• from Boyer (2022:72-77)

There are many references to earthquakes affecting the west side of the city in the published archaeological corpus. The reliability, and hence the value, of the evidence in these references, varies considerably, so a simple matrix was established as follows to enable the data to be evaluated that took into account confidence that the damage/destruction was seismically induced, the existence of a sealed archaeological context, and the chronological constraints.

• High reliability: significant earthquake destruction, sealed archaeological contexts, and reliable chronology.
• Medium reliability: significant earthquake destruction, some chronological constraint, but sealed context doubtful or uncertain.
• Low reliability: destruction not necessarily seismic-induced, no dating evidence or only vague chronological constraint, no sealed context, or generally where there is uncertainty over the reliability of the data.

The application of the matrix was to some extent subjective due to differing professional practices and levels of reporting detail, especially in pre-mid- twentieth-century excavation reports, compared to more recent reports. The references, including the seismic event attributions made by the respective authors, and an estimate of reliability derived from the matrix, are listed in Appendix E (Table E.1), although references to undated or very vaguely dated earthquake events are excluded. The reliability assessment will no doubt need revision once final reports become available for excavations around the city.

The results show that evidence supporting individual earthquake events is often poorly constrained chronologically, and evidence from sealed and well-dated archaeological contexts is available for only a few sites. The most reliable evidence relates to events in the mid-seventh and mid-eighth centuries; evidence of pre-sixth-century earthquakes is both sparse and, with two exceptions, of low reliability. The higher proportion of references to the younger seismic events is influenced by factors such as the limited extent of excavation to Roman levels and the tendency for the foundations of later buildings to be taken down to bedrock, thus removing any evidence of earlier seismic damage.

There is no published evidence of earthquake damage in the city before the third century; however, it has been speculated that the rebuilding of the North Gate in AD 115 followed an earthquake event in the early second century.¹⁴⁴ If so, then the most likely event was the AD 112 earthquake with an epicentre west of the Dead Sea. The c. second-century plastering events recorded at several sites along the main north-west aqueduct (JW01) in the present study may possibly reflect damage from the same event.¹⁴⁵ Aside from damage to the North-West Quarter in the third–early fourth centuries, there is little reliable evidence of strong earthquake activity in the third- to the fifth-century period affecting buildings in the central part of the city, and although there are several attributions to the AD 363 earthquake episode, the evidence cannot be confidently attributed to any particular regionally attested event.¹⁴⁶ The study found evidence of the ruination of columns on the Artemis Temple podium and the deposition of column drums onto the adjacent natural rock surface of the Artemis upper terrace that probably dates to the early period following the spoliation of the temple in the fourth–early fifth centuries.¹⁴⁷ The ruination of structures on the west side of the Nymphaeum and the subsequent construction of a quadriporticus on top of these ruins also appears to predate the mid-fifth century.¹⁴⁸ These findings hint at the possibility that earthquake damage to the city in the fourth–early fifth centuries was more severe than previously supposed and may be associated with the AD 418 event that impacted Palestine.¹⁴⁹

The most significant impacts on the city, and the southern Decapolis region as a whole, occurred in major seismic episodes that recurred at c. one hundred-year intervals between the mid-sixth and mid-eighth centuries, and in particular, the mid-seventh- and mid-eighth-century events. There is an increase in references to damage from the sixth century, and especially the mid to late sixth century, but the reliability is generally poor, and the destruction is limited in extent across the city. Some of this damage could conceivably have been associated with the major AD 551 earthquake episode, estimated to have had a movement magnitude of 7.5, which caused widespread damage along the Lebanese littoral and to a lesser extent inland, but the epicentre is c. 200 km from Gerasa.¹⁵⁰ The collapse of the floor of the large reservoir (Reservoir 1; see Chapter 7) in the city's North-West Quarter in the fifth or sixth century was likely caused by a substantial shock, but this damage has not been attributed to a specific seismic event by the excavators.¹⁵¹

Earthquake events in the seventh century caused widespread and well-documented damage across the city between the Hippodrome and the West Propylaea. Evidence from the Macellum and Propylaea Church is considered to be reliable and is attributed by the excavators to earthquake events in AD 633 or AD 659/60, but the majority of excavators attribute the earthquake damage to the so-called AD 659/60 seismic episode.¹⁵² This episode is now considered to have comprised two separate events two days apart in June AD 659 that affected Palestine and, to a lesser extent, Syria.¹⁵³ Although there appears to be no support for this event affecting the Decapolis from literary sources, there is convincing archaeological evidence for earthquake damage around the mid-seventh century for the western side of the city.

