Reading a drop table: what "where to farm" actually tells you
A complete drop index gives you every place a material drops. That is genuinely useful and it is also not the whole answer — here is what a drop list can tell you, and the part you still have to decide yourself.
A game ships its progression the wrong way round. It tells you a unit needs four Manganese Trihydrate, and then leaves you to work out where Manganese Trihydrate comes from — a question the game never answers in one place, because the information lives in the stage tables, indexed by stage.
Reversing that index — every material, every repeatable stage that drops it, with the stage code and AP cost — is what a "where to farm" page is. It is a real answer to a real question. It is also worth understanding what it does not settle.
Tiers, not percentages
Arknights' own stage tables sort each stage's drops into bands rather than publishing a number per item. Our index carries those bands through unchanged:
- Normal — the stage's regular drop pool.
- Special — a rarer band, and in practice where most of the tier-4 materials live.
- Additional — the extra-drop band, rarer again.
The band tells you the shape of your luck, not its size. If you want an exact expected number of runs, no honest source can give you one from the game's published tables — because the game does not publish one. Anywhere you see a precise drop percentage for an Arknights stage, it is a community sampling estimate, and it should be labelled as such.
A tier is a fact from the game. A percentage is somebody's measurement. Both are useful; conflating them is not.
You can see how consistently the game applies this. Across all 73 farmable materials, the band tracks rarity: tier-1, tier-2 and tier-3 materials never appear in the Special band at all, while for tier-4 materials Special is the most common band — 123 stage entries, against 71 Additional and 36 Normal.
The distribution is far more uneven than it feels
Counting how many repeatable stages drop each material produces a shape most players never see laid out:
| Stages that drop it | Materials |
|---|---|
| 100 or more | 15 |
| 21–100 | 19 |
| 6–20 | 15 |
| 2–5 | 7 |
| Exactly one | 17 |
Orirock drops from 311 different stages. So do Damaged Device (310), Sugar Substitute (305), Oriron Shard (304), Ester (304) and Diketon (302). You will never plan a route for any of them; they accumulate.
At the other end, seventeen materials drop from exactly one stage in the entire game — the class Chip and Chip Pack items, plus Chiral Refractor. For those, "where to farm" has precisely one answer and the only remaining question is whether you can clear it.
Almost everything that actually blocks your progression sits in the thin middle: 22 materials with between two and twenty sources. That band is where a drop index earns its keep, and it is the band a wiki page organised by stage cannot show you.
Why the lowest-AP stage is usually not the cheapest
Manganese Trihydrate is a good worked example. Eight stages drop it, all in the Special band:
| Stage | AP | Band |
|---|---|---|
| 4-7 · Mutual Benefits | 18 | Special |
| 6-2 · Misunderstandings | 18 | Special |
| 9-15 · Scalding Earth | 18 | Special |
| R8-10 · Silent Butcher | 18 | Special |
| 13-20 · Sanguinary Billows | 21 | Special |
| 14-14 · High-Altitude Interception | 21 | Special |
| 10-7 | 24 | Special |
| 13-11 | 24 | Special |
The instinct is to run an 18-AP stage. But AP per run is not AP per material, and three things break the instinct:
- Rate differences inside a band. Two stages can both be "Special" and not drop at the same frequency. A 24-AP stage with a materially better rate beats an 18-AP one.
- What else falls out. A stage you are running for one material is also producing everything else in its pool. A slightly worse stage that also drops the second thing you are short of is usually the better run.
- Whether you can auto-deploy it. A stage you clear reliably at 18 AP is worth more than a stage you fail at 18 AP.
None of that is in the drop table, and no drop table anywhere can supply it. What the table does is reduce eight candidates from three hundred, which is the expensive part of the problem.
How to use it
- Start from the material, not the stage. Open where to farm and search what you are short of. Every repeatable source appears with its code, its AP and its band.
- Check the count first. One source means there is nothing to optimise. Three hundred means it is not your bottleneck. The middle is where a decision exists.
- Shortlist by band, then by AP — never AP alone.
- Cross-check against the other things you need. The best stage is usually the one that appears in two of your searches.
One honest caveat. Because the game publishes bands rather than rates, this page cannot tell you "expect 14 runs". Anyone who tells you that is quoting a community sample, which may be excellent and is still a different kind of claim from the ones on this site. We would rather give you the eight stages and say plainly what we do not know.