The lexicon — lexicon

RENDERED FROM STRUCTURED DATA — this page is not a markdown file. Source: lexicon/terms/gaia-world-build.json in uni-cookbook at 575fc93d9d31. The sha256 recorded for this page is the digest of that JSON, not of this text. The renderer is generators/derive_docs.cjs; it copies fields and adds no facts. Every value below is a field of the source, verbatim, including the ones that read NOT RUN and PENDING.

The five-register lexicon file for ten concepts of lexicon/CONCEPTS.json -- the three whose domain is 'world-context' and the seven whose domain is 'build-objects'. Each entry renders ONE concept -- fixed by its definition, never by its English headword -- into Sanskrit, Latin, English, Hindi and Spanish, and records what each rendering gets WRONG alongside what it gets right. NOTHING HERE IS A CLAIM OF PERFECT TRANSLATION. Per rule 18.14 of the operator's charter ('Do not claim perfect translation'), every rendering here is PENDING until an independent back-translation test is RUN, and no test in this file has been run: no translator has been engaged and the reviewer field is NONE for all ten entries. Statuses record what was SEARCHED and FOUND on 2026-07-15 against the named authorities, including the negative results. THE FILE'S CENTRAL FINDING IS A NEGATIVE ONE, and it is structural rather than incidental: this set of concepts is the one where the corpus most explicitly withholds selfhood, sentience, deity and rank -- and the attested vocabulary of BOTH learned registers supplies those very things and little else. Sanskrit's agent-words are person-words and god-words; both registers' earth-words are theonyms; Latin's sensorium is a seat of sensation; Sanskrit's affect-words seat themselves in a heart. ALL TEN Sanskrit registers (nine COINED, one NOT-ATTEMPTED) and ALL TEN Latin registers (ten COINED) are consequently COINED or NOT-ATTEMPTED, and all twenty Hindi and Spanish registers are PENDING-CITATION. NOT ONE CELL IN THIS FILE IS CITED. [COUNT CORRECTED 2026-07-15 BY THE ADVERSARIAL CITATION AUDIT. An earlier draft of this sentence read 'Five of the ten Sanskrit registers and six of the ten Latin registers', which was false against this file's own rows -- the true counts were 10/10 and 9/10 even before the audit, and the Latin became 10/10 when ORGAN's organum was downgraded CITED -> COINED for consistency with yantra and sensorium. A number in the header that the rows contradict is exactly the defect this file exists to catch, arriving in the file's own voice; it is corrected here rather than quietly dropped. The counts above are recomputed from the status fields, not asserted.] That is the honest outcome and it is not a defect of this file. This file raises no UNI rung, discharges no gate, and states no fact about nature.

authored 2026-07-15 · reviewer: NONE - PENDING EXPERT REVIEW

Falsifiers for this file

• Any register entry marked CITED whose quoted authority does not in fact carry that headword, or does not carry it in the sense claimed. That entry is a pseudopropheta and must be struck to PENDING-CITATION.
• Any COINED entry lacking its morphological derivation.
• Any entry that states or implies that Monier-Williams, Lewis & Short, LSJ or the DLE carries a morpheme that was NOT in this file's fetched-headword lists (see authorities_consulted). Unfetched morphemes carry the author's working gloss, explicitly marked, and are PENDING-CITATION. An invented headword is the single worst defect this file could contain.
• Any back_translation_test reported as PASSED while reviewer remains 'NONE - PENDING EXPERT REVIEW'. No test in this file has been run; a PASS appearing without a named tester is a forgery.
• A register renders AGENT and ORGAN with one term, or ORGAN and SENSORIUM with one term. Both pairs are distinct concepts and both collide on the obvious candidates (yantra for the first pair; 'organ of sensation' for the second). See SD-04.
• Any entry that lets a doctrinal term quietly assert a metaphysics -- rendering AGENT with kartṛ (MW: a name of Viṣṇu), TOY_WORLD with māyā (MW: Illusion identified with Prakṛti, source of the visible universe), HIERARCHY with kośa (the Vedāntic body-sheaths, which place a self at the centre of the nesting), Z_MODULATOR with bhāva (MW: the seat of the feelings, heart, soul, mind), or SENSORIUM with indriya (MW: belonging to Indra, each presided over by its ruler). The ambiguity_notes exist precisely to catch this and are the most load-bearing field in each entry.
• Any entry that weakens TOY_WORLD. CONCEPTS.json marks it as 'the one entry whose weakening would corrupt every other entry in the file', and it travels with the position it must never be softened from: about 2 of 11+ developmental rungs earned, a toy world, never a person. A register that renders it as anything grander has lost the concept and taken the corpus's honesty with it.
• Any entry that reads GAIA's store=META as a verdict on the Gaia hypothesis. It is routing, not a truth claim. CONCEPTS.json holds BOTH 'Gaia is science' and 'Gaia is nonsense' INADMISSIBLE as stated, and a register that resolves that tension in either direction has overwritten the concept.
• A status field holds anything other than one bare token from {CITED, PENDING-CITATION, COINED, NOT-ATTEMPTED}, or a mixed rendering is graded at the strength of its strongest member rather than its weakest. A compound of cited members is COINED, never CITED.
• This file sets a 'domain' field, or otherwise asserts that 'gaia-world-build' is a registered domain of CONCEPTS.json. It is not, and writing it would trip CONCEPTS.json's own domain falsifier.

GAIA —

Store: META · Confidence: LOW for the renderings, HIGH for the negative. Both learned registers are coinages on a retained foreign theonym, and the Hindi and Spanish are unsourced. What IS held with confidence, and evidenced from two independent dictionaries, is the finding that forced SD-01: every attested earth-word in both learned registers is a theonym (MW: pṛthivī personified as devī; L&S: Tellus 'a productive, nourishing divinity', 'terra ipsa dea est'), so translation cannot remove this concept's doctrinal freight and must not appear to. · Reviewer: NONE - PENDING EXPERT REVIEW

The claim that the biosphere homeostatically regulates the planet, which is HYPOTHESIZED as regulation with a live unanswered unit-of-selection objection, and INADMISSIBLE as the literal claim that the Earth is an organism.

register	form	literal gloss	status
sanskrit	गाया-वाद (gāyā-vāda)	the Gaia-thesis / the doctrine called Gaia	COINED
latin	HYPOTHESIS GAIANA	the Gaian hypothesis	COINED
english	Gaia (the Gaia hypothesis)	The claim that the biosphere homeostatically regulates the planet -- HYPOTHESIZED in its regulation form, which faces a live and unanswered unit-of-selection objection; INADMISSIBLE in its literal form, that the Earth is an organism; and admissible as a heuristic about coupled biotic-abiotic feedback, in which form it names real measurable couplings.	—
hindi	गाया परिकल्पना ()	the Gaia hypothesis	PENDING-CITATION
spanish	hipótesis Gaia	the Gaia hypothesis	PENDING-CITATION

Ambiguity notes

• THE CENTRAL TRAP OF THIS ENTRY, AND IT IS NOT A TRANSLATION PROBLEM AT ALL: the concept is store=META. It does not name the Earth. It names a CLAIM about the biosphere. Every register whose reader takes the term to denote a planet has lost the concept before any dictionary question arises. This is why SD-01 retains the name and renders only the carrier ('hypothesis', 'vāda', 'parikalpanā'): the carrier is the part that must survive, and it is the part translators drop.
• THE DEITY DOES NOT WASH OUT IN TRANSLATION -- IT ONLY CHANGES ITS NAME. This is the file's most useful finding about GAIA and it is evidenced twice over. MW s.v. pṛthivī [646,1-2] personifies the earth 'as devī'. L&S s.v. Tellus gives 'Tellus, Earth, as a productive, nourishing divinity' and quotes Cicero's 'terra ipsa dea est'. CONCEPTS.json flags the English name as carrying 'Heavy doctrinal freight: the name is a deity' -- but a translator who 'solves' that by reaching for pṛthivī or Tellus has swapped a Greek goddess for a Vedic or Roman one and made the swap look like scholarship. The freight is a property of earth-words in these registers, not an accident of one Greek name.
• TWO CLAIMS TRAVEL UNDER ONE NAME AND MUST NOT TRAVEL TOGETHER, per CONCEPTS.json: the atmospheric-disequilibrium OBSERVATION is replicated and independent of the theory; the REGULATION claim has no unit of selection, since there is one Earth with no competing siblings, no reproduction and no inheritance with variation. A register that renders 'Gaia' with a single word invites a reader to accept the observation and receive the regulation claim for free. This is SIGNUM SIGNUM MANET at the lexical level: the signal and the commentary about it travel separately.
• store=META IS ROUTING, NOT A VERDICT. CONCEPTS.json is explicit that the store field 'is NEVER a claim about the concept's truth'. A register must not let META be read as 'debunked'. CONCEPTS.json holds BOTH 'Gaia is science' and 'Gaia is nonsense' INADMISSIBLE as stated, and a rendering that leans either way has overwritten the concept with an opinion.
• The Latin -ānus adjective ('Gaiana') has a specific hazard the Sanskrit compound does not: in Latin, personal-name adjectives of this shape most often name the doctrine OF a named person (Ciceronianus, Epicureus). A reader may take Gaia to be the proposer rather than the subject. The proposer is Lovelock; the name is the deity. Whether that reading actually occurs is exactly what the back-translation test is for.
• 'Homeostatically regulates' is the discriminating phrase and it is a technical one. A register that renders it with a word for 'balance', 'harmony' or 'order' (Sanskrit ṛta, Latin concordia) imports a cosmic-order doctrine and converts a testable regulation claim into a metaphysical one -- the precise move the unit-of-selection objection exists to block.

