The week ended at an empty tomb. And the question that presses itself forward is whether it also left a trace outside the texts — whether the earth itself remembered anything of the day on which, according to Matthew, "the earth shook and the rocks were split" (Matthew 27:51). This chapter pursues that question, and it is a ring: it does not carry the case, it colours it. But a ring, precisely, must be weighed with the same rigour as a pillar, and at one point that rigour will work against me. For there is one trace that fits surprisingly well — and there are two "witnesses" who would say exactly what I want to hear, and whom I must nevertheless send away.
Begin with the trace in the ground, for that one is real. In the bed of the Dead Sea lie, layer upon layer, the deposits of two thousand years, and earthquakes leave their signature in them: a disturbed, churned-up layer between the quiet annual laminae. In the section near Ein Gedi lies such a quake layer from the early first century. And the remarkable thing is how it was dated. In 2012 Williams, Schwab and Brauer counted the annual layers back from one uncontested anchor point — the well-known earthquake of 31 BC, which Josephus describes — and they noted expressly that they made no assumption about the presumed date of the layer: "a date was assigned on the basis of counting the annual layers alone". Their result: AD 31 or 32 (±5), depending on the count. The authors give the window as 26 to 36, largely the term of office of Pilate; without a year zero it shifts by one year (Williams, Schwab & Brauer, 2012).
Dossierthe window, and the base rate
The measurement. The window 26–36 comes from the free counting of the annual layers (Williams, Schwab & Brauer, 2012). An anchor-free radiocarbon dating (Kagan et al., 2011) gives a wider window, AD 12–91 at 1σ, which contains it; Kagan then links the layer to the catalogue earthquake of 33. Neither method yields a year — only a window. The oft-heard dating "AD 33" is something else: it is a fitted catalogue date — Migowski and colleagues (2004) shifted their floating chronology so that the layer coincided with an already dated catalogue earthquake, and that catalogue date comes in turn from crucifixion chronologists with Matthew as their only source. Whoever says "33" cites the fit; whoever says "26–36" cites the measurement.
The base rate — obligatory, and the opponent supplies it. Earthquakes are frequent in the Dead Sea rift. In the late Second Temple period the series counts roughly one correlated quake per half-century (Migowski et al., 2004); the chance that an arbitrary eleven-year window (26–36) contains at least one is then roughly one in four to five [CALCULATED]. The seismite is thus corroboration at window level — consistent with Wednesday 25 April 31, but discriminating for nothing within that window. Whoever makes more of it loses by this book's own house rule. And there is a second limit, from the other side: the layer says nothing about what was felt in Jerusalem. A seismite in the Dead Sea rift only registers from about M ≥ 5.5, and that is a threshold for laying down the layer, not for the perceptibility of a shock a hundred kilometres away (Zohar et al., 2016; Williams, Schwab & Brauer, 2012). The layer therefore carries an event within the window; it does not carry a tremor anyone in the city noticed. That is no weakening of this chapter but its sharpening — the chapter did not claim the latter in the first place. [DOCUMENTED]
There is a fine irony in the way this trace has been handled, and it is worth recording because it repeats in miniature the pattern of this whole book. The geologists measured the year 31 — and then gave it back. Immediately after their own dating, Williams, Schwab and Brauer write: "naturally it is not likely that an earthquake of the crucifixion could have taken place in AD 31" — with a reference to the list of candidate years drawn up by the crucifixion chronologists, in which AD 31 did not appear. The geologists found 31 and returned the year to the chronologists, who had never examined it. Each discipline outsources the question to the other, and so the year that no one weighed disappears. That is not evidence that AD 31 is right; it is evidence that AD 31 has never been tested. The trace opens the door; it pushes no one through.
Dossierhow "33" got into the catalogue, and how such entries arise
The fitted catalogue date has a family tree, and it is on record. Williams says in his own earthquake catalogue where the label comes from: "This earthquake is known in most catalogs as the 33 AD earthquake, though some designate it as a 30 AD event. Both labels refer to the same textual passage in the Gospel of Matthew [...] The difference in dates reflects uncertainty surrounding the year of Jesus' death." Karcz has written out the chain: New Testament and Talmud → Arvanitakis (1903/04) → Willis (1928) → "other cataloguers followed suit" → all later catalogues (Karcz, 2004). The church's dating supplied the year to the catalogue, the catalogue gave it to the geologists, and the geologists gave it back to the theologians as independent confirmation. That is a circle, and it is documented. Two data show that the mechanism really works this way. The catalogue entry "earthquake AD 48" goes back to a misread verse, Acts 8:24 instead of 16:26, where a quake stands in Macedonia and not in Jerusalem; Williams points that out. And the counts that gave "33" were tuned to it: Migowski "used varve counts in the masked intervals to match her varve-counted year to historically documented earthquakes", and Ken-Tor's own, untuned dating of the same layer was AD 124 ± 188, only afterwards linked to "33" (Ken-Tor et al., 2001). Whoever says "33" thus cites not only a fit; he cites Matthew, by way of two intermediate stations. And the same measure holds in reverse: Williams' midpoint too is a count with a margin, not a year the layer itself carries. [DOCUMENTED — Williams, Schwab & Brauer (2012) and Williams' catalogue; Karcz (2004); Ken-Tor (2001) and Migowski via Williams]
And honesty demands the other side at once. Only Matthew mentions the quake, and he mentions it together with raised saints coming out of the tombs — an image that scholars, not without reason, read as a theological motif, a topos of nature responding to great events. That explains the literary form. But it does not explain the layer in the lake: there is a physically dated quake in the window, measured by researchers with no stake whatever in the date of the crucifixion. The strongest formulation is therefore doubly modest: the only physical quake trace from the time of Pilate falls in a window that largely covers his term of office and whose arithmetic midpoint in Williams is 31 — no more, and that is enough. The link to Matthew's quake remains a construal.
