Rudra sounds his drum and summons the three worlds to escort him to Kailasa. Everyone comes — Vishnu, Brahma, the gods, the sages, the siddhas, the asuras, and "all the festivals," which is a lovely detail nobody follows up.
And then the chapter stops the narrative to do something it does nowhere else.
There were immensely fortunate leaders of the ganas, revered by all the worlds. I will tell you about their number. Listen attentively.1
What follows is roughly forty named gana-leaders, each with a headcount, in crores.
It is the only census in Volume 1. And it is the last substantial passage of the Srishti Khanda — the closing movement of the whole unit is a list of names with numbers attached.
A stretch of it, to show the texture:
Shankakarna went there, with one crore ganas; Kekaraksha went with ten crores and Vikrita with eight; Vishakha went with sixty-four crores and Pariyatraka with nine; Sarvantaka went with six and the handsome Dundubhi with eight; Jalanka, bull among ganas, went with twelve crores; Samada and the handsome Vikritanana went with seven each; Kapali went with five and the auspicious Sandaraka with six; Kanduka and Kundaka went with one crore each...1
It runs on. Mahakesha with a thousand crores. Kala, Kalaka and Mahakala with a hundred each. Agnika with a hundred crores. Nila, Devesha and Purnabhadra with ninety crores each. Nandishvara and Mahabala with a hundred crores each.
And then, having enumerated for two hundred words:
Surrounded by crores, thousands of crores, hundreds and twenties, Sarvesha went there.1
Hundreds and twenties. After thousands of crores.
The numbers do not behave.
A crore is ten million. So Mahakesha alone brings ten billion ganas. Nila, Devesha and Purnabhadra bring nine hundred million each. Vishakha, Kashthaguda, Sukesha, Vrishabha and nine others bring six hundred and forty million apiece.
Add the list up and you are past a hundred billion, escorting one god up a mountain.
And they are not undifferentiated. Every one of them is described:
Each of them possessed one thousand hands and wore matted hair and diadems. All of them were adorned with the crescent moon. They were blue in the throat and three-eyed... They resembled Brahma, Indra and Vishnu and possessed anima and other qualities. They were as radiant as crores of suns.1
Every leader looks like Shiva. Thousand-handed, blue-throated, three-eyed, crescent-moon. The retinue is a hall of duplicates of the god they are escorting.
Somewhere in the list — and it is worth noticing where — the counting stops being counting.
The first entries are legible. One crore. Ten crores. Eight. Twelve. You can hold the relative sizes; Kekaraksha has ten times Shankakarna.
By the time Mahakesha arrives with a thousand crores, the ratios have stopped landing. And when the summary says "crores, thousands of crores, hundreds and twenties" — mixing magnitudes four orders apart in one clause — the arithmetic has been abandoned inside the sentence that is performing it.
This is what the build has already found about the volume's soteriological numbers: they are magnitude vocabulary rather than quantities, and the test is whether a figure is ever used again.2 None of these is. Nobody totals the ganas, nobody compares two leaders, and no later chapter refers back.
But this list is doing something the price lists are not, and that is why it earns a page rather than a footnote to that finding.
Compare the two enumerations and the difference is instructive.
The lingam price lists in chapters 28 and 46 are instructions. Seven hundred to kill a man, five crore for emancipation. They are addressed to a reader who is supposed to act, and the numbers are the content.3
The gana census is addressed to nobody who will act. There is no instruction in it. Nothing is being priced, promised or required.
So what is the passage for?
Look at what survives the numbers going illegible: the names. Shankakarna, Kekaraksha, Vikrita, Vishakha, Pariyatraka, Sarvantaka, Dundubhi, Jalanka, Samada, Vikritanana, Kapali, Sandaraka, Kanduka, Kundaka, Vishtambha, Chandratapana, Mahakesha, Kundi, Vaha, Parvataka, Kala, Kalaka, Mahakala, Agnika, Abhimukha, Adityamurdha, Dhanavaha, Sannaha, Kumuda, Amogha, Kokila, Sunmantraka, Kakapada, Santanaka, Mahabala, Madhupinga, Pingala, Nila, Devesha, Purnabhadra, Chaturvakra, Kashthaguda, Sukesha, Vrishabha, Chaitra, Nakulisha, Svayamprabhu, Lokantaka, Diptatma, Daityantaka, Bhringi, Riti, Devadevapriya, Ashani, Bhanuka, Sanatana, Nandishvara.
