world/regions/angkor_khmer/page/angkor_khmer_S20.md
CANON SUBORDINATION — this document is PROPOSAL-TIER: it serves canon and never outranks it.
The canon: the CVD · the T1 foundation docs · the T0 registries (registries/) · the spine
(docs/spine/CH_*.md) · the region pages (_source/02_Tier_2_Region_Pages/). Authority order:docs/DOC_MAP.md§ 0.
Canon served (scanned from this document's own citations — widen it by hand where it is thin):
T0_Creature_Roster·T0_Source_Material_Registry·_source/02_Tier_2_Region_Pages/angkor_khmer.md·_source/02_Tier_2_Region_Pages/bali.md
READ THAT CANON FIRST — open it and derive from it before you build anything from this document.
If this document disagrees with canon, CANON WINS and this document is the defect — fix the
document, never the canon. Nothing here is applied until it is ratified into canon.
This section is the build-gating record for the region: which source feeds which authoring session,
what state the retrieval layer is in, and what is honestly still closed. Pass-2 changes one thing
materially. The page was cut with no processed source at all; the angkor-khmer shelf now carries ONE,
deep-read and audited — SRC_00322, Ing-Britt Trankell's ethnography of the Jahed Cham Muslim community
of Kampong Chhnang, Cambodia (fieldwork 1996-2002 with Jan Ovesen), centred on the cay/cai royal-spirit
possession-and-healing cult. Its audit verdict is RATIFY-WITH-CUTS-AND-HOLDS: of the 18 typed rows in
the four status-bearing classes, 13 advance, 3 are HELD by critic MAJORs and 2 were cut, with a further
2 civilizational_marks carrying no status field at all and resolvable only from the record's own
care_findings. The Region Identity header was refreshed to name SRC_00322 at this assemble pass; the deep-research
grounding it names is not superseded, because the new source is a modern ethnography and touches almost
none of the 1200 frame.
SRC_00322 is a journal chapter whose PDF and print numbering differ by a fixed offset, re-derived
across all 12 pages at audit: print page N = PDF page N minus 242. The article runs print pp. 243-254.
CITE PRINT PAGES throughout, in the page's own convention — [SRC: SRC_00322, p. 248]. Two consequences
for a session at the retrieval console. First, the stored chunk metadata carries PDF-relative
page_start and page_end, so a retrieval hit that reports p. 1 is print p. 243; convert before you cite,
or the citation will be off by 242. Second, PDF p. 1 carries the article title block and no printed
numeral, so print p. 243 — cited only by L3_02 — is inferred from the offset rather than read off the
page. It resolves, and the audit confirmed every other citation on the record resolves to the print
page it names, so this shelf carries no anchor drift of the kind the bali and sumatra passes found.
The T0_Source_Material_Registry declares SRC_00322 at 23 chunks, ocr_method native_text — no
OCR-recovery path, nothing quarantine-flagged. The 23 chunks are present in book_substrate under
distinct chunk_ids, so this shelf did not take the chunk_id-collapse failure. The region tag applied by
the 2026-07-19 substrate retag is angkor_khmer with an underscore, while the extraction shelf and the
manifest use angkor-khmer with a hyphen; a region-filtered retrieval must be issued against
angkor_khmer or it returns nothing, and the substrate-region ruling means region-filtering is
mandatory, not optional. Two registry-hygiene items owed at the assemble pass: the row's
extraction_status still reads deep_read_extracted where the record reads deep_read_audited, and the
row's source_title is the filename stub 443 1 4217 1 10 20140825 rather than a bibliographic title
while source_author and source_year are blank — so a title lookup against the registry returns a
plausible-looking non-title, and the record's provenance note is where the real bibliography lives.
