Paper 1 of Five
Abstract
This paper names a form of argumentative behavior that has been noticed piecemeal across several literatures and treated as a general object in none of them: the setting aside of an authoritative text by a party who never denies its authority, never alters its wording, and never argues that it means something other than what it says. The text is left whole and simply not brought forward. I call this functional subtraction and distinguish it from three neighbors with which it is routinely conflated — suppression, which removes the text; mistranslation, which alters it; and disputed interpretation, which engages it. The central claim is that non-citation of an on-point text, by a party who cites freely and densely from the same corpus elsewhere, is a positive act rather than an omission, and is therefore a proper object of description and counting. The demonstration corpus is American proslavery religious literature, roughly 1830–1865, with the fugitive-rendition sermons of 1850–1851 as the paradigm case. The paper states four conditions that must jointly hold before the category applies, steelmans the strongest case against the category at length, and closes with a falsification constraint: if non-citation rates for on-point adverse texts match non-citation rates for on-point favorable texts within the same corpus, the category dissolves and this paper should be withdrawn.
1. The Problem
In September 1850 the Congress enacted a statute requiring citizens of the free states to assist in the return of persons who had escaped from bondage. In the months that followed, ministers across the North and the South preached on the duty of obedience to it. Their sermons were printed, sold, answered, and reprinted. They are among the most Scripture-dense productions of the American pulpit: Romans 13 on the powers ordained of God, 1 Peter 2 on submission to the ordinance of man, the render-unto-Caesar passage, the household codes, Abraham and his servants, Onesimus returned to Philemon.
There is one command in the Bible that addresses, in plain terms and without qualification, what a man is to do when a servant escapes from his master and comes to him:
Thou shalt not deliver unto his master the servant which is escaped from his master unto thee: He shall dwell with thee, even among you, in that place which he shall choose in one of thy gates, where it liketh him best: thou shalt not oppress him. (Deuteronomy 23:15–16)
It is not obscure. It is not textually contested. It is not buried in a genealogy or a building specification. It is a direct prohibition on precisely the act the statute compelled, and it comes with a positive provision for the escaped servant’s dwelling and a closing clause against oppressing him.
In a great many of these sermons it does not appear. Not refuted, not distinguished, not confined to the Mosaic economy, not argued to apply only to servants fleeing heathen masters into Israel — which is what the arguments would have been, and were, in the handful of places where the verse was met head-on. It simply is not there. The preacher who had, on the same page, produced three verses in support of magistracy, and who could be relied upon in other sermons to reach for a passage from Habakkuk or Zephaniah when it served, arrived at the one command on the subject and passed it by.
That is the phenomenon this paper is about. It is not a claim about slavery, which needs no further adjudication, and it is not a claim about anyone’s soul. It is a claim that a certain kind of silence has a shape, that the shape can be described, and that describing it well requires a name that does not already belong to something else.
2. The Act Named
Functional subtraction is the operational removal of a text from an argument by a party who concedes the text’s authority, leaves its wording intact, offers no reading of it, and does not bring it forward at the point where it bears.
The word functional does the work. Nothing is subtracted from the canon; the verse remains printed in every copy in every pew. What is subtracted is the verse’s function — its capacity to constrain the argument in progress. The text keeps its status and loses its force. It becomes, for the duration of the argument, furniture rather than authority.
Four features characterize the act as I define it:
Concession of authority. The subtracting party is not a skeptic. He holds the corpus to be binding and says so, frequently and at length. This is not incidental; it is what makes the act possible. A man who denies the authority of a text has no need to subtract it, because it constrains nothing. Functional subtraction is available only to the believer.
Preservation of the wording. No alteration, no emendation, no preferred variant, no convenient rendering. Where the text is quoted at all — usually in another connection — it is quoted as it stands.
Absence of engagement. No exegesis, no distinction, no dispensational limitation, no argument from context. The absence of engagement is not incidental to the definition; it is the whole of it. A text that has been distinguished has been engaged, and engagement is a different phenomenon requiring a different name.
Non-appearance at the point of bearing. The text is missing where it is on point, in a body of argument that is otherwise citation-dense and that treats texts of comparable directness as decisive on other questions.
