Sword in the Sky Quake

September 634 CE

by Jefferson Williams

Introduction     Textual Evidence     Archeoseismic Evidence     Tsunamogenic Evidence     Paleoseismic Evidence     Notes     Paleoclimate - Droughts     Footnotes     References


Theophanes, Michael the Syrian, and several other late sources describe an earthquake followed by a "sword in the sky" that appeared for 30 days during the beginning of the Muslim conquest of the Levant. Although there are chronological inconsistencies in the various accounts, it is possible to establish that this earthquake very likely struck in September 634 CE. It was described as occurring during the same year that the Rashudin Army invaded Palestine - ie 634 CE. Some authors provided a month (ilul - August/September) and an observation that the earthquake was followed by aftershocks. The "sword in the sky" appears to have been a comet which was observed by ancient Chinese and Japanese sources in September 634 CE.

Textual Evidence

Chronology of Theophanes

Theophanes wrote the Chronicle in Greek during the years 810-815 CE as a continuation of George Syncellus' Chronicle. In the entry for Annus Mundi 6124, we can read:
ANNUS MUNDI 6124 (SEPTEMBER 1, 632— AUGUST 31, 633) a.d. 624

Roman Emperor Herakleios: 31 years: year 23
Arab ruler Abu Bakr: 3 years: year 2
Bishop of Constantinople Sergios: 29 years: year 24
Bishop of Jerusalem Modestus: 2 years: year 2
Bishop of Alexandria George: 14 years: year 14

In this year Abu Bakr sent out four generals who, as I said before, were shown the way by the Arabs— they took Hira and the whole land of Gaza. Sergios had just come from Palestinian Caesarea with a few soldiers; he engaged the Arabs in battle but was the first one killed. So were three hundred of his soldiers. The Arabs withdrew after a decisive victory, having taken many prisoners and much booty.

In the same year there was an earthquake in Palestine. Also, a sign — known as an “apparition” — appeared in the southern sky. It was sword-shaped, and remained for thirty days, stretching from south to north and predicting the Arab conquest.

Theophanes provides a lot of dating information at the top of this quote some of which is, unfortunately, inconsistent. However, the earthquake takes place at the start of the Muslim conquest of the Levant and there are enough clues in the text, particularly regarding the exploits of the Islamic Army, to establish that the earthquake struck in 634 CE.

Muslim Invasion of Syria
Muslim Invasion of the Levant - from Wikipedia

The Islamic Army of this time was known as the Rashudin Army. They were part of the Rashudin Caliphate. The Rashudin Caliphate ruled from the death of Muhammad in 632 CE until it was supplanted by the Umayyad Caliphate in 661 CE. The first Rashudin Caliph was Abu Bakr. He ruled for ~27 months from after Muhammad's death (8 June 632 CE) until his own death (23 August 634 CE). From 632-633 CE, the Rashudin Caliphate fought and won the Ridda Wars, squashing internal opposition from rebellious Arabian tribes. In 634 CE, Abu Bakr sent four armies into Byzantine controlled Syria and Palestine. Each army had a different commander. Theophanes notes this when he states that the earthquake happened in the same year that "Abu Bakr sent out four generals". The battle in which Sergios died fighting the Rashudin Army was the Battle of Dathin which, according to Wikipedia, took place in February 634 CE although it may have occurred later in the year as other sources state that the 4 armies left Medina in April 634 CE and that the southern army crossed from Elat and marched to Gaza in May 634 CE. Either way, the battle happened in 634 CE and Abu Bakr was the caliph until his death in August 634 CE. This anchors the earthquake to 634 CE.

Chronicle by Michael the Syrian

Michael the Syrian (1126 AD - 1199 AD wrote The Chronicle covering "Creation" until his times. According to Ambraseys (2009), Michael provided two accounts of this earthquake. One of the entries, provided by Ambraseys (2009), is shown below
In the year 945 of the Greeks, there was a violent earthquake, in the month of ilul, and after the earthquake there was a sign in the sky; it appeared in the form of a sword stretching from south to north, and remained for thirty days. To many it seemed to signify the coming of the Taiyayz (Arabs)
The month of ilul in the Hebrew Calendar dates this earthquake to August/September. In another passage, also provided by Ambraseys (2009), we can read
`At this time [a.S. 946 (Arabs' defeat of Theodoric), before the fourth year of Omar] there was a great earthquake, and at the moment of the earthquake the sun was eclipsed. In this earthquake the church of the Resurrection and that of Golgotha fell, together with many other places. Modestus, the Chalcedonian bishop, rebuilt them.
The reported destruction of the Church of the Holy Sepulcher (aka Church of the Resurrection and that of Golgotha) during this earthquake is a false synchronicity. The church was destroyed by the Persians in 614 CE and restored in 629 CE by Modestus. It was intact when Omar took possession of Jerusalem in 637 CE (Le Strange (1890: p. 202).

Regarding the solar eclipse, NASA reports one solar eclipse whose path would have passed over the region in 634 CE. This annular solar eclipse occurred on 1 June 634 CE. Michael the Syrian's primary contribution here is he provided a month which, when added to Theophanes account, suggests that the earthquake struck in August or September of 634 CE. The solar eclipse may have been conflated with the earthquake.


al-Makin, a Coptic Christian, wrote al-Majmu al-Mubarak (The blessed collection) in Arabic between 1262 and 1268 CE. Two passages, both provided by Ambraseys (2009), mention the earthquake:
There was a great earthquake in Palestine for thirty days and also a great plague arose in the same place.' (al-Mak. HS. i. 2/19).
`There was a great earthquake in Palestine which lasted thirty days. The shock was violent and was followed by a great plague.' (al-Mak. HM. i. 2/20).
An earthquake lasting 30 days is presumed to describe an earthquake with an extended period of aftershocks.

Other Sources

Gregory Abu 'l-Faraj (Bar Hebraeus) and George Cedrenus also wrote about these events. Gregory Abu 'l-Faraj (Bar Hebraeus) also stated that the earthquake struck during the month of ilul (August/September). More details can be found in the Notes section.


