Repeat the Method, Lose the Text

This is Part 8 of “By What Standard Is This the Text? A Presuppositional Examination of Modern Textual Criticism.”

The Final Test

The series began with a question asked at the end of the second debate between Jeffrey Riddle and James White:

What is your objective, universal, and absolute standard by which you evaluate the Greek manuscripts and their variants?

White answered, “Well, there is no such thing, obviously,” and then referred to several criteria. The seven articles since that answer have shown why criteria are not an ultimate criterion. The accepted canons can conflict. Editors must assign their relative weight. Missing history is supplied by conjecture. The result is provisional, while the church's use of Scripture requires a settled and finally appealable text.

One practical experiment gathers the problem into a single test:

Repeat the method.

Give competent editors the witnesses, the collations, the accepted canons, the software, and freedom to exercise reasoned judgment. Ask them independently to reconstruct the initial or original text. If the same method can responsibly produce another main text, then the method has not uniquely identified the text. It has produced one permitted result among others.

That does not prove that every word in the result is false. It refutes the claim that the result's unique authority has been established by the process that produced it.

What Repeatability Means Here

Textual criticism is a historical discipline, not a laboratory experiment in which an autograph can be placed repeatedly under controlled conditions. “Repeat the method” therefore does not demand that every judgment be mechanical or that all qualified persons must agree before knowledge is possible.

It asks a question fitted to the claim. If an edition is offered merely as a useful scholarly hypothesis, variation among repeated attempts is expected. If it is offered to the whole church as the verbal form of God's final authority, the method must supply a non-arbitrary reason why this result, rather than another result permitted by the same evidence and canons, binds the conscience.

The stronger the authority claimed for the output, the stronger the demand upon the rule that identifies it.

Suppose two editors agree that external and internal evidence must be weighed. Both consider age, geographical spread, coherence, scribal habits, authorial style, difficulty, and explanatory power. At one variant, the first gives greater weight to an early witness and a harder reading. The second judges that the reading arose by error and gives greater weight to broad coherent transmission. Both can state reasons recognized within the field.

The method does not contain a universal numerical rule assigning the decisive weight. The disagreement ends in trained judgment. When that judgment changes, the main text changes.

The process is repeatable. The output is not uniquely entailed.

The Blind Reconstruction Challenge

The issue could be made visible by a controlled scholarly exercise.

1. Select a substantial set of variation units without revealing the reading printed in the current standard edition.
2. Give several independent teams the same transcriptions, dates, versional evidence, patristic evidence, and CBGM data.
3. Require each team to state its canons, prior assumptions, and reasons before seeing another team's conclusions.
4. Prevent appeal to the authority of the existing main text or to the knowledge that one reading is currently preferred.
5. Compare the resulting texts and stated levels of confidence.

If the method independently identifies one text, convergence should be unusually high at the very places selected because they are difficult. If the teams diverge while remaining within accepted method, the underdetermination has been demonstrated.

One can then repeat the experiment with new teams or altered but defensible judgments about local stemmata. If the text changes again, no appeal to “the method” can turn one output into an invariant standard. The method is a field of judgment, not a verbal rule that entails one edition.

This proposal is not a taunt. It is the sort of transparency the authority claim requires. The reader should be able to distinguish conclusions compelled by evidence from conclusions selected among live possibilities.

The Published Record Has Already Run the Experiment

No hypothetical trial is actually necessary to establish the basic point. The history of critical editions is a history of repeated reconstruction.

The German Bible Society explains that the twenty-eighth Nestle-Aland edition incorporated the decisions of the Editio Critica Maior in the Catholic Epistles and thereby made more than thirty changes to the main text, some significantly affecting meaning. Those changes replaced readings printed in the prior standard edition.

The same body now previews the twenty-ninth edition. It will adopt the ECM's textual changes in Mark, Acts, and Revelation, and its product description says that the new main text differs from NA28 at well over one hundred places.

This is not hostile characterization. It is how the publisher advertises scholarly progress.

The changes may be improvements according to the method. That concession strengthens the reductio. If the newer method and evidence correct more than one hundred places in the present standard text, then the present standard text was not uniquely secured by the earlier application of critical principles. If the forthcoming text remains open to later ECM work, NA29 will not become invariant merely by replacing NA28.

