colregs 0.1.1 → 0.2.2

package/docs/verification/2026-08-30-q6-q8.md

@@ -0,0 +1,278 @@
+# Verification: Q-6 and Q-8 primary-source claims
+
+Date: 2026-08-30
+
+Scope: convert four recalled claims underlying Q-6, Q-8, and ADR 0004 into
+cited primary-source evidence or refutations, for PR #4. Verdict
+interpretation (what this means for GATE-1, GATE-6, or ADR 0004) is left to
+the review session — this file records evidence only.
+
+---
+
+## CLAIM 1 (Q-8, load-bearing) — REFUTED
+
+**Claim**: 33 CFR 83 (US Inland Rules) preserves COLREGS Part C paragraph
+structure at the paragraph-path level — i.e. every path in `data/rules.json`
+for Rules 20–31 is either citation-compatible (same path, same text) or
+cleanly absent, never same-spelling-different-text.
+
+**Method**: fetched 33 CFR 83 Subpart C (§§ 83.20–83.31) from eCFR
+(`https://www.ecfr.gov/api/renderer/v1/content/enhanced/current/title-33?part=83`,
+retrieved 2026-08-30, cross-spot-checked against Cornell LII's mirror with no
+contradictions) and diffed it paragraph-by-paragraph against all 90 paths
+this repo's `data/rules.json` carries for Rules 20–31 (source: USCG
+amalgamated international text, retrieved 2026-08-29).
+
+**Verdict: REFUTED.** Of 90 paths, **18 are the dangerous case** (the
+first pass stated 15 but its own table had 14; see the 2026-09-05
+correction) — identical path spelling, materially different denoted text —
+plus one structural path-mismatch, one whole-rule clean absence, and two
+clean intl-only absences. 68 of 90 are citation-compatible
+(90 − 18 − 1 − 1 − 2).
+
+### Diff table
+
+| Path | Int'l | 33 CFR 83 | Match | Note |
+|---|---|---|---|---|
+| 20(a)–20(e) | Y | Y | SAME | Trivial wording only (CFR adds an Annex I cross-ref at 20(e)). |
+| **21(a)** | Y | Y | **DIFFERENT** | CFR appends "except that on a vessel of less than 12 meters in length the masthead light shall be placed as nearly as practicable to the fore and aft centerline of the vessel." Inland-only: the USCG amalgamated page wraps it in `‹‹ ››` and the 1972 treaty text lacks it (see the 2026-09-05 correction below). Repo text is correct. |
+| **21(b)** | Y | Y | **DIFFERENT** | Same pattern: CFR appends the "<12 meters … as nearly as practicable to the fore and aft centerline" clause to the combined-lantern sentence. Inland-only; repo text is correct. |
+| 21(c)–21(f) | Y | Y | SAME | Verbatim. |
+| — (no `21(g)` in intl) | N | Y | US-only insertion | `83.21(g)` defines "Special flashing light" (yellow, 50–70 fpm, forward, 180–225° arc). Confirmed to genuinely live at Rule 21(g) — not misfiled elsewhere. Feeds the `22(a)/(b)/(c)` and `24(f)` mutations below. |
+| 22 | Y | Y | SAME | Verbatim intro. |
+| **22(a)** | Y | Y | **DIFFERENT** | CFR adds "(vi) A special flashing light, 2 miles" — tied to 21(g). |
+| **22(b)** | Y | Y | **DIFFERENT** | Same addition. |
+| **22(c)** | Y | Y | **DIFFERENT** | Same addition. |
+| 22(d) | Y | Y | SAME | CFR (d)(ii) is "[Reserved]" — no material change to (i). |
+| 23(a), 23(a)(i–iv) | Y | Y | SAME | Verbatim. |
+| **23(b)** | Y | Y | **DIFFERENT** | CFR appends "where it can best be seen." Inland-only: `‹‹ ››`-marked on the USCG page, absent from the 1972 treaty text. Repo text is correct. |
+| 23(c) | Y | Y | SAME | Verbatim (WIG craft). |
+| **23(d)(i)** | Y | **N** | **path mismatch** | CFR 83.23 has no lettered sub-paragraphs under (d) — the equivalent content sits at bare `83.23(d)`, not `23(d)(i)`. Content matches; the path spelling doesn't exist in CFR. A structural landmine for a future `us/inland` fixture. |
+| 23(d)(ii) | Y | N | intl-only (clean) | The <7m/≤7kt exception has no CFR 83.23 counterpart. |
+| 23(d)(iii) | Y | N | intl-only (clean) | The off-centerline masthead-light provision has no CFR 83.23 counterpart. |
+| — | — | Y | US-only insertion | `83.23(e)`: Great Lakes vessels may substitute lighting. No intl counterpart. |
+| 24(a), 24(a)(i–v), 24(b) | Y | Y | SAME | Verbatim. |
+| **24(c)** | Y | Y | **DIFFERENT** | Repo/intl (c)(iii) = "a sternlight." CFR 83.24(c)(iii) = "Two towing lights in a vertical line." CFR also references a Western Rivers exception repo/intl lacks. |