Note what is not in the definition: intent, motive, awareness, dishonesty, hypocrisy. These are unmeasurable in the historical record and the category does not need them. Functional subtraction is a description of what a corpus does, not a diagnosis of what an author knew.
3. Boundary Work: Three Neighbors
The category earns its keep only if it is not already covered. Three existing terms come close and each fails in a way worth stating precisely, because the failures are informative.
3.1 Suppression removes the text
Suppression is the destruction, prohibition, or physical withholding of the text itself. It is the burning of the pamphlet, the mails closed against the tract, the statute forbidding the teaching of letters. The antebellum South supplies genuine instances: abolitionist literature was seized from the mails, and laws in several states criminalized teaching enslaved persons to read.
Suppression and functional subtraction are not the same act and do not have the same signature. Suppression is visible from outside — there is a statute, a bonfire, a customs record. It is also, crucially, inefficient against a text the suppressor himself venerates. Nobody proposed to excise Deuteronomy 23 from the Bibles of Charleston. The Bible was the one book that could not be edited, and its distribution to the enslaved was, in many quarters, actively promoted. This is the point: functional subtraction is the technique that becomes necessary precisely where suppression is unavailable. It is what you do when you cannot remove the page.
The two therefore stand in something close to complementary distribution. Where a party controls the text, expect suppression. Where a party venerates the text and cannot touch it, expect functional subtraction. The proslavery religious literature is a limit case of the second condition, which is part of why it makes a good demonstration corpus.
3.2 Mistranslation alters the text
Mistranslation — including tendentious rendering, selective lexical choice, and the quiet preference for a variant that softens a command — alters the words the reader receives. It has been studied, it leaves traces in the philological record, and it is detectable by collation.
It is also, for this corpus, largely beside the point. The men under study worked from the Authorized Version, as did their opponents, and the disputed renderings that did matter (chiefly doulos and the Hebrew ebed) were argued in the open, in print, by both sides. That is engagement, not subtraction. Where translation was contested it was contested loudly. The verses this paper is concerned with are not translation problems. Deuteronomy 23:15–16 says in the Authorized Version what it says in the Hebrew, and no party to the dispute claimed otherwise.
3.3 Disputed interpretation engages the text
This is the near neighbor and the one most often mistaken for the phenomenon. Disputed interpretation argues that the text means less, or other, than the opponent claims. It confines the command to the Mosaic economy. It reads the escaped servant of Deuteronomy 23 as a fugitive from a foreign master, fleeing heathen bondage into the covenant land, so that the command has no bearing on rendition between two jurisdictions of a Christian nation. It reads the manstealer of Exodus 21:16 and 1 Timothy 1:10 as the kidnapper of a free man, leaving lawful acquisition untouched.
These are arguments. Some of them are bad arguments. All of them are the right kind of thing, and they matter here for a reason that cuts against my own thesis: they were in fact made. John Henry Hopkins made the foreign-master argument. Stringfellow and others made the manstealing distinction. The specialized proslavery literature engaged these texts, sometimes at length.
Disputed interpretation therefore honors the text in the only way that finally counts: it concedes that the verse has a claim on the argument and must be answered before the argument can proceed. It puts the text on the field. Whatever else may be said of Hopkins, he had read Deuteronomy 23 and knew he had to deal with it.
The distinction I am drawing is not between good and bad exegesis. It is between exegesis and none. And the existence of the engaged treatments is what makes the unengaged ones countable, because it establishes that the on-point status of these verses was known to the parties. This is developed in §6 and is met directly in the steelman at §7.2.
3.4 A fourth candidate: innocent ignorance
One more neighbor deserves naming and dismissing. A text may go uncited because the citer did not know it, did not recall it, or did not perceive its bearing. This is not functional subtraction and no honest account can pretend otherwise.
