Madain Sari — the Eidolon Wall

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.

The village again, and this time the part of it that was shut.

The room under the kitchen

Go into Eiko’s house, through the kitchen, and down. There are two chests: a Survival Vest and a Phoenix Down.

This is the room several guides tell you to walk past on your way to the fight. Do not.

The Eidolon Wall

Talk to the moogle at the entrance — he will heal you — and go in.

You get an Exploda and an Elixir handed to you in the scene. You cannot miss them.

Scarlet Hair

A one-on-one fight with Zidane.

It spends most of the fight dodging, then announces it is about to attack. Hit it when it announces. That is the rhythm of the whole thing.

Steal first — an Ether and a pair of Poison Knuckles. The Knuckles are worth taking because they are free, not because they are rare; they are sold in six shops for 5,000 gil, including the one in this village. If you bought a pair at Conde Petie, this is the money you did not have to spend.

Afterwards

Amarant joins, automatically.

Go back down to the room under the kitchen — Eiko has a Memory Earring for you, and she joins properly.

And then the game asks you to pick your party for the first time. Up to now it has been choosing for you.

One piece of advice about Amarant

Do not hand him your best accessories the moment he arrives.

He learns abilities from equipment, and some of the ones he wants are on weaker gear he will need to wear later. Two guides warn that kitting him out now makes that more annoying than it needs to be. Nothing is lost either way.

Game Notes

The party selection is the quiet event here, and it arrives with no ceremony at all.

For twenty-two areas the game has decided who walks with you. Zidane and Vivi because the plot needed them, Freya because you were in Burmecia, Quina because Quina followed you. You have never once been asked.

And then Amarant joins — the last one, and the only one who joins by losing a fight and shrugging — and the menu simply opens. No line of dialogue marks it.

It is placed exactly where it should be. The disc is about to end, the party is complete, and the game has just spent two areas showing you that these people have reasons of their own for being here. Then it hands you the list.

How this page was made

Researched across 5 community walkthroughs and written from scratch — no source’s wording appears here.

How we decide what to print

Completionist mode. Full spoilers, in play order, with every permanent loss called out before you reach it. 3 entries below are marked where our sources disagree with each other — we have not picked a winner for you.

A correction to what we wrote two areas ago

On the Madain Sari first-visit page we said we would render Scarlet Hair’s HP as 8,985, on the grounds that “the low camp is the one with a controller in it.”

That was wrong, and it was wrong in a way worth spelling out rather than quietly editing.

A second research pass found the low camp is not playtest reports. It is narrative FAQs that print a stat line and no other granular numbers. Worse, the cluster does not agree with itself: one prints 8,986, and a fourth GameFAQs boss FAQ prints 6,800, which no other source supports at all.

Figure	Sources
8,985	Jegged, ffexodus, Blackestmage’s boss FAQ, BlurYamaneko’s boss FAQ
8,986	bover_87’s walkthrough
18,985	Caves of Narshe, Game8 (two separate pages) — and our field
6,800	Gilga’s boss FAQ

Both 18,985 sources reproduce our entire stat block field for field — level 22, MP 5,865, 4,790 gil, zero experience, 9 AP — and then land on the higher HP.

So we had an argument that sounded like evidence and was not. The Armodullahan case had Jegged saying “deplete about 800 HP”, which is a sentence written by somebody holding a controller and is worth more than any stat table. There is no equivalent here, and we reasoned as though there were, because we wanted the question to be closer to settled than it is.

The page renders the dispute, not a number. That is what the disputed field is for, and this is the first time on this site we have used it on a figure we had already committed to.

The one thing that has got firmer: three consecutive clean boss controls — Hilgigars at 8,106 (six publishers), Soulcage at 9,765 (five publishers, no dissent), and every ordinary encounter through both areas. Whatever the offset is, it is confined to a subset and Scarlet Hair is in it while its immediate neighbours are not.

The item list is the most agreed-upon in the game so far

Nine publishers give an item list for this visit and all nine give the same one: Survival Vest, one Phoenix Down, Exploda, Elixir, the Memory Earring, and a Tent as the boss drop. Nothing else, from anybody.

Our field data matches exactly.

After the Iifa Tree — where two of the largest guides in the corpus were each an item short — this is worth noticing. The difference is that nothing here is conditionally rendered. Two chests in one room, two items handed over in a cutscene, one key item from dialogue. There is nothing to push, nothing to reveal, and so there is nothing to miss and nothing to disagree about.

Coverage quality tracks how the content is presented, not how important it is.

