Aerial View of Avdat Avdat Acropolis (left) Orthophoto of Avdat from Click on Image for high resolution magnifiable map

(right) Oblique Aerial View of Avdat Acropolis from Wikipedia


Transliterated Name Source Name
Avdat Hebrew עבדת‎‎
Abdah Arabic عبدة‎
Oboda Ancient Greek ‎‎Ὀβόδα
Ovdat ‎‎
Obodat ‎‎
Abde ‎‎

Avdat started out in the 3rd or 4th century BCE as a Nabatean way station on the Incense Road (Avraham Negev in Stern et al, 1993). By the 1st century BCE, the town was named Oboba after Nabatean King Obodas I. It was occupied continuously until it was abandoned perhaps as late as the 9th or 10th centuries CE. Situated at the end of a ~4 km. long ridge, Avdat may have suffered from seismic amplification during past earthquakes as it may be subject to a topographic or ridge effect (terrain map ).

Maps and Plans
Maps and Plans

  • Annotated Satellite Image (google) of Avdat from
  • Plan of Avdat from
  • Aerial Overview of Avdat from Zion et al (2022)
  • Avdat Settlement Plan from Zion et al (2022)
  • Avdat/Oboda General Plan from Avraham Negev in Stern et al, 1993

Chronology of earthquakes at Avdat

Archeological excavations have uncovered several earthquakes which struck Avdat/Oboda. Erickson-Gini, T. (2014) noted approximate dates and Intensities:

  1. Substantial destruction in the early 2nd century CE
  2. Some damage due to an earthquake in 363 CE.
  3. A massive earthquake in the early 5th century CE
  4. A massive earthquake in the early 7th century CE [JW: recent work (2022) may suggest this should revised to the 7th century rather than the early 7th century]
Korjenkov and Mazor (1999) conducted two archaeoseismic surveys at Avdat and were able to distinguish between 7th century CE seismic effects and effects from a "previous" earthquake where the "previous" earthquake would likely be the massive earthquake in the early 5th century CE.

Early 2nd century earthquake

Erickson-Gini, T. (2014) described the early 2nd century earthquake as follows:

There is indirect evidence of a more substantial destruction in the early 2nd century CE in which residential structures from the earliest phase of the Nabataean settlement east of the late Roman residential quarter were demolished and used as a source of building stone for later structures. Destruction from this earthquake is well attested particularly nearby at Horvat Hazaza, and along the Petra to Gaza road at Mezad Mahmal, Sha'ar Ramon, Mezad Neqarot and Moyat `Awad, and at `En Rahel in the Arava as well as at Mampsis (Korjenkov and Erickson-Gini 2003).
Erickson-Gini and Israel (2013) added
Evidence of an early second-century CE earthquake is found at other sites along the Incense Road at Nahal Neqarot, Sha'ar Ramon, and particularly at the head of the Mahmal Pass where an Early Roman Nabataean structure collapsed (Korjenkov and Erickson-Gini 2003; Erickson-Gini 2011). There is ample evidence of the immediate reconstruction of buildings at Moyat ‘Awad, Sha'ar Ramon, and Horvat Dafit. However, this does not seem to be the case with the Mahmal and Neqarot sites.
Erickson-Gini and Israel (2013) discussed seismic damage at Moyat ‘Awad due to this earthquake
The Early Roman phase of occupation in the site ended with extensive damage caused by an earthquake that took place shortly before the Roman annexation of the region in 106 CE (Korjenkov and Erickson-Gini 2003). The building in Area C and the kiln works were destroyed, and the cave dwellings were apparently abandoned as well. Reconstruction was required in parts of the fort. At this time, deposition from its floors was removed and thrown outside of the fort and a new bath as well as heating were constructed in its interior. Along its eastern exterior and lower slope, rooms were added. Thus, the great majority of the finds from inside the fort and its ancillary rooms date to the latest phase of its occupation in the Late Roman, post-annexation phase, the latest coins of which date to the reign of Elagabalus (219–222 CE).

Southern Cyril Quake (363 CE)

Tali Erickson-Gini in Stern et al (2008) provided some information on the southern Cyril Quake of 363 CE.

In 1999–2000 an area located east of the Byzantine town wall and the north tower at Oboda was excavated on behalf of the Israel Antiquities Authority.
Some structural damage, probably resulting from the 363 CE earthquake, is evident in the blockage of a few doorways and the collapse of one of the rooms (rooms 4, 7, 17).
one room of the earlier structure appears to have been utilized in the fourth century CE (room 7), and it apparently collapsed in the 363 earthquake.

the numismatic and ceramic evidence uncovered in this third phase indicate that the dwellings were destroyed in a violent earthquake several decades after that of 363 CE. Following this second, local earthquake, the area was abandoned and many of the building stones were robbed.
The second earthquake could be due the Monaxius and Plinta Quake of 419 CE which fits as the early 5th century earthquake discussed below.

