The Reproducible Research Protocol for This Series

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.

From a Thesis to a Test

The first eleven articles established the question, the target, the textual layers, the distinction between survival and derivation, the strongest-case rule, the canonical-coherence test, the logical vocabulary, the separation of text from translation and interpretation, the problem of hidden borrowing, the use of negative controls, and the burden-and-verdict system. Those principles now require a fixed procedure.

Without a protocol, the series could become a sequence of impressive but irreproducible essays. One article might use NA28, another the ESV, another a reading from the apparatus, and another an inherited Reformed harmonization. Favorable cases could be judged by broad doctrinal compatibility while unfavorable cases were judged by exact confessional wording. The conclusion would then reflect shifting standards rather than the capacity of a defined textual corpus.

The protocol exists to prevent that result.

“Reproducible” does not mean that every informed reader must reach the same theological judgment. It means that another reader can identify the sources, recover the readings, retrace the inferential steps, see every control and repair, and locate the precise point of disagreement. Reproducibility turns a verdict from an assertion into an auditable argument.

The controlling question remains:

Can the doctrinal system confessed in the Westminster Standards be reconstructed from the designated modern critical-text ecosystem, with every necessary premise derived from that ecosystem, every material textual difference counted, and every canonical relation tested?

Research Phases

Phase One: Register the Question before Examining the Outcome

Every audit begins with a short research register. This prevents the article from redefining the doctrinal target after seeing the textual evidence. The register must include:

• article number and title;
• Westminster Confession chapter and paragraph;
• related Larger and Shorter Catechism questions;
• the exact proposition or propositions being tested;
• necessary qualifications and exclusions within the proposition;
• original Westminster proof texts;
• preliminary classification of each proof as principal, cumulative, illustrative, or confirmatory;
• relevant canonical dependencies not listed as formal proofs;
• expected textual units requiring inspection;
• edition cutoff date;
• known negative controls;
• known objections and likely alternative proofs;
• conditions that would count against the project’s provisional thesis.

The last field is essential. A test that cannot register evidence against its thesis is an advocacy exercise, not a genuine audit. Examples of disconfirming outcomes include:

• the critical corpus states the precise confessional proposition as directly as the received control;
• the critical reading resolves a tension present in the received text;
• a supposedly lost proof is replaced by an equally explicit and independent proof;
• an alleged hidden premise is actually derivable from the selected readings;
• the received-text control contains the same difficulty to the same degree;
• an ESV rendering thought to depart from its base proves to be a legitimate direct translation.

Such results must be recorded rather than explained away.

Phase Two: Fix the Corpus and Edition Date

The expression critical text does not identify one book. Each article must state which of the following layers is operative:

The article must include an edition date lock. For this initial protocol the lock is July 17, 2026. On that date, NA28 remained the current published Nestle-Aland; UBS6 had supplied the newer Greek main text intended for NA29; NA29 was still described by its publisher as being prepared; BHS remained the last complete scholarly Hebrew edition; BHQ remained incomplete; and the ESV preface continued to name BHS 1997, UBS5 corrected 2014, and NA28 2012 as its textual bases.¹

No article may attribute a UBS6 reading to NA28. No article may call the ESV a translation of UBS6 merely because UBS6 is newer. No article may treat BHQ as though all thirty-nine Old Testament books have been published in that edition.

Phase Three: State the Confessional Proposition Atomically

Before atomization, a historical-meaning record identifies the original wording, contemporary sense of controlling terms, scope, exclusions, polemical setting, primary sources consulted, and remaining uncertainty. Westminster sentences often contain several propositions. For example, a paragraph may state a positive doctrine, its ground, its means, its limitations, and the error it excludes. Testing the paragraph as one undifferentiated unit encourages vague verdicts. Each paragraph will therefore be divided into atomic propositions. An atomic proposition is not a single word; it is the smallest doctrinal claim that can receive a meaningful proof and verdict. For each proposition, record:

1. subject;
2. predicate;
3. logical relation to adjoining propositions;
4. qualifications of time, respect, agency, or extent;
5. explicit negations and exclusions;
6. catechetical parallel;
7. historical proof texts;
8. whether the claim is express or deduced by good and necessary consequence.