+| **24(d)** | Y | Y | **DIFFERENT** | Repo/intl: "...comply with Rule 23(a)(ii)." CFR: "...comply with Rule 23(a)(i) and 23(a)(ii)" — genuinely broader reference. |
+| 24(e) | Y | Y | SAME | Trivial wording. |
+| **24(f)** | Y | Y | **DIFFERENT** | CFR (f)(i)/(f)(ii) each add "and a special flashing light" (tied to 21(g)); CFR adds an entirely new (f)(iii) (vessels towed on both sides) absent from repo/intl. |
+| **24(g)** | Y | Y | **DIFFERENT** | Repo/intl (g)(ii): "two additional all-round white lights." CFR (g)(ii): "four all-round white lights." CFR also adds (g)(v) (searchlight toward the tow), absent from repo/intl. (Repo's own `24(g)(i)` remains a documented gap, orthogonal to this finding.) |
+| 24(h) | Y | Y | SAME | Verbatim. |
+| **24(i)** | Y | Y | **DIFFERENT** | Repo/intl closes "...in particular by illuminating the towline." CFR closes "The searchlight authorized by Rule 36 (§ 83.36) may be used to illuminate the tow," and references a new (j) repo/intl lacks. |
+| — | — | Y | US-only insertion | `83.24(j)(i)/(ii)`: Western Rivers pushing/towing-alongside exception. No intl counterpart. |
+| 25(a)–25(c) | Y | Y | SAME | Verbatim. |
+| **25(d)(i)** | Y | Y | **DIFFERENT** | True IMO text has no "all-round white light" option. CFR inserts "she shall exhibit an all-round white light or" before the existing clause. |
+| **25(d)(ii)** | Y | Y | **DIFFERENT** | Same insertion. |
+| **25(e)** | Y | Y | **DIFFERENT** | True IMO text ends "...conical shape, apex downwards." CFR appends: "A vessel of less than 12 meters in length is not required to exhibit this shape, but may do so." |
+| 26(a)–26(c)(iii) | Y | Y | SAME | Verbatim (trivial wording only). |
+| **26(d)** | Y | Y | **DIFFERENT (structural)** | Repo/intl 26(d) has substantive text (Annex II fishing-proximity signals apply). CFR `83.26(d)` = "[Reserved]" — same path, no text; content relocated to a new Inland-only `83.26(f)`. |
+| 26(e) | Y | Y | SAME | Verbatim. |
+| — | — | Y | US-only insertion | `83.26(f)(i)–(iii)`: full fishing-proximity signal text (trawler shooting/hauling, pair-trawling, purse-seiner), reproducing Annex II content directly inside Rule 26. |
+| 27(a), 27(a)(i–iii), 27(b), 27(b)(i–iv) | Y | Y | SAME | Trivial wording only. |
+| **27(c)** | Y | Y | **DIFFERENT** | Repo/intl: "A **power-driven** vessel...in addition to the lights prescribed in **Rule 24(a)**..." CFR drops "power-driven" (broadens scope) and cites all of Rule 24, not just 24(a). |
+| 27(d), 27(d)(i–iii) | Y | Y | SAME | Verbatim. |
+| — | — | Y | US-only insertion | `83.27(d)(iv)`: new subparagraph on dredge-pipeline lighting. |
+| 27(e), 27(e)(i–ii), 27(f)–27(h) | Y | Y | SAME | Trivial wording only. |
+| 28 | Y | **N ("Reserved")** | whole-rule clean absence | `§ 83.28` is entirely "[Reserved] (Rule 28)" — "constrained by her draft" is not an Inland Rules concept at all. |
+| 29(a), 29(a)(i–iii) | Y | Y | SAME | Trivial wording only. |
+| **29(b)** | Y | Y | **DIFFERENT** | CFR 83.29(b) drops the word "similar" ("prescribed for a vessel of her length"); the IMO text and the USCG international text both carry it, so the repo's "similar" is verbatim, not a quirk. |
+| 30(a)–30(d) | Y | Y | SAME | Verbatim; the 50m/100m thresholds match on both sides. |
+| **30(e)** | Y | Y | **DIFFERENT** | True IMO text: "...not in or near a narrow channel, fairway or where other vessels normally navigate..." CFR inserts "anchorage," into that list — narrows the exemption. |
+| 30(f) | Y | Y | SAME | Verbatim; 12m/aground threshold matches. |
+| — | — | Y | US-only insertion | `83.30(g)–(l)`: special-anchorage exemption, moored-barge lighting, mooring-buoy rules, named river-mile-marker exemptions. Large Inland-only block. |
+| 31 | Y | Y | SAME | Trivial wording only. |
+
+### Correction, 2026-09-05 (issue #6)
+
+The first pass marked 21(a), 21(b) and 23(b) `SAME*` with a note that the
+repo's text had a transcription gap, and called 29(b)'s "similar" a repo
+quirk. Re-checked against primary sources, all four calls were wrong:
+
+- On the USCG amalgamated page, every clause the table said the repo
+  "omits" sits inside `‹‹ ››` — the page's legend defines double angle
+  brackets as Inland Rules text. The repo transcribes the International
+  text only, so their absence is correct, not a gap.