Expertthe brakes on the layer, in full
The margin of five years is no detail but the whole measurement. Of the 62 counted annual layers, 58 percent received the lowest quality rating, "questionable"; the same core gave another year with another counter, Migowski's 33, and that this count was tuned to the catalogue weakens it without removing the uncertainty. The authors themselves leave three possibilities open, of which the third does not flatter this case: that Matthew's quake is allegory and the seismite comes from an unrecorded quake. And the echo of 2012 ("Quake Reveals Day of Jesus' Crucifixion") promised what the article does not say; Williams himself, when asked: "we added no new information about the date of the crucifixion". This chapter says no more than he does. [DOCUMENTED — Williams, Schwab & Brauer (2012); the press statement via the 2012 reporting]
One line was pursued until it ran dead, and it belongs here, because it could have given the seismite its season and with it a test on Passover. An annual layer consists of a winter half of clay and a summer half of aragonite; if a quake writes its stroke at the moment of the shock, the position of the seismite within the annual layer would betray the season, and Passover falls in spring. The method even seemed calibratable, for of the anchor quake of 31 BC Josephus gives the season: "at the beginning of spring". The kill criterion was set in advance: if the season cannot be read from the layer, the line lapses. It lapses on three points. The anchor layer is nine centimetres thick and disturbs several years at once, so that the calibration is not even feasible; Williams reports the position nowhere; and the deepest objection is physical: the seismite arises because the upper laminae liquefy and settle again, and that process destroys precisely the seasonal information it should have preserved. The criterion has been hit. The line stands here as a closed door, with the question that ought to be put to the authors, and nothing is built on it. [kill criterion hit; line lapsed]
Now the two witnesses I must send away, and this is the purest test in this whole chapter — for they would seem to serve my case, and precisely for that reason I must be strict. For centuries apologists have paraded two pagan writers who are said to have recorded the darkness at the crucifixion: Thallus and Phlegon. If they were real, independent witnesses, I would have a gift. But they are not, and I say so before a critic does.
Of Thallus not a single word survives. What is physically at hand is a Byzantine chronicle from around 800 which says that Julius Africanus said that Thallus said something — and even there stands no quotation, only Africanus' summary, with the remark that Thallus "wrongly" called the darkness a solar eclipse. The link to the crucifixion sits in Africanus' demonstrative pronoun — "this darkness" — not in Thallus; the name Jesus does not occur in it; and the dating of Thallus himself has a margin of three hundred years. A witness three hundred years wide, whose words do not survive, is no witness. And even a Christian historian saw that there was nothing astronomical to see: Orosius (around 417) denies expressly that the darkness was an ordinary solar eclipse — as indeed it could not have been, for Pesach falls at full moon, when a solar eclipse is impossible (Orosius, Historiae adversus paganos VII.4).
And Phlegon is worse — for he does not even point the right way. Phlegon names no Jesus, no Judea, no Passover; he names an earthquake in Bithynia, well over a thousand kilometres from Jerusalem, and a solar eclipse. And recompute his own dating: it falls in the fourth year of the two hundred and second Olympiad — the year 32/33. Phlegon supports AD 33, not AD 31. Whoever defends a crucifixion in 31 and cites Phlegon as confirmation would be undercutting his own thesis; he would have to shift the witness two years, and then he has no right left to call him a witness. That is measuring with two measures, and I do not do it. I send Phlegon away — not despite the fact that he names a crucifixion year, but in part because it is the wrong one.
Expertthe astronomical vice, which holds for everyone
Beneath all these witnesses lies one physical impossibility that closes off the whole eclipse explanation, for every year and for every camp. A solar eclipse requires a new moon — sun and moon in conjunction. The Passover falls by definition at full moon — sun and moon opposite each other. A solar eclipse at the Passover is therefore physically impossible, and maximal totality lasts seven minutes, not three hours. The darkness of the ninth hour, whatever it was, was no solar eclipse — and whoever brings in Thallus or Phlegon to make it an eclipse is bringing in an explanation that astronomy forbids. That holds against the apologist who wants to support AD 31 exactly as hard as against whoever wants to support AD 33.