Ho Peng Yoke (1962) has the following entry for a comet in September 634 CE:
20th September, 634.

On the 23rd day in the eighth month of the eighth year of the Chen-Kuan reign-period a (po) comet appeared at the Hsu and the Wei (11th and 12th lunar mansions) for 11 days before going out of sight [30th September].
(CTS 36/5b; HTS 32/5b; THY 43/la; WHTK 286/21b; W177.)

Williams follows the HTS which says that the comet appeared on a chia-tzu day (22nd September).

The Japanese records, "During the eighth month of the sixth year of Jomei-tenno [29th August to 27th September approx.] a (chhang-hsing) comet was seen at the south. During the first month of the seventh year [24th January to 22nd February, 635 approx.] the (hui) comet turned round and appeared at the E."
(Dainihonshi ch. 359; Nihongi tr. Aston vol. II pp. 166 and 167; K.)
This comet appears to date this earthquake fairly conclusively to September 634 CE. Although the sources may have succumbed to creating a forced synchronicity by having the comet appear immediately after the earthquake, even with a delay the earthquake likely struck in September. The description of a "sword in the sky" appears to be a poetic description of a comet where the use of the word sword ties it to the military exploits of the Rashudin Army adding allegory to the account; something which would have been understood by the Christian authors who wrote about it and their intended audience.

Archeoseismic Evidence

Archeoseimic damage is summarized below

Location Status
Qasr Tilah probable
Bet Sh 'ean possible
Tall Hisban/Heshbon possible
Jerash needs investigation
Pella possible and needs investigation
Monastery of Khirbet es-Suyyagh possible I = IX
Caesarea needs investigation
Mount Nebo needs investigation
Ein Hanasiv possible - needs investigation
Giv’ati Junction possible
Avdat/Oboda definitive - I = IX-X
Mizpe Shivta possible
Mezad Yeruham possible
Shivta possible I = VIII to IX epicenter a few tens of km. to the WSW or NE
Rehovot ba Negev possible
Saadon possible
Nessana possible
Mamshit (aka Mamphis) possible - I => IX epicenter to the Southwest
Haluza possible - I = VIII to IX waves arrived on NE-SW axis probably from NE, epicenter a few tens of kilometers away
Haluza possible - I = VIII to IX waves arrived on NE-SW axis probably from NE, epicenter a few tens of kilometers away

Archeoseismic Evidence is examined on a case by case basis below

Qasr Tilah

Qasr Tilah faulted birkeh
Broken Corner of the Birkeh at Qasr Tilah (photo by Jefferson Williams)

Haynes et al. (2006) examined paleoseismic and archeoseismic evidence related to damage to a late Byzantine—Early Umayyad birkeh (water reservoir) and aqueduct at Qasr Tilah and concluded that left lateral slip generated by several earthquakes cut through a corner of the reservoir and aqueduct creating displacement of the structures. They identified 4 seismic events which produced coseismic slip on the Wadi Arava fault and led to a lateral displacement of 2.2. +/- 0.5 m at the northwest corner of the reservoir (aka birkeh) and 1.6 +/- 0.4 m of the aqueduct. The first seismic event was dated to the 7th century. Haynes at al (2006) suggested it was caused by either the Sword in the Sky Quake (633/634 CE) or the Jordan Valley Quake of 659/660 AD - favoring the Jordan Valley Quake. There was a repair after this 7th century destruction indicating that the site was occupied when the earthquake struck. This suggests that the Sword in the Sky Quake struck the location since the location would likely have been occupied at the time - i.e. at the start of the Muslim conquest of the Levant. It is also possible that this location received damage from the Sign of the Prophet Quake (613-622 CE). At some point the site was abandoned. Haynes et al (2006) noted that archeological evidence at the site indicates that it was abandoned and was not occupied past the Early Umayyad Period (661-700 CE). They also noted that
MacDonald (1992) [] collected some Byzantine and Umayyad surface potsherds at the site and documented ruins of Byzantine houses (village) along the fan surface of Wadi Tilah.
It is not known if the location was still occupied or only partially occupied when the Jordan Valley Quake struck in 659/660 CE. If the site was abandoned around the same time as the archeoseismic sites in the Negev (~640 CE ?), it may have been empty enough not to have been repaired if the Jordan Valley Quake caused further damage. Because of the repair, it it is unclear how much lateral slip was produced.

Archeoseismic evidence for the Sword in the Sky Quake at Qasr Tilah is labeled as probable.

Qasr Tilah Trench Log A7
Figure 5 Schematic diagram of Trench A.7 north wall. Stratigraphic units are identified by lowercase letters. Faults are emphasized by heavy lines. Earthquakes are identified by Roman numerals, with IV as the oldest. Dashed lines indicate unexcavated portion of aqueduct floor. Haynes et al. (2006)

Qasr Tilah Trench Log A7 Stratigraphic Column Schematic
Figure 4 Schematic stratigraphic column of Trench A.7. Thicknesses of units are generalized from measurements of unit throughout the trench. Listed artifacts provide age control for constraining deposition and earthquake history in units where they were discovered. Age constraints come from radiocarbon data and typological dating of sherds. Haynes et al. (2006)

Bet Sh 'ean

Russell (1985) reported the following
Fitzgerald (1931:7) uncovered three Byzantine houses that had collapsed and burned in the early 7th century, sealing coins of Anastasius I, Justin II, Maurice Tiberius. and Phocas beneath their destruction debris. a temporal span ca. 491-610.

In the Byzantine monastery at Beth-shan, gold coins of Heraclius (610- 641) were sealed beneath similar collapse debris Fitzgerald (1939:2) .
Such damage could have also been the result of sacking by the Rashudin Army. Archeoseismic evidence at Bet Sh 'ean is labeled as possible.