The "previous" earthquake - Early 5th century CE or 5th century CE

An early 5th century earthquake suggests the Monaxius and Plinta Quake of 419 CE where there appears to be archaeoseismic evidence in Yotvata. Erickson-Gini (2014) described the early 5th century earthquake at Avdat/Oboda:

A massive earthquake took place in the early 5th century CE, substantial evidence of which was uncovered in the late Roman and early Byzantine residential quarter (Erickson-Gini 2010a: 91-93). All of the structures east of the town wall were abandoned and used as a source of building stone for the late Byzantine town. Following this earthquake, massive revetment walls were constructed along the southern wall of the acropolis in order to shore up the heavily damaged walls. In contrast, the late Byzantine citadel adjoining the temenos area of the acropolis has no revetment walls, certainly due to its construction following the earthquake. The two churches inside the temenos area were built using numerous early Roman ashlars and architectural elements originally from the Obodas Temple damaged in the earthquake.
Negev (1989) provided a wider range of dates for this earthquake which entertains the possibility that this archaeoseismic evidence was caused by the hypothesized Negev Quake which, if real, is dated to around 500 CE.
A decisive factor in determining this phase is the dating of a series of earthquakes, one or more of which shattered numerous buildings in some of the towns of the central Negev. Although literary evidence is scarce, there is ample archaeological evidence that testifies to these disasters. At Oboda the entire length of the old southern Nabatean retaining wall was thrust outwards, and for this reason it had to be supported by a heavy, slanting supporting wall. Similarly much damage was caused to a massive tower of the Nabatean period, identified in July 1989 as the temple of Obodas (?), which in the Late Roman - early Byzantine period was incorporated in the citadel occupying the eastern half of the acropolis hill. Most of the damage was caused to the western and southern walls of the temple, and for this reason these too had to be supported by still heavier stone taluses, blocking the original entrance to the temple on the southern wall. It is against this talus that the South Church was built. Similar damage was also caused to some of the nearby buildings in the so-called Roman Quarter south of the temple. We may thus place the date of the earthquake between the end of the third century A.D., when the latest building in this quarter was constructed, and A.D. 541, when the Martyrium of St. Theodore was already being used as a burial ground.
Zion et al (2022:30) reports that the quarters for the soldiers of the Roman army camp, located east of and adjacent to the Acropolis, were destroyed in an early 5th century CE earthquake.

7th century earthquake

Erickson-Gini, T. (2014) discussed the early 7th century earthquake.

The destruction of the town by a massive earthquake sometime in the early 7th century CE was one piece of a puzzle not mentioned by Negev. The earthquake certainly occurred after the latest inscription found at the site in the Martyrion of St. Theodore (South Church) in 617 CE (Negev 1981: 37). Direct evidence of the destruction and abandonment of the site was uncovered by Fabian, with massive destruction evident throughout the site, and particularly along the western face of the site with its extensive caves and buildings (Korjenkov et al., 1996). Mezad Yeruham, several kms further south, was apparently destroyed at the same time (Y. Baumgarten, personal communication), while the earthquake left a trail of damage at numerous sites. This is indicated by the early seventh-century construction of revetment walls around churches and private houses at Sobota (Shivta), Sa'adon, Rehovot in-the-Negev, and Nessana. Compared to other Nabataean sites in the Negev Highlands that indicate a continued occupation through the late Byzantine period well into the early Islamic period in the 9th c., Oboda was devoid of settlement in the early Islamic period. In place of a central town, such as Sobota (Shivta), Rehovot in-the-Negev, or Nessana, a significant number of early Islamic farming villages—many with open-air mosques—were found in close proximity to Oboda.
Bucking et al (2022:1) uncovered evidence in the caves on the slope for substantial human activity at the site during the Early Islamic period, c. AD 650–900. Investigations focused on the 2ha residential sector, which comprises hundreds of rock-hewn dwellings built in terraces along the northern, southern and western slopes immediately below this necropolis (Shereshevski 1991: 38–42; see Figure 2 ). Site chronology was established using radiocarbon dating due to a sparsity of ceramics. These results were described as follows:
Radiocarbon results were surprising: while a sample of straw extracted from mortar in the dipinti-intensive cave vestibule wall yielded dates within the expected Byzantine range, nine samples from the organic, dung-enriched south-western section were of Umayyad–Abbasid date, ranging from 1190–1320±30 BP/AD 650–890 (Figure 4 ) (Bucking & Erickson-Gini 2020). Three newly obtained dates from the northern and western sections yielded Abbasid dates (Figure 5 ): two from a substantial organic layer in the northern section, and one in the western section, where Abbasid Mefjer/Buff Ware (Cytryn-Silverman 2010) further corroborated this result (Figure 6). The emerging extended chronology enriches ongoing debate surrounding the fate of Negev Highland settlement in the Early Islamic period, especially that regarding continuities and discontinuities of site use between the Byzantine, Umayyad and Abbasid periods (e.g. Magness 2003; Avni 2008; Butler et al. 2020).

Seismic Effects
Seismic Effects

In surveys conducted in 1994 and 1996, Korjenkov and Mazor (1999) examined hundreds of deformation features and selected 41 measurements of wall inclinations, 26 of wall collapse, 17 of block rotations, and 96 cases of through-going fractures, where [they] were certain of the non-static origin of dislocations. They divided the features of seismic destructioninto 2 groups based on diagnostic use.

