Diacritic corrections, link styling, and a build lock

Eight diacritic corrections at seven sites, each verified against the page image at 900 dpi: p0049 Nastaʿlīq -> Nasta`līq the mark is a quote, not a ʿayn (Sammay) p0072 Cálidás -> Cálidās p0256 Rajākṛṣṇa -> Rājākṛṣṇa p0260 Īsvaracandra -> Īśvaracandra (x2) p0430 Iśvaracandra -> Īśvaracandra p0430 Gītā -> Gīta (x2) p0430 Munṣī -> Munśī Found by two nets over the transcription: a rare-character inventory (U+02BF appeared exactly once in 431 pages — the tell) and a check for names that appear with more than one accenting, 84 occurrences of which ~32 are now checked. Neither net catches a name mis-transcribed the SAME way everywhere, so this is a partial sweep, not a clean bill. The French accents are all handwritten additions in the scan — pen-drawn cedillas and acutes — applied inconsistently by the author and reproduced per instance. All of that cluster checks out. Clickable regions now carry a 0.4pt hairline rule in RGB(95,125,165) on the Contents/List-of-Plates page numbers and on resolved cross-references; endnote superscripts and whole-line entries stay unmarked. polish.py emits \xref for the former and \hyperlink for the latter so the two differ. Also: plate 1's List-of-Plates entry uses \plx, which build_maps() did not scan, so plate 1 was the one plate of 178 whose "pl. 1" references never linked. build.sh takes an atomic mkdir lock: two builds sharing work/ corrupt main.aux and produce an error that names the wrong file entirely. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-09-15 00:50:04 +06:00
parent 9776331de0
commit 05c41040f7
12 changed files with 70 additions and 21 deletions
@@ -140,11 +140,18 @@ assigned `\hangafter=15590` and the number was gone. Fixed by terminating the as
 **Any numeric assignment in an `\everypar`, a hook, or a macro that can be followed by body text needs
 `\relax`.**

-### One build at a time in `work/`
-`build.sh` runs every pass in `work/`, so two concurrent `make build`s share `main.aux`. Starting a second
-build (or clearing the aux files by hand) while one is running kills the first at `\end{document}` with
-"I can't find file `main.aux'", and the next pass dies at `\begin{document}` in `\openout\@mainaux` — which
-looks like a permissions or disk fault and is neither. Wait for the running build before starting another.
+### One build at a time in `work/` — now enforced by a lock
+Every pass runs in `work/`, so two concurrent builds share `main.aux`: the second reads the first's
+half-written copy and dies at `\begin{document}`. It surfaces as **"Undefined control sequence"** or
+**"File ended while scanning use of `\enotez@note`"**, and the reported line belongs to `main.aux`, not to
+the file the message appears to name — chasing it through `src/main.tex` or the page files wastes an hour.
+The tell is the log's file stack: the last file opened before the error is `(work/main.aux`.
+
+`build.sh` now takes an atomic `mkdir work/.buildlock` and refuses to start rather than corrupting a running
+build. Diagnosed wrongly twice before that (2026-09-14/15): first as a truncated aux left by an interrupted
+run, then as an orphaned container. Neither. The cause both times was simply starting `make` again while an
+earlier build was still going — a build takes several minutes and prints nothing until it ends, so a running
+one is easy to forget. Check `docker ps` first.

