Papyrus 52, the Rylands fragment of John's Gospel showing John 18
|

300,000 Variants and No Originals

Anyone who has followed arguments about the New Testament has probably encountered the number. There are, we are told, somewhere between 300,000 and 400,000 variants in its surviving Greek manuscripts, more variants than there are words in the New Testament itself. None of the original manuscripts survives, either. For some, this proves that centuries of copying have left us unable to know what the authors wrote.

The usual reply is that most variants are spelling mistakes or slips of the pen, so the number tells us little about the reliability of the text. Both arguments begin with a statistic whose meaning is rarely explained. Where did the number come from, what is being counted, and what do the scholars who work with these manuscripts actually conclude from it?

The story begins in 1707 with John Mill, a fellow of Queen’s College, Oxford, who spent thirty years preparing a Greek New Testament and published it a fortnight before his death. His printed text was essentially a reprint of the standard 1550 edition. Beneath it, however, Mill supplied an apparatus recording alternative readings drawn from fewer than one hundred Greek manuscripts.

A later scholar, Gerard von Maestricht, estimated that the apparatus contained 30,000 variants. That estimate, often attributed directly to Mill, caused considerable alarm. Richard Bentley answered with an observation that remains central to the discussion: the number of variants rises as more manuscripts are compared.

Every additional copy provides another opportunity to encounter a different reading. A tradition represented by ten thousand copies may produce far more distinct variants than one represented by ten, even if its scribes were generally more careful. The raw total tells us how many differences have been recorded, but little about the likelihood that a particular word has been lost or changed.

Peter Gurry, then a textual critic at Cambridge, traced the estimates that followed Mill and asked how their authors arrived at them. The history is not especially reassuring. In 1861, Frederick Scrivener suggested that Mill’s 30,000 must have grown at least fourfold, to 120,000, without explaining why fourfold was the right multiplier. Philip Schaff offered 150,000 in 1883. B. B. Warfield reached 180,000 to 200,000 in 1889 and at least explained his method, although it introduced another problem: he counted occurrences of a reading rather than distinct readings. If a hundred manuscripts shared one reading against his standard text, they contributed a hundred variants to the total.

Later estimates rose to 250,000 in 1934, when Louis Pirot and Léon Vaganay offered that figure. Pirot was also the first in Gurry’s survey to point out that the number probably exceeded the words in the New Testament. Estimates reached 250,000–300,000 by 1951. Kenneth Clark’s 1966 estimate of 300,000 drew on new collation work on Luke and remained influential for decades. Eldon Epp later offered what he called a “wild guess” of 400,000 to 750,000. The familiar figure of 400,000 gained a wider audience through a bestselling 2005 book that attributed it to unnamed scholars. Gurry found that estimates had repeatedly circulated without a clear account of who made them, how they were calculated, or what counted as a variant in the first place.

Gurry proposed an estimate with an explicit definition and a method readers could examine. For his purposes, a variant is a word or sequence of words in one Greek manuscript that differs from the reading in another manuscript at the same place, excluding spelling differences and variations in the abbreviation of sacred names. The readings are compared with one another rather than with a supposedly correct printed edition. Where five readings compete, all five count toward the total, including whichever reading an editor eventually judges to be original.

Gurry drew on detailed manuscript comparisons for three passages: John 18, with 3,058 variants across 1,659 manuscripts; Philemon, with 1,185 across 572; and Jude, with 1,694 across 560. Each yielded between three and a half and four distinct variants per word of Greek text. Extrapolated across the 138,020 words of the standard Greek edition, the samples suggested roughly 488,000 to 534,000 variants. Gurry also tested a second way of extrapolating the data and rejected it as inflated. His rounded estimate was about 500,000 in Greek manuscripts alone, excluding ancient translations and quotations by church fathers.

Across Gurry’s three samples, just over half of the variants were singular readings, found in only one of the collated manuscripts. Some may reflect mistakes made by individual scribes; others may be readings that failed to spread or whose other witnesses have not survived. Roughly 31 percent of the variants in those samples were nonsense readings, with John 18 reaching nearly 45 percent. These figures overlap substantially: many nonsense readings are also singular. A large share of the recorded differences therefore consists of readings that no editor would seriously consider printing as the original text.

