diff --git a/docs/LANCES.md b/docs/LANCES.md
index ae66168..7ba13ed 100644
--- a/docs/LANCES.md
+++ b/docs/LANCES.md
@@ -132,27 +132,33 @@ findings with different fixes, and one column cannot tell them apart.
 
 Two corpora of 20,000 at different seeds, same settings, compared row by row:
 
-| row size (satisfies) | spread between seeds |
+| row size (satisfies) | largest disagreement |
 | --- | --- |
-| above 2,000 | within ±1% |
-| 700 to 2,000 | up to ±9% |
-| 200 to 700 | 10% to 15% |
-| under 50 | 10% to 30% |
-
-That is what a binomial says it should be: a row of *k* lances carries a
-relative standard error of about `1/sqrt(k)`, so **±5% needs about 400 lances
-in the row** and 20,000 draws buys that down to a satisfies rate of 2%. A row
-of thirty is ±20% and should be read as *"this happens"* and not as a rate.
-
-The **wins** column is noisier than **satisfies** at the same size, because it
-is what survives precedence rather than what qualifies: `Fire` moved 48 to 35
-between the two seeds and `Light Fire` moved 5 to 1. Quote a wins figure only
+| above 2,000 | 1.7% |
+| 700 to 2,000 | 16.4% |
+| 200 to 700 | 8.7% |
+| under 200 | 28.3% |
+
+**A binomial underestimates this and should not be used to plan a sample.** A
+row of 1,250 would be ±2.8% if every lance were an independent trial, and
+`Hammer` moved 1,246 to 1,449. Lances are clustered: a draw fixes a faction and
+a year and then takes four correlated designs out of one table, so the
+effective sample is a good deal smaller than the lance count. Measure the
+spread with a second seed rather than computing it.
+
+What that leaves usable at 20,000 lances: **rows above 2,000 carry a rate**,
+rows between 200 and 2,000 carry an order of magnitude, and rows under 200 say
+*"this happens"* and nothing more.
+
+The **wins** column is worse than **satisfies** at the same size, because it is
+what survives precedence rather than what qualifies. `Hammer` moved 3 to 12
+between the two seeds and `Light Battle` 143 to 109. Quote a wins figure only
 where the satisfies row above it is large.
 
 ## What this corpus cannot reach
 
 `Order` is the one formation of the thirty-three that no lance satisfied in
-60,000 draws across three settings. It wants **four machines of the same
+80,000 draws across four settings. It wants **four machines of the same
 model**, and the generator draws each slot independently from the faction's
 table, so more samples is not the fix. (Its `faction: DC` gate is not what
 excludes it: `Formation::qualifies` does not read faction at all, because a
@@ -163,3 +169,43 @@ that *satisfies* a named formation, which is the tool for that row and for
 `Artillery Fire`, `Light Recon` and `Horde` beside it. It is not what this
 corpus is - a targeted draw answers "what does an Order lance look like" and
 this answers "what turns up" - and the two should not be mixed in one file.
+
+## What twenty thousand lances came out as
+
+`--seed 20260904 --count 20000`, defaults otherwise. Every lance classified as
+something, because `Support` admits any four ground Meks and is last in the
+order. The rows that matter are the ones above it:
+
+| formation | satisfies | wins |
+| --- | --- | --- |
+| Support | 20,000 | 538 |
+| Command | 12,504 | 1,786 |
+| Security | 7,125 | 4,584 |
+| Ranger | 6,967 | 168 |
+| Urban | 6,427 | 1,615 |
+| Heavy Battle | 5,736 | 25 |
+| Anvil | 5,284 | 3,293 |
+| Berserker/Close | 5,149 | 1,964 |
+| Battle | 5,149 | 0 |
+| Assault | 1,195 | 1,195 |
+| Heavy Recon | 1,366 | 1,344 |
+| Rifle | 781 | 652 |
+
+Two things in there are worth an argument rather than a note.
+
+**`Battle` never wins.** Its satisfies count is identical to
+`Berserker/Close`'s in every arm measured - 5,149 and 5,149 here, 5,080 and
+5,080 at another seed - and `Berserker/Close` is above it in the order, so
+`Battle` is classified as `Battle` never. Either the two criteria are the same
+criteria, or the order is wrong; the corpus says which pair to look at and does
+not say which of the two it is.
+
+**`Ranger` and `Heavy Battle` are large and almost never win** - 6,967 to 168
+and 5,736 to 25 - where `Assault` and `Heavy Recon` win nearly everything they
+qualify for. That is the precedence order doing its job, and it is the number
+to have in hand before changing it.
+
+The two sampling settings are worth less than they look. `--ratings drawn`
+moves nothing above its own noise. `--scope any` gives almost the same
+distribution and wastes 63% of its draws on faction-year pairs that did not
+exist - 54,654 draws for 20,000 lances, against 21,436 for the default.