Final answer — where religions come from, what they share, what they do, and whether the shared stories have one source

The short answer to all four questions at once: every religion whose history can be dated turns out to be a regional development with a traceable formation period — there is no moment where "religion" appears fully formed, and no evidence that all religions descend from one original religion. What they share — floods, sky fathers, dying gods, ancestor rites, the conviction that their own teachings are timeless — they share for three different reasons: some stories were inherited from a common ancestral culture (descent), some were borrowed along trade and scribal routes (contact), and some were reinvented independently because human minds facing the same problems produce similar answers (convergence). The three legs are not equally proven, and saying so plainly is part of the answer: descent and contact are demonstrated at high confidence — sound-law grade for the sky father, textual-dependence grade for the flood — while convergence is the best available reading for specific cases (the Osiris strand, the global flood distribution, the non-Indo-European sky-father figures) but is nowhere proven at the same grade; each of those cases remains an open question (Q5, Q7, Q11). The picture is a braided river — many streams, splitting and rejoining — not a single trunk with branches.

What follows answers the four parts in order, citing the vault note behind every claim and carrying each note's confidence rating with it.

---

Part 1 — Where does each religion come from?

Before any religion we can name, there is ritual we can date. The earliest material plausibly ancestral to religious behavior is mortuary: at Qafzeh Cave (Israel, c. 92,000 BP) anatomically modern humans buried their dead with deliberately transported, heat-treated red ochre and perforated shells — evidence of a socially shared symbolic code in a death context, at medium confidence, because utilitarian uses of ochre cannot be fully excluded (claim-qafzeh-ochre-earliest-symbolic-behavior). At Sungir (Russia, c. 34,000–30,000 BP) a community sank roughly ten thousand hours of labor into grave goods — beads, ivory spears, ochre — for three burials, implying a culturally enforced obligation to the dead, again at medium confidence (claim-sungir-collective-labor-mortuary-ritual). What none of this evidence can deliver is belief content: the inference from "ochre on burials" to "belief in an afterlife" is a cognitive hypothesis, not something readable from stratigraphy (paleolithic-mortuary-religion). The popular idea that shamanism/animism was humanity's "baseline religion" from which all others grew is rated low confidence in this vault: the cross-cultural pattern is real, but it is at least as well explained by independent reinvention from shared neuropsychology as by a single ancestral tradition (shamanism-animism-baseline-religion-hypothesis).

Mesopotamia — the oldest continuous trail. The earliest structure interpretable as a dedicated cult building in Mesopotamia is Eridu temple Level XVIII, c. 5500 BCE — a mud-brick room with altar and offering debris, beginning an unbroken 2,000-year sacred-architecture sequence; the "temple" label is an interpretation, so this is medium confidence (eridu-temple-level-xviii-earliest-cult-building). The named Sumerian pantheon (An, Enlil, Enki, Inanna) enters writing c. 2600–2500 BCE in the Fara god-lists; whether the Ubaid-period builders of Eridu were culturally continuous with historical Sumerians is an open question, so "Sumerian religion" may be a later label on an older substrate (sumerian-religion). This tradition supplied the ancient world's most influential flood narrative, written down by c. 1635 BCE (atrahasis-antedates-gilgamesh-flood-tablet, high confidence on the textual priority).

Egypt. Structured afterlife provisioning is archaeologically visible from the Badarian burials, c. 4400 BCE — contracted bodies, grave goods, a west-facing orientation kept for three millennia (ancient-egyptian-religion). The Pyramid Texts (pyramid of Unas, c. 2350 BCE) are the oldest surviving large-scale written religious corpus — a claim that is high confidence for the surviving inscription, explicitly not for intellectual priority over Sumerian material (pyramid-texts-oldest-large-corpus). The Osiris cycle — the murdered, reconstituted god who rules the dead — is securely attested from the 5th Dynasty, with anything earlier being inference (osiris-cycle-earliest-attestation).

The Indo-European family. Proto-Indo-European religion is a scholarly reconstruction, not an observed tradition: no PIE texts exist, and the reconstruction (sky-father, storm god, dragon-slaying formula, sacred fire) is evidence-class 3 throughout, held at medium confidence (proto-indo-european-religion). Its daughters are dated: Hittite religious texts from c. 1650 BCE, Mycenaean Greek "to Zeus" (di-we) on Linear B tablets c. 1400–1375 BCE (sky-father), and the Rigveda — composed c. 1500–1200 BCE by linguistic inference, externally anchored by the Mitanni treaty naming Mitra, Varuna, Indra and the Nasatyas c. 1380 BCE, but physically attested only in manuscripts from c. 1040 CE (reported) / 1464 CE (secure). That ~2,500-year oral gap is bridged by the most elaborately safeguarded oral transmission known, yet composition dates remain inference, not observation (rigveda-oral-gap-composition-vs-attestation, rigveda-oral-composition-attestation-gap).

