Chapter 28 opens with the sages asking a practical question — how many lingams for what — and the answer is a table.
Not a metaphor. A table.
A man in search of learning must cheerfully worship one thousand Parthiva lingams... A man in search of wealth must fashion five hundred. A person seeking sons must use one thousand and five hundred, while one desiring garments must use five hundred. A person desiring emancipation must use one crore...1
Ten million lingams for liberation. Five hundred for clothes.
There is no currency in this. Nothing is being bought from anyone. But the structure is a price list — a want in the left column, a quantity in the right — and once you notice that, the chapter is difficult to read as anything else.
The full list is worth setting out, because the ordering of what people wanted is itself a document.
| Want | Lingams |
|---|---|
| Learning | 1,000 |
| Wealth | 500 |
| Sons | 1,500 |
| Garments | 500 |
| Emancipation | 1 crore |
| Land | 1,000 |
| Compassion | 3,000 |
| A tirtha | 2,000 |
| Well-wishers | 3,000 |
| Control over someone | 800 |
| Killing someone | 700 |
| Confounding someone | 800 |
| Expulsion | 1,000 |
| Stupefying someone | 1,000 |
| Kindling hatred | 500 |
| Freedom from bondage | 1,500 |
| Safety from a king | 500 |
| Safety from thieves | 200 |
| Safety from dakinis | 500 |
| Escape from poverty | 5,000 |
| Everything desired | 10,000 |
Twenty-one entries. Debroy glosses two of them — expulsion means warding off harmful omens, objects or people; a dakini is a flesh-eating demoness.2
Some of it is touching. Garments at the bottom, sons near the top, poverty costing ten times what a new coat does. And compassion at three thousand — more than land, more than learning, more than a place of pilgrimage. Somebody thought compassion was hard to come by.
Now the four rows that stop the reader.
A person desiring to kill someone must use seven hundred, while a person desiring to confound someone must use eight hundred... A person seeking to stupefy must use one thousand, while a person desiring to kindle hatred must use five hundred.3
Murder. Confusion. Stupefaction. Sowing hatred.
This is abhichara — the harmful ritual tradition — and it is sitting in a devotional worship manual, in a numbered list, between "well-wishers" and "expulsion," with no partition of any kind.
Take the measure of what is not there.
No condemnation. Not a clause. Not a "but the wise avoid this."
No hedge. No "some say," no "in extremity," no attribution to a rival school.
No separate procedure. The rite is identical. Same clay, same Parthiva lingams, same chapter, same speaker. Only the count changes.
No warning about consequences. The volume is entirely willing to describe hells — chapter 6 has a betel-chewer eating excrement and a critic reborn as a dog, chapter 26 impales a lax brahmana on a trident.4 Here, nothing. A man may worship seven hundred lingams to kill his enemy and the text moves straight to the next row.
And the pricing is the part that lands hardest. Killing is seven hundred. Learning is a thousand. Land is a thousand. Compassion is three thousand.
Murder is cheaper than knowledge, cheaper than a field, and less than a quarter the price of compassion.
Nobody in the chapter, and nobody in Debroy's notes, remarks on it.
The one structural observation available is that all four harmful items are clustered together, which suggests they came in as a block from a source that had them as a block — probably a tantric ritual manual of the shatkarma type, where pacification, subjugation, immobilization, enmity, eradication and killing are a standard six-fold set. If so, the Purana absorbed a technical list and did not sort it. That is an explanation of how the passage got there. It is not a mitigation, and the text supplies none.
Then something odd happens, and it is easy to miss because it comes with an apology.
Having finished the table, Suta says:
It progressively gets better and better as more than the preceding number is used. However, there are differences in views among sages and I will tell you about that.5
And gives a second set of numbers:
The second list is between seven and thirty-six times more expensive than the first for the same outcomes.
And the text says so openly. Differences in views among sages. It is not resolving them. It gives both, side by side, and leaves the reader holding two incompatible schedules.
Which is precisely what CLAUDE.md's first constraint is about — the contradiction is preserved, not smoothed, and the preserving is the text's own doing rather than a modern editor's.7
It also tells you what kind of knowledge this is. A number that varies by a factor of thirty-six between authorities is not a measurement. It is a contested tradition, transmitted with its disagreements attached, in a genre that could easily have picked one and suppressed the other.
