Author — Anonymous (traditionally ascribed to Krishna Dvaipayana Vedavyasa); translated by Bibek Debroy
Year — Text consolidated c. 1000–1400 CE. This translation: Penguin Classics, the second of three volumes.
Original file — RAW/books/Shiva Purana Volume 2 - Bibek Debroy.md (converted from EPUB; 3,725 lines, 212,824 words; body at lines 125–1503, translator's endnotes at 1505–3666)
Source type — Primary text in unabridged scholarly translation
Publisher — Penguin Books (Penguin Classics)
Sanskrit source text — Sivamahapuranam, Nag Publishers, Delhi, 1986
Coverage — 154 chapters, ch 151–304, in four sections: Kumara Khanda (Rudra Samhita 3.4, ch 151–170); Yuddha Khanda (Rudra Samhita 3.5, ch 171–229); Shata Rudra Samhita (ch 230–271); Koti Rudra Samhita, first part (ch 272–304). The Koti Rudra Samhita continues into Volume 3, which is unread; nothing on this stub or downstream of it asserts anything about Volume 3's contents.
Endnote layer — Introduction notes 1–14; body notes restart at 1 and run to 1067. The Introduction itself is verbatim identical to Volume 1's.
Volume 2 is the war half of the Shiva Purana and then its catalogue half. The first 79 chapters narrate five demon wars — Taraka, Tripura, Jalandhara, Shankhachuda, and the Hiranya–Andhaka–Bana sequence — and the remaining 75 convert the god into a taxonomy: a hundred named Rudras and avataras, then a geographical gazetteer of lingams with a local legend attached to each.
The argument running under both halves is that Shiva outranks Vishnu and Brahma in every register the text can find — narrative, ritual, etymological, and cartographic — while never quite denying the rival claim. The same volume that has Vishnu's avatara Nrisimha killed and flayed, his hide worn as a garment, also has Shiva conceding that he "emerged from Narayana's body," and closes the flaying chapter by promoting the flayed god to chief among devotees.
The second, quieter argument is procedural: this text is establishing a pilgrimage economy. It supplies Kashi with a cosmogony older than creation, names twelve jyotirlingams with their sites, pairs each with an upalingam, geo-locates episodes from Volume 1, and gives each shrine an origin story in which somebody was injured.
Classification rationale. Filed as primary-text for the same reason as Volume 1: the object of study is the Purana; Debroy's contribution is translation and annotation, not argument.
Chapter titles are Debroy's, not the text's — as he records for Volume 1. Any page anchored to a chapter title attributes the title to the translator.
No Critical Edition of any Purana exists, so verse and chapter numbering vary between editions and citation is inherently softer than for the Mahabharata.
The volume repeats itself without saying so, at least four times, and the repetitions are not compatible. Tulasi has two origins, both presented as the origin (ch 196, ch 211). The Avadhuteshvara episode is narrated twice with a shared distinctive simile and a shared name-etiology (ch 183, ch 259). Nrisimha's rage is resolved twice — cursed and rehabilitated (ch 216) versus killed and flayed (ch 241) — and the two versions disagree about whether Vishnu dies. Kartikeya's conception has two incompatible transmission-chains, and the sworn deposition at ch 154 omits the gods' pregnancy and the six wives that ch 152 narrates. None of the four is flagged by the translator, and this matters because the text has four working mechanisms for disclosing repetition — kalpa-variance (ch 163, ch 298), back-reference-and-decline (ch 239), announced re-telling (ch 286) — and used them elsewhere. Downstream pages should treat the text as a compilation whose seams are visible in the English.
One block is flagged as compositionally distinct by the translator himself. Note 501, on the Andhaka chapters: "The composition of these Andhaka chapters is quite different. The sentence constructions make comprehension difficult. Sometimes, the sentences are incomplete." Two independent confirmations are visible in the English — an attribution collapse at ch 214 ("The ministers said" twice, the second speech in the ascetic's voice) and an incomplete sentence at ch 216. These are corroboration of a translator's flag and should never be reported as unaided findings.
The disputed-Mahapurana tension carries forward from Volume 1 unchanged, and unresolved. Hazra treated this text as a late sectarian Upapurana and gave it no separate treatment, focusing on the Vayu Purana instead; Wilson noted the same substitution pattern between the two; Debroy declines to adjudicate. Any page treating this source as canonically equivalent to, say, the Vishnu Purana is choosing inside a live dispute and should say so.
Volume 2 supplies datable sectarian evidence that Volume 1 did not, and it belongs on the stub rather than buried in a concept page. Chapter 298 curses a group of brahmanas into foreheads smeared with mud, branding with hot coins, refusal of Shiva's singularity ("Shiva is without a second, but you will treat him on par with other devas"), and residence at Kanchi — note 1050 covers the branding. Urdhva-pundra, taptamudra, Kanchipuram: this is an origin myth for Shrivaishnavism written as a curse. Chapters 174–175 and 182 similarly manufacture a Jain/Buddhist heresy to order, with a composite heretic assembled from Jain iconography, "Buddhist agama texts" cited by name, and Carvaka materialism. A polemic this specific is evidence about the text's period and milieu, and narrows the plausible dating window at the late end.
Two standing flags apply corpus-wide and are named on downstream pages rather than re-argued per page.
🚩 MOTIVATED REASONING — structural. This is a sectarian text whose interest is in itself and its god. Its claims about what other traditions teach are the text advertising against a rival, never description.
🚩 TRANSLATION DISTORTION — present but strongly mitigated by Debroy's stated word-for-word method, his Shastri comparisons, and his habit of flagging alternative readings and his own repairs. Volume 2's disclosed-repair set is large: typo corrections (notes 214, 277, 423, 427, 437), a systematic class correction (note 442: "In several places in this chapter, there are typos, with the text saying suras instead of asuras. These have been corrected"), admitted liberties (501, 771), translation against the literal text (935), admitted non-comprehension (300, 387, 511), and a pre-emptive defence of the text's consistency (500).
Sectarian material is reported as the text's polemic, never as description. This governs the Arihat chapters (174–175), the damnation of future believers (182), Buddha's inclusion in the avatara list (186), and the ch 298 Kanchi curse, absolutely. Nothing drawn from these chapters describes Jainism, Buddhism or Shrivaishnavism; it describes what this text says about them.
Gender material must not be averaged. This one volume contains three sexual violations commissioned by the goddess (ch 192, ch 210–211, ch 221–222), a formal adjudication ranking a woman's service above a great sage's austerities (ch 274–275), a woman disqualifying her own sex while exercising authority (ch 198), a closing episode in which a woman argues Shiva out of a killing (ch 304), and a narrator's flat gloss that "the thoughts of co-wives are generally vile and harsh." Pages hold the spread rather than resolving it.
Naming caution — three names each denote two different figures. "Sudharma" is both Bhadrasena's son (ch 195) and Ghushma's husband (ch 303–304). "Sharada" is both a devotee in the ch 195 litany and Parvati's incarnation at ch 250. "Pingala" is a courtesan (ch 195), a gana named Pingalaksha (ch 203), and a Rudra named Pingala (ch 247). Check the referent before citing any of the three.
Confidence tagging. Claims drawn from the body are claims about what this text says, cited by line reference, and carry no web-source confidence tag. Claims about dating, transmission or scholarship drawn from Debroy's Introduction are tagged [SCHOLARLY — translator's introduction].
Evidence tagging for structural findings. Assertions this vault makes about the text — recurring formulae, unflagged doublets, pattern counts — carry [n=1 OBSERVED], [n=2 SUGGESTIVE] or [n≥3 PATTERN], and no rigour-register language below n=3.