Author — Anonymous (traditionally ascribed to Krishna Dvaipayana Vedavyasa); translated by Bibek Debroy
Year — Text dated by the translator to c. 500–1000 CE. This translation: Penguin Books, the second of three volumes.
Original file — EPUB, ingested 2026-08-12 from Downloads/Telegram Desktop/. ⚠ Not yet placed in RAW/books/. The file should be moved there for permanence; every citation on downstream pages currently traces to a download path that will not survive a disk clean.
Source type — Primary text in unabridged translation
Publisher — Penguin Books
Sanskrit source text — Nag Publishers, Delhi, funded by the ministry of human resource development. A transliterated text is available through the University of Gottingen.
Coverage — 119 chapters. Fifth Skandha (26 ch), Sixth (19), Seventh (15), Eighth (24), Ninth (24), and Tenth chapters 1–11. The volume breaks mid-Skandha — it closes "The Tenth Skandha will be continued in Volume 3." Volume 1 (Skandhas 1–4) and Volume 3 are not in this vault, and nothing on this stub or downstream of it asserts anything about their contents.
Endnote layer — 1,505 body notes numbered continuously across the six Skandhas (S5 1–400, S6 401–639, S7 640–859, S8 860–1068, S9 1069–1370, S10 1371–1505), plus 19 separate notes to the Introduction.
The Bhagavata Purana is a Vaishnava text arguing that devotion outranks every other route to liberation — ritual, knowledge, austerity, yoga and caste standing alike — and Volume 2 is where it argues that hardest, because it argues it through characters who should not qualify.
The through-line of these six Skandhas is that the operative variable is contact, not merit and not even intent. A brahmana who has destroyed his life calls his son's name at death, and the name happens to be Narayana, and that is sufficient (6.1–6.2). A demoness disguises herself to poison an infant, offers the breast, and is granted the destination of a mother because she offered it (10.6). Enmity is stated outright to be a faster route than devotion (7.1). Against these, the text repeatedly shows correct doctrine failing to move anyone who holds it: Hiranyakashipu delivers flawless Vedantic grief-counselling in the same chapter he orders the earth burned, and Kamsa argues impeccable non-doership, weeps, is forgiven, and is talked back into infanticide by morning.
The second argument is jurisdictional, and the text stages it as a formal dispute. In the Sixth Skandha, Vishnu's messengers and Yama's argue the law of atonement on its merits in front of a dying sinner, and Yama loses — then concedes his own authority, restricts the class of people his servants may approach, and asks Narayana's pardon. This volume is where a Vaishnava text puts the god of law on the record surrendering.
Running under both is a theological knot the text never unties. It asserts three incompatible accounts of divine agency and asserts each flatly: God as sovereign who takes away riches as a favour (8.22), God as servant who says "I am not independent and am completely controlled by my devotees" (9.4), and God as captive, tied to a mortar by a woman with a rope and staying tied (10.9).
Classification rationale. Filed as primary-text on the same basis as the Shiva Purana volumes: the object of study is the Purana. Debroy's contribution is translation and annotation, not argument.
The translator states his own method and it constrains every citation. From the Introduction: the aim is "to offer a translation, not an interpretation," with interpretation "relegated to footnotes"; and, decisively, "I have consciously done what is akin to a contemporary Manmatha Nath Dutt translation. Without distorting, I have chosen the simplest possible interpretation or translation that fits." He names the schools he is declining — Bhaktivedanta Swami Prabhupada, Swami Tapasyananda, and C.L. Goswami with M.A. Shastri, each of whom "follow one particular line of commentary or interpretation or school" — and rates Ganesh Vasudeo Tagare best from an academic's perspective while declining that too. This is not neutral access to the Sanskrit and must never be presented as such. A doctrinal claim sourced here is sourced to Debroy's simplest-reading English.
No diacritical marks, by deliberate choice — "tends to put off readers who are less academically inclined." Downstream pages use vault diacritics in prose and quote Debroy verbatim without them inside blockquotes.
No Critical Edition of any Purana exists. Debroy says so directly, and adds that his own text carries just over 14,000 shlokas across 335 chapters against the 18,000 traditionally ascribed — partly missing verses, partly a counting difference. Citation is inherently softer than for the Mahabharata.
The volume breaks mid-Skandha. Anything about the rest of the Tenth Skandha, or about Skandhas 1–4 and 11–12, is out of range. Absence claims on downstream pages are bounded to "not found in Volume 2 as read."
The translator's editorial hand is visible and disclosed, and the disclosures are themselves evidence. Volume 2's set:
The long-range cross-reference system closes. The narrative makes at least seven forward promises of the form "I will tell you about the reason later," and every one is resolved by an endnote pointing to the Skandha that pays it off. Six of the seven pay off inside this volume; only Bana's character (note 617) does not. This is a text that keeps its appointments, and downstream pages may rely on it.