It is also, for this corpus, an unusually weak explanation, and the reason is worth stating. These were men of prodigious Scripture memory who produced apt verses on demand as a professional competence, in a culture where the Bible was the common possession of the literate and the illiterate alike. More to the point: the enslaved themselves knew these texts, sang them, and preached them, and their opponents in print cited them constantly and by name. Weld’s The Bible Against Slavery organized itself around Exodus 21:16. Bourne, Barnes, Cheever, and Goodell put Deuteronomy 23:15–16 in the front rank of the antislavery argument for thirty years. A man writing a rejoinder to Barnes cannot plausibly be said not to have known which verses Barnes used.
Ignorance remains a live explanation for any individual silence and is never excluded case by case. It is not a plausible explanation for a corpus-level pattern, and the category is a corpus-level category.
4. Why Non-Citation Is a Positive Act
The load-bearing claim of this paper — and of the series — is that non-citation of an on-point text by a free citer is something done rather than something that failed to happen. This needs argument, because the surface grammar runs the other way.
4.1 The acts-and-omissions problem
Moral philosophy has a long dispute over whether the act/omission distinction carries weight, and I do not need to settle it. My claim is narrower and descriptive: whatever the moral upshot, there is a class of non-doings that are individuated, chosen, and attributable in the way doings are, and non-citation by a free citer belongs to it.
The mark of that class is selection under competence. A man who cannot swim does not choose not to rescue. A man standing at the edge with a rope in his hand does. The relevant question is never “did something fail to occur” but “was the agent operating a selection procedure over the relevant options, and did this option come before the procedure.”
Citation in a Scripture-dense argument is a selection procedure of exactly this kind. The writer is not receiving verses passively; he is retrieving them, sorting them, and deploying them against a specific proposition. Every citation he makes is evidence that the procedure is running. The denser the citation, the stronger the evidence. When the retrieval set demonstrably includes obscure and far-flung passages, the claim that the most on-point verse in the canon simply never came up ceases to describe an absence and starts describing an output.
4.2 The implicature argument
There is a second and independent route. Grice’s maxim of relation holds that a contribution to a cooperative exchange is expected to bear on the matter at hand; Sperber and Wilson generalize this into a presumption of optimal relevance. Both entail that in a genre organized around producing the governing text, the non-production of the governing text carries meaning.
Proslavery pulpit argument was such a genre in an unusually strict sense. Its whole warrant was that Scripture settled the question. A sermon that argued from prudence, economics, or racial theory alone was, by the corpus’s own standard, not a sermon. The genre’s promise to the hearer was: here is what the Bible says about this. Against that promise, the absence of the one verse addressing the case is not neutral. The hearer who left that sermon believing there was no command on fugitive rendition was not misled by anything the preacher said. He was misled by the shape of what the preacher selected, and the shape was the preacher’s.
4.3 The within-passage control
The strongest single argument is not theoretical but structural, and it is the cleanest evidentiary asset the series has.
Consider Ephesians 6. Verse 5 — “Servants, be obedient to them that are your masters according to the flesh, with fear and trembling, in singleness of your heart, as unto Christ” — is among the most cited verses in the entire proslavery corpus. It appears in sermons, pamphlets, catechisms prepared for the enslaved, and denominational reports. Four verses later, in the same paragraph, by the same author, addressed to the same congregation, comes verse 9: “And, ye masters, do the same things unto them, forbearing threatening: knowing that your Master also is in heaven; neither is there respect of persons with him.”
The same structure appears in Colossians (3:22 against 4:1), in 1 Peter, and in 1 Timothy 6.
This is a natural experiment of a quality historians rarely get. A comparison between verse 5 and verse 9 holds constant: the book, the human author, the immediate context, the passage’s familiarity, the reader’s memory encoding, the printing cost of the quotation, the lectionary position, the genre, the occasion, and the audience. Every innocent explanation for differential citation that Paper 5 catalogues — homiletic convention, occasional preaching, lectionary constraint, printing economics — is neutralized by construction, because the two verses share all of these properties.
One variable is left standing. Verse 5 binds the servant. Verse 9 binds the master, who is the man in the pew paying the minister’s salary.
If differential citation rates within such passages prove to be large and consistent, no explanation but direction survives the design. If they prove to be small, the category is in serious trouble. This is the measurement developed and operationalized in Paper 3.
