Sign in
Dylanvir
Account
Atlas

← the contents

appendix · the north, 965 — 1335

Thraldom, and the boundary that moved

The rest of this book asks who won. This appendix asks the question underneath it: who could be taken, on whose authority, and what it took to stop being taken. The answer the sources give is not a law about slavery. It is a rule about MEMBERSHIP — and a rule about membership does not end a trade, it relocates one.

8 doctrine rows across 4 scopes, 4 rungs of freedom, 4 polities, 2 events that publish and 2 that do not. Of the 4 polities, 1 carries a dated ban and none yields a computable lag. That is the shape of the evidence before any argument is made on it.

A guide to the labels and evidence

IThe doctrine that made rank a body

Divine-hereditary: Rig (Heimdallr in disguise) visits three households and fathers Thrael, Karl and Jarl -- ancestors of slaves, farmers and magnates.

The poem does not merely describe hierarchy, it NATURALISES it as divinely instituted. The thrall line is marked PHYSICALLY: the son ugly with a crooked back, his partner Thir described as filthy, descendants given black hair, thick ankles, coarse fingers, low and deformed stature. Status is asserted as heritable and VISIBLE -- not law, which can be amended, but phenotype, which cannot.

sourceRigsthula (Poetic Edda) · late 10th / early 11th c. (composition) · liturgical text
gradeT3 contested — and the dating is the reason
scopeNorse, pre-conversion · a cultural zone

why it is coded contested

THE DATING CARRIES THE OBJECTION. Scholars generally date Rigsthula to the late 10th / early 11th c. -- exactly when a universalist religion is arriving. A myth naturalising hereditary rank composed at that moment reads as a DEFENCE UNDER PRESSURE, not a settled inheritance. So the poem may be evidence the doctrine was CONTESTED rather than evidence it was settled.

falsifier · A secure dating of the poem well BEFORE the arrival of Christianity in Scandinavia would remove the defence-under-pressure reading and let the row stand as settled doctrine rather than contested doctrine.

Coded T3 and contested=1 ON PURPOSE. The temptation is to read the clearest statement in the corpus as the standard belief; the date says it may be the least standard thing in it.

the same field, one token away, and a whole cosmology apart

Heritable-legal: the disability descends unless the both-ales condition is met, so rank is inherited by law rather than by birth-nature.

Norse, pre-conversion

divine hereditary

a visible phenotype. No public act amends it.

the Norse law codes

heritable legal

a legal disability. A condition can be met and it lifts.

THE COMPARISON THE value_base COLUMN EXISTS FOR. divine_hereditary against heritable_legal is one token apart in the schema and a whole cosmology apart in fact: one is amendable and one is not. Prose alone would never have surfaced the pair.

IIThe ladder — and what the rungs actually measure

Freedom in the codes is a gradient, not a switch. There are 4 states, and the thing that moves a man up them is not whether anyone owns him. Ownership ends at rung 1 and the ladder runs to rung 3.

