PWG — Sanskrit-Wörterbuch (Grosses PW, 1855–1875)

Chapter authored per Decision 29 Tier A. Position 2 in the atlas ordering — the multi-volume German philological apparatus that MW collapsed.

Trust Block

• Evidence: CDSL v02 pwg.txt, src/data/cross-dict.json, and linked dictionary-profile/audit notes.
• Limitations: narrative atlas profile; counts summarize committed/generated artifacts and do not replace source edition inspection.
• Validation: checked by npm run build; chapter consistency is tracked in _consistency_audit.
• Owner repo: csl-atlas.
• Next use: use this dictionary profile as context, then open source-linked records or compare the lemma in Reader Lookup.

1. Overview

The seven-volume Sanskrit-Wörterbuch, compiled by Otto von Böhtlingk and Rudolph Roth under the auspices of the Imperial Russian Academy of Sciences in Saint Petersburg (hence "Petersburger Wörterbuch", PW). Published in seven folio volumes 1855–1875, ~9,500 columns total. The reference Sanskrit-German dictionary of the 19th century, exceeding all predecessors in both lemma coverage and citation density. PWG is the proximate source for MW 1899: Monier-Williams worked from PWG (and its shorter seven-part counterpart PWK) as the German philological base, applying his own editorial reduction to produce a single-volume English-language work. Within CDSL, PWG carries the densest citation apparatus of any dictionary (4.6 <ls> per record, ~4× MW's 1.09) and a fully differentiated named-kosha apparatus (821 distinct <ls> sigla including 238,271 distinct values overall) — what MW collapsed into the single <ls>L.</ls> hedge.

2. Profile table (5 primary types × profile)

Per the refactored typology in PAPER.md §5, restricted to the types PWG's data supports (PWG has no <info> tagging, no <bot>/<bio>, no <lang>):

Citation profile spread: 0.4 pts — every typed-bucket is cited at ~98.4 %. PWG can afford to cite every entry uniformly because seven folio volumes give it the print space; MW's 11.3-pt spread is the single-volume editor's selective version of the same impulse (analysis/CROSS_DICT_PROFILES.md).

Block-presence by name (cross-dict-common subset):

3. Citation density and apparatus

4.63 <ls> per record (570,817 citations across 123,366 records) — the densest of any structured bilingual CDSL dict, and ~4× MW. The PWG editors chose to cite every kosha and literary work by name; the 7-volume format made this affordable.

Top 12 <ls> sigla (analysis/LS_HEDGE_CHECK.md):

Striking absence: <ls>L.</ls> = 0. PWG never collapses kosha sources into a generic hedge — it always names which kosha. The top 5 sigla are all named indigenous Sanskrit lexicons (ŚKDR, MED, H. an., RĀJAN, ŚABDAR); together they account for 42,931 citations (7.5 % of PWG's total apparatus). MW's <ls>L.</ls> 40,212-instance hedge (MW chapter §4) is the editorial compression of exactly this kind of named-kosha citation into a binary "named-source vs lexicographer-only" distinction.

4. Hedge analysis — PWG has none, by design

PWG carries 0 generic-hedge tags (of 570,817 <ls> total). The contrast with MW is the central data point of the L.-hedge analysis:

The PWG design treats every indigenous lexicon as a named source on a par with literary works. There is no need for a generic hedge because every kosha citation already carries a specific siglum. The L.-hedge is therefore a single-volume editorial invention — MW could not afford PWG's 821-siglum apparatus in one volume and substituted a binary "named source vs L." compression. The three-stage lineage of the hedge (Benfey 1866 † → MW 1872 preface declaration → Cappeller 1891 * → MW 1899 tagged <ls>L.</ls>) is detailed in analysis/LS_HEDGE_CHECK.md §"MW 1872 preface and body check"; PWG belongs to the named-source tradition that the hedge replaces, not to the hedge lineage itself.

5. Lineage statement

PWG occupies the densest node in the CDSL European-Sanskrit-lexicography lineage. It draws on the indigenous Sanskrit kosha tradition (via direct citation of Amarakośa, Medinīkośa, Hemacandra, Śabdakalpadruma, Rājanighaṇṭu, Halāyudha, etc.) and on the extant European Sanskrit dictionaries of the early 19th century (Wilson 1832, cited as <ls>WILS.</ls> 2,014×). Its 19-year compilation (1855–1875) sets the methodological benchmark that every later European Sanskrit dictionary measures itself against. Its direct successors: PWK (Böhtlingk's own compact seven-part abridgement, 1879–1889, which dropped 90+ % of PWG's kosha apparatus); MW 1899 (the English-language single-volume re-edition with <ls>L.</ls> compression). The four indigenous-kosha repos that resolve PWG's named-kosha citations live at ARMH, ABCH, ACPH, ACSJ.

Full lineage narrative in PWG/DICT_PROFILE.md § Lineage.

6. Cross-references — divergence/convergence with adjacent chapters

7. Decisions log

Per-dict editorial choices for this chapter:

• 5-type profile (not 8+1): PWG's data lacks <info>, <bot>, <bio>, <lang> — so the encyclopedic, IE-cognate, and verbal-lemma primary types from MW's refactored 8+1 typology cannot be detected in PWG. The remaining types (nominal-m/f/n, adjective, other) are reported.
• "Other" 21.7 % is large because PWG verbal lemmas and compounds use grammar in prose (<ab>P.</ab>, <ab>Ā.</ab>) rather than <lex> (same pattern as MW per DOUBTS D18).
• No <ls>L.</ls> reporting because PWG has 0; the contrast itself is the finding.
• Three-stage L.-hedge lineage is named in §4 but PWG sits outside the lineage (it belongs to the named-source tradition the hedge replaces), per D21 resolution.
• Effect-size threshold (D19): all percentages on this page exceed |Δ| ≥ 5 pt vs corpus baseline and would survive FDR at q = 0.05.

8. Data dictionary + reproducibility manifest

• DATA_DICTIONARY.md — full tag inventory for PWG
• Source file: csl-orig/v02/pwg/pwg.txt
• Block-detector script: figures/scripts/export_data.py (functions classify_type, detect_blocks)
• Cross-dict aggregate JSON: src/data/cross-dict.json (includes PWG block matrix data)
• License: CC-BY-SA-4.0

See also (tools)

• Matrix explorer — PWG block × type heatmap (where applicable)
• Lineage Sankey — kosha → PWG → PWK → MW flow
• Cross-dictionary comparison — PWG alongside all 8 other CDSL dicts
• PWG/ROADMAP.md — 53 open issues in quarterly plan

Source: CDSL pwg.txt 2026-05-23 · MWS docs-pass commit reflects audit pipeline as of 2026-05-27 · CC-BY-SA-4.0