+- The deposited 1972 treaty text (UNTS Vol. 1050, I-15824, Rules
+  21, 23 and 29) has none of the three clauses and does carry "similar" in
+  29(b). No later IMO amendment touched 21(a), 21(b), 23(b) or 29(b):
+  A.464(XII) is the only resolution that edits Part C definitions and its
+  items touch Rules 22, 23(c)/(d), 24, 25(b), 27, 28, 29(a)(iii) and 30,
+  among others — none of them 21(a), 21(b), 23(b) or 29(b). The <12 m
+  centreline-displacement provision it introduced lives at what is now
+  Rule 23(d)(iii), which the repo carries.
+- The rows are now classified as CFR-side divergences (`DIFFERENT`), which
+  raises the dangerous-case count from 14 to 18 and leaves the verdict
+  unchanged. (The first pass stated 15, but its own table carried only 14
+  `DIFFERENT` rows; the count is now taken by tallying the table's Match
+  column — 17 `DIFFERENT` plus 26(d)'s `DIFFERENT (structural)` — and every
+  such row is a single path in `data/rules.json`.)
+
+Sources for the correction, all retrieved 2026-09-05: eCFR
+`https://www.ecfr.gov/api/renderer/v1/content/enhanced/current/title-33?part=83`
+(§§ 83.21(a), 83.21(b), 83.23(b), 83.29(b)); USCG
+`https://www.navcen.uscg.gov/navigation-rules-amalgamated` (raw HTML, the
+`‹‹ ››` markers and the legend); UNTS
+`https://treaties.un.org/doc/Publication/UNTS/Volume%201050/volume-1050-I-15824-English.pdf`;
+`https://navcen.uscg.gov/sites/default/files/pdf/navRules/IMO_A464_12_COLREGS_Amd_19NOV81.pdf`;
+`https://wwwcdn.imo.org/localresources/en/KnowledgeCentre/IndexofIMOResolutions/AssemblyDocuments/A.910(22).pdf`.
+
+**Sources**: eCFR Title 33 Part 83 Subpart C, retrieved 2026-08-30 via
+`https://www.ecfr.gov/api/renderer/v1/content/enhanced/current/title-33?part=83`;
+cross-checked against `law.cornell.edu/cfr/text/33/83.24`, `83.27`, `83.30`.
+`data/rules.json` (this repo, USCG amalgamated international text, retrieved
+2026-08-29) as the international side; mutation candidates additionally
+cross-checked against independent COLREGS transcriptions (cultofsea.com, the
+IALA COLREG reference site) to rule out repo transcription artifacts being
+mistaken for CFR-side divergence.
+
+---
+
+## CLAIM 2 (Q-8, secondary) — REFUTED
+
+**Claim**: no IMO amendment since 1972 has renumbered a Part C (Rules 20–31)
+paragraph — insertions only (e.g. WIG craft at 23(c) in 2001).
+
+**Method**: fetched the IMO resolution PDFs for all seven COLREG amendment
+resolutions directly from `wwwcdn.imo.org`, and classified each amendment's
+effect on Rules 20–31 as no change / insertion / renumbering.
+
+**Verdict: REFUTED.** The seven-resolution list itself is correct with no
+corrections needed. But two amendments demonstrably renumbered Part C
+paragraphs — including the WIG amendment the claim cites as the model
+insertion case.
+
+| Amendment | In force | Effect on Part C | Detail |
+|---|---|---|---|
+| A.464(XII), 1981 | 1 Jun 1983 | **RENUMBERING** | Item 16: "Reletter existing paragraph (g), which becomes paragraph (h)" — pre-1983 Rule 24(g) becomes 24(h); new content is inserted as the new 24(g). Item 22: Rule 27(d)(iv) is deleted and (iv) renumbered to (iii). |
+| A.626(15), 1987 | 19 Nov 1989 | no change | Touches Rules 1(e), 3(h), 8(f), 10(a)/(c), Annex I/IV only. |
+| A.678(16), 1989 | 19 Apr 1991 | no change | Sole amendment is Rule 10(d) (TSS). |
+| A.736(18), 1993 | 4 Nov 1995 | no change (lettering) | Rule 26(b)(i)/(c)(i)/(d) text amended in place; no reletter within Rules 20–31. |
+| **A.910(22), 2001** | 29 Nov 2003 | **RENUMBERING** | Item 4, verbatim: "Add a new paragraph (c) as follows **and renumber accordingly**." Pre-2001 Rule 23(c) (small power-driven vessel <12m alternative lights) becomes 23(d); the new WIG-craft paragraph takes over the path `23(c)`. |
+| A.1004(25), 2007 | 1 Dec 2009 | no change | Sole amendment is Annex IV (distress signals). |
+| A.1085(28), 2013 | 1 Jan 2016 | no change | Adds Part F (Rules 39–41); doesn't touch Rules 20–31. |
+
+The WIG amendment specifically is the counterexample: it is not a clean
+append. The resolution's own operative text orders a renumber, and the
+pre-existing small-vessel provision at `23(c)` was displaced to `23(d)` when
+WIG lighting took the `23(c)` path.