The editions form a sequence of responsible outputs:

• A former best text
• A present best text
• A forthcoming improved text
• Later texts anticipated as the ECM proceeds

The inspired words do not pass through those stages. Editorial confidence does.

The CBGM Does Not Make the Textual Decision

The Coherence-Based Genealogical Method may appear to answer the repeatability problem by placing judgment upon a more exact computational foundation. It analyzes patterns of agreement and helps editors explore relationships among readings and witnesses across large quantities of data.

Its own official description prevents the exaggerated claim. Gerd Mink's introduction states that a large proportion of witnesses is missing, that relations among extant states can only be a genealogical hypothesis, and that the method does not reconstruct detailed historical processes. Most decisively, it says that the CBGM “makes no textual decisions.” It presents an image of the tradition and tests the plausibility of philological hypotheses.

The newer INTF introduction likewise says that the non-extant initial text must be reconstructed by applying internal and external criteria passage by passage. The tool supplies structured evidence. The editor still determines the priority of readings.

This is not a defect in the software. It is a limit upon what software can do. A diagram cannot provide the ultimate standard absent from the premises supplied to it. If human decisions shape local stemmata and the resulting coherence leads editors to revisit those decisions, the process is iterative rather than self-authenticating.

Repeat the iteration with different defensible starting judgments, and the method may display another plausible flow.

Revelation Makes the Point Concrete

The INTF's account of the ECM edition of Revelation is unusually candid. It says that in some textual problems the editors found no conclusive evidence for either reading. They therefore printed both readings on equal footing and gave the reasons so readers could decide which they preferred.

That is commendable scholarly honesty. It is also a confession that the reconstruction has not identified one reading at those places.

At Revelation 21:6, the ECM changed the previously published wording and stated that the decision would affect new translations. Whatever judgment one reaches upon that particular variant, the sequence is plain:

1. One Greek reading stood in the accepted main text.
2. The method was reapplied with a fuller project and new judgments.
3. Another reading replaced it.
4. Translations based upon the project are expected to change with it.

The same process that authorized the former wording now authorizes its replacement. Without an invariant external control, the existence of a later preference disproves the unique entailment of the former one but does not grant finality to the latter.

In What Sense the Critical Text Is Refuted

Precision matters. A method can arrive at true readings even when it cannot uniquely warrant its whole output. The critical text agrees with the received text in most places. It contains the books of the New Testament, proclaims Christ, and communicates countless true propositions.

The refutation concerns its identity and authority as the critical text.

If the critical text means “the current editorial committee's best scholarly reconstruction,” repeatability does not refute its usefulness. A later reconstruction may indeed be better by the committee's measures.

If the critical text means “the uniquely identified verbal form of God's inspired and finally binding New Testament,” then non-unique repeatability refutes the claim. The method permits another text without providing an objective, universal, and absolute rule that proves why one permitted output is God's Word at the disputed points and the other is not.

An edition cannot acquire divine finality from a process that withholds human finality.

This is why the argument reaches the critical text itself, not merely a few weak explanations offered on its behalf. The edition's warrant is the process that produced it. When the process is shown not to entail the edition uniquely, the edition loses the warrant claimed for it as a whole.

Again, the conclusion is not that every reading is false. It is that the edition has not been established as the single authentic textual standard by its generative method.

Why Derived Translations Are Refuted in the Same Respect

A translation cannot rise above the textual identity of its base.

Suppose a translation accurately renders NA28. Where NA29 changes the Greek, the translation faces a choice. If it retains the older rendering, it no longer follows the newest result of the method. If it revises the rendering, it acknowledges that the prior edition placed another wording before the church. If editors remain divided, a footnote distributes the uncertainty to the reader.

The translation may remain readable, useful, and doctrinally rich. Those qualities do not settle the textual question. At the contested places it inherits the provisional character of its source.

Thus a successful repetition of the critical process refutes any derived translation in the same precise respect: it refutes the translation's claim to rest upon a uniquely established and finally authoritative base text. It does not prove that the translation contains no Word of God, that every rendering is wrong, or that no Christian may profit from reading it.

The distinction keeps the argument strong without making it reckless.

Four Familiar Replies

“The differences are small.”