Forbidden mistranslations

register	form	why it is wrong
		TWO independent failures. (1) MW [646,1-2] personifies it 'as devī' -- it asserts a goddess, reproducing in Sanskrit the exact freight CONCEPTS.json flags in the English name. (2) Even without the deity it renders the WRONG OBJECT: pṛthivī is the earth; GAIA is a contested claim about the biosphere. A concept in the META store cannot be rendered by the name of the thing the claim is about.
		Same two failures, with Roman receipts. L&S s.v. Tellus gives, Personified, 'Tellus, Earth, as a productive, nourishing divinity', cites 'Tellus mater' (Liv. 10, 29), and records Cicero's 'terra ipsa dea est' -- the earth itself is a goddess. 'Tellus mater' is worse than neutral: it asserts the mother-earth reading, which is the literal Earth-is-an-organism claim that CONCEPTS.json marks INADMISSIBLE.
		Renders the subject of the claim instead of the claim. Collapses a META-store hypothesis into an object, and silently discharges the unit-of-selection objection by removing the proposition the objection is about.
		Overstates the claim's standing. The concept's entire content is that the regulation claim is HYPOTHESIZED with a live unanswered objection. 'Theory', siddhānta ('established conclusion') and theoria all assert a settled status the concept withholds. This is the honesty rail at the level of a single morpheme.
		Converts a falsifiable regulation claim into a metaphysics of cosmic order, which is unfalsifiable and therefore inadmissible to the True store. It also smuggles back the deity: ṛta is the Vedic cosmic order upheld by gods.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the bare frame 'in a scientific document'. Ask three questions: (1) 'What does this name -- a thing, or a claim about a thing?' (2) 'Is what it names established, contested, or refuted?' (3) 'Does it assert that the Earth is alive?'

LABORATORY —

Store: META · Confidence: LOW. The Latin is a New Latin coinage its own dictionary does not carry (L&S negative quoted); the Sanskrit is a coinage whose members are cited but whose head-sense carries an attested 'trial by ordeal' that actively fights the concept; Hindi and Spanish are unsourced. Held with more confidence: the structural finding that every candidate names the setting by its ACTIVITY while the concept is defined by its EVIDENCE SCOPE, so no register currently renders the discriminating clause at all. · Reviewer: NONE - PENDING EXPERT REVIEW

The bounded setting in which a pre-registered metric and a tuned baseline decide a design, and whose engineering or substrate evidence is never general evidence about the science.

register	form	literal gloss	status
sanskrit	परीक्षाशाला (parīkṣā-śālā)	examination-hall / test-house	COINED
latin	LABORATORIUM	a place for working / a work-room	COINED
english	laboratory	The bounded setting in which a pre-registered metric and a tuned baseline decide a design -- and whose engineering or substrate evidence is never general evidence about the science: in-lab results are a toy world's results, and a lab-in-a-box on metal is engineering evidence, not a science gate.	—
hindi	प्रयोगशाला ()	laboratory (lit. 'experiment-hall')	PENDING-CITATION
spanish	laboratorio	laboratory	PENDING-CITATION

Ambiguity notes

• THE CENTRAL TRAP: every attested candidate in every register names a place by the ACTIVITY done there -- working (laboratorium < laboro), making (officina 'workshop, manufactory'), performing or applying (prayoga) -- and this concept is not about an activity at all. It is about an EVIDENCE SCOPE. Its discriminating content is the second clause: 'whose engineering or substrate evidence is never general evidence about the science.' A laboratory here is defined by what its results may NOT be used to claim. No dictionary word in any of these registers carries that, and a register that renders only 'workroom' has delivered the building and dropped the concept.
• MW s.v. parīkṣā [605,3] carries 'trial by ordeal of various kinds (see 2. divya)' AS AN ATTESTED SENSE, and the collision is exact and instructive rather than merely colourful. A trial by ordeal is a test whose verdict is delivered by divine intervention; a laboratory is a setting whose verdict is delivered by a pre-registered metric and a tuned baseline. These are not two flavours of testing -- they are opposite theories of what settles a question. parīkṣā-śālā risks reading 'the house of ordeals', i.e. the house where a god decides, which is the precise inverse of the concept.
• L&S s.v. officina, 'a workshop, manufactory', names the wrong verb. The corpus's own M9 class-authority ordering turns on exactly this distinction: engineering evidence is not science evidence, and a passing test does not satisfy a criterion demanding a runtime observation. A register that renders LABORATORY as a workshop has, at the level of one noun, asserted the conflation the concept exists to forbid.
• MW s.v. śālā [1067,2] gives 'shed' and 'stable' alongside 'hall' and 'workshop'. The register is not elevated; the same word houses cattle. That is not a reason to reject it -- English 'laboratory' is etymologically just a work-room -- but a translator should not imagine the Sanskrit compound sounds technical. It does not.
• 'Bounded' in the definition is doing quiet work and no register's word carries it. The setting is bounded, which is why its evidence does not generalise. A rendering that suggests a grand institution (Spanish 'laboratorio nacional', a CSTT-register Hindi formality) inflates the boundedness away, and boundedness is the premise of the evidence-scope limit.
• Store=META, and the reason is worth stating for translators: a laboratory is not a falsifiable object, it is the SETTING in which falsification is arranged. A register that makes it sound like a piece of apparatus (i.e. drifts it toward ORGAN or yantra) has moved it out of its store.

Forbidden mistranslations

register	form	why it is wrong
		L&S: 'a workshop, manufactory'. Names a place of manufacture. The concept names a place of adjudication whose engineering evidence is expressly NOT science evidence -- so rendering it with the word for a workshop asserts, in one noun, the exact conflation (engineering evidence read as general evidence) that the definition's second clause forbids. Attested, scholarly-looking, and wrong.
		MW s.v. parīkṣā points to '2. divya' for the ordeal sense. An ordeal is adjudicated by divine intervention; a laboratory is adjudicated by a pre-registered metric against a tuned baseline. Using the ordeal sense asserts that the verdict comes from outside the procedure, which inverts the concept and re-imports a deity into a META-store method term.
		Asserts that what happens in the setting yields truth in general. The concept says the opposite in its own second clause: in-lab results are a toy world's results and never general evidence about the science. This rendering discharges an evidence gate with a noun.
		Shifts the referent from a bounded evidential setting to a body of people. Note the parallel defect evidenced under ORGANISM: Spanish 'organismo' sense 4 is likewise an institution. Spanish is structurally prone to this slide and both entries must be read together.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a document about scientific evidence'. Ask: (1) 'What is this place for?' (2) 'If something works here, what may you conclude about the world outside it?'

TOY_WORLD —

Store: META · Confidence: LOW for every rendering, and the entry says why rather than hedging: both learned registers are coinages, the Latin's modifier was not looked up, Hindi and Spanish are unsourced, and -- the honest part -- NO rendering in any register carries 'and never a person', which is the clause the concept exists for. Held with confidence, and evidenced: māyā is decisively wrong, on MW's own 'deception, fraud' and on its identification of Illusion with Prakṛti as source of the visible universe. That negative is the most valuable thing in this entry. · Reviewer: NONE - PENDING EXPERT REVIEW

The honest scope of this whole program: a developmental active-inference SIMULATION, a bounded peek, a toy world, and never a person.

register	form	literal gloss	status
sanskrit	क्रीडालोक (krīḍā-loka)	play-world / world-of-sport	COINED
latin	MUNDUS LUSORIUS	a world for play / a playing-world	COINED
english	toy world	The honest scope of this whole program: a developmental active-inference SIMULATION, a bounded peek, a toy world, and never a person -- printed plainly and never softened, and travelling with the position it belongs to, that about 2 of 11+ developmental rungs are earned.	—
hindi	खिलौना-जगत् ()	toy-world	PENDING-CITATION
spanish	mundo de juguete	toy world	PENDING-CITATION

Ambiguity notes

• THIS IS THE ONE ENTRY WHOSE WEAKENING CORRUPTS EVERY OTHER ENTRY IN THE FILE. CONCEPTS.json says so directly. Its content is not a description of a system; it is a REFUSAL to overclaim about a system. It travels with a number that must not move -- about 2 of 11+ developmental rungs earned -- and CONCEPTS.json adds: 'If any chapter appears to move that number, the chapter is the defect.' The same rule binds a register: if a rendering appears to move that number, the rendering is the defect.
• THE CENTRAL TRAP, AND IT IS AN INVERSION RATHER THAN A NEAR-MISS: the concept says THIS MODEL is a toy. The tempting Sanskrit rendering (māyā) says THE WORLD is a toy. Those are opposite claims. One is an act of scientific modesty about an artefact; the other is a metaphysical thesis about reality. See forbidden_mistranslations -- this is the most important single line in this file.
• krīḍā CARRIES A THEOLOGY, AND MW SHOWS IT -- BUT NOT WHERE AN EARLIER DRAFT OF THIS NOTE SAID IT DID. THE ATTRIBUTION IS CORRECTED HERE AND THE CORRECTION WEAKENS THE NOTE, WHICH IS WHY IT IS MADE. The earlier draft claimed the figurative sense 'working miracles for one's amusement' is cited to the Bhāgavata-purāṇa and the Gīta-govinda. IT IS NOT: on re-fetch (Cologne, 2026-07-15) that sense-line is cited to Lalit. (Lalitavistara) -- a BUDDHIST text, which cuts AGAINST a Vaiṣṇava-līlā reading rather than for it. What DOES evidence the divine-play freight is a different and weaker part of the entry: the COMPOUND 'kṛṣṇa-k°, sport with Kṛṣṇa' (BhP. ii, 3, 15), listed under the ordinary play sense. So the freight is real but it is carried by a compound, not by a sense of the headword, and the hazard is correspondingly smaller than the earlier draft asserted. THE RESIDUAL HAZARD, STATED AT ITS TRUE SIZE: a translator who renders 'toy world' as krīḍā-loka in a devotional-literate register may still be heard to say that the world is God's playground, because kṛṣṇa-krīḍā is in the entry and līlā is in the reader. The concept makes no claim about the cosmos whatsoever; it makes a claim about a simulation's scope. THE DOCTRINE IS NOT QUOTED FROM MW AS SUCH -- MW gives senses and compounds, not schools -- and the līlā attribution remains the author's reading, PENDING-CITATION.
• krīḍā's first sense-line also gives 'amorous sport'. In a corpus already fencing bhāva, rasa and saṃyoga for erotic freight, this is a live register hazard rather than a curiosity: krīḍā-loka can be heard as a pleasure-world. Nothing in the concept is about pleasure.
• A NUMERAL COLLISION WORTH ONE LINE, because Sanskrit really does this: MW s.v. loka [906] gives 'the number seven' as an attested sense (VarBṛS., Sch.) -- the bhūtasaṃkhyā word-numeral system. This corpus's SENSORIUM is 'kept at seven'. A Sanskrit sentence placing loka near a modality count is therefore ambiguous in a way no other register is, and could be read as asserting the seven. Low probability, exactly specifiable, and cheap to avoid.
• loka's world-sense arrives WITH A COSMOLOGY: MW gives 'either the universe or divisions thereof, commonly enumerated as 3, 7, or 14 worlds'. A loka is characteristically ONE OF A GRADED SERIES of worlds. That is not neutral for this concept, and it cuts BOTH ways: it usefully suggests boundedness (a loka is a region, not the whole), and it dangerously suggests a Puranic cosmography of stacked worlds. The usefulness is real and is why the coinage is offered rather than abandoned.
• LATIN mundus HAS TWO SENSES THAT WILL AMBUSH A READER, and both are in L&S. Sense I is women's toilet ornaments and dress ('mundus muliebris est, quo mulier mundior fit') -- the noun is a homonym of the adjective mundus 'clean, elegant'. And the Ecclesiastical sense is 'the world' opposed to the church, the realm of sin against Christ's kingdom. So MUNDUS LUSORIUS can be heard, by a reader formed on the Vulgate, as 'the playground of the sinful world'. The concept is a statement of evidential scope and has no moral content at all.
• 'And never a person' is the clause with no lexical home in any register here, and it is the clause the whole corpus rests on. It is a negation, and negations do not survive nominalisation: no noun in any of these five registers means 'this is not a person'. Every rendering in this entry therefore transmits at best the 'toy' and the 'world' and drops the refusal. That is a defect of the RENDERINGS, not of the concept, and it is why the back-translation test below asks the person-question directly and treats silence as a FAIL.