+
+**Sources**: IMO resolution PDFs, `wwwcdn.imo.org/localresources/en/KnowledgeCentre/IndexofIMOResolutions/AssemblyDocuments/{A.910(22),A.626(15),A.678(16),A.736(18),A.1004(25),A.1085(28)}.pdf`;
+A.464(XII) via `navcen.uscg.gov/sites/default/files/pdf/navRules/IMO_A464_12_COLREGS_Amd_19NOV81.pdf` (full text matches IMO's own).
+
+---
+
+## CLAIM 3 (Q-6) — VERIFIED (Arabic component partially)
+
+**Claim**: COLREG 72 was done in English and French, both equally authentic;
+official Spanish and Russian translations deposited with the original;
+Arabic and Chinese texts exist via IMO's official languages.
+
+**Method**: fetched the actual UNTS deposit — Vol. 1050, registration
+I-15824 (registered by IMCO/IMO 8 August 1977) — and quoted Article IX of
+the Convention proper directly, rather than a secondary summary.
+
+**Verdict: VERIFIED** for the English/French and Spanish/Russian
+components, directly against the deposited treaty text. **VERIFIED for
+Chinese, PARTIALLY VERIFIED for Arabic** on the third component (mechanism
+confirmed, Arabic edition indicated but not independently re-fetched).
+
+Exact quote, Article IX ("LANGUAGES"), immediately before the signature
+clause:
+
+> "The present Convention is established, together with the Regulations, in
+> a single copy in the English and French languages, both texts being
+> equally authentic. Official translations in the Russian and Spanish
+> languages shall be prepared and deposited with the signed original."
+>
+> "DONE at London this twentieth day of October one thousand nine hundred
+> and seventy-two."
+
+The UNTS registration header separately notes the bundled Final Act (a
+distinct instrument, not the Convention/Regulations) is authentic in
+English, French, Russian **and** Spanish — that is a different document and
+doesn't change the Convention's own English/French-only authenticity.
+
+Arabic/Chinese: IMO's own multilingualism page confirms Arabic, Chinese,
+English, French, Russian and Spanish as IMO's six official languages
+(`imo.org/en/about/pages/imo-multilingualism.aspx`). A Chinese COLREG
+edition is directly catalogued on IMO's e-publications platform (ISBN
+9789280160512, "避碰规则"), as is a Russian edition (ISBN 9789280141078,
+"МППСС"). An Arabic edition is repeatedly indicated as available on the
+same platform, but direct fetches to its product page returned HTTP 403
+(bot-blocked); this rests on search-result paraphrase, not a page the agent
+loaded and quoted itself — flagged as the one sub-claim not confirmed to
+the same standard as the rest.
+
+**Sources**: UNTS Vol. 1050, I-15824,
+`https://treaties.un.org/doc/Publication/UNTS/Volume%201050/volume-1050-I-15824-English.pdf`
+(retrieved 2026-08-30); `imo.org/en/about/pages/imo-multilingualism.aspx`;
+`imo-epublications.org/content/books/9789280160512` (Chinese),
+`imo-epublications.org/content/books/9789280141078` (Russian).
+
+---
+
+## CLAIM 4 (ADR 0004) — VERIFIED
+
+**Claim**: signalk-server is Apache-2.0, and the prominent community plugins
+are too; the plugin model loads plugins in-process, which is why ADR 0004
+argues "every layer it consumes has to stay permissive."
+
+**Method**: fetched `signalk-server`'s LICENSE file directly from GitHub;
+identified actual top plugins by npm weekly download count via the npm
+downloads API rather than a guessed list; fetched each plugin's LICENSE
+file or npm registry `license` field; fetched SignalK's own plugin
+developer docs for the loading-model claim.
+
+**Verdict: VERIFIED**, after excluding one self-owned data point (below).
+
+1. **signalk-server license — VERIFIED Apache-2.0.** LICENSE file at
+   `raw.githubusercontent.com/SignalK/signalk-server/master/LICENSE` reads
+   "Apache License / Version 2.0."
+
+2. **In-process plugin loading — VERIFIED.** SignalK's plugin developer
+   docs (`github.com/SignalK/signalk-server/blob/master/docs/develop/plugins/README.md`)
+   state plugins install into the server's own `node_modules`, are
+   standard Node.js/TypeScript projects that export `(app) => {...}`
+   receiving the live server `app` object by reference, with `start`/`stop`/
+   `schema` called directly by the server process — standard `require()`/
+   module-graph loading, same process and memory space, no IPC or sandbox
+   boundary for the standard (non-WASM) plugin path.
