Burden of Proof, Certainty Labels, and the Verdict Scale

Revision note: Substantially revised July 20, 2026. The argument has been reordered according to the classical rhetorical movement, compressed where necessary, and given one consolidated Notes and Further Reading section.

A Scale Cannot Repair an Argument

This series employs grades, coherence codes, and certainty labels. Those devices can discipline judgment, but they can also create an illusion of precision. A letter placed in a table does not become true merely because the table looks exact. A verdict is warranted only when the evidence has met the burden appropriate to the claim. The reconstruction test therefore needs more than a list of outcomes. It needs rules governing who must show what, how strongly it must be shown, what kind of evidence can discharge the burden, and how uncertainty must be reported. The fundamental distinction is this:

Certainty labels qualify our confidence in a claim; derivation grades describe what the designated corpus can bear; coherence codes describe the relation among the operative readings; reach labels identify how broadly the confessional proposition is shared.

These are four different axes. A scholar may be highly certain that NA28 prints a particular reading while judging the doctrinal consequence of that reading to be contested. A doctrine may receive grade B with high confidence. A proposed strict contradiction may receive code C4 only with low confidence and therefore fail publication under that label. Or the textual evidence may remain divided while every viable reading supports the same confessional proposition, leaving doctrinal derivation stable despite textual uncertainty. None of those findings, by itself, tells us whether the proposition is catholic, Protestant, broadly Reformed, or Westminster-specific.

No single score can compress those differences without losing information.

The First Burden: Say Exactly What Is Being Claimed

Safeguard 19 places a prior burden upon that statement: a historical confession must first be interpreted in the sense its words bore in its own setting. The audit must identify contemporary usage, scope, exclusions, and polemical context before reducing the sentence to atomic premises. A modern paraphrase may be evaluated, but it may not silently become the proposition Westminster is said to confess. Many disputes over textual criticism never reach the evidence because the claim changes during the argument. Consider the following statements:

1. NA28 prints a certain reading.
2. The NA28 editors considered that reading the earliest attainable text.
3. That reading is the autograph’s wording.
4. The ESV translates that reading.
5. The ESV’s English necessarily means a particular proposition.
6. That proposition weakens a Westminster proof.
7. The Westminster doctrine can no longer be derived.
8. The critical text contradicts another biblical passage.
9. Critical-text advocates preserved the doctrine only by borrowing it from the received-text tradition.

The first statement is directly observable. The second requires documentation of editorial status and stated purpose. The third is a historical-textual conclusion. The fourth requires comparison of Greek or Hebrew, translation, and footnote. The fifth requires grammatical and semantic argument. The sixth requires a proof-function analysis. The seventh requires a complete canonical reconstruction. The eighth requires formal logical demonstration after legitimate distinctions are exhausted. The ninth requires evidence of an indispensable premise missing from the selected corpus; it cannot be inferred from presumed motives.

An argument that proves statement 1 has not thereby proved statements 3–9. The series will therefore write claims in their narrowest accurate form. “NA28 prints” will not be silently changed to “the original text says.” “The ESV renders” will not be changed to “the Greek says.” “This proof becomes less direct” will not be changed to “the doctrine disappears.” “A prima facie tension arises” will not be announced as “Scripture contradicts itself.”

Precision at the claim level is the first form of intellectual honesty. It is tempting to assign one permanent burden to “the critical-text advocate” or “the received-text advocate.” That is too simple. The burden normally rests on the person advancing the proposition at issue.

If this series claims that a critical reading produces a new contradiction, this series bears the burden of demonstrating the contradiction. If a critic replies that the doctrine is “taught elsewhere,” that critic bears the burden of identifying the other passages and constructing the precise doctrine from them. If the series claims that the ESV departs from NA28, it must place the wording side by side. If an ESV defender says the departure follows a better Greek reading, the defender must identify that reading and its textual warrant.

This reciprocity prevents two asymmetries.

The first is polemical asymmetry: demanding a complete apparatus-level demonstration from one side while accepting a slogan from the other. The second is confessional asymmetry: treating the inherited Reformed system as proof that every selected modern reading must be capable of producing it. The whole point of the reconstruction test is to discover whether that capability exists.

The Confession itself commends this discipline. It permits doctrine to be deduced by “good and necessary consequence,” but it does not call every desired consequence good or necessary.¹ It directs the church to rest in the Holy Spirit speaking in Scripture rather than in councils, ancient writers, doctrines of men, or private spirits.² The inherited confession is the target being tested; it is not a substitute for the textual proof required by the test.