Tall Hisban/Heshbon

Crushed Ceramics at Hesbon due to 7th century CE quake
Figure 16. Ceramic vessels crushed by fallen vault in Early
Islamic Room N.1. from Walker and Labianca (2003)

Walker and Labianca (2003:453-454) uncovered 7th century CE archeoseismic evidence which they attributed to the Jordan Valley Quake of 659/660 CE from an excavation of an Umayyad-period building in Field N of Tall Hisban. They report a badly broken hard packed yellowish clay floor which was pocketed in places by wall collapse and accompanied by crushed storage jars, basins, and cookware. Based on observed archeoseismic damage and the Earthquake Archaeological Effects Chart of Rodriguez-Pascua (2013:221-224), local intensity is estimated to VIII+ based on:
An excerpt from their article follows:
Two roughly square rooms, each approximately 4 x 4 meters wide and built against the inner face of the Hellenistic wall, occupied most of N.l and N.2. Masonry walls, four courses high, delineated the space. The original rooms were separated by what appears to have been an open air corridor; a door in the east wall of N. l and one in the west wall of N.2 allowed passage between the two rooms. The floors of these rooms (N.1: 18, N.2: 16) were made of a hard packed, yellowish clay, which was badly broken and pocketed in many places by wall collapse. Upper courses of the walls of the rooms had fallen onto the floor and crushed several large storage jars and basins and cookware (Fig. 16), dated in the field to the transitional Byzantine-Umayyad period. The only foundation trench identified (N.2: 25) yielded no pottery. The fill above these floors contained pottery that was late Umayyad and Abbasid in date. While it is not possible at this early stage of excavation to determine when this structure was first built, it was clearly occupied in the middle of the seventh century, suffered a catastrophic event, and was reoccupied (at some point) and used into the ninth century. Fallen architecture, crushed pottery, badly damaged floors that appeared to have "melted" around the fallen blocks, and wide and deep ash pits and lenses bare witness to a major conflagration. The most likely candidate for this is the recorded earthquake of 658/9, which was one of the most destructive in Jordan's history since the Roman period, rather than the Islamic conquests of the 630's (El-Isa 1985: 233).
As dating is no more precise than the 7th century CE, archeoseismic evidence at Tall Hisban is labeled as possible.


Although Ambraseys (2009) cites Russell (1985) as stating that there is archeoseismic evidence for this earthquake in Jerash (aka Gerasa), no such statements can be found in Russell (1985). Archeoseismic evidence at Jerash is labeled as needs investigation.


Walmsley (2007) attributes some archeoseismic destruction at Pella due to the Jordan Valley Quake although this date assignment seems tentative.
Excavations in the early 1980s identified six house units destroyed in the earthquake of 749. These houses represented the last phase in a long urban development that commenced with the complete redevelopment of living quarters on Pella's main mound in the first half of the sixth century (Watson 1992). The original arrangement consisted of four-metre wide gravelled streets set out on a formal grid, each street flanked by stone and mudbrick terrace-style houses two storeys high, prefaced in some places by shops. These streets, intended to serve local needs, were not equipped with colonnades or sidewalks. Although modified, the layout remained the same until an earthquake in 659-60 required a rebuilding of the quarter, in which the linear terrace houses were replaced by independent, self-contained units centred on one or more sizeable courtyards.
Walmsley (1982) discussed this in more detail noting that:
only in one trench (IVE) has the Sydney team excavated much below the A.D. 746/7 surface, producing evidence for at least three Byzantine and Umayyad architectural phases. Since an attempt to establish a detailed chronology for the whole Umayyad period on the basis of this one trench would be premature, the following account concentrates on the final phase in the life of urban Pella.


We turn now to a consideration of the layout and use of the buildings in Areas III and IV (figs 28-29 and end-plates 2-3). A dominant feature of Pella in the Byzantine and early Umayyad periods appears to have been streets with packed mud and pebble surfaces. One such street, 5 m wide, ran east-west through Area IV. From it, north and south, doorways gave access to dwellings, hence referred to as the North and South Buildings. But at some stage during the Umayyad period the street was cut by a wall which continued south to form the west wall of the South Building. Before this event it appears that this building had covered a considerably greater area; now to the west of the north-south Umayyad wall the earlier walls were razed level with the new and final occupation surface of a courtyard. Into this surface were dug lightly fired clay tabuns. Although the date of the demolition of the western sector of the South Building and of the construction of the north-south wall is uncertain, the slight build-up of de¬tritus on this surface points to a time not far removed from the final destruction of A.D. 746/7. Tentatively we ascribe these alterations to the period following the earthquake of A.D. 717.
The earthquake of 717 CE refers to an earthquake which Ambraseys (2009) and Guidoboni et. al. (1994) locate in Syria and Upper Mesopotamia. None of the sources mention specific localities except for a conflation mistake by Pseudo-Dionysius of Tellmahre. However, reports from Upper Mesopotamia suggests an epicenter far from Pella indicating that another closer earthquake may have been responsible for this tentatively identified and dated archeoseismic evidence. Archeoseismic evidence at Pella is labeled as possible and needs further investigation.

Monastery of Khirbet es-Suyyagh

Taxel (2009: 186) reports damage, based on ceramic and numismatic finds, around the middle of the 7th century CE concluding that "it is highly likely that the observed damage and subsequent repairs in Khirbet es-Suyyagh were caused by one or more earthquakes." Damage descriptions follow:
Damaged architectural remains can be recognised throughout the site. Signs of destruction and nearly immediate rebuilding combined with absence of signs of man-made violent actives are typical earthquake-related features.

The area of the large courtyard (Fig. 2.1:8-10) had been completely rebuilt after a destructive event. An earlier construction phase, which is observed south of the centre of the courtyard (Fig. 2.1:9), is covered by a later floor. Fallen masonry and subsequent repairs were observed in the southern part of the apse of the church, with its inner face remaining asymmetric. Since the damage is observed close to the foundations of the church it seems that the damage had a pervasive affect on the entire structure. A section of about 10 m in the southern end of W33 seems also to have been rebuilt. Similarly, in W100 there is a warped contact in room 19, where two different styles of masonry meet but are misaligned.

