Drawing the map
1300–1050 BCE · eastern Mediterranean
The Bronze Age collapse, place by place
153 destructions are said to have happened around 1200 BCE. Going back to the excavation reports, Jesse Millek keeps 59. Every claim is put on a clock that shows what dated it; 27 have no known position and appear in the list only.
The usual map
One wave, as the books draw it#
In the common picture the Bronze Age ends in a ring of ruined cities around the eastern Mediterranean: Mycenaean palaces, the Hittite capital, Ugarit, the towns of Canaan, all at about 1200 BCE. Over 150 years of excavation reports and surveys, as Jesse Millek collected them, have been cited for 153 destructions at 148 places at that date. These are claims collected from the literature, not a census of ancient cities or of every destruction that archaeologists have found.
This is that picture. Every claim is lit at once.
Back to the reports
94 of the 153 claims are set aside#
59destructions around 1200 BCE that Millek keeps after going back to the excavation reportsJesse Millek (2023) returned to the original report of every claim. 33 are dated to another time, 35 rest on little or no evidence of destruction, and 26 are not in the report they are cited from. Those 94 are the hollow rings. This does not mean that 94 cities were never destroyed or that 61% of ancient destructions are disproved: they are claims that do not stand at about 1200 BCE in his reading of the reports. The 59 he keeps are not 59 precisely dated catastrophes either; a few rest on thin evidence, and the place cards say which.
A review in the journal Antiquity (2024) calls the recount supportive and notes that what counts as a destruction is not straightforward. To test the rejections, 20 of the 94, drawn by lot, were checked against the excavators’ own reports by AI research agents (Claude models, not human archaeologists) that were given only the site name and could search and read open sources: 11 held, 2 did not (Iria and Tell Abu Hawam, shown on their cards), 7 could not be settled. It is a limited, source-access-dependent check, not a validation of the catalogue.
How big were the fires
Of the 59, 8 were destroyed across a whole site#
The rest are one building (19), several buildings (15) or a patch the report cannot size (17). Ringed here: Pylos, Midea, Troy, Tarsus, Kilise Tepe, Ugarit, Aphek and Maa-Palaeokastro.
The report names no cause for 36 of the 59. Millek reads warfare for 10 and an earthquake for 2; the other 11 are arson, accident, natural causes or ritual closure. The “earthquake storm” hypothesis (a run of earthquakes that struck many places, Nur and Cline 2000) needs more than that.
What is and is not at stake
A town burnt is not a system gone#
The recount concerns physical destruction of places. The end of palace administration in the Aegean, the end of the Hittite imperial system, the loss of writing in some regions and the change in settlement patterns are not the same thing as a town burning. A political or economic system can stop working without its towns being destroyed, and a town can be burnt and recover.
Millek’s recount weakens the picture of one synchronous wave of destruction. It does not settle whether a deep political and institutional collapse took place; the clocks below ask only when places were destroyed, and how well that can be known.
The strongest objection to the framing comes from a review of Millek’s book in the Journal of Greek Archaeology (2025): his window of 1225–1175 BCE is narrow, and many scholars now see the end of the Bronze Age as a process over about a century, roughly 1250 to 1150, not as one event. That is compatible with the recount. A century-long crisis and the absence of one synchronised wave of destruction can both be true; the recount tests only the second.
1300 BCE
A world run from a few palaces#
Hatti, the Hittite kingdom, governs Anatolia and much of Syria from its capital Hattusa. Egypt holds Canaan. Palaces at Pylos, Mycenae and Thebes keep their stores on clay tablets. Ugarit is a harbour kingdom between them, and copper comes from Alashiya, which is Cyprus.
The lights are the places that appear in the lists of destructions. Each stands until its own date.
About 1250 BCE
Fires before the crisis#
Some of the destroyed places are older than the crisis. Hazor was destroyed about 1250 BCE, Tel Harasim and Palaikastro Kastri in the middle of the century, and Arslantepe in the first half of it. Textbook maps of 1200 BCE include them anyway.
Nine contemporary texts refer to grain shortage or famine in Hittite lands; Manning and colleagues (2023) caution that they lack detail and context.
1225–1175 BCE
The window, and what dates it#
Millek accepts a destruction only inside 1225–1175 BCE, which this page calls the window. Of the 89 entries that describe a destruction the page can place in time (the 59 kept, plus the 30 of the 33 misdated claims whose report gives a date the page can place; a few are uncertain even in Millek’s own wording), 51 have no date of their own: they sit on the window because they were put there. 21 more are dated by their pottery phase to another time, and 17 by a clock that does not depend on pottery: a text, tree rings or radiocarbon.