4.4 What the claim does not require
It does not require that any writer thought “I shall omit this verse.” Selection procedures run below the level of deliberate choice constantly, in scholarship as in preaching, and the honest statement is that we cannot see inside. It does not require unanimity — §7.2 concedes that some writers engaged these texts openly. And it does not require that the pattern be conscious to be real. It requires only that the pattern be a product of the corpus’s selection behavior rather than of chance or of a mechanism indifferent to the direction of the text. That is an empirical claim, and §8 states what would refute it.
5. Conditions of Application
The category applies only where four conditions jointly hold. Each is a gate, and failing any one of them disqualifies the case.
C1 — Free citation. The party cites the corpus densely and voluntarily, across a range wide enough to establish retrieval competence. Operationalized in Paper 3 as citation density above the corpus median plus demonstrated reach into low-frequency passages.
C2 — On-point status. A competent contemporary reader would recognize the text as bearing directly on the proposition, and the same corpus elsewhere treats texts of comparable directness as decisive. The standard is deliberately conservative; borderline cases code as not on point.
C3 — Non-engagement. The text receives no exegesis, distinction, limitation, or refutation anywhere in the unit of analysis. A single sentence of engagement removes the instance from the category and places it under disputed interpretation.
C4 — Availability. The text was in the party’s working canon, in an accessible edition, and had been placed in the dispute by identifiable opponents in circulation. Availability is established documentarily, not assumed.
C1 excludes the sparse citer. C2 excludes the tendentious reading-in of relevance, which is the commonest failure mode of absence claims. C3 excludes the engaged case, which is a different phenomenon. C4 excludes genuine ignorance.
6. Detection Signatures
Five patterns mark functional subtraction in a corpus. None is sufficient alone; the argument is cumulative and the floor principle applies — a case is only as strong as its weakest satisfied condition, not as strong as the sum of them.
S1 — Asymmetry within the retrieval range. On-point adverse texts go uncited at rates exceeding on-point favorable texts of comparable familiarity, length, and canonical position.
S2 — The truncated passage. Citation stops at the point where the passage turns and binds the citer. Ephesians 6:5 without 6:9; Colossians 3:22 without 4:1. This is the half-verse warrant of Paper 3 and the sharpest visible signature.
S3 — Reach mismatch. The same author retrieves obscure supporting texts while missing famous adverse ones. Where a writer produces a verse from the minor prophets to establish a minor point and never produces Matthew 7:12 on the major one, the retrieval explanation collapses.
S4 — Adversary-set divergence. The set of texts the opposing literature treats as central is systematically underrepresented, without the divergence being marked as a disagreement about relevance.
S5 — Register-dependent presence. The text appears in the specialized or technical literature, where a professional audience would notice its absence, and vanishes from the popular and homiletic literature reaching the general hearer. This signature is the hinge between this paper and Paper 4, and it is the one that most complicates the simple story. It is treated at §7.2.
7. Steelmanning: The Strongest Case Against
An absence claim is the easiest kind of claim to make and the easiest to make badly, because the evidence is by construction not there. What follows is the best case I can build against my own category. I have tried to make it strong enough that I would be persuaded by it if I could not answer it, and in two places below I cannot fully answer it.
7.1 The argument from silence is a known fallacy
The objection. The argument from silence has been recognized as weak since antiquity and analyzed rigorously in the modern literature. Lange (1966) showed that historical silence arguments require an implausibly strong premise — that the source would have mentioned the thing had it existed or mattered. McGrew (2014) put the matter in Bayesian terms: the inference is licensed only when the probability of silence given the hypothesis is much lower than the probability of silence given its negation, and that ratio is almost never established. Functional subtraction is an argument from silence with a new name and a coat of methodological paint. Naming a fallacy does not repair it.
The strongest form. Worse: the naming may make it more dangerous. A named category with detection signatures and a coding scheme has the apparatus of rigor without necessarily having the thing itself. Readers grant credibility to a taxonomy that they would withhold from the bare observation “he didn’t cite it.” The paper may be manufacturing false confidence.