You want a son and the number is one thousand five hundred.
Except that a second reading of the same chapter says fifty-five thousand, and the man who told you the first number and the man who told you the second are both alive and both learned.
So you go with one thousand five hundred, because fifty-five thousand is not a thing a person does.
You make them from the riverbank. Fifty a morning is achievable if you start before light and do not stop to eat, so you are looking at a month, and then the avahana and the pratishthana and the puja are separate for each one — the chapter is explicit that "the first avahana, pratishthana and puja are separate."8
By the second week your hands have changed. The clay has taken the skin off your knuckles and put a callus where the base of your thumb pushes.
By the third week you are not thinking about a son at all. You are thinking about the count, and whether that one was properly formed, and whether the one that broke counts.
That is the thing about a number this large. It stops being a petition and becomes a job. You do not spend a month asking for something. You spend a month doing something, and at the end of it you are a person who spent a month.
Whether the son arrives is not, by then, the only thing that has happened.
Worth being honest about how much of this is derivable.
Some of the figures have obvious internal logic. One crore for emancipation matches the ceiling elsewhere in the volume — one crore japas for a single-syllable mantra, one crore oblations at the outer limit of a peace rite.9 A crore is the volume's word for the most.
Some are relational and make sense. Freedom from thieves costs two hundred, the least on the list, because thieves are the smallest threat named. Poverty at five thousand is ten times a new coat, which is about right.
But most of them are not derivable from anything. Why is confounding eight hundred and killing seven hundred? Why is control also eight hundred, the same as confounding? Why is compassion three thousand — the same as well-wishers, and double a son?
There is no principle. The chapter offers none and Debroy annotates none.
Which leaves the honest reading: the numbers carry precision without carrying information. Seven hundred is not seven hundred because anyone calculated it. It is seven hundred because a specific figure makes a claim feel like knowledge, and a range would not.
That is not a cynical reading, and it is not unique to this text. It is what happens to any body of practical lore that gets written down — the figures harden, and the second list in this very chapter is the proof, because it shows them hardening differently in a different lineage.
Having priced everything, the chapter weighs the practice against the whole rest of the religion.
If one places all donations, many kinds of vows, tirthas, niyamas and yajnas together on one side and the worship of lingam alone on the other, the two will be equal. In kali yuga, it is evident in the world that the worship of lingam is the best. There is nothing else.10
Everything, versus this. Even.
And then not even: there is nothing else.
The chapter opens by saying the same thing from the other end — "Without fashioning a Parthiva lingam, if a person worships any other deva, that worship is futile. Self-control and devotions are in vain."11
So the two claims bracket the numbers. Before the table: everything else is void. After the table: this alone equals everything else combined.
Which reframes what the table is. It is not a menu of options within a religion. It is the pricing schedule for the only remaining transaction, in a chapter that has just closed every other counter.
And it lands on a practice introduced sixteen chapters earlier as what you do if you cannot manage the real thing.12
Back to the four rows, because the page has to end there.
Why does a devotional manual price murder without comment?
The available answers are all partial.
The absorption reading. The four harmful items came in as a technical block and nobody sorted them. Plausible, and it explains the clustering. It does not explain why an editor who cut five entire Samhitas from this text left these four lines in.13
The completeness reading. A manual claiming to price every human want cannot omit the wants people actually have. Wanting an enemy dead is not rare. A list that skipped it would be a list about a different species.
The mechanism reading, and this is the one the chapter's own logic supports. The rite is repeatedly described as working regardless of the worshipper's state — "even if the worship is done unknowingly and even if it is done because one is addicted to material objects."14 Elsewhere in the volume ash purifies whether worn voluntarily or involuntarily, and a man is liberated for overhearing a recitation while dying of fever.15 If the machinery does not consult intention at all, then it cannot be expected to refuse a bad one. The list is amoral because the mechanism is, and the mechanism was made amoral in order to be maximally merciful.
That is the finding, and it is uncomfortable in the right way. The same doctrine that saves a murderer for being within earshot of a recitation is the doctrine that will kill a man for seven hundred lingams. Grace that does not check credentials is grace that does not check purposes. The volume built one and got both, and never once put them on the same page.
Strongest support is the table's own uninterrupted format — twenty-one wants in one sequence, with murder in position eleven and no typographic, rhetorical or moral separation anywhere.
