Śuka is laying out the cosmic body for Parīkṣit to meditate on. Partway down the enumeration:
maharloka is his neck, janarloka is his mouth, taparloka is his forehead and satyaloka is the crest of this thousand-headed one.1
A few lines later, still the same body, still the same speech:
Scent is his sense of smell and the blazing fire is his mouth.2
And a page after that, closing the enumeration:
The great-souled one has the brahmanas as his mouth, the kshatriyas as his arms, the vaishyas as his thighs and the dark-complexioned ones as his feet.3
Three occupants, one mouth, one chapter. Nobody mentions it.
Once you see it, the collisions are systematic rather than incidental. Set the two main assignment schemes side by side and they overwrite each other at four points.
Feet. Pātāla is read as the soles of his feet.4 The dark-complexioned ones are his feet.3
Thighs. Vitala and atala are the two thighs.4 The vaiśyas are his thighs.3
Arms. Indra and the other resplendent gods are said to be his arms.2 The kṣatriyas are his arms.3
Mouth. Janarloka.1 Fire.2 The brahmanas.3
So there is a loka-map, which assigns the fourteen worlds bottom to top, and a varṇa-map, which assigns the four social orders, and the second lands squarely on top of the first at every point where both have something to say. This is not a case of one map being detailed and the other rough. They are the same length in the places they overlap, and they disagree completely.
A third scheme runs through the same passage and collides with both — the elemental and divine assignments. Fire at the mouth. The firmament as eye-pits and the sun as eyeballs, when the firmament was already the depression of his navel.1 Water as his palate. Yama as his tusks. Brahmā as his penis, when Brahmā is also the resident of satyaloka, which is the crest of the head.2
That last one is worth sitting with. Satyaloka is Brahmā's world, and satyaloka is named as the crest. Brahmā is then separately named as the genitals. The same figure occupies the top and the bottom of the same body, four lines apart.
Register: this is a description of what the chapter says, not an argument about how it got that way. No claim is being made here about redaction, layered composition, or authorial intent. What can be shown is only that the assignments are present, incompatible, adjacent, and unremarked.
The tightest instance is not in Skandha 2 at all. It is in the Fourth, where brahmanas addressing Pṛthu describe their own origin and their own function in consecutive breaths:
…from his mouth and you are full of austerities, learning and yoga, and are free from lust. The one with the one thousand hands and one thousand feet created us for your protection. It is therefore said that brahmanas are his heart and kshatriyas are his limbs. Hence, brahmanas and kshatriyas protect each other.5
Read the joint. The brahmanas came from his mouth. Therefore brahmanas are his heart.
The therefore has nothing to carry. Nothing about origin at the mouth entails residence in the heart, and the standard assignment — the one given in Skandha 2 — puts brahmanas at the mouth and kṣatriyas at the arms, not at the heart and the limbs. Inside a single sentence the speaker uses the mouth-origin as a premise and arrives at a conclusion that contradicts the mouth-assignment.
And it is doing visible work. The heart-and-limbs version produces mutual dependence — hence, brahmanas and kshatriyas protect each other — which is exactly the point the speakers are making to a king whose protection they want. The mouth-and-arms version produces a rank order, which is not the point they want to make here.
So the body is not being reported. It is being used. Whichever assignment yields the needed conclusion is the one that appears, and no accounting is kept.
Debroy annotates the passage and does not flag the discrepancy. That is consistent with the pattern already logged across this volume — internal disagreements glossed locally, none marked as disagreements.
Here is the thing that makes this more than a list of errors: the passage is not incoherent to read. You can go through the whole enumeration without stumbling.
The reason is that the assignments are never used referentially. Nobody in the text ever asks what the mouth is. The practitioner is not looking up parts; they are moving through them, one at a time, and each assignment does its work in the moment it is uttered and is then left behind — which is exactly what the instruction says to do with each limb.
A contradiction requires two claims to be held at once. This text never holds two at once. It holds one, then releases it, then holds the next. Under that discipline the collisions are structurally invisible: they exist in the text and cannot arise in the practice.
Which is a real observation about the genre and not a defence of it. The same feature that makes the passage usable makes it unusable for anything else. You cannot build a cosmology out of it, because a cosmology has to answer questions asked in any order.
The two schemes do different jobs, and that is the best available account of why both are here.
The loka-map establishes scale and orientation. It tells the practitioner that the object is the whole vertical cosmos and gives the traversal its direction — start at the underworld, end at the crest. Its content is spatial.
The varṇa-map establishes social order as divine anatomy. Its content is not spatial at all; it is a ranking, and it is stated in the language of body parts because that is the register that makes a ranking feel structural rather than arranged.
Both need the same four locations because those are the locations that carry rank in a body — head, arms, thighs, feet. There was never room for both.
The collision is therefore not sloppiness about anatomy. It is two different projects competing for the same small number of slots that mean anything.
There is a moment where a reader might expect an accounting and does not get one. Śuka finishes the whole enumeration and says:
I have told you about the way in which the lord's body is spread out. One should focus dharana on this gross form, using one's mind and intelligence. There is nothing beyond this.3
The body. Singular, definite, complete. The summary treats what preceded it as one object that has now been fully described.
