There is a state you have been in that you do not have a word for.
You know its shape. You could describe the physical facts — the sleep, the appetite, the way time behaved — and you have probably described them to somebody, badly, using the nearest available words, which were all slightly wrong. Stressed. Obsessed. Not coping. Each of them true enough to pass and none of them the thing.
And the absence of the word does specific damage. It makes the state feel like a personal failure rather than a condition, because a condition is something a lot of people have had and a failure is something you are doing. It also means nobody can tell you where you are in it, or whether it ends.
Madhya-līlā 23 supplies the vocabulary for the most extreme state this corpus records — and every strange thing in the last six chapters of the book runs on it.1
The corpus divides the highest grade of ecstasy in two.
Mādana — the condition of meeting. Mohana — the condition of separation.2
And mohana subdivides again:
Udghūrṇā — unsteadiness, the physical register. Citra-jalpa — literally variegated talk: the varieties of mad speech.3
Stop on what that structure asserts before going further. Separation is not the absence of the highest ecstasy. It is one of its two divisions — a state with its own name, its own subdivisions and its own contents, sitting alongside meeting rather than beneath it.
That is a structural claim and it is doing all the work. Every tradition has an account of union. This one has built out the other half with its own apparatus, and once you have done that, the person in the second condition is not waiting for the real thing. They are already in it.
There is also an asymmetry worth noting carefully, because it is easy to overstate. In the range read, the subdivisions are given for mohana and not for mādana. Separation gets an internal architecture — a physical register and a speech register, each named; meeting is named and not further divided here.
That may be an artefact of what this passage was for rather than a claim about the structure, and nothing on this page rests on it. But the corpus's development effort visibly went into the far half. Six chapters of the Antya-līlā are spent inside mohana and none inside mādana, and a vocabulary tends to get built where the work is.
Then the line that unlocks the whole Antya-līlā:
*"In separation from Kṛṣṇa, one experiences the manifestation of Kṛṣṇa, and one thinks oneself to be Kṛṣṇa."*4
Two claims, and they are not the same claim.
First: separation produces manifestation. Not memory, not longing, not imagination — the corpus says manifestation, and it means that the object appears in the condition defined by its absence. Absence is the mode in which the thing shows up.
Second: the boundary fails. One thinks oneself to be Kṛṣṇa. The subject who was longing and the object longed for stop being two, and the text records this as a feature of the state rather than as an error inside it.
Nothing in the madness chapters is intelligible without both. A man running at a sand dune because he has taken it for Govardhana, speaking in Rādhā's voice, addressing trees, is not being described as confused. He is being described as somebody in the state whose defining property is that the absent thing manifests and the boundary between seeing and being does not hold.
Which is why the build order matters and why this page had to be written first. The chapters that follow report symptoms; this chapter supplies the framework — and read without it, the symptoms read as pathology, which is precisely the reading the corpus has constructed a vocabulary to prevent.
The mohana material comes with a list of ten transformations, and the list is worth reading with attention because of how it is ordered and where it stops.5
Anxiety · wakefulness · mental agitation · thinness · uncleanliness · talking like a madman · disease · madness · illusion · death.
Three things about that sequence.
It is a progression, not a set. It begins with worry and sleeplessness — states anybody has had — and ends somewhere else entirely. The list does not present ten equivalent features; it presents a direction of travel, with the early items available to any reader and the late ones not.
The middle of it is ordinary. Thinness. Uncleanliness. Not eating and not washing are the two most reliable outward signs of interior collapse in any culture, and the corpus lists them alongside the exalted ones without marking a change of register.
And the tenth item is death.
That is the part a devotional reading has to work to absorb. The taxonomy of the highest ecstasy terminates in dying, stated flatly, as the last of a numbered list — not as a warning, not as a limit case, and not as something to be avoided.
Which tells you what kind of taxonomy this is. It is not a ladder with a summit you arrive at and remain on. It is a description of a condition with a terminal stage, and the corpus's other great statement about the same territory says the process has no terminating condition at all.6 A process that never completes, and a list of its stages that ends in death, are both on the record, and no chapter reconciles them.
The most useful thing on this page is not the content of the classification but what having one does.
A named state has a shape, and a shape can be located within. Somebody who knows the word udghūrṇā knows that the unsteadiness is a documented feature and not a private disintegration. Somebody who knows citra-jalpa knows that the speech is a category with divisions, which means that other people have talked this way and that the talking was recorded rather than treated.
That is the entire practical function of a technical vocabulary for extreme states, and it cuts both ways.
The gain is real: a person in a named condition is not alone in it, is not the first, and can be told where they are. The corpus's community can look at a man on the floor with his joints separated and say this is the fourteenth chapter's condition rather than something has gone wrong with him.
The cost is that the vocabulary decides in advance what the thing is. Once udghūrṇā is available, the unsteadiness is ecstasy — and no observation made from inside a community holding that vocabulary can come out any other way. The corpus is candid enough to have preserved an outsider who read the same body and said ghost, and its own protagonist saying epilepsy to soldiers.7 Three vocabularies, one body, and the vocabulary supplies the verdict before the observation is made.
Worth stating plainly, because the chapter is mostly something else.