 ### A group that ends before its own `\par` loses its leading
 Twice now: the plate captions, and the last paragraph of `src/colophon.tex`. `\setstretch` (like `\small`)
@@ -13,7 +13,7 @@ Character from 1778 to 1978*, Ph.D., SOAS, 1988. ProQuest scan, 431 PDF pages,
 | Structure | `\chapstart{Title}` = new page + centred bold head + PDF bookmark + notes group. One `.tex` per original page in `src/pages/pNNNN.tex` (NNNN = PDF page); `\chapstart` must come *before* `\origpage` in a file |
 | Plates | Cropped from the 300 dpi scan by `tools/crop_plate.py` (scanner bars, page number, caption stripped; gutter shadow swept; deskewed when the page is out of square — the one step that resamples), embedded lossless at native resolution. `\plateop{N}{no}{img}{caption}` for a plate that begins page N; `\plate{no}{img}{caption}` otherwise. Per-page crop recipe recorded in the manifest `notes` column |
 | Colophon | `src/colophon.tex` = leaf 2: description of the conversion + ProQuest notice text. Blank stamped leaf (PDF p.3) dropped and noted there |
-| Editorial | Original spelling and typography stand. **Handwritten corrections** in the scan are adopted silently and logged in QUESTIONS.md. **The author's own slips** are corrected via `\erratum{corrected}{as printed}` and listed in the Errata section. **Slips inside quoted matter** are never corrected — mark `\qslip{as printed}` and table them in QUESTIONS.md so they can be checked against the source. A grammatical slip stands, with an editorial `[sic]`. All confirmed by Sammay, 2026-09-13. The three typographic intentions the typewriter could not realise — en-dashes for numeric ranges, true quotation marks, small caps for institutional acronyms — are applied at set time by `tools/polish.py`; the transcription itself stays as typed (Sammay, 2026-09-13) |
+| Editorial | Original spelling and typography stand. **Handwritten corrections** in the scan are adopted silently and logged in QUESTIONS.md. **The author's own slips** are corrected via `\erratum{corrected}{as printed}` and listed in the Errata section. **Slips inside quoted matter** are never corrected — mark `\qslip{as printed}` and table them in QUESTIONS.md so they can be checked against the source. A grammatical slip in the author's own prose stands **exactly as printed and unmarked** (Sammay, 2026-09-14, reversing the editorial `[sic]` first added on 2026-09-13): the seven `[sic]`s in this edition are all hers, and an eighth in her own notation would have been indistinguishable from them. The edition now makes no interpolation into the text at all. Other decisions confirmed by Sammay, 2026-09-13. The three typographic intentions the typewriter could not realise — en-dashes for numeric ranges, true quotation marks, small caps for institutional acronyms — are applied at set time by `tools/polish.py`; the transcription itself stays as typed (Sammay, 2026-09-13) |
 | Uncertain | `\unsure{...}` → `unsure.log` → QUESTIONS.md; never silently guessed |
 | Contents | Contents and List-of-Plates entries are whole-line hyperlinks to the original page target (`page.N`) |
 | Versioning | **Git** (from 2026-09-14, Sammay). The built PDF is not tracked and not archived — `make` regenerates it from `src/` and `plates/`. The ProQuest scan (`source/`) is untracked too, being third-party; only `make prep` and `make plate` need it. Superseded design drafts stay in `older-versions/` as text |
@@ -47,12 +47,16 @@ table is the authoritative record.)
 ## Resolved
 - **p.295 *compliments* and p.348's dropped *to* join the Errata** (Sammay, 2026-09-14). This overturns the
  p.84 precedent for these two: a word-choice slip and a dropped word are corrected and listed like a
-  misspelling. p.84's own agreement slip keeps its editorial `[sic]` — it was ruled on separately and the
+  misspelling. p.84's own agreement slip is left bare — it was ruled on separately and the
  reading there is the author's grammar, not a slip of the typewriter.
 - **Errata boundary: slips inside quotations stay as printed** and are listed above for verification against
  the source (Sammay, 2026-09-13). Ross's own slips continue to be corrected and listed in the Errata section.
- **p.84 (PDF 86), note 149** — the agreement slip (“the characters \ldots\ shows greater skill”) stands, with an
-  editorial `[sic]` added (Sammay, 2026-09-13). This is the edition's only editorial insertion into the text.
+- **p.84 (PDF 86), note 149** — the agreement slip (“the characters \ldots\ shows greater skill”) stands
+  **as printed and unmarked** (Sammay, 2026-09-14). An editorial `[sic]` was added on 2026-09-13 and has been
+  removed: the edition sets it in the author's own notation (italic *sic* in upright brackets), so a reader
+  could not have told the one editorial interpolation from her seven. Those seven — orig. pp. 47, 105, 106,
+  107, 169, 203, 307 (PDF 49, 107, 108, 109, 171, 205, 309), all inside quoted matter — were each checked
+  against the page image and are the author's. **The edition now interpolates nothing into the text.**
 - **p.50 (PDF 52), note 67** — the handwritten square brackets round the typed parenthesis are ignored; the note
  is set exactly as typed (Sammay, 2026-09-13).
 - **Bengali: modern type is text, historical specimens are images** (Sammay, 2026-09-13). Bengali that shows the
@@ -4,6 +4,22 @@
 set -e
 cd "$(dirname "$0")"
 OUT=ross-1988-retypeset.pdf
+# Two builds must never share work/: they share main.aux, and the second reads
+# the first's half-written copy and dies at \begin{document} with "Undefined
+# control sequence" or "File ended while scanning use of \enotez@note" — which
+# reads like an unbalanced brace in a page file and is nothing of the kind.
+# mkdir is atomic, so it is the lock. The way a second build actually gets
+# started is mundane: someone runs `make` again while an earlier build is still
+# going. Check `docker ps` before starting one — a build takes several minutes
+# and gives no output until it finishes, so a running one is easy to forget.
+if ! mkdir work/.buildlock 2>/dev/null; then
+  echo "build.sh: another build holds work/.buildlock — refusing to start." >&2
+  echo "  check: docker ps    (if nothing is running: rmdir work/.buildlock)" >&2
+  exit 1
+fi
+trap 'rmdir work/.buildlock 2>/dev/null' EXIT INT TERM
+# Three full passes run below regardless, so a stale aux buys nothing.
+rm -f work/main.aux work/main.toc work/main.out
 python3 tools/polish.py >/dev/null   # typographic pass: src/pages -> work/pages
 { cat src/preamble.tex
  printf '%s\n' '\begin{document}\immediate\openout\unsurefile=unsure.log\immediate\openout\erratafile=errata.log\immediate\openout\qslipfile=qslips.log'
@@ -2,7 +2,7 @@
 % note 53: handwritten insertion `a contemporary of' adopted (logged in QUESTIONS.md)
 \noindent\origpage{47}tradition are all important'.\fn{48}{Ruari McLean, \emph{The Thames and Hudson Manual of Typography} (London, 1980), p. 52 taken from \emph{Dossier A - Z 73} ed. Fernand Baudin, Association Typographique Internationale, 1973, p. 18.} It is still an open question as to how Charles Wilkins obtained this education and sense of tradition in the arts of type design, punchcutting, and type casting.