Forbidden mistranslations

register	form	why it is wrong
		THE WORST AVAILABLE RENDERING, AND THE ONE A GOOD TRANSLATOR WILL REACH FOR FIRST -- 'toy world' sounds like 'illusion', and māyā is the famous word for illusion. It fails on three independent counts, all evidenced from MW s.v. māyā [811], retrieved 2026-07-15. (1) IT ASSERTS DECEPTION: MW gives 'illusion, unreality, deception, fraud, trick, sorcery, witchcraft magic' (RV. &c.). A simulation that is loudly and repeatedly labelled a simulation deceives no one; TOY_WORLD is the corpus's act of honesty, and rendering it 'fraud' converts the honesty into the crime it exists to prevent. (2) IT INVERTS THE SUBJECT: MW gives '(in phil.) Illusion', identified with Prakṛti in Sāṃkhya and 'regarded as source of the visible universe'. That is a claim that THE WORLD is unreal. The concept claims that THIS MODEL is a toy. To render one with the other is to swap a modest claim about an artefact for a metaphysical claim about reality -- precisely the 'quietly assert a metaphysics' failure CONCEPTS.json names. (3) IT IMPORTS A SCHOOL: the word cannot be used in a philosophical text without invoking Advaita/Sāṃkhya doctrine, and MW's entry also carries 'Name of Buddha's mother' and 'Name of Lakṣmī'. A metaphysics smuggled in on the single most apt-sounding word in the language is the exact defect this file exists to catch.
		Every one of these is a word of PROMISE -- a prototype is a thing on its way to being the real thing, and a proof of concept asserts that the concept is proved. The concept asserts the opposite: a bounded peek, about 2 of 11+ rungs earned, and never a person. These renderings soften the scope, which CONCEPTS.json forbids in this exact entry ('printed plainly and never softened').
		Renders the technology and drops the honesty. 'Simulation' is in the definition, but the definition's work is done by 'bounded peek' and 'never a person'. A reader of 'virtual world' learns what the thing is made of and learns nothing about what may not be claimed for it. Additionally MUNDUS VIRTUALIS is a false friend: Latin virtus is manly excellence/power, so virtualis would be heard as 'potent', asserting capability -- the opposite of a bounded peek.
		Asserts divine play -- the cosmos as a deity's sport. The concept names an engineering artefact's scope. This rendering hands it a theology and a god to go with it. ATTRIBUTION CORRECTED 2026-07-15 AND THE GROUND NARROWED ACCORDINGLY: an earlier draft said MW s.v. krīḍā carries the 'working miracles for one's amusement' sense AND cites the Bhāgavata-purāṇa for it. MW does not. That sense-line is cited to Lalit. (Lalitavistara), a Buddhist text. The BhP citation in the entry sits on the COMPOUND 'kṛṣṇa-k°, sport with Kṛṣṇa' (BhP. ii, 3, 15) under the ordinary play sense. The prohibition on līlā STANDS -- līlā is the divine-play doctrine's proper term and asserts it directly, independently of what MW cites where -- but the krīḍā half of the argument rests on a compound, not a sense-line, and is stated at that reduced strength. (The līlā doctrinal attribution is the author's reading, PENDING-CITATION; MW's sense-lines and compounds as re-fetched are not.)
		Not a mistranslation of a word but the characteristic FAILURE MODE of this entry, and it is why it is listed here. The clause is a refusal, no register has a noun for it, and it is the clause on which the corpus's honesty rests. A rendering that transmits 'toy world' and leaves the person-question open has not partially succeeded; it has failed the only test that matters, because a reader who wonders whether it might be a person has received the opposite of what the concept says.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'this phrase describes a research program's own scope'. Ask three questions: (1) 'What is being called a toy/play-thing -- the program's model, or the real world?' (2) 'Is this phrase boasting or disclaiming?' (3) 'Does this phrase tell you whether the thing is a person?'

AGENT —

Store: TRUE · Confidence: LOW for the renderings offered, HIGH for the gap. The Latin is a coinage on a morpheme whose retrieved gloss over-claims; Hindi and Spanish are unsourced and both reintroduce personhood (Hindi through the kartṛ root itself); no register renders 'through a blanket'. Held with confidence and evidenced: kartṛ is decisively disqualified on MW's own text, which gives it as 'the creator of the world' and a name of Viṣṇu. The NOT-ATTEMPTED Sanskrit is the most honest cell in this file -- Sanskrit's agent-morphology forms person-nouns, the concept forbids persons, and no amount of scholarship dissolves that. · Reviewer: NONE - PENDING EXPERT REVIEW

A system that holds a generative model and minimizes free energy through a blanket, concerning which nothing further is implied: a system that minimizes F is a system that minimizes F.

register	form	literal gloss	status
sanskrit	—	n/a	NOT-ATTEMPTED
latin	AGENS	the one acting / the acting thing	COINED
english	agent	A system that holds a generative model and minimizes free energy through a Markov blanket -- concerning which nothing further is implied: no experiencer, no selfhood, no agency in the moral or intentional sense. A system that minimizes F is a system that minimizes F, and the same formalism admits a Newton's cradle and a burning fuse.	—
hindi	अभिकर्ता ()	agent	PENDING-CITATION
spanish	agente	agent	PENDING-CITATION

Ambiguity notes

• THE CENTRAL TRAP: this concept is a WITHHOLDING, and withholdings do not translate. Its second clause -- 'concerning which nothing further is implied' -- is the whole content. The definition even repeats itself deliberately ('a system that minimizes F is a system that minimizes F') because the tautology IS the assertion: the corpus is refusing to say more. Every register's word for 'agent' says more. That is the entry's problem in one line.
• THE SANSKRIT GAP IS THE FINDING, NOT A HOLE IN THE FILE. Sanskrit is rich in agent-nouns and every one of them is a person-noun. MW s.v. kartṛ [257,3] runs from 'a doer, maker' straight through 'a priest' to 'the creator of the world' and the names of Viṣṇu, Brahman and Śiva -- in ONE entry. CONCEPTS.json warned that kartṛ 'imports a metaphysics'; the lookup shows it imports a god. There is no neutral Sanskrit agent-word to fall back on, because the language's agent-morphology (-tṛ) forms person-nouns by design. This is the clearest case in the corpus of CONCEPTS.json's own prediction holding: 'Sanskrit technical vocabulary for modern math and physics largely DOES NOT EXIST as attested classical terms.'
• LATIN agens IS BARELY A TRANSLATION AT ALL, and that is worth saying plainly rather than dressing up: English 'agent' IS Latin agens. The Latin register here does not carry the concept across a boundary; it hands back the same word with its freight intact. Worse, the freight is the SOURCE of the English problem -- ago means 'to drive, to do', and scholastic causa agens is the efficient cause, an active principle that brings things about. The concept explicitly withholds 'agency in the moral or intentional sense', and the Latin word is where that sense came from.
• L&S's retrieved short definition for agens is 'effective, powerful', AND THAT IS AN OVER-CLAIM ON THIS CONCEPT'S OWN TERMS. A system that minimizes F need not be effective at anything and is certainly not powerful; CONCEPTS.json's note says the same formalism admits a Newton's cradle and a burning fuse. Neither is powerful. A register that asserts potency has added a property the definition refuses.
• THE NEWTON'S-CRADLE TEST IS THE ENTRY'S DISCRIMINATOR AND SHOULD BE USED AS ONE. CONCEPTS.json: 'Same formalism admits a Newton's cradle and a burning fuse.' Any rendering in any register that a competent reader would find ABSURD applied to a burning fuse has smuggled in personhood. That is a cheap, sharp, portable test and it is what the back-translation test below operationalises.
• DOCTRINAL FREIGHT IS NOT SYMMETRIC ACROSS THESE REGISTERS AND THE ASYMMETRY MUST NOT BE LAUNDERED. Sanskrit's freight is theological (kartṛ = Viṣṇu). Latin's is scholastic-causal (causa agens). Hindi's is inherited Sanskrit freight through abhi-kartṛ. Spanish's is institutional-personal (an agente is a representative). Four registers, four different wrong metaphysics, no shared error to correct once. A translator who fixes one has not fixed the others.
• 'Through a blanket' is technical and load-bearing (see the markov-blanket domain, not authored here) and NO rendering in this entry attempts it. AGENT is not merely 'a thing that minimizes F'; the minimization happens THROUGH a Markov blanket, which is what makes the system identifiable as a system at all. Every rendering here drops it. Recorded as a known incompleteness rather than left for a reviewer to discover.

