Alexandria — the Coronation

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.

Disc 3 opens at home, with nothing chasing you. There is no boss and no dungeon.

Do everything here before you leave for Treno. Almost all of it closes when you go, and the chapter gives you no warning at all.

The footrace is the big one

There is a boy on Main Street called Hippaul. Race him, beat him, and his level goes up. Keep going.

Prizes at levels 10, 20, 30, 40, 50, 60 and 70 are cards. Level 80 gives the Athlete Queen, a key item.

That is eight prizes, and most players stop after two or three because nothing tells them there is a ladder. If you are going to grind one thing this chapter, grind this.

The steeple

Stiltzkin is here — Phoenix Pinion, Hi-Potion and Elixir for 777 gil. Fifth stop.

There is also a bell. Pull the rope and you get a Shiva Card and a Ramuh Card — but it does nothing at all if you are already holding 100 cards. It fails silently. Clear some space first.

And there are card players down there, by the steeple and on Main Street, if you play.

The shell game

The Nero brothers run one in the weapon shop. Cards as prizes, and winning nine in a row is worth an achievement.

Things lying around

• 3,927 gil and a Phoenix Pinion at the docks
• 2,680 gil on the mini-theatre floor
• Lapis Lazuli behind the fountain in the castle grounds
• an Ether and two more Phoenix Pinions around the castle exterior
• Leo, a Stellazzio coin, in the room with the Neptune statue

Garnet’s room

Doctor Tot gives her Opal, Topaz and Amethyst, which teach Shiva, Ifrit and Atomos. This happens in a scene; you do not have to find anything.

The ATEs

Six of them. They are the whole point of the chapter, and they go when you do.

Game Notes

This is the calmest hour in Final Fantasy IX and the game knows exactly what it is doing with it.

Disc 2 ended with a queen’s fleet, a summoned dragon and a character dying. Disc 3 opens with a boy who wants to race you down the street, a shell game in a shop, a bell you can ring for no reason, and a princess being handed three rocks by her old tutor.

None of it is required. All of it is on a timer.

That combination is unusual and it is deliberate — the chapter is about people getting one last ordinary afternoon, and it makes you feel the shape of that by quietly taking the afternoon away when you leave. You do not get told. You just go to Treno, and the footrace was yesterday.

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.

Our label for this area was wrong, and the field block covers three visits

Our area table called this “Alexandria — the Bahamut attack”. Six publishers, unanimously, say Disc 3 runs:

Alexandria (coronation) → Treno → Alexandria (the Bahamut attack) → Lindblum.

So the Bahamut attack is not the opening of Disc 3. It is the third chapter of it, with an entire other city in between — and our single field block contains both Alexandria visits, plus a later free-roam period once you have a ship and the city opens again for a second wave of pickups.

Three visits, one block. We checked the neighbouring block to be sure and it holds sixteen screens with no item awards at all — a pure cutscene set, the Alexander sequence and its aftermath. So every interactive thing in Disc 3 Alexandria lives in the one script set, gated by story flags.

This page is the coronation visit only. The Bahamut attack gets its own page, in its own place in the order, after Treno.

That is the third area-boundary correction in eight areas, and the pattern has now shown every face it has: blocks ordered against play (Conde Petie and the Black Mage Village), one place split across two blocks (the Iifa Tree), and now three visits collapsed into one block. The rule stands and has stopped being surprising: field id blocks record how the game was built, and tell you nothing reliable about what a player walks through.

Hippaul Racing is eight prizes and our data can count them

This is the densest missable in the game and the corpus undersells it.

Beat Hippaul, his level climbs, and prizes come at levels 10 through 80: Wyerd, Carrion Worm, Tantarian, Armstrong, Ribbon, Nova Dragon and Genji cards, then the Athlete Queen key item at 80.

Our field data for Main Street carries exactly seven card-range ids and one key-item id. Seven cards, one key item, one script, one location. That is the ladder, and it is the same shape of evidence that named the four Coloured Stones — a run of unknowns whose count and location match a set the corpus documents.

It also lets us say something the guides do not quite: the prizes are a single ladder in one script, not eight separate rewards scattered through the chapter. Stop racing at level 30 and the remaining five are simply never awarded.

We are not naming the individual card ids from this. The count matches; the order does not follow from it, and no public source gives a numeric card table to check against. Same discipline that has kept 286 a candidate.

The bell fails silently at 100 cards

Pull the rope at the steeple and you get a Shiva Card and a Ramuh Card.

If you are holding 100 cards, nothing happens. No message, no explanation. One publisher and several player reports cover this; the two largest guides do not mention the bell at all.

This is the most user-hostile thing in the chapter and it is documented almost nowhere. A player who has been collecting cards — exactly the player who wants these two — is the player most likely to be at the cap.

A ladder we thought nobody had documented, and a correction

A note here about the next Alexandria visit, because we found it while reading this block — and because the first version of this section was wrong.