-With regard to the \emph{Grammar,} it is scarcely credible that Wilkins `with the exception of the paper and Roman type, not only formed every article requisite for the work but literally printed it, with no other assistance or advice than the directions for the several branches in Chambers's Cyclopaedia'.\fn{49}{IOL MS EUR 30, A.J. Keily, \emph{A Memoir of Sir Charles Wilkins}, f. 48.} Not only has Wilkins been credited with the achievement of cutting, casting, and printing with the Bengali types for the \emph{Grammar} within one year single-handed, but he is also reputed to have designed a fount of Nastaʿlīq types in as short a time\fn{50}{IOR: H/207/2 \emph{Home Miscellaneous Series,} Extract of Bengal Revenue Consultations, 13 Nov 1778, p. 467: `everything necessary for printing in the Bengal and English Character have been provided, and a fount of Persian Types is nearly compleated [\emph{sic}]'.} by `the labour of his own hand, from the metal in its crudest state, through all the different stages of engraving and founding'.\fn{51}{Francis Balfour, \emph{The Forms of Herkern} (Calcutta, 1781), p. 7.} A report published by the Royal Society of Arts in 1819 casts doubt on such claims:
+With regard to the \emph{Grammar,} it is scarcely credible that Wilkins `with the exception of the paper and Roman type, not only formed every article requisite for the work but literally printed it, with no other assistance or advice than the directions for the several branches in Chambers's Cyclopaedia'.\fn{49}{IOL MS EUR 30, A.J. Keily, \emph{A Memoir of Sir Charles Wilkins}, f. 48.} Not only has Wilkins been credited with the achievement of cutting, casting, and printing with the Bengali types for the \emph{Grammar} within one year single-handed, but he is also reputed to have designed a fount of Nasta`līq types in as short a time\fn{50}{IOR: H/207/2 \emph{Home Miscellaneous Series,} Extract of Bengal Revenue Consultations, 13 Nov 1778, p. 467: `everything necessary for printing in the Bengal and English Character have been provided, and a fount of Persian Types is nearly compleated [\emph{sic}]'.} by `the labour of his own hand, from the metal in its crudest state, through all the different stages of engraving and founding'.\fn{51}{Francis Balfour, \emph{The Forms of Herkern} (Calcutta, 1781), p. 7.} A report published by the Royal Society of Arts in 1819 casts doubt on such claims:

 \begin{extract}Caslon, the letter-founder, thinks \ldots\ it would be scarcely possible for one person to complete a fount of letters from first to last.\fn{52}{Although he is referring to Latin diamond type, Bengali would have been no simpler, and would have comprised a greater number of sorts.} In the ordinary course of business the mere preparation of the types after the punch cutter has finished his process,\fn{53}{According to a contemporary of the letter-founder Vincent Figgins (1766-1844), the greatest number of punches that an artist can cut in a day is two; see below, chapter 5.} goes through the following eight different hands; 1st, the justifier, who strikes the matrices; 2nd, the mould maker; 3rd, the caster; 4th, the breaker off; 5th, the rubber; 6th, the kerner; 7th, the setter up; 8th, the dresser. Two of these are boys; and although the work might no doubt be executed by fewer than eight different hands \ldots; yet that the whole could be gone through by one person he [Caslon] thinks barely
 % the quotation continues on the next page — closed in p0050.tex
@@ -1,5 +1,5 @@
 % PDF page 72 — printed page 70 — Chapter 1 (cont.)
 % handwritten `from' inserted over a caret after `different' (adopted)
-\origpage{70}The design of Wilkins's third Bengali fount (CW3), which was employed by the Company's Press for printing translations, regulations, forms, and other sundry items including the first work ever printed in Sanskrit,\fn{128}{Kālidāsa [Ṛtusaṃhāra]. \emph{The Seasons: a Descriptive Poem, by Cálidás, in the Original Sanscrit}, ed. Sir William Jones (Calcutta, 1792). A Sanskrit work in Bengali characters. The first regulation to be printed in Bengali characters by the Honourable Company's Press was the \emph{Translation of the Regulations for the Administration of Justice in the Courts of the Dewanny Adawlut} by Jonathan Duncan (Calcutta, 1784).} is manifestly different from that of CW1 (and CW2).\fn{129}{See pls. 24-26.} This typeface, whilst retaining a diagonal stress achieved by such characters as \ig{plates/inline/p0072-u01.png} , \ig{plates/inline/p0072-u02.png} , and \ig{plates/inline/p0072-u03.png} , is characterized by its heavy headline and the strong vertical strokes emphasized by the tight intercharacter spacing and the relatively narrow counters of characters like \ig{plates/inline/p0072-u04.png} and \ig{plates/inline/p0072-u05.png} . The main characters of the syllabary have been reduced in height as well as width, although when set with subscripts and superscripts the size of the typeface has remained much the same. This modification renders it more suitable for mixing with Latin types;\fn{130}{For which purpose it was often employed by the East India Company.} it therefore also begins to approximate the height of the script in manuscript form.
+\origpage{70}The design of Wilkins's third Bengali fount (CW3), which was employed by the Company's Press for printing translations, regulations, forms, and other sundry items including the first work ever printed in Sanskrit,\fn{128}{Kālidāsa [Ṛtusaṃhāra]. \emph{The Seasons: a Descriptive Poem, by Cálidās, in the Original Sanscrit}, ed. Sir William Jones (Calcutta, 1792). A Sanskrit work in Bengali characters. The first regulation to be printed in Bengali characters by the Honourable Company's Press was the \emph{Translation of the Regulations for the Administration of Justice in the Courts of the Dewanny Adawlut} by Jonathan Duncan (Calcutta, 1784).} is manifestly different from that of CW1 (and CW2).\fn{129}{See pls. 24-26.} This typeface, whilst retaining a diagonal stress achieved by such characters as \ig{plates/inline/p0072-u01.png} , \ig{plates/inline/p0072-u02.png} , and \ig{plates/inline/p0072-u03.png} , is characterized by its heavy headline and the strong vertical strokes emphasized by the tight intercharacter spacing and the relatively narrow counters of characters like \ig{plates/inline/p0072-u04.png} and \ig{plates/inline/p0072-u05.png} . The main characters of the syllabary have been reduced in height as well as width, although when set with subscripts and superscripts the size of the typeface has remained much the same. This modification renders it more suitable for mixing with Latin types;\fn{130}{For which purpose it was often employed by the East India Company.} it therefore also begins to approximate the height of the script in manuscript form.

