The worship catalogue in chapter 2(3) is usually read the way lists are read — at speed, for the gist. Read at speed it says there is a god for everything. Itemised, it says several things that are harder to summarise.
Here is the whole of it, in the order given: radiance to Brahmā. The senses to Indra. Offspring to the Prajāpatis. Prosperity to the goddess Māyā, whom Debroy interprets as Durgā.2 Energy to Vibhāvasu, the fire god.3 Wealth to the Vasus. Valour to Rudra. Food to Aditi, heaven to Aditi's sons — the twelve Ādityas.4 A kingdom to the Viśvadevas. Control over people to the Sādhyas. Long life to the Aśvins. Health to Ilā, interpreted as the earth.5 A steady position to earth and sky, the mothers of the world. Beauty to the gandharvas. Women to the apsarā Urvaśī. Suzerainty to Parameṣṭhī. Fame to Yajña. Treasures to Prācetas, interpreted as Varuṇa.6 Learning to Girīśa. A happy marriage to Umā. Dharma to Uttamaśloka. Offspring and their protection to the ancestors. Protection to virtuous people. Spirit to the Maruts. A kingdom to the divinity Manu. Control of spells to Nirṛti. Pleasure to Soma. And freedom from desire to the supreme Purusha.1
Twenty-nine entries. Now the things that only appear on the second pass.
Entry ten: one who desires a kingdom should worship the Viśvadevas.
Entry twenty-six: one who desires a kingdom must worship the divinity Manu.
Same desire, two suppliers, sixteen entries apart, in a single continuous list delivered in a single breath. No preference is expressed. No distinction is drawn between kinds of kingdom.
If the catalogue were a directory in any working sense, this is the entry a user would hit and stop at. It is the one place where the list fails to do the only job a list of this shape can do — tell you where to go.
This is the larger finding, and it is Debroy's own identifications that produce it.
Fame → Yajña. Debroy: "Interpreted as Vishnu."7
Dharma → Uttamaśloka. Debroy: "The one who is spoken about in excellent shlokas, Vishnu."8
Freedom from desire → the supreme Purusha.
So the deity this entire volume exists to establish as the source of everything appears three times in a catalogue of alternatives — twice as a vendor of particular goods, once as the exit from vending altogether. Between the fame entry and the dharma entry sits treasures to Varuṇa, and after the dharma entry come the ancestors, virtuous people, the Maruts, Manu, a demoness, and the moon.
He is on the menu. Under two names. Interleaved with everyone else, selling fame and dharma the way Urvaśī sells women.
Register: the identifications are Debroy's, made locally at each site, and he does not connect them. What the Sanskrit intends by yajña and uttamaśloka at these two points is a question this page does not settle. What can be shown is that the apparatus a reader is holding names Viṣṇu three times in one list and never remarks on it.
Once you start counting, the list turns out to double back on itself in at least four different ways, and they are worth separating because they mean different things.
One desire, two suppliers. The kingdom, to the Viśvadevas and to Manu. This is the only true collision — a user of the list cannot proceed.
One supplier, two names, two goods. Rudra for valour and Girīśa for learning — Debroy glosses Girīśa as Śiva.9 The same god sells two unrelated products under two names, and nothing marks them as the same vendor.
One supplier, two names, one of them the frame. Brahmā for radiance and Parameṣṭhī for suzerainty. Parameṣṭhī is Brahmā, and Brahmā's abode is the destination the previous chapter spent its ascent reaching.
One supplier appearing as both an item and the exit. Viṣṇu, three times, as described above.
Only the first is a defect in the list. The other three are something more interesting: the catalogue is not organised by who at all. It is organised by what you want, and the suppliers are whatever name attaches to that want in whichever ritual idiom it comes from. Nobody built this by starting with a pantheon and distributing goods. It was built by starting with goods.
Which is why the duplications pass unnoticed. A list indexed by desire has no reason to check whether two entries point at the same being, any more than a shop's aisles check whether two products share a manufacturer.
The sequence is not ranked. Radiance opens and freedom from desire closes, and in between there is no visible ascent — women at sixteen, suzerainty at seventeen, fame at eighteen, treasures at nineteen.
Nor is it grouped. Health sits next to a steady position, which sits next to beauty. The two Aditi entries are adjacent, which is the only place adjacency tracks anything.
What the order does have is a terminus. The last entry is different in kind from all twenty-eight before it, and its position is doing the work its content could not do alone. Arriving at freedom from desire after twenty-eight desires is a different experience from encountering it first.
So: unranked, ungrouped, and ending somewhere. That is the shape of a list assembled from ritual practice rather than derived from doctrine — with one editorial decision made, about what goes last.
Three items in the middle deserve naming, because summaries of this passage never include them.
Protection → virtuous people. Not a deity. A class of persons, sitting in the list between the ancestors and the Maruts, worshipped for the same kind of return.