+
+3. **"Prominent community plugins are permissively licensed" — VERIFIED**,
+   once one data point is excluded on conflict-of-interest grounds, not
+   evidentiary ones. The initial download-based sample turned up one
+   apparent counterexample, `signalk-noaa-space-weather`
+   (`github.com/mark-brannan/signalk-noaa-space-weather/blob/main/LICENSE`,
+   AGPL-3.0-or-later, 4,351 weekly downloads), which sits in the same
+   download band as several permissively-licensed independents also
+   checked (`signalk-to-nmea2000`, ISC; `signalk-n2kais-to-nmea0183`, ISC;
+   `@signalk/udp-nmea-plugin`, Apache-2.0). That plugin is maintained by
+   this repo's own author (Mark Brannan), was only recently relicensed to
+   AGPL, and carries 1 GitHub star — it is not third-party community
+   evidence either for or against the ADR's claim, so it is excluded here
+   rather than counted as a refutation. With it excluded, every
+   independently-installed plugin checked in this pass is permissively
+   licensed (Apache-2.0 or ISC); one further plugin, `sk-ais-status-plugin`,
+   claims `"Apache-20"` in package.json (not a real SPDX identifier) but
+   carries no LICENSE file in its repo — unverifiable, not counted either
+   way. This is also consistent with the independent, larger stars-based
+   sample cited in ADR 0004 itself (19 of the top 20 SignalK-ecosystem
+   projects by GitHub stars are Apache-2.0, one MIT — GitHub comment
+   `mark-brannan/colregs#4` at `issuecomment-5467185331`), which used a
+   different metric and a different population and did not include
+   `signalk-noaa-space-weather` (1 star, well outside a top-20-by-stars
+   cutoff).
+
+**Bearing on ADR 0004**: all three premises the "every layer it consumes
+has to stay permissive" argument needs — signalk-server's own license,
+in-process loading, and ecosystem permissiveness among independent,
+third-party plugins — check out. No open tension with ADR 0004's evidence
+comment remains once the self-owned outlier is set aside for the reason
+given above, not because its license doesn't count.
+
+**Sources**: `raw.githubusercontent.com/SignalK/signalk-server/master/LICENSE`;
+`raw.githubusercontent.com/SignalK/signalk-server/master/docs/develop/plugins/README.md`;
+npm downloads API (`api.npmjs.org/downloads/point/last-week/<pkg>`, window
+2026-08-23–2026-08-29); per-package LICENSE files and npm registry `license`
+fields as cited inline above; `github.com/mark-brannan/signalk-noaa-space-weather`
+repo metadata (1 star, AGPL-3.0 license) via `gh api`.

package/docs/verification/2026-09-05-rule27f-rule28-anchor-scenarios.md

@@ -0,0 +1,330 @@
+# Verification: anchored-vessel scenarios under Rule 27(f) and Rule 28
+
+Date: 2026-09-05
+
+Scope: corroborating real-world evidence, for issue #14, on two questions the
+rule text alone cannot settle — (Q1) whether mine countermeasures vessels
+genuinely operate and signal under Rule 27(f) while **at anchor**, and (Q2)
+whether "constrained by her draught" is understood to apply, lapse, or persist
+for a vessel **at anchor**. Rule text is treated as already established and is
+quoted only where the quotation is itself the evidence. Verdict interpretation
+for `data/applicability.json` is left to the implementing session; this file
+records evidence.
+
+---
+
+## QUESTION 1 (Rule 27(f) at anchor) — CORROBORATED
+
+**Question**: is the "at anchor" branch of Rule 27(f) genuine operational
+practice, or a legally-provided corner case never seen in the water?
+
+**Verdict: CORROBORATED as a deliberate, purposive provision, and partially
+corroborated as observed practice.** The strongest evidence is legislative
+rather than operational: the anchored branch does not date from 1972 — it was
+*added* in 1981, in the same amendment that widened the rule from
+"minesweeping" to "mine clearance". Direct documentary evidence that an
+anchored MCM vessel actually hoisted the three balls was not found; evidence
+that MCM vessels routinely anchor inside their operating area was.
+
+### 1a. The anchored branch is a 1981 addition, not original drafting
+
+Original 1972 text, Rule 27(f), verbatim:
+
+> "(f) A vessel engaged in minesweeping operations shall, in addition to the
+> lights prescribed for a power-driven vessel in Rule 23, exhibit three
+> all-round green lights or three balls. One of these lights or shapes shall be
+> exhibited at or near the foremast head and one at each end of the fore yard.
+> These lights or shapes indicate that it is dangerous for another vessel to
+> approach closer than 1,000 metres astern or 500 metres on either side of the
+> minesweeper."
+
+No Rule 30 branch. No "as appropriate". The anchored alternative arrives with
+IMO (IMCO) resolution **A.464(XII), adopted 19 November 1981** (in force
+1 June 1983), item 24, which replaces the whole paragraph:
+
+> "A vessel engaged in mineclearance operations shall in addition to the lights
+> prescribed for a power-driven vessel in Rule 23 **or to the lights or shape
+> prescribed for a vessel at anchor in Rule 30 as appropriate**, exhibit three
+> all-round green lights or three balls..."
