Who Made the End: Societies, Scribes, and the Invention of the World's Last Day

A survey of why some societies invented an end of the world and others did not, from Egyptian maintenance and Mesopotamian lament through Zoroastrian direction, Jewish judgement under empire, Hindu arithmetic, Norse battle, and the catastrophes that were actually delivered. With forty images and a working bibliography.

Ask where the end of the world comes from and you will usually be handed a genealogy. Zoroaster to Daniel, Daniel to John of Patmos, John to Joachim of Fiore, Joachim to William Miller, Miller to the pamphlet in your letterbox. The line is tidy, it is taught, and it is roughly half true. What it hides is the more interesting fact: most societies that have left us records did not expect the world to end at all, and the ones that did expect it wanted very different things from the expectation.

This is an attempt to describe those differences without flattening them, and to be specific about what produced them. Not the human condition, not anxiety in the abstract. Particular states, particular defeats, particular bureaucracies of time.

The societies that did not need an end

Pharaonic Egypt ran for roughly three thousand years on a cosmology with no terminal date. The world was made once, it is threatened continually, and it is maintained by ritual and kingship against a chaos that never wins and never disappears. Maat, the order, is not a destination. It is a daily obligation.

The reckoning in Egyptian religion is real, precise, and individual. The heart is weighed against the feather in the Hall of Two Truths; Anubis adjusts the balance; Thoth records; Ammit waits for the failures. Everything an apocalypse promises is present here except scale. One person is judged. The cosmos is not.

There are Egyptian texts that sound end-adjacent, and they repay care rather than excitement. The Prophecy of Neferti and the Admonitions of Ipuwer describe the land inverted, the poor rich, the Nile dry, the country unrecognisable. These are almost certainly not prophecies of a final catastrophe. They belong to a genre of political lament, and in Neferti's case a piece of retrospective propaganda: chaos is described so that a named king can be shown ending it. The disorder is a rhetorical setup. It is what strong rule abolishes, not what history culminates in.

Mesopotamia is closer to the border and still on the far side of it. The flood in Tablet XI of Gilgamesh is a past event, survived. Atrahasis gives us the same story with the divine committee minutes attached: the gods are irritated by human noise, they resolve on destruction, one god leaks the plan, a boat is built, and afterwards the gods regret the decision and install limits so it does not recur. The lesson is that annihilation was tried and found to be bad policy.

Mesopotamian divination was enormous, systematic, and almost entirely short-range. Omen series ran to thousands of entries; a professional class read livers, birth defects, planetary positions and dreams to advise on this harvest, this campaign, this king's health. There was a well-developed sense that a dynasty could be doomed and a city could be abandoned by its god, and the great city laments mourn exactly that. What the tradition lacks is a last day. The world is chronically endangered and structurally permanent.

The point is worth stating plainly because it undercuts the assumption that apocalyptic thinking is what humans do under stress. Southern Mesopotamia had floods, plagues, invasions, salinised fields and collapsing states, and it produced a literature of maintenance and lament rather than a literature of the end. Stress is not sufficient. Something else has to be added.

What Zoroaster added

The thing that gets added, on the standard account, is a direction.

Norman Cohn's Cosmos, Chaos and the World to Come (1993, second edition 2001) put the case at book length. Ancient Near Eastern cosmology, he argued, was static: order under permanent threat, held by ritual, going nowhere. Somewhere between roughly 1500 and 1200 BCE, his dating and it is contested, the Iranian prophet Zoroaster reworked the inherited combat myth into something structurally new. Conflict was no longer a cycle to be endured but a process with an outcome. Good would win, once, finally, and the world afterwards would be free of the disorder that currently defines it. Cohn's phrase for the endpoint was cosmos without chaos.

If that is right, the consequences are large. A future that differs in kind from the present becomes thinkable. History acquires a plot. Suffering acquires a term. And crucially, the individual acquires a stake in an outcome that arrives after their death, which is what makes resurrection a necessary doctrine rather than a decorative one.