The more useful denominator is also much larger than the 138,020 words in a printed New Testament. Those variants arose across the combined words of hundreds or thousands of handwritten copies. In Gurry’s calculations, the manuscripts of John 18 introduced roughly one distinct variant for every 430 words copied, while the rates for Philemon and Jude were about one for every 150. His conclusion is that the large total reflects “the frequency with which scribes copied more than their infidelity in doing so.”

Nor does each new manuscript add variants at the same rate. Because later Byzantine copies are often closely related, the number of newly discovered readings grows more slowly as more of them are collated. Wasserman’s collation of Jude included more than three times as many witnesses as an earlier critical edition but produced fewer than twice as many variants.

Kurt and Barbara Aland approached the evidence from another direction. Their handbook describes the tenacity of the New Testament’s textual tradition: once a reading entered circulation, it often persisted somewhere. That helps explain why differences accumulated, but it exists alongside extensive agreement. The Alands compared seven major critical editions made by scholars of different schools and found complete agreement, apart from spelling, in 4,999 of the New Testament’s 7,947 verses, or about 63 percent. This is a comparison of edited texts, not a claim that every surviving manuscript agrees in those verses.

The Alands also observed that the Byzantine and Alexandrian forms of the text, often presented as sharply opposed, agree in perhaps 80 percent of their wording. Some disagreements remain significant, including readings that affect the interpretation of particular passages. The manuscript tradition contains both a substantial shared text and differences that have to be judged individually. Neither the size of the variant count nor the extent of agreement can settle those judgments by itself.

Then there is the second part of the familiar charge: none of the original manuscripts survives. That is true. The absence of an author’s own copy is familiar throughout the study of ancient literature, and the surviving evidence does not let us inspect the original page.

Textual critics have become more precise about the point their methods can reconstruct. Gerd Mink called it the Ausgangstext, usually translated as the initial text: “the reconstructed form of the text from which the manuscript transmission started.”

The scholars responsible for the Editio Critica Maior, a critical edition of the Greek New Testament produced at the University of Münster, regard their reconstruction of the Catholic Letters as a hypothesis about what the original authors wrote. They believe there is little reason to assume a substantial gap between the reconstructed text and the authors' wording, although the identification cannot be established with certainty.

The label marks the limit of the reconstruction without assuming that every reading printed in a modern edition must have stood on the author’s page. Critics still produce continuous Greek texts and argue over individual readings; the absence of an autograph has not made that work impossible. It does leave questions that the surviving manuscripts cannot always answer.

The age of those witnesses brings its own uncertainty. Consider P52, the small fragment of John 18 now held by the John Rylands Library in Manchester. Its first editor, Colin Roberts, dated its handwriting in 1935 to the first half of the second century AD. The Alands later gave its date as approximately 125 AD, with a possible range of about twenty-five years on either side, and reported an earlier scholarly tendency to treat that date as something close to the latest plausible limit.

On such a dating, the fragment could stand within a generation or two of the Gospel’s composition. But handwriting is dated by comparison with other surviving examples, and the same letter forms can remain in use for decades. Brent Nongbri examined the dated documents used to compare P52, including examples from the later second century and one dated to 184 AD. He argued that an honest range must allow for the later second and early third centuries as well. Other scholars have continued to favor earlier ranges, including a proposed date of 125–175 AD. A date of approximately 125 AD may be defensible; treating it as a precise measurement gives the fragment more chronological weight than paleography can bear.

P52 is only one part of the early manuscript evidence. The Alands cataloged other early papyri, including P46, a substantial witness to Paul’s letters commonly dated to around 200 AD. They also described the text preserved in P52, so far as its few surviving lines permit a comparison, as a normal form of John’s text rather than an unexpectedly different version. Nongbri’s criticism narrows what the fragment can prove about when John’s Gospel was already circulating; it does not, by itself, show that the surviving wording of John is unreliable.