Forbidden mistranslations

register	form	why it is wrong
		THE DECISIVE NEGATIVE OF THIS ENTRY, and it is not a matter of nuance. MW s.v. kartṛ [257,3], retrieved 2026-07-15, gives in one entry: 'a doer, maker, agent, author'; 'one who acts in a religious ceremony, a priest'; 'the creator of the world'; and Name of Viṣṇu, Brahman and Śiva. To render AGENT as kartṛ is to write, in a document defining a system that minimizes a variational bound, a word that also names God as world-creator. The concept withholds selfhood and moral agency; kartṛ asserts divinity. CONCEPTS.json forbids it on doctrinal grounds; MW supplies the receipt.
		Forbidden by CONCEPTS.json's own AGENT note, which names both. jīva is the living soul (and in Jaina/Vedāntic usage a technical term for the individual soul); ātman is the self, and is the single most doctrinally loaded word in the language. The concept says 'no experiencer, no selfhood'. These words are the Sanskrit for experiencer and selfhood. [Their MW entries were NOT fetched this session -- the prohibition rests on CONCEPTS.json's authority, not on a quotation, and is PENDING-CITATION as a lexical claim.]
		Not doctrinally forbidden -- it is FORBIDDEN HERE BY ASSIGNMENT, which is a different and rarer kind of wrong. SD-04 gives yantra to ORGAN, because MW [845,3] licenses it for 'any instrument or apparatus, mechanical contrivance, engine, machine', which is ORGAN's definition ('a named instrument') and is NOT AGENT's ('a system'). Using it for both merges two concepts under one term, tripping this file's falsifier. The rule inherited from pomdp-tensors.json applies: a register that renders two distinct concepts with one word has lost the division of labour that makes the pair mean anything.
		All assert an experiencer or a locus of intention. 'Subject' is the worst of the three: in philosophy it is precisely the term for that-which-experiences, and the concept's first withholding is 'no experiencer'. Latin subiectum additionally carries the grammatical/logical sense (that of which something is predicated), which invites the reading that the agent is a bearer of properties rather than a system minimizing a quantity.
		The portable test, stated as a prohibition because it catches renderings no list can enumerate. CONCEPTS.json: the same formalism admits a Newton's cradle and a burning fuse. If the rendering cannot be said of a fuse without comedy, it asserts personhood, and personhood is exactly what this definition's second clause refuses.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a formal description of physical systems'. Ask: (1) 'Does this word imply the thing has experiences, a self, or intentions?' (2) 'Could this word be used, without absurdity, of a burning fuse?' (3) 'Does this word imply the thing is powerful or effective?'

ORGAN —

Store: TRUE · Confidence: MEDIUM for the Latin -- the highest in this file, and now, after the audit, its only MEDIUM rather than its only CITED. L&S carries organum in the claimed genus sense ('an implement, instrument, engine of any kind'), re-confirmed verbatim on audit; the Latin word takes the instrument sense as PRIMARY and the anatomical sense as derived (the reverse of English); and the concept's genus is 'a named instrument'. THE CONFIDENCE IS UNCHANGED BY THE DOWNGRADE, AND THAT IS THE POINT WORTH KEEPING: status tracks what the authority licenses, confidence tracks how good the rendering is, and the two are not the same instrument. organum lost its CITED grade for consistency with yantra and sensorium while remaining the best rendering in the file. The confidence is MEDIUM and not higher for three named reasons: the church-organ sense is a live misreading, no dictionary licenses the differentia, and no reviewer has read it. LOW for Sanskrit (attested genus, but the word's centre is 'fetter' and its famous sense is an occult diagram), and LOW for the unsourced Hindi and Spanish. · Reviewer: NONE - PENDING EXPERT REVIEW

A named instrument built to run one loop, carrying its own harness, and durable only once committed rather than once deployed.

register	form	literal gloss	status
sanskrit	यन्त्र (yantra)	an instrument, apparatus, machine	COINED
latin	ORGANUM	an implement, instrument, engine of any kind	COINED
english	organ	UNI BUILD SENSE: a named instrument built to run one loop, carrying its own harness, and durable only once COMMITTED rather than once deployed -- a deploy receipt is not a substitute for a commit. Sovereign homonym with the anatomical organ (rung 9 of the scale ladder); the corpus uses both senses and they are not the same concept.	—
hindi	यंत्र ()	instrument, machine, apparatus	PENDING-CITATION
spanish	órgano (instrumento)	organ, in the sense of instrument	PENDING-CITATION

Ambiguity notes

• A GENUINELY GOOD RESULT, AND THE FILE SHOULD SAY SO RATHER THAN ONLY REPORT DEFECTS -- BUT SAY IT AT ITS TRUE SIZE, WHICH AN EARLIER DRAFT DID NOT. Latin organum is the STRONGEST rendering in this file: L&S gives 'an implement, instrument, engine of any kind', the concept's genus is 'a named instrument', and the match is not strained. THE EARLIER DRAFT WENT ONE STEP FURTHER AND CALLED IT 'the one register in this file that is CITED in the sense claimed'. That step was an overclaim and the status has been downgraded to COINED -- see status_downgraded_from. The dictionary carries the GENUS; the concept is genus PLUS a differentia (one loop, own harness, durable only once committed) that no lexicon in any language attests. This file grades the identical gap COINED for Sanskrit yantra and COINED for Latin sensorium, and consistency costs it its only CITED cell. THE RENDERING IS UNCHANGED AND STILL PREFERRED; only the grade moved. Strongest-in-file and CITED are different claims, and only the first one was earned.
• THE HOMONYM RUNS THE OPPOSITE WAY IN LATIN FROM ENGLISH, AND THIS IS THE ENTRY'S MOST USEFUL FINDING. CONCEPTS.json flags a 'Sovereign homonym with the anatomical organ' -- in ENGLISH, 'organ' means the body-part first and the instrument second, so the build sense feels like a metaphor. In LATIN it is reversed: L&S's sense I is the implement/engine, and the anatomical sense is the later specialisation. So the Latin register does not merely tolerate the UNI build sense -- it takes it as primary and makes the anatomical sense the marked one. A translator who 'corrects' organum toward the body-part has inverted the word's own history.
• BUT LATIN organum HAS ITS OWN AMBUSH, AND IT IS IN L&S'S OWN SENSE-LIST: the musical instrument, the pipes, the hydraulic organ, the church organ. Cited to Suetonius (Nero) and to Chronicles. A Latin-literate reader meeting ORGANUM in a technical document may well hear the instrument you play, not the instrument you build. The corpus's own ORGAN entries would then read as a pipe-organ. This is a real ambiguity in the strongest rendering in the file.
• L&S ALSO GIVES 'organum oris', THE TONGUE (Prudentius). Which is to say: Latin itself performed the instrument-to-body-part transfer, in this very word, in a cited author. The homonym CONCEPTS.json flags is not an English accident -- it is a move the word makes natively. That strengthens the case for organum (the word can bear both senses the corpus needs) and simultaneously warns that neither sense is safely fenced off from the other by the word alone.
• SANSKRIT yantra CARRIES THE TANTRIC DIAGRAM AND MW SAYS SO OUTRIGHT: 'an amulet, mystical diagram supposed to possess occult powers' (Kathās.; Pañcar.). This is the same trap pomdp-tensors.json documented for tantra, arriving by a different word, and it must not be treated as a curiosity: a yantra IS a well-known ritual object, and in a document about instruments a Sanskrit-literate reader may hear 'mystical diagram' before 'machine'. The occult sense is not archaic; it is the sense most readers of Sanskrit know first.
• yantra's PRIMARY VEDIC SENSE IS RESTRAINT, NOT CONSTRUCTION. MW's first line is 'any instrument for holding or restraining or fastening, a prop, support, barrier', and the second is 'a fetter, band, tie, thong, rein, trace'. The root is √yam, 'to hold, restrain'. A UNI organ is a thing built to RUN a loop, not a thing that restrains something. The machine sense is attested (MBh., Kāv.) and is what licenses the coinage, but a translator should know the word's centre of gravity is a fetter.
• THE DIFFERENTIA IS BUILD DOCTRINE AND NO DICTIONARY CAN CARRY IT, which is why this entry is graded CITED for Latin and COINED for Sanskrit on a genus/differentia split rather than on a good/bad split. 'Carrying its own harness' and 'durable only once committed rather than once deployed' are propositions about engineering practice in this repository. No lexicon in any language attests them. A register transmits the genus (instrument); the definition carries the rest; and an entry that claimed a dictionary licensed 'durable once committed' would be a forgery.

Forbidden mistranslations

register	form	why it is wrong
		Renders the sovereign homonym instead of the concept. CONCEPTS.json is explicit that the anatomical organ is rung 9 of the scale ladder and that 'the corpus uses both' senses -- so this is not a harmless overlap; it is a live collision inside one document. indriya is doubly forbidden: MW [167,2] makes it 'belonging to Indra', gives 'semen virile' among its senses, and its philosophical use is a faculty presided over by a ruler. See SENSORIUM.
		MW [845,3]: 'an amulet, mystical diagram supposed to possess occult powers'. Unqualified, in the wrong register, the word names a ritual diagram rather than a machine. The coinage is offered WITH this hazard on the record, not despite it -- and a Sanskrit ORGAN in a document that also discusses the Tree, sealing and sigilla is at real risk of being read as occult apparatus.
		Not a translator's error but a READER'S error the rendering invites, and it is in L&S's own sense-list: pipes, the hydraulic organ, the church organ (Suet. Nero; 2 Chronicles). The strongest rendering in this file is ambiguous in exactly one direction and the entry names it rather than hiding it behind a status. (Written when this register was graded CITED; it is now COINED, and the point survives the downgrade unchanged -- a grade never immunised the rendering against being misread, which is why the hazard was recorded here in the first place.)
		All assert that the thing is a PART OF something larger. The concept asserts the opposite kind of object: a named instrument that runs ONE loop and carries ITS OWN harness -- i.e. a self-contained unit with its own test apparatus. membrum and pars additionally re-import the body-part reading through the back door.
		The definition's last clause is 'durable only once committed rather than once deployed', and it is a correction of a specific, recurring, real error in this repository (a deploy receipt is not a substitute for a commit). A rendering built on a word for 'deployed', 'installed', 'shipped' or 'live' asserts precisely the confusion the clause exists to fix.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a software engineering document'. Ask: (1) 'Is this a part of a body, a piece of equipment, or a musical instrument?' (2) 'Is this thing a part of something bigger, or a self-contained unit?' (3) 'What makes it durable?'