  1. Seismic-related features, which can be used for the determination of the seismic origin of the destruction, and degree of seismic shaking - seismic intensity
    1. joints crossing through a few adjacent blocks
    2. rotation of arch or roof slabs around horizontal axis
    3. hanging stones in the arches
    4. later built supporting walls for the tilted walls and columns
    5. non-coincidence of lower rows of masonry with upper building construction
  2. Seismic indicators which can be used for the determination of epicentral direction
    1. inclination of walls
    2. shifting of complete walls or wall fragments
    3. collapse of arches and wall fragments
    4. rotation of building fragments in arches and walls around the vertical axis
Examples and summaries of observations are presented below:
Damage Type
7th century
Location Figure Comments
JOINTS AS AN INDICATION OF THE SEISMIC NATURE OF THE DESTRUCTIONS 7th century Northern Church 4 Joints are mode 1 (dilatation) fractures developed as a result of extension (Engelder and Fisher. 1996). Joints confined to stone breaks often appear in old buildings. Interpretation of such joints is somewhat ambiguous: they could be erected tectonically, they could also be the result of weathering, i.e., repeated heating and cooling events. In contrast, joints passing through two or more adjacent blocks (through-going joints) could be formed only under high strains. Such joints require the application of tremendous amounts of energy to overcome the stress shadows, appearing along free surfaces at the block margins (Fisher et al., 1995: Engelder, and Fisher, 1996; Becker and Gross, 1996) and therefore cannot be related to the weathering process.
Numerous examples of through-going joints were observed during the study of the ruins of Avdat town. One such joint was found in the WSW external wall of the Northern Church (trend azimuth is 150°) in a corner of a small ledge (Figure 4). The joint crosses two adjacent blocks with a thickness of 50 cm each. What is most important in this case, is that the joint has passed straight through cement between the two blocks, without any bends. The length of the joint is 1 m. It starts 30 cm in from the upper corner of the upper block and it finishes 70 cm in from the lower corner of the lower block. The joint is inclined by an azimuth 174° L59° in its upper part, dip azimuth is 173° L68° in its lower part.
All of the above is evidence of an earthquake which took place in the region of Avdat town in the 7th century A.D., probably 631-633 A.D. However, there is other evidence in the town, dating back to the Late Roman period, of at least one more strong seismic event, probably the well known earthquake of 363 A.D. (Amiran, 1950-1952; Russell, 1980; Amiran et al., 1994), which terminated the Late Roman settlement of the city. Several years later, a new town was rebuilt on the ruins of the old one. This idea was suggested by P. Fabian (1996, 1997). Our study has confirmed his suggestion.
Strange discordance of trends of first lower rows of masonry (usually one or two rows) and upper wall fragments is visible in some parts of Avdat. For example, there is counterclockwise rotation of the whole NW wall of room No. 10 of the court (see, Figure 3). Horizontal displacement was 45 cm. During rotation around the vertical axis the NW wall was not collapsed and townsmen, who settled there after the 363 A.D. shock, used the rotated wall for rebuilding (Fabian 1996, 1997). The original trend of the wall was 50°, preserved first and second lower rows testify about that building (Figure 5). Modern trend azimuth of rotated wall is 41°.
In some places, one can see a sharp deviation of trends for separate walls joining to each other perpendicularly. Such deviations can sometimes amount to an angle of 11° (see, for example, SE wall of room No. 2 of the court on the Figure 3).
The shift of the building elements without rotation may be used in a similar manner to wall inclination or block collapse. The upper element of a construction is shifted toward or away from an epicenter due to inertia. In the Avdat such a displacement, of 80 cm, can be observed for the upper fragment of the NW wall of room No. 8 of the court (see, Figure 3) in a NW direction (Figure 6). Its former position (trend azimuth is 41°) is marked by one stone row of 20 cm height. The width of the shifted wall fragment is 70 cm, length is 165 cm, height of preserved fragment is 55-60 cm, its trend azimuth is 45°.
These facts apparently testify to the adaptation of the lower non-destroyed rows of masonry and preserved walls (only rotated slightly) for the regeneration of the town in Byzantine times. During Roman times at the same place, there was a settlement which was destroyed by an earthquake. Later the town was, again rebuilt on the site of the former settlement using the preserved lower rows of masonry and preserved whole walls (Fabian, 1996, 1997).
Additional indirect evidence of possible seismic activity in the studied territory is non-coincidence of lower stone rows with upper building structures. Such patterns occurred when a building was partly destroyed during an earthquake, but ancient people decided not to restore it. They removed still standing preserved fragments of the destroyed building and smoothed out the piles of rubble. They built a new building on the site of the old one. Later, during recent archeological excavations, researchers discovered strange non-coincidence of lower stone rows with upper building structures (Fabian, 1996, 1997).
For example, such non-coincidence can be observed in the northern yard of the bath-house, which is located near the foot of the Avdat hill (Figure 7). The bottom row of the NW corner of the wall is pulled out to the west 13 cm if compared with the upper fragment of the wall, with the trend azimuth of 159° (see, Figure 7(a)). This non-coincidence is even larger - 28.5 cm if compared with the SE part of the wall, with the trend azimuth of 167°. The lower pulled row of the northern fragment of the wall continues to the NW over the perpendicular external wall of the yard (see Figure 7(b)). The probable explanation of this case is given in the previous paragraph.
SUPPORT-WALLS "Previous" Southern Church 8 Indirect evidence of more old shocks are special support-walls which were built solely for this purpose. One such wall was built to support the eastern corner of the Southern Church (P. Fabian, 1994, personal communication). The wall which needed support had an ENE trend (Figure 8). One more support-wall was built to support the external wall (with NE strike) of the South Quarter of the town, opposite the eastern corner of the Fort, later it was dismantled by archeologists during excavation (P. Fabian, personal communication, 1996). This building of supporting walls for city walls of the same trend is not isolated. Apparently, during the Roman earthquake these city walls were slightly tilted, but they were not collapsed. Ancient people built those support-walls specifically to prevent them from possible future collapse (Fabian, 1996, 1997).