The problems with the thesis are real and should not be waved through. Zoroaster's dates range across a millennium depending on which scholar you read and which criteria they weight: linguistic archaism in the Gathas, the Greek tradition placing him absurdly early, or the later Zoroastrian chronology. The developed eschatology, with the saviour Saoshyant, the resurrection, the molten-metal ordeal and the frashokereti renewal, is documented in detail mainly in Pahlavi texts such as the Bundahishn, compiled in the ninth century CE. That is well over a thousand years after any plausible Zoroaster, and after extended contact with Judaism, Christianity and Islam. Reading the Bundahishn back into the Gathas is a real methodological hazard, and the direction of influence between Iranian and Jewish eschatology has been argued both ways for a century.

What survives the objections is narrower and still substantial. Iranian religion holds, earlier and more thoroughly than its neighbours, that cosmic time is going somewhere and that the destination is a judgement. Whether that idea reached Jerusalem from Persia, or developed in parallel under the same imperial pressures, the Persian period is when Jewish texts start doing something new.

Defeat, and the scribes who wrote from inside it

Apocalyptic literature in the Jewish tradition does not emerge from the confident monarchy. It emerges from its wreckage.

The pre-exilic prophets threaten the Day of YHWH, and the threat is aimed inward: Amos tells his audience that the day they are looking forward to will be darkness and not light. This is covenantal indictment, not cosmology. The addressee is a functioning state that is behaving badly and can stop.

Then the state stops existing. Jerusalem falls in 587 BCE, the temple is destroyed, the elite is deported. What returns from Babylon is a community without political sovereignty, holding texts that promise it. The gap between promise and circumstance is the engine.

Daniel is the clearest case, and its dating is the key to reading it. The visions run through a sequence of empires and land, with sudden granularity, on the reign of Antiochus IV Epiphanes (the desecration of the temple in 167 BCE, the suppression of the cult, the resistance). The historical detail is accurate up to that point and then goes wrong: Daniel predicts Antiochus's death in a manner and place that did not match. This is why the scholarly consensus dates the final composition to the mid-160s BCE, during the crisis, and reads the earlier "predictions" as history written backwards to lend authority to the one prediction that was genuinely future.

Read that way, Daniel is not a fortune-telling document. It is a work of political theology produced by literate men under a persecuting empire, arguing that the empire is the fourth and last of a series, that its arrogance is scheduled, and that sovereignty is about to transfer. The beasts are states. The chronology is the argument.

Two innovations in Daniel become permanent features of the genre. First, resurrection: chapter 12 promises that many who sleep in the dust will awake, some to everlasting life and some to shame. This is what makes martyrdom coherent. If the faithful are being killed precisely for faithfulness, the accounting has to happen somewhere. Second, the numbers. 1,290 days. 1,335 days. A time, times and half a time. Once the end has been given arithmetic, the arithmetic can be recalculated forever, by anyone, in any century. Daniel's numbers were still being run in Massachusetts in 1843.

The pattern holds outside Judaea. The Egyptian Potter's Prophecy and the Oracle of Hystaspes come out of populations under Greek and then Roman rule. Pseudo-Methodius is written under early Islamic rule in Syria. The genre is a scribal response to living inside somebody else's empire: literate enough to write, powerless enough to need to.

Where the end is arithmetic instead of hope

South and East Asian traditions solved the problem of an unbearable present by a different route. Rather than promising that history will end, they made it enormous.

The yuga scheme gives four ages of declining virtue, totalling 4,320,000 years, and locates us in the last and worst of them, the Kali Yuga, which alone runs 432,000 years. On the conventional reckoning it began in 3102 BCE. Do the subtraction: roughly five thousand years elapsed, well over four hundred thousand remaining.

That number changes the emotional content of the doctrine completely. An apocalypse you can be told is imminent produces movements; a decline with four hundred millennia left to run produces resignation, or ethical adjustment, or an argument about the arithmetic. Kalki will come, the cycle will restart, and none of it will happen within any conceivable political horizon.