The manumission ladder: four states of freedom, ordered by publicity rather than by ownership.A ladder of four rungs beside a vertical axis labelled publicity — how many people saw it — running from level 0 at the foot to level 3 at the top. Level 0 is the thrall, owned, where there is no act to witness because nothing has been granted. Level 1 is the frjalsgjafi, freedom given but not made public, the lowest rank of freedmen: without the freedom ale he cannot earn above a third of an eyrir, the master controls where he lives and whom he may marry, he cannot leave the master's property without permission, and his children cannot inherit. Between level 0 and level 1 a dashed line marks where ownership ends. Level 2 is the leysingi, reached by giving a freedom ale, after which more freedoms follow; his children still cannot inherit unless he had married a freedwoman and both had given their ale. Between level 1 and level 2 a solid rule and a burst of witness marks carry the freedom ale itself, drawn as a witnessing event. Level 3 is the freedman with further freedoms, earned by further work or granted, who is still consistently regarded as inferior in the codes, whether he committed an offence or was the victim of one. A coloured bar runs down the left edge of all four rungs, the top one included, because a constraint still binds at every level. The figure argues one thing: ownership ends at rung 1 while the ladder runs two rungs further, so ownership is not what the rungs measure — publicity is.The manumission ladderfreedom is a gradient, not a binary — and the operative variable is not ownership, it is publicityPUBLICITY — HOW MANY PEOPLE SAW ITFreedman, further freedomsleysingi — more freedoms earnable by further work, or by further grantsSTILL BINDS consistently regarded as inferior in the codes, whether he committed an offence orwas the victim of one. The top rung of the ladder is not equality.3Freedman after the freedom aleleysingi — reached by giving a freedom ale, after which more freedomsSTILL BINDS his children cannot inherit unless he had married a freedwoman and BOTH had given their ale.A freedman is still ranked below a freeborn man in the codes.2Freedman, grant not made publicfrjalsgjafi — freedom given, but not announced. The lowest rank of freedmenSTILL BINDS without the freedom ale he cannot earn above a third of an eyrir.The master controls where he lives and whom he may marry; he cannot leave the master'sproperty without permission; his children cannot inherit.1Thrallthrall — ownedSTILL BINDS there is no act to witness. Nothing has been announced, because nothing has been granted.OWNED0OWNERSHIP ENDS HERE — AND THE LADDER RUNS TWO RUNGS FURTHERa private grant — freedom given, but not made public. Nobody was told, so nothing was witnessed.THE FREEDOM ALE a witnessing event — status is a function of how many people saw itthe same structure as marriage banns, or a landnam claim: the act is public, or it is not an actfurther freedoms — more earnable by further work, or by further grants, and each one announced againOwnership ends at rung 1. The ladder runs two rungs further — so ownership is not what the rungs measure.Manumission had to be confirmed by an oath and announced publicly, at an assembly or in the church.
the ladder — four states, and the constraint that still binds at each
  1. 0ThrallthrallownedT2

    publicity 0 · none -- no act to witness

    Owned. Rigsthula places the estate inside the created order and marks it physically -- black hair, thick ankles, coarse fingers, low and deformed stature -- which asserts a status no public act can amend.

    ceiling · No ceiling question arises: nothing has been granted.

  2. 1Freedman, grant not made publicfrjalsgjafinot ownedT2

    publicity 1 · private grant -- freedom given but NOT made public

    Lowest rank of freedmen. WITHOUT THE FREEDOM ALE HE COULD NOT EARN ABOVE A THIRD OF AN EYRIR -- an earnings cap attached not to his labour, not to his owner, but to the fact that nobody watched.

    ceiling · Capped at a third of an eyrir until the ale is given.

  3. 2Freedman after the freedom aleleysinginot ownedT2

    publicity 2 · the freedom ale -- a witnessing event

    Still constrained: the master controlled where the freedman lived and whom he could marry; he could not leave the master's property without permission; his children could not inherit unless he had married a freedwoman and BOTH had given their ale.

    ceiling · Consistently regarded as inferior in the codes -- whether he committed an offence or was the victim of one.

  4. 3Freedman with further freedoms earned or grantedleysinginot ownedT3

    publicity 3 · further public grants and further work

    Whatever has not yet been earned or granted still binds. The substance does not enumerate what is lifted first.

    ceiling · NO TERMINAL RUNG IS ATTESTED. The codes never make him freeborn: a freedman was consistently regarded as inferior. The ladder has no recorded top, and inventing one would be the easiest error in this table.

the operative variable

THE OPERATIVE VARIABLE IS NOT OWNERSHIP. IT IS PUBLICITY. Ownership ends at rung 1 and the ladder continues for two more rungs, so ownership cannot be what the rungs measure. Status is a function of how many people saw the act.

Marriage banns and a landnam claim have the identical structure: a private fact that becomes a legal fact only once witnessed. The ale is not a celebration attached to the grant, it is the mechanism of the grant.

falsifier · THE NATURAL EXPERIMENT IS RUNGS 1 AND 2. Same grant, same grantor, same loss of ownership -- different audience, different rank. A code in which an unwitnessed grant conferred the SAME rights as one confirmed by the freedom ale would kill the publicity thesis outright and put ownership back in the operative slot.

IIIThe boundary rule — a supply constraint in vestments

Chattel slavery was phased out in Europe under the doctrine that Christians should not enslave other Christians. As Europe became increasingly Christian, pagan slaves could no longer be acquired. The doctrine does not say slavery is wrong. It says the pool is smaller.