Every identity, deity, personality, distinction, number, exclusivity, unity, equality, relation, universality, modality, and necessity claim is preserved when indispensable. A shortened approximation is not the registered target. This protects both sides. A critical-text corpus should receive full credit for every proposition it can establish. It should not receive credit for the entire Westminster paragraph merely because it supports the paragraph’s broad theme.

Phase Four: Map Every Proof and Its Function

A proof-text list is not self-interpreting. The Assembly may cite several passages for different clauses, and later editions of the Standards may vary in their proof lists. The protocol therefore maps the relation between proposition and passage rather than merely copying references. Each proposition–passage row must state:

• passage reference;
• the exact clause doing the work;
• proof function: principal, cumulative, illustrative, confirmatory, or contextual;
• whether the proof is explicit or inferential;
• received reading;
• NA28 or UBS6 reading;
• BHS or BHQ status where relevant;
• ESV rendering and footnote;
• alternative critical-text proof proposed by an opponent;
• result after the reading change.

The protocol forbids double counting. If one textual variant affects five Westminster citations, the dataset records five proposition–passage relationships but only one underlying textual event. Cumulative articles will distinguish affected proof relationships from unique textual variants.

Phase Five: Transcribe the Readings before Interpreting Them

The article must record the relevant original-language forms exactly enough to permit verification. The record includes:

• lemma or variation unit;
• critical main-text reading;
• punctuation when material;
• brackets, double brackets, diamonds, or other editorial signs when material;
• principal alternative readings;
• identified received reading;
• literal gloss;
• morphology and syntax relevant to the argument;
• source edition and apparatus location;
• manuscript, versional, or patristic evidence only as supported by a reliable apparatus or primary transcription.

Transcription precedes theological characterization. The article first says what the editions contain. It then analyzes why the readings differ and what they mean. The following formulations are prohibited unless the evidence specifically warrants them:

• “all the oldest manuscripts omit”;
• “the majority text says” without defining the counted tradition;
• “the critical text removed” when the historical claim is only that an edition omits;
• “the Textus Receptus added” when the genealogical direction has not been shown;
• “the Greek says” when the statement comes from an English translation;
• “the original says” when the conclusion is merely the main text of one edition.

Phase Six: Separate Text, Translation, and Interpretation

Every case receives a three-layer table.

This separation prevents an ESV decision from being blamed on NA28 and prevents a critical reading from being insulated by a translation that follows another text. It also reveals where the dispute actually lies. Two scholars may agree on the Greek reading and disagree on syntax. Or agree on syntax and disagree on its confessional consequence. The ESV’s own preface requires this discipline. It identifies preferred base editions while expressly allowing divergence in exceptional Old Testament cases and in a few New Testament cases.² Therefore, the ESV is an accessible witness, not a transparent mechanical reproduction of its named bases.

Phase Seven: Construct the Strongest Critical-Text Case

Before issuing criticism, the article must build the best complete case available from the designated critical corpus. The reconstruction must:

1. use every unchanged passage relevant to the proposition;
2. include alternative proofs proposed by responsible Reformed critical-text advocates;
3. employ legitimate grammatical and canonical inference;
4. acknowledge when a doctrine is directly or cumulatively derivable;
5. avoid dependence on readings excluded from the controlling main text unless that dependence is openly labeled;
6. distinguish an apparatus option from the edition’s main text;
7. distinguish the ESV from the Greek or Hebrew it translates;
8. state the strongest objection to the series’ provisional thesis.

The case is not considered complete when it says, “The doctrine is taught elsewhere.” It must identify the elsewhere, state the contribution of each passage, and display the valid inferential chain. The purpose is not theatrical fairness. A weak opposing case cannot tell us what the critical corpus is actually capable of bearing.