Quantity does not answer identity. One disputed word is enough to require a rule for identifying that word. Small variation may secure broad doctrinal stability; it does not produce final verbal authority.

“All sciences revise their conclusions.”

That is true and precisely the point. A revisable scholarly product should be treated as a revisable scholarly product. It should not be made the constituting ground of the church's unrevisable rule. Scripture judges sciences; a science does not reconstruct the judge into existence.

“Scholarly consensus provides stability.”

Consensus is testimony, not an ultimate standard. The movement from NA28 to NA29 is itself a change in consensus. A majority conclusion remains corrigible and cannot bind merely because experts hold it.

“The confessional position also makes judgments.”

It does. It compares editions, corrects printing errors, studies manuscripts, and reasons about hard readings. The difference is the order and office of judgment. Confessional criticism receives the authentic Hebrew and Greek Scriptures as a providential whole and uses evidence ministerially within that received identity. Restorationist criticism begins with divided witnesses and constructs the text that will thereafter be treated as authentic.

One judges witnesses by the rule. The other builds the rule by judging witnesses.

The Four-Part Verdict

The series can now state Bahnsen's analysis in one view.

Arbitrariness: The method lacks an invariant rule governing the relative weight of its accepted canons. At difficult places, learned judgment ends the appeal. Rival conclusions remain possible.

Inconsistency: The canons can reverse direction; terms such as preserved, original, initial, and available shift; critical Christians treat the text as settled in worship and provisional in method; professed presuppositionalism makes an exception for the identification of Scripture.

Consequences: The perfect text becomes an unavailable ideal, the available text a revisable reconstruction, certainty a practical posture built upon graded probability, ecclesiastical dependence a chain terminating in editorial expertise, and translations heirs of their base text's uncertainty.

Preconditions of intelligibility: The discipline presupposes an ordered creation, stable causation, meaningful language, universal logic, reliable testimony, binding truth, moral duties, a determinate original, providential history, and a publicly recognizable rule. Those conditions are grounded in the triune God and his self-attesting Word, not in autonomous criticism.

The critical method uses the capital of the Christian worldview to construct a tribunal in which the Word's verbal identity is decided without the Word's rule.

The Text Is Received Before It Is Corrected

The alternative is not ignorance, hostility to Greek, or inspiration of the Authorized Version. The final authority is the preserved Hebrew and Greek Scripture. Translations, including the Authorized Version, are faithful and valuable insofar as they render that text.

Neither is the alternative the infallibility of Erasmus, Stephanus, Beza, or any printer. Human editions and copies may contain faults. Ministerial textual work remains necessary.

The difference is constitutional. The church does not begin with a heap of undecided readings and commission a guild to create its rule of faith. It receives the Word God inspired, preserved, and entrusted to his people. It then uses every lawful historical and linguistic tool to guard, compare, print, and translate that Word faithfully.

The preserved text judges the witnesses. The witnesses do not constitute the preserved text.

That order makes collation useful, correction possible, translation accountable, and final appeal meaningful. It also answers the question James White said had no answer.

There is an objective, universal, absolute, and invariant standard by which manuscripts, variants, canons, editors, editions, translations, and this very series must be judged.

That standard is the Word of God.

Notes and Further Reading

• Chris Thomas, “Argument Analysis Checklist”.
• Chris Thomas, “Why New Testament Textual Criticism Can Never Arrive at Truth”, presenting the fuller reconstruction thought experiment.
• German Bible Society, “The 28th Edition of the Novum Testamentum Graece”, reporting more than thirty main-text changes in the Catholic Epistles.
• German Bible Society, “Preview: The 29th Edition of the Nestle-Aland”, and the NA29 product description, reporting the incorporation of ECM changes and well over one hundred differences from NA28.
• Gerd Mink, “The Coherence-Based Genealogical Method—What Is It About?”, especially its statements concerning missing witnesses, genealogical hypotheses, iterative revision, and the fact that the CBGM makes no textual decisions.
• Institut für Neutestamentliche Textforschung, “The Coherence-Based Genealogical Method”.
• Institut für Neutestamentliche Textforschung, “Publication of Four New Volumes of the Editio Critica Maior”, describing the Revelation text, equally weighted readings in unresolved places, and expected effects upon translations.
• The Orthodox Presbyterian Church, Westminster Confession of Faith 1.4–10.