The Iifa Tree

Story mode. Written for a first playthrough and kept free of spoilers — it will not tell you what happens here, only what the game is asking of you.

You climb down the inside of the tree. It is a puzzle of platforms and branches and it is the densest set of chests in the game so far.

None of it opens again. Sweep it properly.

Nine things on the way down

• Ruby at the entrance, before you go in
• Phoenix Down, Hi-Potion and a Healing Rod in the inner roots
• Ether, Remedy and Lamia’s Flute further down
• Brigandine and an Elixir at the bottom

Some of these do not appear until you push the right side of a branch. That is the whole puzzle, and it is why people reach the bottom three items short and never know it. If a screen looks empty, push things.

The Lamia’s Flute is Eiko’s weapon and Eiko is not in your party. Take it anyway.

Before you fight Soulcage

Two things will kill a character outright if you walk in unprepared.

LV5 Death kills anyone whose level is a multiple of five. Check your party and change it if you need to.

Mustard Bomb inflicts Heat, and a character with Heat who takes any action is killed. Body Temp prevents it.

Soulcage

9,765 HP. It is undead, which is the whole fight.

Casting Life on it kills it instantly. A Phoenix Down almost does. This is unanimous across every guide that discusses the fight.

So the order is: steal first, then end it. Take the Oak Staff, the Magician Cloak and — if you are lucky — the Brigandine, then Life it. The drops come either way; only the steals are lost by winning too fast.

Do not cast Fire. Fire is technically a weakness and it triggers a counter that hurts you more than the spell hurts it. This is the single most common mistake people make here.

If Quina is with you

Stroper and Dracozombie both teach LV5 Death. Eat one.

This is the only thing in the area you can come back for — the lower root screens stay reachable later, even though everything above them does not.

Game Notes

The Life trick is one of the best-known things in FFIX and it is worth noticing what the game is actually doing with it.

Soulcage is undead. Undead things take damage from healing. That rule has been true since the first screen of the game, and it has been quietly relevant every time you threw a Phoenix Down at a zombie in the roots on your way down here.

So the boss is not a puzzle with a trick answer. It is a rule you have been taught, applied at a scale you did not expect. The game never tells you it will work on a boss. It just makes the boss undead and waits.

And then it lets you steal three items off it first, which means the optimal play is to know the answer and deliberately not use it for two turns.

Completionist mode. Full spoilers, in play order, with every permanent loss called out before you reach it. 4 entries below are marked where our sources disagree with each other — we have not picked a winner for you.

We nearly shipped this area with two items instead of nine

Our field data for the Iifa Tree showed one item and two moogles across fourteen screens. Seven publishers list eight or nine chests.

That is the first time the corpus has accused our extraction of missing things rather than over-counting them, and it is much the more alarming direction. An over-count is a tick box. A gap means the checklist is quietly incomplete, and a reader has no way to tell.

Our extraction was complete. The items are all there, in a second block of field scripts we had filed as a different area.

The game splits the Iifa Tree into two script sets. One covers the exterior — the Outer Seal, the trunk, the roots, the screens where the fleet arrives later. The other covers the interior descent, the inner roots and the bottom, and that is where every chest and the boss live. Our area table had read the second block as a later, separate underground location and placed it two areas further on, with the return to Madain Sari in between.

So the failure was ours and it was an area-boundary failure, not a reader failure. It is the second one in five areas, after Conde Petie and the Black Mage Village came out of the field ids in the wrong order.

The rule generalises past ordering. We wrote earlier that field id order approximates play order and is not it. The fuller version: field id blocks approximate areas and are not them. They record how the game was built. Two blocks can be one place, one block can be two visits — we hit that at Madain Sari last area — and neither fact is visible without asking the corpus what a player walks through.

Every remaining area now gets its item list checked against publishers before it is written, not after. That check is what caught this one, and it caught it because a researcher was asked whether an item-poor area was really item-poor.

The full list, and a tie our data breaks

Our field data carries all nine, in exactly those rooms.

The Ether is the interesting one. Only three publishers list it — GameFAQs, Game8 and RPGClassics. Jegged and Gamer Guides, the two largest and most careful guides in the corpus, have item tables for this area that do not include it.

It is in the script, at a known offset, in the same room as the Remedy and the Flute. So the minority of three is right and the two majors are short an item.

