Introduction
The previous two papers surveyed the principal bodies of fragmentary evidence — the Hebrew witness to the Old Testament and the Greek witness to the New — and assessed what each does and does not establish. Those surveys assumed, but did not closely examine, a set of practical disciplines by which fragmentary evidence is actually read, reconstructed, and reported. The present paper turns to those disciplines.
The work of fragmentology is, at its most practical, the work of moving from a piece of physical material to a published claim about a text. The path from artifact to claim passes through several stages, each of which admits its own characteristic errors and offers its own opportunities for honest service. A method paper is, in a real sense, a paper about temptations — the temptations to see what one wishes to see, to fill silence with one’s own voice, to dress conjecture in the language of certainty, to crown the critic’s judgment in the place of the church’s reception. The aim of the present paper is to set out, in plain terms, the disciplines by which those temptations are resisted.
The structure follows the natural flow of the work itself. First, the move from artifact to reading: how the worker actually arrives at a transcription. Second, the move from reading to reconstruction: how, when, and with what restraint the worker proposes to fill what is missing. Third, the weighing of internal and external evidence: how readings are assessed against one another. Fourth, the reporting of findings to those who are not specialists: how the worker tells the truth to the church without overstating either the certainty or the uncertainty of what he has found.
I. From Artifact to Reading
A. Imaging, transcription, and collation
The work begins with the physical artifact. In earlier generations, this meant traveling to the holding institution, sitting before the manuscript with such tools as a magnifying glass and a careful eye, and producing a transcription by hand. In the present generation, much of the work begins with high-resolution digital images, often produced under several lighting conditions and sometimes by multispectral imaging that recovers ink no longer visible to the unaided eye. The conditions of the work have improved; the underlying discipline has not changed.
A transcription is the worker’s representation, in modern letters and conventions, of what is visible on the artifact. A good transcription is faithful to what is there. It does not silently regularize spellings, complete abbreviations, or supply missing letters; or, if it does any of these things, it marks them clearly so that the reader can distinguish what was on the artifact from what the transcriber has supplied. The discipline of transcription is, at its heart, the discipline of separating one’s own contribution from the artifact’s testimony.
Collation is the comparison of one witness’s reading with the readings of others. A collation may be made against a chosen base text, or among a set of witnesses pairwise; the choice of method depends on the purpose. What matters, for the present paper, is that collation is the procedure by which the agreement and disagreement of witnesses is made visible. Without careful collation, claims about textual relationships rest on impression rather than evidence; with careful collation, those claims rest on an examinable record.
The honest worker keeps his transcriptions and collations in a form that others can check. This is not a procedural nicety; it is a substantive commitment. A transcription that no one can verify is a transcription whose accuracy must be taken on the worker’s authority alone, and the discipline of this series, formed by the hermeneutics of humility set out in Paper Three, does not encourage taking any worker’s authority alone in matters that can in principle be checked.
B. Distinguishing what is visible, what is reconstructed, and what is conjectured
A central discipline in fragmentary work is the discipline of triple distinction — already named in Paper Three but worth restating here in its concrete application to transcription.
What is visible is what the surviving material plainly shows. A letter that is fully formed, present, and unambiguous is visible. A word that can be read without doubt is visible. The category of the visible is the category of direct attestation, and it is the category in which confidence is highest.
What is reconstructed is what the worker proposes for portions of the artifact where the evidence is partial — a damaged letter that retains traces sufficient to identify it, a word at the edge of a tear where part of the letters survive and part do not, a line whose beginning or end is lost but whose middle is preserved. Reconstruction is a reasoned inference from partial evidence to a probable reading. It is not the same as direct attestation, and it must not be presented as such.
What is conjectured is what the worker proposes for portions where the artifact itself does not provide a basis for reconstruction — a missing line whose absence is total, a damaged region where no traces remain to constrain what was there, a word for which the surviving evidence is consistent with multiple possibilities. Conjecture is sometimes appropriate, but it is conjecture, and the line between it and reconstruction must be kept visible.
The honest worker keeps these three categories distinct in his own mind and in his published work. He does not allow visible readings to be eroded by the fashions of reconstruction; he does not allow reconstructions to harden into visible readings by repetition; he does not allow conjectures to be cited later as though they were reconstructions or readings. Each category has its proper place; each must be marked for what it is.
C. Bracket conventions and why they matter for the non-specialist reader
The discipline of triple distinction is supported, in published transcriptions, by a set of conventional markings — chiefly square brackets and related symbols — that indicate to the reader exactly what kind of evidence underlies each portion of the printed text. The specific conventions vary somewhat among editions, but the principle is constant: the printed page makes visible, by typographic means, the difference between what is on the artifact and what the editor has supplied.