Another type of damage appears in two broken door thresholds, that of the main gate and that of the small courtyard in the south of the monastery. The large, monolithic and nicely carved stones are placed in-situ but broken by a width wise crack into two pieces. Assuming the thresholds were carved from intact rocks without significant fractures, we can envision strong vertical acceleration, perhaps of the order of lg, which caused the fracturing. Such strong shaking is known based on modern earthquakes to occur either near the epicentre of strong earthquakes (of the order of magnitude 7 and above) or in places with strong local amplification of seismic waves.

Each of the damaged elements alone would not suffice to indicate an earthquake as the damaging agent. However, the occurrence of many such elements, the extensive repair and reconstruction of features without any sign of human violence and in short time, together with the frequent occurrence of earthquakes in the region supports the association of the damage to earthquake/s.
This suggests that the Jordan Valley Quake, the Sword in the Sky Quake and/or possibly even the Sign of the Prophet Quake damaged the site with the Jordan Valley Quake the most likely candidate. Archeoseismic evidence at the Monastery of Khirbet es-Suyyagh is labeled as possible . Seismic Intensity is estimated at IX (1 g).


Without citing a source, Ambraseys (2009) states that
stratigraphic analysis of the site of Caesarea Maritima shows a destruction level dating to c. AD 630 it is not certain whether this can attributed to an earthquake or to a Persian invasion.
Archeoseismic evidence at Caesarea is labeled as needs investigation.

Mount Nebo

Ambraseys (2009) notes that
Indeed, Russell remarks that it is impossible to ascertain the effects of this and the AD 632 (634) earthquake on the Mt Nebo monastery owing to the manner in which the excavations were conducted.
However Russell (1985) correlates archeoseismic destruction at Mount Nebo to the Mount Lebanon Thrust Quake of 551 CE and the Sabbatical Year Earthquake of 746/749.

July 9, 551 CE entry - p. 45

This earthquake also appears to have been responsible for the destruction and subsequent abandonment of the Town of Nebo (Saller and Bagatti 1949: 217, n. 2).
January 748 CE entry - p. 49

The final destruction of the basilica at Mt. Nebo also appears to correlate with this earthquake (Schneider 1950: 2-3),
Notes - p.54

At Mt. Nebo (Sailer 1941: 45-46) and Aereopolis (Zayadine 1971) in the region of ancient Moab, recovery after the 551 earthquake apparently did not occur until the end of the century. Related to this delayed recovery is the possibility that an influx of southeastern populations from decaying urban centers like Petra subsequent to the 551 earthquake was responsible for the intensified building during the late 6th and early 7th centuries in both Moab (Sailer 1941: 248) and the Negev (Kraemer 1958: 23. 28-29; Colt 1962: 21-22).
This archeoseismic evidence is labeled as needs investigation.

Ein Hanasiv

Karcz et. al. (1977) list archeoseismic evidence (oriented collapse, alignment of fallen masonry) in Ein Hanasiv in the 7th century AD based on Vitto (1975). Archeoseismic evidence at Ein Hanasiv is labeled as possible and needs investigation.

Giv’ati Junction

Baumgarten (2001) excavated a round pottery kiln at Giv' ati Junction dated to the 4th-7th century CE (Shmueli (2013)). Langgut et al (2015) report that four fired Late Roman Amphora (similar to those at Yavne) "were found inside the kiln’s collapsed firing chamber" covered by a thick layer of aeolian sand. Langgut et al (2015) noted that while "the excavator suggested that the kiln was destroyed during operation, possibly due to some technical fault, and was consequently abandoned (Baumgarten 2001)", they believe an earthquake should also be considered as a cause of destruction.

(Shmueli (2013)) excavated Stratum III in a rectangular building (L109, L119) at Giv'ti Junction in 2011 where, on the floor, they found three Gaza jars which were set upside down (Fig. 4) and broken. A fourth jar was found upright but also broken. Based on numismatic finds, they dated the beginning of the settlement to the fourth or fifth century CE. Construction and use of the rectangular building was dated to the fifth to seventh centuries CE. In the seventh century the installation and building went out of use.

Archeoseismic evdience at Giv’ati Junction is labeled as possible.


In surveys conducted in 1994 and 1996, Korjenkov (1999) identified and examined seismic features such as


Avdat Wall Bulges
Bulges in SW Wall of Avdat Tower - photo by Jefferson Williams

Korjenkov (1999) identified a number of seismic features at Avdat and was able to produce an estimate of local Intensity and other information as follows : Unfortunately, this estimate is derived from multiple earthquakes. Korjenkov estimated that three seismic events created the features and that the first seismic event was the southern Cyril Quake of 363 CE and and the last event was the Sword in the Sky Quake of 634 CE. The middle event might be the hypothesized Negev Quake or the Monaxius and Plinta Quake.

However, Korjenkov and Mazor (1999) following Fabian ( 1996, 1997/8) estimated that
the Byzantine village of Avdat was destroyed by an earthquake between 631 and 633 AD with a local MSK-64 intensity of IX-X.
Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE. Avdat (aka Oboda) was abandoned in the 7th century; likely after the Sword in the Sky earthquake.

Mispe Shivta

Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE.

Mezad Yeruham

Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE.


Arch Deformation at Shivta
Fig. 3. (a) Asymmetric deformation of an arch trending ENE in a room at the SE corner of the Southern Church;
(b) A photograph of two parallel arches—deformation of the near one is portrayed in Fig. 3a. The two arches were
deformed to opposite directions, indicating that these arches responded to different phases of the seismic waves.
from Korzhenkov and Mazor (1999a)

Korzhenkov and Mazor (1999a) provided the following conclusions from a study of archeoseismic evidence at Shivta.