Tension one, unresolved by the text and unresolvable here. Abhichara sits inside a devotional manual, uncommented, priced below learning and land.
Tension two, and the text raises it itself. Two incompatible schedules for the same outcomes, differing by up to thirty-six times, presented as differences in views among sages with no adjudication.
Tension three. "There is nothing else" in kali yuga, against chapter 14's Shiva personally granting Vishnu equal worship and his own festivals.
Tension four. The practice that now outweighs all donations, vows, tirthas, restraints and sacrifices combined was introduced as the fallback for people incapable of the three means.
Open question. Do the four harmful entries appear in other Purana recensions, or are they specific to this transmission? Debroy compares against the J.L. Shastri edition throughout the volume and flags divergences where he finds them — he flags nothing here, which suggests the passage is stable across at least two witnesses. Volume 1 cannot take it further.
The vault's Śaiva corpus has nothing like this, and the absence is the finding.
Puraścaraṇa: The Lakh-System is the vault's existing counting page — 100,000 repetitions per syllable, consistency over intensity. Its outcomes are all interior: siddhi, steadiness, the mantra becoming live.
Chapter 28's outcomes are almost entirely exterior, and four of them are aimed at other people.
God Is Not a Fish Market makes the practitioner's objection — you cannot compel grace with quantities. This chapter is the strongest possible case for that objection, and it is drawn from the tradition's own primary text rather than from a critic.
And The Nine Curse Types holds the vault's existing material on ritual harm, where cursing is at least classified, motivated and situated. Chapter 28 does none of that. It gives a number.
The comparison is worth stating plainly: the vault's practitioner-tier Śaivism has no harm-magic at all, and the primary text has it in a table. Any account of this tradition built only from modern teachers is missing a page that its own scripture prints without embarrassment.
Disguising Numbers and Direction (behavioral-mechanics). The influence corpus documents how specific figures function persuasively — that a precise number reads as evidence of a process, and that precision and accuracy come apart in exactly the situations where nobody can check.
Chapter 28 is that mechanism operating without any intent to deceive. Seven hundred, eight hundred, one thousand five hundred, fifty-five thousand. Nothing generates them. Nothing verifies them. Two lineages disagree by a factor of thirty-six and both are transmitted.
What the pair produces: precision is a property of transmission, not of knowledge — a body of practical lore, written down and copied, will harden into exact figures whether or not there was ever anything to be exact about, because a range does not survive recitation and a number does. The behavioral pages read false precision as a technique. This shows it arising with no technician, as a side-effect of a genre that is spoken aloud to a room. Which is a considerably more disturbing finding, because it means precision-as-authority does not require anyone to have decided to deploy it.
Mental Accounting (psychology). The behavioral-economics material shows people keeping separate mental ledgers for different categories of value, and applying different rules to each — money for rent reasoned about differently from money for a treat, with the partitions doing real moral work.
Chapter 28 has no partitions. Emancipation and murder are in one ledger, denominated in one unit, with murder priced lower.
The insight neither domain produces alone: a moral partition is not implied by having a moral doctrine — it has to be built, and a system can hold sophisticated ethics elsewhere while running a completely flat accounting here. The volume has a full ethics: repentance as supreme atonement, sins that ash cannot erase, a hell for a man who chews betel while listening.16 None of it is in this room. So the partition failure is local, not a sign of an amoral text — which suggests moral partitioning is a property of individual passages and their sources rather than of whole traditions, and that reading a tradition's ethics off its most articulate page will systematically mislead you.
Sharpest implication. Chapter 28 prices twenty-one human wants in Parthiva lingams — a thousand for learning, three thousand for compassion, one crore for emancipation — and slots four items of harm magic into the middle of the list with no separation, no condemnation, no hedge and no altered procedure. Killing a man costs seven hundred: less than learning, less than a field, less than a quarter of compassion. Then the chapter gives a second, incompatible schedule, up to thirty-six times more expensive, introduced as "differences in views among sages," and adjudicates nothing. The amorality is not an oversight but the cost of the volume's own mercy doctrine: a mechanism built not to consult the worshipper's intention — so that a murderer overhearing a recitation while feverish can be saved — is a mechanism that cannot refuse a murderer's intention either.
Generative questions.