The chronology itself has a history worth knowing. The Puranic material divides the generations from Manu to the Bharata war into ages of a scale that works out to something like a few hundred years each, plausible dynastic time. The astronomical yuga of hundreds of thousands of years is a different construction, tied to planetary-conjunction epochs and elaborated by mathematicians. Āryabhaṭa set his Kaliyuga epoch at dawn on 18 February 3102 BCE, at the intersection of the equator and the Ujjain meridian, and adjusted the scheme's internal divisions for computational convenience. That is not prophecy. It is a calendrical zero point chosen so that planetary tables would resolve cleanly. The two systems, genealogical and astronomical, got conflated afterwards, and the conflation is now what most people mean by the Kali Yuga.

Buddhist decline works on a shorter and more anxious scale, and it produced consequences. The doctrine of mappō in Japan held that the Dharma passes through phases of decreasing efficacy until, in the final age, the teaching survives as text but practice can no longer reach awakening. Japanese calculation set the onset at 1052 CE. The dating landed in a period of visible institutional disorder (factional violence, armed monasteries, the erosion of central authority) and educated Japanese read the disorder as confirmation.

What followed was not passivity. It was a redirection of religious effort on a massive scale. If self-powered practice cannot work in a degenerate age, salvation has to come from elsewhere, and Pure Land devotion supplied it: rebirth in Amida's western paradise through reliance on Amida's vow rather than one's own capacity. Byōdō-in was built in 1052 itself. Hell scrolls were painted as instruments of persuasion. An eschatological calculation reorganised the devotional life of a country.

The Kālacakra system, which times a war against barbarian forces from Shambhala with genuine precision, then performs a characteristic move: the outer war is read as an inner one, and the enemy armies as afflictions. The eschatology survives while its literal referent is dissolved.

Mesoamerica is the case most thoroughly misrepresented in popular writing, and the correction is simple. The Long Count is a positional calendar. It rolled over on 21 December 2012 in the same sense that a mileage counter rolls from 99,999. Maya inscriptions project dates far past that boundary: Palenque's texts run into the fifth millennium CE, and Coba records cycles of absurd magnitude. The Dresden Codex is dense with Venus tables and eclipse arithmetic and contains nothing resembling a terminal date.

The genuine Mesoamerican world-age doctrine is not calendrical panic but the Aztec account of five suns, four of which have already ended, each by a distinct catastrophe, with the fifth to end in earthquake. It is a claim about the fragility of the present order and about what the present order owes in return, which is why the fifth sun's maintenance was framed as requiring blood. The catastrophe is structural, not scheduled.

The end as a battle report

Norse eschatology sits awkwardly in every typology, which is the most interesting thing about it.

Ragnarök is the most narratively detailed end in world mythology. We have the sequence, the participants, the pairings, the specific wounds. Fimbulwinter, three winters without summer. The wolf takes the sun. The serpent thrashes the sea. Naglfar sails, built from dead men's nails. Heimdall blows the horn. Odin falls to Fenrir. Thor kills the serpent and walks nine steps before the venom stops him. Surtr burns everything. Then the earth rises green.

Nothing else has this density. The Egyptian material tells you what happens to you; the Qur'an tells you what happens to everyone; the Völuspá tells you the name of the field and the colour of the rooster.

And it does not judge. No one is weighed, no verdict is issued, no righteous remnant is selected. The gods fight and lose. That single feature disqualifies Ragnarök from the judgement family it is usually filed beside.

The source problem is severe. The Völuspá survives in the Codex Regius, copied around 1270, and in Hauksbók from the early fourteenth century. Snorri Sturluson's systematising account was written around 1220, in Christian Iceland, roughly two centuries after conversion. Snorri is our best informant and an interested one.

The manuscripts themselves record the discomfort. The Codex Regius text ends with the dragon Níðhöggr rising over the corpse-strewn plain, and stops. Hauksbók adds a final stanza about a hall fairer than the sun, thatched with gold at Gimlé, where good men dwell, a heaven in a tradition that has no obvious use for one, in a metre that does not sit right. A copyist wanted a verdict and supplied one. The older text refuses.