  1. The ancient city of Shivta is situated on flat low-land, built of massive carbonate bedrock. Hence, no site-effects are expected to have affected the patterns of seismic damage.
  2. Walls of buildings and agricultural fences trending SE (130°±15°) reveal collapse in a preferential direction towards the SW (Fig. 8), whereas walls oriented NE (40°±20°) reveal random collapse.
  3. This key observation indicates that the seismic waves arrived either from the SW (in the case of a compression wave), or from the NE, if the collapse happened in an extensional quadrangle (Korjenkov and Mazor, 1999a). In any case, the SE and NW directions of seismic wave propagation can be excluded.
  4. Rotations of blocks are observed at the Shivta ruins to be clockwise at walls trending NE (40°-50°), and counterclockwise at walls trending SE (115°-130°), as shown in Fig. 6c. Such rotations could be caused only by push movements by compression waves. Thus, the seismic waves arrived from the SW.
  5. The Shivta ruins disclose two main perpendicular directions of walls: NE (30°-60°) and SE (120°-150°), as can be seen in Fig. lc. Hence, all the buildings of the Byzantine city can be modeled via a "representative room" depicted in Fig. 16. Three possible scenarios warrant discussion: (a) seismic waves arrived parallel to the NE-trending walls (Fig. 16a) — the shear stresses along the walls would be minimal, and hence no rotation would be caused, and only collapse of NW walls would be systematic; (b) seismic waves arrived from the west, i.e., along a line of the bisector between the wall directions—both NE and SE trending walls would reveal oriented collapse to the NW and SW sides respectively; walls with a NE trend would reveal clockwise rotation, and walls with a SE trend would reveal a more or less equal number of counterclockwise rotations; (c) seismic waves arrived from the WSW, i.e., at a different angle to each of the wall directions — the SE walls would manifest systematic collapse generally toward the SW, whereas the NE walls would show random collapse; rotations of elements of walls trending NE would be clockwise, whereas rotations of stones of the SE-trending walls would be counter-clockwise. The field observations fit this solution (c).
  6. A few hundred individual observations, made at almost one hundred locations at the ancient city of Shivta, revealed the 19 types of damage patterns reported above. Part of these observations are useful in determining the axis along which the seismic waves propagated (WSW—ENE), other observations point out that the epicenter was located WSW of the city, and yet another group of observations points to an intensity of I= 8-9 (MSK-64 scale) of the earthquake that destroyed the Byzantine city in the 7th century.
  7. The distance of the epicenter of the earthquake that destroyed Byzantine Shivta can be estimated from the following boundary conditions and considerations: (a) the systematic pattern of destruction indicates dominance of horizontal seismic movements, which in turn rules out the possibility that the hypocenter was beneath the city (i.e., Shivta was not at site A of Fig. 17); (b) on the other hand, the dominance of a horizontal component of the seismic movements implies that the epicenter was at a distance that at least equaled the depth of the hypocenter (i.e., Shivta was at site B of Fig. 17); (c) the intensity I = 8-9 (MSK-64 scale) limits the distance of the epicenter probably to less that 30 km, a conclusion that has to be checked by data from more sites from the Negev, applying the "triangulation method".
  8. An attempt to locate the epicenter of the post-Byzantine earthquake at Shivta is made by applying the reconstructed WSW direction of the epicenter, and the concluded epicenter distance of a few tens of kilometers. These boundary conditions were projected on the geological map of Israel: the concluded direction of the epicenter crosses the Zin fault at a distance of 10 km, and the adjacent Nafha fault crosses with the direction of the concluded epicenter at a distance of 50 km. In any case, the results clearly point out that the epicenter was in the Negev highlands and not in the Dead Sea Rift Valley.
  9. The seismic damage patterns described so far were observed on buildings built in the Byzantine period and in secondary walls added later on, leading to the conclusion that at least two earthquakes damaged the Byzantine and post-Byzantine constructions.
  10. The described variety of seismic damage patterns provides tools to establish certain characteristics of the involved earthquakes, e.g., seismic intensity, axis of seismic waves propagation, and in the case of systematic rotation, also the specific direction of the epicenter. In a more advanced stage of the archeoseismological study, the investigations in individual sites can be put together into a regional picture that provides more definite answers on the nature of the studied earthquakes. For example, the Negev data from several ancient ruin centers may be compiled, e.g., Mamshit, Avdat, Rehovot, Haluza, Hurvat Sa'adon, Shivta, and Nizzana (Fig. 1). In other words, the triangulation approach is recommended (Korjenkov and Mazor, 1999a,b).
  11. The common descriptions of damage patterns typifying different earthquake intensities are based on the inventory of modern buildings. The present work brings up additional damage patterns observed in ancient architectural complexes, e.g., damage pattern of stone arches, systematic tilt, collapse and rotation of stone building elements, the distance to which collapse debris is thrown away from the respective foundation, as well as preferential collapse of colonnades observed in many published case studies.
  12. The described archeoseismological study has modern applications in regard to specifications of seismic safety to be taken into account in new constructions in the Negev highlands.
  13. Finally, the described archeoseismological work lends itself to inter-regional and international collaboration in the construction of a seismic archive that goes back thousands of years.
Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE. Archeoseismic evidence at this location for the Sword in the Sky Quake is labeled as possible.

Rehovot ba Negev

Tilted Walls at Rehobot ba Negev
Fig. 12 Rehovot ba Negev, northward tilting and shifting of the southern wall of the Northern Church
Khorzhenkov and Mazor (2014)

Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE.