That is the moment where a text keeping accounts would reconcile, or rank the schemes, or say which one to hold. Instead it asserts completeness and moves to the next instruction. The absence is not concealment — there is nothing to conceal from a reader who is not checking — but it is the point at which the passage commits to being a practice rather than a description.
You maintain something with parts — a strategy doc, a set of personas, a taxonomy your team files things under, a framework you teach. It has never given you trouble.
The Bhāgavata's failure mode is the one to check for, because it does not announce itself. Ask: has anyone ever queried two parts of this at the same time?
Most working models are used the way Śuka's body is used — sequentially, one entry at a time, each retrieved for a local purpose and released. Under that usage a model can carry flat contradictions for years without a single symptom, because the symptom requires simultaneity and nothing in ordinary use supplies it.
So the audit is not "is each entry right." Each entry is usually right, locally, for what it was written for. The audit is: pull two entries that touch the same slot and put them next to each other on purpose. Brahmanas at the mouth, brahmanas at the heart. The premium tier is for enterprises, the premium tier is for individual power users. Two definitions of an active account.
Then the harder half, which the Pṛthu passage demonstrates: when you find a collision, look at what each version was for. Usually neither is wrong. Usually each was built to reach a conclusion someone needed, and the collision is a record of two purposes, not two errors. Deciding which to keep means deciding which purpose you are still serving.
The strongest evidence that these are two distinct schemes rather than one confused one is that the overlaps are exactly at the four rank-bearing sites. Feet, thighs, arms, mouth. The maps do not collide randomly across fifty correspondences; they collide at precisely the places where a body encodes hierarchy, which is where each scheme needed to be.
Tension — the volume states its cosmic anatomy at least three times incompatibly and calls the result "the lord's body," singular. Neither text nor apparatus reconciles them or marks them as variants. Preserved unresolved.
Tension — the Pṛthu passage derives "brahmanas are his heart" from "brahmanas came from his mouth" with a therefore that does not hold. It is not an incidental slip; the conclusion reached is the one the speakers need. Preserved.
Open question. Does the practice described actually require a consistent body? If dhāraṇā proceeds one limb at a time with each released before the next, consistency may be doing no work at all — in which case the collisions are not a defect of the text but a property of what this kind of text is for.
This vault's material on Purāṇic cosmology tends to read these body-maps as evidence of composite authorship — different strata assigning the same anatomy differently, the seams left in. That reading is available here and this page does not dispute it.
But it explains the wrong thing. Composite authorship explains how incompatible assignments got into one corpus. It does not explain how they sit four lines apart in a single speech and pass without comment through a modern annotated translation whose translator flags plenty else. The Arithmetic That Does Not Fit documents the same apparatus reporting a failed numerical check and then routing around it — so the capacity to notice is present.
The better account is the one the passage itself suggests: nobody notices because nobody queries. A text used sequentially does not generate the comparison that would produce the objection, and that is as true of the translator working through it as of the practitioner sitting in it.
With Compartmentalized Morality in Honor Systems. That page works the bushidō contradiction — absolute honesty toward peers, systematic deception toward enemies — and names the mechanism as epistemic compartmentalisation: different domains running different rules, held as non-contradictory because the domains never touch.
The Bhāgavata's body is the same mechanism operating on description rather than on conduct. The loka-map and the varṇa-map are two domains with two sets of rules for the same four sites, and they are held as non-contradictory because nothing in the practice ever runs both.
What the pairing produces: compartmentalisation is normally analysed as a moral phenomenon — a way people avoid the cost of their own inconsistency, with the boundary doing protective work. This case has no moral content and no one to protect. It is a text and a set of correspondences, and the compartments still hold. Which suggests the boundary is not maintained by motivated reasoning at all but by usage pattern: any system consulted one entry at a time will compartmentalise by default, motive or no motive. The bushidō case may be the same structure with a motive laid over it, and the motive may be doing less of the work than it appears to.
With Cognitive Dissonance: Festinger. That page works the finding that holding two incompatible cognitions produces a real, aversive pressure, and that the pressure is what drives belief change.
This passage should be a dissonance engine. Three occupants of one mouth, the same god at the crest and the groin, a therefore that inverts its own premise. And it produces nothing — no discomfort, no revision, no note.
What the pairing produces: the dissonance account has a precondition usually left unstated because laboratory conditions supply it automatically — the two cognitions must be co-present. The experiment puts them in front of the subject together; that is what the experiment is. Śuka's enumeration guarantees they never are, by instructing the practitioner to release each limb before taking the next. So the interesting variable is not what people believe but what the retrieval discipline permits them to compare, and a practice can be designed — deliberately or not — to make dissonance unavailable. Neither field sees this alone: the psychology has no case where co-presence is structurally prevented, and the text has no vocabulary for what it is doing.
Sharpest implication. Consistency is a property demanded by lookup, not by use. A system consulted sequentially can carry flat contradictions indefinitely and function perfectly — which means the arrival of a reader who checks is not a neutral event. It converts a working practice into a broken description, and the text does not change at all.
Generative questions.