The enslaveability boundary as a moving line, 10th to 13th centuryA time axis running from the tenth to the thirteenth century. A boundary line steps downward at Denmark's conversion around 965 and Iceland's around 1000, and descends again along a dashed, undated segment for Sweden; above the line is Latin Christendom, below it the enslaveable pool, which narrows but never closes. An arrow leaves the shrinking pool and enters a band at the foot of the figure marked as the pagan Baltic rule, which admits any captive including Christians. In 1226 an arrow runs upward out of that band, across the Christian line, to a marker for Sweden.THE LINE MOVES · THE POOL SHRINKS · THE TRADE RELOCATESINSIDE THE LINE — LATIN CHRISTENDOM“Christians should not enslave other Christians” · T2?BEFORE THE LINEany captive —no bar attested · T3Denmark ~965 · Harald Bluetooth declares itIceland ~1000 · by arbitration at the Althing1226 — THE REVERSALSaaremaa takes the newlyChristianised SwedenSwedenOUTSIDE THE LINE — THE ENSLAVEABLE POOLit narrows as the line descends. it never closes.THE TRADE DOES NOT STOP — IT RELOCATESthe Viking slave trade ends mid-11th c.; the route is takenover by pagan Baltic traders · T2THE PAGAN BALTIC RULEany captive, Christians included — everything above this line is takeable · T296510001104115411641226conversion year — steps the linearchdiocese — A DIFFERENT CLOCK, and it does not step the lineNorway and Sweden have NO conversion year in the substance. The dashed descent is that gap, not a date.NOT DRAWN: the competing economic explanation is T0 — unsourced, quarantined, not publishable.
the boundary, not the wall — one line moves, a second line never did

The doctrine does not say slavery is wrong. It says the pool is smaller. Chattel slavery was phased out in Europe under the rule that Christians should not enslave other Christians, so as Europe became increasingly Christian, pagan slaves could no longer be acquired. The proof that this was a boundary and not an abolition is that the trade moved: when the Viking slave trade ended mid-11th c., the Baltic-to-Black-Sea route via the Russian rivers was taken over and maintained by pagan Baltic traders, and became the only remaining slave-trade route in Europe after the western markets declined in the 12th c. T2 — modern scholarship, item 4 of the research brief; the underlying scholarship is not named in the substance.

1226 is the row that makes the mechanism visible. Pagan Baltic pirates from Saaremaa raided the newly Christianised Sweden and carried off women and girls to sell. The raiders became the raided within two centuries of conversion, and the mechanism is the boundary itself: Sweden crossed the line and landed on the enslaveable side of somebody else's version of the same doctrine. Conversion does not remove a polity from the slave trade; it moves it from one side of a boundary to the other, and the other side is only safe if everyone around you holds the same boundary. T2 — item 5 of the research brief; the underlying chronicle is not named, which is why this is graded T2 and not T1.

The weakest line on the page is the leftmost one. That the pre-conversion Norse rule admitted any captive with no coreligionist bar is inferred, not attested — the boundary rule only bites once Europe is Christian, which presupposes the prior rule had no such bar. It is drawn dotted and faint because it is T3, inferred from practice. Any located pre-Christian Norse rule limiting who could be taken would replace it outright.

What the axis deliberately does not carry. Conversion and archdiocese are two clocks, and abolition is a third that is not on this figure at all. Denmark's abolition year — the load-bearing datum that would decide whether doctrine or economics drove the end of thraldom — is MISSING. It has not been located, it has not been estimated, and it must not be interpolated from Sweden's 1335 or from the regional statement that slavery had disappeared in most Scandinavian countries by the 14th c.

The competing explanation is stated, not shown. An economic account — land abundance rather than doctrine, with manumitted thralls treated as leaseholders in the late 12th-c. Norwegian collections — is coherent, specific, and unsourced. It is machine-generated material with no citable basis in what was retrieved, so it is held at T0, quarantined, and refused entry to this figure until it is traced to a named scholar. That refusal is the finding, not a gap in the drawing.