This is the first time our data has settled a dispute in favour of the minority rather than confirming a majority or correcting ourselves. It is worth being precise about why we trust it here: the Ether sits in the same award block, with the same structure, as six items nobody disputes. There is no branch pair, no purchase signature, no duplicate. Believing the six and disbelieving the seventh would take an argument nobody has.

Why an area like this gets miscounted

Several publishers describe the descent as a puzzle of hidden switches on the branches — chests that do not exist on screen until the correct side of a branch is pushed.

That explains the shape of the disagreement. This is not an area where you walk past a chest; it is one where the chest is not drawn. Two guides missing one item each, in an area whose items are conditionally rendered, is exactly what you would predict, and it is why the Healing Rod is missing from two lists and the Ether from four.

It also means the practical advice matters more than usual: if a screen looks empty, push things.

Soulcage is a clean boss record

9,765 HP at level 26. Five publishers, no dissent. Jegged, Game8, Caves of Narshe, Blackestmage’s GameFAQs boss FAQ and bover_87’s walkthrough all print the same number, and it is ours. Nobody prints 19,765.

That is now three consecutive clean boss controls — Hilgigars 8,106, Soulcage 9,765, and the ordinary encounters throughout — sitting either side of Scarlet Hair, which splits the corpus four ways. The offset is confined to a subset of records and we still cannot say which, or why.

Everything else about it matches: weak to fire and holy, halves shadow; steal Oak Staff, Magician Cloak, Brigandine; drops a Phoenix Pinion and Elixirs.

One publisher’s page read as though Soulcage resisted fire on one fetch and was weak to it on another. We are treating that as a page-rendering artifact rather than a claim, since the same publisher’s own stat table gives 150% fire.

The Life kill, and what it actually costs

Unanimous — every publisher that discusses the fight documents it. Soulcage is undead, Life kills it outright, a Phoenix Down takes it to almost nothing.

What it costs is the steals and nothing else. No publisher claims the trick forfeits AP; the fight gives no experience regardless; the drops fire on defeat however it dies. So the optimal line is to know the answer and deliberately not use it for two or three turns.

Fire is the trap. It is a listed weakness, and using it triggers a Fire Blades counter that costs more than the spell gains. Multiple guides flag casting Fire here as the most common first-timer mistake in the area — a rare case of the data-level fact (150% fire) being actively misleading as advice.

Two element reads we cannot settle

Our bitmasks and the one publisher with a full element table disagree twice:

All four of the area’s encounters match that publisher exactly on HP (Dracozombie 2,179, Stroper 1,840, Zombie 973, Myconid 1,372), which makes the element disagreement harder to dismiss as a different record. No second publisher with numeric element data was reachable, so it stands unresolved.

Worth noting in passing that forum players describe killing both Zombies and Dracozombies with Life, which is consistent with undead and says nothing about fire or ice either way.

The moogle letter nobody can reconcile

There are two Mocchis at the Iifa Tree, one on each pass through. A moogle database says the letter here is from Mogryo; three walkthroughs describe a letter from Stiltzkin.

These are probably the two different moogles rather than a contradiction, but no single source covers both, so nobody can say. Filed as unresolved — the same shape of problem as everything else that spans two visits to one place.

What closes

Everything except the lower roots. The force-field puzzle, the branch descent, the elevator, the Soulcage arena — all one-time. The nine chests are all permanently missable and no publisher labels them as such, because the area is mandatory and guides reserve the word for optional content.

The lower Roots screens stay reachable on later discs, which matters for one reason: Stroper and Dracozombie both teach LV5 Death to Quina, and they live on the screens that survive.

Game Notes

The near-miss here is the part worth keeping, and it nearly went the other way.

Our data said two items. Every instinct built over twenty-one areas says that when our data and the corpus disagree, we are the ones who are wrong — that rule came out of the Ozma retraction and it has been right every time since. So the expected move was to trust the guides and write the area from them.

That would have produced a correct page for the wrong reason, and left a five-hundred-row hole in our own data that nobody would have noticed until some later area came up short in the same way.

The thing that actually worked was asking the question as a question: is this area really item-poor? Not “who is right”, which invites a verdict, but “what would explain both”. The explanation turned out to be a boundary we had drawn ourselves, two areas away, for reasons that made sense when we drew it.

Most of what this project has found has been at a seam between two guide pages. This one was at a seam we introduced.