Dossierand no volcano: what the ice cores say
The eclipse is astronomically ruled out; there remains the question whether a volcanic veil could have quenched the sun for three hours. That door too swings shut, and now with physical data. The ice-core reconstruction eVolv2k (version 4) has been consulted at first hand for the window. Around the year 31 the cores carry precisely one sulphur signal — VSSI 0.95 ± 0.71 Tg sulphur — and it points the wrong way: unipolar-Antarctic (Greenland 0.0), originating from around 45° southern latitude. A weak, uncertain eruption in the southern hemisphere cannot darken the sky above Jerusalem. More broadly, the whole period from roughly 40 BC to AD 160 lies as the "Roman Quiet Period": volcanic sulphur output there runs 55 to 70 percent below the long-term average. And from close by there is nothing: the Harrat Ash Shaam volcanic field has no historical eruptions — the youngest dates from about 2670 BC and was an effusive, basaltic flow, not an ash-shedding eruption — and no first-century tephra layer has been reported in Israel or Judea, nor a named source volcano anywhere in the window (eVolv2k: "N/A"). A tephra layer not found is no demonstrated absence; but a stratospheric veil dims for months in any case, not three sharply bounded hours, and ashfall darkens only beneath the plume, close to the crater. On each of these points the volcanic explanation fails. [DOCUMENTED — eVolv2k v4 consulted at first hand] Together with the astronomical vice this means: the midday darkness cannot be explained away as an ordinary natural phenomenon — not as eclipse, not as volcano, not as khamsin (a khamsin sky is diffusely luminous, not dark).
What then remains of the darkness, if neither the eclipse nor the volcano carries it? The most honest line is not physical but scriptural — and the evangelists lay it down so themselves. The midday darkness stands in the language of prophetic judgement, and one prophecy sounds unmistakably within it. "I will make the sun go down at noon and darken the land in broad daylight", says Amos — and in the same breath: "I will turn your feasts into mourning" (Amos 8:9-10). It is a judgement over a feast-keeping people, and it does not come alone: two verses earlier the land trembles at it (Amos 8:8), and the book opens with the dating "two years before the earthquake" (Amos 1:1). Darkness at noon and a trembling earth, together, over a people at feast — that is precisely the conjunction Matthew sets down at the ninth hour (Matt. 27:45, 51). I give that reading its mark without ado: CONSTRUED, not CALCULATED — a typological echo, not a measurement. And precisely as a construal it is worth more than a fabricated physics: it places the darkness in the category the text itself already sets it in — that of the sign.
Excursushow a "proof" accretes, and the mirror it holds up
The most instructive thing about Thallus and Phlegon is not that they fail, but how they were made into "witnesses" — sentence by sentence, traceably. Phlegon (around 130) reports an eclipse and a quake in Bithynia. Africanus (around 221) adds that it was "at full moon" and lasted "until the ninth hour" — two details that come straight out of Mark 15:33, and which Origen himself denies that Phlegon said. Eusebius (around 320) adds that "all this agrees with what happened at the Saviour's passion". Malalas recalculates it to "the eighteenth year of Tiberius". And Michael the Syrian, in the twelfth century, has Phlegon say that the dead rose, entered Jerusalem and cursed the Jews — and that it was a Friday. Each link adds precisely that which the argument needs. That is the signature of apologetic accretion, and the table belongs in this book not as an attack on others but as a mirror: test every chain of sources of your own against this pattern. Modern source criticism draws from it the only honest conclusion: Thallus nowhere named Jesus and should be struck from the witness lists (Carrier, 2012, p. 191). And that measure holds in all directions — towards mine as well: test every chain of sources of your own against this same pattern of accretion.
And yet beneath all this rubble lies one thing that shines, and it is better than anything the apologists have made of it. That same Africanus, around the year 221, rejects the solar-eclipse explanation on purely astronomical grounds — though it would have suited him excellently, for it would have furnished him a pagan witness. He knows that the Passover falls on the fourteenth of the lunar month, that this means the full moon, and that an eclipse requires the new moon. And he refuses the argument: "how could one assume an eclipse when the moon stands almost diametrically opposite the sun?" He sacrifices evidence congenial to him because astronomy forbids it. That is exactly the discipline this whole book demands — carried out eighteen centuries earlier. Africanus is not a witness. He is an example.
Where does the inquiry stand, then, at the end of this ring? The earth has left one trace that fits — a quake layer in the window, its arithmetic midpoint on the year 31, measured by people who never even considered that year. More it does not carry: a base rate of one in four to five keeps it in its place, as colour, not as pillar. And the two literary "witnesses" who had seemed most likely to help, I have sent away, because on closer inspection they are phantoms — and one of them even points to the opponent's year. That sending away is not a weakness of this chapter; it is its integrity. A case that rejects its own weak witnesses earns the trust that a case which seizes everything never gets.
Thus far everything has argued for the date: the day, the year, the calendar, the week, and now the earth. It is time for the reverse movement — the case against, at its strongest. The heaviest verse, the strongest rival, and the reigning date measured by its own standard. That is the next, and decisive, movement.