⚑ ENRICHED 2026-08-13 from the Volume 3 build. The finding above is correct for this volume read alone, and it does not survive contact with Volume 3. The apparatus does not close across volumes — it disagrees with itself backward, at Volume 3's notes 503, 512, 1307 and 1354. So "this is a text that keeps its appointments" should be read as within Volume 2, and downstream pages relying on it must not extend the reliance across the volume boundary.
Volume 3 also supplies the harder version of the same problem. Its front matter lists chapters 10(85) and 10(86) at 59 shlokas each; neither is printed in the EPUB as ingested; and endnote 673 cross-references 10(86) as though it were present. That is the apparatus disagreeing not with another volume but with its own body — an appointment the system announces and does not keep, unflagged. Volume 3 further carries six internal arithmetic failures in its Twelfth Skandha alone, two flagged and four silent, every one checkable against material on the same page.
⚠ The correct downstream statement is therefore narrower than this section originally supported: the apparatus is reliable for what it contains, is not complete, and is bounded per volume. No claim here is made about the Sanskrit or about any print edition — all of it is bounded to these EPUBs as ingested.
Two findings about the translator that are the vault's, not his. Neither is a correction; both are observations he did not make.
First, the four-direction dakshina at 9(11) and 9(16) is word-identical across five chapters and two different avataras, and inside that identical sentence he renders the udgatri by function in one place ("the one who chanted the Sama hymns") and by transliteration in the other. He annotates the first instance and leaves the second unannotated and uncross-referenced.
Second, at 6(18) the sentence "Bali had one hundred sons and Bana was the eldest. Through excellent shlokas, we will later describe his praiseworthy character" is resolved by note 617 to Bali. The nearer referent is Bana. In the Eighth Skandha as read here in full, Bali's character is described at length and Bana appears only as a name in the battle roster at 8(10). The pronoun is genuinely ambiguous and the reading is defensible; it is recorded, not asserted as an error.
Three standing flags apply corpus-wide and are named on downstream pages rather than re-argued per page.
🚩 MOTIVATED REASONING — structural. This is a sectarian Vaishnava text. Its rankings of paths, its placement of Shiva in subordinate or compromised positions (8.12, where Shiva loses control before Mohini; 7.10, where a Vaishnava trick wins the Tripura victory attributed to Shiva), and its accounts of what rival positions hold are the text advertising, never description.
🚩 TRANSLATION DISTORTION — present and only partly mitigated. Mitigated by Debroy's word-for-word aim and his habit of flagging alternatives and his own repairs. Not mitigated by his stated method, which is the opposite of a scholarly apparatus: he selects the simplest reading that fits and says so. On any passage where doctrine turns on a fine distinction, the page must say that a commentarial tradition would read it otherwise, and that this translation declines to.
🚩 SECONDARY WITHOUT PRIMARY. No Sanskrit is consulted anywhere in this vault's use of this source. Claims are about what Debroy's English says.
Gender and caste material must not be averaged or softened. This volume contains the etiology of menstruation as absorbed murder-guilt (6.9); the cosmic-body varna hierarchy with "the ignorant shudras emerged from his feet" (8.5); a womb-taught son retaining what his mother forgets "because she is a woman" (7.7); "A mother is a mere receptacle" as the ruling that legitimises a repudiated child (9.20); Kashyapa's razor-hearted-woman speech (6.18); and two instances of a female character voicing the text's misogyny about herself — Mohini's "A learned person never trusts a woman" (8.9) and Urvashi's "The hearts of women are like those of wolves" (9.14). Against these sit Lakshmi choosing her own husband after rejecting every other being by name of defect (8.8), Vindhyavali's objection to the whole premise of ownership (8.22), and Yashoda binding God with a rope (10.9). Pages hold the spread rather than resolving it.
A speech-attribution asymmetry is recorded as an open question, not a finding about the Purana. Counting the text's formal discourse-opening device, Narada receives it 16 times, no other speaker exceeds 3, and named women receive roughly 8 between them. Yashoda — the domestic centre of the whole Tenth Skandha material here — never receives it once; her speech arrives through narration. Whether this tracks the Sanskrit uvaca attributions cannot be determined from this volume, and no page may claim it does.
Confidence tagging. Claims drawn from the body are claims about what this text says, cited by chapter reference, and carry no web-source confidence tag. Claims about dating, transmission, shloka counts, or rival translations drawn from the Introduction are tagged [SCHOLARLY — translator's introduction].
Evidence tagging for structural findings. Assertions this vault makes about the text — recurring formulae, repeated blocks, pattern counts — carry [n=1 OBSERVED], [n=2 SUGGESTIVE] or [n≥3 PATTERN], and no rigour-register language below n=3.