The Burden Map

The following burdens govern the series.

The burden grows with the reach of the claim. “This word is omitted from NA28” is local and readily verified. “Reformed Christology cannot be reconstructed from the critical text” is systemic and requires the entire relevant corpus. The more comprehensive the conclusion, the more comprehensive the evidence must be.

Five Certainty Labels

The series will use verbal labels rather than invented percentages. Percentages can suggest a mathematical calibration the evidence does not possess. The labels qualify discrete claims, not whole articles.

Established

The claim is directly observable or supported by convergent primary documentation, and no material uncertainty presently affects it. Examples include: NA28 prints a specified string; the German Bible Society states that NA28 is the current published Nestle-Aland edition; the ESV preface names BHS, UBS5, and NA28 as its textual bases; UBS6 adopts ECM textual changes in Mark, Acts, and Revelation.³ “Established” does not mean metaphysically infallible. It means the claim is secure enough to function as a datum in this project.

Strongly supported

The evidence substantially favors the claim, responsible alternatives have been considered, and the conclusion is unlikely to change without significant new evidence or argument. Many textual and exegetical judgments will fall here. The claim remains an inference rather than a directly observed fact, so the supporting argument must remain visible.

Probable

The claim currently explains the evidence better than its alternatives, but a material unresolved consideration remains. A probable claim may guide a provisional verdict, but the article must identify the uncertainty.

Contested

Responsible alternatives remain live because the evidence is divided, the decisive premise is disputed, or the conclusion depends upon an unresolved interpretive choice. The article may still state which option it prefers, but it must not present that preference as settled fact.

Undetermined

The available evidence does not warrant a responsible choice, or the research necessary to choose has not been completed. “Undetermined” is a result, not an embarrassment to be hidden. These labels apply separately to textual reading, translation, interpretation, proof function, and doctrinal effect. A case record may therefore read:

The separate labels expose precisely where disagreement belongs.

Observation, Inference, and Judgment

Every article will distinguish three levels of statement.

Observation

An observation reports accessible data: wording, punctuation, brackets, apparatus entry, translation, footnote, confession paragraph, or historical citation.

Inference

An inference explains what those data most likely mean: why a scribe may have produced a reading, how a syntax functions, whether a translator followed another witness, or which proposition a clause asserts.

Judgment

A judgment evaluates the result under the project’s theological rubrics: whether a proof is weakened, whether a premise is borrowed, whether the system is coherent, and which grade or code applies. Confusing these levels produces premature certainty. “The word is absent” is an observation. “A scribe omitted it by accident” is an inference. “Its absence destroys the doctrine” is a judgment requiring a canonical audit. Each must carry its own evidence and certainty.

Textual Certainty Is Not Doctrinal Certainty

A reading can be textually uncertain while its doctrinal effect is stable. Suppose two variants differ lexically but both affirm the deity of Christ in the respect under examination. The textual question remains important, yet that particular doctrine does not depend on resolving it. The reverse is also possible. Editors may be confident in a reading whose doctrinal effect is difficult to state. Main-text certainty does not automatically settle syntax, referent, or canonical relation.

The UBS tradition historically made the distinction visible by attaching A–D evaluations to selected main-text decisions. The German Bible Society’s account of the first UBS edition explains that the rating indicated how certain the committee considered its text at each apparatus entry.⁴ Whatever changes UBS6 makes to presentation, the conceptual lesson remains: confidence in the chosen reading is a separate question from the theological consequence of choosing it. The series’ A–F derivation scale must never be confused with those editorial ratings. A in this series means that a confessional proposition is fully and directly derivable; it does not mean that a textual reading has received an editorial A rating.

The Verdict Scales

A — Exact proposition explicitly established

Every indispensable element of the historically controlled confessional proposition is explicitly and coherently established from the designated corpus. Identity, deity, personality, distinction, number, exclusivity, unity, equality, relation, universality, modality, and necessity are included wherever the proposition requires them. Grade A does not require one proof text to state an entire chapter of the Confession. It requires that the proposition under review be directly warranted by the corpus and not merely compatible with it.

B — Exact proposition necessarily entailed

No single passage states the whole proposition, but a valid and sound synthesis necessarily entails it. Every indispensable premise must be textually available, exegetically warranted, mutually coherent, and sufficient to close materially different rival models. Grade B is not a demotion in truth. Much systematic theology is necessarily cumulative. The grade records the form of derivation.

