Vīrabhadra sets out, and the volume stops the story to do arithmetic.
O Nārada! Hear about the numbers of the gaṇas. There were large numbers. The foremost among them were extremely strong and full of fortitude. Gaṇeśvara Śaṅkukarṇa advanced with ten crores. Kekarākṣa had ten crores and Vikṛta had eight. Viśākha had sixty-four crores and Pariyātrika had nine…1
It goes on for a full paragraph. Sarvāṅkaka six crores, Jvālakeśa twelve, Samadjjimān seven, Dundubha eight, Kapālīśa five, Sandāraka six.
Then it stops being countable at all:
Koṭikuṇḍa had crores and crores. O son! Sannāda and Pippala each had one thousand crores… Mahāveśa, leader of gaṇas, had one thousand crores. Kāla, Kālaka and Mahākāla advanced towards Dakṣa's sacrifice, each with one hundred crores.1
A crore is ten million. A thousand crores is ten billion. Two named leaders bring ten billion each, and the paragraph does not treat that as a climax — it is a middle item in a list.
The summary at the end abandons the pretence of a total:
Following Śiva's command, the valiant Vīrabhadra, lord of all those who are brave, advanced, surrounded by crores and thousands of crores, hundreds and twenty times that number.1
Two details in the march are worth extracting because they are the concrete ones in a paragraph of abstraction.
As the valiant one rapidly marched, thousands of crores of bhūtas and three crores of canine gaṇas born from the body hair proceeded with him.1
Thirty million dog-gaṇas, generated from body hair.
The volume has an established habit of producing beings from the discarded parts of a body — Vīrabhadra himself comes from a thrown strand of matted hair, and Mahākālī from the other half of it.2 Here the same mechanism runs at scale and downward: not a champion from a strand, but an army from the hair generally.
And the instruments:
There was the loud sound of drums and many kinds of sounds from conch shells. There were many kinds of horns. All these emerged from Hara's mouth and matted hair.1
The music is emitted by Śiva's body too. The march is not equipped; it is exuded — troops, dogs and orchestra alike, from hair and mouth.
Before the counting starts, the chapter says what the army looks like, and the answer is: him.
These were companions in the form of time and destiny and in appearance, all of them resembled Rudra. The great-souled Vīrabhadra wore attire and ornaments that were like those of Hara, and he was with these gaṇas.1
Not an army serving Śiva. An army of copies of Śiva, led by a being wearing his clothes.
That is consistent with everything else about this force — the dogs from his body hair, the drums from his mouth — and it completes the picture. The march is Śiva replicated, not Śiva accompanied. Which is also, quietly, the answer to Dakṣa's whole argument: the man who said the ceremony was complete without Rudra is about to be visited by an uncountable number of him.
The transport is on the same scale and is worth recording because it is the one place the chapter gives a spatial measure rather than a headcount:
There were so many chariots that two thousand nalvas of ground were covered. Ten thousand lions made efforts to pull the vehicles. Like that, there were many powerful lions, tigers, makaras, fish and thousands of elephants to act as those who protected the flanks.1
A nalva is 400 hastas.11 Two thousand of them is a footprint, not a count — the one figure in the chapter that measures something in the world.
And there is a discrepancy here that Debroy does not flag. One chapter earlier, at his creation, Vīrabhadra "possessed two thousand arms." Here, on the march, "He possessed one thousand firm arms, resembling the king of serpents."12
Same being, adjacent chapters, halved. [n=1 OBSERVED] — recorded as found. It is the second unflagged numerical inconsistency this build has found in the Satī Khaṇḍa, after the twenty-thousand/ten-thousand gaṇa death count.13 Two is not a pattern and neither is being written up as one; what they establish jointly is only that the translator's flagged arithmetic failures are not the complete set, which matters for any claim resting on his catches.
This section exists to correct something, and the correction matters.
An earlier page in this build noted that Mahākālī is created in the same gesture as Vīrabhadra, from the same strand, and then receives no instruction — Śiva commissions him at length and says nothing to her.2 That is accurate about chapter 84.
It is not the whole picture, and chapter 85 supplies the rest:
Along with the large number of bhūtas, Mahākālī advanced to destroy Dakṣa. The nine Durgās were with her — Kālī, Kātyāyanī, Īśānī, Cāmuṇḍā, Muṇḍamardinī, Bhadrakālī, Bhadrā, Tvaritā and Vaiṣṇavī.1
She marches under her own heading, with a named retinue of nine goddesses, and her own class of followers:
To destroy Dakṣa's sacrifice, ḍākinīs, śākinīs, bhūtas, pramathas, guhyakas, kūṣmāṇḍas, parpaṭas, caṭakas, brahma-rākṣasas, bhairavas and kṣetrapālas emerged swiftly. These brave ones were intent on obeying Śiva's command. There were sixty-four categories of yoginīs.1
So she is uninstructed and not unaccompanied. She gets no commission and mobilises anyway, with nine named Durgās, eleven classes of being and sixty-four categories of yoginī — and the text says they are "intent on obeying Śiva's command", a command he never gave her.