+
+The same resolution (items 19 and 23) replaces "minesweeping" with "mine
+clearance" throughout, which Cockcroft & Lameijer explain was done "to include
+other operations such as minehunting" — i.e. exactly the class of operation
+(sonar hunting, ROV/diver disposal) in which a vessel holds a fixed position
+rather than towing a sweep. A drafting body that inserts an anchored branch at
+the same moment it extends the rule to station-keeping operations is legislating
+for a case it expected to occur, not tidying prose.
+
+Cockcroft's commentary on Rule 27 confirms the wider design intent behind
+"special operations at anchor" signals — the 1960 Regulations' Rule 11(d)
+already required certain hampered vessels to show hampered-vessel signals *in
+addition to* anchor signals, "to warn other vessels to keep well clear", and
+Rule 27(b)(iv) of the 1972 Regulations carries that forward.
+
+**Confidence: high** for the legislative-history claim. Both texts were read
+directly (1972 original and the 1981 resolution's own operative text), not via
+a secondary summary.
+
+### 1b. MCM vessels do anchor in the operating area
+
+USS *Guardian* (MCM-5) Command Operations Report for calendar 1993 (US Navy
+Naval History and Heritage Command archives), chronology entries, verbatim:
+
+> "Underway Corpus Christi OPAREA for mine counter measures (MCM) training.
+> Anchored during evening hours."
+>
+> "Underway Corpus Christi OPAREA for MCM training. Anchored each evening."
+>
+> "Underway Corpus Christi OPAREA for MCM training. Anchored at night."
+> (twice)
+
+Four separate periods in one year in which an MCM ship anchored inside its MCM
+operating area rather than returning to port. This is a US Navy primary source
+and directly supports the "pause in operations while anchored in the operating
+area" scenario. It does **not** state what lights were shown: a ship anchored
+overnight with hunting suspended is arguably no longer "engaged in mine
+clearance operations", in which case Rule 30 alone applies. So this
+corroborates the operational posture, not the signal.
+
+**Confidence: high** that MCM vessels anchor in their operating areas;
+**low** that this particular case involved the 27(f) display.
+
+### 1c. Practitioner/training material treats anchored MCM as a standard case
+
+eColregs (COLREGs training resource, ecolregs.com) carries four
+dedicated entries specifically for the anchored case — "A vessel engaged in
+mine clearance operations at anchor — lights" and "— shapes", alongside the
+underway pair. The anchored-lights page renders three all-round green lights
+per Rule 27(f) *plus* the two Rule 30(a) anchor lights for a 50–100 m vessel.
+Where its catalogue thinks a status only exists underway, it says so in the
+entry title (see Q2 below) — so the existence of a separately-titled "at
+anchor" entry for mine clearance is a positive signal about how the case is
+understood.
+
+**Confidence: medium.** Authoritative-flavoured training material, but it is
+teaching the rule, not reporting a sighting.
+
+### 1d. What was NOT found
+
+- No NATO / USN / RN mine warfare doctrine publication (unclassified) stating
+  that an anchored MCM vessel displays the 27(f) signal. NWP 3-15 / MCWP
+  3-3.1.2 and JP 3-15 were located but neither addresses COLREG signalling for
+  anchored MCM units in the portions reachable publicly.
+- No IMO NCSR/NAV correspondence specific to Rule 27(f) and anchoring.
+- No casualty or incident report involving an anchored MCM vessel's lights.
+- UK Admiralty Annual Notice to Mariners **No. 10** ("Mine-laying and mine
+  countermeasures exercises — Waters around the British Isles", source: Ministry
+  of Defence (Navy)) is the closest operational instrument found. It directs
+  that ships engaged in operations "will show the lights or shapes prescribed in
+  the International Regulations for Preventing Collisions at Sea 1972, Rule
+  27(f)" and that approaching within 1000 metres is dangerous, and it describes
+  minehunters "controlling a wire-guided submersible or working in conjunction
+  with an inflatable rubber dinghy from which divers may be operating at up to
+  1000 metres from the minehunter" — a station-keeping posture. It never
+  mentions anchoring either way. **Absence of a prohibition, not evidence of the
+  practice.**
+
+**Net**: the anchored branch of 27(f) is a purposive 1981 insertion tied to the
+extension of the rule to minehunting; MCM vessels demonstrably anchor in their
+operating areas; no source found says the anchored display is a dead letter, and
+none was found documenting it in use. Modelling the anchored branch is
+well-founded; claiming it is commonly seen is not.
+
+**Sources**:
+- Original 1972 COLREG text, Centre for International Law, NUS,
+  `https://cil.nus.edu.sg/wp-content/uploads/2019/02/1972-Convention-on-Regulations-for-Preventing-Collisions-at-Sea.pdf`
+  (retrieved 2026-09-05; Rule 27(f) and Rule 28 read verbatim).
+- IMCO resolution A.464(XII), 19 November 1981, via USCG NAVCEN,
+  `https://www.navcen.uscg.gov/sites/default/files/pdf/navRules/IMO_A464_12_COLREGS_Amd_19NOV81.pdf`
+  (items 19, 23, 24; retrieved 2026-09-05).
