title: Bühler's Laws of Manu, verse by verse summary: How Georg Bühler's 1886 English was attached to the Manu-smṛti from a proofread chapter and a scanned volume — and why two chapters were refused. date: 2026-09-23
The sources
Georg Bühler's The Laws of Manu (Sacred Books of the East XXV, Oxford 1886) is public domain. English Wikisource holds a proofread transcription of chapter 1 only. For the rest, the Internet Archive's scan text of the University of Toronto copy was used — raw OCR, with all the faults that implies: digits misread (a "1" as "i", a "7" as "j"), running headers mixed into the text, and every page's footnotes numbered from 1 again, in the same style as the verses.
The rule
Bühler numbers verses exactly as the Sanskrit does, so alignment is by number. The question is only whether a numbered paragraph in the scan really is that verse. Three guards decide it:
- Sequence. Within a chapter, a paragraph is accepted only if its number is exactly the next expected one. A misread digit therefore cannot move a verse onto its neighbour: an out-of-sequence number is simply not accepted.
- Page anchors. Every page of the print carries a running header naming the chapter and the first verse on that page. Those headers re-anchor the expected number on every page, so after a run of unreadable numbers the sequence recovers on the next page rather than dying, and footnotes, which are numbered below the page's first verse, are recognised for what they are.
- Note detection. A paragraph that opens with a commentator's abbreviation (Kull., Medh., Gov., Nār., Nand.), a quotation mark or a cross-reference is a note, not a verse, and is dropped along with everything until the next accepted verse. Any accepted paragraph that still mentions a commentator in its first line, is shorter than a verse could be, or longer than a verse could be, is left out.
Finally, the sparse Sanskrit words Bühler does keep (Brāhmaṇa, Soma, Śrāddha, Snātaka…) are compared verse by verse against the Sanskrit; a chapter whose evidence favours a shifted numbering is refused whole.
The result
| Chapter | Attached | Note |
|---|---|---|
| 1 | 118 of 119 | Wikisource, proofread |
| 2 | 175 of 249 | scan text |
| 3 | 238 of 286 | scan text |
| 4 | 174 of 256 | scan text |
| 5 | 127 of 168 | scan text |
| 6 | 60 of 97 | scan text |
| 7 | — | fewer than 60% of verses parsed; refused |
| 8 | 284 of 420 | scan text |
| 9 | 242 of 336 | scan text |
| 10 | 84 of 131 | scan text |
| 11 | — | evidence favoured a shifted numbering; refused |
| 12 | 86 of 126 | scan text |
1,588 verses in all. Everything from the scan is labelled Georg Bühler (scan text, unproofread) wherever it appears, and the Sanskrit beside it is always the Library's own.
What would improve it
A proofread transcription of the volume — anyone can help with that on Wikisource — would let chapters 7 and 11 in and close the gaps in the others with the same confidence as chapter 1.