No Brute Facts: Manuscripts Do Not Interpret Themselves

Part I — The Method Bahnsen Defended. Article 4 of 24.

A parchment does not arrive with its theological meaning attached. It may display ink, letter forms, corrections, omissions, marginal notes, and a particular reading. These features are real. They are not self-interpreting.

Greg Bahnsen learned from Cornelius Van Til to reject “brute facts”—facts imagined as uninterpreted atoms whose meaning can be established without reference to God. In the Christian account, every fact is what it is because God created, governs, relates, and knows it. Human knowers never view the fact from nowhere. They receive and interpret it within basic commitments about reality and knowledge.¹

This principle is often invoked in debates over science, history, and apologetic evidence. It belongs equally in textual criticism.

The date of a manuscript is a fact. Its age does not announce the weight it should receive. A shorter reading is a fact. Its brevity does not tell us whether an author wrote it or a scribe omitted words. A difficult expression is a fact. Its difficulty does not reveal whether it generated a smoother variant or resulted from corruption. A genealogical relation is a fact established by a model. Its relevance depends upon what the model represents and what history is assumed.

Manuscripts speak through their words. They do not interpret their own place in the doctrine of Scripture.

Brute Facts and Created Facts

The denial of brute facts is sometimes caricatured as though Christians think evidence has no stable identity until theology invents one. That is the opposite of Bahnsen’s point.

A brute fact is unstable precisely because it has been detached from the God whose plan gives it identity and relation. A created fact has objective reality. The manuscript was written at a time, by a scribe, from some exemplar or combination of sources, within a history fully known and governed by God. Our interpretation may be incomplete or mistaken, but the fact itself is not fluid.

Christian objectivity therefore rests on two truths:

1. The object is determinate before our judgment. A reading belongs or does not belong to the text regardless of an editor’s vote.
2. Our knowledge is creaturely and covenantal. We know by God-given faculties, through evidence, within moral and religious commitments, never from an autonomous standpoint.

This account avoids both positivism and relativism. Facts do not interpret themselves, and interpreters do not create the facts.

Observation Is Already Selective

Even the collection of evidence involves prior questions. A researcher must decide what counts as a witness, what unit of variation to record, how to distinguish a correction from the first hand, which versional evidence is sufficiently precise, and how to represent uncertainty.

These decisions can be made responsibly. They are not necessarily sinister or consciously theological. They are nevertheless shaped by a conception of the object.

Consider a palimpsest. The scholar may recover an erased biblical text beneath a later writing. Is the recovered layer one witness among many? Does its early date grant presumptive priority? Does isolation from ecclesiastical use increase independence or decrease representative value? Does a distinctive reading preserve antiquity or idiosyncrasy?

The ultraviolet image does not answer. It supplies data for an argument.

The same is true when building an apparatus. The apparatus is an extraordinary tool, but it is not a transparent photograph of “all the evidence.” It selects readings, witnesses, and degrees of detail. Its sigla encode editorial decisions. Scholars who consult it receive evidence already organized by a theory of relevance.

None of this discredits the apparatus. It reminds the reader to ask what work the organization is doing.

Age Is Not a Verdict

Older manuscripts are important because they stand closer in time to the first writing. All else being equal, fewer generations of copying may mean fewer opportunities for alteration. That is a rational historical consideration.

But all else is rarely equal.

An early witness may have been poorly copied. A later witness may descend through a careful line from an excellent exemplar. An old manuscript may survive because it was set aside in a dry climate; a heavily used accurate manuscript may wear out. A reading found in later copies may be attested much earlier in a version or patristic citation. A localized early form may not represent the text received more broadly.

Therefore the fact “Manuscript A is older than Manuscript B” does not entail “A’s reading is original.” The entailment requires a transmissional account.

A providential doctrine adds further questions. Does God’s preservation imply only the survival of enough ancient fragments for modern restoration, or the public availability of the authentic text in the church’s continuous use? How should long reception be weighed against material age? Those questions cannot be answered by radiocarbon dating.

Confessional Bibliology does not make lateness a mark of truth. It denies that age alone is a self-authenticating canon.

Shorter and Harder Are Not Oracles

Two familiar internal considerations are often summarized as preferences for the shorter and harder readings. Both reflect real observations about scribal behavior.

Scribes may expand a text by harmonization, explanatory gloss, liturgical addition, or accidental repetition. They may also omit by eye-skip, similar endings, fatigue, doctrinal discomfort, or deliberate abbreviation. Brevity can result from originality or loss.

Scribes may smooth grammar, resolve an apparent difficulty, or conform an unusual expression. They may also introduce nonsense by mistake, reproduce a damaged exemplar, misunderstand an abbreviation, or create a harder reading unintentionally. Difficulty can result from originality or corruption.

Accordingly, responsible criticism does not apply slogans mechanically. It asks which reading best explains the origin of the others within a specific context. That explanatory judgment draws upon language, authorial style, known scribal habits, parallel passages, and the external distribution of witnesses.

