The twelve jyotirliṅgas are the most portable object in this volume. Twelve names, each with a place attached, compressed into a single passage a person can memorise in an afternoon:
"Somanatha in Sourashtra; Mallikarjuna in Shrishaila; Mahakala in Ujjayini; Amareshvara in Omkara; Kedara on the slopes of the Himalayas; Bhimashankara in Dakini; Vishvesha in Varanasi; Tryambaka on the banks of Goutami; Vaidyanatha in a cremation ground, Nagesha in the forest of Daruka; Ramesha in Setubandha; and Ghushmesha in Shivalaya."1
Debroy's note identifies what this is: "one variant of the famous shloka that lists the twelve jyotirlingams."2 A verse older than the chapter carrying it, quoted in.
And the reward for saying it is stated immediately, before any of the twelve is described: "If a person wakes up in the morning and reads these, he is freed from all sins and obtains success in everything as a fruit."3
⚑ Note the order of operations. The list is issued complete, with its recitation-reward attached, before a single story is told. Everything that follows — twenty chapters of episodes — is filling in a structure that already works without them.
The volume gives the list twice in quick succession, at the end of one saṃhitā and the start of the next.
Ch 271, closing the Śata Rudra Saṃhitā: "Amareshvara in Omkara."1 Ch 272, opening the Koṭi Rudra Saṃhitā: "Parameshvara in Omkara."3
Eleven names identical, one different, across a section boundary — and the chapter that eventually tells the story is headed Greatness of Omkareshvara Jyotirlingam.
⚑ Three names for one liṅga inside forty pages, and no acknowledgement of variance.
The translator's note explains what is underneath: at Oṃkāra there are in fact two liṅgas — "Omkara or Omkareshvara is on the island of Mandhata… The second lingam is on the bank of the river and is known as Amareshvara (Mamaleshvara)."2 So the slot itself is contested on the ground, and the variance in the shloka is the tradition's unresolved question surfacing as a variant reading.
That is a small thing, and it is the kind of small thing that shows what a list of this sort actually is: not a description of twelve places, but a verse with twelve slots, some of which the tradition had not settled.
Now the check that produces the page's real finding.
The list is issued at ch 271–272. The stories arrive at ch 285–304, in a different saṃhitā, after twelve intervening chapters on other matter. Do they follow the shloka?
Exactly, and completely:
| # | shloka | chapters |
|---|---|---|
| 1 | Somanātha | 285 |
| 2 | Mallikārjuna | 286 |
| 3 | Mahākāla | 287–288 |
| 4 | Oṃkāra | 289 |
| 5 | Kedāra | 290 |
| 6 | Bhīmaśaṅkara | 291–292 |
| 7 | Viśveśa (Kāśī) | 293–294 |
| 8 | Tryambaka | 295–298 |
| 9 | Vaidyanātha | 299 |
| 10 | Nāgeśa | 300–301 |
| 11 | Rāmeśa | 302 |
| 12 | Ghuṣmeśa | 303–304 |
Twelve for twelve, in sequence, with chapter counts varying from one to four according to how much material each attracted.4
⚑ Set that beside what the volume does not keep stable across the same span — a name that drifts three ways, two locations nobody can pin down, and an upaliṅga list that comes up short. The sequence is the most robust thing in the whole apparatus. It survives a saṃhitā boundary, a change of narrator's frame, and twenty chapters of intervening material, without a single transposition.
Which is what you would expect if the shloka is the primary object and the chapters are commentary on it. The order is not the compiler's choice; it is the verse's, and the verse is older.
Chapter 272 does something odd before it gives the list, and the oddity is worth pausing on because the chapter does not notice it.
It opens by declaring the set uncountable:
"The total number of all Shiva's lingams cannot be enumerated. Everything on earth is full of lingams. Everything in the universe is full of lingams. All the tirthas are full of lingams. Everything is established on lingams. Since they have no number, I will speak about a few… Everything that is seen, can be seen, described or remembered, is Shiva's form. Nothing exists without him."3
Then, four sentences later, the canonical twelve, with a recitation-reward attached.
⚑ Two claims sit side by side without a word of reconciliation. Every object in the universe is a liṅga — a strong non-dual position, stated plainly. And these twelve are the ones that discharge your sins if you say their names in the morning — a strong particularist one.
The bridge the chapter offers is procedural rather than theological: "I will speak about the major ones. Even among the major ones, I will speak about those that are most important." A selection from an infinite set, ranked, with no criterion given for the ranking.
And the mechanism of instantiation is stated too: *"Wherever Shambhu's devotees remember him with devotion, he assumes an avatara there. Having performed the task, he remains established there."*3 So the twelve are not places the god chose. They are places where somebody's devotion summoned him and he stayed — which makes the canon a record of where the traffic was, dressed as a record of where he is.