+- A. N. Cockcroft & J. N. F. Lameijer, *A Guide to the Collision Avoidance
+  Rules*, 6th ed. (Elsevier/Butterworth-Heinemann), commentary on Rules 3(g),
+  3(h), 3(i), 18(d), 27 and 28. Consulted 2026-09-05 from an unlicensed
+  online copy; excerpts below are kept short and should be re-checked against
+  a licensed edition before being quoted downstream.
+- USS *Guardian* (MCM-5) Command Operations Report 1993, Naval History and
+  Heritage Command,
+  `https://www.history.navy.mil/content/dam/nhhc/research/archives/command-operation-reports/ship-command-operation-reports/g/guardian-mcm-5-ii/pdf/1993.pdf`
+  (retrieved 2026-09-05).
+- UKHO Admiralty Annual Notice to Mariners No. 10, NP247(1),
+  `https://msi.admiralty.co.uk/nms/annual/10.pdf` (retrieved 2026-09-05).
+- eColregs, "A vessel engaged in mine clearance operations at anchor — lights"
+  (item 360) and "— shapes" (item 362), `https://www.ecolregs.com/`
+  (retrieved 2026-09-05).
+
+---
+
+## QUESTION 2 (constrained by her draught, at anchor) — REFUTED as a valid state
+
+**Question**: is `activity: constrained-by-draught` + `position: anchored` a
+real operational state anyone models or lights, or is it inapplicable the
+moment the vessel stops being underway?
+
+**Verdict: REFUTED.** Every source located that speaks to the point at all
+treats the Rule 28 signal as an underway-only signal. No source was found —
+commentary, IMO guidance, case law, P&I bulletin, pilotage guidance or training
+material — that contemplates a vessel at anchor showing the Rule 28 signal or
+holding CBD status. The refutation rests on four independent lines, none of
+which is merely a restatement of Rule 3(h).
+
+### 2a. The standard commentary says so directly
+
+Cockcroft & Lameijer, comment on Rule 28, after noting that the signal is
+optional and that without it the Rule 18 privilege does not apply:
+
+> "The three red lights must be shown in conjunction with the lights for a
+> power-driven vessel under way."
+
+That sentence is the direct answer: the Rule 28 signal is defined only as
+an addition to the *underway* power-driven light set. There is no anchored
+composite.
+
+The same commentary, on Rule 3(h), also gives the status an explicit
+termination condition tied to navigation, not to a vessel's dimensions:
+
+> "The signals should only be shown when the ability to alter course is severely
+> restricted. On passing clear of the area of relatively shallow water the
+> signals must no longer be displayed."
+
+**Confidence: high.** This is the most-cited English-language COLREG
+commentary, and the sentence is unambiguous. (Caveat as above on the copy used.)
+
+### 2b. IMO's own approved guidance presupposes a vessel making way
+
+Two "guidance for the uniform application" items approved by the IMO Maritime
+Safety Committee, as reproduced in Cockcroft. The first (on Rule 3(h)) says
+navigable width as well as depth counts, and that a vessel with small underkeel
+clearance but adequate room to take avoiding action is not CBD. The second (on
+Rule 18(d)) says a CBD vessel
+
+> "should, when showing the signals prescribed by Rule 28, have her engines
+> ready for immediate manœuvre and proceed at safe speed as required by Rule 6."
+
+The second is decisive on its face: IMO's condition for *showing* the Rule 28
+signal is that the vessel proceed at a safe speed with engines ready — a
+description of a vessel underway and making way. An anchored vessel cannot
+satisfy it. The first repeatedly frames the test in terms of *navigating* and
+*manœuvrability*.
+
+**Confidence: high on content, medium on citation form.** These items were read
+in Cockcroft's reproduction, not fetched from an IMO document; the IMO document
+number carrying them was not located in this pass. Anyone citing them formally
+should chase the underlying MSC circular / guidance annex first.
+
+### 2c. The 1987 amendment strengthened, not loosened, the underway framing
+
+Assembly resolution A.626(15) (1987, in force 19 November 1989) changed Rule
+3(h) from "available depth of navigable water" to "available depth **and width**
+of navigable water". Cockcroft: "The purpose of this amendment was to confirm
+and strengthen the above clarification." The direction of travel over 15 years
+of amendment is toward *manœuvring room while navigating*, and away from a
+draught-based property a hull carries around with it. A ship at anchor has no
+"course she is following" to deviate from, and the amendment history shows IMO
+tightening rather than relaxing that reading.
+
+**Confidence: high** for the resolution number and in-force date (confirmed in
+this repo's 2026-08-30 verification pass, CLAIM 2); **medium** for the specific
+wording change, which rests on Cockcroft alone; **medium** for the inference
+drawn from it.
+
+### 2d. Structural: Annex I positions the Rule 28 lights relative to masthead lights
+
+Annex I, as amended by A.464(XII) item 38, new §2(f)(ii):
+
+> "When it is impracticable to carry the all-round lights prescribed by Rule
+> 27(b)(i) or Rule 28 below the masthead lights, they may be carried above the
+> after masthead light(s) or vertically in between the forward masthead light(s)
+> and after masthead light(s)..."