The judgment also rests upon assumptions about authorial coherence and textual history. A critic who expects early diversity may explain the evidence differently from one who expects a stable ecclesiastical text. A critic who treats doctrinally fuller readings with suspicion may assign motives not applied symmetrically to omissions. A defender of a received text may reverse the bias and dismiss every shorter reading as mutilation.

No canon substitutes for argument. No argument is worldview-free.

Genealogy Is a Model of Relations

Genealogical reasoning seeks to avoid counting manuscripts as though every surviving copy were an independent vote. If ten manuscripts descend from one recent exemplar, their agreement may represent one ancestral reading rather than ten independent streams. This is sound in principle.

The challenge is that manuscript relationships are rarely simple family trees. Scribes may copy from multiple exemplars. Texts may be corrected against another tradition. A witness may be related differently in different passages. Lost intermediaries must be inferred. Modern methods therefore construct models of textual flow from shared readings and patterns of agreement.

Such models can reveal genuine relationships. They also embody definitions and thresholds:

• Which similarities are genealogically significant?
• How is contamination represented?
• What prior probability is assigned to one reading producing another?
• Does the model seek a historical manuscript, a textual state, or an editorial initial text?
• How are unattested ancestors inferred?

The output can look mathematically impersonal. Mathematics orders the premises it is given. It does not provide a theology of the text.

This is especially important for contemporary claims that a computer has “proved” a reading. A computational tool may process more relationships than a person can track and expose patterns otherwise missed. Its result remains an inference within a model. The model does not become religiously neutral because it is complex.

Distribution and the Meaning of Catholicity

A reading found across distinct regions and textual streams may have strong external support. Wide distribution can suggest antiquity because a reading must have arisen early enough to spread. Yet distribution is not simply counted.

Geographical labels may be uncertain. Surviving witnesses are unevenly distributed by climate and discovery. Later ecclesiastical standardization can produce broad agreement. Early local forms may survive disproportionately.

Theological interpretation enters when the church’s catholic use is considered. Historic Reformed writers did not identify truth by majority vote. They nevertheless believed God committed his oracles to his people and kept the original-language Scriptures for their use. A reading’s durable presence in the worship, preaching, controversy, and translation of the churches may therefore be providential evidence, not merely social inertia.

Reception cannot make a non-scriptural reading divine. But a method that excludes reception in principle has already decided that God’s ecclesiastical providence is irrelevant to textual identity.

Scribal Motives Require Evidence

Textual arguments sometimes attribute a variant to doctrinal alteration. Such alteration did occur. A scribe may have expanded, omitted, or changed wording for theological reasons. The possibility must not become a universal explanation.

Motive claims require evidence. A reading favorable to an orthodox doctrine may be original. A shorter reading may reflect accidental loss rather than deliberate suppression. A scribe may reproduce an inherited form without sharing any motive that first produced it.

Bahnsen’s work on self-deception offers a useful restraint here. One should not infer hidden motivation merely from an outcome that serves an interest. The same caution owed to a modern theologian is owed, proportionately, to an ancient copyist. Establish the textual event first. Classify motive only when contextual evidence supports it.

This rule protects textual criticism from polemical storytelling.

The Fact of a Variant and the Doctrine of Corruption

A variant proves that witnesses differ. It does not by itself prove that Scripture has been lost.

Historic Reformed writers distinguished a fault in a particular copy from a universal corruption of the sources. Francis Turretin acknowledged spelling differences, irregularities, and variants while denying that the Hebrew and Greek text had become so corrupted that it could no longer serve as judge. John Owen acknowledged diverse readings while arguing that the whole Scripture, as given, remained preserved in the copies.²

This distinction changes the meaning of the evidence.

Under a restorationist model, the variant is a point at which extant witnesses compete to approximate an absent original. Under a preservationist model, the variant is ordinarily a local disorder within a textual estate whose genuine reading remains available through comparison. Both models see the difference. They disagree about the state of the whole.

The variant does not announce which model is correct. Scripture, confession, and history must supply the larger account.

Bahnsen’s Appeal to Textual Results

Bahnsen believed that textual criticism had produced comforting results. Present Bibles, he argued, are substantially identical with the autographic text. The abundance and agreement of witnesses confirm that the church possesses an adequate Bible.³

This confidence is preferable to textual skepticism. It recognizes providence and refuses the claim that transmission destroyed Christian doctrine.

Yet the phrase “results of textual criticism” can hide the interpretive chain. Which witnesses were prioritized? Which canons governed? Which edition represented the result? What degree of uncertainty remained? How did those conclusions relate to Westminster’s present-tense authenticity?

Bahnsen’s doctrine of facts requires those questions. He cannot appeal to textual science as though its conclusions emerged from uninterpreted data. The same presuppositional scrutiny applied to naturalism, evidentialism, and historical criticism must be applied to the canons of textual reconstruction.

The Christian Alternative

A Christian account of manuscript facts should affirm the following:

1. Every witness belongs to providence. No manuscript is an accident outside God’s government.
2. Every observation is answerable to truth. Confession cannot license false transcription or selective reporting.
3. Every criterion is subordinate. Age, distribution, brevity, difficulty, and genealogy are considerations, not autonomous laws.
4. The text has determinate identity. Editorial judgment recognizes or mistakes that identity; it does not create it.
5. Preservation governs the whole. Local faults do not entail universal loss of the authentic text.
6. The church has a ministerial history. Reception is neither infallible legislation nor theologically meaningless sociology.
7. Reason serves revelation. Critical judgment operates beneath the Word whose witnesses it studies.

