Skip to main content

Where MW's Supplement Came From

Monier-Williams' 1899 dictionary carries a supplement — the "annexure", entries tagged <info n="sup"/> in the Cologne digitization of mw. Where did those entries come from? The obvious candidate is Böhtlingk's Nachträge und Verbesserungen (the supplement sections of the great Petersburg dictionary, PW 1855–1875, digitized separately as pwkvn and tagged sup_1sup_7 inside pw.txt).

That question is measurable rather than rhetorical, because the sources carry dates and the Cologne digitizations are queryable.

The enrichment metric

Take a headword set S. Compare its share among MW99's annexure entries with its share among MW99's main entries:

enrichment(S) = [ |S ∩ annexure| / |annexure| ] ÷ [ |S ∩ main| / |main| ]

A ratio ≫ 1 means the supplement is disproportionately stocked from S.

source set Sn(S)share in annexureshare in mainenrichment
PW Nachträge, sup_7 (letzte Nachträge)13,09429.6 %3.8 %7.79×
PW Nachträge, all layers sup_1sup_714,40132.2 %4.1 %7.91×
pwkvn standalone Nachträge digitization14,99532.5 %4.1 %7.84×
Edgerton, Buddhist Hybrid Sanskrit Dictionary (1953)17,7776.8 %3.3 %2.09×

Convention: k1-only headwords; MW99 annexure = 6,067 k1, main = 188,016 k1.

Reading. The 1899 annexure is ~7.8× enriched for PW-Nachträge vocabulary and only ~2.1× enriched for Edgerton's BHS vocabulary — i.e. it is about 3.7× more "PW-flavoured" than "BHS-flavoured". Edgerton (1953) post-dates MW99 by 54 years, so its low enrichment is expected; the PW layers pre-date MW99 and their enrichment is high. The asymmetry is the quantitative form of the long-standing suggestion that MW's supplement was built on the Nachträge.

The dates are the argument

sourceyearrelation to MW99 (1899)
Böhtlingk–Roth, Sanskrit-Wörterbuch (PW), vols 1–7 + Nachträge1855–1875pre-dates — available to MW
Böhtlingk, kürzere Fassung1879–1889pre-dates — available to MW
Monier-Williams, MW721872earlier edition, no annexure
Monier-Williams, MW991899the dictionary with the annexure
Edgerton, BHS Dictionary1953post-dates — not available to MW

Because every Nachträge layer post-dates MW72 and pre-dates MW99, a Nachträge word present in MW99 but absent from MW72 could only have been taken from the Nachträge. Of the 13,094 unique sup_7 headwords, 4,112 appear in no MW edition at all, 27 were dropped between the editions, and MW99 took 6,090 (≈60 %) of the pool MW72 lacked — 1,474 of them through the annexure.

A worked example

kāritra (abstract noun "activity", = ceṣṭita) is absent from mw.txt entirely and from MW72 (1872), while it stands in the Nachträge at PW 7-331-d (pw L216013, pwkvn L16013), citing Mahāvyutpatti 245, 844 (ed. 1849) and glossed by Edgerton's BHS as a Buddhist-Hybrid term. kārāpaka, its alphabetical neighbour, did enter the MW99 annexure — the shape of a supplement that was compiled from the Nachträge and skipped some of its entries.

Caveats

  • The metric ranks provenance likelihood, not citation: set overlaps do not prove that a given annexure entry was copied from a given Nachträge page. The reverse share — ~30 % of annexure headwords are verbatim sup_7 entries — is the direct check.
  • The candidate lists are a worklist, not verdicts: every proposed addition needs printed-scan verification before filing.
  • Near-form string matches are review flags, not matches (kāritra fuzzy-hits kārita at 0.92 — a different word).

Reproduce / read more

Full study (methods, tables, DCS-ranked samples, adjudication workbook): MW-vs-Nachträge trial · independent recheck · uptake & provenance details · adjudication workbook · proposal list. Discussion: csl-corrections#119.