1 files changed, 102 insertions, 0 deletions
diff --git a/docs/reference/canals.md b/docs/reference/canals.md
new file mode 100644
index 0000000..738f9ee
--- /dev/null
+++ b/docs/reference/canals.md
@@ -0,0 +1,102 @@
+# Real canals: a calibration reference for great works
+
+Purpose: ground the sim's canal model ([[t-great-works]], [[t-canal-site-scan]]) in real projects. It covers what
+canals were for, what they cost, who could build them, how long they took, which failed and what made them decay.
+Gathered 2026-10-03 at the author's request.
+
+**Figures are approximate.** They come from the standard references (mostly each canal's Wikipedia article and the
+works it cites), recalled rather than re-checked line by line. Lengths are the navigable route; "dug" is the
+artificial part where a canal also uses lakes or rivers. Use them for orders of magnitude, not as data.
+
+## 1. Pre-industrial (closest to High Fantasy technology)
+
+| Canal | Built | Length | Summit / locks | Joins | Purpose | Builder, labour |
+|---|---|---|---|---|---|---|
+| Canal of the Pharaohs (Nile–Red Sea) | begun c. 600 BC (Necho II), finished c. 500 BC (Darius I) | ≈ 150–200 km | near sea level | Nile delta ↔ Red Sea | trade, prestige | Egyptian, then Persian state; Herodotus claims 120,000 dead under Necho |
+| Xerxes Canal (Athos) | 483–480 BC | ≈ 2 km, ≈ 30 m wide | sea level | across the Athos peninsula | military: the fleet avoids the cape where a Persian fleet was wrecked in 492 BC | Persian army, ≈ 3 years |
+| Diolkos (Corinth) — *portage, not a canal* | c. 600 BC, used to the 1st c. AD | ≈ 6–8.5 km paved haul road | over the isthmus | Gulf of Corinth ↔ Saronic Gulf | ships or cargo dragged across instead of sailing round the Peloponnese | city of Corinth; a cheap alternative used for centuries |
+| Lingqu (Qin) | 214 BC | ≈ 36 km | contour canal over a watershed; flash locks later | Xiang (Yangtze basin) ↔ Li (Pearl basin) | military supply for the conquest of the south | Qin state |
+| Grand Canal (China) | sections from 486 BC; Sui trunk 605–610; Yuan route to Beijing 1280s–1290s | ≈ 1,800 km | summit ≈ 40 m (Shandong) | Yellow River ↔ Yangtze ↔ Beijing | grain tribute to the capital, army supply, unity | Sui state; millions of conscripts (figures vary) |
+| Fossa Carolina | 793 | ≈ 3 km dug, never finished | over the Rhine–Danube divide | Rezat (Rhine) ↔ Altmühl (Danube) | link two great river systems | Charlemagne; abandoned (rain, soft ground) — a *failure* |
+| Stecknitz Canal | 1391–1398 | ≈ 95 km | summit canal, ≈ a dozen locks | Elbe ↔ Trave → Lübeck (Baltic) | salt trade (Lüneburg) | city of Lübeck; one of Europe's first summit canals |
+| Briare Canal | 1604–1642 | ≈ 55 km | summit level, ≈ 40 locks | Loire ↔ Seine | inland trade to Paris | French crown (Sully, Henri IV), then a company |
+| Canal du Midi | 1666–1681 | ≈ 240 km | summit ≈ 190 m, ≈ 60–100 locks (counts vary) | Garonne (Atlantic) ↔ Mediterranean | avoid the long, hostile route round Spain via Gibraltar | Riquet with Louis XIV's backing; ≈ 12,000 workers |
+
+## 2. Early industrial (locks, steam; still fits a magic-assisted age)
+
+| Canal | Built | Length | Summit / locks | Joins | Purpose | Notes |
+|---|---|---|---|---|---|---|
+| Bridgewater Canal | 1759–1761 | ≈ 65 km | level | coal mines ↔ Manchester | coal | halved coal prices in Manchester |
+| Caledonian Canal | 1803–1822 | ≈ 97 km (≈ ⅓ dug; rest is lochs) | summit ≈ 32 m, 29 locks | North Sea ↔ Atlantic across Scotland | avoid the stormy Pentland Firth; naval use; work for the jobless | built for frigates; ships soon outgrew it, so it was a commercial *disappointment* |
+| Göta Canal | 1810–1832 | ≈ 190 km (≈ 87 km dug) | summit ≈ 92 m, 58 locks | Baltic ↔ Kattegat via lakes Vättern and Vänern | avoid the Danish Sound tolls and route | ≈ 58,000 soldiers, ≈ 7 million man-days; obsolete as a trunk route soon after (railways) |
+| Erie Canal | 1817–1825 | ≈ 580 km | rise ≈ 170 m, 83 locks | Hudson ↔ Great Lakes | open the interior to the sea | freight cost Buffalo–New York fell by ≈ 90 %; made New York the leading port |
+| Welland Canal | 1829 (rebuilt 1932) | ≈ 43 km | ≈ 100 m, 8 locks (today) | Lake Ontario ↔ Lake Erie | bypass Niagara Falls | a lake-to-lake access canal |
+
+## 3. Ship canals (sea-going ships)
+
+| Canal | Built | Length | Summit / locks | Saves or opens | Notes |
+|---|---|---|---|---|---|
+| Suez | 1859–1869 | ≈ 164 km (now ≈ 193) | sea level, no locks | London–Bombay ≈ 7,000 km shorter (≈ 40 %) | company with state backing; forced labour early on; deaths in the tens of thousands (contested) |