CAVE DESTRUCTIONS "Previous" Caves As stated above, on the slope of Avdat hill there are many caves which were inhabited for living during Nabatean—Byzantine times. However, below the caves there are huge piles of rubble, which consist of debris from Avdat hill's rocks and from remains of domestic objects (pieces of Nabatean earthenware vessels, for example - T. Gini, personal communication, 1996). This fact also indicates a possible earthquake in 363 A.D. during which the collapse of inhabited caves took place. After that event ancient people cleaned out the caves and used them for living in for the second time. However, some of the caves were not cleaned after the 363 A.D. shock.
The caves near the top of the hill were the most severely damaged (T. Gini, 1996, personal communication). This fact can be explained by the "sky-scraper effect - maximum oscillation during earthquakes is in the upper part of the building (or the hill in the Avdat case).
A study of habitable (in the past) caves was made. They were dug up on a hill slope, on top of which there are main town buildings. This study shows numerous collapses of walls and cave vaults, and also considerable long fractures. The displacement of chisel traces on the cave ceilings was observed, where those traces are crossed by long fractures in limestone massif . The latest ones show subsidence on the first few centimeters of the middle parts of the limestone hill compared to the external parts. It is the opposite to what one would expect due to gravitation forces. Such graben-like subsidence of watershed parts of mountain ridges was observed during strong earthquakes in the Baikal Rift area (Khromovskikh, 1965) and in the Tien Shan seismic belt (Korjenkov and Chedia, 1986; Korjenkov and Omuraliev, 1993; Ghose et al., 1997). These seismogenic features are indicators of an earthquake intensity of IX—X.
The new Byzantine town existed until the beginning of the seventh century A.D., probably 633 A.D., and was then totally destroyed by an earthquake never to be rebuilt (Fabian, 1996, 1997). This may explain the absence of any Early Muslim period finds at the site in spite of the continued occupation of other Negev sites such as Nessana and Shivta (see Figure 1) that existed until the tenth century A.D. (E. Oren, personal communication, 1996). These towns were located west of Avdat and were probably less affected by the earthquake.
The following are the seismic features belonging to group 2, used for the determination of the seismic wave propagation direction. They belong to the seismic event which occurred in the 7th century.
INCLINATION OF BUILDING AND CONSTRUCTION ELEMENTS mostly 7th century ? various locations 9
As in strong earthquakes throughout the world, a large number of structural elements were found to be preferentially inclined (Richter, 1958; Cloud and Scott, 1969; Bolt, 1978; Polyakov, 1978; Omuraliev et al., 1993a and others). A similar destruction was found in the ancient city of Avdat: forty one cases of preferentially inclined walls (Figures 9 and 10) and inclination of single stones within walls can be seen there. As seen in Figure 5, walls trending SE 130°-140° are systematically inclined to the SW. In contrast walls trending NE 40°-60° are inclined to NW and SE with no preferential direction. This observation seems to indicate that the seismic shock arrived along the NE—SW direction: the walls oriented roughly normal to the seismic wave direction were systematically collapsed or inclined, whereas walls oriented parallel to the seismic waves lost support, were tilted and collapsed randomly.
COLLAPSE FEATURES 7th century ? Agricultural Fences 11a
Numerous ruins of agricultural fences remained on the top (Figure 11(a)) and near the foot of the Avdat hill (Figure 11(b)). The fences trending about EW reveal a clear systematic picture of the collapse: the lower part of the wall is intact (easily seen from its northern side), whereas the upper part of the fences fell southward (see Figure 11). Azimuth of preferred collapsed features are plotted in Figure 12 versus wall trend. One group of walls trending SE 90°-140° reveals collapse toward SW 180°-240°, whereas walls oriented in other directions fell on both sides of the original wall position, they did not show a systematic pattern of the collapse, and so they were not shown on the graph. This observation indicates that the direction of seismic wave propagation was roughly perpendicular to the SE-trending walls.
It is necessary to mention the cases of wall drags (rotations) because of wall collapse. Many rotated blocks or block fragments in Avdat were caused by the drag due to the collapse of a wall (Figure 13). Such rotations cannot be used to determine shear stresses, however the patterns of drag-caused rotations enable us to reconstruct the direction of wall collapse.
ROTATION OF BUILDING ELEMENTS 7th century ? various locations 13
Field study of the epicentral zones of the well-known strong earthquakes revealed that some building constructions or rock fragments were rotated clockwise, whereas others were rotated counterclockwise (Richter, 1958; Cloud and Scott, 1969; Bolt, 1978: Polyakov, 1978; Omuraliev et al., 1993b and others). Horizontal rotation of arch supports, separate blocks in arch supports and walls, or rotation of a large fragment of a wall with tens to hundreds of stones were measured in the ruins of Avdat town. Clockwise and counterclockwise patterns of rotation were observed. Some examples of the rotated elements are shown in Figure 14.
For the case of the Avdat ruins the pattern and degree of rotations were plotted against the wall trends (Figure 15 ). As can be seen in the graph, the only one case of clockwise rotation was found in a wall fragment with trend SE 140°, whereas counterclockwise rotations were found on walls trending NE 40°-60°.
The rotations described above were measured in well-preserved walls at some distance from the corners, so that a researcher could be confident, that the rotations were caused by a shear couple. However, many rotated blocks or block fragments in Avdat were caused by a drag which occurred due to collapse of a wall (see Figure 13). Such rotations cannot be applied to determine shear stresses, however, the patterns of drag-caused rotations enable us to reconstruct the direction of wall collapse, which, as described above, is an independent kinematic indicator.

