The Anukramani names a 'devata' for every hymn, and 57 of the 214 are not gods: 22 human patrons and seers (Vasistha, Visvamitra, Brbu the carpenter), 7 danastuti labels naming the gift rather than a recipient ('praise of the gift of Sudas son of Pijavana'), and 28 abstractions ruled not to name an addressee, each with its reason recorded. This response excludes all 57 and KEEPS the 5 labels ruled UNDECIDED (Atma, Annam, Brahma, Rati, Saci), because an undecided label removed is a decision made by default. Ask with population=all_ascriptions to see the Anukramani's slot as it stands, with each row's ruling stated.
deity_population_contractDeity counts here include DEITY_CERTAIN + DEITY_PROBABLE. The product default is DEITY_CERTAIN + DEITY_PROBABLE, measured at 0.9742 precision against the gold sample versus 0.6142 for what DEITY_AMBIGUOUS holds; all three tiers are reported regardless, because a deity whose name is an ordinary noun is flattered or penalised by that choice.
measuredATTRIBUTION IS NOT MENTION. The `attributed_*` figures come from the two resolved dedication predicates and from no other: HAS_DEVATA, the Anukramani's own hymn-level dedication, 10,558 edges every one Rigvedic and 8,329 of them a sukta's label projected onto each of its mantras rather than a statement about the verse; and HAS_DEVATA_DERIVED, 882 Atharvavedic dedications over 851 passages reaching 35 deities, recovered from the Anukramani's Sanskrit adjective by its own morphology under Panini 4.2.24 sasya devata. The dedication there is the source's and the resolution is this project's. Two figures, deliberately apart: 47 of the Atharvaveda's 324 descriptors RESOLVE to a deity, and the 882 dedications above were derived from 39 of them -- R4 widened the resolver and the derived layer has not been rebuilt from the 8 it gained (R4-RESIDUAL-ATTRIBUTION-002), so a descriptor can be resolved and still contribute no dedication here. The other 277 are refused with a typed reason each, so an Atharvavedic zero here may be a refused descriptor, a resolved-but-unpropagated one, or no dedication at all. A Samavedic or Yajurvedic zero means those corpora carry no Anukramani apparatus of any kind, NOT that the deity is absent from them -- for that read `mentions_by_veda`, which spans all four.
measuredTHREE DEITY-ATTRIBUTION PREDICATES, NONE COMPLETE. HAS_DEVATA resolves a dedication to a :Devata node and is Rigveda-only: 10,558 edges, every one on the RV. The Atharvaveda carries its dedication under HAS_DEVATA_ASCRIPTION, 5,385 edges over 4,665 of its 6,590 PASSAGES -- 4,816 of those edges on 4,160 of its 5,839 mantras, the rest on hymn-level containers -- pointing at a :DevataAscription that holds Whitney's verbatim descriptor rather than a deity name. Those two share no label value, by category rather than by coverage. The bridge between them is now BUILT and is partial: HAS_DEVATA_DERIVED carries 882 Atharvavedic dedications over 851 passages, resolved from the descriptor's own morphology under Panini 4.2.24 sasya devata, derived from 39 of them and reaching 35 deities. 47 of the 324 descriptors now RESOLVE -- R4 widened the resolver and the derived layer has not been rebuilt from the 8 it gained (R4-RESIDUAL-ATTRIBUTION-002) -- so resolution and dedication are two figures here and not one. The other 277 are refused rather than unprocessed, each carrying a typed reason: 202 reach no canonical deity surface, 67 are compounds naming two ascriptions, 4 are hymn subjects rather than deities, 3 name a plurality, and 1 is the Anukramani's own deferral marker lingokta. So a zero here means this predicate does not reach that corpus, NOT that the corpus records no dedication and NOT that the deity is absent from it. The Samaveda and Yajurveda carry no dedication layer of any of the three kinds, which is GAP-ATTRIBUTION-001 and a source block, not an unbuilt projection. MENTIONS_DEVATA answers 'is this deity named here?' across all four (17,165 edges: RV 10,284, AV 3,582, YV 1,964, SV 1,335). Counts include CONTAINER_INHERITED attributions: a sukta's label projected onto each of its mantras. See the *_strict variant for source-stated attribution only.
deity_profileThe interpretive-claim layer holds 6 claims for the whole corpus and one of them concerns a deity, so an empty `interpretive_claims` is an unbuilt layer rather than an undisputed deity.
measured