A destruction is dated by what lies in its layer. A pottery style is dated by comparison with other places; a text by its own year; tree rings and radiocarbon by measurement. On the map a dotted ring is a date by convention and a bright flash a date from a text, tree rings or a radiocarbon model.
About 1208 BCE
Merneptah names five peoples from the north#
In year 5 of King Merneptah of Egypt, about 1208 BCE, a Libyan chief reaches the western Delta with Ekwesh, Teresh, Luka, Sherden and Shekelesh, bringing his wife and children. The king defeats them. The text calls them “northerners coming from all lands”; it does not say they came by sea.
The same inscription has Merneptah sending grain by ship “to keep alive” Hatti.
About 1200 BCE
The Greek palaces are dated by one pottery label#
Mycenae, Tiryns, Midea, Thebes, Gla and Pylos were all destroyed, most of them by fire, and the reports date each to the end of the pottery phase called Late Helladic IIIB (the last palace-period style of the Greek mainland) or its change into IIIC; a 2024 reassessment of the pottery puts the final destruction of Pylos in IIIC Early 1. That is one label, not six measurements. At Pylos, tablets list coast-watchers and rowers; none names an enemy.
Radiocarbon, which dates when a plant or tree stopped living, was measured for Pylos, Mycenae and Midea in 1962–1970. Most are on charcoal, which can be older than the fire. Calibrated, each spans more than three centuries, too wide to place a fire in a decade.
1198–1196 BCE
Three dry years at Gordion, and a capital emptied before it burnt#
The rings of juniper beams at Gordion show an unusually dry run from about 1198 to 1196 BCE (±3 years). Gordion lies some 230 km from Hattusa; the authors argue the drought contributed to the collapse alongside other pressures, not that it was the single cause.
At Hattusa the city was emptied by the royal administration, and only afterwards were some monumental buildings burnt. The fire has no date of its own.
After 1194 BCE
Ugarit and Emar: two ends tied to a text#
A letter from the chancellor of Egypt, dated by Egyptian reigns to somewhere in 1194–1190 BCE on the low Egyptian chronology (higher chronologies move it about eleven years earlier), reached the last king of Ugarit, Ammurapi, and the city fell after it. Emar’s archive stops after a tablet of the second year of Meli-Shipak of Babylon, 1185 or 1175 BCE by the two chronologies; the city fell after that.
Both give only the earliest possible date: the end came after them, by an unknown number of years. Ammurapi’s own letters speak of enemy ships and of a land in danger. One names “seven ships”. No text says who they were.
About 1177 BCE
Medinet Habu: what the king says, and what it cannot date#
In year 8 of Ramesses III (1177 BCE on the low Egyptian chronology, 1188 on the higher), the text carved on his temple at Medinet Habu says that no land could stand before them, from Hatti, Kode, Carchemish and further, and that a camp was set up in Amor. His ships and troops meet them at the Nile mouths.
The verb is “cut off”. Carchemish did not fall. The sweep is a boast; it does not date the end of any place on the list.
1160–1120 BCE
Some southern towns are dated later#
Lachish Level VI, Jaffa, Tel Azekah, Tel Dan, Beth-Shean, Tell Deir Alla and Megiddo VIIA are put by their excavators between about 1150 and 1050 BCE. Millek counts all of these as misdated: they were destroyed, but not at 1200.
Radiocarbon puts three of them late. Jaffa’s last fortress was destroyed most probably in 1127–1098 BCE (at 95% not before 1189), Tel Azekah’s last phase ended in 1160–1030, and Megiddo Stratum VIIA, a separate and later fire, in 1080–1041 (68% ranges). Lachish Level VI, at 1210–1126 (95%), is wide enough to hold a date in the 1190s and one in the 1130s.
Taking stock
What the clocks that do not depend on pottery add up to#
Take every destruction, in whichever class, whose own radiocarbon date belongs to the destruction itself: 7 places, 6 of them in the southern Levant. Five have a 95% range in print, 84 to 245 years wide, and their outer limits overlap between 1189 and 1148 BCE, and the years all five sets actually contain are 1189–1175 and 1159–1148 (Jaffa’s own set leaves out 1174–1160). Overlap of separate ranges is not evidence that five places fell together; it only shows that the ranges do not rule it out. Of the two printed only at 68%, Tel Azekah overlaps and Megiddo VIIA (1080–1041) lies after. 1200 BCE falls outside Jaffa’s 95% set, a firmer statement, and outside Megiddo’s 68% range, a weaker one.