Archaeoseismic Analysis
Archeoseismic Analysis

Korjenkov and Mazor (1999) provided an extensive discussion regarding the analysis of their data. This discussion provides information for Avdat and explains the methodology used to examine archaeoseismic observations from other sites in the Negev. Due to it's value as a reference, much of the discussion is repeated below:

Study of the destruction in the Avdat ruins reveals a systematic type of dislocation:
  1. Walls of buildings trending SE 120° revealed strong preferential collapse or inclination toward south, whereas walls trending NE 20°-50° tilted and fell without a noticeable systematic pattern (see Figure 10 ). A similar structure of collapse was observed for the ruins of agricultural fences (see Figure 12 ). These observations indicate that the seismic shock arrived from the south in the case of a compressional wave, or from the north, if the wave causing the collapse was extensional. Thus, by this exercise the eastward and westward propagating seismic waves can be excluded.
  2. Most rotated blocks in the Avdat ruins are turned counterclockwise and they were found exclusively on NE-trending walls (see Figure 15 ). The only case of clockwise rotation was found in a wall fragment with trend SE 140°. The fact of the appearance of rotated blocks, as described above, indicates push movements (compression wave approaching the buildings). Thus, the only possibility left is a compressional seismic wave coming from the south. Rotation itself involves shear stresses acting along the walls, thus the seismic wave must have arrived at some angle to the walls.
Following the well-known strong earthquakes a large number of structural elements were found to be preferentially inclined toward the epicenter, however, in some cases the inclination was in the opposite direction. As in the case with the wall inclinations, the walls facing the seismic wave collapsed systematically toward the seismically induced compression strain, whereas the walls aligned parallel to the seismic wave lost support and collapsed in a random manner. Therefore, one has to look for a correlation between the trend of a construction element and the direction of collapse. The collapse debris form the shape of a cone, because the central part of a collapsing wall segment undergoes maximum oscillation during the seismic event (Figure 16 ).

The preferred direction of collapse or inclination of building elements may be either toward an epicenter or away from it. If the damaged site is located in the quadrangle of compression strain (Figure 17 ), the deformation will be caused by a push movement exerted on the ground, resulting in inclination and collapse toward the epicenter. In contrast, in the sites located in a tensional quadrangle, the deformations are induced by a pull movement causing inclination and collapse away from the epicenter. In either case, the line of collapse or relative motion can be determined. This line connects the original position of an object and its position after an earthquake, or corresponds to the dip azimuth of an inclined element. The intersecting points of the collapse lines measured in many places will converge at the area of the epicenter (Figure 18 ).

Shear stresses applied to an elongated element cause its rotation. The direction of rotation depends on two factors:
  1. orientation of principle stresses in a location and
  2. the orientation of the elongated element
Field study of the epicentral zones of the world-known strong earthquakes revealed that some building constructions or rock fragments were rotated clockwise, whereas others were rotated counterclockwise. A seismic wave approaching a building parallel or normal to its walls will result in collapse, shift or inclination with no rotation (Figure 20(a) ). The rotation should take place in the cases where the principle stresses are oblique to a construction element, and the resolved shear stresses are high (Figure 20(b) ). Thus, rotated elements situated on perpendicularly oriented walls should have an opposite direction of rotation, if the seismic shock came along the bisector of the two walls (Figure 20(c) ).

Two mechanisms of rotation, caused by tectonic movements, are known in geology (Figure 21 ):
  1. book-shelf structures, or synthetically rotated blocks, and
  2. asymmetric pull-aparts, or antithetically rotated blocks (Jordan, 1991)
As can be seen in Figure 21 , the same direction of rotation can be obtained by the different stress setups. These rotated blocks are termed "antithetical" or "synthetic" because with respect to the same simple shear couple two directions of rotation are possible. A synthetic structure is formed as a result of compression acting parallel to an element along axis, whereas the antithetical structure is developed when extension is parallel to an elongated element. Thus, in tectonics the interpretation of the rotation structures should be proceeded by a determination of the strain that occurred parallel to a rotated element. Such an ambiguity does not exist in seismic interpretations. Any lateral extension applied to a construction should lead to its collapse or inclination, whereas rotation could occur only under horizontal compression. This provides an additional criterion for the determination of strain accompanying an earthquake: the appearance of rotated blocks is an indication of a push movement. A scheme showing the direction of rotation, with respect to the direction of seismic wave propagation, is shown in Figure 20 .

This discussion leads to an additional conclusion explaining the lack of oriented inclination and collapse features in an epicentral area (and additionally, to the assumption that the point seismic source is not valid in the epicentral zone): the shock wave moving from a hypocenter under a high angle to the surface, results in a lateral extension applied to constructions. This explains why in recent earthquakes (Acapulco, 1962; Scopje, 1963; Tashkent, 1966 and others) the areas above a hypo-center do not reveal systematic inclination and collapse patterns (Muto et al., 1963; Binder, 1965; Medvedev, 1966; The Scopje Earthquake of 26 July 1963, 1968; Mirzoev et al., 1969; Liquidation of Consequences of Tashkent Earthquake, 1972), whereas some distance away inclination and collapse have pronounced directional patterns (Figure 22 ).

All said above is true for the features of destruction found in building constructions built on an isotropic massive foundation without a strong preferential orientation of the fabric in the basement rocks. In the studied case, Avdat was built directly on massive limestones. Thus, an input caused by rock anisotropy could be neglected. To avoid gravitational reasons for the city's destruction, the authors did not conduct the measurements on the slope of Avdat hill.

Avdat ruins have two perpendicular directions of walls (—NE 50° and —SE 140°), so the overall model can be represented as a single building (or room). To cause south-directed wall collapse by a compressional seismic wave, the shock should have come from south side. If the shock arrived exactly perpendicular to the NE-trending walls (i.e., from SW, Figure 23(a) ), the shear stresses along walls should be minimal and the rotations should appear only occasionally.