The archaeology supports a pre-Christian core without settling the ending. The Rök stone, around 800 CE, carries the longest known runic inscription and language that may involve a wolf, the sun and a reckoning. The Ledberg stone shows a figure with a beast at his foot. The Gosforth Cross, around 950 in Cumbria, carves bound Loki and a horn-blower on the same monument as a crucifixion, placed alongside, not in replacement.

Then the detail that makes the renewal unreadable. The surviving gods find the old golden game pieces in the grass and resume the game. They do not make new ones. Baldr returns, and so does Höðr, who killed him. The fratricide comes back with his victim. Whatever this is, it is not resolution, and the Völuspá declines to say which it is: the seeress says she sees further and does not say what.

The end that actually happened

The clearest case for reading eschatology as social history is one where the catastrophe was not predicted but delivered.

In 1856, a girl of about fifteen named Nongqawuse, living near the Gxarha river mouth in the eastern Cape, reported meeting strangers who told her the dead would rise. The condition was that the Xhosa must destroy their cattle and stop cultivating. Her uncle Mhlakaza carried the message. Over the following year an enormous number of cattle were killed and fields left unplanted. The famine that followed killed tens of thousands and broke Xhosa independence, and the Cape administration used the collapse to take land and labour.

The colonial reading was mass suicide by superstition, and it persisted because it was useful. J. B. Peires's The Dead Will Arise (1989) made the case that has framed the scholarship since: the prophecies landed in a society already under compound assault. Nine frontier wars. Progressive dispossession. And, immediately before the movement, bovine lung sickness, a European-introduced epidemic tearing through herds that were the society's wealth, kinship currency and ritual centre. Cattle were dying anyway, filthily and beyond use. A prophecy that instructed believers to kill cattle in order to bring back a purified world was answering a specific catastrophe with a logic internal to Xhosa thought about pollution and renewal.

Peires is explicit that no conspiracy is needed on either side to explain the movement. The revisionist literature has pushed further and in productive directions: Helen Bradford and Msokoli Qotole have argued for reading the affair through gender and through Xhosa-language sources, and for granting Nongqawuse considerably more agency than a purely materialist account allows.

What the case establishes is that millenarian action can be strategic without being correct. The choice was not irrational relative to the information available; it was catastrophic. Those are different findings, and colonial historiography spent a century collapsing them.

Comparable dynamics run through the Hopi prophecy material, where the years around 1900 saw Oraibi split over how to answer American schooling, allotment and missionary pressure, with prophecy supplying the vocabulary for a genuine political disagreement. They run through Tupi-Guarani migrations toward a Land Without Evil, recorded in the sixteenth century by observers whose framing needs constant correction. And they run through the Dogon case, where an ethnographer's account hardened into a global myth about ancient astronomy that Dogon informants would not have recognised.

The comparative literature has been arguing about the mechanism for decades. Relative deprivation, the idea that millenarianism erupts from felt loss of status, security or wealth, has been the dominant explanation and has taken sustained criticism for being loosely predictive. Deprivation is common; millenarian movements are not. The theory does not explain why the same pressure produces a prophet in one place, a riot in another, and nothing at all in a third. It also fails a straightforward empirical test: movements have not reliably appeared where deprivation was most acute.

The variable that seems to matter more is availability. A society needs an existing framework that makes a total renewal thinkable, someone positioned to voice it, and authority structures too weak to suppress it. That is why the Xhosa case needed lung sickness and a Xhosa cosmology of pollution and ancestral return and a frontier administration that benefited from the outcome.

Arithmetic in a print culture

The nineteenth-century American case shows what happens when Daniel's numbers meet cheap printing and a literate public.

