Reliability of Scripture

Correct, and it is worth being sharp about it: most Greek manuscripts are Byzantine, from the ninth century onward, and the early papyri number only around 140. But the late copies are not worthless. They preserve readings and let scholars trace how the text was transmitted in different regions. The point of the count is comparative: even restricted to early witnesses, the New Testament has more manuscripts within two centuries of composition than any classical text has in total. The raw number is a starting point; how early the copies are is the exhibit that carries the weight.

The non-Greek totals are older, widely repeated figures that are not audited the way the Greek count is, and a range is safer than a hard number. On usefulness, translations cannot settle fine points of Greek grammar, but they are excellent for detecting large-scale changes such as added or missing verses: a verse present in second-century Syriac and Latin cannot have been inserted into Greek manuscripts in the fourth century. That is the specific work the versions do, and it is the work that matters for the claim that the text was corrupted.

The comparison is not meant to show the New Testament is more true than Homer. It shows that the common objection, that we cannot know what the text originally said, would apply with far greater force to every other ancient text people happily rely on. Motivated copying cuts both ways: it produces many copies, and it also produces the theological variants critics point to. Both effects are real, and textual criticism exists to sort them out. That the sorting can be done with unusual confidence for the New Testament is the whole point.

A fair challenge, and the only honest answer is to keep correcting. The old chart was retired publicly, the Greek manuscript count was updated, the 24,000 total is flagged as unaudited, and the 95 percent word-for-word figure for the Dead Sea Scrolls was dropped. Every claim on this site carries a correction link for exactly this purpose. A site that corrects its own side’s overstatements should be more trustworthy on the figures it keeps, not less.

Nongbri’s caution is well founded, which is why P52 is given a range rather than a single date. The larger point survives the correction. Even at AD 200, the gap between composition and earliest fragment is about a century, and substantial papyri such as P46 (Paul’s letters) and P66 and P75 (John and Luke) date to around 200. For Plato and Tacitus the gap is over a thousand years. P52 is a symbol; the case rests on the whole body of early papyri.

This is Ehrman’s strongest point and it deserves a straight answer. First, the early period is not blank: there are papyri from around 200, quotations in second-century writers such as Justin and Irenaeus, and early translations, all showing the same text. Second, the theory requires a coordinated change before any surviving copy was made, across communities that were often at odds with one another and had no central authority to enforce an edit. Where early editing did happen (the long ending of Mark, for instance), it left fingerprints in the manuscript tradition that scholars have found. The absence of such fingerprints for the core text is evidence, not merely absence of evidence.

True of each fragment individually; the argument is cumulative. Around twenty copies covering different portions of Genesis, all agreeing with the later text where they overlap, give a broad sample of the book across two centuries and many scribes. No single scrap proves stability, but two dozen scraps from different hands and decades, none of which shows a different Genesis, is strong sampling evidence. The claim is that nothing in the sample points to another version, not that any one piece proves the whole.

Correct, and it is a good example of the textual plurality the Scrolls reveal: in the last centuries BC several text forms circulated side by side. The expansions in the Samaritan-type Exodus are mostly harmonizing additions that repeat material from elsewhere in the Pentateuch; they do not introduce new content or change the narrative. Seeing that form beside the proto-Masoretic one is exactly how scholars learned which readings are older. The claim is not that only one Exodus existed but that the one that became standard was faithfully preserved.

The script matters for a different reason: paleo-Hebrew was a deliberately archaic choice reserved for texts regarded as especially sacred, which shows the care the community took with the Torah. On accuracy, 11QpaleoLev is a substantial scroll whose text is close to the Masoretic tradition with some Samaritan-type readings, which is exactly the kind of witness that lets scholars check both traditions against a third.

A fair refinement. 4QNum-b is a mixed text that sometimes agrees with the Samaritan tradition and sometimes with the Septuagint against the Masoretic. That is a real finding, and stating it plainly strengthens the case rather than weakening it. The narrative and legal content of Numbers is stable across all of them; the differences are the kind textual critics use to reconstruct the earliest form of the text, which is what a recoverable tradition looks like.

That is accurate. 4Q41 is an excerpt collection, probably for devotional use, and its Decalogue blends the Exodus and Deuteronomy forms. The honest description is the oldest surviving text of the Ten Commandments in an excerpted and slightly harmonized form. The harmonization itself is telling: scribes working from both books presupposes that both were already fixed enough to harmonize, which supports rather than undermines the stability of the underlying texts.