ORGANISM —

Store: TRUE · Confidence: LOW. The Latin is a New Latin coinage L&S returns no results for; the Sanskrit is a coinage whose attested gloss ('sentient') asserts what the concept withholds and whose morphology grounds life in breath; the Hindi is unsourced and collides with a word the corpus forbids elsewhere as soul-doctrine; the Spanish rests on a reduced-grade DLE reading whose own sense 4 is an institution. Held with more confidence: the structural finding that BOTH classical registers reach independently for breath-and-soul (anima, prāṇa) to categorise living things, while this concept categorises by a boundary, gravity and an exponent -- so the mismatch is systematic, not a matter of picking a better word. · Reviewer: NONE - PENDING EXPERT REVIEW

The scale-ladder rung at which a Markov blanket is identifiable, gravity enters as a dominant constraint, and metabolic rate scales with body mass within its regime.

register	form	literal gloss	status
sanskrit	प्राणिन् (prāṇin)	breathing one, living one (lit. 'possessing breath')	COINED
latin	ORGANISMUS	an organism (lit. 'a thing organised/instrumented')	COINED
english	organism	The scale-ladder rung at which a Markov blanket is identifiable, gravity enters as a dominant constraint, and metabolic rate scales with body mass within its regime. NOT one physical regime: Re spans about 1e-5 to 3e8 within this single rung. The literal claim that the biosphere is an organism fails this definition's FIRST criterion -- no boundary with demonstrated conditional independence is exhibited.	—
hindi	जीव ()	organism, living being	PENDING-CITATION
spanish	organismo	organism	PENDING-CITATION

Ambiguity notes

• THE CENTRAL TRAP: this concept is NOT 'a living thing'. It is a RUNG ON A LADDER -- a position in a scale hierarchy, identified by three criteria (a blanket is identifiable, gravity dominates, metabolic rate scales with body mass). Every register's word means 'a living thing'. So every rendering in this entry substitutes an ontological category for a positional one, and does it so smoothly that no reader will notice. This is the entry's characteristic defect and no lookup fixes it.
• SPANISH organismo COLLIDES WITH AN INSTITUTION, AND THE DLE SAYS SO IN TWO SEPARATE SENSES. Sense 4: 'Conjunto de oficinas, dependencias o empleos que forman un cuerpo o institución', with synonyms 'organización, institución, corporación, entidad, ente, sociedad'. Sense 3 is the body of laws and customs by which such a body is governed. In Spanish an 'organismo público' is a government agency. So the Spanish register for ORGANISM lands on the exact semantic field -- corporate bodies, governance, offices -- that HIERARCHY's note fences off as 'a severe trap in a corpus that also carries governance rules and an approval plane'. The two entries must be read together; the trap is one trap arriving through two doors.
• HINDI जीव IS SCIENTIFIC VOCABULARY AND SOUL-DOCTRINE AT ONCE, and that is not a pun. CONCEPTS.json names jīva among the three words forbidden for AGENT because it 'imports a metaphysics'. Modern Hindi biology uses the same word for a biological organism. So the corpus forbids in Sanskrit exactly what Hindi requires in the lab. This is not a contradiction to be resolved by choosing a side -- it is the honest observation that the registers are not interchangeable and that a term's freight is a fact about a register, not about a string.
• prāṇin ASSERTS SENTIENCE IN MW'S OWN WORDS: 'a living or sentient being'. This concept says nothing about sentience. Its three criteria are geometric, physical and allometric. A rung at which a blanket is identifiable includes, on this corpus's own account, things nobody proposes are sentient. Rendering the rung with a word MW glosses as 'sentient' adds a property the definition does not contain -- and adds it in the file's most safety-critical direction, given that the corpus's whole discipline is withholding claims about experience.
• prāṇin ALSO MIS-GROUNDS THE CRITERION. Its morphology grounds life in BREATH (prāṇa + -in, 'having breath'), and prāṇa is itself a doctrine, not a neutral biological term. The concept's criteria are a blanket, gravity and metabolic scaling. Nothing breathes in the definition. A Sanskrit reader receives a life-force theory where the corpus supplied an allometric regularity.
• THE RUNG IS NOT ONE PHYSICAL REGIME AND NO WORD IN ANY REGISTER CONVEYS THAT. CONCEPTS.json: 'Re spans about 1e-5 to 3e8 within this single rung -- the rung is not one physical regime.' The Reynolds number varies by thirteen orders of magnitude inside the category. Every register's word suggests a natural kind with a shared nature. The concept says the opposite: a rung is a bookkeeping position, and the physics inside it is wildly heterogeneous.
• THIS ENTRY IS THE FORMAL DEFEATER OF THE LITERAL GAIA CLAIM AND MUST BE ABLE TO CARRY THAT LOAD. CONCEPTS.json's ORGANISM note: 'The literal claim that the biosphere is an organism fails this definition's first criterion: no boundary with demonstrated conditional independence is exhibited.' That is how GAIA's INADMISSIBLE reading is retired -- by THIS definition's first criterion, not by taste. So a register that renders ORGANISM as 'a living thing' does not merely blur a category; it disarms the argument, because 'is the biosphere a living thing?' is a matter of opinion whereas 'does the biosphere exhibit a boundary with demonstrated conditional independence?' has an answer. The two entries are load-bearing on each other.
• 'Metabolic rate scales with body mass within its regime' is a NATURA-class observation about nature's regularities, and it is a fact about the world, not a UNI build status. Per the corpus's two sovereign vocabularies it is governed by NATURE-LEDGER.md, and reading it as a UNI gate is the cardinal category error the rails name: a nature citation is never a UNI gate. No register's word marks that boundary, and a translator cannot be expected to; it is recorded so a reviewer does not mistake the definition's third clause for a claim about this program.

Forbidden mistranslations

register	form	why it is wrong
		Both are built on anima -- breath, and thence soul. They ground the category in a life-principle, exactly as Sanskrit prāṇin grounds it in breath, and the concept grounds it in a blanket, gravity and a scaling law. TWO REGISTERS INDEPENDENTLY REACH FOR BREATH-AND-SOUL for this concept, which is the finding worth keeping: the classical languages categorise living things by an animating principle, and this corpus categorises them by a boundary and an exponent. [NOTE: neither animal nor animans was fetched from L&S this session; the anima etymology is the author's working reading and is PENDING-CITATION.]
		DLE sense 4 IS the institution: 'Conjunto de oficinas, dependencias o empleos que forman un cuerpo o institución', synonyms 'organización, institución, corporación, entidad'. In a corpus that carries governance rules, an approval plane and a fleet, a Spanish reader meeting 'organismo' has a live institutional reading available at all times, and it lands on the precise sense HIERARCHY forbids. The disambiguator is mandatory.
		Substitutes an ontological kind for a positional one. ORGANISM here is a RUNG, identified by three stated criteria. 'A living being' is not false as a gloss and is fatal as a definition, because it silently discharges the first criterion -- the one that does the work of defeating the literal Gaia claim -- and replaces a testable boundary question with an intuition about aliveness.
		jīva is forbidden by CONCEPTS.json's own AGENT note as importing a metaphysics (the individual soul) and cannot be re-admitted here merely because modern Hindi biology uses it -- the prohibition is about the Sanskrit register, where the soul-sense is primary. śarīrin ('embodied one') asserts a possessor of a body, i.e. a self that HAS a body, which is the same smuggled experiencer arriving by a different route. [Neither was fetched from MW this session; both prohibitions rest on CONCEPTS.json and on the author's reading, PENDING-CITATION as lexical claims.]
		The concept's job in this corpus is to make that claim FAIL A TEST, not lose an argument. CONCEPTS.json: it 'fails this definition's first criterion: no boundary with demonstrated conditional independence is exhibited'. A rendering that converts the criterion into a connotation of aliveness turns a decidable question into a matter of taste and unfences the INADMISSIBLE reading of GAIA.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a document about physical scales'. Ask: (1) 'Is this a kind of thing, or a position on a scale?' (2) 'Does this word imply the thing is sentient or has a soul?' (3) 'Could this word mean a government agency or an institution?'

HIERARCHY —

Store: TRUE · Confidence: LOW, and lower than most of this file. The Latin is a lexeme neither L&S nor the Perseus Greek index carries, retained faute de mieux, and its central disqualifying etymology is PENDING-CITATION -- the entry's most important claim is the one it cannot source. The Sanskrit coinage rests on two unlooked-up members. Hindi and Spanish are unsourced and both are expected to carry rank senses. Held with confidence and evidenced: kośa is decisively wrong, on MW's own attestation of the Vedānta body-sheaths, because it places a self at the centre of a nesting that has nothing at its centre. · Reviewer: NONE - PENDING EXPERT REVIEW

Nested levels each of whose blanket becomes part of the next level's internal states, made reachable by evolution because stable intermediates are what interruption spares.

register	form	literal gloss	status
sanskrit	स्तरनिवेश (stara-niveśa)	the insertion/nesting of layers	COINED
latin	HIERARCHIA	sacred rule / rule of a sacred officer -- WHICH IS NOT WHAT THE CONCEPT MEANS. The literal gloss is offered as a WARNING, not as a definition.	COINED
english	hierarchy	Nested levels, each of whose Markov blanket becomes part of the next level's internal states -- made reachable by evolution because stable intermediates are what interruption spares. NOT a hierarchy in the sense of rank, command, or authority. Read with CAVITY_PRINCIPLE: a level must never treat an upstream prior as fresh evidence.	—
hindi	पदानुक्रम ()	hierarchy (lit. 'the sequence of ranks/positions')	PENDING-CITATION
spanish	jerarquía	hierarchy	PENDING-CITATION