Response. McGrew’s own analysis supplies the conditions under which the inference is licensed, and the four conditions in §5 are an attempt to instantiate them. The likelihood ratio is exactly what the within-passage control at §4.3 estimates: given a writer who cites Ephesians 6:5, what is the probability he does not cite 6:9? That is a measurable quantity with a measurable comparison. The objection is correct that most silence arguments never establish the ratio. It does not follow that none can. What it establishes is that the burden sits on me, which I accept, and that Paper 5 is not optional garnish but a load-bearing member.
I do not think this objection is fully answered. It is answered to the extent that a measurement can answer it, and it will remain partly open until the control corpus is built.
7.2 The texts were engaged — the premise is false
The objection. This is the serious one, and it is factually grounded. Deuteronomy 23:15–16 was not universally passed over. Hopkins engaged it. So did others, arguing that the escaped servant in view was one fleeing a heathen master into the covenant land, or that the provision belonged to the civil code of a particular polity and did not transfer. Exodus 21:16 was met with the manstealing distinction repeatedly. Stringfellow, Thornwell, Bledsoe, and Dabney all handled adverse material at length. Fuller argued the whole question with Wayland across a full volume.
So the category rests on a false premise about the corpus. The proslavery divines did not decline to engage. They engaged and lost. What the paper is calling functional subtraction is the ordinary distribution of arguments across a literature: not every writer makes every argument, and no one demands that every sermon rehearse the whole controversy.
The strongest form. Push further. If the texts were engaged in the literature, then any given sermon’s silence is licensed by the availability of the argument elsewhere. A preacher may reasonably assume the exegetical work has been done and need not redo it from the pulpit. Physicists do not rederive thermodynamics in every paper. Silence in the presence of an available published answer is not subtraction; it is citation economy, and treating it as subtraction would convict every scholar in every field.
Response. I concede the premise and think it changes the shape of the claim rather than defeating it — but the reader should judge whether that is a repair or a retreat.
First, the concession is what makes the phenomenon countable. C4 requires availability, and the engaged treatments establish it. A writer who published in a denominational quarterly that Deuteronomy 23 concerned heathen masters, and who then preached a rendition sermon that never raised the verse, satisfies every condition in §5 with the availability question closed by his own hand. The engagement does not exonerate the silence; it dates it.
Second, the citation-economy analogy fails on the audience. A physicist writing for physicists addresses readers who can locate the derivation. A minister preaching rendition to a general congregation addresses hearers whose entire access to the question is the sermon in front of them. The published answer is not available to them in any operational sense. This is signature S5, and it is why Paper 4 exists.
Third — and this is where the objection retains real force — the shift I have just made moves the claim from “the texts were not engaged” to “the texts were not engaged in the register where engagement would have reached the people bound by the outcome.” That is a narrower claim and a harder one to measure, because it requires classifying registers, matching author identities across venues, and establishing audience reach. I regard the narrower claim as the true one and the original framing as loose. A reader who thinks the narrowing is ad hoc is entitled to that judgment, and the honest test is whether the register classification was fixed before the counting or after. It was specified before; the pre-registration record will say so or it will not.
7.3 Citation is driven by mechanisms indifferent to direction
The objection. Citation frequency in homiletic literature tracks familiarity, memorability, meter, lectionary position, catechetical use, and the plain economics of a printed page. Ephesians 6:5 may be cited more than 6:9 because it is the opening of a passage, because it was in the catechisms, because it scans, because it was what everyone had heard — a self-reinforcing frequency effect entirely indifferent to whom the verse binds. The bibliometric literature has documented at length that citation reflects convention and visibility rather than probative weight (MacRoberts & MacRoberts, 1989), and Merton’s Matthew effect describes exactly the accumulation dynamic that would produce the observed asymmetry with no directional mechanism at all.
The strongest form. The within-passage control at §4.3 does not defeat this, because verse position is itself a variable. Opening verses of household codes are structurally more citable than closing ones regardless of content. To break the tie, one would need passages where the citer-binding half comes first — and if the asymmetry persists in those, the position explanation dies; if it reverses, my explanation dies.