 The most striking change in the design of the types is the alteration in character to character relationships. Proportions have altered considerably. For instance, the character \ig{plates/inline/p0072-u06.png} , which in CW1 was of similar proportions to \ig{plates/inline/p0072-u07.png} , has been so condensed in CW3 as to render it hardly comparable to its precursor; it is very poorly designed. The sort \ig{plates/inline/p0072-u08.png} also suffers from the excessive reduction of its internal counter, and the narrow typeform \ig{plates/inline/p0072-u09.png} , whilst kerning skilfully under the character it modifies, has become stiff and angular, losing the flow of its earlier design.\fn{131}{See Halhed, \emph{Grammar}, p. 23.} Variations in the design of one sort, as in CW1, no longer appear: the medial \ig{plates/inline/p0072-u10.png} has returned to its form \ig{plates/inline/p0072-u11.png} , but discrepancies exist in the stroke weight of certain characters, which are not wholly attributable to unevenness in the printing.
@@ -1,6 +1,6 @@
 % PDF page 86 — printed page 84 — Chapter 1 (cont.); continues across plates 29–32
 \noindent\origpage{84}been completely revised. The anomalous form of \ig{plates/inline/p0086-x01.png} in the previous fount has been expanded, and the stroke weight in this typeface is more consistent, particularly with regard to the flared vertical strokes. The intercharacter spacing has been opened up, as have the counters of \ig{plates/inline/p0086-x02.png} , \ig{plates/inline/p0086-x03.png} , and similar sorts, thereby diminishing the condensed appearance of CW3. The thickness of the headline also appears to have been reduced, but this may be the result of lighter inking of the imprints inspected. Some of the inherent liveliness of Wilkins's first Bengali fount has returned to CW4, which can be described as a legible and elegant typeface.

-Wilkins's ability as an engraver has been demonstrated in Halhed's \emph{Grammar} by the reproduction of a hand-written petition.\fn{148}{See above, pl. 23, see also pls. 33 and 34 of Devanagari characters; yet Wilkins maintained they were `designed by a mere amateur in the art of writing'; Wilkins, \emph{Grammar of the Sanskrita Language,} p. xiv.} Haughton's \emph{Rudiments of Bengali Grammar} also dispels any doubts of Wilkins's putative skills as a letterer. Three of the four plates in the 1821 grammar were designed by Wilkins,\fn{149}{Presumably after Shepherd's death (see pl. 36), although the characters designed by J.H. Patton shows [\emph{sic}] greater skill (see pl. 35). These plates were reproduced by Stephen Austin in Duncan Forbes's \emph{Grammar of the Bengālī Language} (London, 1861).} whose skill and knowledge of Bengali letterforms had undoubtedly increased since his first attempts at Bengali typography. These plates engraved by J. Swaine bear, for the most part, the style of lettering that Wilkins strove to imitate in metal, rather than the cursive lettering penned in documents of the eighteenth and nineteenth centuries.
+Wilkins's ability as an engraver has been demonstrated in Halhed's \emph{Grammar} by the reproduction of a hand-written petition.\fn{148}{See above, pl. 23, see also pls. 33 and 34 of Devanagari characters; yet Wilkins maintained they were `designed by a mere amateur in the art of writing'; Wilkins, \emph{Grammar of the Sanskrita Language,} p. xiv.} Haughton's \emph{Rudiments of Bengali Grammar} also dispels any doubts of Wilkins's putative skills as a letterer. Three of the four plates in the 1821 grammar were designed by Wilkins,\fn{149}{Presumably after Shepherd's death (see pl. 36), although the characters designed by J.H. Patton shows greater skill (see pl. 35). These plates were reproduced by Stephen Austin in Duncan Forbes's \emph{Grammar of the Bengālī Language} (London, 1861).} whose skill and knowledge of Bengali letterforms had undoubtedly increased since his first attempts at Bengali typography. These plates engraved by J. Swaine bear, for the most part, the style of lettering that Wilkins strove to imitate in metal, rather than the cursive lettering penned in documents of the eighteenth and nineteenth centuries.

