Testing Keister’s Method: Matthew 1:7–8, 10–19 and the Actual Variance

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 concrete comparison of Scrivener’s Received Text and NA28 across Matthew 1:7–8 and 10–19 yields neither a textual catastrophe nor a vindication of critical neutrality. There are eight comparison units affecting thirteen printed word tokens. Three are spelling or transliteration differences that normally disappear in English; three are minor lexical or discourse differences; and two change the printed names of kings in Matthew’s genealogy. Those two—Asa/Asaph and Amon/Amos—are an unusually clean test because the principal canons point in opposite directions. Early witnesses and the “harder reading” favor Asaph and Amos. Historical coherence and the broad Byzantine reception favor Asa and Amon. The method becomes visible precisely where its canons disagree.

Now the terminology must face a text.

Keister has correctly written that “the issue is whether a given reading is original” in his chapter-by-chapter review of Why I Preach from the Received Text. That statement is true. It is also the question, not yet the method of answering it. A theory of preservation must explain how the church recognizes the original reading when extant witnesses contain rival forms. A textual method must show how it ranks competing evidence when its ordinary canons do not converge.

Matthew 1 supplies exactly that test.

What Is Being Compared

The comparison below uses F. H. A. Scrivener’s 1894 Greek text as the Received Text reference and the Nestle-Aland twenty-eighth edition as the critical reference. Accents, breathing marks, capitalization, and punctuation are excluded unless they correspond to a lexical difference. A recurring proper name is counted as one comparison unit but as two affected tokens when it appears both as the object of “begat” and as the subject of the next clause.

The readings were checked against the editions and against STEP Bible’s Translators Amalgamated Greek New Testament, a dataset that identifies Scrivener 1894 as its TR base and records membership in NA28, TR, the Byzantine text, and several other editions. This is not an apparatus-based attempt to count every singular spelling in every manuscript. It is a defined comparison of two continuous printed texts.

Here is the result:

Reference: Matthew 1:7–8, twice

Scrivener 1894 TR: Ἀσά / Ἀσὰ

NA28: Ἀσάφ / Ἀσὰφ

Classification: Different proper name

Ordinary English effect: Asa / Asaph

Reference: Matthew 1:10, twice

Scrivener 1894 TR: Ἀμών / Ἀμὼν

NA28: Ἀμώς / Ἀμὼς

Classification: Different proper name

Ordinary English effect: Amon / Amos

Reference: Matthew 1:13, twice

Scrivener 1894 TR: Ἐλιακείμ / Ἐλιακεὶμ

NA28: Ἐλιακίμ / Ἐλιακὶμ

Classification: Orthography/transliteration

Ordinary English effect: Normally Eliakim in both

Reference: Matthew 1:14, twice

Scrivener 1894 TR: Ἀχείμ / Ἀχεὶμ

NA28: Ἀχίμ / Ἀχὶμ

Classification: Orthography/transliteration

Ordinary English effect: Normally Achim in both

Reference: Matthew 1:17, twice

Scrivener 1894 TR: Δαβίδ / Δαβὶδ

NA28: Δαυίδ / Δαυὶδ

Classification: Orthography/transliteration

Ordinary English effect: David in both

Reference: Matthew 1:18

Scrivener 1894 TR: γέννησις

NA28: γένεσις

Classification: Related but distinct nouns

Ordinary English effect: Usually birth; possible “origin” nuance in NA28

Reference: Matthew 1:18

Scrivener 1894 TR: γάρ present

NA28: γάρ absent

Classification: Presence/absence of conjunction

Ordinary English effect: “For,” often left implicit in English

Reference: Matthew 1:19

Scrivener 1894 TR: παραδειγματίσαι

NA28: δειγματίσαι

Classification: Related verbs

Ordinary English effect: Publicly shame/expose/disgrace

The first result is not a crisis but a count. There are eight comparison units affecting thirteen printed tokens. Verses 11–12 and 15–16 have no lexical difference between these two editions. The changes in verses 13, 14, and 17 concern Greek representations of the same names and normally vanish in translation. The two nouns in verse 18 are close enough that both may be rendered “birth.” The conjunction affects the explicit connection of clauses but no doctrine. The verbs in verse 19 differ in form and nuance but substantially overlap in this context.

No doctrine of Christ is lost in these verses. That does not make the words unreal. A spelling difference that disappears in English should not be preached as a doctrinal loss. A conjunction is still a word even when doctrine does not turn upon it. The question under examination is verbal identity and public text, not whether Christianity survives the comparison.

The percentage-of-agreement argument therefore receives its due. Agreement is extensive. Most differences here are minor. But neither fact identifies the original reading at a disputed locus. A method is not tested where every kind of evidence points the same way. It is tested where its principles compete.