Phase Eight: Open the Premise Ledger

Every doctrinal construction will be represented as premises and conclusion. The article may express the reasoning in prose, but the underlying ledger must make dependencies visible. Example form: “ text
P1. Passage A asserts X.
P2. Passage B asserts Y.
P3. X and Y stand in relation R.
P4. Relation R excludes alternative Z.
C. Therefore, Westminster proposition W follows.
“

For each premise, the dataset records its source:

• direct textual statement;
• translation judgment;
• lexical or grammatical inference;
• canonical inference;
• historical information;
• received reading not present in the critical main text;
• ESV departure;
• inherited confessional synthesis;
• conjecture or emendation.

Each indispensable premise also receives one status: explicitly stated, necessarily entailed, cumulatively probable, merely compatible, not established, or contradicted. If the conclusion fails without a premise unavailable from the selected corpus, the construction is a candidate for Grade D. That finding concerns logical dependence, not the interpreter’s motives.

Phase Eight-A: Run the Closure Test

The article constructs the strongest materially different rival model that can affirm the cited passages. It asks what the proof does not exclude and identifies the premise that would be required to close the alternative. Named instances do not establish an exhaustive class. If the rival can affirm every premise without contradiction, the exact doctrine has not been reconstructed, however orthodox or probable the preferred model may be.

Phase Eight-B: Run the No-Borrowing Test

The researcher removes every premise obtained from the Received Text, the confession, inherited tradition, an ESV departure, an apparatus reading not adopted into the declared corpus, or the completed conclusion itself. The inference is then rerun. The corpus receives credit only for the conclusion that survives this subtraction.

Phase Nine: Run the Local Exegesis

The local analysis must address, as relevant:

• morphology;
• syntax;
• lexical range;
• discourse function;
• immediate argument;
• genre;
• speaker and referent;
• quotation formula;
• chronology;
• historical setting;
• punctuation;
• scribal tendencies and transcriptional probabilities.

No appeal to “the whole Bible” may bypass the local wording. Canonical synthesis begins with, rather than replaces, exegesis. Where several interpretations are grammatically possible, the article identifies them, gives the evidence for each, and applies a certainty label. A preferred theological outcome does not turn an available reading into the necessary reading.

Phase Ten: Run the Canonical-Coherence Tests

The protocol tests three directions:

Old Testament with Old Testament

Do the relevant Hebrew readings agree in proposition, chronology, number, agency, and covenantal relation? If they differ, is the difference one of perspective, scope, transmission, or strict conflict?

New Testament with New Testament

Do the selected Greek readings cohere across authors and genres? Does a qualification absent locally appear clearly elsewhere? Does the proposed use of one passage respect the argument of the other?

Old Testament with New Testament

How does the New Testament cite, translate, interpret, typologize, or apply the Old? Does the critical ecosystem supply one coherent account of the relation, or does it alternate among Hebrew, Septuagintal, conjectural, and translational options without a stated rule? Each apparent conflict is tested for subject, time, respect, sense, agency, genre, and perspective before receiving a coherence code. Scripture interpreting Scripture must be demonstrated, not invoked as a slogan.

Phase Eleven: Run the Negative Control

Every critical-specific allegation must compare the received reading. The control record asks:

• Is the relevant wording identical?
• If not, is the operative proposition identical?
• Does the same difficulty exist in the control?
• Is its severity the same?
• Does each text require the same repair?
• Does the proof perform the same function?
• Does either text improve or worsen the construction?

The result is labeled:

• shared;
• critical-aggravated;
• critical-alleviated;
• critical-specific;
• received-specific; or
• undetermined.

This step precedes any charge that the critical text “creates” a contradiction or doctrinal loss.

Phase Twelve: Distinguish Exegesis, Harmonization, and Repair

The protocol uses these terms narrowly.

Exegesis

An interpretation arising from the grammar, syntax, discourse, genre, and historical setting of the operative reading.

Harmonization

A relation between passages established by a warranted distinction that preserves what each passage actually asserts.

Repair

A procedure that changes, bypasses, or supplements the controlling reading in order to obtain coherence. Repairs include:

• adopting an apparatus reading instead of the main text;
• conjectural emendation;
• following a translation that departs from its base;
• importing a qualification from another textual tradition;
• changing punctuation without adequate support;
• supplying an inherited premise not derivable from the selected corpus.