This account does not solve every variant in advance. It tells the critic what kind of reality a variant inhabits.

A Test Case in Logical Form

Consider the claim: “Reading A is found in the earliest surviving manuscript; therefore Reading A is original.”

The premise may be factually correct. The conclusion does not follow without additional premises:

• the earliest surviving witness is more likely to preserve the earlier reading;
• no evidence of local corruption defeats that likelihood;
• later witnesses do not preserve an older independent form;
• versional, patristic, internal, and reception evidence do not outweigh the manuscript’s date; and
• the doctrine of preservation permits the relevant displacement of the long-used reading.

Once those premises are stated, debate becomes possible. Some are historical. Some are probabilistic. The last is explicitly theological. The inference was never a bare fact speaking for itself.

The same analysis can be applied to arguments from majority, difficulty, length, or geographical spread. Make the suppressed premises visible. Then test them.

Underdetermination Is Not Indeterminacy

Sometimes the surviving evidence does not force one local conclusion. Two readings may have serious support, and the critic’s ordinary historical canons may not settle the matter with equal confidence for every reader. This is evidential underdetermination: the available considerations leave more than one live judgment at a particular stage of inquiry.

It does not follow that the text itself is indeterminate. One wording was given. God knows it. The fact has determinate identity even if a particular scholar has not recognized it correctly.

This distinction prevents an epistemic limitation from becoming an ontological doctrine. “I cannot presently decide” is a claim about the knower. “No stable textual fact exists apart from critical construction” is a claim about the object. The first may be humble and true; the second requires an argument that variants alone do not provide.

The same distinction appears in other fields. A jury may lack enough evidence to identify a culprit while a definite person committed the act. Historians may dispute the date of an event that occurred on one day. Limited knowledge does not dissolve reality.

For Scripture, providential preservation adds a corporate and historical dimension. One editor’s uncertainty does not establish the church’s non-possession. The reading may be clear in the wider apographic estate, in prior ecclesiastical judgment, or through evidence the editor has undervalued. Conversely, long reception does not make a difficult question unreal. It becomes one consideration within the public history of the text.

This gives the critic disciplined language:

• “This witness is defective here” does not mean “the Word was lost everywhere.”
• “This reading remains disputed among scholars” does not mean “Scripture has no wording here.”
• “My conclusion is fallible” does not mean “the final rule is fallible.”
• “Further evidence may correct me” does not mean “further evidence must reconstruct an absent norm.”

Bahnsen’s apologetic depends upon a similar distinction. Human reasoning is fallible, yet Christian truth is not merely probable. Creaturely limitations do not reduce God’s revelation to a hypothesis. Applied textually, the critic may be modest about himself while confessing certainty about God’s work of preservation.

Criteria Work in Concert

Because no individual consideration interprets itself, textual judgment ordinarily weighs a convergence of evidence. Date, quality, distribution, transcriptional explanation, authorial usage, immediate context, parallel passages, versional support, citations, and ecclesiastical history may reinforce or qualify one another.

Convergence does not eliminate presuppositions. It makes their role more visible. The critic must decide which considerations are admissible, how much weight each receives, and what kind of conclusion the combination may support. A transparent method states these judgments rather than hiding them behind one slogan.

Confessional control contributes the account of the whole: the object is the preserved, authentic text, not an open-ended literary ancestor. Technical criteria then serve local recognition within that whole. They remain genuine evidences, but none is allowed to redefine Scripture’s providential status in silence.

Objectivity Through Submission

Modern thought often imagines objectivity as detachment: the knower becomes objective by bracketing ultimate commitments. Bahnsen offers a different account. Creaturely objectivity comes through submission to the God who knows the object exhaustively and reveals the principles by which it is rightly understood.

That submission does not make the critic omniscient. It makes him teachable. He may revise a local judgment without revising God’s faithfulness. He may say “I do not yet know” without saying “God has not kept.” He may distinguish a disputed witness from an indeterminate Scripture.

The confession supplies certainty at the level God has promised; technical inquiry supplies judgments at the level of particular evidence. Confusing those levels creates either false precision or unnecessary skepticism.

Conclusion

Manuscripts do not interpret themselves. Their dates, readings, corrections, and relationships are real created facts, and they deserve rigorous study. Their place in the doctrine of Scripture depends upon more than their bare existence.

Bahnsen’s rejection of brute facts requires textual critics to expose every suppressed premise. Age is not a verdict. Brevity is not an oracle. Difficulty is not self-authentication. Genealogy is not metaphysics. Reception is not infallible, and it is not meaningless. A variant is not the same thing as a lost Word.

The Christian critic sees the evidence under the God who inspired Scripture, governs history, keeps covenant, and gives his church a rule. That vision does not diminish facts. It is the condition for receiving them as facts rather than fragments in an autonomous reconstruction.

Sources