C — Substantial but incomplete support

The corpus supplies substantial or cumulatively probable support, but at least one indispensable element is not explicitly stated or necessarily entailed. The article must name the missing or merely probable premise and the rival construction left open. Grade C is not a qualified reconstruction success.

D — Completed by an external premise

The exact conclusion is reached only when an indispensable premise from outside the declared corpus is supplied—from the Received Text, confession, tradition, a translation departure, an apparatus reading treated as main text, or another external source. Grade D is not assigned merely because an interpreter knows church history. The missing premise, its source, its indispensability, and the failure of the designated corpus to supply it must all be shown.

E — Compatible or underdetermined

The corpus is compatible with the Westminster proposition but does not establish it, or it permits more than one materially different construction and does not adjudicate among them. An interpreter may prefer the Westminster result, but the edition-exclusive reconstruction cannot claim to have derived it.

F — Contradicted or excluded

One or more operative readings contradict or exclude an indispensable element of the precise confessional proposition under the best-supported natural reading. This is the strongest grade and carries the heaviest burden. It is not warranted by proof-text loss alone. Only Grades A and B are successful exact reconstructions. Before a whole-doctrine grade is assigned, every indispensable premise receives one status:

1. Explicitly stated: the corpus directly asserts the premise in the required sense and scope.
2. Necessarily entailed: the premise follows by a valid and sound inference from explicitly stated propositions.
3. Cumulatively probable: the combined evidence favors the premise but does not make its denial contradictory.
4. Merely compatible: the premise can coexist with the evidence, but the evidence does not favor or entail it sufficiently.
5. Not established: the required premise lacks adequate warrant in the declared corpus.
6. Contradicted: an operative reading denies or excludes the premise in the relevant respect.

The grade may not outrun the weakest indispensable premise. A Grade B argument cannot contain a premise that is only probable, compatible, or unestablished. The closure test asks what the proof fails to exclude. If a materially different model can affirm every cited proposition without contradiction, the exact doctrine has not been proved. Named examples do not establish an exhaustive class; repeated triads do not, by repetition alone, prove that the class contains exactly three members.

The no-borrowing test removes every premise learned from the Received Text, confession, church tradition, translation departure, or conclusion itself and reruns the inference. If the argument fails after removal, the external premise must be disclosed and the result cannot receive A or B. The derivation grade asks what doctrine can be built. The coherence code asks how the passages relate.

C0 — No relevant tension

The readings cohere without special pleading.

C1 — Resolvable tension

A grammatically, historically, or theologically warranted distinction explains the appearance of conflict. The distinction must arise from the text rather than merely preserving the desired result.

C2 — Repair-dependent

Coherence requires an emendation, a translation departure, a reading absent from the controlling main text, or another explicit repair. A legitimate appeal to another canonical passage is not automatically C2; the code applies when the controlling reading itself must be altered or bypassed.

C3 — Systemically unstable

Different cases require mutually inconsistent repair principles. For example, the reconstruction may treat the main text as binding when it supports the doctrine, the apparatus as binding when the main text does not, and the ESV as decisive when neither Greek option is convenient. One isolated inconsistency does not establish a system; the pattern must be demonstrated.

C4 — Strict contradiction

The operative readings yield P and not-P of the same subject in the same respect, time, and sense, after legitimate distinctions have been exhausted. C4 is never inferred merely from surprise, ethical discomfort, different numbers, or apparently divergent narrative details. Article 3 fixed four reach labels so that a Westminster-specific result is not inflated into a claim about all Reformed or catholic Christianity. Article 11 retains those labels unchanged:

The labels describe doctrinal breadth, not importance, textual location, or degree of damage. More than one may apply. A variant can affect a W-level formulation of a doctrine whose core is C-level, while leaving the catholic proposition intact. The article must also state the scope of the demonstrated effect—local passage, proof function, confessional proposition, or larger system—in ordinary prose. That scope statement does not replace the locked C/P/R/W reach codes.

Applying the Burdens

Proof-text loss is often overstated and often minimized.

To establish it, the article must first state the proposition for which the Westminster Standards cite the passage. It must then determine the proof’s function: principal, cumulative, illustrative, or confirmatory. Next it must show what the received reading contributes and what the critical reading retains. Finally, it must identify alternative passages and construct the strongest remaining case.

Three outcomes are possible:

1. The cited passage no longer proves the proposition, but another passage proves it directly.
2. The cited passage becomes less direct, and the doctrine becomes cumulative.
3. The cited passage was indispensable, and the precise proposition is no longer derivable.