William Miller was a New York farmer and Baptist lay preacher who worked through the prophetic numbers and concluded that Christ would return around 1843–44. His method was not eccentric by the standards of its time; day-for-year interpretation of Daniel's periods was widely practised. What made Miller different was distribution. Joshua Himes, a publicist of real ability, produced newspapers, tracts, camp meetings and, decisively, large lithographed charts that rendered the entire prophetic scheme as a visual timetable.

The chart is the artefact that matters. It converts an interpretive argument into an object that can be hung on a wall, pointed at, and checked. It makes the end of the world look like a railway schedule.

The date passed. Then a revised date, 22 October 1844, passed. The event is called the Great Disappointment, and the mockery was immediate and durable: the enduring image of a believer leaping off a roof in homemade wings is a hostile invention that outlived the movement's own account of itself.

What is analytically interesting is the aftermath rather than the failure. The movement did not simply evaporate. Some participants reinterpreted the date as referring to an event in heaven rather than on earth, and out of that reinterpretation came Seventh-day Adventism, now a global denomination of many millions. Disconfirmation, handled institutionally, can be more generative than fulfilment. This is the pattern Leon Festinger and colleagues later theorised as cognitive dissonance, using a small millenarian group as their case, and the theory has been criticised since for overgeneralising from it, but the basic observation holds up: prophetic failure is routinely survived, and sometimes rewarded.

The end without a god

The twentieth century produced the first eschatology that requires no cosmology at all.

After July 1945 the end of the world became a description of available capability. It needed no prophet, no calculation from Daniel, no degenerate age. It needed warheads and a decision procedure. The novelty is not the fear; the novelty is that the mechanism is public, engineered, and in principle subject to policy.

Castle Bravo in 1954 supplied the appropriately humbling detail. The yield came in at roughly two and a half times prediction, contaminating an area far beyond the planned zone. The people who built the device did not fully understand the device. That is a specifically modern eschatological condition: the danger is not that a god is angry but that the calculation is wrong.

The Doomsday Clock is the honest emblem of this shift, and it began as a magazine cover illustration. Its hands are set by a committee, by argument, from evidence, which makes it the first apocalyptic instrument that openly admits to being an interpretation. Every previous tradition claimed its timing came from elsewhere.

What the comparison shows

Four things survive this survey. First, the end of the world is not a human universal. Egypt and Mesopotamia had all the raw material of catastrophe, judgement, divine anger and cosmic threat, and did not assemble it into a terminal event. The world's end is a specific construction with a history, not a default of the species.

Where the world does end, the shape of the ending tracks the situation of the people describing it. Communities under imperial rule with literate elites produce judgement and transfer of sovereignty. Communities with vast calendrical schemes produce decline without urgency. Communities facing an actual delivered catastrophe produce prophecy that instructs action. Ragnarök, from a society without an imperial centre or a scribal bureaucracy until Christianity supplied one, produces a battle report with no verdict in it.

The arithmetic is not decoration. Daniel's day-counts, Āryabhaṭa's epoch, the 1052 dating of mappō, the 1843 chart: in each case a number turned a general conviction into something that could be checked, argued about, and acted on. Give an end a date and you have created a social fact.

And every one of these traditions reaches us through interested hands. Snorri writing two centuries after conversion. A Hauksbók copyist adding a heaven. Colonial officials explaining Xhosa famine as superstition. Griaule making the Dogon into astronomers. Popular authors making the Maya into doomsayers. The sources are not windows. They are arguments, and reading them as arguments is most of the work.

---

Sources and further reading

General and comparative

• Norman Cohn, Cosmos, Chaos and the World to Come: The Ancient Roots of Apocalyptic Faith, Yale University Press, 1993; 2nd edn 2001. The Zoroastrian-origin thesis, with the expanded chapter on Jews, Zoroastrians and Christians added in response to criticism of the first edition.
• Norman Cohn, The Pursuit of the Millennium, 1957. Still the standard account of medieval European millenarian movements.
• Encyclopedia of Religion, "Millenarianism: An Overview" : the review of relative-deprivation and contact theories and their critics summarised above.
• D. Aberle, "A Note on Relative Deprivation Theory as Applied to Millenarian and Other Cult Movements," in S. Thrupp (ed.), Millennial Dreams in Action, Mouton, 1962.
• Michael Barkun, Disaster and the Millennium, Yale University Press, 1974 : includes the standard critique of deprivation reductionism.
• Lamont Lindstrom, Cargo Cult: Strange Stories of Desire from Melanesia and Beyond, University of Hawai'i Press, 1993 : on the analytic and ethical problems of the "cargo cult" category itself.