Fifty-seven of them.
The numbers are the metre and the names are the payload. The crore-counts vary to keep the list from becoming a drone, and what a listener actually receives across two hundred words is that Shiva's household is enormous and populated by specific individuals with names you have now heard.
That is a different achievement from a price list, and it is the reason this passage sits where it does — at the end of the unit, as the god takes possession of his mountain.
You are watching credits roll and you do not leave.
Not because you are reading them — nobody reads them. But something happens over ninety seconds of scrolling names that a title card saying made by 1,400 people does not do.
The scale arrives as duration. You cannot feel fourteen hundred as a number; you can feel a minute and a half of names not stopping.
That is what this passage is doing, and it explains a feature that looks like a defect. The numbers are inconsistent — crores, thousands of crores, hundreds and twenties — because precision is not the mechanism. Length is.
A reciter delivering this out loud produces two hundred words of proper nouns with varying figures attached, and the audience's experience is not an arithmetic total. It is: this goes on, and on, and every one of them has a name, and the god they belong to is about to take his seat.
Which is why the passage is placed at the consecration and not in a chapter about cosmology. It is a procession rendered in words — and if you have ever stood on a route watching one, you know that the number of people who passed is not what you came away with.
Two details confirm that reading, and they are both things a purely numerical passage would not bother with.
The leaders are individuated. Not by size but by character — the handsome Dundubhi, Jalanka, bull among ganas, the auspicious Sandaraka, the divine Bhringi, the handsome Vikritanana, the immensely strong Chaturvakra. Small epithets, distributed unevenly, doing nothing arithmetically.
And Vishvakarma is put to work housing them.
The lord instructed Vishvakarma to construct mountains as residences, as each deserved, for the many devotees, his own as well as those of others.1
As each deserved. The same phrase — the same entitlement principle — that governed the lingam distribution eight chapters earlier, where nineteen gods received materials by rank and nobody objected.4
So the census is not floating. It has a consequence: these are beings who get accommodation, assigned by standing. Which is what you do with a household, and not what you do with a number.
The general shape, and it is what the passage contributes.
There is a mode of religious writing in which counting is the devotional act — the thousand names, the hundred and eight epithets, the litany where the point is not the information but the sustained attention that reciting it requires.
This volume is full of counting that is not that: quotas, prices, thresholds, tariffs. Crores of japa with a stated return. A hundred thousand flowers weighed on a scale.3 Counting as accountancy.
Chapter 52 is the other kind, and it is the only clear instance in Volume 1.
Nothing is being bought. No fruit is attached, no requirement is stated, and the reciter and the listener get nothing for it except the time it takes.
Which is worth marking because the vault has been reading this volume's numbers as one phenomenon and they are two. A number that prices something and a number that lengthens something look identical on the page — both are large, both are specific, both are unverifiable. The test is not whether the figure is used again but whether anything is being asked of the reader.
Here nothing is. The passage is the closing of a unit, and its work is to make a mountain feel occupied.
Strongest support is the summary clause — "crores, thousands of crores, hundreds and twenties" — which mixes magnitudes four orders apart inside a sentence performing an enumeration, and the fifty-seven proper nouns that survive when the figures stop landing.
Tension one. The volume's other large numbers are tariffs addressed to an actor. This one asks nothing of anyone. Two kinds of counting, indistinguishable on the page, and the vault has been treating them as one.
Tension two. Every gana-leader is described as thousand-handed, blue-throated, three-eyed and crescent-moon-adorned — each one a duplicate of the god they escort. A retinue that is a hall of copies is a strange thing for a text arguing for a unique supreme deity, and nothing comments on it.
Tension three. Vishvakarma houses them "as each deserved," the same entitlement language as the lingam distribution.4 So the census has a rank-order attached which the numbers do not straightforwardly encode — Mahakesha has the largest following and is not the senior figure; Nandishvara is.