Repair is not automatically illegitimate. An editor or translator may have good reasons to prefer another reading. The protocol’s concern is disclosure: a repaired construction must not be represented as derivation from the unrepaired main text.

Phase Thirteen: Compare the Received-Text Construction Symmetrically

The received-text comparison follows the same stages:

1. identify the exact edition;
2. transcribe the reading;
3. translate it;
4. perform local exegesis;
5. test canonical coherence;
6. identify required harmonizations or repairs;
7. map proof function;
8. issue the same bounded judgments.

The Authorized Version is not permitted to settle the Greek or Hebrew question. Theological familiarity is not permitted to lower the burden. A received reading that appears to solve a problem must actually supply the relevant proposition and fit the canon. The comparison then states the delta:

• no material difference;
• greater directness;
• greater precision;
• greater coherence;
• reduced directness;
• increased ambiguity;
• different but nonmaterial wording;
• unresolved.

Phase Fourteen: Apply Certainty, Grade, Code, and Reach

Every final record contains four independent fields.

Certainty

Established, strongly supported, probable, contested, or undetermined.

Derivation

A through F, as fixed in Article 11.

Coherence

C0 through C4.

Confessional reach

C for catholic or ecumenical, P for magisterial Protestant, R for broadly Reformed, and W for Westminster-specific precision. More than one label may apply. The article separately states in prose whether the demonstrated scope is confined to a local passage, changes a proof function, reaches the proposition, or extends across a larger system. The verdict must include reasons. Bare labels are not publishable findings. An example of an acceptable bounded verdict is:

It is established that NA28 omits the received word at this location. It is strongly supported that the omission removes a local qualification. Two unchanged passages make the broader qualification cumulatively probable but do not necessarily entail the exact universal boundary, and a stated rival model remains open. The proposition therefore receives Grade C, which records substantial but incomplete support rather than successful reconstruction. The canonical tension is resolvable without changing the critical main text, so the coherence code is C1. The greatest warranted conclusion, the missing premise, and the demonstrated proof-textual scope are stated separately.

---

1. Deutsche Bibelgesellschaft, “Novum Testamentum Graece (Nestle-Aland)”; Deutsche Bibelgesellschaft, “The NA 29”; Deutsche Bibelgesellschaft, “The Sixth Edition of the Greek New Testament (UBS6)”; Crossway, “Preface to the English Standard Version”. All accessed July 17, 2026. ↩
2. Crossway, “Preface to the English Standard Version,” “Textual Basis and Resources.” ↩

The form makes clear what was found and what was not.

Phase Fifteen: Write “What This Does Not Prove”

Every article must identify the boundaries of its evidence. At minimum, this section asks:

• Does the case prove originality or only consequence?
• Does it affect one proof or the whole doctrine?
• Does it show dependence or motive?
• Does it show tension or strict contradiction?
• Does it concern NA28, UBS6, BHS, BHQ, the ESV, or an interpreter?
• Does it establish a received-text advantage in this case only or a general rule?
• What evidence would be needed for the stronger claim?

This section is not a ritual disclaimer. It is part of the argument’s validity.

Phase Sixteen: Preserve the Evidence Record

The living dataset will contain at least these fields:

The published prose may summarize the dataset. It may not contradict or conceal it.

Source Hierarchy

Sources will be used in the following order.

Primary textual sources

Printed or official digital editions of NA28, UBS6, ECM, BHS, BHQ, identified received texts, and the ESV 2025.

Primary edition documentation

Edition introductions, publisher documentation, official INTF/ECM materials, and textual commentaries written for the edition.

Primary confessional and historical sources

The Westminster Standards and their proof texts; Reformation and post-Reformation works in verified editions.

Secondary scholarship

Critical commentaries, grammars, monographs, and Reformed systematic theology representing the strongest responsible alternatives.

Tertiary aids

Indexes, databases, search tools, and summaries used to locate evidence. They may guide research but do not bear disputed claims when a primary source is available. Quotations must be checked against the cited source. Search snippets are not quotations. A modern transcription of an early work must be identified as such.