Asa or Asaph?

Matthew’s genealogy moves from Solomon to Rehoboam, from Rehoboam to Abijah, and then to the king known from the Old Testament as Asa. Scrivener prints Ἀσά. NA28 prints Ἀσάφ, the ordinary Greek form associated with Asaph. This is not merely the difference between two harmless ways of spelling the same name. The final phi distinguishes the Greek names. Asa is the king of Judah; Asaph is the name associated especially with the psalmist and Levitical line. Whatever Matthew intended his reader to understand, the two printed forms are lexically distinguishable.¹

The external case for Asaph is serious. Representative support includes Papyrus 1, Codex Sinaiticus, Codex Vaticanus, Codex C, Family 1, Family 13, and additional Greek, versional, and lectionary witnesses. Papyrus 1, commonly dated to the early third century, supports the final phi in both occurrences, though the edge of the fragment requires restoration in the first. The Penn Museum’s images of Papyrus 1 make the antiquity of the reading a physical fact rather than a nineteenth-century invention.

The external case for Asa is also substantial, though its extant Greek support is generally later. It is represented by such uncials as E, K, L, W, Delta, Pi, and Sigma; by many minuscules and lectionaries; and by the Byzantine textual tradition reflected in Scrivener. The reading became broad, stable, and ecclesiastically dominant in the Greek transmission.

Those facts establish two boundaries. The papyrus proves that Asaph is early; it does not prove that Matthew wrote Asaph. The Byzantine prevalence of Asa proves broad transmission and reception; bare numerical superiority does not, by itself, prove originality. Internal evidence creates the sharper contest.

The standard critical explanation is immediately plausible. A scribe encountering Asaph in a list of kings would recognize an apparent historical problem and correct it to Asa. On that account, the Received reading is smoother because copyists repaired the harder form. The canon commonly called lectio difficilior—prefer the more difficult reading when other things are equal—therefore favors Asaph. So does the familiar transcriptional question: which direction of change is easier to explain? Asaph becoming Asa by historical correction appears easier than Asa becoming Asaph without an obvious trigger.

That is the critical case at its strongest. It should not be dismissed as arbitrary excision, hostility to Scripture, or blind allegiance to two codices. It combines early attestation with a coherent account of scribal behavior. Yet it is not conclusive. A hard reading may be original; it may also be hard because it is corrupt. Lectio difficilior is a comparative probability, not an oracle. A very early exemplar could acquire a final phi, and descendants could reproduce it. Multiple surviving witnesses may then attest the antiquity and spread of the alteration without carrying us back to the autograph. Age establishes antiquity, not apostolicity.

Intrinsic coherence also matters. Matthew is not supplying a free association of biblical names. He is tracing the royal line. The expected person is King Asa, the son of Abijah and father of Jehoshaphat. The reading Asa does not merely make the sentence easier; it makes the name agree with the historical identity the genealogy unmistakably requires.

Some interpreters have proposed that Matthew intentionally wrote a form evoking Asaph, perhaps creating a literary or theological allusion to the Psalms. That hypothesis is possible as an explanation of an intentional difficulty. But possibility is not evidence of authorial intent. Nothing in the immediate syntax signals a detour from the kings to the psalmist, and the genealogy continues as though one royal ancestor has followed another. A conjectured allusion should not be converted into a license to impute a historical misidentification to the inspired evangelist.²

The theological consideration is not an embarrassment smuggled into neutral philology. If Matthew wrote under divine inspiration, intrinsic probability properly includes the coherence of his assertion with the historical testimony of Scripture. That does not mean every superficially difficult reading must be rejected. Scripture can contain purposeful difficulty, irony, wordplay, compressed genealogy, and unfamiliar forms. It means that a reading which appears to make the inspired author name the wrong biblical figure bears a real explanatory burden.

Thus the canons divide. External antiquity and the scribal-correction hypothesis favor Asaph. Historical coherence, the absence of a demonstrated allusion, and broad ecclesiastical transmission favor Asa. Calling one side “evidence” and the other “theology” would falsify the actual argument. Both readings require a cumulative judgment.

Amon or Amos?

Matthew 1:10 repeats the pattern. The genealogy moves from Hezekiah to Manasseh, from Manasseh to the king Amon, and from Amon to Josiah. Scrivener prints Ἀμών. NA28 prints Ἀμώς, the Greek form Amos. Again, this is not a difference that can be reduced to accents or transliteration. Nu and sigma distinguish Amon from Amos. In ordinary biblical reference, Amon is the king of Judah and father of Josiah; Amos is the prophet associated with the book bearing his name.