Response. This is a fair objection and it supplies its own decisive test, which I adopt. The reversed-order passages exist — Colossians places the masters’ obligation at 4:1, after the servants’ at 3:22, but Malachi, Job 31, and several prophetic passages state the obligation of the powerful first. Paper 3 will report position-controlled comparisons, and I commit here to reporting them whichever way they fall. If the asymmetry tracks position rather than direction, this paper is wrong and I will say so.
7.4 “Positive act” smuggles in intent
The objection. The paper disclaims intent and then relies on it. “Positive act,” “selection,” “choice,” “declined to bring forward” — this vocabulary carries agency, and agency carries culpability. Either the claim is about what men did knowingly, in which case it is unprovable from the printed record; or it is a claim about statistical patterns, in which case the moralized vocabulary is doing illegitimate rhetorical work.
Response. I accept the criticism of the vocabulary more than the criticism of the claim. “Positive act” is a term of art here meaning an output of a running selection procedure, and nothing more. That is a mechanical notion, not a mental one; it applies to any system that sorts options against a criterion, and it does not require that the sorter be able to report the criterion. But the ordinary-language pull of the phrase is real, and a reader who finds the vocabulary prejudicial has identified a genuine defect in presentation rather than in substance — I have not found a better term, and if one exists this paper should adopt it.
7.5 The category is unfalsifiable in practice
The objection. The stated falsification constraint requires comparing non-citation rates for on-point adverse and on-point favorable texts. But “on-point” is coded by the investigator, and an investigator committed to the thesis will code adverse texts as on-point more readily than favorable ones. The falsification test is therefore hostage to the very judgment it is meant to check.
Response. Correct, and it is the reason for the blinded coding procedure, the two-of-three concordance rule, and the pre-registered alpha floor of 0.67 specified in Paper 3. On-point status is coded from the proposition alone, with the direction of the text masked where masking is feasible. Where masking is not feasible — and for famous verses it often is not — the procedure records that and the finding is reported as weaker. I do not claim the test is airtight. I claim it is a real test with real ways to fail, run by coders who can and should overrule me.
7.6 The historiography already knows this
The objection. Noll, Haynes, Genovese, Tise, and Swartley have all noted that particular texts went missing from particular defenses. The observation is not new. A new name for an old observation is not a contribution.
Response. I agree the observations exist and this paper is additive rather than revisionist — it overturns nothing in that literature and depends on it throughout. What has not been done, so far as I can find, is to treat the behavior as a general object with conditions of application, detection signatures, a measurement, a control corpus, and a stated way to be wrong. The difference between a recurring observation and a category is that a category can be transported, tested, and refuted. Whether that difference is worth a paper is the reader’s call to make.
8. Falsification Constraint
The category dissolves under the following condition, stated in advance:
If, within the same corpus and after the controls specified in Paper 5, non-citation rates for on-point adverse texts do not exceed non-citation rates for on-point favorable texts of comparable canonical familiarity, position, and length, then functional subtraction has no referent and this paper should be withdrawn.
Three subsidiary constraints, each independently sufficient to defeat the paper:
F1 — Position dominance. If the within-passage asymmetry (§4.3) tracks verse position rather than the direction of obligation when tested on reversed-order passages, the mechanism is bibliometric and not directional.
F2 — Control corpus parity. If a matched control corpus addressing questions unrelated to the institution — matched for denomination, region, decade, and venue — shows comparable adverse/favorable citation asymmetries on its own contested questions, then the asymmetry is a property of homiletic argument generally and carries no specific weight here.
F3 — Coder failure. If blinded coding of on-point status fails to reach Krippendorff’s alpha of 0.67, the coding is reported as failed. It is not recoded until it passes. A failed coding is a published result.
The commitment is to report all three whichever way they fall, and to publish the abandonment as readily as the finding. A methodology series that cannot survive one of its own papers being wrong is not a methodology series.
9. Scope and What Is Not Claimed
This paper does not claim that the men studied were insincere. Intent is not measurable here and the argument does not require it.
It does not claim that silence is generally guilty. Most silence is innocent. Paper 5 exists because the difference between guilty and innocent silence is the whole methodological problem, and a reader who takes this paper without that one has taken a knife without its guard.
It does not claim novelty against the historiography of American slavery, which is mature and to which this adds nothing.