Ancient Near East and Iran

• The Epic of Gilgamesh, Tablet XI (British Museum K.3375); Atrahasis.
• Bundahishn (Pahlavi, 9th c. CE) : the developed Zoroastrian eschatology, with the dating caveat noted above.
• Prophecy of Neferti; Admonitions of Ipuwer : political lament rather than eschatology.

Jewish and Christian apocalyptic

• Daniel, esp. chs. 7–12. Mid-160s BCE composition is the mainstream critical position, on the grounds of the accurate-then-inaccurate Antiochus material.
• Revelation; 1 Enoch; Pseudo-Methodius, Revelations; the Oracle of Hystaspes; the Egyptian Potter's Prophecy.

South and East Asia

• Mahābhārata, Śānti Parva, on the yuga scheme; Āryabhaṭa, Āryabhaṭīya, for the 3102 BCE epoch and its computational purpose.
• Genshin, Ōjōyōshū (985 CE), on rebirth in Amida's paradise; the mappō dating of 1052 and the founding of Byōdō-in that same year.
• The Kālacakra tantra literature on the Shambhala war and its internalising reading.

Mesoamerica

• The Dresden Codex (Sächsische Landesbibliothek) : Venus and eclipse tables, no terminal date.
• Palenque and Coba inscriptions projecting dates long past the 2012 Long Count rollover.
• The Aztec five-suns account, with the Sun Stone (Museo Nacional de Antropología) as its monumental statement.

Norse

• Völuspá, Codex Regius (GKS 2365 4to, c. 1270) and Hauksbók (AM 371 4to, early 14th c.); note the Gimlé stanza present only in Hauksbók.
• Snorri Sturluson, Prose Edda, Gylfaginning 51–53, c. 1220.
• Rök runestone (Ög 136); Ledberg stone (Ög 181); Gosforth Cross, Cumbria, c. 950.

Southern Africa

• J. B. Peires, The Dead Will Arise: Nongqawuse and the Great Xhosa Cattle-Killing of 1856–7, Ravan Press, 1989; 2nd edn Jonathan Ball, 2003.
• J. B. Peires, "The Central Beliefs of the Xhosa Cattle-Killing," Journal of African History 28 (1987), 43–63.
• Helen Bradford and Msokoli Qotole, "Ingxoxo enkulu ngoNongqawuse (A Great Debate about Nongqawuse's Era)," Kronos (2008).
• Adam Ashforth, "The Xhosa Cattle Killing and the Politics of Memory," Sociological Forum 6:3 (1991), 581–592.

Modern

• Leon Festinger, Henry Riecken and Stanley Schachter, When Prophecy Fails, University of Minnesota Press, 1956 : and the subsequent methodological criticism of it.
• The Bulletin of the Atomic Scientists for the Doomsday Clock and its setting procedure.

What remains unsettled

• Zoroaster's date, and therefore whether Iranian eschatology could have influenced Jewish apocalyptic or developed alongside it. The Pahlavi evidence is too late to settle it and the Gathas are too sparse.
• Whether the Völuspá's green-earth stanzas are original. The metre is irregular and the argument has run for over a century.
• Whether Líf and Lífthrasir are an old motif or a late rationalisation supplying human survivors.
• How much agency to assign Nongqawuse herself, as against the material conditions Peires emphasises : the live question in the current literature.
• Why deprivation produces a prophet in one place and nothing in another. No theory currently predicts this well.

The taxonomy it draws is treated under What Is Eschatology.