The two texts give only earliest possible dates: Ugarit fell after a letter dated within 1194–1190 BCE, so no earlier than 1194; Emar after 1185 or 1175. The result is a shortage of resolution, not a newly found date. Five separate ranges 84 to 245 years wide cannot by themselves tell destructions bunched within a few decades from a sequence spread over a century or more; that needs a joint model of the sites.
1050 BCE
What was left#
Of the places in these lists, Byblos, Sidon, Tyre, Carchemish, Kition and Ashkelon show no end-of-Bronze-Age destruction in their reports. The Phoenician coast, Kition, Carchemish and Egypt itself continued; the Philistine towns rise on the coast with new pottery, loomweights and cooking pots. Hattusa is gone as a capital, Luwian-speaking successors rule from Carchemish, and Egypt has lost Canaan. Assyria and Babylon fight Arameans for generations.
The lists record places whose end archaeologists noticed. They say nothing about the much larger number of places that stayed.
Look up a place
Any of the 153 claims, in the words a report would use: what is said to have been destroyed, what the report shows, how the date is known.
Type a place: Troy, Ugarit, Pylos, Lachish, Hattusa, Byblos.
Drag the map, pinch or scroll to zoom.
The names the Egyptians wrote
The label “Sea Peoples” is modern: a 19th-century coinage, used by Maspero and others. The Egyptian texts say “northerners” and “isles”, and attach “of the sea” to a few names only: Ekwesh in Merneptah’s texts, Sherden and Teresh at Medinet Habu, Sherden and Weshesh in Papyrus Harris I.
Nine names recur across the texts. Lukka is also known as a land on Hittite maps; Denyen is often linked with the land Danuna, though some scholars connect it with the Greek Danaoi. The others are names in Egyptian spelling. The places they are matched with rest mostly on likeness of sound, not on a text that says “from here”; for the Philistines, archaeology and DNA add to it.
| Name | Where it appears | Proposed home | How firm |
|---|---|---|---|
| Lukka (Luka) | Kadesh 1274; Merneptah year 5; a letter of Ammurapi of Ugarit | Lycia, south-west Anatolia | Firm: a land on Hittite maps |
| Sherden | Amarna letters c. 1350; Ramesses II year 2 and Kadesh; Merneptah; Medinet Habu | Sardinia | Argued; open |
| Ekwesh | Merneptah only | Ahhiyawa, the Achaeans (Greeks) | Argued; contested |
| Teresh | Merneptah; Medinet Habu | Tyrsenoi (Etruscans) or Lydia | Weak: likeness of name |
| Shekelesh | Merneptah; Medinet Habu; a Hittite letter to Ugarit that names the Šikalayu, read as Shekelesh or as Tjeker | Sicily, or Sagalassos in Pisidia | Weak |
| Peleset | Medinet Habu; Papyrus Harris I | The Philistines | Firm on the name; origin argued (Aegean) |
| Tjeker | Medinet Habu; Papyrus Harris I; Wenamun, at Dor, about 1075 | Teucrians of the Troad, or Crete | Weak |
| Denyen | Medinet Habu; Papyrus Harris I; Danuna in Amarna letter EA 151 | Adana, Cilicia (Danuna); or the Danaoi of Greece | Divided; the link to 1177 is an inference |
| Weshesh | Medinet Habu; Papyrus Harris I | Unknown | Weak |
What the Medinet Habu pictures show
Five ships hard pressed by four Egyptian vessels; a land battle in which the other side flees in chariots and on foot while women, children and baggage move away in heavy ox-carts. These are the editors’ descriptions of the 1936 Chicago publication, and the oldest reading of them (Breasted 1906) was a migration of families with a fleet.
Both readings stand on the same wall. The page does not choose between a raid and a migration. What the wall does not show is a city burning.
What the clocks that do not depend on pottery can and cannot say
Three things can date a destroyed level without a pottery typology: a text with a regnal year or a known correspondence, tree rings, and radiocarbon on short-lived seeds. Of the 89 entries that describe a destruction the page can place in time, 51 sit on the window by classification, 21 are dated by pottery phase to another time, and 17 have a clock of one of those three. Of those 17, 7 give a radiocarbon range for the destruction itself, 2 are texts that give only an earliest possible date, 4 give a level boundary (the change from one occupation layer to the next) or a single wood sample, and 4 date something far earlier or later.