Norse, pre-conversiona cultural zoneany captiveT3

Any captive -- no coreligionist bar is attested.

pre-conversion · inferred from practice

Latin Christendoma doctrinal scopenon coreligionist onlyT2

Non-coreligionist only: Christians should not enslave other Christians.

the phase-out runs to the decline of the western markets in the 12th c. · modern scholarship

the pagan Baltic tradersa trade networkany captiveT2

Any captive, Christians included.

mid-11th c. onward; demonstrated in 1226 · modern scholarship

One field, opposite tokens. The boundary is the engine of this lane because it is the only field on which two parties can hold incompatible values and both act on them at the same time.

the proof it was a boundary and not an abolition

When the Viking slave trade ended mid-11th c., the Baltic-to-Black-Sea route via the Russian rivers was taken over and maintained by PAGAN BALTIC traders, and became the only remaining slave-trade route in Europe after the western markets declined in the 12th c.

A SUPPLY CONSTRAINT WEARING THEOLOGICAL CLOTHES. If the doctrine had said slavery is wrong, the route would have closed. It said the pool is smaller, so the route changed hands. Displacement, not abolition -- and displacement is a falsifiable prediction the abolition reading does not make.

whenViking slave trade ends mid-11th c.; the route persists as the only remaining slave-trade route in Europe after the western markets decline in the 12th c.
whereThe Baltic to the Black Sea, via the Russian rivers
whoPagan Baltic traders, taking over from the Viking-age operators
gradeT2modern scholarship · item 4 of the research brief (underlying scholarship not named in the substance)
falsifierEvidence that the route also closed in the 11th-12th c., or that its later operators were Christian, would break the displacement argument and let the Christian doctrine be read as an abolition after all.

year_from is NULL and date_text carries the hedge VERBATIM. 'Mid-11th c.' is not 1050; writing 1050 would create a precision the substance does not have and a later reader could not tell apart from a sourced year.

IV1226 — the raiders become the raided

In 1226 pagan Baltic pirates from Saaremaa raided the newly Christianised Sweden and carried off women and girls to sell. The raiders became the raided within two centuries of conversion.

the mechanism

THE BOUNDARY ITSELF IS THE MECHANISM. Sweden crossed the line and landed on the ENSLAVEABLE side of somebody else's version of the same doctrine. Conversion does not remove a polity from the slave trade; it moves it from one side of a boundary to the other, and the other side is only safe if everyone around you holds the same boundary.

directionPagan raiders take Christian captives -- the reverse of the direction the Viking-age trade ran in.
whereSaaremaa (Osel) and Sweden
whoPagan Baltic pirates from Saaremaa; the newly Christianised kingdom of Sweden
gradeT2modern scholarship · item 5 of the research brief (the underlying chronicle is NOT named in the substance, which is why this is graded T2 and not T1)
falsifierA source showing the 1226 raiders were themselves Christian, or that the captives were not taken for sale, would remove the boundary mechanism and leave an ordinary raid.

THE SINGLE BEST ROW IN THE LANE. It is the reversal that makes the boundary rule visible: the doctrine has two sides, and a polity does not get to choose which side it is on.

VThe lag — conversion on one clock, unfreedom on another

If doctrine drove the end of thraldom, conversion date should predict it. The table below is that test laid out, and most of it is empty. Every cell is read from the database; the two lag columns are computed in the database as generated columns and are simply printed here, so a lag cannot exist where an end date does not.

The lag between conversion and the end of thraldom, drawn as four timelines — most of which have no end.A timeline from 950 to 1350 with one horizontal line per polity. Denmark converts about 965, takes an archdiocese in 1104, and its last thralls are attested as existing in 1201; the line then runs on and frays, because no end date for Danish thraldom is held in this register. Iceland converts about 1000; slavery is last attested as existing in 1133 but remains legally permitted under Gragas until 1270, and the 137 years between those two marks are shaded, because legal permission and social existence are not the same fact. Norway and Sweden have no conversion year in this register at all, so their lines fray at the left instead: Norway runs from an archdiocese in 1154 to 1274, Sweden from an archdiocese in 1164 to an official ban in 1335. Only Sweden has a hard mark at both ends of any single clock. The figure argues that the doctrinal test cannot be run on this evidence: two lines have no beginning and three have no agreed end.95010001050110011501200125013001350YEARDenmarkTHE OPEN CASENO END DATE IN THIS REGISTERCONVERTS965ARCHDIOCESE1104LAST ATTESTED1201Iceland137 YEARS — PERMITTED, NOT ATTESTEDCONVERTS1000LAST ATTESTED1133PERMITTED UNTIL1270NorwayNO CONVERSION YEAR HEREARCHDIOCESE1154PERMITTED UNTIL1274SwedenNO CONVERSION YEAR HEREARCHDIOCESE1164PERMITTED UNTIL1335BANNED1335Two of four lines have no left end. Three have no agreed right end. The doctrinal test needs both.An archdiocese is institutional consolidation, not conversion — a different clock, drawn as a different mark.A shaded band is a span in which the code still permits what nobody records as happening.
the lag — 4 polities, and the ends that do not close
polityconvertsarchdioceseendslaglag / arch
Denmark~9651104missingmissingmissing
Iceland~1000missingmissingmissingmissing
Norwaymissing1154missingmissingmissing
Swedenmissing11641335missing171 y