The finding survives in a sharper form: the volume gives her a force and never gives her an order.
The yoginī number invites a check, and Debroy pre-empts it with an unusually candid note:
The number of yoginīs varies, but is usually given as sixty-four, or groups of sixty-four. The names also vary. Often, yoginīs are linked with mātṛkās, though the names/numbers of mātṛkās also vary. Each mātṛkā is associated with seven other yoginīs. With eight mātṛkās, there are 8 × 8 yoginīs. But this taxonomy is not consistent.3
This taxonomy is not consistent.
That is a translator declining to reconcile, and it is the second time in this volume's apparatus he has done so on a numbered scheme — the other being maṇḍala, which he leaves untranslated "since the word can have different meanings."4
It matters here because the chapter is making a claim out of counting. If the sixty-four is a conventional figure with variable contents rather than an enumeration, then the yoginī number is doing the same work as the crores: signalling completeness rather than reporting a quantity.
And that is consistent with what the build has already found about this volume's use of numbers — it counts when something is being established, not when something is being measured.5
There is a defect inside the census and Debroy catches it.
Partway through the list: "Kāṣṭhagūḍha, Sukeśī, Vṛṣabha and Sumantraka, lord of gaṇas, each advanced with sixty crores." A few lines later: "Kāṣṭhaguḍa, Sukeśa and Vṛṣabha had sixty-four crores each."1
Debroy's note on the second: "This sentence is unnecessary repetition."6
Two things are wrong at one site. The same three leaders appear twice — a repeated block. And the figure changes, sixty to sixty-four, for the same beings.
This is the only place in Volume 1 where the translator hands the build a duplication rather than an address error, and it is worth stating precisely what it does and does not support.
What it supports: a repeated block with a numerical discrepancy exists here, flagged, in a passage whose whole function is enumeration.
What it does not support: any connection to the vocative lens. There is no frame violation anywhere near this passage. The build once carried a "compound signature" pairing numerical failure with frame violation, stated in the register of rigour on n=2, which collapsed to 2/5 as more instances arrived.7 This site has two of the three signals and not the third, which is what independent damage looks like rather than a syndrome.
[n=1 OBSERVED] for a translator-flagged duplication in this volume. One instance.
Somebody presents a capability as a number, and the number is enormous.
Three checks, in order, and this passage fails all three in an instructive way.
Is the unit stable? The census runs in crores, then "crores and crores", then "one thousand crores", then "crores and thousands of crores, hundreds and twenty times that number." When the unit dissolves partway through, the figure stopped being a measurement before it stopped being quoted — and the exact point where precision is abandoned is usually the point where the real information ran out.
Does anything downstream depend on it? Nothing in the Khaṇḍa turns on whether Viśākha brought sixty-four crores. No battle is resolved by weight of numbers; the fights are settled by mantras, tridents, kuśa grass and prior arrangement.8 A figure that no later step consumes is a claim about scale, not an input to anything.
And does the source itself hold the figure steady? Here the same three names get sixty crores and then sixty-four, in one paragraph, and the translator has to say so. An inventory that cannot keep its own entries consistent is telling you what kind of document it is, and that is more useful than the numbers.
The residue is small: numbers in a persuasive register should be read as a claim about the speaker's confidence, not about the world — and the honest way to cite this passage is the volume describes a force it declines to total, which is both accurate and the only defensible thing to say.
So what is the census doing, if not counting?
Two things, and the chapter is fairly clear about both.
It is establishing that the outcome is not in doubt. This is the volume's answer to a structural problem it created for itself: Śiva's existing sixty thousand gaṇas were routed by one priest's mantras, and the survivors came home frightened.9 Having shown Śaiva force to be defeasible, the text needs the second expedition to be beyond question — and it does that by arithmetic rather than by argument.
And it is making a point about source. Every element of the march is Śiva's body or Śiva's command: the dogs from his body hair, the drums from his mouth, the beings "intent on obeying Śiva's command" including the ones he never commanded. The census is not a headcount. It is an inventory of what one god can emit.
Which puts it beside the volume's other counted passage. Dakṣa's sacrifice is enumerated — eighty-six thousand ṛtvijas, sixty-four thousand udgātṛs — and that inventory is used by Dakṣa himself as an argument: "since they are here, what is the need for Rudra?"5
Two armies of numbers, twenty chapters apart, both deployed as premises. One says the ceremony is complete without Śiva. The other says the response is beyond counting. The volume answers an enumeration with a bigger enumeration, and the second one is the one it wants believed.