Revision Protocol

A revision is triggered by:

• publication of NA29;
• a new or corrected ECM volume;
• publication of a relevant BHQ fascicle;
• correction of an apparatus or transcription error;
• a new ESV edition or footnote;
• a demonstrated logical or exegetical error;
• a previously unknown alternative reconstruction;
• changed source access that resolves an undetermined claim.

The revision record must preserve the earlier finding and explain the change. Cumulative totals in Articles 148–156 must be recalculated from the dataset, not adjusted impressionistically.

Replication Test

Before an article is considered complete, a competent reader should be able to answer “yes” to the following:

1. Can I identify the historical meaning and exact confessional proposition?
2. Can I recover every cited reading from the named edition?
3. Can I distinguish main text, apparatus, translation, and interpretation?
4. Can I see the complete strongest critical-text reconstruction?
5. Can I classify every indispensable premise under the six-status scale?
6. Can I reproduce the closure countermodel and no-borrowing test?
7. Can I reproduce the negative control?
8. Can I see why the proof function changed or remained stable?
9. Can I reconstruct the path to the grade, coherence code, and greatest warranted conclusion?
10. Can I identify certainty, reach, source form, verification status, response decision, and correction path?
11. Can I state what evidence would overturn or revise the verdict?

If not, the article is not reproducible enough for the cumulative verdict.

Quality-Control Checklist for Publication

Before a Markdown article enters its WordPress import package, verify:

• complete YAML front matter;
• unique slug and correct series number;
• draft status and no unintended date;
• exact shared featured-image path for its production part;
• no H1 title repeated in the body;
• all Greek and Hebrew preserved in UTF-8;
• tables render in Markdown;
• footnote references and definitions match;
• no local-only workspace or filesystem links;
• all web citations use stable public links;
• required final sections appear in order;
• bibliography contains every substantial source cited;
• historical-meaning record precedes proposition grading;
• every indispensable premise has one of the six statuses;
• closure, rival-model, no-borrowing, and greatest-warranted-conclusion fields are complete;
• source artifact form, access, and verification status are recorded;
• named-person response decision, jurisdiction, and pastoral purpose are recorded where applicable;
• revision history states the edition-control date;
• package manifest agrees with the Markdown file;
• ZIP preview shows the intended post and image.

The importer validates format. It cannot validate theology. Both forms of review remain necessary.

The Strongest Objections

“The protocol prejudges the critical text by making Westminster the target”

The target is openly confessional because the question is whether the Westminster system can be reconstructed. The protocol does not assume that the Standards are derivable from every corpus. It allows A through F and requires the series to concede every successful reconstruction. A different project could test a different target.

“No theologian begins with an empty mind”

Correct. The protocol does not demand psychological amnesia. It demands an open premise ledger. An interpreter may know the doctrine in advance while still demonstrating whether the selected text supplies it.

“Edition-exclusive reconstruction is artificial”

It is deliberately controlled. Churches ordinarily use manuscripts, editions, translations, commentaries, and inherited theology together. The experiment isolates the textual base in order to test a particular historical and confessional claim: whether the modern critical ecosystem can independently reproduce a system formed before its present readings and editions existed.

“Reproducibility imports a secular scientific standard into theology”

Reproducibility here means public reasons and recoverable evidence. That demand is congenial to the Protestant appeal to Scripture as the public rule by which religious claims are examined. It does not make divine revelation an experiment or subordinate Scripture to naturalism.

“The protocol is too demanding for ordinary doctrine”

The final claim is unusually large: the reconstruction or failure of an entire confessional system. A correspondingly careful method is appropriate. Shorter articles may summarize routine cases, but the underlying ledger must remain complete.

Verdict

The reproducible research protocol is adopted for Articles 13–156. Each audit will register the proposition, fix the corpus and date, map proof functions, transcribe readings, separate text from translation and interpretation, construct the strongest critical-text case, expose premises, perform local and canonical exegesis, run negative controls, distinguish harmonization from repair, compare the received construction symmetrically, and issue a certainty-qualified, bounded verdict. The protocol is designed to make the project capable of producing a result contrary to its provisional thesis. That possibility is not a weakness. It is what permits a favorable or unfavorable cumulative verdict to carry evidentiary weight.

Notes and Further Reading