For the specialist, these conventions are second nature. For the non-specialist — the pastor, the seminary student, the interested believer — they can be opaque, and worse, they can be invisible. A reader who does not know the conventions may take a printed transcription as a uniform text and miss the indications that part of it is reconstructed and part is conjectured. The same reader may then quote what is, in fact, a conjectural reading as though it were the unambiguous testimony of the manuscript.
The remedy for this is twofold. Editors should explain their conventions plainly, in language that a non-specialist can understand, in introductions that the non-specialist will actually read. And teachers — pastors, professors, and writers of expository works — should educate their hearers to look for the conventions and to weigh transcribed material accordingly. A small effort in this direction would yield large returns in honest understanding of what fragmentary evidence does and does not show.
II. Reconstruction with Restraint
A. The legitimate uses of conjecture
It would misrepresent the discipline to suggest that all conjecture is illegitimate. There are circumstances in which conjecture is appropriate, useful, and even necessary, provided it is named for what it is.
The most common legitimate use of conjecture is in the reconstruction of damaged passages where the surviving evidence constrains the possibilities to a small set, sometimes to one. If the surviving traces of a word are consistent with only one Greek or Hebrew word that fits the context, and if the parallel evidence from related witnesses supports that word, then proposing that word as the reading is reasonable. The conjecture remains a conjecture; it is not yet visible reading; but it is a conjecture with substantial constraint, and reporting it as such serves the larger work.
A second legitimate use of conjecture is in the proposal of explanations for difficult or apparently corrupt readings. A worker who encounters a reading that does not fit the context may propose, as a hypothesis, that an earlier scribe misread or miscopied an underlying word, and may propose what the underlying word was likely to have been. Such proposals are properly conjectural, are properly subject to evaluation against other witnesses and against the larger evidence, and are sometimes vindicated by later discoveries.
A third legitimate use of conjecture, more limited, is in the construction of working hypotheses about the larger documents from which fragments have been separated. A worker assembling several fragments that appear to come from the same original codex may conjecture an arrangement, a number of lines per page, a relationship among the fragments, and so on, in order to test those conjectures against further evidence. Provided the conjectures are named as such and are revised when the evidence requires, this work is part of the legitimate practice of the discipline.
B. The dangers of speculative reconstruction
The dangers of conjecture, however, are real, and they have produced enough bad work in the history of the discipline to warrant sustained attention.
The first danger is the danger of confirmation. A worker who has formed a hypothesis about what a fragment must contain — for theological reasons, or for reasons of professional reputation, or simply because the hypothesis is interesting — may read traces in the artifact as confirming the hypothesis when the traces do not, in fact, do so. This kind of error is rarely deliberate. It is more often the product of expectation: the worker sees what he expects to see, and his expectation has shaped his perception of the surviving marks.
The second danger is the danger of accumulation. A reconstruction proposed tentatively in one publication is repeated in a second, cited in a third, treated as established in a fourth, and built upon in a fifth. By the time a careful reader examines the chain, the original tentativeness has been forgotten, and a chain of inference has acquired the appearance of a chain of evidence. This pattern is not unique to fragmentology, but fragmentology, with its dependence on partial evidence, is particularly susceptible to it.
The third danger is the danger of agenda. A worker with a theological or polemical interest in a particular outcome may, without acknowledging the interest, allow it to shape his reconstructions. This danger runs in every direction of theological commitment. The biblicist is not exempt from it, and the present series would be dishonest if it pretended otherwise. The only safeguard is the discipline of distinguishing between what one wishes the fragment showed and what the fragment actually shows, and of submitting one’s work to readers, including unsympathetic ones, who can check the difference.
C. Test cases of overreach in the history of the discipline
The history of fragmentology contains its share of cautionary cases, and a few patterns are worth naming without identifying particular controversies.
There have been cases in which fragments have been confidently dated, on paleographic grounds, to a particular decade or even a particular year, only for later analysis to demonstrate that the evidence supported only a much wider range. Such cases have sometimes entered popular literature in the form of dramatic claims about how early certain biblical writings can be attested, claims that did not survive careful re-examination. The lesson is not that paleography is useless; the lesson is that paleography, like other methods, has limits, and that publication of results in advance of careful peer review can cause durable harm.