Strongest support is the census itself — named leaders with crore-counts, escalating to "one thousand crores" and then to an uncountable summary — set beside Debroy's flag that part of it repeats itself with a changed figure.
Tension one, and it is the page's own. The passage counts and then abandons counting. The figure is never totalled and nothing downstream consumes it.
Tension two. Mahākālī marches with nine named Durgās, eleven classes of being and sixty-four categories of yoginī — having received no instruction from Śiva, while the text says the force is "intent on obeying Śiva's command."
Tension three. Debroy states the yoginī/mātṛkā taxonomy "is not consistent" — so the sixty-four is a conventional completeness-figure, not an enumeration, inside a passage built on enumerating.
Tension four. [n=1 OBSERVED] — one translator-flagged duplication with a numerical discrepancy, and no vocative anywhere near it. Two signals, not three; recorded as such.
Open questions. Is the "born from the body hair" generation mechanism used anywhere else in the three volumes at army scale? And do the nine Durgās named here appear together elsewhere in this volume, or only in this march?
Eighty-Six Thousand Ṛtvijas holds the volume's other counted force. The two are a matched pair — an enumerated ceremony arguing that Śiva is unnecessary, answered by an enumerated army that declines to total itself.
Vīrabhadra and Mahākālī From One Hair holds the creation. This page corrects it in one respect and sharpens it in another: Mahākālī is not narratively discarded — she marches with a named retinue — and she still receives no instruction.
Bhṛgu's Mantra Beats the Gaṇas holds the defeat this census is answering. The first Śaiva force lost to one priest, which is why the second one has to be uncountable.
Six Disclosed Repairs holds the flagged-versus-fixed distinction. Endnote 859 is the one place Debroy hands this build a duplication rather than an address error, and he flags rather than deletes it.
The Gaṇa Census holds the volume's earlier count of Śiva's forces, and the two should be compared before either figure is cited.
One handshake. A second was drafted against the vault's costly-signalling material and cut: that connection is already made by eighty-six-thousand-ritvijas, which is this page's direct counterpart, and repeating it would have split one argument across two pages.
Mass Movement Deployment Architecture (behavioral-mechanics). Hoffer's account is a sequence rather than a list of ingredients — an ordered chain where each step creates the conditions the next requires, with no optional elements, so that a movement failing at step two never becomes a mass movement no matter how well it handles everything downstream. Its analytic value is that it tells you where the load-bearing walls are.
The census is the opposite kind of description of a mass, and setting them side by side shows what enumeration cannot do.
The insight neither produces alone: a mass described by size tells you nothing about whether it can act, and this passage is the pure case. Hoffer's sequence would ask what the march is for at each stage, what it needs from the stage before, and where it would break. The census answers none of that; it answers how many, and then stops answering even that. And the volume has already supplied the evidence that size is the wrong variable — sixty thousand gaṇas were routed by one brahmana performing a correct oblation, so the text's own record shows the load-bearing wall was ritual authority rather than headcount. The second expedition is bigger and the difference that actually matters is that Vīrabhadra carries an explicit commission and a pledge-based warrant.10 An organisation that reports its scale when asked about its capability is describing the variable it can measure rather than the one that decides, and the tell is exactly the one here: the number is never totalled, and nothing downstream ever uses it.
Sharpest implication. The volume stops the story to count Vīrabhadra's army, and the count dissolves while you watch it: named leaders with ten crores, sixty-four crores, then "crores and crores", then two leaders with one thousand crores each — ten billion apiece, offered as a middle item — and finally a summary that abandons arithmetic altogether, "crores and thousands of crores, hundreds and twenty times that number." The unit fails partway through and the figure is never totalled, and nothing downstream in the Khaṇḍa ever consumes it. Inside the list, Debroy flags a defect: the same three leaders appear twice with the figure changed from sixty crores to sixty-four — "this sentence is unnecessary repetition." A repeated block and a numerical discrepancy at one site, with no vocative anywhere near it [n=1 OBSERVED] — two of the build's three seam-signals and not the third, which is what independent damage looks like. Meanwhile Mahākālī marches under her own heading with the nine Durgās named, eleven classes of being, and sixty-four categories of yoginī — on which Debroy notes the taxonomy "is not consistent", so the sixty-four is a completeness-figure and not a count. She was never given an order, and the text still says the force is "intent on obeying Śiva's command." What the census is actually for is legible from what it answers: Śiva's previous sixty thousand were routed by one priest chanting into a fire, so the volume responds to a defeat it cannot argue away with an army it declines to total — and every element of it, dogs from his body hair, drums from his mouth, is not equipment but emission.
Generative questions.