In contrast, maximal shear stresses would result if the seismic wave approached the buildings along a bisector line between the walls (Figure 23(b) ), i.e., from south. In this case rotations on both wall directions should be clearly pronounced, whereas both NE and SE-trending walls should reveal oriented collapse and inclinations to the south (SE and SW sides correspondingly).

In the case of Avdat the only NE-trending walls revealed oriented collapse and inclinations, and SE-trending walls demonstrate systematic counterclockwise rotations. Such a situation is possible if the compressional wave came from SSW (Figure 23(c) ).

Thus, the epicenter was located somewhere SSW from the Avdat settlement, and the scale of destruction indicates that the epicenter was situated 15 km south of Avdat, probably in the area of the Nafha Fault zone. The force (seismic intensity) of a shock resulting in the destruction of buildings was determined using the scale of earthquake intensity MSK-64. Buildings in Avdat town according to this scale are classed as B type - buildings from natural hewed stones. Quantitative characteristics of destruction: most buildings were destroyed (more then 75%). According to the degree of destruction Avdat town is classified as fourth degree:
  • through cracks and breaks in the walls
  • collapse of building parts
  • breaking of connections between separate parts of buildings
  • collapse of internal walls and walls of framework filling
All these features of destruction show on IX-X intensity of seismic shock on territory of Avdat town.
The destruction was caused by a compressional seismic wave and the epicenter was located SSW of Avdat somewhere in central Negev. The degree of town destruction during the historical earthquake according to Seismic Intensity Scale MSK-64 was IX-X.

Intensity Estimates
Distinguishing 7th century effects from "previous" earthquake effects

Korjenkov and Mazor (1999) did not produce an Intensity or directional estimate for any of the earthquakes that preceded the 7th century CE event. However, by making use of their detailed descriptions of seismic effects and the Earthquake Archeological Effects chart, I produced Intensity estimates for both the 7th century CE earthquake and the "previous" one. "Previous" earthquake seismic effects were presumed to come from seismic effects associated with rebuilding as limited rebuilding should be associated with the 7th century earthquake. Although Bucking et al (2022) produced evidence of Umayyad and Abbasid occupation on the slopes below the acropolis, the upper acropolis area may have been destroyed and largely abandoned as archaeologists (e.g. Peter Fabian) have posited in the past. Although I cannot rigorously distinguish whether my "previous" earthquake Intensity estimate is for the southern Cyril Quake of 363 CE or the early 5th century CE earthquake, if Erickson-Gini, T. (2014) is correct that the southern Cyril Quake only caused some structural damage and the 5th century earthquake was massive, my Intensity estimate for the "previous" earthquake is likely effectively for the 5th century quake. So, it is labeled as such. An intensity estimate for the "363 CE earthquake" was derived from Cave dwellings which the archaeologists beleive were damaged or destroyed during this event.

Topographic or Ridge Effect

Citing a personal communication with Tali Erickson-Gini in 1996, Korzhenkov and Mazor (1999), noted increased seismic damage in upslope caves adjacent to the Avdat acropolis after the 363 CE earthquake. This suggests that a ridge effect may present at Avdat. A terrain map shows that Avdat is situated at the end of a ~4 km. long ridge Avdat. Orientation of the ridge further indicates that seismic energy arriving from the NE or the SW (orthogonal to the ridge) would be most likely to produce seismic amplification at the site. A slope effect may also be at play as Avdat surrounded by steep slopes on 3 sides.

Southern Cyril Quake (363 CE)

Effect Location Intensity Comments
Collapsed Vaults Caves in the slopes adjacent to the Avdat Acropolis VIII + numerous collapses of walls and cave vaults
Collapsed Walls Caves in the slopes adjacent to the Avdat Acropolis VIII + numerous collapses of walls and cave vaults
These effects, dated to the 363 CE earthquake, were observed in the caves furthest upslope and suggest a site effect or what Korzhenkov and Mazor (1999) call a "sky-scraper effect". Either way, seismic amplification is indicated so while this archaeoseismic evidence requires a minimum Intensity of VIII (8) when using the Earthquake Archeological Effects chart of Rodríguez-Pascua et al (2013: 221-224 big pdf), it is downgraded one unit to VII (7).

The "previous" earthquake - Early 5th century CE or 5th century CE

Effect Earthquake
Location Intensity
Displaced Walls "previous"
prob. 5th century
Room 10 in court in S Quarter
Fig. 5
Room 8 in court in S Quarter
Fig. 6
Displaced Walls "previous"
prob. 5th century
N yard of bath-house
Fig. 7a
Fig. 7b
Tilted Walls "previous"
prob. 5th century
Support Walls of Southern Church
Fig. 8
VI +
Collapsed Walls "previous"
prob. 5th century
Caves VIII +
Collapsed Vaults "previous"
prob. 5th century
Caves VIII +
This archaeoseismic evidence requires a minimum Intensity of VIII (8) when using the Earthquake Archeological Effects chart of Rodríguez-Pascua et al (2013: 221-224).