Representative support for Amos includes Sinaiticus, Vaticanus, C, Gamma, Delta, Theta, Family 1, and minuscule 33. Amon is represented by L, W, Family 13, and the broad Byzantine tradition. Papyrus 1 cannot decide this locus because its surviving leaf has a gap after verse 9 and resumes later in the chapter. The case for Amos is early and weighty, but it should not be padded with a papyrus that does not contain the verse.

The same transcriptional explanation returns. If Amos stood in an early copy, a scribe familiar with the royal history had a reason to change it to Amon. If Amon was original, why would a scribe replace a correct king’s name with that of a prophet? The harder-reading canon and the correction hypothesis favor the NA28 form.

Historical coherence again favors the Received form. Matthew’s sequence requires Amon. Nothing in the clause tells the reader that the prophet Amos has entered the royal line. An intentional literary allusion has again been proposed: the altered name may evoke a prophet as the earlier alteration evokes a psalmist. But the pairing can be read in two directions. It may support a deliberate Matthean pattern, or it may reveal that an early branch twice transmitted corrupted royal names and thereby invited later correction. The mere existence of two difficulties does not decide whether the pattern is authorial or transmissional.

This cumulative feature makes the test unusually revealing. The critical editor is not choosing one isolated oddity. He prints two forms that appear, on their face, to identify the wrong Old Testament men within a royal genealogy. His defense is that their difficulty and early support make scribal correction to the familiar kings more probable than two accidental corruptions in the other direction. The Received Text defender answers that the inspired and historically coherent royal names, reinforced by broad Greek reception, should not be displaced on the strength of a canon that can mistake early corruption for authorial difficulty.

Neither argument is answered by counting manuscripts alone. Neither is answered by counting centuries alone. Neither is answered by repeating “harder reading.” The question is which evidential judgments may govern the text and how the doctrine of preservation orders those judgments.³

What the English Reader Is Allowed to See

The editorial decision becomes an ecclesiastical consequence when the Greek text is translated. The official ESV text of Matthew 1 prints Asaph in verses 7–8 and Amos in verse 10. The English reader therefore sees the names corresponding to its critical Greek base. The official NET text, however, prints Asa and Amon, even though the NET’s textual method is not a Received Text method. Its translation communicates the historical persons intended by the genealogy rather than reproducing the distinctive NA28 spellings as different English names.

That contrast is instructive. It does not prove dishonesty or inconsistency in either committee. It shows that the textual and translational decisions are distinguishable. An editor chooses Ἀσάφ and Ἀμώς for the Greek main text. A translator must then decide whether English readers will see Asaph and Amos or the kings Asa and Amon. The apparatus preserves the alternatives; the main text chooses; the translation either exposes or normalizes the result.

The difference is not doctrinal ruin. It is nevertheless more than an invisible spelling adjustment. A reader of one translation encounters two apparent historical anomalies that a reader of the other does not. A commentary or footnote may explain them, but explanation presupposes that an editorial decision has already reached the page. Formal preservation in the aggregate does not resolve this. An aggregate contains both answers. It gives the church the variation unit; it does not, by itself, give the church the reading.

The Remaining Variants: Neither Inflated nor Erased

The other six units prevent the analysis from becoming a polemic constructed around only its most dramatic examples. The forms of Eliakim, Achim, and David are orthographic or transliterational differences. They refer to the same persons and ordinarily produce the same English names. Their existence matters to an exact collation, but they should not be represented as three separate attacks upon biblical content. This is precisely where Keister’s insistence upon proportion is salutary.

At Matthew 1:18, γέννησις in the TR is the more specific noun for birth or begetting, while NA28’s γένεσις can denote origin, beginning, or birth. Context makes “birth” a natural rendering of either. The difference is lexical, not merely typographical, but its ordinary English effect is small.

In the same verse, the TR and Byzantine text contain γάρ, “for,” after the participle describing Mary’s betrothal; NA28 does not. Before deciding ancient direction, the accurate description is presence in Scrivener and absence in NA28. One may later argue that a scribe supplied a connective or that another lost it. Editorially, NA28 excludes the conjunction from its main text. Ecclesiastically, many translations express the relation naturally without an explicit “for.” Small consequence does not mean no decision.

At verse 19, Scrivener’s παραδειγματίσαι and NA28’s δειγματίσαι are related verbs concerning public exposure, disgrace, or shame. The prefixed TR form can carry the force of making someone a public example; the shorter form can mean exposing or disgracing. Context draws their semantic fields close together. Here too, one should resist both exaggeration and disappearance: this is a genuine lexical substitution with limited translational effect.⁴

The honest result is therefore mixed. Three units are orthographic. Three are minor but lexical or syntactical. Two are proper-name readings with unmistakable historical and translational significance. “Vast agreement” accurately describes the passage globally. It does not adjudicate the two places at which the method is under the greatest pressure.