The plan is per-session, not per-source: a session opens one section, pulls the named rows, and stops
at the section boundary. Nothing below reaches the ~1200 CE present frame as scene material — the cult
as documented is a post-Khmer-Rouge phenomenon, so this source's home is the ELEVATED modern-Cham block
and the not-yet-pinned modern-revisit node, with 1200-frame use limited to witness-depth or to
separately-cited period-plausible antecedents.
whose Islam layers over a worldview named simply custom, prapeni; the cult's near-erasure in the
mahandori and its post-1980s revival to near-weekly ceremonies; the bilateral chour-sambour kinship
scrutinised in divination; and the aristocrat-warrior versus riess caste memory levelled by the war
[SRC: SRC_00322, p. 245-246, p. 249, p. 251-252]. EC_03 the On g'nur, the Cham religious-political
leader who mourns the mahandori as the loss of the true ritual and dismisses the revival as theatre
[SRC: SRC_00322, p. 246], and EC_04 the mediums — predominantly female, some male, some children
initiated as mediums — join the same session [SRC: SRC_00322, p. 245, p. 251-252]. L3_03, the
mediums' own memory that UNTAC revived all the spirits that had disappeared, is Section 19's
[SRC: SRC_00322, p. 252].
chief among them venerated as the teacher, their history sung in the pi cay
[SRC: SRC_00322, p. 245, p. 247, p. 250-251]; and L3_01, the Hindu-substrate royal-ancestor cult
surviving inside a Muslim congregation [SRC: SRC_00322, p. 245-246]. EC_01 routes to
T0_Creature_Roster at species_class folklore with the §2.6c authenticity firewall already enforced on
the row: living documented tradition, not a Western cryptozoology construct, revered and LEFT.
collected herbs in the forest, boiled them, and prescribed drinking the medicine for seven days
[SRC: SRC_00322, p. 248]. The gather-boil-timed-regimen loop is fully grounded; the plant list is not,
and the audit's MINOR stands — no plant is named anywhere in the source. A Section-2/14 session must
bring a separate ethnobotanical citation for lowland Cambodia before it writes named materia, and the
source scopes the kru khmer to spirit afflictions specifically, so any widening to bodily illness is
the build's inference and not the source's scoping.
raised before the host's house, with banners and weapons set out as signs of power, women selling
snacks along the edges, and children keeping the royal orchestra's beat on bamboo stems laid on the
ground [SRC: SRC_00322, p. 247, p. 249-250]; and S03, the exorcism-divination in which the possessing
spirit refuses to give its name and the elders rule it an ordinary ancestor spirit rather than a true
cay [SRC: SRC_00322, p. 248, p. 250]. Both render at reference register: the consecrated performance
is referenced, never reproduced. S03 carries a standing authorial firewall — the un-named-spirit motif
is the tradition's own ritual mechanic and must never bleed into the antagonist reveal ladder in any
surfaced text.
name against the roster: the Cham are known in Khmer society to practise magic and are sometimes
visited by Khmer wanting to buy their knowledge [SRC: SRC_00322, p. 252]. It ships with the same
page's both/and or it becomes the sinister-sorcerer-race trope — the Cham are also known for
solidarity and kindness extended to neighbours, poverty and sorcery accusation are long-linked, and
the community's own self-naming on that page is neak sre neak prey, people making a living from
fields and forest.
corpus in the Cham's own language and ancient script, now displaced by free standardised prayer-books
[SRC: SRC_00322, p. 245-246]. The name akhar thrah is correct but is NOT attested in this source; it
comes from this page's own British Library EAP531 revival citation, and a session must attach that
citation rather than letting a Swedish ethnography stand as the name's source. EC_05's culture is
Cham-wide, not Jahed — the script is not a sub-group's property.
it, the theatre-state reading in which processions and signs of power constitute authority in the act
of performing it [SRC: SRC_00322, p. 243, p. 249, p. 251]. This is grounding for the page's own
ritual-kingship staging, never a player lecture.
Five audit corrections each session carries rather than re-derives. Four are citation-coverage: add
print p. 249 to S04 (chour-sambour), p. 251 to EC_01 (male and child mediums), p. 251 to L3_02 (Bell
1997:136) and p. 244 to L3_03 (the UNTAC dating). The fifth is attribution: S04's care_flag conflates
two self-naming communities as the Jahed/Cham Bani strand, which this source does not attest — name the
Jahed alone and ground any Bani distinction on the page's own Po Dharma citation. Separately, the two
civilizational_marks on this record carry no status field at all, so a Section-4 or Section-16 session
reads the record's care_findings before consuming either: both were verified and kept, with one wording
trim — the pi cay preserves Cham mythical history and the founding myth of the Cham King, not a
dynastic history.