 As already stated, the exact measure of Wilkins's involvement in the production of any of the Bengali founts ascribed to him is unknown. Despite the numerous laudatory remarks regarding Wilkins's typographic successes, scant information is available on the actual procedures he employed to manufacture the founts. It must be assumed that he was aided in his typographic activities, which by no
@@ -1,5 +1,5 @@
 % PDF page 256 — printed page 254 — Chapter 7 (cont.)
-\origpage{254}sufficient distinction in character formation to ensure readability. These faces share the method of letter structure common to formal Bengali penmanship\fn{31}{See above, chapter 3ii, and Lambert, \emph{Devanagari Script,} pp. 177 and 178.} which is lacking in the designs produced by the European typefoundries and also in GV2. The elegance and vibrancy of these founts employed in editions of \emph{Chandaḥkusuma}\fn{32}{Bhuvanamohana Rāya Caudhuri \emph{Chandaḥkusuma} (Calcutta, 1864); see pl. 111.} and \emph{Naradeha Nirṇaya}\fn{33}{Rajākṛṣṇa Rāya Caudhuri, \emph{Naradeha Nirṇaya} (Human Physiology in Bengali) (Calcutta, 1859); see pl. 112.} are aided by the deliberate effort given to the design of each sort, as instanced by \ig{plates/inline/p0256-m01.png} , whose finial varies in length and curvature when in conjunction with another consonant.\fn{34}{See pl. 102, lines 7, 12 and 14 (GV3); see also pl. 107, \emph{sandhi} section.} Each sort, whether basic character, conjunct, or ligature appears carefully designed in harmony with the remainder of the typeface, rather than fused together from common components. A number of conjuncts, however, are still generated by the \emph{phalā} system created by Wilkins for his first Bengali typeface.
+\origpage{254}sufficient distinction in character formation to ensure readability. These faces share the method of letter structure common to formal Bengali penmanship\fn{31}{See above, chapter 3ii, and Lambert, \emph{Devanagari Script,} pp. 177 and 178.} which is lacking in the designs produced by the European typefoundries and also in GV2. The elegance and vibrancy of these founts employed in editions of \emph{Chandaḥkusuma}\fn{32}{Bhuvanamohana Rāya Caudhuri \emph{Chandaḥkusuma} (Calcutta, 1864); see pl. 111.} and \emph{Naradeha Nirṇaya}\fn{33}{Rājākṛṣṇa Rāya Caudhuri, \emph{Naradeha Nirṇaya} (Human Physiology in Bengali) (Calcutta, 1859); see pl. 112.} are aided by the deliberate effort given to the design of each sort, as instanced by \ig{plates/inline/p0256-m01.png} , whose finial varies in length and curvature when in conjunction with another consonant.\fn{34}{See pl. 102, lines 7, 12 and 14 (GV3); see also pl. 107, \emph{sandhi} section.} Each sort, whether basic character, conjunct, or ligature appears carefully designed in harmony with the remainder of the typeface, rather than fused together from common components. A number of conjuncts, however, are still generated by the \emph{phalā} system created by Wilkins for his first Bengali typeface.

 The later imprints of the Girīśa Vidyāratna Press\fn{35}{See 1899 edition of Nīlamaṇi Vāsaka's \emph{Navanārī} (Calcutta, 1899); see pl. 113.} show that, in time, it was competent to produce very presentable and readable typefaces in the smaller range of type sizes while still avoiding the European style of letter structure. However, none of the founts can be considered outstanding for its period, and the employment of the accepted method of composition did not encourage other native foundries to seek alternative solutions to the problem of the reproduction of the Bengali script by pre-fabricated letterforms.

@@ -3,4 +3,4 @@

 No absolute character set had been determined for Bengali composition;\fn{36}{This remains the case today. See above, chapter 3iii.} even the simplest textual matter required several hundred sorts to achieve a fair representation. The time and cost of engraving so many punches\fn{37}{`About one per day being the average output' [for Latin]; Whetton, \emph{Practical Printing,} p. 8; see also chapters 1 and 5.} prohibited the production of many different weights and type sizes. Thus the emphasis from the outset was placed on reducing the number of characters by various means. Wilkins confined his first Bengali fount to some 170 sorts by employing common components to generate types that were wanting.\fn{38}{Using what later became known (particularly for Devanagari composition) as the `Degree' system; see below, chapter 8.} An increasing number of indigenous designs, however, sought to improve the quality of handsetting by extending the number of sorts to include the most frequent conjuncts, combinations with \emph{raphalā}, and some with subscript vowel signs. The customary number of sorts grew to just over 500.