What Would Keister’s Method Have to Do?

Keister identifies himself as Sturzian rather than as an unqualified defender of every NA28 reading. He gives greater weight to Byzantine evidence than some older forms of critical eclecticism and has publicly accepted received readings at important loci. It would therefore be unfair to announce, without an explicit statement from him on Matthew 1, that he must choose Asaph and Amos.

The proper question is conditional: what would his stated approach have to do with the evidence? If he selects Asaph and Amos, early attestation and the scribal-correction hypothesis have overridden historical coherence and broad Byzantine reception. His preservation doctrine then locates both alternatives within the aggregate while critical judgment installs the harder names in the continuous text. A translation may preserve those forms for the reader, as the ESV does, or normalize them to the historical kings, as the NET does.

If he selects Asa and Amon, he has allowed historical coherence and Byzantine continuity to override the readings printed in NA28 and supported by several of the earliest extant witnesses. That conclusion is possible within a Sturzian approach. But it demonstrates that NA28 is not self-authenticating, that the oldest extant reading is not a synonym for the original reading, and that the harder reading is not automatically the inspired reading.

The point is not to trap Keister. Either answer may be argued. The point is that “preserved in all manuscripts collectively” does not generate the answer. The aggregate supplies Asa/Asaph and Amon/Amos. To move from the aggregate to a continuous text, Keister must rank antiquity, geographical distribution, genealogical relation, scribal probability, intrinsic coherence, Byzantine continuity, and ecclesiastical reception. His textual judgment does the identifying work that his definition of preservation leaves undone.

This is the structural concern developed throughout the series. Keister explicitly affirms preservation, and his conservative intentions should be granted. Yet if preservation means the survival of the inspired words somewhere among all extant readings, the doctrine guarantees an evidentiary field, not a publicly identified text. The critical process must still constitute the wording presented to the church at every disputed locus.

The Received Text position also faces a demand. It may not answer these variants by merely saying, “The received reading is received.” It must defend the ministerial significance of reception, explain why the early difficult forms are secondary, and show why historical coherence is evidence rather than special pleading. Its case is cumulative: the royal identities required by the context, the possibility of early corruption, the limits of lectio difficilior, broad Byzantine continuity, and the confessional expectation of a publicly preserved text.⁵

That case does not make every copyist, manuscript, printer, TR edition, or KJV printing perfect. It does not make Scrivener inspired. It does claim that criticism should minister to the authentic text preserved and received in the church rather than place that text under perpetual reconstitution by an aggregate that contains mutually exclusive readings.

What This Test Proves—and What It Does Not

Matthew 1:7–8 and 10–19 does not prove every critical reading wrong. It does not prove originality by majority vote. It does not establish that every TR difference has equal significance. It does not permit accusations about the motives of Keister, critical editors, or translators. It establishes five narrower conclusions.

First, the agreement between Scrivener and NA28 is genuinely extensive. A preservationist weakens his case if he conceals that fact. Second, actual variance cannot be summarized as spelling alone. Most units in this sample are minor, but Asa/Asaph and Amon/Amos name different biblical figures in ordinary Greek and English usage.

Third, critical canons can pull in opposite directions. Early attestation and a plausible scribal-correction account favor the critical forms; historical coherence and broad ecclesiastical transmission favor the received forms. “Follow the evidence” cannot decide the issue until one explains how different kinds of evidence are ordered.

Fourth, the doctrine of preservation is not completed by the survival of both readings. Preservation in the aggregate tells us that the rival forms remain available. It does not identify which form belongs to the text. Fifth, editorial and ecclesiastical consequences remain even when doctrinal consequences are small. The main text chooses; translations either display or mediate that choice; churches read one wording rather than a cloud of variants.

Keister’s strongest point—that originality is the question—survives the test. But the test also exposes the deficiency of treating preservation as collective survival. The church needs more than access to the contest between Asa and Asaph. It needs Matthew’s word. A confessional doctrine must account for how that word was not only retained somewhere, but kept, received, and publicly available as Scripture.

That brings the series to its final question. If the authentic text is neither an infallible physical artifact nor a permanently unconstituted set of possibilities, where does a Reformed doctrine locate it? How do inspiration, apographic preservation, ecclesiastical reception, and ministerial criticism fit together without collapsing into either KJV-onlyism or modern critical reconstruction? The final article will be “Preservation with a Preserved Text: The Confessional Verdict.”

Notes and Further Reading