Yes, and it is one of the best-known substantive variants, where the Scrolls and the Septuagint preserve the older reading and the Masoretic scribes apparently altered it. Modern translations such as the ESV now follow the Scrolls here. This is textual criticism doing its job in public: the difference is visible, can be studied, and has been corrected in the Bibles people read. A hidden corruption would leave no such trail. Faithful transmission includes preserving the evidence of where the text once differed.

Correct, and this is a case where Qumran may preserve an older arrangement, since Josephus also seems to know the altar episode in that position. With only two copies the base is thin, and Joshua is one of the books where the text form genuinely varied in this period. It does not undermine the claim that the book’s content was preserved: the same events appear in both, in a different order at one point, and the difference is on the table for anyone to weigh.

It may mean that, and the variant is of real interest to scholars studying how the book took shape. The claim is not that no word ever changed but that the transmission was faithful and its variations are recoverable. A missing paragraph in one early copy, detectable by comparison, is a variant scholars can weigh, which is the opposite of an undetectable corruption. Faithful transmission includes preserving the evidence of where the text once differed.

The list is complete on purpose. The claim is about the whole Old Testament, so every book’s status is reported, including the thin ones and the two that are missing. Ruth’s fragments do agree with the later text where they overlap, which is a small data point in a large sample. Omitting the minor books would invite the charge of hiding gaps; reporting them, thin or not, lets you see the real shape of the evidence.

Samuel is the strongest case for the objection, which is why it is stated up front rather than buried. The Masoretic text of Samuel is widely agreed to have suffered more accidental omissions than most books, and 4QSam-a lets scholars repair some of them; the NRSV and other translations now include the Nahash paragraph. This is faithful transmission in the sense that matters: the older form survived in another line, was recovered, and the damage was reversible. The claim would be false only if the corruption were undetectable, and here it was detected.

Fair. The fragments agree with the Masoretic tradition where they overlap, and the coverage is limited; that is all that can be said. Kings also has a well-known chronological divergence between the Masoretic and Septuagint traditions that the Qumran fragments are too small to settle. The honest summary is present, fragmentary, and consistent where checkable.

Agreed on both counts; the identification of 4Q118 has been questioned, and the safest wording is probable rather than certain. This is a thin data point. The near-absence of Chronicles at Qumran is more likely a fact about the community’s interests than about the book’s existence, since Chronicles is quoted and known elsewhere in the period. It is listed here for completeness, not as support.

That is the situation for Ezra, and nothing more than present in the collection is claimed. Comparing one early witness to the medieval text is still a comparison across a thousand years; it just cannot detect variation within the Qumran period itself. The claim as a whole rests on the well-attested books such as Isaiah, Psalms, and the Pentateuch. The thin ones are listed so that you can see where the evidence is thin.

It is a real wrinkle in wording. Ezra-Nehemiah is one book in the Hebrew canon, so every book except Esther is defensible on a technicality, but a plain reader hears it as Nehemiah is there, and it is not. The accurate statement is that Nehemiah is unattested at Qumran apart from its pairing with Ezra. A small Nehemiah fragment has been reported from the antiquities market, but its provenance is disputed and the fragments once held by the Museum of the Bible proved to be modern forgeries, so nothing should be built on it.

That is one plausible explanation: Esther never mentions God and centers on a festival, Purim, that the Qumran community did not observe. It bears on the history of the canon, not on transmission, which is what this claim is about. Nobody here argues that the canon was fixed at Qumran, only that the books the community did copy were copied carefully. Esther’s later text is checked against the Septuagint and the Masoretic tradition instead.

The targum is evidence that Job was valued and read, which is what it is cited for. For textual stability the Hebrew fragments (4QJob-a and 4QpaleoJob) are the relevant witnesses, and they align with the Masoretic tradition. The targum’s differences and early ending are interesting for the history of interpretation and should not be counted toward the stability claim.

The debate between liturgical collection and alternative Psalter is real and unresolved, and wrinkle undersells it. What the scroll shows is that the last third of the Psalter was still being arranged in the first century BC while the individual psalms themselves were stable. Content was fixed; ordering was not fully fixed. That distinction matters, because it is exactly the sort of nuance a hostile reader will suspect is being hidden, and it applies to the Book of the Twelve as well.

Both are true. The early date of 4QQoh-a refutes the older scholarly view that Ecclesiastes was written in the second century BC, which is a separate point worth having. For stability, the fragment is compared forward to the medieval text and agrees closely, and that thousand-year comparison is what this claim is about.

