The Symbolic World
families · doctrinal-attribution

relation family

doctrinal-attribution

assigned, not verified What this family covers, what it fails to cover, and every relation string in the graph that lands in it. A family is assigned: two models read the definitions document and named the same kind of relation, or a person ruled where they did not. Verified is the other layer: a quote checked letter-exact against its paragraph and judged by a model of a different family. That was already done before any of this began.

119 assigned relations 82 effective relation strings 66 queued items carry a string this family covers 2 declared pairs
graph-wide: 1669 assigned · 354 awaiting a ruling · 65 no family · 3 text undecided · 2091 relations

Definition

Quoted from docs/FAMILY_DEFINITIONS.md v0.7.2, family 21. The atlas quotes that document; it never paraphrases it.

A carrier names, defines, translates, teaches, interprets or classifies — the relation records an act of predication by someone, not an ontology between subject and object.

Boundary: the decisive test — remove the carrier and the claim collapses. Voice often co-varies but never decides (rule 3). Tradition-to-tradition transmission is derivation; a single carrier's predication is attribution.

A carrier is "named" by definite reference, not only by proper name: (v0.7, ruling 5): "this remarkable text explains…" identifies a carrier; collective ascription ("the alchemists say") stays with the content family unless the claim collapses without the ascription. Genre note, from the external check: dictionary-genre sources state relations carrier-free, so dictionary evidence systematically under-reports attribution — weigh attribution questions toward discursive sources.

Attribution outranks the surface verb.: When a NAMED carrier performs the predication, the family is attribution no matter what the relation string says — a symbolizes/symbolized-by/parallels string does not send the edge to symbolization or correspondence if the quote's claim is the carrier's act of reading. Worked example: the-opus symbolized-by peregrinatio (14 §199, "Maier imagined the opus as a wandering") is doctrinal-attribution, not symbolization — Maier's act; remove Maier and this quote's claim collapses. Contrast a symbolization edge whose quote asserts the image relation with no carrier named in the quote itself.

Relation strings that land here

proposes 9defines 5symbolizes 5denies 3describes 3interprets 3names 3adopts 2affirms 2amplified-by 2articulates 2characterizes 2and 70 more, 78 relations between them

The number beside a string is how many relations carry it. 62 of these strings occur exactly once. The tail is the informative part: a string used once usually carried its own qualifier in the sentence it came from. A string can appear under more than one family, and does: assignment is made to an edge in its paragraph, never to a string in the abstract.

seeds: the examples the definition names

declared pairs

A declaration licenses a row in the other direction when the query panel's closure toggle is on, and nothing else. It is read at query time and never written into the graph.

pending a ruling: items carrying a string this family covers

Nigredo amplified-by The Shulamite CW 14 §592
arms-disagree
Coniunctio amplified-by Tifereth & Malchuth CW 14 §18
arms-disagree
Rebis / Hermaphrodite amplified-by Anthropos / Adam Kadmon CW 14 §652
arms-disagree
The shadow amplified-by Sol niger (black sun) CW 14 §172
arms-disagree
Sol amplified-by Gold (aurum) CW 14 §134
arms-disagree
Rex (the King) amplified-by Lion (leo) CW 14 §409
arms-disagree

60 more. These are edges whose two model arms did not agree, or whose agreement the calibrated gate disputed. They carry a relation string this family covers, which is not the same as belonging here. The ruling has not been made, and the queue is not a backlog to be cleared by machine.

Checked against readers of Jung

Checked against 8 works of scholarship: across the sample, 8 of 16 treated items agreed with our family. This family: 1/2 agreed, 1/2 disagreed, 0/2 split, 1 not treated.the whole table
familyagreesdisagreessplitnot treated
archetypal-identity (incl. 26)0/31/32/32
correspondence (incl. 27)2/31/30/32
symbolization3/30/30/31
doctrinal-attribution1/21/20/21
amplification0/33/30/30
controls (personif., union)2/20/20/20
All8/166/162/166

A row this small is the items themselves; no rate over a handful of items carries precision worth reporting. The check asked which kind of relation a reader of the literature would say Jung asserts, not whether Jung's claim is true, which the verification chain had already asked of the paragraph. The disagreements stand beside the agreements because hiding them would make the table an argument rather than a record. Full worksheets: docs/experiments/external_validity_2026-08/.

Agreement is not endorsement: no scholar reviewed this vocabulary, and the sample is small enough to name in a sentence.

receipts counts read from seed.json sha256:9dc668bfff36 via web/family-map.mjs;
definition and seeds quoted from docs/FAMILY_DEFINITIONS.md v0.7.2;
external-validity row from docs/experiments/external_validity_2026-08/results.json (report v3, closed 2026-08-12);
built 2026-09-05

These families are this project's classification. They are calibrated inside this repository, checked once against outside scholarship on a small sample, and endorsed by nobody. What the layer records is how we sorted relations Jung wrote: a map of assertions, not a reading of him.