Each bar below is one published range. Where a paper prints a 95% range it is the grey bar. Where it prints only the 68% range, only the orange bar is drawn: it holds about two chances in three, so the real 95% range is wider and was not printed. These clocks are independent of pottery, not of modelling: each range is a modelled inference tied to the stratigraphy and to the authors’ reading of which layer is the destruction.
Radiocarbon is bluntest where it is most needed. The calibration curve is flat in the 12th century: a measurement of 1200 BCE with a ±20 year error calibrates to a 95% range of 162 years in total, spread between 1260 and 1057 BCE. At 1500 BCE the total is 81: half as much. Bayesian models narrow that in the best cases to 45–70 years at 68%. None of the 5 ranges printed at 95% is narrower than 84 years.
The outer limits of the 5 ranges printed at 95% overlap between 1189 and 1148 BCE; the years all five sets actually contain are 1189–1175 and 1159–1148, because Jaffa’s own set leaves out 1174–1160, which carry less probability than the cut-off for the set. Overlapping marginal ranges do not establish a high joint probability of simultaneity, and do not rule it out. Of the two printed only at 68%, Tel Azekah overlaps and Megiddo VIIA, at 1080–1041, lies after. 6 of the 7 are in the southern Levant; the Aegean, the Hittite heartland and Egypt have no range of this kind in Millek’s catalogue, though chronometric work exists there.
So “about 1200 BCE” is, for most places, the label of a pottery phase. 1200 BCE itself is outside Jaffa’s 95% set and outside Megiddo’s 68% range, which are statements of different strength, and inside the four wider 95% ranges, which cannot tell. The finding is a shortage of resolution: these ranges neither show a common date nor exclude one, and by themselves they cannot test one wave.
| Place | What the range dates | Material | How the range was made | Ranges as printed | Source |
|---|---|---|---|---|---|
| Gezer | The modelled end of the Stratum 12B residence, tied to its destruction | Seeds, olive pits and grape pips; collagen of a victim | Bayesian model (IntCal20) ordered by the stratigraphy; samples come mostly from the last years of use, so the end boundary follows them | 68.3%: 1218–1172; 95.4%: 1244–1148 | Webster et al. 2023, PLOS ONE 18(11), Table 4 |
| Lachish Level VI | The modelled end of Level VI, the destruction of the final city | Cereal seeds and legumes from burnt jars | Bayesian model (IntCal13) of the level sequence; the end boundary follows the samples | 68.2%: 1203–1130; 95.4%: 1210–1126 | Garfinkel et al. 2019, Radiocarbon 61(3), Table 5 |
| Megiddo Stratum VIIA | A modelled boundary between phases H-11 and H-10; the authors tie the end of H-11 to the final VIIA palace destruction | Olive pits and lentils | Bayesian model (IntCal13); that the boundary is the destruction is the authors’ correlation | 68.2%: 1080–1041; 95.4% not printed | Martin, Finkelstein and Piasetzky 2020, BASOR 384, Table 5 |
| Jaffa | The modelled end of fortress phase RG-3a, a destruction | Wheat seeds and grain from floors and debris | Bayesian model (IntCal13) of the phase sequence; the sets are in pieces | 68.2%: 1151–1147, 1127–1098 (64.3%), 1096–1095; 95.4%: 1189–1175 (4.3%), 1159–1140 (8.4%), 1129–1047 (82.7%) | Burke et al. 2017, AJA 121(1), Fig. 35 |
| Tel Azekah | The modelled end of phase T2-3a, which ends in destruction | Carbonized olive pits | Bayesian model (IntCal13); only the 68% range was read | 68.2%: 1160–1030; 95.4% not read | Webster et al. 2018, Radiocarbon 60(1), Fig. 8 |
| Apliki-Karamallos | The modelled destruction of House A, a boundary placed just after the last-use samples | Annual cereals, short twigs and brushwood of up to five rings | Bayesian boundary model (IntCal13) | 68.2%: 1193–1190 (2.0%), 1176–1118 (66.2%); 95.4%: 1210–1104 | Manning, Kearns and Lorentzen 2017, Fig. 13 |
| Tell Deir Alla | One sample from the Phase E destruction debris | A single grape pip | One calibration (IntCal20), with no model of the sequence | 68.3%: 1261–1131; 95.4%: 1366–1121 | Halbertsma et al. 2026, Levant 58(1), Tables 3–4 |
The weather, and what is known about it
The sharpest climate record is the Gordion juniper beams: about 1198 to 1196 BCE (±3), three years of ring widths 34% narrower than in the four years before. Its authors, Manning and colleagues (2023), argue that the drought contributed to the fall of Hatti alongside other pressures, and reject a single-cause reading.
