The volume names three kinds of transgressive woman and never defines them.
The definitions are in an endnote:
A svairini is a loose and wanton woman. There were different types though. Svairinis are loose women who have sex with anyone they want, but only with those from the same varna. Kaminis are loose women who have sex with anyone they want, irrespective of varna. Pumshchalis have no sense of discrimination and are almost like harlots.1
Three terms, and the grading between the first two is stated with unusual precision.
Same appetite, same freedom, same absence of restraint in both. The only variable distinguishing a svairini from a kamini is whether her partners are of her own varna.
Set the three out and the axis becomes visible.
Svairini — anyone she wants, within her varna. Kamini — anyone she wants, regardless of varna. Pumshchali — no discrimination at all; "almost like harlots."
The first two are distinguished by exactly one thing and it is not frequency.
Not how often. Not how many. Not whether she is married, or whether she initiated it, or whether money changed hands — the last of those only enters at the third term, and Debroy hedges it (almost like harlots), which suggests the third category is not really about payment either.
The grading variable is varna-crossing. A woman who does the same acts the same number of times with the same lack of restraint is one category or the next depending entirely on who the other person was.
That deserves stating plainly because it is counter-intuitive and it is the finding.
A taxonomy of sexual transgression could sort by almost anything. Frequency. Consent. Marital status. Deception. Public knowledge. Whether children resulted. Every one of those is a plausible axis and several are what a modern reader would expect.
This one sorts by the partner's varna, and it does so at the point where the first two categories are otherwise identical.
Which tells you what the classification is measuring. Not the woman's conduct in itself — her conduct is held constant across svairini and kamini — but the boundary her conduct crosses.
So the object of concern is not the sexual behaviour. It is the varna line, and the woman is being graded by how much damage her behaviour does to it.
That is a system protecting a boundary, using a woman's classification as the instrument.
And now the finding that neither list produces alone, and that this build flagged during the Cluster A endnote pass as the strongest structural result of that lens.2
The volume has a second social-deviance taxonomy, also in the endnotes, also absent from the body: the eight shudra sub-classes.
Eight shudra sub-classes are mentioned — vyadha, vyalagrahi, vagatita, chandala, and others. A chandala is the offspring of a brahmana mother and a shudra father.3
And endnote 450 supplies the general term: a pratiloma marriage is one "where the mother's varna is superior to that of the father's."4
Put the two taxonomies side by side.
The women's list runs downward from a woman's behaviour: what she did, and specifically whether her partner crossed a varna line.
The shudra list runs upward from a child's parentage: who his parents were, and specifically whether they crossed a varna line.
They index on the same single variable. Sexual contact across varna boundaries, viewed from the two ends of one act.
One classifies the woman. The other classifies the child.
Between them they cover the complete consequence set of a single event — and neither list mentions the man.
That is not a small omission. In the chandala definition the father is named — a shudra father — and he is named only to establish the direction of the crossing. He has no category of his own. There is no term for a man who fathers a chandala, and no term for a man who is a kamini's partner.
The system produces a permanent status for the woman and a permanent status for the child, and nothing at all for the third party.
You will never hear either of these words used about anybody.
That is the first thing to understand about them. They are not village vocabulary. Nobody at the tank says she is a kamini.
What happens instead is that a family's marriage prospects change, and nobody can say precisely when or on what basis, and a household two generations later finds certain doors are simply not open.
The taxonomy operates as the reasoning behind decisions nobody articulates, and it survives in a translator's endnote because it was never something anyone needed to say out loud.
And here is what the pairing does to how you read the volume.
The text you actually hear at a recitation says none of this. Chapter 20 gives shudras two grades in a single clause and moves on. The women's terms do not appear in the body at all.
So the entire architecture — the grading of women by whom they slept with, the fixing of a child's status by his grandparents' varna — was too obvious for the text's own audience to require stating.
You are not reading a text that argues for this. You are reading one that assumes it so completely that it needed no words.
Worth being precise about what the volume does say, because the gap is the evidence.
The body text mentions women constantly and almost always in a ritual-permission register: women may worship the lingam; women should end with namah; some women should begin with it; brahmana women take instruction from gurus; women use the Tryambaka mantra for ash; widows follow the shudra rule; a Parthiva lingam for those with husbands, Sphatika for pravritti-widows, Rasa for nivritti-widows.5
That is a great deal of specification about access to practice, and it is on the whole permissive — chapter 28's flat statement is "everyone, women and others, have a right to worship the lingam."6
And there is not one word in the body about svairini, kamini or pumshchali.
So the volume's stated position on women is the ritual one, and it is inclusive. The classification of their sexual conduct exists in the layer the volume did not consider itself to be transmitting.
Two registers, and the reader without endnotes gets only one.
Return to the asymmetry, because it is the load-bearing structural fact and it should not be softened.
In an act involving two people that may produce a third:
He appears in the chandala definition only as a varna label establishing direction — a shudra father — and nowhere in the women's list.
Two consequences.
The costs are loaded onto the two parties with the least agency in the arrangement. A child had no part in it whatever. And the grading of women is precisely a grading of choices made about partners, which in the social arrangement the volume elsewhere describes were not, for most women, freely made.