Madhya 23 is largely Rūpa Gosvāmī's rasa apparatus — vibhāva, anubhāva, the sāttvika transformations, the thirty-three vyabhicārī bhāvas — and the vault already holds all of it from the Bhakti-rasāmṛta-sindhu material.8 This page does not restate it and does not compete with it.
What is net-new here is the top of the structure: that the highest grade divides in two, that the separation half has its own internal architecture, and that the state's defining claim is manifestation-in-absence with the boundary failing.
That distinction matters for reading the corpus. The rasa taxonomy is a system for classifying devotional emotion in general. This is a description of one specific extreme condition, and it exists because a particular man spent twelve years in it and his secretary took notes.9
The system came first and the condition was fitted into it. A vocabulary built to classify everybody's emotion turns out to have a compartment that exactly fits one person's twelve years — and whether that is because the system is good or because the notes were taken by people holding the system is not a question this corpus can answer about itself.
Layer A supplies the two divisions, the subdivisions of mohana, the manifestation-in-separation sentence and the ten transformations.10
Layer B expands the material into instruction on the stages of devotional development and on the exalted position of the gopīs' separation.11
Worth marking the difference in what each is for. Layer A is a taxonomy — terse, structural, offered as classification. Layer B is exhortation, and its interest is in the reader's position on the scale rather than in the scale.
Neither distorts the other here, and this is one of the cleaner chapters in that respect. But the ten transformations ending in death is a Layer A fact that Layer B has no use for, and it survives because the verse lists it, not because the commentary needed it.
You are in a state you cannot name. Find out whether somebody has already named it, in any tradition, clinical or otherwise. The name will not change the state and it will change two things that matter: whether you think you are the first, and whether you can tell where in it you are.
You are supplying somebody with a vocabulary for what they are going through. Notice that you are also supplying the verdict. Every technical name for an interior state carries a judgement about what kind of thing it is, and the judgement arrives before any observation. That is not a reason to withhold the word; it is a reason to know which word you are handing over.
You are reading a report of an extreme state. Ask what vocabulary the reporter held. The corpus's own material shows the same body producing ecstasy, possession and epilepsy from three observers, and none of the three was lying or careless. The verdict tracked the frame.
You are building a taxonomy of anything experiential. Include the terminal stage and do not soften it. This one ends its list at death and does not editorialise, which is why the list still reads as a description four centuries later rather than as an advertisement.
The evidence is a compact technical passage: two named divisions, two named subdivisions, one doctrinal sentence about manifestation in separation, and a ten-item enumeration.12
The tension the source doesn't close: the ten transformations end in death and the corpus's other account of the same territory describes a process with no terminating condition. A rate and a terminus are different claims, and both are made.
A second one. The taxonomy makes the state unfalsifiably devotional from inside — and the corpus itself preserves two other readings of the identical physical presentation, one from a fisherman and one from its own protagonist. The vocabulary decides the verdict, and the corpus supplies the evidence that it does.
Open: whether citra-jalpa was a category before this man or was built to hold him. The corpus presents it as classification and its only extended instance is Rādhā's bhramara-gīta recited by somebody in the state being classified.
This page and The Mirror Competition are the same territory described as structure and as motion. That page has a process with no terminating condition; this one has a ten-item list ending in death — and the corpus asserts both without noticing that a rate and a destination are different offers.
Against Rati-ābhāsa it is the same apparatus doing the opposite job. There the taxonomy is used to disqualify — a convincing imitation named as a mere glimpse; here it is used to legitimate, and in both cases the operative variable is one only an expert can inspect.
And it belongs beside Prema-vilāsa-vivarta, which is the same collapse of union and separation stated as a summit rather than as a classification. There a hand goes over the singer's mouth; here the identical content is written down as a category with subdivisions — and the corpus never remarks that it silenced the thing in one chapter and catalogued it in another.
State vs Trait and the FDIC Framework — that page's mechanism is the drowning man who shouts "Help!": his compressed, unhedged, commanding register reads on any diagnostic as high status, and he has no status at all. He has a state, and reading a state as a trait produces confident conclusions that are systematically wrong.
The mādana/mohana structure is a taxonomy built entirely of states, and the pairing shows what that buys and what it hides.
The diagnostic literature's problem is that observers default to trait readings. Somebody behaves strangely and the inference runs to what kind of person they are, because traits are what we are built to infer and states require knowing the situation.
This corpus does the opposite by construction. Every term in the taxonomy names a condition of a relationship — meeting, separation — and none of them names a property of the man. Nobody in the Antya-līlā ever asks what sort of person does this. They ask which state he is in.
Which gives the finding neither page reaches. A vocabulary made only of states makes trait inference structurally unavailable to the people using it — which is a genuine protection, and it is why this community could watch a man deform on a beach for twelve years and never once produce the sentence there is something wrong with him.
And it is the same property that makes the frame unfalsifiable. If there is no trait vocabulary in the room, there is no way to say that a condition has become a characteristic — and he is like this now is exactly the observation the ten transformations, ending where they end, would have needed somebody to be able to make.
Sharpest implication. A technical vocabulary for an extreme state gives the person inside it a location and gives everybody around them a verdict, and the verdict arrives before the observation. This corpus's version is built entirely of relationship-states with no trait terms available — which is why nobody in it ever concluded that something was wrong with him, and why nobody could have.
Generative questions.