I get asked this a lot, so let me answer it once, with the whole map: where ESVA's nine incursions come from, what each one uses as a reference, and why the order between them isn't random.
The hypothesis holding all of this up, canonical from the second incursion on, is easy to write down and uncomfortable to accept: the Ordem works from the idea that myth, folklore, and theology are the record of earlier Brechas. The creatures were never supernatural — they were sighted, and whoever sighted them had no vocabulary for "physical violation," only vocabulary for haunting. This isn't proven fact within the fiction itself. It's the Ordem's working hypothesis, and the game treats it as such — never omniscient narration confirming that "the angels were esvas."
The rule that keeps this from becoming "generation with an accent"
Every incursion has one dominant physical law, and the test I run before approving any of them is: describe the incursion without naming the country. If nothing distinctive survives that, it doesn't have an identity — it has an accent. That's why no incursion gets an "anchor class": the law is a lens that changes how the thirteen classes express themselves there, not a quota that hijacks the roster's distribution to fit a mythology.
1 — Primeiro Registro. The beta's baseline, where the Ordem's method was calibrated. No cultural lens — it's the control the other eight are measured against.
2 — The Interior. Heat with no source, biomass with no substrate. Reference: fauna, flora, and testimony from the Brazilian sertão, 18th through 20th centuries — often from a single person, never corroborated by a second witness. This is where the central hypothesis is born, not as a decree, but as the conclusion of an expedition that cross-referenced dozens of rural accounts with the old archive of the founding disaster and found a repetition too statistically uncomfortable to ignore. It leaves behind a silent Open Case: a Registro entry with no field name, because the only witness died before repeating the description to anyone else.
3 — The Witness. Non-local action, impossible geometry. Reference: angelological text twenty-five hundred years old — Ophanim, Seraphim, Cherubim, Thrones. This is the incursion that proves the method holds for sacred scripture, not just sertão tales, and it's where the Vigília becomes, for the first time, the most uncomfortable voice among the three clans — not because it disagrees with the reading, but because it disagrees with how coldly it's made.
4 — The Northern Margin. Impossible proportion and scale. Reference: Norse tradition — Sleipnir and his eight legs, Fenrir growing past every containment, Huginn and Muninn as "thought" and "memory" flying apart from the body. This is where the Ordem notices something that won't become central until the eighth incursion: a witness with no technical vocabulary can still report an exact number, and that precision survives centuries of oral transmission better than any other detail in the account.
5 — The River of No Return. Fabric that doesn't decay, time that doesn't pass. Reference: Egyptian iconography stripped of the artist's convention — Anubis, Sekhmet, Sobek, Thoth (I already wrote a whole post on this one, worth reading on its own). This is where the Ordem confronts, for the first time, the possibility that it isn't original — that it's reading the archive of a sister institution, dead for millennia, that arrived at the same institutional solution with no one left alive to explain why.
6 — The Closed Island. Metamorphosis — fabric that changes kingdom with no intermediate step. Reference: Greek mythology, but only the transformation axis — Daphne becoming a laurel, Arachne becoming a spider, Actaeon becoming a stag. The Chimera stays out because it combines three violations in a single body, against the one-violation-per-concept rule. To me, this is the emotionally costliest incursion in the whole arc: the Ordem's reading requires erasing a woman from her own story to fit the catalog, and it's the only time the Vigília and the Raiz unite against the Índice.
7 — The Hundred Years. An object that gains metabolism. Reference: Japanese tsukumogami — an object that reaches a hundred years of use comes alive — plus kitsune and bakeneko. Kappa and oni stay out for being humanoid. This is the law closest to a measurable clock the Ordem has ever catalogued — and also the first to break the line between "creature" and "manufactured object."
8 — The Dark Current. Distributed body, colony consciousness. Reference: the Chinese pantheon, mostly animal — Long, Qilin, Fenghuang, Pixiu, and Bai Ze, the beast that catalogued and dictated to the Yellow Emperor the eleven thousand five hundred and twenty known supernatural creatures. In the Ordem's reading, Bai Ze is the oldest existing description of an institution just like the Ordem — a Registro that walks. If it exists and is found, it would solve the roster's pool-coverage problem overnight. That stays on record as lore irony, never as an actual mechanical hook.
9 — The Open Case. The only incursion with photography, radar, and a living witness — and still the only one whose case never closes. Reference: modern cryptids. This is where the hypothesis stops evolving and starts collapsing on itself: if the instrumented present confirms nothing, the never-instrumented past can't honestly have been confirmed either — it was only accepted under a cheaper standard, because no one alive could contest it.
The mistake that nearly hijacked the whole roster
I actually proposed, at one point, giving each incursion fixed dominant classes — it felt natural to lean the Norse incursion heavy on CINÉTICO and GEO, for instance. I got corrected in time: that would have worsened the game's pool coverage, which was already calculated at 22.78% against a statistically derived benchmark — and worse, that benchmark's threshold has to be recalculated from scratch when the roster grows from 120 to 300 species, because the universe of pool combinations changes size in a non-linear way. Authoring the 180 missing species aimed at a number calculated for the wrong universe would repeat, with data, the same mistake "anchor classes" nearly caused with rules.
What got left out, on purpose
Australian Aboriginal mythology stayed out because it's sacred material under a living cultural protocol, not public-domain folklore — it would need formal consultation I haven't done yet. Slavic stayed out because its main figures are almost all human, without the metamorphosis escape hatch that saved the Greek one. Aztec and Maya stayed out because, even though Quetzalcoatl is an almost perfect fit, the pantheon around him is human and the context of sacrifice demands a care that a whole incursion doesn't repay — I'd rather use Quetzalcoatl loose, somewhere else, than force an incursion built around him.