Repository path: workshop/translations/alf-layla/register.md · rendered 2026-09-09

«ألف ليلة وليلة» — the binding register

R05 §4. Every entry cites the log decision that created it. Revisions require an erratum on R05-v1/translation.md, not an edit here. Opened at span A (D1–D14); extended at span B (D15–D28), which discharged three of the five open questions and added one; extended at span C (D29–D42), which added V10, T14–T15, withdrew a false statement of fact from open question 2, and added one question; extended at span D (D43–D55), which added V12–V14 and T16–T19, discharged open question 2 — the night formulae, and added one question; extended at span E (D56–D69), which added V15–V16, S5 and T20–T23, discharged nothing, and added one question — the copy-text's night boundaries do not all look the same; extended at span F (D70–D84), which added V17 and T24–T28, and discharged open questions 7 and 10; extended at span G (D85–D101), which added V18 and T29–T32, settled the priority between T13 and S1, and reported on open questions 4, 6, 8, 9 and 10 without discharging any of them except by retirement (question 9); extended at span H (D102–D118), which added V19, V20 and T33–T35, carried the first three-level nesting in the work with nothing supplied at any seam, and added one question; extended at span I (D119–D134), which added V21 and T36–T39, gave open question 12 its first instance and it runs for T2, reported on question 10 without reopening it, and added no question; extended at span J (D135–D150), which added V22, V23 and T40–T44, discharged open question 12 in the register's favour — the case the question named arrived and T2 held — made the first refusal of V20 in the arm, and added no question.

V — voice and register

S — saj' policy

T — terms and names

Discharged at span B

1. انشراح — the calque split. ~~Policy or inconsistency?~~ Inconsistency. The rule is grammatical, not positional, and is now V7 (D25).
2. Verse. ~~No policy exists yet.~~ Settled as S3, S4 and V8 (D15–D18). Fourteen rhyme positions in three poems, fourteen carried, three paid for in sense.
3. Whether الملك before a name is translated at all. ~~May become English clutter.~~ Settled as V9: follow the source, which varies.

Discharged at span D

2. The night-closing formula. ~~Not yet met; the decision will be contaminated and must say so.~~ Met and settled as V12, with V13 for its twin, the night-opening formula, and V14 for the general rule the pair implies (D43–D45). The contamination was handled by excluding Burton's rhyme by name and writing down the three candidates refused; the Arabic end-rhyme is recorded lost, with a compensating polyptoton at the same seam.

Discharged at span F

7. Whether the taṣrīʿ of a poem's first bayt is worth an English rhyme. ~~A later span with a longer poem should test whether the cost scales.~~ Settled: the cost does not scale, and it was a property of one poem rather than of the device. Span F has three poems with taṣrīʿ, including the span's longest at four abyāt; all three carried and none paid for in sense. Over spans B and F the device stands at five instances, five carried, one paid for — span B's poem 1 (D18). Attempt it.

8. Does the copy-text's night-boundary irregularity recur? ~~One irregular pair is an edition's slip and two would be a property of the recension.~~ Settled: it is a property of what is being printed. Four boundaries have now been rendered and three are marked three different ways (D70's table): formula + full frame + heading; heading displaced a page ahead of the formula; and a bare formula with no frame passage at all, resumed with بلغني alone and no Dunyazad exchange. The English carries three different seams because the copy-text has three, and D56 / V15 are confirmed rather than merely carried. And witness 2 settles it independently: it thins the night 3 → 4 seam exactly where the copy-text does — bare formula, no frame, bald resumption — and twenty lines later, at night 4 → 5, prints the frame in full (../collation.md §11.4). Two unrelated witnesses vary their boundaries and vary them at the same place, so the thinning is in the tradition and not in an editor's typesetting. What a later span should watch instead is whether a boundary is thinned in a fourth way, which would say the marking is not merely various but unsystematic.

Unresolved — questions the next span is required to settle

4. الخرزة (D6). If the bead returns later in the work, the crux resolves; if it does not, the flag stands.

5. قرن in على غفلة قرن هذا العفريت (D24, new at span B). Rendered literally as the horn of this ifrit and not solved. If قرن recurs in this sense the crux resolves; the English cuckold's-horn association may be an import and is flagged as one.

6. Does the copy-text systematically lack epithet chains that other witnesses have (new at span C)? Witness 2 describes Shahrazad with a five-member rhymed chain (ذات حسن وجمال وبهاء وقد واعتدال) that the copy-text has not got at all — the first difference in three spans that runs toward witness 2, and it is saj', the device this arm's study limbs measure (../collation.md §7 row 1). Nothing can be translated where the copy-text is silent, so this is not a translation decision; it is a question about the copy-text that a later span should watch, because two instances of a missing rhymed chain would be a pattern and one is not.

7. Does V12's compensating figure survive repetition (new at span D)? — RETIRED at span G, 2026-08-21, with the reason on the record. The question's own bar, written at span D, is ten or more nights rendered. Four nights are done and this arm renders about one night a visit, so the bar is six further visits away, and the arm's declared scope — the frame tale and the first night — was met three spans ago. The bar is also the wrong instrument. What would settle whether broke upon / broke off has gone stale is a reader's report at the fortieth occurrence, and this project has no human readers and no instrument shown to detect a figure wearing out; ten rendered nights would produce ten printings of one sentence and no evidence about any of them. Retiring the question is not a finding that the figure does not wear. V12 stands exactly as written, R05 §2 keeps its first form visible, and a later hand with readers can reopen it.