+
+and item 44, new §3(c), likewise positions the Rule 28 lights "vertically
+between the forward masthead light(s) and the after masthead light(s)". The
+positioning rules for the Rule 28 lights are written *entirely* in terms of
+masthead lights, which a vessel at anchor does not exhibit. There is no Annex I
+provision positioning the Rule 28 lights relative to Rule 30 anchor lights —
+unlike, e.g., the treatment mine clearance gets in Rule 27(f) itself.
+
+**Confidence: high.** Read directly from the resolution text.
+
+### 2e. Nobody models the combination
+
+eColregs enumerates the light/shape catalogue exhaustively and titles each
+entry by status. It carries "A vessel restricted in her ability to manœuvre when
+at anchor — lights" and "A vessel engaged in mine clearance operations at anchor
+— lights/shapes", i.e. it does model anchored variants where they exist. For
+Rule 28 it carries only:
+
+- "A vessel constrained by her draught under 50 m **underway** — lights"
+- "A vessel constrained by her draught over 50 m **underway** — lights"
+- "A vessel constrained by her draught **underway** — shapes"
+
+No anchored variant exists in the catalogue, and the word "underway" is in the
+title of every CBD entry — a distinction the same catalogue does not make for
+statuses it considers position-neutral.
+
+**Confidence: medium-high.** Absence of an entry is weaker than a positive
+statement, but this catalogue demonstrably *does* create anchored entries where
+it thinks they exist, which makes the absence meaningful.
+
+### 2f. On the "anchored deep-draught tanker awaiting the tide" scenario
+
+Searched for, not found: no P&I club circular, flag-state guidance, pilotage
+instruction, IMO interpretive document or admiralty judgment was located that
+addresses a deep-draught vessel's CBD status while anchored awaiting sufficient
+rise of tide. **No corroborating evidence found** that anyone treats this as a
+named operational state. The nearest adjacent authority found is the boundary
+rule on when a vessel counts as anchored at all — Cockcroft, comment on Rule
+3(i):
+
+> "A vessel is only considered to be at anchor when the anchor is down and is
+> holding. Vessels using an anchor to turn in the river, or riding to their
+> chains with anchors unshackled, or dredging with the tide, or dragging their
+> anchors have been held to be under way."
+
+That is a case-law-backed statement (Cockcroft cites it as held, though the
+cases are not named at that point in the text) and it cuts the other way from
+the modelling question: it narrows "at anchor" rather than widening CBD. A ship
+dredging down a channel on her anchor is *underway* and may therefore be CBD;
+one brought up and holding is not underway and, on 2a–2d, is not CBD.
+
+### 2g. What was NOT found
+
+- No admiralty case in which a vessel's CBD status while anchored was in issue.
+- No P&I club loss-prevention bulletin on Rule 28 and anchoring.
+- No IMO NCSR/NAV correspondence on the point.
+- No practitioner-forum thread (gCaptain, YBW or similar) squarely on "can you
+  be CBD at anchor" — searches surfaced only general Rule 28 and general
+  anchoring threads. Practitioner testimony is therefore **absent**, not
+  supporting or contradicting.
+
+**Net**: the combination `constrained-by-draught` + `anchored` has no support in
+any source located, and four independent lines of evidence — commentary, IMO
+approved guidance, amendment history, and Annex I's light-positioning rules —
+point the other way. It should be treated as an invalid combination, not as an
+under-specified one.
+
+**Sources**:
+- Cockcroft & Lameijer, *A Guide to the Collision Avoidance Rules*, 6th ed.,
+  comments on Rules 3(g)–3(i), 18(d), 27, 28 (caveat as cited in Q1).
+- IMCO resolution A.464(XII), 19 November 1981, items 38 and 44 (Annex I), via
+  USCG NAVCEN, URL as cited in Q1.
+- Original 1972 COLREG text, CIL NUS, URL as cited in Q1 (Rule 3(h) pre-1987
+  wording; Rule 28 unchanged since 1972).
+- IMO Assembly resolution A.626(15) — number and date confirmed in
+  `docs/verification/2026-08-30-q6-q8.md`, CLAIM 2; wording change per Cockcroft.
+- eColregs Rule 28 catalogue entries (items 89, 90, 91) and Rule 27/30 anchored
+  entries (items 294, 360, 362), `https://www.ecolregs.com/`
+  (retrieved 2026-09-05).
+- Negative results: WebSearch across IMO, P&I, admiralty-reporter and
+  practitioner-forum sources, 2026-09-05, returned nothing on point (see 2f,
+  2g).
+
+---
+
+## Method note
+
+All searching was done on 2026-09-05 via WebSearch/WebFetch plus direct
+retrieval of source PDFs. Where a search engine's own prose summary asserted a
+conclusion (several did, e.g. "Rule 28's provisions cease to apply when
+anchored"), that summary was **discarded** and not used as evidence: it is model
+output, not a source. Only text read directly out of a fetched document is
+quoted above.