The Abrahamic line. Israelite religion did not appear from nowhere: it emerged inside the Canaanite religious world documented by the Ugaritic texts (c. 1400–1200 BCE). The earliest possible external trace of the divine name is Egyptian — "the land of the Shasu of yhwꜣ," Soleb, c. 1390–1352 BCE — read as Yahweh by most specialists but not all (israelite-religion-origins). Early Israel identified Yahweh with the Canaanite high god El, absorbing El's titles and council: high confidence that the identification is present in the texts, medium for the mechanism (inherited unity vs. post-settlement merger — Day's school argues the latter) (el-yahweh-identification-canaanite). At the popular level, inscriptions from Kuntillet Ajrud (c. 800 BCE) bless "by Yahweh of Samaria and his Asherah" — evidence, at medium confidence (goddess vs. cult-object reading), that early Yahwism was not yet monotheism (kuntillet-ajrud-yahweh-asherah-popular-religion). The arc from national god to explicit "no god but me" monotheism runs through Josiah's reform and crystallizes in the Babylonian exile (6th century BCE); Christianity and Islam are downstream developments of this Second Temple matrix — the former applying messianic and Greek philosophical categories to Jesus within 1st-century Judaism, the latter drawing on the monotheistic traditions present in 7th-century Arabia (israelite-religion-origins).

The Dharmic line. Vedic religion descends from Proto-Indo-Iranian religion (shared with the Iranian Avesta — the soma/haoma ritual is the clearest cognate), carried into northwest India by steppe-descended Indo-Aryan speakers in the mid-second millennium BCE, with contested substrate contributions from post-Harappan populations (vedic-religion, medium confidence). What most people now think of as the heart of Indian religion — karma, rebirth, liberation — is not in the early Vedas: it first appears, tentatively, in the Upanishads c. 700–600 BCE, and the balance of evidence (at medium confidence, genuinely contested) supports an origin in the non-Vedic śramaṇa culture of the eastern Ganges basin, the same milieu that produced Buddhism and Jainism, later absorbed by Brahmanism (karma-rebirth-doctrine-origins).

East Asia. China's earliest direct religious evidence is spectacular in its precision: oracle bones from Anyang, radiocarbon-dated to 1254–1197 BCE, recording royal divination addressed to named ancestors and the high power Di — simultaneously the earliest dated Chinese writing and the earliest documented Chinese religion (high confidence; oracle-bones-earliest-dated-chinese-religious-writing). Ancestor-oriented ritual is archaeologically visible far earlier — Longshan mortuary assemblages from c. 3000 BCE — but whether that is the same tradition or a parallel invention is unresolved (ancestor-veneration-antiquity-china, medium). The Zhou conquest (c. 1046 BCE) replaced the Shang ancestor monopoly with Tian and the Mandate of Heaven — a documented case of theology redesigned for legitimation (shang-religion). Japan is the cautionary tale in dating: kami worship is textually attested only from 712–720 CE (Kojiki, Nihon Shoki — both court political projects), with material kami-site ritual dated about three centuries earlier and everything before that being inference (shinto-textual-attestation-gap, high confidence in the gap itself).

Indigenous oral traditions — the limit case. Australia shows the method's boundary honestly. Human occupation with ochre reaches back c. 65,000 years, rock art c. 28,000 years — but the Dreaming's theological content is attested only by 20th-century ethnography, and the assumption that modern belief preserves Pleistocene belief is structurally unfalsifiable with current methods (australian-aboriginal-dreaming, medium). The celebrated claim that Aboriginal stories preserve memories of post-glacial coastal flooding 7,000–13,000 years ago is rated low confidence: the geological events are real, but the dating of the stories assumes the very transmission fidelity it needs to prove (aboriginal-oral-tradition-coastal-flooding-time-depth-claim).

The pattern across all domains: in every domain with archaeological coverage, the earliest evidence is ritual before theology, archaeology before text — and when text arrives, it arrives in the service of states and elites (pyramids, king lists, royal divination, court chronicles). No documented religion begins as a finished system; each one forms over centuries from identifiable predecessors.

Part 2 — What common traits do they share?

Across eight evidence domains (the ninth, 09_comparative, is the synthesis layer over them, not an independent trail), six traits recur. (These are observed regularities in the vault's documented sample — a strong cross-cultural pattern, not a law.)

1. The dead remain socially present. From Qafzeh's ochre (c. 92,000 BP) to Egyptian funerary corpora, Shang ancestor petitions, and Osiris's kingship over the dead, treating the dead as still mattering — still owed labor, goods, ritual attention — is the oldest and most universal documented symbolic mortuary behavior; what the actors believed while doing it is a class-5 inference, not something the evidence reads out directly (paleolithic-mortuary-religion, ancestor-veneration-antiquity-china, osiris-cycle-earliest-attestation).

