence — not because the scholars are corrupt, but because the field has a demonstrated record of holding numbers whose arguments have died.  So go and look.  Count the words of Greek.  Read the inscription.  Check the folio.

Common Objections

"You have picked fifteen classical works to make a point. A different fifteen would give a different answer."

Correct, and said above.  The table cannot support a claim about what classics does on average.  It supports a weaker claim that is still sufficient:  classics both accepts and rejects attributions resting on Gospel-quality evidence, so no standard is being withheld from the New Testament.  If someone assembles a properly sampled version and it shows classicists accept manuscript attributions at a much higher rate, that would be a real finding and it would change this article.

"The Gospels matter more, so of course they get more scrutiny. That is not bias."

Agreed, and stated in the article.  Extra scrutiny of a claim carrying more weight is ordinary practice.  The tests here are not about how much scrutiny but about what kind:  whether the inference rules differ, and whether conclusions survive the failure of their arguments.

"Three of your four 'conclusions outliving arguments' cases are decades old. Has the field not improved?"

Partly.  The 2012 volume on the criteria and the mid-1990s retreat from Martyn are themselves the field self-correcting, and Nongbri's demolition of the P52 date was published in a mainstream journal and has been widely accepted.  That is a discipline working.  What the cases show is the lag, not paralysis:  forty years for the criteria, roughly fifteen for the birkat ha-minim's consequences, seventy for the genre question.  A field can be self-correcting and still be a poor guide to truth on any given decade's consensus.

"Arguments from silence are legitimate when the silence is loud enough."

Yes.  If a document that would certainly have mentioned an event does not, that is evidence.  The politarch numbers do not show that arguments from silence are always bad;  they show how to price one.  The right question is always what fraction of the relevant evidence survives, and the answer for Greek epigraphy in 1898 was:  a third of what a laptop can reach today, for that one word.

"You are just excusing the critics."

The article's central finding is that the critics' rule is the same rule classicists use — which is a defence of the method, not of every conclusion drawn with it.  On the specific conclusions, this series disagrees with the critical consensus in several places:  it holds that Mark's attribution probably stands, that Luke–Acts is by a companion of Paul, that no Gospel has a demonstrable floor after AD 70, and that a formally anonymous work is not thereby less likely to be an eyewitness's.  A method can be sound and its practitioners still wrong.

"The ETS material is whataboutism."

It would be, if the article's thesis were "conservatives are worse."  It is not.  The ETS vote is here because Test 5 asks whether the field's consensus figures mean what they are said to mean, and 116-to-41 is a consensus figure produced by an explicit ballot in a society whose members sign a doctrinal statement.  If that number does not establish a historical fact about Matthew — and it does not — then neither does any other percentage of scholars, quoted by anyone.

"If a consensus is worth nothing, why cite scholars at all?"

Scholars are cited here for two things:  facts they have established that anyone can check (Burton's seventeen inscriptions, Dessau's fourth-century office titles, Nongbri's genealogy of a date), and admissions against interest (Langer on the retreat from Martyn, Nongbri on an apologetic favourite).  Neither use is an appeal to authority.  What the article declines to do is treat "most scholars hold X" as an argument for X.

Sources