+| Corinth | 1881–1893 | ≈ 6.4 km | sea level, cut up to ≈ 90 m into rock | ≈ 400 km round the Peloponnese | tried by Periander and Nero (begun AD 67, abandoned); too narrow for big ships, so it never paid |
+| Kiel | 1887–1895 | ≈ 98 km | sea level, locks at both ends | ≈ 460 km round Denmark | military first: the German fleet between the Baltic and the North Sea |
+| Manchester Ship Canal | 1887–1894 | ≈ 58 km | 5 locks, ≈ 18 m | ocean ships reach an inland city | built to bypass Liverpool's port dues |
+| Panama | French 1881–1894 (failed), US 1904–1914 | ≈ 80 km | summit lake ≈ 26 m, 3 lock flights | New York–San Francisco ≈ 13,000 km shorter (vs Cape Horn) | disease (yellow fever, malaria): ≈ 22,000 dead in the French attempt; a sea-level design failed, locks worked |
+| Volga–Don | 1948–1952 (attempted 1697 by Peter I) | ≈ 101 km | ≈ 88 m, 13 locks | Caspian ↔ Black Sea river systems | Peter's attempt was abandoned |
+| White Sea–Baltic | 1931–1933 | ≈ 227 km (≈ 48 km dug) | 19 locks | White Sea ↔ Baltic | forced (Gulag) labour, ≈ 12,000–25,000 dead; too shallow (≈ 3.5 m) for the ships it was meant for — a *cautionary* case |
+| Rhine–Main–Danube | 1960–1992 (Ludwig Canal 1836–1846 before it) | ≈ 171 km | summit ≈ 406 m, 16 locks | North Sea ↔ Black Sea by inland water | finally achieved Charlemagne's link, 1,200 years later |
+
+## 4. What this suggests for the sim
+
+**Why a polity digs** — four kinds of gain, often mixed:
+
+1. *Shorter route* (Suez, Panama, Kiel, Corinth): gain ∝ traffic × distance saved.
+2. *Access*, where no water route existed (Erie, Welland, Manchester, Lingqu, the Grand Canal joining river systems):
+   the largest gains. A whole hinterland or lake joins the sea economy; freight costs fall by up to ≈ 90 %.
+3. *Avoiding a hazard or toll* (Xerxes: a storm cape; Caledonian: the Pentland Firth; Göta: the Danish Sound dues;
+   Manchester: Liverpool's dues).
+4. *Military* logistics or fleet movement (Lingqu, Sui Grand Canal, Kiel, Xerxes). States accept a poor commercial
+   return for this.
+
+**Who can dig:** centralised states that can command mass labour (Persia, Qin, Sui, Louis XIV's France, Sweden's
+army). Later, chartered companies with state backing (Suez, the second Panama). Cities in leagues manage smaller works
+(Lübeck's Stecknitz). A polity must hold, or control, both ends.
+
+**Scale of effort, as a rough guide:**
+
+- Big pre-industrial projects take **3–15 years** of work by **10,000 to 100,000** people, or millions for the
+  Grand Canal.
+- Cost rises with length, with every lock (summits of 30–200 m are common; ≈ 400 m is the extreme), with rock
+  (Corinth), and with disease and climate (Panama).
+- Deaths can be large and are a story hook (the Pharaohs' canal, Panama, White Sea).
+
+**Failure is common:**
+
+- The Fossa Carolina and Nero's Corinth were abandoned; the first Panama attempt went bankrupt.
+- The Caledonian and White Sea canals were finished too small to matter.
+- Peter I's Volga–Don stalled.
+
+So the sim should let works fail or come out undersized, not just succeed.
+
+**Cheaper alternatives:** a *portage road* (Diolkos, used for centuries) gives part of the gain at a fraction of the
+cost. It is a natural first stage before a canal.
+
+**Decay:** canals need upkeep.
+- The Pharaohs' canal was reopened and silted up several times (finally closed 767).
+- The Grand Canal declined when the state weakened and the Yellow River moved (1855).
+- A canal should decay or close when its polity collapses. That fits the decline from High into Low Fantasy.
+
+**Ship size matters:** a canal that only fits barges (the Midi, the Göta) serves inland trade. One that fits sea
+ships (Suez, Kiel, and the Republic's neck cut onto a deep lake) joins two seas for the same fleet. The site scan
+should record depth and level on both sides so the sim can tell which kind of canal a site allows.
+
+## 5. Mapping to the model sketch
+
+- **gain** = traffic × days saved, plus an *access* bonus when no water route exists, plus a hazard/toll term, plus
+  a military weight for strong expansionist polities.
+- **cost** = length × (1 + climb / k_lock) × terrain (rock, marsh, disease) — labour-years. Labour-years / available
+  labour gives the build time, and the polity must stay stable that long, or the work fails.
+- **options:** a portage road (cheap, partial gain) → a barge canal → a ship canal.
+- **upkeep:** a yearly cost; it decays without a strong owner.
+
+Placeholder numbers only; tune against the cases above (e.g. the Erie access gain, the Suez distance gain, the
+Göta labour-years) when sub-project 2 is built.