Tension four. Fifty-seven names are given and the volume never uses most of them again in Volume 1. Whether they are inherited from a fuller tradition or generated for the passage is not decidable here.
Open question, and it is cheap to check. Do any of these fifty-seven recur in Volumes 2 and 3 — particularly at Daksha's sacrifice, where Virabhadra leads the ganas against it? If a substantial subset reappears with consistent characters, the list is a real inherited roster. If none does, it is a procession written for one occasion.
The Prices That Do Not Match established that most of the volume's large numbers are rhetorical rather than operational, and proposed the test: is the figure ever used again?
This page complicates that usefully. The gana counts also fail that test, and they are not doing what the price lists do. So the test separates operational from non-operational numbers and does not distinguish the two kinds of non-operational number.
A sharper test: is anything being asked of the reader? Prices ask. Litanies do not.
The Flowers Weighed on a Scale is the extreme of the accountancy mode — devotion converted into prasthas and palas so it can be verified on a balance. This chapter is the extreme of the other mode. They are eight chapters apart and there is no vocabulary in the volume that distinguishes them.
The Retinue and the Order of Worship holds the parivara-devata layout — Ishana, Nandi, Chanda, Mahakala, Bhringi and the rest, positioned spatially around the lingam, with the hierarchy encoded as a walk from the north-east.
Bhringi and Mahakala appear in both. So the census and the worship layout overlap, which means at least part of this roster is liturgically real rather than decorative — these are beings someone actually worshipped in a fixed order.
The Equivalent Album Sales Metric (business). The problem there is that a single headline number has to absorb incommensurable things — streams, downloads, physical sales — and the conversion ratios are chosen rather than discovered, so the resulting figure is precise, comparable, universally quoted, and not really a measurement of anything.
The gana census is that with the pretence dropped.
Nobody is converting anything. Crores, thousands of crores, hundreds and twenties sit in one clause without a common denominator, and the passage does not care, because the figure was never going to be totalled.
The insight neither gives alone: a number survives in a culture when it is quotable, not when it is meaningful, and the two come apart in opposite directions here. Equivalent album sales is meaningless and quotable, so it thrives; the gana counts are meaningless and un-quotable — nobody could repeat them — so they function as texture instead. Which suggests the useful question about any large figure in circulation is not is it accurate but what does it survive on: precision, repeatability, or duration. Only the first is about the world.
Misaligned Optimisation Metrics (business). The failure documented there is a measure adopted as a proxy for a goal and then optimised directly, so that the number improves while the thing it was standing in for degrades.
The census is a case where no such risk exists, and the reason is worth extracting.
You cannot game a number nobody acts on. There is no incentive attached to Mahakesha's thousand crores, no threshold, no comparison anybody is scored against. The figure is inert, and inertness is what makes it safe.
What the pair produces: the danger in a measurement is not imprecision, it is consequence — a wildly imprecise number with nothing riding on it does no harm at all, while a well-constructed one with a bonus attached will be gamed within a year. Which inverts the usual instinct to improve a metric's accuracy first. The prior question is what happens to someone when the number moves, and if the answer is nothing, precision is a luxury rather than a requirement. This volume has both cases side by side: a census nobody can game, and a price list for killing a man.
Sharpest implication. The Srishti Khaṇḍa closes with the volume's only census — roughly fifty-seven named gaṇa-leaders with headcounts in crores, escorting Rudra to Kailāsa. The numbers stop working almost immediately: Mahākeśa brings ten billion, and the summary clause reads "crores, thousands of crores, hundreds and twenties" — four orders of magnitude in one phrase, inside a sentence that is performing an enumeration. The arithmetic is abandoned mid-count, and it does not matter, because precision was never the mechanism — length was. What survives when the figures go illegible is fifty-seven proper nouns, each leader given a small epithet, every one of them thousand-handed, blue-throated and three-eyed like the god they escort. And Viśvakarmā is then set to build them houses "as each deserved." These are not a number, they are a household — which is why the passage sits at the consecration rather than in a cosmology chapter. It also splits a finding this build had treated as single: a number that prices something and a number that lengthens something look identical on the page, and the test is not whether the figure is used again but whether anything is being asked of the reader.
Generative questions.