7th century CE earthquake

Effect Earthquake
Location Intensity
Penetrative fractures in masonry blocks 7th century many locations
an example from Northern Church
Figure 4
Tilted Walls 7th century various locations VI +
Collapsed Walls 7th century various locations
Fig. 9
Collapsed Walls 7th century Agricultural Fences
Fig. 11a
Fig. 11b
Arch damage 7th century various locations VI +
This archaeoseismic evidence requires a minimum Intensity of VIII (8) when using the Earthquake Archeological Effects chart of Rodríguez-Pascua et al (2013: 221-224)
Korjenkov and Mazor (1999)'s seismic characterization of the 7th century earthquake

As mentioned previously, Korjenkov and Mazor (1999) were able to sort a number of seismic effects by earthquake event - distinguishing whether the observed damage was due to the 7th century earthquake or one of the "previous" earthquakes (i.e the southern Cyril Quake of 363 CE and/or the 5th century CE earthquake). As such, one can have confidence in the Intensity estimate Korjenkov and Mazor (1999) produced for the 7th century earthquake. Korjenkov and Mazor (1999)'s conclusion for the 7th century CE earthquake is that

The destruction was caused by a compressional seismic wave, the epicenter was located SSW of Avdat somewhere in central Negev, and the degree of town destruction [] according to Seismic Intensity Scale MSK-64 was IX-X.

Discontinuous Deformation Analysis by Kamai and Hatzor (2005)

Kamai and Hatzor (2005) performed Discontinuous Deformation Analysis (DDA) on a model

for displaced blocks on the western wall of the Roman Tower of Avdat. The tower, dated to 294 AD, was founded directly on bedrock, and has risen to a height of 12 m, from which only 6 m are left standing today. (Kamai and Hatzor, 2005 citing Negev, 1997). The best-fit simulation (Fig. 16A ) was run with the following seismic parameters:
  • Ah = l g
  • Av = 0
  • f =3 Hz.
  • Dh_avmax = 8 cm.
Kamai and Hatzor (2005:133-134) did not present single best fit parameters due to various limitations so this parameterization, though consistent with other estimates of Intensity, should only be considered approximate. A PGA of 1 g converts to an Intensity of 9.3 using Equation 2 of Wald et al (1999). Although Korjenkov and Mazor (1999) did not explicitly attribute the bulges in the Roman Tower to the 7th century CE earthquake, the high PGA that comes from Kamai and Hatzor (2005)'s simulations suggests that this is the case as the 7th century earthquake was apparently a powerful and destructive earthquake which both destroyed Avdat and led to its abandonment.

Kamai and Hatzor (2007) noted that seismic amplification can be at at play at higher parts of a structure (i.e. the "Sky-scraper effect" mentioned by Korzhenkov) leading to potential amplification of bedrock PGA by as much as 2.5. This could in turn lead to a bracket of PGA values for The Roman Tower from 0.4 and 1.0 g. These PGA values convert to Intensities of 7.8 - 9.3 using Equation 2 of Wald et al (1999). A final result can thus be that DDA modeling of the Roman Tower suggests bedrock Intensities between 8 and 10 during this earthquake. Note that this ignores seismic amplification due to a ridge effect over the entire site. The ridge effect could add an additional amplification factor.
Variable Input Units Notes
g Peak Horizontal Ground Acceleration
Variable Output - Site Effect not considered Units Notes
unitless Conversion from PGA to Intensity using Wald et al (1999)

Model and Lab derived properties

Model was run in qk.mode using a sinusoidal input function. The authors noted that in the case of Avdat the obtained ground-motion parameters may be higher than reasonably expected (e.g. l g at Avdat). Therefore, they do not argue at this stage for exact historical ground motion restoration. Soil-structure and rock-structure interactions were not part of the analysis and considering that Avdat may be subject to a ridge effect, 1 g could be reasonable and could explain the unusual wall bulge at the Roman Tower at Avdat which appears to have been generated by a significant seismic force. Although the authors date this seismic effect to the 3rd or 4th century CE, Erickson-Gini (2014)'s characterization of the 363 CE earthquake as causing the least damage to the site of the 4 recognized earthquakes suggests that this is not the case.

Lab Measurements of original stones from Avdat

Property Value Units
Density 2555 kg./m3
Porosity 5 %
Dynamic Young's Modulus 54.2 Gpa
Dynamic Shear Modulus 20.4 Gpa
Dynamic Poisson's Ratio 0.33 unitless
Interface friction angle 35 degrees

Notes and Further Reading

Articles and Books

Bucking, S., et al. (2022). "The Avdat in Late Antiquity Project: uncovering the Early Islamic phases of a Byzantine town in the Negev Highlands." Antiquity 96(387): 754-761.

Bucking, S., et al. (2022). "The Avdat in Late Antiquity Project: uncovering the Early Islamic phases of a Byzantine town in the Negev Highlands." Antiquity: 1-8.

Erickson-Gini, T., Crisis and Renewal: Settlement in the Negev in the 3rd and 4th Centuries CE, with an Emphasis on the Finds from the New Excavations in Mampsis, Oboda and Mesad ‘En Ḥazeva (Ph.D. diss.), Jerusalem

Erickson-Gini, T. (2000). Nabataean or Roman? Reconsidering the date of the camp at Avdat in light of recent excavations. XVIIIth International Congress of Roman Frontier Studies, Amman, Jordan.

Erickson-Gini, T. and I. Yigal (2013). "Excavating the Nabataean Incense Road." Journal of Eastern Mediterranean Archaeology & Heritage Studies 1(1): 24-53.

Erickson-Gini, T. (2014). "Oboda and the Nabateans." STRATA - Bulletin of the Anglo-Israel Archaeological Society 32.

Fabian, P. (1998). Evidence of earthquakes destruction in the archaeological record–the case of ancient Avdat. Pp. 21E-26E in The Annual Meeting of the Israel Geological Society, Mitzpeh Ramon.

Goren, Y. and P. Fabian (2008). "The Oboda Potter's Workshop Reconsidered." Journal of Roman Archaeology 21.

Kamai, R. and Y. Hatzor (2005). Dynamic back analysis of structural failures in archeological sites to obtain paleo-seismic parameters using DDA. Proceedings of 7th International Conference on the Analysis of Discontinuous Deformation (ICADD-7).

