Returning models on the Spawn line (T-747)

6 Oct 2026, Pacific. Jordan approved the card on 6 Oct: the fix T-745 left open. Since T-745, rules that bring slain models back to a unit count for nothing (× 1). This card measures them on the Spawn line, the way the Tervigon's and the Apothecary's are measured. Scored with tools/recipestep.js against Track A2 (step 41), on snapshot 1b88e955b9b3. The scratch numbers are in build/recipe/t747-spawn-returns.json and build/t747/ (git-ignored). The live site's rankings don't read the ledger and are untouched.

In plain English

• What was built. Each army with returning rules now has its rules written down, in our own words:
  • the Genestealer Cults, Astra Militarum, Agents of the Imperium, Death Guard, Dark Angels, Drukhari, Thousand Sons, World Eaters and Space Wolves;
  • a returned model is priced at its own unit's cost per model, and the Spawn line counts it.
• What it did: nothing a reviewer would notice. On the held-out exam the gain over Track A2 is +0.0 [+0.0, +0.0]. All 15 armies are flat, and every fold picks the old ledgers.
  • Returning squads gain 15 to 37 points of Value, on Values of 350 to 1,550.
  • Poxwalkers: 814 → 831, still 16th. Neophytes: 28th → 27th.
  • The leaders don't move at all: Slaughterbound, the Exalted Sorcerer, the Haemonculus, the Wolf Priest and the Ravenwing Command Squad.
• Verdict: reverted. The keep rule needs a positive gain. The rules files are kept but switched off, and Track A2 on 1b88e955b9b3 stays the benchmark.
• No re-freeze was needed. The new files are the ledger's own live files, so no frozen input moved:
  • --verify passes 27 of 27, and --drift lists the same eleven files before and after;
  • the snapshot is still 1b88e955b9b3, so there is no new hash.
• Why so little: two reasons, both about the ledger, not the rules.
  1. The Spawn line is scaled army by army. Each line is divided by its median (T-608), and for Spawn that is the median over the army's own spawning units. If Death Guard has one spawner, its Spawn line is exactly "1 typical spawner", whatever the rule does. So every returning unit gains about wSpawn (15.65 in Track A2), against a Kill weight of 498.
  2. A leader's returns land on its joined forms. Each unit is ranked at its best solo form. The Wolf Priest alone brings nobody back, so its rank can't move. The Apothecary has always worked this way.
• What this means for the Cults and World Eaters. Their T-745 falls came from the old ×1.805 heal multiplier on the whole unit. The Spawn line, as scaled today, can't give that back.

1. The rules, as counted

Each rule was read on Wahapedia's 11th-edition datasheet page, and the numbers are written in our own words. Every rule can fire once a battle round. Availability over rounds 1 to 3 is 0.5, 1, 1, as for the Tervigon: the unit needs a model destroyed first. A returned model is priced at its unit's points per model at its smallest size.

• Curse of the Walking Pox. The trigger rate comes from the ledger's own Kill. Each form's kill['6'].byClass is enemy points destroyed per 100 of its points, by target class. Dividing by the typical field's points per model in each class gives models destroyed. Monsters and vehicles are left out, and the line takes the form's charge-model lands.
  • The points per model by class (from the field the ledger reads): chaff 8.75, elite infantry 14.3, cavalry and beasts 30, heavy infantry 35, character 95.
  • Ten Poxwalkers destroy about 2.3 enemy models a fight, so about 2.3 Poxwalkers come back.
• Not counted, and stated:
  • The rule's bearer is always alive.
  • There is always a dead model to return once the rule is available (the Starting Strength cap isn't modelled).
  • Typhus's Eater Plague counts through the joined form's whole Kill.
• The Dark Angels keep the Space Marines' Apothecary through include, unchanged. Legends datasheets are left out: the Deathwing Command Squad and A Hearty 'Pick Me Up'.

2. The prediction (ledger-steps.md, row spawn-returns, committed before the rebuild: 3afd22d156)

Right:

• The size: 10 to 30 Value; it came out 15 to 37.
• The leaders don't move.
• The Tyranids and the six unaffected armies are unchanged to the bit.
• Held out about 0, 12 to 15 of 15 not falling (15).
• Shuffled letters about 0.
• The verdict: revert.

Wrong:

• Neophytes rose 1 place, not 1 to 5.
• The Command Squads rose 0 to 1 places, not 1 to 5.
• The Genestealer Cults recovered +0.0, not "0 to +1.5".

3. The scores (tools/recipestep.js, base step 41 on the T-745 ledgers, candidate step 41 on the rebuilt ones)

• Fixed, over the 15 exam armies: pooled 60.8 → 60.8. Astra Militarum −0.1: Krieg's rise from 38th to 37th passes a graded unit the wrong way. Every other army is +0.0.
• Tyranids: 82.6 (three lists), 79.4 (five), 90.4 consensus, all unchanged.

Every unit whose Value moved (Track A2, rank and Value before → after):

A few other units move by 1 or 2 points of Value, with no change of rank: Deathleaper and the Red Terror in the Cults' pool, and Astra Militarum's characters. Step 41's character and size layers read the army's median Value, which moved by a hair.

4. What it would take

These are open questions for Jordan, and each would be its own declared step:

• A cross-army Spawn scale. Price a returned point against a killed point, which is what wSpawn meant before T-608 ("what a spawned point is worth against a point killed").
  • Ten Poxwalkers bring back about 15 points a round (20 per 100 of their points).
  • At half a killed point, that is roughly 10% more Value, not the 2% they get today.
  • It would also move the Tyranids' spawners, so it touches the training army.
• A leader's credit for what it returns. For example, the Wolf Priest's best joined form's Spawn, credited to its solo rank. The leaderCredit-style dials have failed before (−0.8), so expect little.

How to run it again

• Rebuild with the files on: LEDGER_SPAWN_RETURNS=1 node tools/ledger-all.js --army <armies>. First move the army's burst, burstfx, life, prefer and supportfx files aside, as T-745 says.
• Score it: node tools/recipestep.js --base tools/ledger-benchmarks/2026-10-06-step41-stable.json --base-dir <old ledgers> --cand tools/ledger-benchmarks/2026-10-06-step41-stable.json --cand-dir build/ledger.
• Without the variable every tool reads what it read before T-747, byte for byte. The Cults' and Dark Angels' spawn files were checked.

Doubts

• Half the Command phases on a held objective (the Cult Icon) and half the Pain tokens going to Fleshcraft are our estimates. They don't matter here, because the line's scale removes their size.
• The Medi-packs' returns into a joined squad are priced at the led squad's models. That is also our reading, though the datasheets say "PLATOON models".