lag is end minus conversion. lag / arch is a DIFFERENT CLOCK — end minus archdiocese — and the two must never be read as one. An archdiocese is institutional consolidation, not conversion.

why each blank is blank

Denmark · end date*** THE LOAD-BEARING MISSING DATUM OF THIS ENTIRE LANE. *** The Danish abolition date is NOT KNOWN HERE. It has not been located, it has not been estimated, and it must not be interpolated from Sweden's 1335 or from the regional 14th-c. statement. lag_years is therefore NULL for Denmark, which is the correct value and not a gap to be filled.
Iceland · archdioceseNOT IN THE SUBSTANCE. Item 6 lists archdioceses for Denmark, Norway and Sweden only.
Iceland · end dateNOT IN THE SUBSTANCE. Item 7 gives only a REGIONAL statement -- slavery had disappeared in most Scandinavian countries by the 14th c. -- which is not a dated act for this polity and must not be turned into one. 'Most' is also not 'all', so the statement does not even establish that this polity is inside it.
Norway · conversionNOT IN THE SUBSTANCE for this lane. Item 6 gives Norway's ARCHDIOCESE (1154) and no conversion year. The archdiocese is institutional consolidation, not conversion, and substituting one for the other would manufacture a date.
Norway · end dateNOT IN THE SUBSTANCE. Item 7 gives only a REGIONAL statement -- slavery had disappeared in most Scandinavian countries by the 14th c. -- which is not a dated act for this polity and must not be turned into one. 'Most' is also not 'all', so the statement does not even establish that this polity is inside it.
Sweden · conversionNOT IN THE SUBSTANCE. Item 6 says only that Sweden HELD OUT LONGEST -- politically fragmented, with deep-rooted pagan strongholds such as Uppsala -- and gives no year. The 1164 archdiocese is a DIFFERENT CLOCK and is not a conversion date.

a stated lag is not a computed one

Sweden. Item 8 states the Swedish lag runs ON THE ORDER OF THREE AND A HALF CENTURIES. THIS TABLE DOES NOT COMPUTE THAT. With conversion_year NULL, lag_years is NULL by construction, and the stated figure is carried here as TEXT so it can never be read back as a derivation. To compute it, somebody must source a Swedish conversion year.

VIThe one number that was supposed to decide it

This lane was built around a hole. The block below is the model it was built on, printed as it was written; the block after it is that model being withdrawn. Both rows are live in the register and neither has been deleted, because the thing worth reading here is not the answer but the correction.

Read them in that order. What the retraction takes away is the DECISIVENESS, not the hole: the end date for Denmark is still not held here, and what arrived in its place is a last attestation, which is a different question.

the open question · Denmark

Denmark converts ~965. The year thraldom ended there is not held in this register.

DENMARK IS THE DECIDING CASE. If DOCTRINE drove abolition, conversion date should predict abolition date. If ECONOMICS drove it, population density and land availability should predict it and conversion date should not. Sweden does NOT discriminate: converted last, abolished last, consistent with either. Denmark converted FIRST by a wide margin (~965), so if Danish thraldom persisted to a comparable late date, the doctrinal explanation is badly damaged and the economic one survives. One number decides the lane, and it is the one number that is missing.

*** THE LOAD-BEARING MISSING DATUM OF THIS ENTIRE LANE. *** The Danish abolition date is NOT KNOWN HERE. It has not been located, it has not been estimated, and it must not be interpolated from Sweden's 1335 or from the regional 14th-c. statement. lag_years is therefore NULL for Denmark, which is the correct value and not a gap to be filled.