Tsafrir (1988: 26) excavated the Northern Church (aka the Pilgrim Church) of Rehovot ba Negev and came to the following conclusions regarding its initial construction :
A clear terminus ante quem for the building of the church is given by a burial inscription (Ins. 2) dated to the month Apellaios 383, which falls, according to the era of the Provincia Arabia, in November- December 488 C.E. The church probably was erected in the second half of the fifth century. (See below the subsequent general discussion of the triapsidal basilicas beginning on p. 47.). Although it is clear that several parts of the complex were built later than the main hall, such as the northern chapel (see 111. 1 15), there is no doubt that the entire complex was constructed within the same few year.
Later on he noted that
A date of approximately 460-470 for the building activity therefore seems reasonable, although the calculation remains hypothetical.
After initial construction, additional architectural elements were added; foremost among them a a revetment or support wall which is described and discussed below by Tsafrir (1988: 27).
The most important architectural addition was the talus, or sloping revetment, that was built around the walls of the church from the outside to prevent their collapse. Such revetments were common in the Negev. They supported the walls of churches as well as of private houses. They are found, for example, around the walls of St. Catherine's monastery in Sinai. At Rehovot such walls may have been erected following an earthquake, but more probably it was necessary to reinforce them just because of poor quality masonry. To explain these retaining walls as having created a military defense post (as has been done in the case of the northern church at Shivta) is awkward.
Khorzhenkov and Mazor (2014: 84) identified what they believed were three (or more) earthquakes which had expressions in the walls of the northern church. The first two earthquakes struck after construction of the church around 465 CE and before the site was abandoned by its Christian inhabitants around 640 CE (when the Byzantine Empire permanently lost power in the area and could no longer support these peripheral outposts). A later earthquake struck during the Early Arab period - after ~640 CE.
The existence of revetment walls, supporting the southern wall of the Church from the south, indicates that the southern wall’s tilt occurred during the first of the Late Roman earthquakes. It seems that the southern wall began to tilt northward inside the building during the Early Arab earthquakes; additional evidence for this is the shift northwards of the upper part of the revetment wall. Stones of the perpendicular eastern wall are cracked in the small room marked on the plan. Nevertheless, this wall is better preserved (it is much higher) than the main southern wall of the North Church. This indicates that the seismic shocks during both earthquakes acted perpendicular to the main Church wall: it had freedom of oscillation and was significantly destroyed. The small eastern wall, oriented parallel to the effect of the seismic movements, withstood the seismic oscillations better, although many of its stones were significantly damaged. The whole northern wall of the Church (field station 12 in fig. 3) has a significant tilt to the south (figs. 8 a. b).
Khorzhenkov and Mazor (2014:84) discussed the two late Byzantine quakes (between 465 CE and 640 CE) further
The destruction event (an earthquake), which deformed the original wall, occurred before the decline of the Byzantine Empire. There was then another seismic event which led to the destruction of the revetment wall itself. The last event was probably an end of ›civilized‹ life here.
Rodkin and Khorzhenkov (2019) noted that

Another strong earthquake occurred during the 7th century AD. This could be the same earthquake that destroyed Avdat [ Korjenkov and Mazor (1999), Fabian (1998)]. This earthquake could also drive the inhabitants out of Rehovot, which occurred soon after (or slightly before) of the Arab conquest.
This suggests that the Late Byzantine earthquakes that struck Rehovot ba Negev could include some combination of the following

Archeoseismic evidence at this location for the Sword in the Sky Quake is labeled as possible.


Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE.


Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE.

In the New Encyclopedia of Archeological Excavations in the Holy Land Urman related the following regarding archeoseismic evidence form around this time:
Settlement at the site reached its highest point in the seventh century CE. The Muslim conquest of the area apparently did not harm Nessana’s Christian population. On the contrary, the largest church discovered at the site—the “Central” Church in the lower town (area F)—was built in the late seventh or early eighth century. It continued to function without interruption during the eighth century, and only at the end of that century, or early in the ninth century, did it pass into Muslim hands. Incidentally, there is no clear indication of a violent Muslim conquest of Nessana during the seventh and eighth centuries. While it is true that some of the residential buildings uncovered in the new excavations contained evidence of structural changes dated to those centuries, such as blocked doorways and the construction of new floor levels (in areas B, C, and R), it has not yet been determined whether these changes were a consequence of earthquakes in the region or were motivated by other circumstances. Nevertheless, the copious amounts of sherds and glass fragments, including luxury items recovered from these houses, clearly indicate that the economic status of Nessana’s population remained stable under Muslim rule.
Archeoseismic evidence at Nessana is labeled as possible.

Mamshit (aka Mampsis)

Tilted Wall in Mamshit
Fig.12 b. Tilting westward of upper stones of the N-S (174 degrees) trending east wall in a room
south of the main premises of the West Church – stone A has a dip of 61 degrees and stone B has a dip
of 74 degrees. "Seismic push arrived from SW"
from Korzhenkov and Mazor (2003)

Korzhenkov and Mazor (2003) estimated that an earthquake struck Mamshit in the 7th century. Although they also saw evidence that a late 3rd or early 4th century earthquake also struck the site, they found statistically meaningful directional preferences in the damage patterns that allowed them to separate the effects of the two different quakes. They estimated that the epicenter of the late 7th century earthquake was to the southwest of Mamshit and the minimum local intensity was IX.

They report that the percentage of collapsed buildings could be well estimated as the ruins were left untouched. Their estimates were that practically all of the buildings in Mamphis were damaged and more than 50% were destroyed in this earthquake. Almost no second floor structures survived without severe damage.

Erickson-Gini (personal correspondence, 2021) relates that this site in the Negev suffered seismic damage in the 7th century CE - sometime after 620 CE. Because this damage was dated to the 7th century, archeoseismic evidence for the Sword in the Sky earthquake is labeled as possible.


Fallen Column at Haluza
Fig. 13 Haluza. Collapse of one of the columns of the Cathedral in a NE (45º) direction.
Note lead bundle on the surface of the pedestal and traces of lead on the bottom of the fallen column
from Korzhenkov and Mazor (2005)

Korzhenkov and Mazor (2005) summarized archeoseismic evidence for a second destructive earthquake at Haluza as follows:
The Early Arab – "Second" Ancient Earthquake

Negev (1976) suggested that a strong earthquake caused the final abandonment of Haluza. He summed up his observations at one of the excavated courtyards:
Voussoirs of the arches and extremely long roof slabs were discovered in the debris, just above the floor. It seems that either the destruction of the house occurred for a very short time after its abandonment or the house had to be abandoned because of its destruction by an earthquake.
The date of destruction could be 632/3 A.D., i.e. at the time the Byzantine city of Avdat was destroyed and other ancient towns of the Negev desert were ruined. However, archaeological data demonstrate that occupation of the site continued until at least the first half of the 8th cent. A.D.. So, most probably, the date of destruction by an earthquake was 749 A.D., as the best match with the earthquake catalog of Amiran et al (1994)[1]

Joints crossing several adjacent stones (e. g. Fig. 4) indicate destruction by a high-energy earthquake, as the energy was sufficient to overcome the stress-shadow of the empty space between the building stones.