The Knights of Pluto sequence during the Bahamut attack gives an Angel Earrings for assigning all four soldier pairs correctly. Our script for that scene carries four awards at even spacing: Hi-Potion, Ether, Elixir, Angel Earrings. The research pass we had at the time found two publishers describing the outcome as binary — right, or an Ether — so we wrote that the graded ladder was something “nobody has ever observed”, and built an argument on it about a corpus that cannot see outcomes nobody chooses.

That was wrong, and a second pass found it. Jegged publishes the exact mapping — nothing correct gives a Hi-Potion, one gives an Ether, two give an Elixir, all four give the Angel Earrings — and Final Fantasy Wiki gives the same four-way table independently. It is documented, precisely, and has been for years.

What survives is better than what we claimed, in two directions.

Our data corroborates a subtle detail. Jegged also explains why the ladder has four rungs rather than five: you cannot get exactly three correct, because assigning three pairs correctly forces the fourth. Our script has four award sites, not five. We read the shape without knowing the rule, and the rule explains the shape exactly.

And the narrower claim holds. The second pass searched specifically for players reporting a first-hand Ether or Elixir result — forum posts, trophy guides, talk pages — and found none. So the guides document the ladder and the player-report corpus does not corroborate it. That is a real and much more interesting gap than the one we asserted.

The lesson is ours: “no publisher covers this” is a claim about the sources we checked, not about the world, and we have now over-reached on it twice. The fix is procedural — a corpus-gap claim gets a second research pass aimed specifically at falsifying it before it goes on a page.

Card players nobody covers

One card-location FAQ names Michelle and an old lady on Main Street and Tom and a boatman by the steeple. No walkthrough we reached mentions any of them.

Our data supports the steeple pair — a five-card block in that script, listed twice, which is the branch-pair signature we have seen at Burmecia, Cleyra and Lindblum. One set of cards, two presentation paths.

Filed on the FAQ’s authority with the thinness stated. Card content is the least covered thing in this corpus by a distance, and this chapter is where that starts to bite.

When does it close?

Disputed, and the difference matters.

One publisher says the optional content closes “as soon as you leave this area and continue with the plot” — i.e. leaving Alexandria. A player report about the bell cards says they survive “until you leave for Treno”, which would give you the whole chapter.

Nobody resolves it. Our advice is to treat the stricter reading as true, because the cost of being wrong is the entire Hippaul ladder and the cost of being cautious is nothing.

Worth noting that a Gamer Guides achievement page lists the six ATEs and marks none of them missable, which is not the same claim as “they survive the chapter” and should not be read as one.

Leo, and a prediction that held

Three publishers put the Leo Stellazzio in the Neptune statue room here.

Our id table said the fifth coin — 308, by zodiac order — would be in Alexandria Castle on this disc, and it has no award row we can read, which we noted at the time as one of three coins handled by some other mechanism.

So the decode predicted the location and the corpus supplies it. Not a confirmation in the strong sense — we did not read 308 anywhere — but it is consistent, and the zodiac ordering has now survived every placement we have been able to test it against.

Everything else

Garnet’s room: Doctor Tot gives Opal, Topaz and Amethyst, teaching Shiva, Ifrit and Atomos. Three publishers, no dissent, and our field data has the three gems in that script.

Stiltzkin’s fifth stop at the steeple — Phoenix Pinion, Hi-Potion, Elixir for 777 gil, at the same 250-byte signature as the other seven bundles.

Pickups: 3,927 gil and a Phoenix Pinion at the docks, 2,680 gil on the mini-theatre floor, Lapis Lazuli behind the fountain, an Ether and two more Phoenix Pinions around the castle exterior — though publishers put those two Pinions at three different spots between them.

No boss. The library boss belongs to the next visit.

The city reopens later, once you have a ship, for a second wave — more gems, two large gil finds, and an Alexandria Card from a little girl. Two publishers document it, and they disagree about whether the city visibly reads as ruined by then. Different visit, different page.

Game Notes

The correction above is the thing worth keeping, and it is about us rather than about Alexandria.

We had a script showing four rungs where our sources showed two outcomes, and we reached straight past “the sources we checked are incomplete” to “nobody has ever seen this.” It is a more interesting sentence. It is also a claim about the whole world made from a handful of pages, and it took one more research pass to knock over — a pass we only ran because the content belonged to a later page and got looked at again.

The uncomfortable part is that the wrong version was load-bearing. It carried a whole argument about a corpus that cannot see outcomes nobody chooses, and that argument sounded right, and some version of it may well be true. But we had reached it from an error, and an argument reached from an error is not evidence even when its conclusion survives.

So the rule we are adding is narrow and procedural: a claim that the published record contains no account of something gets a second pass aimed at falsifying it, before it goes on a page. Not because the gaps are not real — the refilling chests at Lindblum are real, and no publisher has written that sentence yet — but because “nobody knows this” is the most satisfying thing a project like this can say, and satisfying is exactly where the checking should get harder rather than easier.