2. Specialists mediate. Sumerian temple personnel, Egyptian lector-priests, Vedic ritualists with hereditary recitation disciplines, Shang diviners, Dreaming knowledge-holders: every documented tradition develops people whose role is privileged traffic with the unseen (sumerian-religion, vedic-religion, shang-religion, australian-aboriginal-dreaming).

3. The social order is cosmically anchored. Kingship "descended from heaven" (Sumer), pharaoh as Horus/Osiris (Egypt), the Mandate of Heaven (Zhou), covenant election (Israel), Dreaming Law (Australia): in every state-level case, and in non-state cases too, religion grounds authority and norms in something bigger than any living person (sumerian-religion, ancient-egyptian-religion, shang-religion, israelite-religion-origins, australian-aboriginal-dreaming).

4. Shared story-stock. Specific narrative motifs recur across traditions that had no obvious connection: the world-destroying flood with a warned survivor, the sky addressed as father, the deity who dies and is mourned and in some sense returns (great-flood, sky-father, dying-rising-god). Part 4 traces where each actually came from.

5. Emic timelessness, etic development. Strikingly, the literate traditions all deny their own historical origins from the inside: the Vedas are apauruṣeya — not of human authorship (rigveda-oral-composition-attestation-gap); biblical monotheism presents itself as monotheism from creation, not a development (israelite-religion-origins); the Kojiki grounds the present in a primordial age of the gods (shinto-textual-attestation-gap); the Dreaming is "everywhen" (australian-aboriginal-dreaming). The vault records these self-accounts respectfully as emic data; the etic evidence in every dateable case shows gradual formation. The gap itself — every tradition experiencing its truths as eternal while leaving a paper trail of development — is perhaps the most consistent shared trait found.

6. Ritual response to uncertainty. Divination (Mesopotamian extispicy, Shang pyromancy), petition and sacrifice (Vedic yajña, temple offerings), and crisis ritual cluster precisely around what people cannot control — weather, war, illness, death (shang-religion, sumerian-religion, vedic-religion, what-religions-do).

Part 3 — What are religions mainly used for in culture?

The vault's dedicated synthesis (what-religions-do, medium confidence) answers this with a layered model, because the classic single-function theories each answer a different question:

• Why do god-ideas exist at all? The cognitive by-product account: human minds over-detect agents and easily entertain invisible persons, so religious representations arise everywhere without needing a purpose — the supply side is free.
• Why do some get institutionalized and kept? Because they do group-level work, and the dated record shows four kinds: cohesion (collective ritual bonds groups — millennia-long lament cults are organized collective emotion), cooperation-enforcement (moralizing watchful gods correlate with large-scale trust, though the causal arrow is genuinely unresolved since the flagship 2019 study was retracted), legitimation (the best-dated function in the whole record — the earliest large corpora are royal instruments: Pyramid Texts, Sumerian King List, oracle bones, Mandate of Heaven), and identity-marking (costly boundary practices — dietary law, circumcision — intensify exactly where group boundaries are under pressure, post-exilic Judah being the dated example).
• When does religious behavior intensify? Under uncertainty and danger — Malinowski's anxiety-reduction answers the when, not the why.

One honest caveat carried from that synthesis: functionalism risks hindsight reasoning (anything that persists can be called "functional" afterwards), and highly secular societies sustain cooperation without big gods — so these functions are real but substitutable, which is why this part stays at medium confidence.

Part 4 — Do the shared stories trace back to common origins? (The heart of the answer)

This is the question the vault was built to answer, and the answer is: it depends on the story — and we can actually tell which is which. Three motifs were traced end-to-end, each with dated attestations and an explicit transmission verdict.

The Great Flood: BORROWED (contact) — high confidence, scoped to the Near Eastern → biblical lineage. The flood story of Genesis demonstrably descends from Mesopotamian literature by cultural borrowing, not common ancestry and not coincidence. Every link in the chain is attested: the story is canonical Mesopotamian literature by c. 1635 BCE (Atra-ḫasīs, colophon-dated tablets — atrahasis-antedates-gilgamesh-flood-tablet); it physically circulated in the Levant in the Late Bronze Age (an Atra-ḫasīs fragment at Ugarit, a Gilgamesh fragment at Megiddo); Judah's literate elite then lived in Babylonia during the exile; and Genesis matches Gilgamesh XI in sequence, not just theme — warned hero, pitch-sealed vessel built to specification, animals, mountain grounding, serial bird release, sacrifice, the deity smelling the offering (great-flood). Sequential, near-verbatim correspondence is the signature of textual dependence — the same signature by which we know Gilgamesh XI itself incorporated Atra-ḫasīs (the borrowed text even keeps the older hero's name in two lines). What remains open is only when the borrowing happened (Late Bronze vs. exilic — register Q4). Crucially, the verdict is scoped: flood myths in the pre-contact Americas and Oceania cannot derive from Mesopotamia by any documented route, so the global flood-myth distribution is a separate, unresolved question — convergence is the default there, with deep descent (Pleistocene population movement) a live but speculative alternative (Q5).