-Two works by the founders of the Sanskrit Press,\fn{39}{Founded in 1847; there were several Sanskrit Presses at this time.} Īsvaracandra Vidyāsāgara and Madanamohana Tarkālaṅkāra, first published in the mid-nineteenth century assisted in establishing a standard Bengali character set for printed works. \emph{Śiśuśikṣā}\fn{40}{Madanamohana Tarkālaṅkāra, \emph{Śiśuśikṣā} (The Infant Teacher), (Calcutta, 1849).} and \emph{Varṇaparicaya}\fn{41}{Īsvaracandra Vidyāsāgara, \emph{Varṇaparicaya,} Pts I and II (Calcutta, 1855).} enjoyed immense popularity, both running to well over a hundred editions before the turn of the century. In typographic terms Vidyasagara's Bengali primer, \emph{Varṇaparicaya,} is of greater importance.
+Two works by the founders of the Sanskrit Press,\fn{39}{Founded in 1847; there were several Sanskrit Presses at this time.} Īśvaracandra Vidyāsāgara and Madanamohana Tarkālaṅkāra, first published in the mid-nineteenth century assisted in establishing a standard Bengali character set for printed works. \emph{Śiśuśikṣā}\fn{40}{Madanamohana Tarkālaṅkāra, \emph{Śiśuśikṣā} (The Infant Teacher), (Calcutta, 1849).} and \emph{Varṇaparicaya}\fn{41}{Īśvaracandra Vidyāsāgara, \emph{Varṇaparicaya,} Pts I and II (Calcutta, 1855).} enjoyed immense popularity, both running to well over a hundred editions before the turn of the century. In typographic terms Vidyasagara's Bengali primer, \emph{Varṇaparicaya,} is of greater importance.
@@ -14,11 +14,11 @@ Dās, Sajanī Kānta. \emph{Bāṅglā Gadya Sāhityera Itihāsa.} Calcutta, 196

 ------. \emph{Ādibhāga.} [Old Testament in Bengali.] Vol. 1, Baptist Mission Press: Calcutta, 1842.

------. \emph{Dāudera Gītā (The Psalms of David).} Serampore, 1803.
+------. \emph{Dāudera Gīta (The Psalms of David).} Serampore, 1803.

 ------. ------. Baptist Mission Press: Calcutta, 1826; and Baptist Mission Press: Calcutta, 1840.

------. \emph{Gītā Saṃhitā...The Book of Psalms.} Bishop's College Press: Calcutta, 1858.
+------. \emph{Gīta Saṃhitā...The Book of Psalms.} Bishop's College Press: Calcutta, 1858.

 ------. \emph{(Old and New Testaments.)} 4th edn, Baptist Mission Press: Calcutta, 1867; and 11th edn, Baptist Mission Press: Calcutta, 1909.

@@ -44,7 +44,7 @@ Morton, William. \emph{Dvibhāṣārthakābhidhāna or a Dictionary of the Benga

 Mukhopādhyāya, Rājīvalocana. \emph{Mahārāja Kṛṣṇacandra Rāyasya Caritraṃ.} Serampore, 1805.

-Munṣī, Caṇḍīcarana. \emph{Totā Itihāsa} [Translated from Hindustani version of Haidar Bhakṣā]. Serampore, 1805.
+Munśī, Caṇḍīcarana. \emph{Totā Itihāsa} [Translated from Hindustani version of Haidar Bhakṣā]. Serampore, 1805.

 Pearson, John David. \emph{Vākyavalī. Idiomatical Exercises, English and Bengalee...,} Calcuttta, 1825.

@@ -64,7 +64,7 @@ Vasu, Rāma Rāma, \emph{Lipimālā.} Serampore, 1802.

 Vidyālaṅkāra, Mṛtyuñjaya, \emph{Rājāvalī.} Serampore, 1808.