It does not claim that the difficulty lay in the text. The commands were whole and in force, and they were not brought forward. Locating the defect in the citer rather than in Scripture is not a rhetorical flourish; it is what the measurement is for, and it is the reason the reciprocity balance index of Paper 3 counts what it counts.
What the paper does claim is narrow: that there is a distinguishable act, that it has conditions and signatures, that it can be counted, and that it can be shown not to exist.
Notes
- Unit of analysis. The unit for C1–C4 is the individual published work — sermon, pamphlet, address, or article — not the author’s collected output. An author may functionally subtract in one work and engage in another; that pattern is itself signature S5 and is measured, not averaged away.
- The 1850–1851 rendition sermons. These are treated as the paradigm case because the proposition at issue is unusually narrow (whether a fugitive is to be delivered up), the on-point text is unusually direct, and the corpus is bounded by a datable statute. Narrowness of proposition is a methodological asset for absence work: the wider the proposition, the more candidate texts, and the weaker any single non-citation.
- “On point” and conservatism. The coding standard resolves ties against the thesis. A verse that a modern reader finds decisive but that a contemporary reader could reasonably have thought oblique codes as not on point. This will understate the phenomenon. Understatement is the correct direction of error for a category whose failure mode is over-finding.
- Deuteronomy 23:15–16 and the heathen-master reading. I do not adjudicate the reading in this paper. Its status here is as an engaged treatment, which places it under disputed interpretation and outside the category by C3. Whether the reading survives examination belongs to a different kind of paper than this one.
- The within-passage control and reversed-order passages. Identification of reversed-order reciprocal passages must be completed and fixed before citation counting begins. If the set is assembled after the primary result is known, F1 is not a test.
- Relation to Paper 4. Signature S5 sits on the boundary between this paper and Domain Quarantine. The rough division: this paper measures whether a text is present; Paper 4 measures whether the register in which it is present is one that reaches the party bound by the outcome. Cases satisfying S5 will be counted in both and should not be treated as two independent confirmations.
- Relation to Paper 5. No counting claim in this paper is established until the control corpus specified in Paper 5 is built. Publishing the two together is a design decision, not a courtesy.
- Scripture. Quotations follow the Authorized Version, both because it is the text the corpus used and because the argument turns on what a nineteenth-century reader had before him.
- Vocabulary. The term “positive act” is used throughout in the restricted sense given at §4.4 and §7.4: an output of a running selection procedure. Readers who find the term prejudicial are asked to substitute that gloss wherever it appears; nothing in the argument depends on the ordinary-language connotation, and if a better term exists this paper should take it.
References
Barnes, A. (1846). An inquiry into the Scriptural views of slavery. Parry & McMillan.
Bennett, J. (1995). The act itself. Oxford University Press.
Bledsoe, A. T. (1860). Liberty and slavery. In E. N. Elliott (Ed.), Cotton is king, and pro-slavery arguments (pp. 269–458). Pritchard, Abbott & Loomis.
Bourne, G. (1845). A condensed anti-slavery Bible argument. S. W. Benedict.
Cheever, G. B. (1857). God against slavery, and the freedom and duty of the pulpit to rebuke it as a sin against God. Joseph H. Ladd.
Cover, R. M. (1975). Justice accused: Antislavery and the judicial process. Yale University Press.
Dabney, R. L. (1867). A defence of Virginia, and through her, of the South. E. J. Hale & Son.
Douglass, F. (1845). Narrative of the life of Frederick Douglass, an American slave, written by himself. Anti-Slavery Office.
Elliott, E. N. (Ed.). (1860). Cotton is king, and pro-slavery arguments. Pritchard, Abbott & Loomis.
Finkelman, P. (1981). An imperfect union: Slavery, federalism, and comity. University of North Carolina Press.
Foot, P. (1967). The problem of abortion and the doctrine of the double effect. Oxford Review, 5, 5–15.
Fox-Genovese, E., & Genovese, E. D. (2005). The mind of the master class: History and faith in the Southern slaveholders’ worldview. Cambridge University Press.
Fuller, R., & Wayland, F. (1845). Domestic slavery considered as a Scriptural institution. Lewis Colby.