Other records are slower. Pollen from the Sea of Galilee shows the driest stretch of the Bronze and Iron Ages between about 1250 and 1100 BCE (Langgut and colleagues 2013), but re-dated with a different age model the same record places 1200 BCE anywhere in a span of 150 depth centimetres (Knapp and Manning 2016). A stalagmite from the south-west Peloponnese shows little sign of drier conditions when Pylos was destroyed, and a gradual drying after about 1150 BCE (Finne and colleagues 2017).
The page draws only the tree rings, because only they are tied to a calendar year within a few years. The rest are a general drift towards dryness whose onset cannot be pinned to a decade.
The causes, and where each stands
Each explanation, at its strongest, with the strongest published objection.
| Explanation | At its strongest | The objection | Status |
|---|---|---|---|
| Systems collapse, a “perfect storm” (Cline 2014) | Drought, earthquakes, revolt and trade failure arriving together | Not a single testable cause; some of the “synchronous” evidence is now counted as false | Hypothesis |
| Earthquake storm (Nur and Cline 2000) | Most destroyed sites lie in zones shaken by 20th-century earthquakes | A map of present-day shaking zones plus damage read as seismic at some sites; the sites’ dates are pottery phases, so synchrony is not shown, and Millek reads only 2 of 59 destructions as earthquake | Hypothesis |
| Drought and famine (Kaniewski and colleagues) | Dry pollen phases from Syria and Cyprus after about 1200 | Chronologies too coarse to match the event year (Knapp and Manning 2016) | Inferred; contested |
| Warfare, new infantry tactics (Drews 1993) | Raiders overwhelm chariot-based kingdoms | Millek reads warfare at only 10 of the 59 he keeps | Hypothesis |
| Sea Peoples as the cause | Egyptian texts name them; ships in Ugarit’s letters | Knapp and Manning: a symptom, not the cause; no real evidence at 16 southern Levantine sites (Millek) | Disputed |
| Internal revolt | Social change altered how resources were handled | No direct text or layer dates an uprising | Hypothesis |
| Plague | Yersinia pestis strains already in Eurasia | No eastern Mediterranean genome from the period shows it | Hypothesis |
| Trade network failure | Tin, copper and Aegean exchange declines with the palaces | A consequence of internal change, argued Millek | Inferred |
Knapp and Manning (2016) conclude that a complex, many-sided scenario must be accepted. Millek’s recount removes part of the “everything at once” evidence these hypotheses were built to explain.
What the genomes say
No individual in any published study is labelled a Sea Person; the label is archaeological. The one direct signal near 1200 BCE is at Ashkelon: four infants from Iron Age I layers carry a component that resembles southern European populations, which is no longer found two centuries later (Feldman and colleagues 2019). Four infants, one site.
Inland, at Megiddo and Abel Beth Maacah, two post-1200 individuals show continuity. Cretan genomes show gradual admixture from the 17th to the 12th century BCE; the mainland evidence is limited, with Tiryns the main case, and cannot show the pace of change across Greece. There are no genome-wide data for Cyprus, Egypt, Troy, Hattusa or Ugarit at that date.
The Carpathians and the Danube
Two separate things are often joined. The tell settlements of the Carpathian Basin end between 1600 and 1500 BCE, some 300 years before the Aegean crisis. The later network of earthwork forts on the plains of the Banat and the Mureș (Cornești-Iarcuri, Sântana, Csanádpalota) shows ditch burning and infilling in the earlier 13th century and wide abandonment around 1200, with a mixture of absolute and relative dates. The excavators of Sântana date the burning of its whole fortification to about 1250 BCE, followed by abandonment; for the network as a whole, destruction and abandonment are told apart with difficulty.
The Romanian sites examined for this page give no destruction layer securely dated to about 1200 BCE. Sava (2020) puts the destruction of Sântana enclosure III around 1250, while another regional framework (Stavilă and Floca 2025) puts the destruction of the forts in 1200/1150–1050; the synchronisation is disputed. The idea of a northern homeland of the Sea Peoples, by way of the Urnfield culture or Naue II swords (a type of bronze sword), rests on the shapes of objects and on place names, and has been argued from both sides for decades.