The Memory Earring, and id 258

Our hidden-room script carries an unnamed key item alongside the Survival Vest and the Phoenix Down. Last area we could not name it and listed the Memory Earring as an unconfirmed candidate.

Eight publishers are unanimous that the Memory Earring exists and that Eiko gives it to you in dialogue, in that room, after the fight — not from a chest. Gamer Guides and GameFAQs use near-identical wording: “return to the room under the kitchen to receive a Memory Earring Key Item.”

So the identification is about as safe as it gets, and the mechanism explains the data: a key item granted by dialogue sits in the same room script as that room’s chests, because it is the room’s script. Our table has no column that separates “opened a chest” from “was handed something here”.

We have written it into the checklist as the Memory Earring with the mechanism flagged, rather than adding 258 to the named-id table — because what nine publishers confirm is the item’s existence and location, not the number.

That distinction has mattered before. It is how the Coloured Stones got named and it is why 286 is still only a candidate.

Do the village’s chests come back?

Our second-visit script re-declares the Tent, the Ore, the Phoenix Pinion, the Libra coin and the Kirman Coffee — every first-visit pickup.

Fourteen publishers list none of them as available now. Not one of them discusses the question in either direction.

That is the same silence that hid the Lindblum refill, so we went looking properly this time. At Lindblum the corpus had numbers — three different gil figures for the same three chests, which is evidence even when nobody writes the sentence. Here the items are identical on both visits, so a refill would leave no trace in the published record at all. There is nothing to read.

So the honest answer is that we do not know, and the available evidence is entirely negative: nobody documents taking a second Tent here. That is consistent with the second script simply re-declaring the same one-time chests, which is the ordinary explanation, and it is what we have written.

Worth recording the asymmetry, because it will come up again: a refill is only detectable from the corpus when the contents differ between visits. Lindblum was findable because it is gil. An area that refills identical items is invisible to this method, and there may be several.

One single-source wrinkle: ffextreme alone frames the Exploda and Elixir as the same two chests you saw but could not open on the first visit, with a moogle opening them for you now. Every other guide describes a plain story hand-off with no chest involved. Interesting if true, because it would be a third thing a script can do, and uncorroborated.

Poison Knuckles, and the end of an argument

Stealable from Scarlet Hair — eight sources list it, one of them giving a 25% rate. Which closes the routing question we opened at Conde Petie two areas back, but not in the direction we said.

We wrote that the guides telling you not to spend 5,000 gil on Poison Knuckles at Conde Petie were right, and that the steal here was the reason.

They are right, and the reason is better than ours: the Poison Knuckles are sold in six shops, including Morrison’s, in this village, on this visit. Two publishers make the point directly. So it was never a scarce item at all — the steal is convenient rather than necessary, and the 5,000 gil at Conde Petie was avoidable for a much more boring reason than the one we gave.

We were right about the conclusion and wrong about the mechanism, twice in one page. Both are recorded.

Everything else

Amarant joins automatically — nine publishers, no choice involved. Three flag this as the first time the game lets you choose your own party.

Two publishers warn against loading Amarant up with good accessories immediately, since he learns abilities from weaker gear he will have to wear later. Not missable; just annoying to undo.

No ATEs on this visit. All five of the village’s ATEs belong to the first trip — three publishers itemise them there and none carries any forward. That is a genuine absence rather than a coverage gap.

Nothing here is permanently missable. No publisher flags anything, the Exploda is synthesisable later, and the Knuckles are in six shops. After the Iifa Tree, where every one of nine chests is gone forever, this is a rest stop in more ways than one.

The village is reachable once more, on the last disc with the airship, for a sidequest at the Eidolon Wall involving Lani and Garnet’s real name — which needs a party without Dagger, Eiko or Amarant. Three sources document it. No publisher says whether anything else is available then, so we do not either.

Game Notes

Two of this page’s three corrections are corrections to us, and both have the same shape: we reached a right answer through a wrong argument.

We said Scarlet Hair’s low HP figure came from people who played the fight. It didn’t. We said the Poison Knuckles steal was why you shouldn’t buy a pair at Conde Petie. It wasn’t — they are in six shops.

Neither error changed what a reader would do. That is exactly what makes them worth writing down, because an error that produces the right advice is one nobody complains about and nobody catches. It survives, and then it gets reused somewhere the conclusion does not hold.

The Ozma rules were both about distrusting our data. This is the third area running where the thing that needed distrusting was our reasoning about somebody else’s.

How this page was made

Researched across 8 community walkthroughs and written from scratch — no source’s wording appears here. 3 are marked where sources disagree with each other — we have not picked a winner for you.

How we decide what to print