The Desert Palace

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.

Kuja’s palace is a candle puzzle. Light things in the right order, statues move, staircases appear.

The six bloodstones are the point

Each room you finish gives you a bloodstone, and each bloodstone hands you a piece of equipment.

More importantly, each one disables one of the boss’s buffs.

Valia Pira opens the fight by raising its magic power, its defence, its evasion, its magic defence, its magic evasion, and its immunity to every element. Every bloodstone you found beforehand takes one of those away.

Find all six and the fight is easy. Find none and it is the hardest thing on the disc. Six guides say this and none of them disagrees.

The six:

• Promist Ring — the lobby
• Anklet — the brazier room, after lighting all three
• Shield Armor — the hallway
• N-Kai Armlet — the library, upper level
• Black Hood and Venetia Shield — the shadow chamber

And one actual chest

Fairy Earrings, in the room with the statues. Light the candle to raise the staircase. It is the only ordinary chest in the palace.

Valia Pira

12,119 HP. It is a pure caster and never attacks physically.

Carbuncle turns the fight off. Its Reflect bounces Valia Pira’s own magic straight back. You can also just drain its MP if you would rather.

Do not plan to steal anything — Zidane is at Oeilvert and nobody here can.

If Quina is with you

Level 5 Death clears three of the four encounters. Torama, Drakan and Grimlock are all level 30. Ogre is level 29 and is the only one it misses.

The moogles

Mojito runs the shop in the lobby and has a letter for Mogsam, who is in the library behind a wall that opens once you have lit enough candles.

Mogsam is your last save before the boss.

Come back for the card

Once you return from Oeilvert there is a Namingway Card on a table in Kuja’s room. It is not behind anything, which is exactly why people miss it.

Game Notes

The bloodstone mechanic is the best-designed thing in the second half of this game and almost nobody notices it is happening.

You are not told the bloodstones do anything. They look like what they are on the surface — a reward for solving a room, one accessory each. A player who solves four rooms and gets bored takes four accessories and walks into a boss fight with two-thirds of its buffs intact, and concludes the boss is hard.

A player who cleared the palace walks into the same fight and finds a caster with no defences that dies to its own spells.

Same boss. Same party. The difference was made an hour earlier by something the game never explained, and both players come away with a completely coherent and completely different impression of how difficult Final Fantasy IX is.

Completionist mode. Full spoilers, in play order, with every permanent loss called out before you reach it. 2 entries below are marked where our sources disagree with each other — we have not picked a winner for you.

Valia Pira is the cleanest +10,000 case we have found

Five independent publishers against one, and the one is the one that has matched our raw field four times before. This is the fifth.

The ±1 spread inside the low camp — 12,119 against 12,120 — is the ordinary signature of old fan FAQs copying each other with a typo, and it is exactly the kind of internal noise the high camp never shows, because the high camp is reading a file.

Two things make this better evidence than Scarlet Hair, where the split was four ways and we ended up rendering the dispute:

The low camp is five deep and includes Game8 — which printed the high figure for Scarlet Hair. So Game8 is not on a side. It aggregates, and it aggregates to 12,119 here and to 18,985 there. A publisher that switches camps by record is evidence that the camps are about the record, not about the publisher.

And Ark, one area over, is a clean 20,002 with five publishers and no dissent. A twenty-thousand-HP boss with no offset, in the same chapter as a twelve-thousand-HP boss that has one. Whatever selects the affected records, it is not magnitude.

We render 12,119.

Running tally: confirmed carriers are Zaghnol, Deathguise, Lani, Armodullahan and Valia Pira. Confirmed clean are Hilgigars 8,106, Soulcage 9,765, Ark 20,002, and every ordinary encounter we have checked. Scarlet Hair remains disputed. Still no structural discriminator, and the playtest stands.

Every chest in this palace is scripted four times, and nowhere else in the game

Our extraction shows each bloodstone reward appearing exactly four times in its room’s script:

Eleven publishers give an item list for this palace and every one of them lists one of each. Nobody has ever found two Promist Rings.

So we went looking for the mechanic. There isn’t one. No publisher describes a rotating room, a switch that changes the layout, a randomised element, or a room enterable in several states. The palace is a linear candle puzzle and everybody describes it the same way.

Then we ran the pattern against the whole game:

Six items in Final Fantasy IX are awarded four times inside a single field script. All six are in the Desert Palace. All six are bloodstones.

Out of 318 item-and-field groupings in the entire table, there is nothing else at four. And the one award in this palace that is not a bloodstone — the Fairy Earrings, in an ordinary chest — appears once, like everything else in the game.