There have been cases in which fragments have been identified, at first glance, with one biblical work, and then later re-identified with another, or with no biblical work at all. This kind of correction is part of the normal life of the discipline, but the corrections sometimes lag well behind the initial reports, and in the interval the initial identification has been treated as established in places it should not have been.
There have been cases in which conjectured reconstructions of damaged passages have been printed without sufficient marking and have entered the citation chain as though they were directly attested. The harm done by such cases is not that the conjectures were offered; conjecture is, as noted, sometimes appropriate. The harm is that the conjectural status was not consistently maintained, and that the line between artifact and reconstruction was allowed to blur.
In each such case, the remedy is the same: the discipline of triple distinction, the conventions of bracket marking, and the willingness to publish corrections as conspicuously as the original claims were made. None of these remedies is glamorous. All of them are necessary.
III. Internal and External Evidence
The reading of fragmentary evidence inevitably leads to the comparison of variant readings, and the comparison of variant readings inevitably involves the weighing of internal and external considerations. A brief account of these categories, and of the biblicist’s appropriate use of them, belongs in any method paper.
A. How biblicists weigh the two
External evidence concerns the manuscripts themselves — their date, their geographical distribution, their textual character, their genealogical relationships. A reading attested by many manuscripts, by early manuscripts, by manuscripts from a wide geographical range, and by manuscripts of demonstrated reliability is, on external grounds, well supported.
Internal evidence concerns the readings themselves — what kind of variant has been produced, how it might have arisen, which reading better explains the others, which reading better fits the author’s known style and the immediate context. A reading that plausibly explains how the other variants in the tradition might have arisen from it is, on internal grounds, well supported.
Biblicists, like other careful workers, weigh both kinds of evidence. There is no responsible textual judgment that ignores either. The differences among biblicist positions, including the differences sketched in the previous paper, often reduce in part to differences in how the two kinds of evidence are weighted relative to each other.
The biblicist commitments of this series do not dictate a single answer to the question of weighting, but they do impose constraints. A weighting that effectively reduces external evidence to one or two favored manuscripts, treating their readings as decisive in advance of examination, is not a sound use of the evidence. A weighting that effectively reduces internal evidence to the critic’s own sense of what the apostle would have written, treating the critic’s literary intuition as decisive, is also not a sound use of the evidence. Between these errors, a careful balance is possible, and is, in practice, what the best textual work has long sought.
B. Where canons of criticism aid, and where they substitute the critic’s judgment for the church’s reception
Over generations, textual scholarship has developed a body of canons — general principles for the evaluation of variant readings. Some of these canons are sound and useful. Others are more questionable. The biblicist’s task is to use the sound ones and to be cautious about the others.
Among the more reliable canons is the principle that, all else being equal, the reading more likely to have produced the others is more likely to be original. This principle, properly applied, accounts well for the kinds of scribal errors that any worker familiar with hand-copied texts will recognize: eye-skip, dittography, harmonization to parallel passages, expansion of pious phrases, and so on.
Among the more questionable canons is the principle, sometimes invoked in the form lectio difficilior potior — the more difficult reading is to be preferred. The principle has a kernel of truth: scribes did sometimes smooth difficult readings into easier ones, and this tendency must be reckoned with. But the principle is sometimes pressed into a form that effectively requires the critic to prefer whatever reading is harder, on the assumption that any easier reading must be a smoothing. Pressed in this direction, the principle elevates the critic’s own sense of difficulty into a textual decision, and it can be used to override broad and ancient attestation in favor of readings whose support is narrow.
The deeper concern, for a biblicist, is when canons of criticism are used not as aids to judgment but as substitutes for the church’s long reception of the text. The wording that the church has read, preached, and lived by for many centuries has a kind of standing that the canons of criticism, however refined, ought not to be permitted to override casually. This does not mean that the received wording is beyond examination; the present paper is, after all, a paper about how examination is properly conducted. It means that examination should be conducted with appropriate weight given to the historical reception of the text, and that a canon of criticism is not, by itself, a sufficient basis for setting that reception aside.
C. A humble use of critical canons that does not crown them
The right posture toward the canons of criticism is the posture of a careful workman toward his tools. The canons are useful; they are not authoritative. They embody generations of accumulated wisdom about how scribes actually behaved; they are not infallible rules. They aid judgment; they do not replace it.
A humble use of the canons, suited to the commitments of this series, would observe several disciplines. It would apply the canons consistently rather than selectively. It would weigh internal arguments against external attestation rather than allowing one to silence the other. It would acknowledge, where the canons yield no clear verdict, that no clear verdict has been yielded, rather than producing a verdict by adding emphasis to a weak argument. And it would refuse to use the canons as instruments for setting aside readings whose ecclesial reception is broad and whose attestation is substantial, in favor of readings whose appeal is chiefly that they conform to the modern critic’s literary intuition.