HeartHEART
There is a curious comfort in sending away a witness you would have liked to keep. It means you love the truth more than the win — and that, in the end, is the only posture fit for this subject. The earth shook, says Matthew, when her Maker died; and perhaps there, in the mud of the Dead Sea, lies a quiet echo of that day. But I will not make it larger than it is, and I will summon no phantom to strengthen it. For if it is true that He died on that Wednesday, then that truth has no need of my exaggeration — and if it is not true, I want to know. The earth whispers; she does not shout. And a whisper honestly heard is worth more than a chorus of fabrications.
Note on the sources
The Ein Gedi seismite has been checked at first hand in the geological literature; the window 26–36 comes from free varve counting (Williams/Schwab/Brauer 2012; AD 31 or 32 ±5, depending on the count); the anchor-free ¹⁴C dating (Kagan et al. 2011) gives a wider window, AD 12–91 at 1σ, which contains it, and then links the layer to the catalogue earthquake of 33; the dating "33" is demonstrably a fitted catalogue date, not a measurement. The base rate (~1 in 4 to 5) stands just as prominently beside it; the link to Matthew's quake is and remains construed. The Thallus and Phlegon passages have been checked in the source itself (via Syncellus, Jerome, Origen); they are expressly not fielded as witnesses but as a case study in source criticism, and Phlegon's own dating (Ol. 202.4 = AD 32/33) has been recomputed. The astronomical vice — a solar eclipse at Passover is physically impossible — holds against every eclipse explanation, on all sides. The volcanic route has been consulted at first hand with the ice-core reconstruction eVolv2k (v4) and ruled out: around AD 31 there is only one weak, unipolar-Antarctic sulphur signal, the wider period is the "Roman Quiet Period", and there is no tephra and no named source volcano (Sigl et al., 2015; Toohey & Sigl, 2017). The darkness thus cannot be explained away physically; what remains as a reading is the prophetic intertext Amos 8:9-10 — darkness at noon and an earthquake over a feast-keeping people — expressly marked as CONSTRUED, not as a measurement. This chapter is a corroborating ring; it does not carry the date. It claims no proof.
Literature (APA)
Carrier, R. (2012). Thallus and the darkness at Christ's death. Journal of Greco-Roman Christianity and Judaism, 8, 185–191.
Kagan, E., Stein, M., Agnon, A., & Neumann, F. (2011). Intrabasin paleoearthquake and quiescence correlation of the late Holocene Dead Sea. Journal of Geophysical Research, 116, B04311.
Karcz, I. (2004). Implications of some early Jewish sources for estimates of earthquake hazard in the Holy Land. Annals of Geophysics, 47(2–3), 759–792. [The chain New Testament/Talmud → Arvanitakis → Willis → later catalogues.]
Ken-Tor, R., Agnon, A., Enzel, Y., Stein, M., Marco, S., & Negendank, J. F. W. (2001). High-resolution geological record of historic earthquakes in the Dead Sea basin. Journal of Geophysical Research, 106(B2), 2221–2234. [Event C, AD 124 ± 188; read via Williams (2012).]
Migowski, C., Agnon, A., Bookman, R., Negendank, J. F. W., & Stein, M. (2004). Recurrence pattern of Holocene earthquakes along the Dead Sea transform. Earth and Planetary Science Letters, 222(1), 301–314.
Sigl, M., Winstrup, M., McConnell, J. R., Welten, K. C., Plunkett, G., Ludlow, F., [...] Woodruff, T. E. (2015). Timing and climate forcing of volcanic eruptions for the past 2,500 years. Nature, 523(7562), 543–549.
Toohey, M., & Sigl, M. (2017). Volcanic stratospheric sulfur injections and aerosol optical depth from 500 BCE to 1900 CE (eVolv2k). Earth System Science Data, 9, 809–831.
Van Voorst, R. E. (2000). Jesus outside the New Testament: An introduction to the ancient evidence. Eerdmans. [consulted, not cited in this chapter]
Williams, J. B. (n.d.). Dead Sea earthquake catalog [website]. deadseaquake.info. ["Both labels refer to the same textual passage in the Gospel of Matthew"; header date "26-36 AD".]
Williams, J. B., Schwab, M. J., & Brauer, A. (2012). An early first-century earthquake in the Dead Sea. International Geology Review, 54(10), 1219–1228.
Zohar, M., Salamon, A., & Rubin, R. (2016). Reappraised list of historical earthquakes that affected Israel and its close surroundings. Journal of Seismology, 20, 971–985. [used for the recording threshold ≈ M 5.5]
Primary texts: Matthew 27:45, 51–54 · Mark 15:33 · Amos 8:8–10 · Amos 1:1 (cf. Exodus 10:21–23; Joel 2:10, 31 as intertext of the darkness) · Julius Africanus, Chronographiae (via Syncellus) · Phlegon of Tralles (via Jerome, Chronicon; Origen, Contra Celsum 2.33, 59) · Orosius, Historiae adversus paganos VII.4. Astronomical control on the darkness (at a full-moon Passover a solar eclipse is impossible): author's own calculation; cf. Stephenson (1997), Historical eclipses and Earth's rotation, Cambridge University Press.