Control of spells → Nirṛti, whom Debroy notes may be Nairṛti, a demoness.10 A malevolent figure, listed without warning among the Ādityas and the Aśvins.
Women → the apsarā Urvaśī. A named individual rather than a class, and the only entry where the desired object is a category of person.
These matter because the smooth reading — worship any god for any boon — is not quite what is written. The suppliers include humans, a demoness, and a specific nymph. Whatever principle organises this, it is not divinity.
Śuka introduces the catalogue with an explicit statement of what it answers:
This is what you had asked me about, about what learned men should do, especially men who are about to die.1
Then he lists suppliers of women, beauty, wealth, treasures, a kingdom, suzerainty, fame, a happy marriage, offspring, long life and health — to a man with days left.
Only the final entry is addressed to Parīkṣit's actual situation. The other twenty-eight are answers to a question he did not ask and could not use, delivered under a heading that says they are.
That gap is either the strongest evidence that the catalogue is inherited material dropped in whole, or the strongest evidence that its function is rhetorical rather than informational — twenty-eight things you can no longer want, recited to a man who can no longer want them. The passage supports both and distinguishes neither.
You have a list you know the gist of — a set of company values, a competitor's feature grid, a policy's exemptions, a taxonomy your field uses. You have read it many times and never once counted it.
Do the mechanical passes, in this order, because each one catches what the last cannot.
Count. Just the number. You will usually be wrong about it, and being wrong about the number is a reliable sign you have been reading the gist.
Check the left column for repeats. Same input, two outputs — the kingdom problem. This is the failure that stops a user cold, and it is invisible at reading speed because the two entries are never adjacent.
Then check the right column for repeats, including aliases. This is the pass almost nobody runs. Rudra and Girīśa are the same god; Brahmā and Parameṣṭhī are the same god; Yajña and Uttamaśloka are the same god. Nothing in the list says so, and it takes an annotator working entry by entry to surface it — and even then only if someone reads the annotations together, which is precisely what did not happen here.
Finally, ask what the list is indexed by. Not what it contains — what its left column is. That is what determines which duplications the list is structurally incapable of noticing, and it is nearly always the thing the list's users assume they know.
The strongest evidence that the catalogue is indexed by desire rather than by deity is that it tolerates three separate cases of one supplier under two names while failing on the single case of one desire under two suppliers. A deity-indexed list would have the opposite error profile.
Tension — Viṣṇu appears in the catalogue of alternatives twice by the translator's own identifications, and as the exit from it a third time. No note connects the three. Preserved unresolved.
Open question. A kingdom is assigned to the Viśvadevas and to Manu. Is there a distinction in the Sanskrit — two different words for kingdom, or two different senses of rule — that the translation has flattened, or is the duplication in the source?
This vault's material on Purāṇic ritual lists tends to treat them as doctrinal statements in disguise — the ordering read as a ranking, inclusion read as endorsement, omission read as polemic.
The itemisation does not support that here. There is no ranking to read, the inclusions run from Ādityas to a demoness to ordinary virtuous people, and the one clear editorial act is the placement of the last entry. Whatever this list is, it is not a compressed doctrine.
Two Incompatible Maps of the Same Body documents the same volume assigning one anatomical site three times over without noticing. This is the same behaviour in list form, and it has the same cause: entries consulted one at a time never meet, so nothing in ordinary use produces the comparison that would raise an objection.
With Human Needs Map: Six Social Drives. That page works a compressed taxonomy of what people want — a small number of drives claimed to underlie the enormous variety of things people actually pursue, with the compression itself as the analytical payoff.
Set the catalogue beside it. Twenty-nine wants, uncompressed, unranked, ungrouped: radiance, senses, offspring, prosperity, energy, wealth, valour, food, heaven, a kingdom, control over people, long life, health, position, beauty, women, suzerainty, fame, treasures, learning, marriage, dharma, protection, spirit, spells, pleasure.
What the pairing produces: the two disagree about whether desire has structure, and the disagreement is methodological. A six-drive map is built from outside the wants, looking for what they share. The catalogue is built from the inside, by accretion, from what people actually walked into shrines asking for — which is why it keeps offspring and offspring and their protection as separate entries, a distinction no taxonomy would preserve and no petitioner would collapse. The compressed map is better for prediction and worse as evidence; the uncompressed one is nearly useless operationally and is a record of what was genuinely asked. Neither yields the other, because compression is lossy in the direction that matters: six drives follow from twenty-nine wants, and the twenty-nine do not follow back.
Sharpest implication. A list's index determines which of its errors it can see. Indexed by desire, this catalogue cannot notice that it sells the same god under four names, and can notice — but does not — that it sells kingdoms twice. The blindness is structural, not careless, and it transfers intact to the apparatus: an annotator working entry by entry inherits the index and therefore inherits the blind spot.
Generative questions.