How this was made
The frame of 148 places and 153 destruction claims is the appendix of Jesse Millek, “Destruction and Its Impact on Ancient Societies at the End of the Bronze Age” (Lockwood Press, 2023). The page recounts it: 59 + 35 + 33 + 26 = 153. Its text is not reproduced.
Coordinates come from the excavations’ own gazetteers and from Pleiades (CC BY). 27 places have none that could be checked and appear only in the list. Relief is ETOPO 2022 (public domain), coasts and rivers Natural Earth.
The blind check
20 of the 94 rejected claims were drawn with a fixed seed. The readers were AI research agents (Claude models); each was given only the site name, not Millek’s verdict, and searched and read open sources for the excavators’ own account. “Agreed” means no destruction was found reported, or one dated outside 1225–1175; “differed” means a report describing a destruction inside the window was found; the rest could not be settled: 11 agreed with Millek (no destruction reported, or one at another date), 2 differed (Iria and Tell Abu Hawam: the reports the readers found describe a destruction at about 1200, by pottery phase at Iria and in Balensi’s re-study at Tell Abu Hawam), and 7 could not be settled from sources that open. For Tell Abu Hawam, a later chapter by Millek (2025) reports the excavator Balensi’s own reading, which bears on the difference without settling it; the original result is kept and the reading is added to the card. The check can show the count of 94 to be too high; it cannot show it too low.
The clocks
Radiocarbon was calibrated on IntCal20 (Reimer and colleagues 2020) with the page’s own code; the widths printed are the total years inside the 95% highest-density set. Two single dates in Kaniewski and colleagues (2019) were recalculated: the page’s outer 95% limits are 1263–1054 and 1282–1055 BCE against 1260–1070 and 1275–1105 as printed there. The lower limits differ by 16 and 50 years; the published ranges may be trimmed and may rest on another version of the curve. The ranges of the destructions are as printed in each paper, read a second time from the page itself; nothing was re-modelled and no range was widened. A modelled level boundary, a single wood sample or a text that gives only an earliest possible date is not a destruction date and is kept out of the shared-years count. A 68% range is not an interval of the same strength as a 95% range, and the page never mixes them without saying so.
Limits of the editions
Short-lived seeds fix the year a plant stopped growing; they can predate the destruction by the time they were stored or used, and a later intrusion cannot be excluded. The Spanish and Chinese editions were translated by Codex and checked against the English by Claude models; no native reader has reviewed them. The Romanian was written for this page and has had no outside editor.
What this does not claim
It does not say there was no crisis, no migration, or no Sea Peoples. It says that the one date, one wave and one cause drawn on the usual map are not what the excavation reports and the clocks that do not depend on pottery, taken one by one, give. It says nothing about the places that were never excavated.
- Millek, J. 2023. Destruction and Its Impact on Ancient Societies at the End of the Bronze Age. Lockwood Press.
- Knapp, A. B. and Manning, S. W. 2016. Crisis in context: the end of the Late Bronze Age in the Eastern Mediterranean. American Journal of Archaeology 120(1): 99–149.
- Manning, S. W., Kocik, C., Lorentzen, B. and Sparks, J. P. 2023. Severe multi-year drought coincident with Hittite collapse around 1198–1196 BC. Nature 614: 719–724.
- Reimer, P. J. and colleagues 2020. The IntCal20 Northern Hemisphere radiocarbon age calibration curve. Radiocarbon 62(4): 725–757.
- Webster, L. and colleagues 2023, PLOS ONE 18(11): e0293119 (Gezer); Garfinkel, Y. and colleagues 2019, Radiocarbon 61(3): 695–712 (Lachish); Martin, M. A. S., Finkelstein, I. and Piasetzky, E. 2020, BASOR 384: 211–240 (Megiddo); Burke, A. A. and colleagues 2017, AJA 121(1): 85–133 (Jaffa); Webster, L. and colleagues 2018, Radiocarbon 60(1): 309–331 (Azekah).
- Feldman, M. and colleagues 2019. Ancient DNA sheds light on the genetic origins of early Iron Age Philistines. Science Advances 5(7).
- Millek, J. M. 2025. Phoenicia at the Late Bronze–Iron Age Transition. In van der Veen, P., Wallenfels, R. and James, P. (eds), Assyria and the West. Archaeopress: 208–210 (Tell Abu Hawam). Review of Millek 2023: Journal of Greek Archaeology 10 (2025), from p. 451.
- Edgerton, W. F. and Wilson, J. A. 1936. Historical Records of Ramses III (SAOC 12). Oriental Institute.