Those copies also resist the simple family trees used for many other ancient texts. The classical genealogical method associated with Karl Lachmann identifies relationships through distinctive shared errors: if two manuscripts contain a mistake unlikely to have arisen independently, that may point to a common ancestor.

New Testament manuscripts are unusually difficult to arrange this way because scribes often corrected one copy against another. Codex Sinaiticus, the fourth-century biblical manuscript, contains an estimated 23,000 corrections made by different hands. A later copyist using a corrected manuscript could encounter readings from more than one textual line, leaving descendants with mixed ancestry. Across centuries of copying, the same problem recurs. A manuscript might preserve an older reading at one verse and a younger reading at the next.

The Coherence-Based Genealogical Method, developed by Mink and colleagues at Münster, attempts to account for this mixture. It compares patterns of agreement among witnesses and tests their possible relationships at individual places where readings differ. Editors judge which readings could have given rise to which others; computer tools then help them see how those decisions fit the broader network of witnesses.

The method can trace relationships that do not fit a single, clean family tree, although its results still depend on editorial judgments. Its use in the Catholic Letters led to thirty-four changes in the printed Greek text. In Jude 5, the reading now identifies Jesus as the one who saved a people from Egypt, where the previous edition read “the Lord.” In 2 Peter 3:10, the editors printed that the earth and its works “will not be found,” a reading supported by some ancient translations but not attested in the Greek manuscripts known to them. These are real editorial choices with consequences for interpretation, even if most of the thirty-four changes are comparatively minor.

The method also led the editors to question the older division of manuscripts into Alexandrian, Western, and Byzantine text-types. Those categories did not reproduce the detailed relationships found across the collated readings, although the Byzantine tradition continued to receive distinctive treatment. The difficulty of drawing neat families had already troubled scholars such as Bruce Metzger, who traced how the proposed “Caesarean text” became harder to sustain as new papyrus evidence entered the discussion.

Doubt at particular points is now marked differently as well: the edition of the Catholic Letters discussed by Wasserman and Gurry used a diamond where its editors declined to decide between readings, marking forty-three passages compared with thirty-two formerly marked by brackets. Greater precision about relationships has not removed uncertainty from the text, and the editors have made some of that uncertainty more visible.

The problem with citing 500,000 as an argument is that it treats every recorded difference as equivalent. A copying mistake found in one manuscript and a disagreement that affects the wording of Jude 5 do not carry the same weight, although both contribute to the total. Nor does the absence of the autographs make every contested word impossible to recover.

The surviving copies contain long stretches of agreement, a great many small differences, and some passages whose wording remains genuinely uncertain. Textual critics can reconstruct much of the New Testament with confidence while acknowledging where the evidence leaves them divided. The serious argument is over those particular readings. A count of 500,000 does not tell us how many words are beyond recovery.

Sources

Peter J. Gurry, “The Number of Variants in the Greek New Testament: A Proposed Estimate,” New Testament Studies 62 (2016): 97–121.

Peter J. Gurry, “How Your Greek New Testament Is Changing,” Journal of the Evangelical Theological Society 59 (2016): 675–689.

Tommy Wasserman and Peter J. Gurry, A New Approach to Textual Criticism: An Introduction to the Coherence-Based Genealogical Method (SBL Press, 2017).

Kurt Aland and Barbara Aland, The Text of the New Testament: An Introduction to the Critical Editions and to the Theory and Practice of Modern Textual Criticism, 2nd ed. (Eerdmans, 1989).

Bruce M. Metzger, Chapters in the History of New Testament Textual Criticism (Eerdmans, 1963).

Brent Nongbri, “The Use and Abuse of P52: Papyrological Pitfalls in the Dating of the Fourth Gospel,” Harvard Theological Review 98, no. 1 (2005): 23–48.

Pasquale Orsini and Willy Clarysse, “Early New Testament Manuscripts and Their Dates: A Critique of Theological Palaeography,” Ephemerides Theologicae Lovanienses 88, no. 4 (2012): 443–474.

Similar Posts

Leave a Reply