And the enforcement runs through classification rather than through penalty. Nobody is punished in either list. There is no prescribed sanction, no atonement, no rite of return. There is a word, applied to a person, permanently.
Which is the same mechanism as vagatita — "one with whom one does not speak" — the shudra sub-class constituted entirely by other people's withholding.7 No enforcer, no appeal, no procedure. A status maintained by everybody and administered by nobody.
The last observation, and it is about method rather than about the doctrine.
Both taxonomies survive because a word happened to appear in the body that Debroy chose to gloss. He is a word-for-word translator who relegates interpretation to notes, so the note exists because the term did, not because he set out to describe the system.8
Which means the recovery is accidental, and its scope is exactly as wide as the vocabulary that happened to surface.
This build has now catalogued six such recoveries: the three kinds of bhakti, the three hardships, the six vices, the eight bonds, the five grades of emancipation, jati versus varna — plus these two social taxonomies.9
Every one is a system the body invokes and never states. And the pattern is consistent enough to license a general claim about this text: the Purana transmits applications and assumes structures, because its audience had the structures and wanted to know what to do.
Which has a hard implication for reading it. A reader without the annotation layer is not encountering a mystical or elliptical text. They are encountering a text with its assumptions removed — and the assumptions are where the social architecture lives.
Strongest support is the convergence: two independently-glossed taxonomies, in different endnotes, keying on one variable — varna-crossing — from opposite ends of the same act.
Tension one. Svairini and kamini are distinguished by nothing except the partner's varna. Frequency, consent, marital status and payment are all absent from the distinction.
Tension two. Neither taxonomy classifies the man. The chandala definition names a shudra father only to fix the direction of the crossing.
Tension three. The volume's body text is markedly permissive about women's ritual access — "everyone, women and others, have a right" — and this material is nowhere in it.
Tension four. Neither list prescribes any penalty. The mechanism is classification alone, maintained by everyone and administered by nobody.
Open question. Are svairini, kamini and pumshchali a standing dharmashastra scheme or a local one? Debroy notes "there were different types though," which suggests he is compressing a longer classification, and the pumshchali gloss is hedged ("almost like harlots") in a way the other two are not. Whether the third term belongs to the same axis or was appended from a different source is not determinable from the note.
The vault has very little on women in this domain, and what it has does not touch this.
Nirvāṇa-Ṣaṭkam: I Have No Caste, No Bond is quoted as the tradition's refusal of these categories — and it is a statement about what Śiva is, not about how a woman is classified. The vault should not let the hymn stand in for a social position it does not take.
Japa That Changes Varna and Sex is the volume's own most direct statement about womanhood — five hundred thousand repetitions "to counter the state of being women" — and it belongs beside this page. There, womanhood is a status to be exited. Here, a woman's sexual conduct is graded by the varna of her partners.
Read together: the volume offers women a ritual exit from being women, and classifies them meanwhile by whom they crossed with. Both in the same Samhita, neither aware of the other.
And The Seven Infertilities is the third piece of this — childlessness classified in seven kinds, one of which is a woman who delivers only daughters. Three endnote taxonomies about women, none in the body.
The Chandala's Wood Block: Caste Architecture (history). The historical record shows caste maintained through physical infrastructure and required signalling — a system that needs markers precisely because the categories are not visible.
These two taxonomies supply what the infrastructure was built around, and they show the boundary being guarded at its reproductive edge rather than at its social one.
What the pair produces: an endogamy rule is enforced twice — once on the woman who crosses and once, permanently, on the child — and the second enforcement is what makes the first credible, because a penalty that ends with the offender is survivable and one that runs to your descendants is not. That is the mechanism, and it is visible only when the two endnotes are read together, which is why neither the body text nor either note alone reveals it.
Identity Scripts: How Shame Becomes Self-Concept (psychology). The developmental material distinguishes a status conferred by naming from one imposed by sanction — and finds that named statuses without procedure are more durable than punished ones, because there is no verdict to appeal and no sentence to complete.
Neither taxonomy prescribes a penalty. There is a word, applied, permanently.
The insight neither gives alone: classification without sanction is the more permanent instrument, and a system that grades rather than punishes has chosen durability over enforcement — nobody has to act, nothing has to be proved, and there is no point at which the matter is settled. Which explains why these schemes survive in vocabulary rather than in law, and why they were never written into the body of a text that legislates defecation in detail. A rule needs stating. A category only needs using.
Sharpest implication. Svairinī and kāminī are distinguished by exactly one variable — whether her partners are of her own varna — with frequency, consent, marital status and payment all absent from the grading. And the volume's other endnote taxonomy, the shudra sub-classes, defines a chandala as the child of a brahmana mother and a shudra father: pratiloma, against the grain. Both systems index on the same single variable — sexual contact across varna lines — from the two ends of one act. One classifies the woman, the other classifies the child, and neither mentions the man, who appears in the chandala definition only as a varna label fixing the direction. Neither prescribes any penalty: the mechanism is a word, applied permanently, maintained by everybody and administered by nobody. And neither list is in the body of the Purana — both survive only because a translator glossed a term.
Generative questions.