Tilts of the walls (Fig. 10), fallen columns (Fig. 13), shifted collapse of an arch (Fig. 11), shift of a stone row of the wall (Fig. 15) – all these observations disclose that the destructive seismic waves arrived along a NE-SW axis (~40º), most probably from NE.

Although all of the buildings in the city were well built and had one or two floors, all of them were severely damaged by an earthquake. The significant seismic deformations observed in the buildings indicate a local seismic intensity of at least I = VIII–IX (MSK Scale). This requires a strong shock arriving from a nearby epicenter, most probably a few tens of kilometers from Haluza. This supposition is based on the fact that short-period seismic waves, which tend to be destructive to low structures (which have short-period harmonic frequencies), attenuate at short distances from the epicenter.

[1] Note by Jefferson Williams : While Amiran et al (1994) is one of the error prone catalogs, 749 A.D. refers to well attested earthquake.
Archeoseismic evidence for the Sword in the Sky Quake at Haluza is labeled as possible.

Tsunamogenic Evidence

Paleoseismic Evidence

Paleoseimic damage is summarized below

Location Status
al-Harif Syria possible
Bet Zayda possible
En Feshka probable
En Gedi no evidence
Nahal Ze 'elim no evidence
Taybeh Trench possible
Qatar, Jordan no evidence

Paleoseismic Evidence is examined on a case by case basis below

Displaced Aqueduct at al Harif, Syria

Sbeinati et. al. (2010) report a seismic event Y which they dated to 657 AD +/- 32 years at a dispalced aqueduct at al-Harif, Syria (close to Masyaf, Syria).

Al Harif Aqueduct Seismic Events
Figure 13. Correlation of results among paleoseismic trenching, archaeoseismic excavations, and tufa analysis. In paleoseismic trenching, the youngest age for event X is not constrained, but it is, however, limited by event Y. In archaeoseismic excavations, the period of first damage overlaps with that of the second damage due to poor age control. In tufa analysis, the onset and restart of Br-3 and Br-4 mark the damage episodes to the aqueduct; the growth of Br-5 and Br-6 shows interruptions (I) indicating the occurrence of major events. Except for the 29 June 1170 event, previous events have been unknown in the historical seismicity catalogue. The synthesis of large earthquake events results from the timing correlation among the faulting events, building repair, and tufa interruptions (also summarized in Fig. 12 and text). Although visible in trenches (faulting event X), archaeoseismic excavations (first damage), and first interruption of tufa growth (in Br-5 and Br-6 cores), the A.D. 160–510 age of event X has a large bracket. In contrast, event Y is relatively well bracketed between A.D. 625 and 690, with the overlapped dating from trench results, the second damage of the aqueduct, and the interruption and restart of Br-3 and onset of Br-4. The occurrence of the A.D. 1170 earthquake correlates well with event Z from the trenches, the age of third damage to the aqueduct, and the age of interruption of Br-4, Br-5, and Br-6. Sbeinati et. al. (2010)

Al Harif Aqueduct Radiocarbon
Figure 12. (A) Calibrated dating of samples (with calibration curve INTCAL04 from Reimer et al. [2004] with 2σ age range and 95.4% probability) and sequential distribution from Oxcal pro-gram (see also Table 1; Bronk Ramsey, 2001). The Bayesian distribution computes the time range of large earth-quakes (events W, X, Y, and Z) at the Al Harif aqueduct according to faulting events, construction and repair of walls, and starts and interruptions of the tufa deposits (see text for explanation). Number in brackets (in %) indicates how much the sample is in sequence; the number in % indicates an agreement index of overlap with prior distribution. Sbeinati et. al. (2010)

Bet Zayda

Wechsler at al. (2014) may have seen evidence for this earthquake as Event CH3-E1 in paleoseismic trenches just north of the Sea of Galilee (aka Lake Kinneret).

Bet Zeyda Earthquakes
Figure 9. Probability density functions for all paleoseismic events, based on the OxCal modeling. Historically known earthquakes are marked by gray lines. The age extent of each channel is marked by rectangles. There is an age uncertainty as to the age of the oldest units in channel 4 (units 490-499) marked by a dashed rectangle. Channel 1 refers to the channel complex studied by Marco et al. (2005).

En Feshka
Kagan et. al. (2011) assigned a 634 AD date to a 1 cm . thick Type D seismite at a depth of 172.0 cm. depth.
En Gedi
Migowski et. al. (2004) did not assign a date of 634 AD to any of the seismites in the En Gedi Core (DSEn) but did assign a 0.5 thick seismite at a depth of 1.99 m to a date of 660 AD.

Nahal Ze 'elim
Kagan et. al. (2011) did not assign any seismites to a date of 634 AD.


Taybeh Trench, Jordan
LeFevre et al. (2018) might have seen evidence for this earthquake in the Taybeh Trench (Event E3).

Taybeh Trench Earthquakes
Figure S5: Computed age model from OxCal v4.26 for the seismic events recorded in the trench

Qatar, Jordan
Klinger et. al. (2015) did not identify any seismic events the Qatar trench which might correspond to this earthquake.

