Hilda Garde 3 and the Blue Narciss

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.

A timing note first. You get this ship at Lindblum on Disc 3, before Ipsen’s Castle. This page sits late in our numbering because the game files its walkable airship interiors at the very end of its own script order — not because the chapter happens late.

So: as soon as the ship is yours, walk the deck.

Two chests

• Elixir — up the ladder, at the top
• Tent — to the left of the same ladder

Both are reachable once Erin has left.

If you check a guide and it tells you there is nothing aboard, or only the Elixir, it is not lying to you — the published record genuinely disagrees about this ship. Two minutes at the ladder settles it.

Take Mogki’s letter

The moogle at Lindblum has a letter for Kumool, who is in Ipsen’s Castle. Pick it up before you fly — the castle is a one-way trip and the delivery is easier to do on the way in than to come back for.

What the ship can and cannot do

It lands on grass. That is the whole list.

Much later you get a ship that sets down on barren ground, desert, swamp and snow. If somewhere looks like it should be reachable and this ship will not land, it is probably not you.

The Blue Narciss

Nothing aboard it. It is a boat.

Game Notes

There is something quietly funny about the fact that the most disputed treasure count in this entire project is two consumables on an airship deck.

Not a boss’s hit points. Not a hidden key item at the end of a dungeon. A Tent.

Three guides, three answers — nothing, one thing, two things — about a room you walk through on your way to the controls. It is not carelessness. It is that nobody stops in a vehicle. You get an airship and the airship is a verb; the inside of it is a corridor between the door and the map.

Which is exactly why the game put things there.

Completionist mode. Full spoilers, in play order, with every permanent loss called out before you reach it. 1 entries below are marked where our sources disagree with each other — we have not picked a winner for you.

Three publishers, three counts, on two consumables

Our field data holds two chests on the deck: a Tent and an Elixir. It agrees with the publisher who has both, and the pairing is ordinary — both are tagged-union chests with gil partners that resolve cleanly.

One room-label note for our own records: our field layer calls the screen the Deck and Jegged calls it the bridge. Our zone names have missed seven times and we do not render them as locations; the ladder is the landmark that matters.

The shape of the gap

This is the second vehicle in three chapters where the published record thins out, and the pattern is consistent enough to name.

The Invincible holds exactly one chest in seven screens — the Pisces Stellazzio, which is to say the coin that completes a thirteen-part sidequest running the length of the game. Two publishers place it only as “aboard the Invincible”; one says the deck. It is routinely the coin people finish the game without.

The Hilda Garde 3 holds two chests and the corpus cannot agree whether there are none, one or two.

The Blue Narciss holds nothing, and our data confirms that, so at least one vehicle is honestly empty.

Coverage in this project has tracked attention throughout, and attention tracks the story — one screen off the critical path and the record thins. A vehicle is a special case of that, and a sharper one. A room is a place. A vehicle is a verb. You acquire an airship in order to go somewhere, and the interior is the corridor between the door you came in by and the controls you are heading for. Nobody stops. Nobody searches. Guide writers included.

Which is presumably why the designers put a sidequest-completing key item in one of them.

Timing, and an ordering problem we should own

This chapter is Disc 3. You board at Lindblum, before Ipsen’s Castle.

This page is numbered 42, after Daguerreo and the Invincible, because our area list follows the game’s own field-script order — and the game files its walkable vehicle interiors at the very end of that order, in a block with the Blue Narciss’s bridge screens. Field-id order has matched play order almost everywhere else in this game, which is why we use it; the vehicles are where it breaks.

We are not renumbering the list over two consumables, and we are not going to leave it unmarked either, because an unmarked ordering error is exactly the failure we criticised the corpus for two chapters ago. The timing warning is the first entry on the checklist and the first line of the short guide.

The broader lesson for anyone reading our method: field-script order is a very good proxy for play order and it is not play order. This is the fifth boundary problem we have hit and the first one caused by a category of screen rather than by a flag.

Everything else in the chapter

Mogki’s letter. The Lindblum moogle hands you one addressed to Kumool, who is in Ipsen’s Castle — a one-way dungeon. Carry it in; do not plan to come back for it. Both moogles are marked in our field data, which is the Mognet index we decoded four areas ago doing its job.

The Garnet. Beatrix gives it in Alexandria and it teaches Bahamut; it belongs to that chapter and is on that page.

Dagger rejoins, having cut her hair.

Landing: grass and only grass. The Invincible, much later, sets down on barren land, desert, swamp and snow — so anywhere this ship bounced off is worth another attempt in that one. Neither lands in forest or on mountains.

Game Notes

We came out of this small chapter with a new gate in the toolchain, and it was not because of anything on this page.

Writing it meant opening the Disc 4 Alexandria page to check how the two visits lined up — and that page said Stiltzkin’s final stop was the Alexandria steeple, 5,555 gil, while the page we had published four chapters earlier said his final stop was Bran Bal. Both were live. Both had passed the schema validator and the spoiler gate, because neither of those can see that two pages disagree.

Bran Bal is the seventh of eight. We had inferred “final” from a local fact — our data showed a three-item set in that room, we knew there were eight sets, and the sentence wrote itself. That is the second time in four chapters we have promoted a local observation to a global claim without checking it against an ordered list; the Mover superlative was the first.

So we wrote tools/claim-lint.py, which scans every page for superlatives, ordinals and exclusivity claims and flags any subject that carries the same kind of claim in two different areas. It is a lint, not a proof — it cannot tell you which page is right. It only tells you that two of them cannot both be.

On its first run it found a third instance of the same error, on the Terra page, that we had already missed while fixing the other two.

The general point: a validator that checks each page in isolation cannot catch the class of error that comes from writing pages one at a time. That is most of what this project does, so it was a hole worth closing — and it took a chapter about two consumables on a deck to notice it.