These disciplines are not novel. They are the long-standing habits of the best textual workers, biblicist and otherwise. Naming them here is not to claim novelty but to commend the continued practice of habits that have served the discipline well.
IV. Reporting Findings to Non-Specialists
The final stage of the work is the stage at which fragmentological findings reach those who are not specialists in the discipline. This stage is, in some respects, the most consequential, because it is at this stage that the work of the worker meets the life of the church.
A. Plain-language reporting of confidence
The first discipline of reporting is the discipline of plain language about confidence. The taxonomy proposed in Paper Three — certain, probable, possible, conjectural — was offered as a working tool for the calibration of claims. Its application in published and spoken reporting requires that the worker actually use the categories rather than collapsing them into a uniform tone of authority.
A claim that is certain may be reported with confidence, and the church should hear it as such. A claim that is probable may be reported with appropriate qualifying language, and the church should hear it as a strong but not final judgment. A claim that is possible may be reported as one option among several, and the church should hear it accordingly. A claim that is conjectural may be reported as a hypothesis, and the church should hear it as the speculation it is.
The discipline here is, in part, a discipline of vocabulary. Words like demonstrated, suggested, proposed, and speculated carry different weights, and the careful worker uses each for the kind of claim it actually expresses. A worker who reports speculation in the language of demonstration misleads his hearers; a worker who reports demonstration in the language of speculation underclaims what the evidence has actually established. Both errors are forms of inaccurate reporting.
B. Avoiding both alarmism and triumphalism
The second discipline of reporting is the discipline of avoiding two opposed temptations.
Alarmism is the tendency to report fragmentological findings in ways that destabilize the church’s confidence in Scripture without warrant. A new fragment is presented as though it overturns long-settled questions; a variant is presented as though it touches the substance of the faith; an unresolved methodological debate is presented as though the underlying text were itself in doubt. Reporting of this kind sometimes arises from skeptical commitment, sometimes from the pursuit of attention, sometimes from carelessness. Whatever its source, it does real harm to the believing reader who is not in a position to evaluate the claims for himself.
Triumphalism is the opposite tendency, and it is the temptation to which biblicist reporters are perhaps more susceptible. A new fragment is presented as though it settles questions it does not settle; an early date is asserted with more confidence than the paleographic evidence supports; the variants in the tradition are dismissed in language that overstates the case. Reporting of this kind, however well-intentioned, also does harm. It teaches the church to expect that every advance in fragmentology will be a vindication, and it leaves the church poorly prepared when an advance turns out to be more complicated than the triumphalist presentation suggested.
The right path is between these errors. It is the path of accurate reporting: confidence where confidence is warranted, modesty where modesty is owed, and willingness to revise when revision is required. Reporting of this kind serves the church best, because it tells the church the truth, and the truth is in fact good for the church to hear. The substance of the biblical text is well-established; the work of fragmentology supports rather than undermines that establishment; the small body of unresolved questions is genuinely small and does not touch what the faith requires.
A pastor handling these matters in his teaching, a writer handling them in his books, a scholar handling them in his publications — each owes to the church the same accurate reporting. None should be alarmist; none should be triumphalist; all should be honest.
Conclusion
The method of fragmentology, as set out in this paper, is a method of disciplined movement from artifact to claim. It begins with careful imaging, transcription, and collation. It maintains the triple distinction among what is visible, what is reconstructed, and what is conjectured. It uses the conventions of bracket marking to keep that distinction visible to readers. It permits conjecture in its appropriate place while guarding against the dangers of confirmation, accumulation, and agenda. It weighs internal and external evidence in balance, using the canons of criticism as aids rather than authorities. It reports its findings in plain language, with calibrated confidence, avoiding both alarmism and triumphalism.
These disciplines do not glamorize the work, but they protect it. The worker who follows them produces results that the church can use with confidence. The worker who neglects them, however gifted otherwise, produces results that the church cannot rely on, because the church has no means of knowing where in the work the worker has spoken from the evidence and where he has spoken past it.
The next paper turns from method to a particular set of cases that lie at the edges of the canonical question: fragments of apocryphal, pseudepigraphal, and unidentified material, and the questions they raise for biblicist fragmentology. The disciplines set out in this paper will be required there in particularly demanding form, because at the canonical edges the temptations both to overclaim and to underclaim are unusually strong.