Korzhenkov, A. and E. Mazor (1998). "Seismogenic Origin of the Ancient Avdat Ruins, Negev Desert, Israel." Natural Hazards 18: 193-226.

Korzhenkov, A. and E. Mazor (1999). "Structural reconstruction of seismic events: Ruins of ancient buildings as fossil seismographs." Science and New Technologies 1: 62-74.

Negev, A. (1974). The Nabatean Potter's Workshop at Oboda, Habelt.

Rodkin, M. V. and A. M. Korzhenkov (2018). Estimation of maximum mass velocity from macroseismic data: A new method and application to archeoseismological data. Geodesy and Geodynamics.

Zion, O., Ashkenasi, Eli, Erickson-Gini, Tali , (2022). Byzantine Oboda / Avdat and the Surrounding Agricultural Regime. Archaeological Excavations and Research Studies in Southern Israel Collected Papers. A. Golani, Varga, Daniel, Tchekhanovets, Yana, Birkenfeld, Michal. 5.

Excavation Reports

Negev, A. (1997). "THE ARCHITECTURE OF OBODA: FINAL REPORT." Qedem 36: III-214..

A. Negev, The Architecture of Oboda: Final Report (Qedem 36), Jerusalem 1997; ibid. (Reviews) BAR 24/6 (1998), 56. — NEA 61 (1998), 182. — BAIAS 17 (1999), 93–94. — AJA 104 (2000), 154. — BASOR 318 (2000), 84–85;


Avdat at

Caves of Avdat

Description Image Reference
Caves on north slope Fig. 4 - Zion et al (2022)
Caves on south slope Fig. 5 - Zion et al (2022)
Caves Plans North 1 Fig. 6 - Zion et al (2022)
Caves Plans North 1 Fig. 7 - Zion et al (2022)
Caves Plans North 2 Fig. 8 - Zion et al (2022)
Caves Plans South 2 Fig. 9 - Zion et al (2022)
Caves Plans South 1 Fig. 10 - Zion et al (2022)
Caves Plans South 2 Fig. 11 - Zion et al (2022)
Caves Plans South 2 Fig. 12 - Zion et al (2022)
Aerial view of the southern slope Fig. 2 - Bucking et al (2022)

Notes on the so-called Potter's Workshop

Russell (1985) cited archeoseismic evidence for the Incense Road Quake at Avdat citing Negev (1961:123,125) and Negev (1974:24) where Russell (1985) states

At Avdat, an imperial coin struck at Alexandria and tentatively identified as Trajanic was apparently found in association with the collapse of the potter's workshop (Negev, 1974:24).
Ambraseys (2009) supplied the following comments:
Negev argues instead that these destructions were caused by invading Safaitic and Thamudic hordes in the mid first century (Negev 1976), basing his thesis on the period of pottery debris found in a workshop at Oboda. This solution might seem preferable, since it is best not to assume an earthquake unless there is written evidence for it. However, apart from the complexity of the multiple dates of the pottery discovered by Negev (and the fact that later potters often imitated earlier styles), the appearance of a second-century coin among the pottery (Russell 1981, 8) seems to refute his thesis. Of course, this coin does not prove that Oboda was destroyed by an earthquake; it merely shows that Negev has made a mistake. What may suggest an earthquake is the sheer severity and extent of the destruction. Russell believes that neither a Roman annexation of the territory nor sacking by Safaitic or Thamudic hordes could, in any case, have done so much damage.
Negev (1976:229) states
Several years ago I suggested, on account of the results of the excavations at Oboda, a new chronological division for the archaeological history of the Nabateans in the central Negev, based on three phases, focusing at that time my attention on what I named the Middle Nabatean Period. The archaeological data indicated that this period, which began at the end of the reign of Obodas II, terminated abruptly during the generation following the death of Aretas IV, after the middle of the first century CE. I attributed the destruction of Oboda and several road stations along the Petra-Gaza road to attacks of Arab tribes who penetrated from Arabia, and left their imprints in the thousands of Safaitic and Thamudic graffiti in the central Negev, to the east of the Arabah, and also in northern Arabia itself.

The evidence on which I based this chronological scheme was purely archaeological — pottery and coins under a destruction layer, and on the basis of the finds in the Nabatean potter's workshop at Oboda 145 which all pointed to a break in the settlement of the central Negev sometime after the middle of the first century CE.
Goren and Fabian (2008) re-examined the so-called Potter's workshop at Avdat/Oboda and concluded that it was probably a 2nd to early 3rd century CE mill-bakery in the Roman Quarter of town. They noted, among other things, that the original excavations by Negev of the "Potter's workshop" were in unstratified deposits, had coins dating from Hellenistic to the 3rd-4th centuries CE, and geochemical and minerological analysis indicated that the pottery found there appeared to be imported rather than made locally. This suggests that Negev's original hypothesis that the so-called Potter's workshop at Avdat/Oboda showed a break in occupation in the 1st century CE due to invasion (as Negev suggested) or an earthquake (as Russell (1985) proposed) is not supported by the evidence.

Negev's Notes on Retaining Walls and a post 3rd century CE Earthquake

Negev (1961) identified several phases of occupation at Avdat one of which, dated by inscriptions, began in the third century CE. Negev (1961: 126) noted that during this Late Roman/Byzantine occupation phase, the retaining walls were "probably shattered by a strong earthquake" and were repaired by "adding a second, rounded wall, screening the original one". A precise date for the archeoseismic damage was not supplied.

Wikipedia page for Avdat