The week ended at an empty tomb. And the question that presses itself forward is whether it also left a trace outside the texts — whether the earth itself remembered anything of the day on which, according to Matthew, "the earth shook and the rocks were split" (Matthew 27:51). This chapter pursues that question, and it is a ring: it does not carry the case, it colours it. But a ring, precisely, must be weighed with the same rigour as a pillar, and at one point that rigour will work against me. For there is one trace that fits surprisingly well — and there are two "witnesses" who would say exactly what I want to hear, and whom I must nevertheless send away.
Begin with the trace in the ground, for that one is real. In the bed of the Dead Sea lie, layer upon layer, the deposits of two thousand years, and earthquakes leave their signature in them: a disturbed, churned-up layer between the quiet annual laminae. In the section near Ein Gedi lies such a quake layer from the early first century. And the remarkable thing is how it was dated. In 2012 Williams, Schwab and Brauer counted the annual layers back from one uncontested anchor point — the well-known earthquake of 31 BC, which Josephus describes — and they noted expressly that they made no assumption about the presumed date of the layer: "a date was assigned on the basis of counting the annual layers alone". Their result: AD 31 or 32 (±5), depending on the count. The authors give the window as 26 to 36, largely the term of office of Pilate; without a year zero it shifts by one year (Williams, Schwab & Brauer, 2012).
Dossierthe window, and the base rate
The measurement. The window 26–36 comes from the free counting of the annual layers (Williams, Schwab & Brauer, 2012). An anchor-free radiocarbon dating (Kagan et al., 2011) gives a wider window, AD 12–91 at 1σ, which contains it; Kagan then links the layer to the catalogue earthquake of 33. Neither method yields a year — only a window. The oft-heard dating "AD 33" is something else: it is a fitted catalogue date — Migowski and colleagues (2004) shifted their floating chronology so that the layer coincided with an already dated catalogue earthquake, and that catalogue date comes in turn from crucifixion chronologists with Matthew as their only source. Whoever says "33" cites the fit; whoever says "26–36" cites the measurement.
The base rate — obligatory, and the opponent supplies it. Earthquakes are frequent in the Dead Sea rift. In the late Second Temple period the series counts roughly one correlated quake per half-century (Migowski et al., 2004); the chance that an arbitrary eleven-year window (26–36) contains at least one is then roughly one in four to five [CALCULATED]. The seismite is thus corroboration at window level — consistent with Wednesday 25 April 31, but discriminating for nothing within that window. Whoever makes more of it loses by this book's own house rule. And there is a second limit, from the other side: the layer says nothing about what was felt in Jerusalem. A seismite in the Dead Sea rift only registers from about M ≥ 5.5, and that is a threshold for laying down the layer, not for the perceptibility of a shock a hundred kilometres away (Zohar et al., 2016; Williams, Schwab & Brauer, 2012). The layer therefore carries an event within the window; it does not carry a tremor anyone in the city noticed. That is no weakening of this chapter but its sharpening — the chapter did not claim the latter in the first place. [DOCUMENTED]
There is a fine irony in the way this trace has been handled, and it is worth recording because it repeats in miniature the pattern of this whole book. The geologists measured the year 31 — and then gave it back. Immediately after their own dating, Williams, Schwab and Brauer write: "naturally it is not likely that an earthquake of the crucifixion could have taken place in AD 31" — with a reference to the list of candidate years drawn up by the crucifixion chronologists, in which AD 31 did not appear. The geologists found 31 and returned the year to the chronologists, who had never examined it. Each discipline outsources the question to the other, and so the year that no one weighed disappears. That is not evidence that AD 31 is right; it is evidence that AD 31 has never been tested. The trace opens the door; it pushes no one through.
Dossierhow "33" got into the catalogue, and how such entries arise
The fitted catalogue date has a family tree, and it is on record. Williams says in his own earthquake catalogue where the label comes from: "This earthquake is known in most catalogs as the 33 AD earthquake, though some designate it as a 30 AD event. Both labels refer to the same textual passage in the Gospel of Matthew [...] The difference in dates reflects uncertainty surrounding the year of Jesus' death." Karcz has written out the chain: New Testament and Talmud → Arvanitakis (1903/04) → Willis (1928) → "other cataloguers followed suit" → all later catalogues (Karcz, 2004). The church's dating supplied the year to the catalogue, the catalogue gave it to the geologists, and the geologists gave it back to the theologians as independent confirmation. That is a circle, and it is documented. Two data show that the mechanism really works this way. The catalogue entry "earthquake AD 48" goes back to a misread verse, Acts 8:24 instead of 16:26, where a quake stands in Macedonia and not in Jerusalem; Williams points that out. And the counts that gave "33" were tuned to it: Migowski "used varve counts in the masked intervals to match her varve-counted year to historically documented earthquakes", and Ken-Tor's own, untuned dating of the same layer was AD 124 ± 188, only afterwards linked to "33" (Ken-Tor et al., 2001). Whoever says "33" thus cites not only a fit; he cites Matthew, by way of two intermediate stations. And the same measure holds in reverse: Williams' midpoint too is a count with a margin, not a year the layer itself carries. [DOCUMENTED — Williams, Schwab & Brauer (2012) and Williams' catalogue; Karcz (2004); Ken-Tor (2001) and Migowski via Williams]
And honesty demands the other side at once. Only Matthew mentions the quake, and he mentions it together with raised saints coming out of the tombs — an image that scholars, not without reason, read as a theological motif, a topos of nature responding to great events. That explains the literary form. But it does not explain the layer in the lake: there is a physically dated quake in the window, measured by researchers with no stake whatever in the date of the crucifixion. The strongest formulation is therefore doubly modest: the only physical quake trace from the time of Pilate falls in a window that largely covers his term of office and whose arithmetic midpoint in Williams is 31 — no more, and that is enough. The link to Matthew's quake remains a construal.