The Sky Father: INHERITED (descent) — high confidence, scoped to the Indo-European cognate set. Vedic Dyáuṣ pitṛ́, Greek Zeû páter, and Latin Iū-piter are not similar gods — they are the same two words, "Sky Father," pronounced according to each language's own sound laws, preserved as a fixed prayer formula across branches separated by thousands of kilometers and years (sky-father). This is the benchmark case because it has a discriminator no other motif has: regular sound correspondence, the same machinery that makes the Indo-European language family a fact rather than a guess. Borrowing would leave loan phonology (it doesn't); chance cannot produce systematic correspondence across five-plus branches. Earliest attestations: Hittite root reflexes in 16th-century BCE copies, Mycenaean di-we ("to Zeus") c. 1400–1375 BCE. Two deflations keep this honest: what is securely inherited is the name, formula, and sky-identity — the fuller reconstructed PIE deity with theology attached stays at medium confidence (dyeus-sky-father-cognate-set); and inheriting a name did not preserve importance — Zeus became king of his pantheon while his exact Vedic cognate Dyaus shriveled to a genealogical fossil. Outside the IE family, sky-father-like figures (Chinese Tian, Turkic Tengri, ethnographic High Gods) show the concept recurs widely — most plausibly by independent invention, though steppe-contact routes remain open (OPEN_QUESTIONS Q11); the descent verdict covers the cognate set only.

The Dying-and-Rising God: UNRESOLVED — because the category itself fragments under scrutiny (medium confidence in that verdict). What looks like one motif is three different things wearing one 19th-century label (dying-rising-god): (1) a continuous Mesopotamian Dumuzi/Tammuz tradition — descent within its own lineage for two millennia — which then demonstrably spread west by contact (his mourning rites practiced in Jerusalem per Ezekiel 8:14; Greek Adonis carrying his Semitic loan-name adōn, "lord"); (2) the Ugaritic Baal narrative (c. 1400–1200 BCE), the one unambiguous pre-Christian die-and-return text, possibly part of a wider Northwest Semitic regional pattern on thinner evidence; and (3) the Egyptian Osiris complex — the earliest attested of all as narrative content (c. 2350 BCE, osiris-cycle-earliest-attestation; Dumuzi's bare name appears earlier still in mid-third-millennium god-lists, but name-attestation is not motif-attestation) — with no documented link to Mesopotamia and a "rising" that means kingship over the dead rather than return to life: best read as independent invention. A single verdict over all of that would be a false unity; the per-strand verdicts are the real finding.

What the three verdicts jointly imply — the braided river. One vault, one method, three different answers: the Near Eastern flood was borrowed, the sky father was inherited, the dying god is partly borrowed, partly inherited within single lineages, partly reinvented — and partly a scholarly construct. Two of the three mechanisms — descent and contact — are demonstrated outright, at sound-law and textual-dependence grade; the third, convergence, is the best available reading of specific cases (the Osiris strand, the global flood distribution, the non-IE sky-father concepts) but is nowhere yet proven at that grade — each such case is an open register question (Q5, Q7, Q11), and a pre-contact New World test bed is the cleanest route to closing it (Q27). That asymmetric result is still decisive, because it rules out both extreme stories the question invites. The romantic extreme — all religions remember one primordial revelation or one ur-myth — finds no support: descent is only provable as far back as language families reach (for Indo-European, roughly the fourth millennium BCE), and every deeper reconstruction in the vault (a Paleolithic shamanic baseline, Witzel's proto-mythologies, deep-time oral memory) sits at low confidence or is explicitly speculative (shamanism-animism-baseline-religion-hypothesis, australian-aboriginal-dreaming, aboriginal-oral-tradition-coastal-flooding-time-depth-claim). The reductive extreme — the parallels are all coincidence, or all borrowing — is equally dead: the sky-father cognates are not coincidence, and the Genesis flood is not independent invention. The truth has the shape of a braided river: humanity's religious streams split (descent), flow into each other (contact), and carve parallel channels in similar terrain (convergence), and for each well-documented story we can now say which. Where the evidence ends — before writing, beyond the reach of sound laws — the honest verdict is "unresolved," and this vault says so rather than extending any of the three mechanisms on faith.

None of this bears on whether any tradition's truth-claims are true. The analysis is of the human, historical side — texts, dates, transmission routes — and the traditions' own self-understandings are recorded alongside it as emic data, not as targets.