⚑ That reading is not the chapter's own, and it should not be attributed to it. But it is what the chapter's stated mechanism produces if you take it literally: an infinite field of liṅgas, twelve of which became famous because people went there, and a verse that then fixes those twelve permanently.
Chapter 272 continues past the twelve into their subordinate liṅgas, and this is a third instance of the volume's characteristic arithmetic failure.
The upaliṅgas given: Antakeśa (for Someśvara) · Rudreśvara (Mallikārjuna) · Dugdheśa (Mahākāla) · Kardameśa (Oṃkāra) · Bhūteśa (Kedāreśvara) · Bhīmeśvara (Bhīmaśaṅkara) · Bhūteśvara (Nāgheśvara) · Gupteśvara (Rāmeśvara) · Vyāgreśvara (Ghuṣmeśa).
Nine. Then: *"O brahmanas! I have spoken about the upalingams of the jyotirlingams."*3
⚑ Three have none — Viśveśa, Tryambaka and Vaidyanātha — and the chapter closes as though the set were complete. No note addresses it.
Two smaller oddities in the same passage. Bhīmeśvara is given as the upaliṅga of Bhīmaśaṅkara, while chapters 291–292 are headed Greatness of Bhīmeśvara Jyotirliṅgam — the same name serving as both the principal and its subordinate. And the last two entries, Gupteśvara and Vyāgreśvara, are the only ones with no location attached at all, where the previous seven each name a river, a confluence or a mountain.
The list is running out of information as it goes, and it ends anyway with a completion formula.
The translator is explicit about which slots are secure, and this is the most useful thing in the note layer here:
"Geographically, ten can unambiguously be identified. But the geographical identification of Bhimashankara (in the region of Dakini) and Nagesha (in the forest of Daruka) is not that robust."2
And the reasons are specific rather than vague.
For Bhīmaśaṅkara, the text's own internal evidence pulls against the standard identification: "The Bhimashankara lingam is usually identified as the one near Pune. However, the mention of the king of Kamarupa places this in Assam and there is a Bhimashankara temple near Guwahati." The note adds that the volume names the king Sudakṣiṇa where the tradition names Priyadharma.2
For Nāgeśa, the problem is upstream: "The crux is the location of Darukavana."2 The liṅga is located by a forest whose location is itself unknown.
⚑ These two are exactly the two whose slots in the shloka name a region or a forest rather than a city or a river — Ḍākinī, Dārukavana — while the ten secure ones name Saurāṣṭra, Ujjayinī, Vārāṇasī, Setubandha and so on. The verse's vagueness and the tradition's uncertainty are the same vagueness, preserved for a thousand years because the verse was what got transmitted.
The rewards attached to the twelve are stated at length, and they describe something other than a pilgrimage.
"If men see them, or touch them, they obtain every kind of bliss" (271). "If a person wakes up in the morning and reads these, he is freed from all sins" (272). "As soon as one hears about them, all sins are dispelled." And the graded ladder: worship any one of them for six months and escape rebirth; a person of inferior birth who sees a jyotirliṅga is "reborn in a sparkling and virtuous lineage"; and — "If a person is born as a mleccha, an antyaja or a eunuch, if he sees a jyotirlingam, he is reborn as a brahmana and attains emancipation."3
⚑ Seeing, touching, reading, hearing, and worshipping one of the twelve are all sufficient, at descending cost, for overlapping rewards. The list is designed so that the cheapest possible engagement still pays — and the cheapest is reciting the verse on waking, which requires no travel, no priest, no money and no liṅga.
Which is why the verse is the durable artifact and the twenty chapters are not. A structure whose minimum viable act is saying twelve names will outlive any particular account of why those twelve.
You are looking at a canonical list — the seven habits, the twelve steps, the four noble truths, a company's five values — and you want to know how it will behave over time.
Ask what the minimum engagement is. If reciting the list is itself sufficient to claim participation, the list will outlive every explanation attached to it, and it will be reproduced by people who cannot supply those explanations. That is not decay; it is the design working.
Then check what has actually stayed fixed. ⚑ Across two statements in different sections and twenty intervening chapters, this list kept its order perfectly and lost a name three ways. Sequence is the most robust carrier; labels drift and referents go vague. If you need something to survive transmission intact, encode it in the order rather than in the wording.
Find the slots that were never specific. Ten of these twelve name a city, region or river; two name a forest and an unlocated district, and those two are precisely the two nobody can find today. Vagueness at the moment of composition does not get resolved later — it gets preserved, because the vague version is the one that travels.
And count the sub-lists. The subordinate list here declares itself complete at nine of twelve. Derived lists are where the care runs out, because nobody recites them.