-Vidyāsāgara, Iśvaracandra, \emph{Varṇaparicaya.} Calcutta, 1855.
+Vidyāsāgara, Īśvaracandra, \emph{Varṇaparicaya.} Calcutta, 1855.

 ------. \emph{Vidhavā Vivāha Pracalita Haoyā Ucita Kina.} 2nd edn, Girīśa Vidyāratna Press: Calcutta 1857.

@@ -277,7 +277,21 @@
 \newcommand{\ig}[1]{\raisebox{-0.3ex}{\includegraphics[scale=\igscale]{#1}}}

 % ---- contents / list-of-plates lines --------------------------------------
-\newcommand{\pg}[1]{\hyperlink{page.#1}{#1}}
+% ---- clickable regions ----------------------------------------------------
+% This edition adds links the original never had, so their cue is as quiet as it can
+% be and still be seen: a hairline rule under the text in a blue-grey close to the ink.
+% Marked: the page numbers of the Contents and List of Plates (\pg), and the
+% cross-references polish.py resolves (\xref).
+% Not marked: the endnote superscripts — already the conventional affordance, 986 of
+% them, and a rule beneath a raised numeral reads as a defect; and the whole-line
+% Contents and plate entries, which stay clickable across their full width while the
+% page number at the right carries the visible cue.
+\definecolor{linkink}{RGB}{95,125,165}
+\newsavebox{\lkbox}
+\newcommand{\linkface}[1]{\sbox\lkbox{#1}%
+  \rlap{\raisebox{-1.8pt}{\textcolor{linkink}{\rule{\wd\lkbox}{0.4pt}}}}\usebox\lkbox}
+\newcommand{\xref}[2]{\hyperlink{#1}{\linkface{#2}}}
+\newcommand{\pg}[1]{\hyperlink{page.#1}{\linkface{#1}}}
 % Contents lines: the whole entry is a link to the chapter's opening page,
 % not just the page number.
 \newcommand{\tocl}[4]{\noindent\hspace*{#1}\hyperlink{page.#4}{\makebox[0.55in][l]{#2}%
@@ -31,6 +31,12 @@ def build_maps():
        t = open(f).read()
        for no, pg in re.findall(r"\\pl\{(\d+)\}\{.*?\}\{(\d+)\}", t):
            plate[int(no)] = int(pg)
+        # \plx is the one List-of-Plates entry whose location is not a bare page
+        # number (plate 1 is "page 18", a plate facing the text), so its page
+        # reference is wrapped in \pg. Without this, plate 1 was the single plate
+        # whose in-text "pl. 1" references never linked.
+        for no, pg in re.findall(r"\\plx\{(\d+)\}\{.*?\}\{.*?\\pg\{(\d+)\}", t):
+            plate[int(no)] = int(pg)
        for no, pg in re.findall(r"\\tocl\{[^}]*\}\{(\d+)\}\{.*?\}\{(\d+)\}", t):
            chap[int(no)] = int(pg)
        m = re.search(r"\\subchap\{(\d+)\.([ivx]+)", t)
@@ -43,13 +49,15 @@ def build_maps():
 PLATE, CHAP, SUBSEC = build_maps()

 def link(target, shown):
-    return r"\hyperlink{page.%d}{%s}" % (target, shown)
+    # \xref, not \hyperlink: the preamble gives \xref a visible hairline rule,
+    # while \hyperlink stays unmarked for the whole-line Contents and plate entries.
+    return r"\xref{page.%d}{%s}" % (target, shown)

 PROTECT = [
    re.compile(r"\\(?:erratum|qslip)\{[^}]*\}(?:\{[^}]*\})?"),
    re.compile(r"\\ig\{[^}]*\}"),
    re.compile(r"\\includegraphics(?:\[[^\]]*\])?\{[^}]*\}"),
-    re.compile(r"\\(?:origpage|fn|hyperlink|hypertarget|pg)\{[^}]*\}"),
+    re.compile(r"\\(?:origpage|fn|hyperlink|hypertarget|pg|xref)\{[^}]*\}"),
    re.compile(r"\\(?:plateop|plate)\{[^}]*\}\{[^}]*\}\{[^}]*\}"),
    re.compile(r"\\(?:pl|tocl|toclnp|erratumline)(?:\{[^}]*\})?"),
    # heading macros: their arguments become PDF bookmarks and running-head