Converted roughly 370 years before Sweden's ban. Whether the abolition gap is 150 years or 370 is exactly what is unknown.

superseded — and kept, because it was wrong in a useful way

🔴 DENMARK IS NOT THE DECIDING CASE AND I SAID IT WAS. Conversion order and economic-integration order are the same order here, so with n=3 both hypotheses predict identically. Withdrawn as a discriminator in v48.

last attested1201 — Bishop Absalon's will, 1201 — last attested EXISTING thralls.
legally untilmissingNo code date located that ENDS it. Absalon's will of 1201 manumits several of his own thralls; a letter between the Pope and Anders Sunesen is the other attestation. After the early 13th c. they stop being mentioned. ⚠ SEE rejected_value / the Codex Holmiensis contradiction.
what it becamecontractual tenancy
gradeT2Brink, Thraldom (Oxford 2021); Absalon's will

The withdrawal takes the outcome variable with it. The measurable question is not when slavery ended — nobody agrees — but when unfreedom stopped being heritable and became contractual. On every row in the register, what it became is the same thing.

and this is why the end date is still not settled

Jyske Lov / Codex Holmiensis (1241) shows slavery still existing in Denmark

UNRESOLVED CONTRADICTION, NOT A REJECTION. The National Museum of Denmark states the practice was apparently still in existence in 1241 because the Codex Holmiensis of that year contained rules regarding slaves. Another source states Jyske Lov of 1241 has NO mention of any slavery in existence. Codex Holmiensis IS the Jyske Lov manuscript — two accounts of one text, flatly opposed. It may reconcile (rules about FORMER thralls read as rules about thralls, the same ambiguity Norway shows) but as stated it does not.

refuse on · It sits exactly on the Danish end-date, so any Danish date that leans on 1241 inherits the contradiction. RESOLVE BY READING THE TEXT, not another summary.

the case where the two clocks can actually be compared · Iceland

permitted until 1270 · last attested as existing 1133 · 137 years apart

the codeSlavery remains PERMITTED under Gragas, which applies until 1270.
the practiceKristinna laga thattur, 1122-1133 — the last text mentioning slavery as EXISTING. No existing thralls appear from the late 12th c.
gradeT2Gragas (Finsen); Kristinna laga thattur

🟢 THE DISCRIMINATING CASE. Converted within a generation of Norway and finished ~a century earlier, so conversion is not doing the work alone. And the 137-year gap between legal permission and last attestation is a finding in itself: a register reading law codes as evidence of practice gets Iceland wrong by more than a century.

Legal permission and social existence are two columns, not one. It is the only row in the register where both are populated, which is precisely why it is the row that forced them apart.

VIIWhere the register disagrees with itself

Two passes over this material are live in the database and they do not agree. The conflicts below are not written down anywhere — they are computed by joining the two tables at export, so a third pass that resolves one of them will make it disappear from this list without anybody editing this page.

Denmark · when it ends

polity_conversion · MISSING

thraldom_status · last attested as existing 1201

Norway · conversion year

polity_conversion · MISSING

thraldom_status · 1030

Norway · when it ends

polity_conversion · MISSING

thraldom_status · legally permitted until 1274

Sweden · conversion year

polity_conversion · MISSING

thraldom_status · 1164

That number is its own archdiocese year (1164) — a different clock, recorded as a different clock, and carried here as a conversion date.

Iceland · when it ends

polity_conversion · MISSING

thraldom_status · legally permitted until 1270

VIIIThe gap — what is not a finding here

not publishable · stored, quarantinedT0thraldom_event#3

There is a competing explanation for the end of thraldom, and it is not the doctrinal one. It is stated here as a gap in this register, never as a result of it.