Expertthe brakes on the layer, in full
The margin of five years is no detail but the whole measurement. Of the 62 counted annual layers, 58 percent received the lowest quality rating, "questionable"; the same core gave another year with another counter, Migowski's 33, and that this count was tuned to the catalogue weakens it without removing the uncertainty. The authors themselves leave three possibilities open, of which the third does not flatter this case: that Matthew's quake is allegory and the seismite comes from an unrecorded quake. And the echo of 2012 ("Quake Reveals Day of Jesus' Crucifixion") promised what the article does not say; Williams himself, when asked: "we added no new information about the date of the crucifixion". This chapter says no more than he does. [DOCUMENTED — Williams, Schwab & Brauer (2012); the press statement via the 2012 reporting]
One line was pursued until it ran dead, and it belongs here, because it could have given the seismite its season and with it a test on Passover. An annual layer consists of a winter half of clay and a summer half of aragonite; if a quake writes its stroke at the moment of the shock, the position of the seismite within the annual layer would betray the season, and Passover falls in spring. The method even seemed calibratable, for of the anchor quake of 31 BC Josephus gives the season: "at the beginning of spring". The kill criterion was set in advance: if the season cannot be read from the layer, the line lapses. It lapses on three points. The anchor layer is nine centimetres thick and disturbs several years at once, so that the calibration is not even feasible; Williams reports the position nowhere; and the deepest objection is physical: the seismite arises because the upper laminae liquefy and settle again, and that process destroys precisely the seasonal information it should have preserved. The criterion has been hit. The line stands here as a closed door, with the question that ought to be put to the authors, and nothing is built on it. [kill criterion hit; line lapsed]
Now the two witnesses I must send away, and this is the purest test in this whole chapter — for they would seem to serve my case, and precisely for that reason I must be strict. For centuries apologists have paraded two pagan writers who are said to have recorded the darkness at the crucifixion: Thallus and Phlegon. If they were real, independent witnesses, I would have a gift. But they are not, and I say so before a critic does.
Of Thallus not a single word survives. What is physically at hand is a Byzantine chronicle from around 800 which says that Julius Africanus said that Thallus said something — and even there stands no quotation, only Africanus' summary, with the remark that Thallus "wrongly" called the darkness a solar eclipse. The link to the crucifixion sits in Africanus' demonstrative pronoun — "this darkness" — not in Thallus; the name Jesus does not occur in it; and the dating of Thallus himself has a margin of three hundred years. A witness three hundred years wide, whose words do not survive, is no witness. And even a Christian historian saw that there was nothing astronomical to see: Orosius (around 417) denies expressly that the darkness was an ordinary solar eclipse — as indeed it could not have been, for Pesach falls at full moon, when a solar eclipse is impossible (Orosius, Historiae adversus paganos VII.4).
And Phlegon is worse — for he does not even point the right way. Phlegon names no Jesus, no Judea, no Passover; he names an earthquake in Bithynia, well over a thousand kilometres from Jerusalem, and a solar eclipse. And recompute his own dating: it falls in the fourth year of the two hundred and second Olympiad — the year 32/33. Phlegon supports AD 33, not AD 31. Whoever defends a crucifixion in 31 and cites Phlegon as confirmation would be undercutting his own thesis; he would have to shift the witness two years, and then he has no right left to call him a witness. That is measuring with two measures, and I do not do it. I send Phlegon away — not despite the fact that he names a crucifixion year, but in part because it is the wrong one.
Expertthe astronomical vice, which holds for everyone
Beneath all these witnesses lies one physical impossibility that closes off the whole eclipse explanation, for every year and for every camp. A solar eclipse requires a new moon — sun and moon in conjunction. The Passover falls by definition at full moon — sun and moon opposite each other. A solar eclipse at the Passover is therefore physically impossible, and maximal totality lasts seven minutes, not three hours. The darkness of the ninth hour, whatever it was, was no solar eclipse — and whoever brings in Thallus or Phlegon to make it an eclipse is bringing in an explanation that astronomy forbids. That holds against the apologist who wants to support AD 31 exactly as hard as against whoever wants to support AD 33.