Evidence. Ch 271's list and its per-liṅga glosses read in place;1 ch 272's list, upaliṅga passage and reward-ladder read in full;3 the chapter-by-chapter sequence at 285–304 checked against the shloka's order directly from the chapter headings;4 endnotes 938–950 read at their own lines.2
Tension 1 — Amareśvara / Parameśvara / Oṃkāreśvara. Three names for one slot within forty pages, unremarked in the body. The note explains the underlying situation (two liṅgas at Oṃkāra) without connecting it to the variance.2
Tension 2 — the upaliṅga list declares completion at nine of twelve.3 The third instance of announced-complete-versus-enumerated-short in this volume, and the only one no endnote touches.
Tension 3 — the text's own detail contradicts the standard geography. The Kāmarūpa king in the Bhīmaśaṅkara material points to Assam against an identification near Pune, and the volume's name for that king differs from the tradition's.2 Preserved, not resolved.
Open question. The chapters supplying stories for the twelve run 285–304 and the volume ends there. Whether any of the twelve receives a second, competing account elsewhere in the corpus is not determinable from the range read, and no claim is made about Volume 3.
Against The Liṅga That Had to Be Caught: that page holds a single liṅga's episode at full narrative weight. This one holds the frame those episodes were written into, and the relationship runs the unexpected way — the slot existed first and the story came to fill it.
Against Kāśī Promises Everything and Takes It Back: Viśveśa is item seven, gets two chapters, and is one of the three with no upaliṅga listed. The volume's most theologically loaded site is under-served by its own cataloguing apparatus.
Against The Five Faces and the Eight Forms: the same defect, a third time — a set announced complete and enumerated short. Three instances now, in three different schemas, which moves the observation from anomaly toward habit.
Against The Two Gautama Accounts: Tryambaka takes four chapters, more than any other of the twelve, because the Gautama material is attached to it. Chapter allocation tracks accumulated story rather than doctrinal rank.
A contemporary Śaiva teaching corpus makes a claim about this circuit that the Purāṇic material has no occasion to make: every one of the twelve has a Bhairava attached to it, and two of the positions are held underground, in Pātāla.5
⚑ What's worth carrying is not the overlay but the observation attached to it. People who have completed all twelve — years of travel, the thing you mention afterwards — are asked whether they saw the Bhairava at any of them. "They don't have the information at all."5
Not turned away. Not refused. Standing in the kṣetra, with the thing there, and it did not register.
That's a claim about how a famous circuit is actually walked, and it cuts against the way this page's sources treat the twelve. A Purāṇic list is a list of destinations, complete when visited. This says a destination has a layer in it that completion does not touch — you can finish the circuit and have missed a whole order of what is standing there.
⚠️ Nothing here establishes that the Bhairava attachments are old, textual, or uniform. 🚩 NO LOCATOR — no text is named for the overlay. 🚩 SINGLE SOURCE. [PARAPHRASED] — auto-generated captions. Recorded as a living tradition's mapping onto this circuit, not as a claim about the Purāṇic material above.
Four-Anchor Memorization — that page holds Dicks's discipline for stage storytelling: memorise only the first line, the last line, the scene bookmarks and the moment of change; everything between the anchors is remembered and improvised live.
The jyotirliṅga material is that principle observed at the scale of a millennium rather than an evening, and the direction of evidence is what makes the pairing worth having.
Dicks prescribes holding the skeleton and letting the flesh vary. This volume demonstrates that a skeleton is what actually survives when nobody is prescribing anything. Across a saṃhitā boundary and twenty intervening chapters, the order of twelve came through with no transpositions — while a name drifted three ways, two locations went permanently vague, and the derived list ran out at nine of twelve. Nobody decided that sequence would be the durable element. It was durable because it was the thing the verse encoded, and the verse was the thing people said.
⚑ Which sharpens the prescription into a design rule rather than a performance tip: if you need something to survive being passed along by people who will not have your explanations, put it in the order, not in the wording. The anchors are not merely easier to hold — they are the only part that transmission does not corrode. And the corollary is uncomfortable for anyone building a framework: whatever you leave in the flesh, you have decided to lose.
One handshake, not two. The recitation-reward material would pair with symbolic-substitution work in behavioral-mechanics, but every candidate there frames the substitution as manipulation, which is a reading this material does not support.
Sharpest implication. The twelve jyotirliṅgas arrive as a memorisable verse with its reward already attached — read these on waking and you are freed from all sins — before a single one of the twelve stories is told. Twenty chapters then supply those stories, in a different saṃhitā, in exactly the shloka's order, twelve for twelve, with no transposition. And across that same span the volume loses a name three ways (Amareśvara, Parameśvara, Oṃkāreśvara, for a slot where the tradition genuinely has two liṅgas), leaves two of the twelve at addresses nobody can now locate — precisely the two whose slots named a forest and a region rather than a city — and closes a list of subordinate liṅgas at nine of twelve with a completion formula. The sequence is the most robust object in the apparatus and the wording is the least. Which follows from what the list is built to do: its minimum viable act is saying twelve names, so the names and their order are what travelled, and everything explaining them is commentary that could be, and was, replaced.
Generative questions.