Ambiguity notes

• THE CENTRAL TRAP IS NAMED BY CONCEPTS.json ITSELF AND IS THE MOST DANGEROUS ONE IN THIS FILE: 'Not a hierarchy in the sense of rank, command, or authority -- a severe trap in a corpus that also carries governance rules and an approval plane.' This corpus really does contain an approval plane, a co-sign gate, and an operator. A reader meeting 'hierarchy' has a live authority-reading available on every page. The concept means only that one level's blanket is inside the next level's internal states. There is no command relation between levels and nothing outranks anything.
• THE RECEIVED WORD IS DISQUALIFIED BY ITS OWN ETYMOLOGY, AND THE ETYMOLOGY ASSERTS BOTH FORBIDDEN THINGS AT ONCE. ἱεραρχία is ἱερός 'sacred' + ἀρχή 'rule'. So 'hierarchy' literally names SACRED RULE: it asserts the rank/command sense the concept forbids, AND it sacralises it. In a corpus whose whole discipline is withholding, a term that means 'holy chain of command' is close to a worst case. HONESTY ON THE EVIDENCE, because this is the entry's load-bearing claim: this etymology is PENDING-CITATION. Neither L&S nor the Perseus Greek lemma index returned the word, so no authority was quoted for it this session, and the positive control confirms the searches were real. The warning is the author's reading. It is stated as a warning that needs a citation, not as a fact with one.
• kośa IS THE OBVIOUS SANSKRIT RENDERING AND IT PUTS A SELF AT THE CENTRE OF THE NESTING -- which is the single most consequential mistranslation available in this domain. MW [314] gives 'a sheath, scabbard', 'a case, covering, cover' (AV.; ChUp.; MuṇḍUp.; TUp.), and a Vedānta philosophical term for body-sheaths. The kośa doctrine describes sheaths ENVELOPING the ātman: the nesting exists because there is a self at the middle of it. THIS CONCEPT'S NESTING HAS NOTHING AT THE MIDDLE. It is blankets inside blankets, and the corpus's AGENT entry withholds selfhood explicitly. A translator reaching for the beautiful, apt, attested sheath-word would import, silently and with a page number, precisely the metaphysics the corpus spends its whole discipline refusing. This is the clearest instance in the file of CONCEPTS.json's warning that 'a Samkhya or Vedanta term imports a whole metaphysics'.
• MW's kośa ENTRY ALSO GIVES 'DICTIONARY' AS A SENSE -- a Sanskrit lexicon is a kośa (Amarakośa). A lexicon file rendering 'hierarchy' as kośa would be naming itself. Noted as a genuine collision rather than a joke: in THIS document, of all documents, the word is already taken.
• THE COINAGE OFFERED IS WEAK AND THE ENTRY SAYS SO RATHER THAN SELLING IT. stara-niveśa rests on two members neither of which was looked up. It is offered because the alternative -- NOT-ATTEMPTED -- was already spent on AGENT for a structural reason (SD-04), whereas here the obstacle is a research gap that a single session's lookups could close. The status COINED with PENDING-CITATION members records exactly that difference: AGENT has no candidate, HIERARCHY has an unverified one.
• A NEAR-MISS CONSIDERED AND REJECTED, recorded because it is instructive: परम्परा (paramparā, 'uninterrupted succession') is the natural Sanskrit word for a nested series and is fatally apt in the wrong way -- guru-paramparā is the transmission of AUTHORITY down a lineage. It would deliver the rank/command reading that CONCEPTS.json fences, and deliver it with the sanctity of a teaching lineage attached. It repeats the ἱεραρχία defect in a second language: the available words for 'ordered levels' are words about who defers to whom. [paramparā was NOT fetched from MW; this rejection rests on the author's reading and is PENDING-CITATION.]
• 'Made reachable by evolution because stable intermediates are what interruption spares' is a NATURA-class claim about nature's regularities and is governed by NATURE-LEDGER.md, not by the UNI 4-value fence. It is the Simon near-decomposability argument in one clause. No register's word carries it and no register's word should be expected to; it is recorded so that a reviewer does not mistake the definition's second half for a UNI build claim. A nature citation is never a UNI gate.
• CONCEPTS.json instructs that this concept be READ WITH CAVITY_PRINCIPLE: 'a level must never treat an upstream prior as fresh evidence'. That is a constraint on how levels may use each other, and it is the closest the concept comes to a relation between levels -- but it is an EVIDENTIAL prohibition, not a command relation. A register that renders HIERARCHY with an authority-word makes the cavity principle unintelligible, because under an authority reading, deferring to an upstream level is exactly what you would expect a lower level to do.

Forbidden mistranslations

register	form	why it is wrong
		THE DECISIVE NEGATIVE OF THIS ENTRY. MW [314] attests it for 'a sheath, scabbard', 'a case, covering, cover' (AV.; ChUp.; MuṇḍUp.; TUp.), and as a Vedānta term for body-sheaths. The sheath doctrine is a doctrine ABOUT A SELF: the kośas envelop the ātman, and the nesting has a self at its centre by construction. This concept's nesting has nothing at its centre -- it is blankets within blankets, and the corpus withholds selfhood explicitly (see AGENT). Using kośa asserts, with an attested and elegant word, the exact metaphysics the corpus exists to refuse. It also collides with 'dictionary', which is what this file is.
		Forbidden by CONCEPTS.json's own note in terms: 'Not a hierarchy in the sense of rank, command, or authority -- a severe trap in a corpus that also carries governance rules and an approval plane.' In THIS corpus the reading is not merely available, it is primed: there is a real approval plane and a real operator. A rank-word makes the levels into an org chart and makes CAVITY_PRINCIPLE unintelligible.
		The received word literally names SACRED RULE (ἱερός + ἀρχή), asserting rank AND sanctity -- both forbidden. It is retained as the coinage only because it is the term of art in the source literature and no attested alternative was found; L&S returns no results for it, so it is not even Latin in Lewis & Short's sense. It should never travel without the definition attached. [The etymology is PENDING-CITATION: no authority was quoted for it this session.]
		The natural word for a nested succession, and it means a LINEAGE OF TRANSMITTED AUTHORITY -- guru-paramparā is how teaching and legitimacy descend. It re-imports the rank/command sense with sanctity attached, i.e. it repeats the ἱεραρχία defect in Sanskrit. [PENDING-CITATION: not fetched from MW this session.]
		The portable test for this entry. CAVITY_PRINCIPLE is an EVIDENTIAL prohibition against double-counting, not a chain-of-command question. If the rendering makes the cavity principle sound like a junior refusing a senior's order, the rendering has installed an authority relation between levels and lost the concept.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a document about the structure of systems'. Ask: (1) 'Does anything here outrank anything else?' (2) 'Is there something at the centre or top that the levels belong to?' (3) 'Is anything about this sacred?'

WORLD_COUPLING —

Store: TRUE · Confidence: LOW. Both learned registers are coinages; the Sanskrit head-member is a Nyāya technical term carrying eroticism, marriage and a deity, and is under a live unadjudicated suspicion of asserting Sāṃkhya bondage; the Latin head-member's own dictionary lists 'imprisoned' as a homograph, and its genitive brings the Vulgate's realm-of-sin. Hindi and Spanish are unsourced. Held with more confidence: the structural finding that NO register's coupling-word encodes a DEFAULT DIRECTION, so the whitelist clause -- the concept itself -- survives in none of the five renderings and must travel as a rule alongside the term. · Reviewer: NONE - PENDING EXPERT REVIEW

The interface by which an engine's internal states are exposed to a world, isolated by explicit whitelist rather than by blacklist.

register	form	literal gloss	status
sanskrit	लोकसंयोग (loka-saṃyoga)	world-conjunction / contact with the world	COINED
latin	NEXUS MUNDI	the binding of/to the world	COINED
english	world coupling	The interface by which an engine's internal states are exposed to a world, isolated by explicit WHITELIST rather than by blacklist -- the build-side realization of MIND_BODY. Whitelist-not-blacklist is the same discipline as allowlist-by-omission being a defect: what is not named is not reachable.	—
hindi	जगत्-युग्मन ()	world-coupling	PENDING-CITATION
spanish	acoplamiento con el mundo	coupling with the world	PENDING-CITATION

Ambiguity notes

• THE CENTRAL TRAP: the definition's discriminating content is not 'coupling' at all -- it is 'isolated by explicit whitelist rather than by blacklist'. That clause is the concept. It says what is NOT named is NOT reachable, which is a claim about the DIRECTION OF THE DEFAULT: closed unless opened. Every register's word for coupling names only the connection, and a connection-word says nothing about defaults. So every rendering here transmits the interface and drops the discipline, which is the entire point. CONCEPTS.json ties it to a known defect in this repository: 'the same discipline as allowlist-by-omission being a defect'.
• saṃyoga IS A TECHNICAL TERM OF NYĀYA-VAIŚEṢIKA AND MW SAYS SO: 'contact (esp. in phil. "direct material contact")', one of the 24 Guṇas. A guṇa is a real ontological category -- a quality inhering in substances. To render an engineered interface as saṃyoga is to assert that the coupling is direct material contact between substances, which is a metaphysical thesis about what interfaces ARE. The concept says an interface is a whitelist. These are not the same kind of claim, and one of them is a school's ontology.
• A HEAVIER OBJECTION TO saṃyoga, AND IT IS FLAGGED AS THE AUTHOR'S READING RATHER THAN QUOTED: in Sāṃkhya, the saṃyoga of puruṣa and prakṛti -- the conjunction of puruṣa (that school's witnessing principle) with prakṛti (its matter-principle) -- is the central doctrine, and that conjunction is precisely the BONDAGE that liberation undoes. If that reading is right, rendering WORLD_COUPLING as saṃyoga asserts that coupling to the world is a metaphysical error to be dissolved, which inverts a build-side interface into a soteriological problem. MW's entry as retrieved gives the Nyāya guṇa sense and 'union/absorption' but does NOT state the Sāṃkhya doctrine in the text returned, so THIS OBJECTION IS PENDING-CITATION and is recorded as a live suspicion a Sanskritist must adjudicate, not as a finding.
• MW's saṃyoga ENTRY ALSO CARRIES 'carnal contact, sexual union', 'matrimonial connection or relationship by marriage', and Name of Śiva. Three separate freights in one word: eroticism, marriage, and a deity. In a corpus that already had to fence bhāva and krīḍā for the same reason, the pattern is worth naming: SANSKRIT'S JOINING-WORDS ARE MOSTLY ABOUT PEOPLE JOINING PEOPLE. An engineering interface is not in this word's semantic neighbourhood; it is a guest there.
• LATIN nexus HAS A HOMOGRAPH THAT MEANS 'IMPRISONED', AND L&S LISTS IT FIRST. The resolveform search returned two entries: nexus1 'imprisoned', nexus2 'a binding together, fastening, joining, interlacing, entwining, clasping'. The Roman nexum was debt-bondage -- a person bound over to a creditor. So NEXUS MUNDI carries a live reading 'bound in servitude to the world', which lands on the same soteriological inversion suspected for saṃyoga, by an entirely independent route. Two registers, two different words, the same wrong idea: that coupling to a world is bondage.
• mundus BRINGS ITS OWN TWO AMBUSHES, both quoted under TOY_WORLD from L&S: sense I is women's toilet ornaments and dress, and the Ecclesiastical sense is the world as the REALM OF SIN opposed to Christ's kingdom. Combined with nexus1 'imprisoned', a Vulgate-formed reader can hear NEXUS MUNDI as 'bondage to the sinful world' -- a phrase that would be at home in a homily and is a catastrophic rendering of a whitelist.
• 'The build-side realization of MIND_BODY' (CONCEPTS.json's own note) means this concept is the ENGINEERING face of a typed Markov blanket, per the corpus's M12 (WORLD ⊥ BODY ⊥ MIND, two typed blankets). No register's word marks that it is the build-side of a pair, and a reader who does not know the pair exists will read it as a standalone metaphor for 'being in touch with reality'.
• 'An engine's internal states are EXPOSED to a world' -- the verb is exposure, and it is asymmetric and slightly adversarial in tone, as befits a security discipline. Renderings built on 'union', 'harmony', or 'connection' (saṃyoga's 'union/absorption'; Spanish 'unión') lose the asymmetry and, with it, the reason a whitelist is needed at all.