The specificity that must reach the surface undiminished is a routing instruction, not a licence to
trim: the record's care_findings enumerate it in full, and a session that softens it is committing the
scrub rather than exercising care. None of it is protected content and all of it is the community's own
— the On g'nur's contempt for extremely unintelligent mediums, the mediums' open factional fight and
the demand that a sitting Imam abandon his office, the transgendered gender-twice-over reading, the
child mediums, the caste memory, the spirit-stories older mediums guard from younger ones, the sorcery
accusations between envious neighbours [SRC: SRC_00322, p. 248], and the money, of which the failed
90,000-Riel heart medication is the anchored instance [SRC: SRC_00322, p. 248]. The same rule binds
causation. The documented causes here are structural and community-internal — aid vehicles leaving
nothing but dust [SRC: SRC_00322, p. 244], pressure to conform to Arab Muslim standards and free
prayer-books displacing the hand-copied ones [SRC: SRC_00322, p. 244-245], the long-established link
between poverty and sorcery accusation [SRC: SRC_00322, p. 252], and the military's claim on as much as
a third of the rice-fields [SRC: SRC_00322, p. 253] — and the true-cause clause forbids replacing them
with an invented puppet-master. The 1200 frame's antagonist routing through BE_0005_S4 does not travel
into the modern block.
until they are re-audited. S01, the Cham healing-path questline, imported a period arogyasala
temple-hospital that the source never contains — the only hospital in twelve pages is the modern
Calmette hospital in Phnom Penh [SRC: SRC_00322, p. 248]. The fix is re-label, not removal: state the
modern triad as the source has it, mediums and kru khmer and hospital, and give any Angkorian
temple-hospital its own separate citation. S06 inverted its own cited page, which attests the King of
Champa going to war rather than a later loss, with defeat and conquer returning zero hits across the
source; the direction of that correction must survive so the error is not repeated, and the eventual
Vietnamese conquest is real history that is not to be hidden but needs its own citation. W04 asserted
a Khmer-Cham neak ta link the source does not carry — neak ta occurs exactly once, at print p. 248,
where the beings named are Chinese ancestor spirits angered by a mosque built over a former Chinese
cemetery. The Khmer-Cham sharing of that register is probably true and worth having; it needs a cited
Khmer-side attestation from a different source.
model rather than a cited tradition and which is correctly homed at L3_02 instead, and the
religious-layering parallel, whose far end names no tradition and carries no citation. Section 15
does not resurrect either without a new source.
that gap; it remains an open build-time RCM addition, and this page still rests largely on the four
deep-research files (angkor_city, khmer_cham_war, cham_living, khmer_myth_care) rather than on
processed substrate.
Trankell is an outside Swedish ethnographer writing about the Cham, however sympathetically, and the
obligation names Po Dharma and the living-community record. Landing this record must not be read as
discharging it.
so the region_pages gate's use_class rule bites here for the first time: every SRC id cited on the
page must be declared in Section 16 on a line carrying its use_class in explicit bracket form. Every
other section simply cites normally.
Build-ready date: still gated on the MUST-ACQUIRE hard authorities (Higham, Coedes, Maspero, Golzio,
Po Dharma) for the Angkor and Champa depth, on the RCM Khmer/Cham section, and now additionally on an
ethnobotanical source for lowland Cambodia before the Section-2/14 Herbalism node can name its plants.
The pre-existing deep-research population plan stands unchanged for the 1200 frame: Sections 1, 7 and 9
draw the Greater Angkor Project, Groslier and Evans hydraulic and mapping anchors with the
Coedes/Chandler/Higham Khmer frame; Sections 4 and 5 draw the Kulke/Vickery devaraja correction and the
Kaundinya-Soma and Reamker traditions, with Po Dharma and the My Son and Po Nagar record on the
elevated Cham surfaces; Section 6 draws the Nautical Angkor vessel iconography and the Bayon and
Banteay Chhmar relief corpus; the modern-Cham session draws Po Dharma, Minority Rights Group and
Kiernan at witness-depth, now joined by SRC_00322 as its first processed source. The machine layer
(Zone Catalog, Derived Indexes, the FORK_CH05_A / BE_0005_S4 / scene rows) is build-ready now against
the v3.1 spine. The v1.0 book-cost fields are superseded by the closed acquisition ledger.