Qatar Trench
Figure 6. Age model computed for the trench stratigraphy using OxCal v4.2 (Bronk-Ramsey et al. 2010) and IntCal13 calibration curve (Reimer et al. 2013). Light grey indicates raw calibration and dark grey indicates modelled ages including stratigraphic information. Phases indicate subsets of samples where no stratigraphic order is imposed. Klinger et al (2015)


Chronicon by Bar Hebraeus (Gregory Abu 'l-Faraj)

Gregory II Abu 'l-Faraj bar Ahron (aka Gregory Bar Hebraeus) wrote Chronicon in Syriac in the 13th century where one can read:
Heraclius transgressed the Law and took Martina, his brother's daughter, to wife, and begat by her an illegitimate son, Herakluna. In the same month of Elul [September], an earthquake took place. And a sign, like unto a spear, appeared in the heavens, and it remained there for 30 days. This manifestly made known the victory of the Arabs.
The passage relates an earthquake striking in September followed by a "sword (or spear) in the sky" which lasted for 30 days. There is no synchronism of the earthquake with Heraclius marriage to his niece in Bar Hebraeus' account. The author stated that the earthquake occurred in September which was the same month as when Heraclius married his niece. He does not mention that these two events are separated by ~21 years. The events are juxtoposed in the text by use of a common month to make a theological point - that Heraclius' transgression with his niece is linked somehow with the way the earth reacted to these events with the assumption that the earth was manifesting something from the supreme being.

Synopsis Historion by Cedrenus

George Cedrenus wrote Synopsis Historion (aka A concise history of the world) in the 1050's in Greek. In Cedrenus (1838: 745), he noted that a comet was observed for 30 days some time after Muhammed died.

Ambraseys (2009) notes

Ambraseys (2009), citing Grumel (1934), noted that Theophanes started his years (ie indictions) in September rather than March during this time period. He also provided the following chronological notes

Russell (1985) notes

The comet mentioned by Theophanes, Cedrenus, and Michael the Syrian would not have been Halley's. since it would have returned to earth ca. March 13.607 and again ca. September 28, 684. The changing form of the comet, which first appeared in the shape of a cross and later assumed the form of a striking sword reflects its movement at perihelion.

Eclipse Path

Annular Solar Eclipse on 1 June 634 CE comes from Eclipse Predictions by Fred Espenak and Chris O'Byrne (NASA's GSFC).

Paleoclimate - Droughts



Ambraseys, N. (2009). Earthquakes in the Mediterranean and Middle East: a multidisciplinary study of seismicity up to 1900. Cambridge, UK, Cambridge University Press.


Fabian, P.: 1995, The Roman and Byzantine Near East: Some Recent Archaeological Research, J. Roman Archaeology, Suppl. Series, No. 14, pp. 235-240.


Fabian, P.: 1996, Evidence of earthquake destruction in the archaeological records the cast of ancient Avdat, In: Big Cities World Conference on Natural Disaster Mitigation in Conjunction with the 10th International Seminar on Earthquake Prognostics, Abstracts, Jan. 5-10, 1996, Cairo, Egypt, p. 25.

Guidoboni, E., et al. (1994). Catalogue of ancient earthquakes in the Mediterranean area up to the 10th century. Rome, Istituto nazionale di geofisica.

Guidoboni et. al. (1994)

Haynes, J., et al. (2006). "Evidence for ground-rupturing earthquakes on the Northern Wadi Araba fault at the archaeological site of Qasr Tilah, Dead Sea Transform fault system, Jordan." Journal of Seismology 10(4): 415-430.


Kagan, E., et al. (2011). "Intrabasin paleoearthquake and quiescence correlation of the late Holocene Dead Sea." Journal of Geophysical Research 116(B4): B04311.


Karcz, I., et al. (1977). "Archaeological evidence for Subrecent seismic activity along the Dead Sea-Jordan Rift." Nature 269(5625): 234-235.


Ken-Tor, R., Agnon, A., Enzel, Y., and Stein, M. (2001). "High Resolution Geological Record of Historic Earthquakes in the Dead Sea Basin." Journal of Geophysical Research 106(B2): 2221-2234.

Korjenkov AM, M. E. (1999). "Seismogenic origin of the ancient Avdat ruins, Negev Desert, Israel." Natural Hazards 18: 193-226.


Migowski, C., et al. (2004). "Recurrence pattern of Holocene earthquakes along the Dead Sea transform revealed by varve-counting and radiocarbon dating of lacustrine sediments." Earth and Planetary Science Letters 222(1): 301-314.


Russell, K. W. (1985). "The Earthquake Chronology of Palestine and Northwest Arabia from the 2nd through the Mid-8th Century A.D." Bulletin of the American School of Oriental Research 260: 37-59.


Wechsler, N., et al. (2014). "A Paleoseismic Record of Earthquakes for the Dead Sea Transform Fault between the First and Seventh Centuries C.E.: Nonperiodic Behavior of a Plate Boundary Fault." Bulletin of the Seismological Society of America.


Zohar, M., et al. (2016). "Reappraised list of historical earthquakes that affected Israel and its close surroundings." Journal of Seismology: 1-15.




al-Makin. HM: al-Makin b. al-`Amid, Kitab al-majmu' al-mubarak, ed. and Latin trans. Th. Erpenius, Historia Saracenica, Leiden, 1625; also French trans. Histoire Mahomatane, trans. P. Vattier, Paris, 1657.

al-Makin. HS: al-Makin b aliAmid, Ta'rikh al-muslimin min sahib shari` at al-islam ila al-dawlat al-atabiakiyya, Historia Saracenica, trans. Th. Erpenius, Lugd. Batavorum, 1625.

Cedrenus, G., et al. (1838). Georgius Cedrenus, Ioannis Scylitzae ope, E. Weber.


Michael The Syrian Chronicle.


Syrian, Michael the (1963). Chronique 4 volumes N. Chabot. Brussels.

https://archive.org/details/ChroniqueDeMichelLeGrand https://archive.org/details/ChroniqueDeMichelLeSyrienT.1Fasc.1translation https://archive.org/details/MichelLeSyrien2 https://archive.org/details/MichelLeSyrien3 http://www.roger-pearse.com/weblog/2009/02/26/michael-the-syrian-preface-to-his-history/ http://en.wikipedia.org/wiki/Michael_the_Syrian

Mango, C. A., et al. (1997). The chronicle of Theophanes Confessor: Byzantine and Near Eastern history, AD 284-813, Clarendon Press.