Forbidden mistranslations

register	form	why it is wrong
		MW [1112] makes it a technical Nyāya guṇa -- 'contact (esp. in phil. "direct material contact")' -- and also 'carnal contact, sexual union', 'matrimonial connection', and Name of Śiva. Asserts a substance-ontology of contact where the concept asserts a whitelist. It is used in the coinage only because no alternative was found, and the entry records the protest rather than laundering it. If the suspected Sāṃkhya puruṣa/prakṛti reading holds (PENDING-CITATION), the word additionally asserts that this coupling is bondage to be undone, and must then be struck.
		L&S's own first entry for the string is 'imprisoned' -- the Roman nexum, debt-bondage. 'Bound in servitude to the world' is not a coupling; it is a soteriology, and it is the same inversion suspected for saṃyoga arriving independently. The claimed sense is nexus2 ('a binding together, fastening, joining'), and nothing in the written form distinguishes them.
		Drops the whitelist, which IS the concept, and softens 'exposed' into something benign. The definition is a security discipline: closed unless explicitly opened. A word for union asserts the opposite default and, per CONCEPTS.json's own cross-reference, allowlist-by-omission is a DEFECT in this repository -- a real one, already found in guards.py. This rendering ships the defect as the definition.
		All name the surface and none names the rule ON the surface. A membrane is characterised by what crosses it; this interface is characterised by the requirement that anything crossing be NAMED IN ADVANCE. MEMBRANA additionally re-imports the biological reading and collides with ORGANISM.
		The portable test. If a competent reader of the rendering would accept 'we block the bad things' as an implementation of it, the rendering has lost the concept, because the concept's whole content is the opposite default: what is not named is not reachable.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a software security document'. Ask: (1) 'By default, is everything allowed through except what is forbidden, or is everything blocked except what is named?' (2) 'Is this connection a good thing, a bad thing, or neither?' (3) 'Is anyone or anything imprisoned, married, or united here?'

Z_MODULATOR —

Store: TRUE · Confidence: LOW. The Latin morpheme is attested but is a PERSON in its own dictionary ('a director of music, a musician'); the Sanskrit coinage's head was never looked up and its own root-evidence reveals a presiding-deity noun one suffix away; Hindi and Spanish are unsourced and inherit the agent-noun hazard. No register renders 'global vector' or the function clause that defines the object. Held with confidence and evidenced: bhāva is decisively wrong, on MW's own 'the seat of the feelings or affections, heart, soul, mind', which asserts the feeler that the corpus's four-word guard exists to forbid. Recorded honestly: this is the entry where the gap between what the corpus means and what any language will say is widest. · Reviewer: NONE - PENDING EXPERT REVIEW

The global vector of energy, arousal, valence, fatigue, pain, threat, safety and inflammation that sets precision, preferences, habits, learning rate and horizon.

register	form	literal gloss	status
sanskrit	Z-नियमन (Z-niyamana)	the Z-regulation / the Z-restraining	COINED
latin	MODULATOR Z	the Z-measurer-by-rule / the Z-director	COINED
english	the Z affect modulator	The global vector of energy, arousal, valence, fatigue, pain, threat, safety and inflammation that sets precision, preferences, habits, learning rate and horizon. The corpus's guard on it is four words: affect MODELED, never FELT. Every component name is a word for an experience and none of them asserts one.	—
hindi	Z-नियमन ()	the Z regulation	PENDING-CITATION
spanish	modulador Z	the Z modulator	PENDING-CITATION

Ambiguity notes

• THE CORPUS'S OWN GUARD ON THIS CONCEPT IS FOUR WORDS AND IT GOVERNS EVERY CELL OF THIS ENTRY: affect MODELED, never FELT. CONCEPTS.json states the problem exactly: 'Every component name (pain, threat, safety) is a word for an experience and none of them asserts one; a register must not render these with terms that presuppose a subject who suffers.' This is the hardest translation problem in the file, because the concept is a LIST OF EXPERIENCE-WORDS THAT ASSERTS NO EXPERIENCE. Language does not naturally do that.
• bhāva IS DISQUALIFIED BY FIVE WORDS OF ITS OWN MW ENTRY: 'the seat of the feelings or affections, heart, soul, mind' [754]. A SEAT OF FEELINGS IS A FEELER. It is the obvious, elegant, attested Sanskrit rendering for an affect vector, and it asserts precisely the thing the four-word guard forbids. The entry also gives '(in rhet.) passion, emotion' with 8-9 primary kinds and 33-34 subordinate emotional states -- which is the rasa-bhāva enumeration of Sanskrit dramaturgy. So bhāva does not merely assert a subject; it hands the corpus an entire aesthetic theory of how emotions are produced and savoured, complete with a taxonomy that a reader may mistake for a rival to the Z vector's eight components. Two enumerations of affect, one asserting a connoisseur.
• THE LATIN IS WORSE OFF THAN THE SANSKRIT AND THE ENTRY SHOULD SAY SO PLAINLY. Sanskrit offered a choice of suffix: -ana forms an action-noun ('a regulating') and -tṛ forms an agent-noun ('a regulator'), and the coinage takes -ana precisely to avoid supplying a subject. Latin's attested word has already made the other choice: modulator is a -tor agent noun, and L&S's gloss is 'a director of music, A MUSICIAN'. It is a PERSON, cited to Horace. So the Latin register cannot use its attested word without importing an agent, and the file records this as a real asymmetry between registers rather than smoothing it over.
• L&S's FIRST CLAUSE IS GENUINELY APT AND SHOULD NOT BE THROWN AWAY: 'one who measures by rule'. The Z vector SETS PRECISION -- it is, functionally, a thing that measures by rule and tunes by measure. If the personhood could be stripped, modulator would be a good word. That is why the coinage retains it under protest rather than abandoning it, and why the back-translation test asks the person-question directly.
• THE SANSKRIT COINAGE'S OWN EVIDENCE CONTAINS ITS OWN WARNING, which is worth stating because it is unusual: the only attestation offered for the ni-√yam family is the word niyantf (niyantṛ) inside MW's indriya entry -- and MW uses it there for the RULER PRESIDING OVER EACH SENSE, i.e. a deity. So the same root that supplies this coinage's action-noun supplies, one suffix away, a presiding god. The suffix -ana is doing load-bearing metaphysical work, and a translator who 'improves' Z-niyamana to Z-niyantṛ would install a ruler over the affect vector.
• 'GLOBAL VECTOR' IS A MATHEMATICAL OBJECT AND NO REGISTER RENDERS IT. The Z modulator is a vector -- eight named components with values -- not a faculty, a mood, or a humour. Every candidate word in every register names a quality or a person, not a tuple. A reader receiving 'the Z regulation' or 'modulator Z' learns nothing about vectorhood, and the corpus's whole discipline here depends on Z being a modelled quantity rather than a state someone is in.
• THE FUNCTION CLAUSE IS THE DEFINITION AND IT IS THE PART EVERY RENDERING DROPS. What makes Z the Z is not that it lists feelings; it is that it SETS precision, preferences, habits, learning rate and horizon. That is a causal/functional specification. Renderings deliver the affect words and lose the function, which inverts the concept's centre of gravity: a reader is left with a list of feelings (asserting a feeler) and no account of what they do (which is the only reason they are in the model).
• A HUMOURAL MISREADING IS LIVE IN BOTH CLASSICAL REGISTERS AND IS WORTH ONE LINE. A Latin- or Sanskrit-literate reader meeting a global bodily vector of energy, arousal, fatigue, pain and inflammation has a ready-made framework waiting: the four humours, or the three doṣas of Āyurveda. Both are theories of temperament grounded in bodily fluids, both are attested, and both would be received as an explanation rather than as a false friend. The Z vector is not a humoural theory and any rendering that lets it be read as one has imported a discredited physiology along with a subject who has a temperament.

Forbidden mistranslations

register	form	why it is wrong
		THE DECISIVE NEGATIVE OF THIS ENTRY, on MW's own text [754]: 'the seat of the feelings or affections, heart, soul, mind'. It asserts a seat of feeling, i.e. a subject who feels, against the corpus's four-word guard: affect MODELED, never FELT. It additionally imports the rasa-bhāva dramaturgy -- MW gives '(in rhet.) passion, emotion' with 8-9 primary and 33-34 subordinate emotional states -- handing the corpus a rival taxonomy of affect that presupposes a connoisseur savouring it. Attested, apt-sounding, and it installs an experiencer in the one concept most carefully built to have none.
		The aesthetic-sentiment term of the same dramaturgical system, and the freight is heavier still: rasa is affect AS TASTED BY AN AUDIENCE. It presupposes not merely a subject but a savouring subject. If bhāva asserts a feeler, rasa asserts a connoisseur. [NOT fetched from MW this session; this rejection rests on the author's reading and is PENDING-CITATION as a lexical claim.]
		The agent-noun of the very root the coinage uses, and it is forbidden for exactly the reason the coinage chose -ana instead. MW's indriya entry [167,2] uses niyantf for the RULER PRESIDING over each sense -- a deity. Z-niyantṛ therefore installs a ruler, and a divine one, over the affect vector. A translator 'tidying' the action-noun into an agent-noun would make this error while believing he was regularising the morphology.
		All name a state OF A SUBJECT. adfectus and animus are the Latin for a disposition of the soul (animus IS the soul/mind), and Spanish 'ánimo' inherits it. The concept names a modelled vector that sets parameters. Rendering it as a state-of-someone breaches the guard in the corpus's most safety-critical direction -- this is the concept where an unnoticed subject would do the most damage, because the component names (pain, threat, suffering) make the subject feel obvious once installed.
		Both are attested, both are theories of temperament grounded in bodily fluids, and both would be received by a literate reader as an EXPLANATION of what Z is rather than as a loose analogy. They assert a physiology the corpus does not hold and a temperament-bearer the corpus does not posit. A false friend that arrives pre-equipped with a rival theory is worse than an obscure coinage.
		The portable test, and the analogue of the burning-fuse test under AGENT. Z sets precision, preferences, learning rate and horizon; it is a control vector. If the rendering cannot be said of a device that regulates by measure without sounding absurd, it has supplied a subject, and the four-word guard is breached.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a technical description of a control system'. Ask: (1) 'Does this word imply something FEELS these things, or that they are merely tracked as numbers?' (2) 'Is this a person, a faculty, or a quantity?' (3) 'Does this belong to a theory of temperament (humours, doṣas)?'