Dossierand no volcano: what the ice cores say
The eclipse is astronomically ruled out; there remains the question whether a volcanic veil could have quenched the sun for three hours. That door too swings shut, and now with physical data. The ice-core reconstruction eVolv2k (version 4) has been consulted at first hand for the window. Around the year 31 the cores carry precisely one sulphur signal — VSSI 0.95 ± 0.71 Tg sulphur — and it points the wrong way: unipolar-Antarctic (Greenland 0.0), originating from around 45° southern latitude. A weak, uncertain eruption in the southern hemisphere cannot darken the sky above Jerusalem. More broadly, the whole period from roughly 40 BC to AD 160 lies as the "Roman Quiet Period": volcanic sulphur output there runs 55 to 70 percent below the long-term average. And from close by there is nothing: the Harrat Ash Shaam volcanic field has no historical eruptions — the youngest dates from about 2670 BC and was an effusive, basaltic flow, not an ash-shedding eruption — and no first-century tephra layer has been reported in Israel or Judea, nor a named source volcano anywhere in the window (eVolv2k: "N/A"). A tephra layer not found is no demonstrated absence; but a stratospheric veil dims for months in any case, not three sharply bounded hours, and ashfall darkens only beneath the plume, close to the crater. On each of these points the volcanic explanation fails. [DOCUMENTED — eVolv2k v4 consulted at first hand] Together with the astronomical vice this means: the midday darkness cannot be explained away as an ordinary natural phenomenon — not as eclipse, not as volcano, not as khamsin (a khamsin sky is diffusely luminous, not dark).
What then remains of the darkness, if neither the eclipse nor the volcano carries it? The most honest line is not physical but scriptural — and the evangelists lay it down so themselves. The midday darkness stands in the language of prophetic judgement, and one prophecy sounds unmistakably within it. "I will make the sun go down at noon and darken the land in broad daylight", says Amos — and in the same breath: "I will turn your feasts into mourning" (Amos 8:9-10). It is a judgement over a feast-keeping people, and it does not come alone: two verses earlier the land trembles at it (Amos 8:8), and the book opens with the dating "two years before the earthquake" (Amos 1:1). Darkness at noon and a trembling earth, together, over a people at feast — that is precisely the conjunction Matthew sets down at the ninth hour (Matt. 27:45, 51). I give that reading its mark without ado: CONSTRUED, not CALCULATED — a typological echo, not a measurement. And precisely as a construal it is worth more than a fabricated physics: it places the darkness in the category the text itself already sets it in — that of the sign.
Excursushow a "proof" accretes, and the mirror it holds up
The most instructive thing about Thallus and Phlegon is not that they fail, but how they were made into "witnesses" — sentence by sentence, traceably. Phlegon (around 130) reports an eclipse and a quake in Bithynia. Africanus (around 221) adds that it was "at full moon" and lasted "until the ninth hour" — two details that come straight out of Mark 15:33, and which Origen himself denies that Phlegon said. Eusebius (around 320) adds that "all this agrees with what happened at the Saviour's passion". Malalas recalculates it to "the eighteenth year of Tiberius". And Michael the Syrian, in the twelfth century, has Phlegon say that the dead rose, entered Jerusalem and cursed the Jews — and that it was a Friday. Each link adds precisely that which the argument needs. That is the signature of apologetic accretion, and the table belongs in this book not as an attack on others but as a mirror: test every chain of sources of your own against this pattern. Modern source criticism draws from it the only honest conclusion: Thallus nowhere named Jesus and should be struck from the witness lists (Carrier, 2012, p. 191). And that measure holds in all directions — towards mine as well: test every chain of sources of your own against this same pattern of accretion.
And yet beneath all this rubble lies one thing that shines, and it is better than anything the apologists have made of it. That same Africanus, around the year 221, rejects the solar-eclipse explanation on purely astronomical grounds — though it would have suited him excellently, for it would have furnished him a pagan witness. He knows that the Passover falls on the fourteenth of the lunar month, that this means the full moon, and that an eclipse requires the new moon. And he refuses the argument: "how could one assume an eclipse when the moon stands almost diametrically opposite the sun?" He sacrifices evidence congenial to him because astronomy forbids it. That is exactly the discipline this whole book demands — carried out eighteen centuries earlier. Africanus is not a witness. He is an example.
Where does the inquiry stand, then, at the end of this ring? The earth has left one trace that fits — a quake layer in the window, its arithmetic midpoint on the year 31, measured by people who never even considered that year. More it does not carry: a base rate of one in four to five keeps it in its place, as colour, not as pillar. And the two literary "witnesses" who had seemed most likely to help, I have sent away, because on closer inspection they are phantoms — and one of them even points to the opponent's year. That sending away is not a weakness of this chapter; it is its integrity. A case that rejects its own weak witnesses earns the trust that a case which seizes everything never gets.
Thus far everything has argued for the date: the day, the year, the calendar, the week, and now the earth. It is time for the reverse movement — the case against, at its strongest. The heaviest verse, the strongest rival, and the reigning date measured by its own standard. That is the next, and decisive, movement.