LAND ABUNDANCE, NOT DOCTRINE. In Norway the maintenance costs of chattel slaves outweighed returns versus incentivising tenants; by the late 12th c. the Gulathing and Frostathing collections increasingly treated manumitted thralls (leysingar) as smallholders or leaseholders (leiglendingar) with heritable rights to farmsteads in exchange for dues, with a parallel move in the Icelandic Gragas toward tenant units. On this account abolition is an accounting outcome and conversion is co-occurrence.

the provenance problem

THIS CAME FROM GROKIPEDIA, WHICH IS MACHINE-GENERATED. It matches standard historiography but has NO CITABLE BASIS in what was retrieved. It enters the register UNSOURCED (T0) or not at all, until traced to Iversen, Karras or Brink directly. Saying so IS THE POINT: this is exactly the class of plausible-and-unattributed material that passes because it sounds right, and the only defence is a register that will hold a row nobody may cite.

whenlate 12th c. (Gulathing and Frostathing); the Gragas parallel is undated in the substance
whereNorway; Iceland
evidencenoneno evidence class, no source reference. Both are null in the row, and the schema refuses to mark it publishable while they are.
falsifierIT IS ALREADY FALSIFIABLE AND THAT IS NOT THE PROBLEM: if economics drove abolition, population density and land availability should predict abolition date and conversion date should not. The problem is that nothing here can be checked against a source. Tracing it to Iversen, Karras or Brink directly would move it to T2 and let it be tested.

publishable = 0, evidence NULL, source_ref NULL. The row is stored so the competing explanation is not silently dropped from the lane, and it is quarantined so it cannot be quoted as a finding. A stated gap, never a finding.

covered and empty · the Norse law codes · Enslaveability boundaryT0thraldom_doctrine#7

The brief carries the ladder and the boundary rule from different places. It says nothing about whom the Norse codes themselves permitted to be taken, so the field is stored UNKNOWN rather than borrowed from the pre-Christian row above.

T0, UNPUBLISHABLE, and deliberately present. An absent row reads as an oversight; a row that says UNKNOWN is a research task with an address.

IXThe five fields, by scope

The comparable token, not the prose. A dash is NOT COVERED — nobody has asked. An UNKNOWN cell is covered and empty, which is a research task with an address, and is the only kind that ever gets done.

fieldthe pagan Baltic tradersLatin Christendomthe Norse law codesNorse, pre-conversion
Status heritability——heritable legal T2divine hereditary T3 contested
Manumission form——assembly and church T2—
Freedman obligation——perpetual T2—
Inheritance right——conditional on both parents freed T2—
Enslaveability boundaryany captive T2non coreligionist only T2UNKNOWN T0any captive T3

XNumbers that circulate and should be refused

A register that only holds right answers cannot stop a wrong one arriving twice. These are held so the next pass recognises them on sight.

Icelandic abolition of slavery in 1117

1117 is the year GRAGAS WAS FIRST WRITTEN DOWN. It is a codification date, not an abolition. Slavery remained permitted under that same code until 1270 — the number is not merely wrong, it is off in the direction that makes the doctrinal story look stronger.

seen in · circulates in several secondary sources

refuse on · Any Icelandic abolition date earlier than the last attestation (1133) or earlier than the code's own expiry (1270). Reject 1117 on sight.

Jyske Lov / Codex Holmiensis (1241) shows slavery still existing in Denmark

UNRESOLVED CONTRADICTION, NOT A REJECTION. The National Museum of Denmark states the practice was apparently still in existence in 1241 because the Codex Holmiensis of that year contained rules regarding slaves. Another source states Jyske Lov of 1241 has NO mention of any slavery in existence. Codex Holmiensis IS the Jyske Lov manuscript — two accounts of one text, flatly opposed. It may reconcile (rules about FORMER thralls read as rules about thralls, the same ambiguity Norway shows) but as stated it does not.

seen in · National Museum of Denmark vs. a second summary

refuse on · It sits exactly on the Danish end-date, so any Danish date that leans on 1241 inherits the contradiction. RESOLVE BY READING THE TEXT, not another summary.

About this title

This appendix is now titled Thraldom, naming the historical subject directly. Earlier links still reach it, and its earlier models and corrections remain part of the account.

Every figure on this page is read from war-atlas/db/atlas.db through seed-v46-thraldom-doctrine.py and seed-v48-thraldom-retraction.py, exported 2026-09-20. Nothing is computed in the page: the lag columns are generated columns in the database, so a lag cannot be written where an end date is missing. Every number on /war/thraldom comes from here. The page computes no history. Two seeds are live and they disagree; both are emitted and the conflicts are computed.

← the contents