Genovese, E. D. (1998). A consuming fire: The fall of the Confederacy in the mind of the white Christian South. University of Georgia Press.
Glancy, J. A. (2002). Slavery in early Christianity. Oxford University Press.
Goodell, W. (1852). The American slave code in theory and practice. American and Foreign Anti-Slavery Society.
Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics 3: Speech acts (pp. 41–58). Academic Press.
Gutjahr, P. C. (1999). An American Bible: A history of the Good Book in the United States, 1777–1880. Stanford University Press.
Harrill, J. A. (2006). Slaves in the New Testament: Literary, social, and moral dimensions. Fortress Press.
Hayes, A. F., & Krippendorff, K. (2007). Answering the call for a standard reliability measure for coding data. Communication Methods and Measures, 1(1), 77–89.
Haynes, S. R. (2002). Noah’s curse: The biblical justification of American slavery. Oxford University Press.
Holifield, E. B. (2003). Theology in America: Christian thought from the age of the Puritans to the Civil War. Yale University Press.
Hopkins, J. H. (1864). A Scriptural, ecclesiastical, and historical view of slavery. W. I. Pooley.
Krippendorff, K. (2018). Content analysis: An introduction to its methodology (4th ed.). SAGE.
Lange, J. (1966). The argument from silence. History and Theory, 5(3), 288–301.
MacRoberts, M. H., & MacRoberts, B. R. (1989). Problems of citation analysis: A critical review. Journal of the American Society for Information Science, 40(5), 342–349.
Mathews, D. G. (1977). Religion in the Old South. University of Chicago Press.
McGrew, T. (2014). The argument from silence. Acta Analytica, 29(2), 215–228.
Merton, R. K. (1968). The Matthew effect in science. Science, 159(3810), 56–63.
Morris, T. D. (1996). Southern slavery and the law, 1619–1860. University of North Carolina Press.
Noll, M. A. (2006). The Civil War as a theological crisis. University of North Carolina Press.
Nord, D. P. (2004). Faith in reading: Religious publishing and the birth of mass media in America. Oxford University Press.
Nosek, B. A., Ebersole, C. R., DeHaven, A. C., & Mellor, D. T. (2018). The preregistration revolution. Proceedings of the National Academy of Sciences, 115(11), 2600–2606.
Proctor, R. N., & Schiebinger, L. (Eds.). (2008). Agnotology: The making and unmaking of ignorance. Stanford University Press.
Raboteau, A. J. (1978). Slave religion: The “invisible institution” in the antebellum South. Oxford University Press.
Rachels, J. (1975). Active and passive euthanasia. New England Journal of Medicine, 292(2), 78–80.
Ross, F. A. (1857). Slavery ordained of God. J. B. Lippincott.
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology: Undisclosed flexibility in data collection and analysis allows presenting anything as significant. Psychological Science, 22(11), 1359–1366.
Smylie, J. (1836). A review of a letter from the Presbytery of Chillicothe to the Presbytery of Mississippi on the subject of slavery. William A. Norris.
Snay, M. (1993). Gospel of disunion: Religion and separatism in the antebellum South. Cambridge University Press.
Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Blackwell.
Stout, H. S. (1986). The New England soul: Preaching and religious culture in colonial New England. Oxford University Press.
Stringfellow, T. (1856). Scriptural and statistical views in favor of slavery. J. W. Randolph.
Swartley, W. M. (1983). Slavery, Sabbath, war, and women: Case issues in biblical interpretation. Herald Press.
Thornwell, J. H. (1850). The rights and the duties of masters. Walker & James.
Tise, L. E. (1987). Proslavery: A history of the defense of slavery in America, 1701–1840. University of Georgia Press.
Trouillot, M.-R. (1995). Silencing the past: Power and the production of history. Beacon Press.
Tushnet, M. V. (1981). The American law of slavery, 1810–1860: Considerations of humanity and interest. Princeton University Press.
Weld, T. D. (1838). The Bible against slavery. American Anti-Slavery Society.
Wiecek, W. M. (1977). The sources of antislavery constitutionalism in America, 1760–1848. Cornell University Press.