So the ×4 signature belongs to the bloodstone award mechanism, and bloodstones exist in exactly one dungeon. That is a complete account of the pattern’s scope, which matters more than knowing why the number is four: it means this is not a reader fault that might be quietly inflating counts elsewhere. It is confined, it is explained by an award type, and it stops at the door.

We cannot say why four. One plausible reading is that a bloodstone is an examinable set-piece rather than a chest, and a set-piece usable from several camera angles may need its trigger duplicated per screen segment — but no publisher says anything of the kind, and we are not going to dress a guess as a finding.

This is now the seventh distinct duplicate pattern the instruction-count rule has produced, and the first tied to a specific kind of award rather than to a branch or a visit.

The bloodstones defuse the boss, and that is the actual dungeon

Six publishers, no dissent, and it is the strongest practical finding in the chapter: each bloodstone you collect disables one of Valia Pira’s opening buffs.

It opens by raising magic power, defence, evasion, magic defence and magic evasion, and nullifying every element. Six buffs, six bloodstones.

Guides that discuss it call the fight “very easy” with all six and describe a much harder one without. Which means the difficulty of this boss is set an hour before you meet it, by something the game never explains, and two players can come away with completely coherent and completely opposite impressions of it.

Carbuncle is the other answer. Valia Pira is a pure caster with no physical attack, so Reflect bounces its own magic back. Two publishers also note you can win by draining its MP, and one suggests Quina’s Night plus Steiner if you have no Carbuncle.

And nobody can steal: Zidane is at Oeilvert, which two publishers point out and which is easy to forget when every other boss page in the game opens with a steal list.

The weakness is split, and the split is informative

Our field reads null earth, weak wind, matching the two stat-table sources.

We checked whether that mask means anything, because Ark next door carries the identical one. Eighteen distinct enemy records in the game share it — a real recurring mask, not a default, and one of the eighteen is Zemzelett, whose earth immunity and wind weakness two publishers confirmed independently back at the Black Mage Village.

So our reader handles this mask correctly where it has been checkable. The water recommendation circulating in three guides looks like received wisdom rather than a stat table — possibly seeded by Water Sword being a good answer to a caster for reasons that have nothing to do with elements.

We render the dispute, because three publishers is not nothing.

One claim we are not carrying: a single GameFAQs FAQ says Valia Pira drops a Shiva card. Seven other publishers mention no card at all.

Our data corroborates a design coincidence

Torama, Drakan and Grimlock all have 3,292 HP at level 30. Ogre has 3,300 at level 29.

Three different enemies sharing an HP figure to the digit is exactly the sort of thing that looks like an extraction fault, so it is worth stating that it is not. Two publishers print the same figures — and a third confirms the consequence without printing any numbers at all: three of the four enemies here have levels divisible by five, and Ogre is the exception.

That is a qualitative claim about a design decision, written by someone thinking about Quina’s Level 5 Death, that independently confirms our numeric read. Rather a good check, arriving sideways.

Everything else

Mojito runs the lobby shop and carries a letter for Mogsam in the library, behind a wall that opens as the candles are lit. Mogsam is the last save before the boss and several guides recommend a Tent there.

No Stiltzkin — a dedicated moogle guide lists his stops and this is not one.

No ATEs. Two purpose-built ATE catalogues jump from Lindblum straight to Bran Bal with nothing in between for either half of this chapter.

The Namingway Card sits on a table in Kuja’s room once you return from Oeilvert. Two publishers flag it missable, and it is missable in the dullest possible way — no puzzle, no trigger, just a table nobody looks at.

The party split is free and consequence-free. Zidane must go to Oeilvert; you pick three to go with him. No publisher reports any lasting effect. The only real consideration is that Oeilvert seals magic and this place does not, so the mages belong here.

The palace closes after Terra, per one publisher — single-sourced, and uncontradicted.

Game Notes

The bloodstones and the +10,000 question are the same shape of problem from opposite sides, and this area has both.

Valia Pira’s difficulty depends on six things a player may or may not have done, and the published record handles that beautifully: six guides independently worked out the mechanism and all say the same thing. It is the kind of fact you get from playing, twice, badly and then well.

Valia Pira’s HP depends on nothing at all, and the published record splits five to one, with the one matching a file. It is the kind of fact you get from reading.

Neither method is better. The guides know what happens when you fight it; the files know what is stored. This project has spent thirty areas learning that the gap between those two things is where everything interesting lives — and that when they disagree, the right question is almost never “who is right” but “what would produce both.”