8. Does the copy-text differ from the received English tradition at other proper names (new at span G)? D85 found that the sage of the fourth night is رويان in the copy-text and in witness 2, against دوبان — Doobán, Douban — in every English Nights there is. Two unrelated Arabic digitisations against the whole English tradition is a recension fact, and one instance is not a pattern. A later span should watch whether the copy-text's names diverge again; if they do, the divergence is a property of this recension worth stating on the collation page, and if they do not, رويان is a single interesting reading and stays one.

Discharged at span J

12. Is T2's جارية → slave-girl right outside the frame tale's slave vocabulary? ~~If a later span uses جارية of a woman who cannot possibly be a slave, T2 needs an erratum.~~ The case arrived at span J and T2 needs no erratum (D135). The queen of the Black Islands — the king's cousin and wife, who cannot possibly be a slave — says هل عندك ما تأكله جاريتك؟, have you anything for your slave-girl to eat?, to the slave she has walked across a city to lie with. The bar the question set is met, and the answer goes the other way: the use is not descriptive but self-abasing, and slave-girl does in English exactly what جارية does in Arabic. The whole point of the sentence is that a woman of royal blood calls herself a slave to a slave, and young woman or any softening deletes the inversion the scene is built on. T2 stands unchanged and strengthened.

The cost is elsewhere and it is real. At span H (D104) the same term fell on a woman at the head of the road who says in her next breath that she is a king's daughter, and there it decided something the Arabic left open. One frozen term, two later spans, one place it costs and one place it pays. The pair is the material of E-20260826c-register-cost and of the arm's craft report.

Unresolved — questions the next span is required to settle (continued)

(question 12, above, is discharged; the original text of it follows for reference)

12b. (the original text of question 12, kept for reference) Is T2's جارية → slave-girl right outside the frame tale's slave vocabulary (new at span H)? D104 kept the frozen term for the woman at the head of the road who tells the king's son, in her next breath, that she is بنت ملك من ملوك الهند — and who is in fact a ghoulah. In this work the word has so far stood beside عبد and سرية and meant property; in Middle Arabic generally it can mean no more than young woman. The register was allowed to bind because a register that yields at the first hard case is not one, and the cost is that the English decides something the Arabic leaves open. If a later span uses جارية of a woman who cannot possibly be a slave, T2 needs an erratum — and this decision is why it will be needed. If no such use turns up, the term stands.

Span H reports on the standing questions without discharging any. الخرزة (4) and قرن (6) do not occur in span H, so both flags stand unchanged. Question 8 gains no third instance: what witness 2 lacks in this span is not a rhymed epithet chain but two whole poems of the eight, 81 tokens (../collation.md §13.1) — the largest verse loss of the arm, and evidence that what thins this tradition is verse rather than any one figure. Question 11 gains no second instance. Span H prints رويان six more times and السندباد twice, and the whole English tradition still writes Douban / Doobán / Duban for the first and Sindbad for the second; T31's divergence recurs but does not multiply, so رويان remains a single interesting reading. What span H does turn up is a recension difference of another kind: the tale King Yunan tells his vizier is the falcon in the copy-text and both its witnesses, the parrot in Būlāq and in Galland's line, and both in Calcutta II (RS-20260823b-embedding-carriage §5).

Span G reports on the standing questions without discharging any: الخرزة (4) does not occur in span G and قرن (6) does not occur in span G, so both flags stand unchanged. Question 8 gains a counter-instance rather than a second instance: at span C the rhymed epithet chain was in witness 2 and missing from the copy-text; at span G the dawn formula's three-member chain is in the copy-text and missing from witness 2 (D90). Whatever thins rhymed chains in this tradition is not thinning one witness in particular. Question 10, discharged at span F, gains its fourth boundary and it is not a fourth kind of thinning — the night 4 → 5 seam is marked in full, by every means the copy-text has, and witness 2 marks it in full too.

9b. (the original text of question 9, kept for reference) Does V12's compensating figure survive repetition (new at span D)? The night-closing formula is written once and printed about a thousand times. broke upon / broke off is a figure at a seam where the Arabic has one, which is why D43 took it; but a figure read once is a felicity and the same figure read for the fortieth time may be a tic, and the Arabic's own device — an end-rhyme — is one that repetition is known to survive, which is why formulae rhyme. A later span that has rendered ten or more nights should ask whether the English formula has worn, and R05 §2 makes the answer an erratum rather than an edit, so the first form stays visible.

Span J reports on the standing questions and discharges one. Question 12 is discharged — see above. الخرزة (4) and قرن (6) do not occur in span J, so both flags stand unchanged; الخرزة has now been absent for eight consecutive spans and a later hand may reasonably retire it. Question 8 gains a third instance and it runs for the copy-text: what witness 2 lacks here is not a rhymed chain but fourteen tokens of narrative, the prince's descent from the dome and the taking of the sword, lost to a haplography between two بنت عمي (../collation.md §15.2). Over spans C, G, H and J the thinning has run toward witness 2 once and away from it three times, so question 8's premise — that the copy-text systematically lacks what other witnesses have — is not holding, and a later span should consider retiring it rather than watching it. Question 11 gains no second instance: span J prints محمود and مسعودة, and the whole English tradition writes Mahmud and Mas'udah too, so T31's divergence at رويان remains a single interesting reading. Question 10, discharged at span F, gains its fifth and sixth boundaries and neither is thinned: the seams into and out of the seventh night are both marked by every means the copy-text has, and witness 2 marks both too. What span J turns up instead is a fourth kind of irregularity that is not a thinning at all — a running head on p. 49 that names the wrong night — and that is a fact about this edition's apparatus, not about the tradition, so it goes to the collation page and not here (D136).