HeartHEART
There is a curious comfort in sending away a witness you would have liked to keep. It means you love the truth more than the win — and that, in the end, is the only posture fit for this subject. The earth shook, says Matthew, when her Maker died; and perhaps there, in the mud of the Dead Sea, lies a quiet echo of that day. But I will not make it larger than it is, and I will summon no phantom to strengthen it. For if it is true that He died on that Wednesday, then that truth has no need of my exaggeration — and if it is not true, I want to know. The earth whispers; she does not shout. And a whisper honestly heard is worth more than a chorus of fabrications.
Note on the sources
The Ein Gedi seismite has been checked at first hand in the geological literature; the window 26–36 comes from free varve counting (Williams/Schwab/Brauer 2012; AD 31 or 32 ±5, depending on the count); the anchor-free ¹⁴C dating (Kagan et al. 2011) gives a wider window, AD 12–91 at 1σ, which contains it, and then links the layer to the catalogue earthquake of 33; the dating "33" is demonstrably a fitted catalogue date, not a measurement. The base rate (~1 in 4 to 5) stands just as prominently beside it; the link to Matthew's quake is and remains construed. The Thallus and Phlegon passages have been checked in the source itself (via Syncellus, Jerome, Origen); they are expressly not fielded as witnesses but as a case study in source criticism, and Phlegon's own dating (Ol. 202.4 = AD 32/33) has been recomputed. The astronomical vice — a solar eclipse at Passover is physically impossible — holds against every eclipse explanation, on all sides. The volcanic route has been consulted at first hand with the ice-core reconstruction eVolv2k (v4) and ruled out: around AD 31 there is only one weak, unipolar-Antarctic sulphur signal, the wider period is the "Roman Quiet Period", and there is no tephra and no named source volcano (Sigl et al., 2015; Toohey & Sigl, 2017). The darkness thus cannot be explained away physically; what remains as a reading is the prophetic intertext Amos 8:9-10 — darkness at noon and an earthquake over a feast-keeping people — expressly marked as CONSTRUED, not as a measurement. This chapter is a corroborating ring; it does not carry the date. It claims no proof.
Literature (APA)
Carrier, R. (2012). Thallus and the darkness at Christ's death. Journal of Greco-Roman Christianity and Judaism, 8, 185–191.
Kagan, E., Stein, M., Agnon, A., & Neumann, F. (2011). Intrabasin paleoearthquake and quiescence correlation of the late Holocene Dead Sea. Journal of Geophysical Research, 116, B04311.
Karcz, I. (2004). Implications of some early Jewish sources for estimates of earthquake hazard in the Holy Land. Annals of Geophysics, 47(2–3), 759–792. [The chain New Testament/Talmud → Arvanitakis → Willis → later catalogues.]
Ken-Tor, R., Agnon, A., Enzel, Y., Stein, M., Marco, S., & Negendank, J. F. W. (2001). High-resolution geological record of historic earthquakes in the Dead Sea basin. Journal of Geophysical Research, 106(B2), 2221–2234. [Event C, AD 124 ± 188; read via Williams (2012).]
Migowski, C., Agnon, A., Bookman, R., Negendank, J. F. W., & Stein, M. (2004). Recurrence pattern of Holocene earthquakes along the Dead Sea transform. Earth and Planetary Science Letters, 222(1), 301–314.
Sigl, M., Winstrup, M., McConnell, J. R., Welten, K. C., Plunkett, G., Ludlow, F., [...] Woodruff, T. E. (2015). Timing and climate forcing of volcanic eruptions for the past 2,500 years. Nature, 523(7562), 543–549.
Toohey, M., & Sigl, M. (2017). Volcanic stratospheric sulfur injections and aerosol optical depth from 500 BCE to 1900 CE (eVolv2k). Earth System Science Data, 9, 809–831.
Van Voorst, R. E. (2000). Jesus outside the New Testament: An introduction to the ancient evidence. Eerdmans. [consulted, not cited in this chapter]
Williams, J. B. (n.d.). Dead Sea earthquake catalog [website]. deadseaquake.info. ["Both labels refer to the same textual passage in the Gospel of Matthew"; header date "26-36 AD".]
Williams, J. B., Schwab, M. J., & Brauer, A. (2012). An early first-century earthquake in the Dead Sea. International Geology Review, 54(10), 1219–1228.
Zohar, M., Salamon, A., & Rubin, R. (2016). Reappraised list of historical earthquakes that affected Israel and its close surroundings. Journal of Seismology, 20, 971–985. [used for the recording threshold ≈ M 5.5]
Primary texts: Matthew 27:45, 51–54 · Mark 15:33 · Amos 8:8–10 · Amos 1:1 (cf. Exodus 10:21–23; Joel 2:10, 31 as intertext of the darkness) · Julius Africanus, Chronographiae (via Syncellus) · Phlegon of Tralles (via Jerome, Chronicon; Origen, Contra Celsum 2.33, 59) · Orosius, Historiae adversus paganos VII.4. Astronomical control on the darkness (at a full-moon Passover a solar eclipse is impossible): author's own calculation; cf. Stephenson (1997), Historical eclipses and Earth's rotation, Cambridge University Press.