Only the third outcome approaches grades D–F, and even then the exact reason must be stated. The slogan “no doctrine is affected” fails because it ignores proof function. The slogan “a word disappeared, therefore the doctrine disappeared” fails because it ignores the rest of the canon.

Article 9 defined hidden borrowing as the unacknowledged use of textual, translational, semantic, canonical, confessional, historical, or inferential capital not supplied by the selected corpus. The accusation carries a significant burden because it can easily become a claim about motives. The series will not say that an author “secretly borrowed” merely because the author is Reformed. It will state an argument of this form:

1. The desired conclusion requires premises A, B, and C.
2. The designated corpus supplies A and B.
3. C cannot be derived from the relevant readings under the stated rules.
4. The reconstruction nevertheless uses C.
5. C is traceable to an inherited synthesis, another textual reading, or a translation departure.

That demonstration establishes dependency, not dishonesty. Grade D is a claim about the logical source of a premise, not the moral character of the interpreter. Because contradiction is rhetorically powerful, it is often used imprecisely. A valid C4 finding must specify:

• proposition P;
• proposition not-P;
• the subject of each proposition;
• the relevant time;
• the respect in which each predicate is asserted;
• the semantic sense of the disputed terms;
• why genre, agency, chronology, ellipsis, and perspective do not resolve the conflict;
• why the received-text negative control does or does not contain the same conflict.

If any of those entries remains open, the case may be C1, C2, C3, or undetermined. Calling every tension a contradiction does not strengthen the confessional case. It teaches readers to discount the verdicts that deserve attention. The common reply that a doctrine rests on many passages is methodologically sound but incomplete. The reply bears its own burden:

1. Identify the passages.
2. State the proposition supplied by each.
3. Demonstrate the inferential link among them.
4. Show that no indispensable premise comes from the displaced reading.
5. Test the passages for canonical coherence.
6. Establish the precise Westminster proposition, not merely a broadly orthodox approximation.

Until that construction is performed, “elsewhere” names a research task rather than an answer.

Writing the Verdict

Every substantive article will end with a bounded statement in this form: “ text
Textual finding: [claim + certainty]
Translation finding: [claim + certainty]
Proof-function finding: [claim + certainty]
Derivation grade: [A–F + explanation]
Premise statuses: [status of every indispensable premise]
Closure result: [rival models excluded / rival model still open]
No-borrowing result: [independent / external premise named]
Greatest warranted conclusion: [the strongest proposition actually established]
Coherence code: [C0–C4 + explanation]
Confessional reach: [C/P/R/W; multiple allowed]
Demonstrated scope: [local passage / proof function / proposition / system]
Control result: [shared / aggravated / alleviated / unique / undetermined]
Revision trigger: [new edition, new evidence, or unresolved check]
“

The prose article may express this naturally, but the evidence ledger will preserve the fields. A reader should be able to disagree with the doctrinal judgment while still seeing which textual and translational facts are established. The following rules are binding:

• No F without demonstrating incompatibility with the precise confessional proposition.
• No C4 without formal contradiction after legitimate distinctions fail.
• No D without naming the missing premise and its external source.
• No A or B while an indispensable premise is only probable, compatible, not established, or contradicted.
• No exhaustive conclusion from a list of named examples without a closure premise.
• No credit to the declared corpus for a premise borrowed from the Received Text, confession, tradition, translation, or apparatus.
• No proof-text-loss claim without mapping the proof’s original function.
• No critical-specific charge without a received-text negative control.
• No ESV-departure claim without comparing the English to the actual Greek or Hebrew base.
• No systemic claim from one local variant.
• No certainty label attached to an entire side or tradition.
• No motive claim inferred from logical dependence.
• No silent upgrade from “probable” to “established” in the conclusion.

These constraints may produce fewer dramatic verdicts. They will also make the genuine verdicts harder to dismiss.

Verdict

The reconstruction test will use separate axes for certainty, doctrinal derivation, canonical coherence, and C/P/R/W confessional reach, while stating the demonstrated scope in prose. Burdens attach to particular claims, not permanently to one theological camp. The stronger and broader the conclusion, the greater the evidence required.

The operational verdict is adopted: all subsequent articles must distinguish observation, inference, and judgment; apply the five certainty labels at the claim level; use grades A–F only for doctrinal derivation; use codes C0–C4 only for coherence; retain the C/P/R/W reach labels fixed in Article 3; state the scope of the demonstrated effect; and stop the conclusion where the evidence stops.

Notes and Further Reading