The excerpted copies omit sections deliberately, apparently for study or liturgical use, the way a modern devotional booklet might. The portions they contain agree with the later text. Excerpting is a fact about how the book was used, not about how it was transmitted, and the two should not be confused.

A fair refinement, and it strengthens the case. 1QIsa-b’s closeness to the Masoretic Text is the better evidence for stability, while 1QIsa-a’s fuller spelling and popular features show the range of copying practice. Together they show two scribes, a century apart and working in different registers, transmitting the same Isaiah. The Great Isaiah Scroll’s showcase status comes from its completeness, not from being the best textual witness, and both things are true at once.

You can have two editions, which is what most scholars conclude: a shorter earlier form and a longer expanded form both circulated before the Christian era. The claim is about transmission, and both editions were transmitted faithfully in their respective lines. Jeremiah is the clearest case where the phrase the original text is more complicated than for most books, and saying so costs nothing. It also shows the tradition preserving both forms rather than quietly suppressing one.

Correct on both points; the honest summary is fragmentary and consistent where checkable, not broad confirmation. The Masada Ezekiel scroll, found outside Qumran, is a better witness and aligns closely with the Masoretic tradition. The Septuagint’s shorter Ezekiel remains a live question that the Judean desert finds do not settle.

Only textual stability is claimed here, and on that point the fragments are unambiguous: the Hebrew-Aramaic-Hebrew structure and the wording match the later text. On dating, the earliest Daniel fragment (4QDan-c) is usually placed around 125 BC, which some scholars argue is too soon after a 165 BC composition for the book to have reached scriptural status, while others disagree. That is a separate debate that belongs in a discussion of prophecy, and it is deliberately not smuggled in here.

Correct, and it is the same pattern seen in Psalms: the individual books are stable while the arrangement of the collection was still settling. The Qumran and Murabba’at copies attest the Twelve with the Masoretic content, and the Greek Nahal Hever scroll shows the same text being carefully revised toward the Hebrew in the first century. Order of books and wording of books are different questions, and only the second bears on the claim.

The pesher covers only chapters 1 and 2, and scholars debate whether that reflects the commentator’s interest or the state of the book. Chapter 3 is present in the Murabba’at scroll of the Twelve and in the Greek Nahal Hever scroll, so the fuller book was known in the period. The pesher’s variants in its quoted text are minor and are used in critical editions, which is what one expects from a real early witness.

Both points are fair and should stay visible. The House of David reading is accepted by the large majority of epigraphers because house of X is the standard Aramaic and Assyrian way of naming a dynasty, the parallel king of Israel appears in the same inscription, and no place called Beth-Dod is otherwise attested. On the second point, the stele is offered only as evidence for David’s dynasty, which is exactly what it shows: within about 130 years of David’s traditional dates, an enemy king recognized a Judahite royal house named after him. That is what you would expect if David existed, and not what you would expect if he was invented centuries later.

Fair in one sense: Pilate’s existence was never in serious doubt. The stone’s value is that it is a contemporary inscription rather than a later literary reference, that it fixes his correct title (prefect), and that it confirms the Gospels’ picture of a Roman governor based at Caesarea. It is a small corroboration and should be weighed as one.

All three cautions are real and should stay prominent. The strongest counter is the 2011 discovery of an ossuary inscribed Miriam daughter of Yeshua son of Caiaphas, priest of Ma’aziah, which independently confirms a priestly Caiaphas family in the right period and place. The identification with the Gospel Caiaphas remains probable rather than certain, and that is how it should be presented.

The rules do need dates attached. The Masoretes compiled letter and word counts and noted the middle letters of books, and those counts survive in the margins of medieval manuscripts, so the core of the claim is documented. The rule that a whole scroll is discarded for a single error is a Talmudic-era standard for Torah scrolls and should be described that way, not as a general practice from the beginning. These rules explain the stability of the medieval text; the earlier stretch from the authors to Qumran is known only from the manuscripts themselves, which show careful copying alongside real variation between text types.

Ehrman’s own textbook, the standard one in the field, says the number of variants that are both meaningful and viable is small and that the essentials are not in doubt. The places that matter most, where a teaching could hinge, are few and are discussed openly in the footnotes of any modern Bible. Uncertainty about whether a verse reads Lord or Lord Jesus is real, and it is not the same as uncertainty about what the New Testament teaches.

The other gospels are real and worth reading, but most are considerably later, second century and after, and many carry a clear theological program the churches did not recognize as apostolic. The four canonical Gospels are the earliest and the most widely attested. Recognition tracked age and reach, not political victory. You can read Thomas and see for yourself how different in kind it is.