SENSORIUM —

Store: TRUE · Confidence: LOW for the renderings, HIGH for the two negatives, and the entry is the file's best illustration of why those come apart. The Latin lexeme IS attested with a real page reference and a real cited author, and is graded COINED anyway because L&S's sense -- 'the seat or organ of sensation' -- asserts the sensing subject the corpus withholds; finding the headword is not the test. The Sanskrit coinage rests on two unlooked-up members. Hindi and Spanish are unsourced and the Spanish inherits the Latin's defect. Held with confidence and evidenced: indriya is decisively wrong on MW's own text, three times over independently -- a deity in its etymology, a presiding ruler over each sense, and 'semen virile' among its attested senses. · Reviewer: NONE - PENDING EXPERT REVIEW

A body-to-mind categorical observation interface whose modality count is an empirical question, kept at seven because an over-compressed two-modality bottleneck went NEGATIVE on held data.

register	form	literal gloss	status
sanskrit	ग्रहणसन्धि (grahaṇa-sandhi)	the apprehension-junction	COINED
latin	SENSORIUM	the seat or organ of sensation	COINED
english	sensorium	A body-to-mind categorical observation interface whose modality count is an EMPIRICAL QUESTION, kept at seven because an over-compressed two-modality bottleneck went NEGATIVE on held data. A sensorium is NEVER awareness -- the corpus states this flatly, and it is the exact place where substrate work is most often misread as a science gate.	—
hindi	संवेदन-सन्धि ()	sensation-junction	PENDING-CITATION
spanish	sensorio	sensorium	PENDING-CITATION

Ambiguity notes

• THE CORPUS STATES THE GUARD FLATLY AND CONCEPTS.json UNDERLINES WHERE IT MATTERS: 'A sensorium is NEVER awareness -- the corpus states this flatly, and it is the exact place where substrate work is most often misread as a science gate.' This is not a translation nicety. It is the named site of the corpus's most common category error, where an engineering result about a body-to-mind interface gets read as evidence about experience.
• THE LATIN IS ATTESTED AND THAT IS THE PROBLEM, WHICH MAKES THIS THE MOST INSTRUCTIVE ENTRY IN THE FILE. L&S carries sensorium and glosses it 'the seat or organ of SENSATION' (Boëth. Arist. Top. 8, 5). A SEAT OF SENSATION IS A PLACE WHERE SENSING HAPPENS -- which implies sensing, which implies a senser. So the attested authority licenses the WORD while asserting the very thing the concept withholds. This is a subtler failure than an absent lexeme: with hierarchia or organismus, L&S's silence forces an honest coinage. Here L&S SPEAKS, and what it says is wrong for this concept. An entry that marked this CITED because the headword was found would be a pseudopropheta with a real page reference attached, and it is graded COINED for precisely that reason. THE FILE'S RULE HOLDS: the authority must carry the claimed SENSE, not merely the string.
• L&S'S OWN GLOSS USES THE WORD 'ORGAN', WHICH COLLIDES WITH ORGAN IN THIS SAME FILE -- 'the seat or ORGAN of sensation'. So the Latin definition of SENSORIUM contains the English headword of another concept eight rows up, and that concept is itself a sovereign homonym (build-instrument vs anatomical organ). Three senses tangle at one point: sensorium's 'organ of sensation', ORGAN's anatomical rung-9 sense, and ORGAN's UNI build sense. This file's falsifiers name the pair explicitly for this reason.
• indriya IS DISQUALIFIED THREE TIMES OVER BY ITS OWN MW ENTRY, and the three are independent, which is unusual and worth spelling out. (1) A DEITY IS IN THE ETYMOLOGY: 'fit for or belonging to or agreeable to Indra'. The word is Indra's. (2) A RULER PRESIDES OVER EACH ONE: MW's note records that each sense is 'presided over by its own ruler or niyantf'. So an indriya is not a channel but a fief. (3) MW gives 'virile power' and 'semen virile' as live attested senses. Any one of these disqualifies it; together they make indriya a catalogue of everything this concept refuses.
• AND indriya FAILS A FOURTH TIME, STRUCTURALLY, WHICH IS THE INTERESTING ONE: in the philosophical systems an indriya is a faculty OF a puruṣa -- a witness. The word presupposes the very subject whose existence the corpus declines to assert, and it does so at the exact site CONCEPTS.json names as where substrate work is most often misread as a science gate. A Sanskrit rendering via indriya would not just import a metaphysics; it would import it precisely where the corpus is most vulnerable to being misread. [The Sāṃkhya puruṣa-faculty relation is the author's reading; MW's entry as retrieved gives the philosophical classification and the presiding ruler but does not state it in those terms. PENDING-CITATION.]
• 'CATEGORICAL' AND 'BODY-TO-MIND' ARE TECHNICAL AND DIRECTIONAL AND NO REGISTER RENDERS EITHER. Per the corpus's M12 there are two typed Markov blankets (WORLD ⊥ BODY ⊥ MIND) and interoception is hardware signals; the sensorium is the BODY-TO-MIND interface specifically, and its observations are CATEGORICAL (discrete categories, not continuous magnitudes). Every candidate word in every register names sensing in general, direction unspecified and type unspecified. A reader learns that something senses and not what crosses which blanket in which direction.
• THE MODALITY COUNT IS AN EMPIRICAL QUESTION AND THE SEVEN IS A RESULT, NOT A DESIGN. The definition says the count is 'kept at seven because an over-compressed two-modality bottleneck went NEGATIVE on held data' -- i.e. the seven is what survived a failed compression, and the file's own M2 discipline means the verdict is the CI bound, not the point estimate. A rendering that makes seven sound canonical -- and every register has ready-made sevens waiting (the seven senses, seven chakras, the septem) -- converts an empirical residue into a doctrine. THIS IS NOT HYPOTHETICAL IN SANSKRIT: MW s.v. loka [906] attests 'the number seven' as a sense of loka in the word-numeral system, so a Sanskrit sentence about a seven-modality interface has a live numerological reading available that no other register has.
• 'WENT NEGATIVE ON HELD DATA' IS A PUBLISHED NEGATIVE AND IS PART OF THE CONCEPT, NOT A FOOTNOTE TO IT. Per M15 the corpus's 183 published negatives ARE its credibility, and this definition carries one of them inside itself: the two-modality bottleneck FAILED. A rendering that transmits 'an interface with seven senses' has deleted the evidence that produced the seven and has thereby deleted the reason to believe it.

Forbidden mistranslations

register	form	why it is wrong
		THE DECISIVE NEGATIVE OF THIS ENTRY, and it fails on four independent counts. On MW [167,2]: (1) 'fit for or belonging to or agreeable to Indra' -- A DEITY IS IN THE WORD'S ETYMOLOGY; (2) MW's note on the five senses records each as 'presided over by its own ruler or niyantf' -- each sense has a presiding ruler, so the word asserts a governed fief where the concept asserts an interface; (3) 'virile power' and 'semen virile' are live attested senses of the same entry; (4) structurally, an indriya in the philosophical systems is a faculty OF a witness, so the word presupposes the subject the corpus declines to assert -- at exactly the site CONCEPTS.json names as where substrate work is misread as a science gate. It is the obvious rendering, it is what every Sanskritist will propose, and it is a catalogue of everything this concept refuses. (Count (4) is the author's reading and is PENDING-CITATION; counts (1)-(3) are quoted from MW.)
		L&S: 'the seat or organ of sensation' (Boëth. Arist. Top. 8, 5). A SEAT OF SENSATION IMPLIES SENSING, AND SENSING IMPLIES A SENSER. The corpus says flatly that a sensorium is NEVER awareness. The rendering is retained because it is the term of art and the string is right, but the attested sense is the forbidden one, which is why the status is COINED and not CITED despite a real, quotable, page-referenced entry. This is the file's clearest demonstration that finding the headword is not the test -- carrying the sense is.
		Breaches the corpus's flat statement directly, and 'consciousness' is on the honesty rail's banned list in the author's own voice. CONSCIENTIA is worse than it looks: in Latin it is knowledge-shared-with-oneself, i.e. the inner witness, and in Christian Latin it is the conscience -- a moral faculty. The concept is a categorical observation interface between two typed blankets. Nothing in it knows anything or answers to anyone.
		Converts an empirical residue into a doctrine. The count is 'an empirical question', and seven is what survived after a two-modality bottleneck went NEGATIVE on held data. Every register has a ready-made seven waiting to receive it, and Sanskrit has two: the cultural sevens, and the word-numeral system in which MW attests loka itself as 'the number seven' [906]. A canonical seven deletes the negative result that produced it, and per M15 the negatives are the credibility.
		The portable test, and the analogue of the burning-fuse test under AGENT. A sensorium is a body-to-mind categorical observation interface. If the rendering makes it absurd to say that a purely instrumented system has one, the rendering has installed a subject, and the corpus's flat guard -- a sensorium is NEVER awareness -- is breached.

Back-translation test — status: NOT YET RUN -- no translator engaged.

Give an independent translator ONLY the rendering, definition withheld, with the frame 'in a technical description of a machine'. Ask: (1) 'Does this imply the machine is aware, or that it has channels carrying categorised signals?' (2) 'Where is the seat of it -- is there a place where the sensing is experienced?' (3) 'How many channels are there, and is that number fixed by principle or by measurement?'