The effects of earthquakes in the mid-eighth century were felt in all the cities of the southern Decapolis, and there is substantial evidence of earthquake damage in Gerasa related to these events. The Gerasene references attribute the damage to events that range in date from AD 717 to AD 749. While the majority of the more recent references attribute the damage to an event in January AD 749, the evidence only supports an approximate mid-eighth-century date and does not preclude the possibility of several separate but closely spaced events. Tsafrir and Foerster argued that events previously dated to AD 746/47/748 actually refer to a major event in January AD 749;¹⁵⁴ however, the matter is not settled, as is shown by evidence for three separate seismic episodes affecting the region between AD 746–757.¹⁵⁵ The Gerasene sites with damage from the mid-eighth-century seismic episode are found throughout the excavated parts of the western side of the city, and the reliability of the evidence is medium to high in the majority of cases. While evidence from the South-West Quarter is sparse for the study period, it seems likely that the mid-eighth-century earthquakes affected the entire western side of the city. Particularly reliable evidence is available for sites in the North-West Quarter and in the South Decumanus–South Tetrakionion precinct, where the suddenness of earthquake impact on domestic life is dramatically represented in the archaeological record.¹⁵⁶ Evidence from the eastern side of the city is currently limited to two sites. The first site was identified in the study adjacent to the substructio carrying the Qairawan aqueduct. The site (JWP111) was briefly exposed during foundation excavations for a new building, and radiocarbon dating of charcoal from horizons above and below a tumble horizon up to 2 m thick over a low, ruined masonry structure (possibly an early water channel) indicated that the tumble dates to the mid-eighth century period (Fig. 3.37). The second site is the Large East Baths, where earthquakes are said to have caused ‘a brutal and almost complete destruction of the monument’.¹⁵⁷

Whatever the actual date of the seismic episodes of the mid-eighth century, the damage they caused was substantial. The massive damage to the Large East Baths and the damage evidenced in domestic buildings around the South Tetrakionion and the North-West Quarter show that buildings both large and small were dramatically affected. Evidence from the North-West Quarter shows that this part of the city was devastated and abandoned after the mid-eighth-century earthquakes and not reoccupied until the Ayyubid–Mamluk period;¹⁵⁸ however, evidence from the vicinity of the Umayyad Congregational Mosque in the central city area shows that this part of the city was resettled shortly after these earthquakes.¹⁵⁹

The catalogue published by Grigoratos and others includes many significant seismic events (i.e. Mw > 6) in the region after the mid-eighth century. Earthquake damage has been identified in the city's South-West Quarter dating to the late ninth–early tenth centuries,¹⁶⁰ and Seigne and Tholbecq attributed a destruction layer on the lower terrace of the Zeus Temple to an earthquake around AD 1200.¹⁶¹ As already noted, an earthquake affecting the city in January 1837 was witnessed by a European visitor, George Moore.

The clearances of the 1920s and 1930s and subse­quent building restorations have removed much of the visible evidence of structural damage to buildings in the city; however, there is sufficient evidence from his­torical photographs and from surviving unrestored buildings to provide an overview of the types of dam­age incurred. Large-scale damage in the form of groups of columns falling in the same direction was revealed in the excavation of St Theodore and Synagogue Churches by the Yale Mission (Figs 3.38a-3.38b), and similar evidence has been reported from churches elsewhere in the Decapolis.¹⁶² Smaller-scale impacts of horizon­tal shear stress on columns are preserved in the colon­nades surrounding the Oval Piazza and on the Artemis podium (Figs 3.39a-3.39c), and there are also exam­ples of the vertical rotation of Ionic capitals of up to seventy degrees in the western colonnade of the piazza (Figs 3.40a-3.40b).

Mortar was rarely used on the larger city buildings, and there are many examples of seismically induced block separation, especially in masonry pilasters and col­umns (Figs 3.41a-3.41b). Seismically induced shearing is also evident across ashlars in extensive walls such as those forming the Artemis cella (Fig. 3.41c).

• from Boyer (2022:77,81,82)

By the Late Byzantine period, the city had experienced several major earthquake episodes, and the populace would have been well aware of their damaging effects. In the case of the Nymphaeum, the risk of further dam­age was mitigated by the construction of buttress walls on the northern and possibly also the southern elevations of the monument, and these walls are well pre­served. Similar buttress walls were also built against the southern wall of the southernmost taberna across the laneway (the so-called Museum Street) to the north of the Nymphaeum, but such walls do not appear to have been common in the city.

The original Nymphaeum was a two-storey structure with a relatively slender profile that was built on the cor­ner of the Cardo and Museum Street by AD 190-191 ¹⁶³. The lack of excavation to the rear of the Nymphaeum means that nothing is known of structures built against it on the west elevation, but, contrary to the findings of the Yale Mission, the Nymphaeum was not originally joined to second-century rooms to the south ¹⁶⁴. The Nymphaeum's central semicircular facade was reason­ably well protected against earthquake stress by thicker walls and flanking lateral projecting wings, but the lat­eral wings themselves had thinner walls that supported huge masonry pilasters and were less well protected. This exposure was later demonstrated by severe damage to upper storeys of both lateral wings above the level of the buttress walls yet much of the facade interior undam­aged.

The evidence suggests that the buttress walls were built in two stages. In the first stage, a well-constructed spolia wall of hard limestone ashlars, including one block with a second-third-century inscription, was added to the northern elevation ¹⁶⁵. In the second stage, a thicker, less well-constructed wall of large nari blocks was built against the earlier stage 1 wall of the Nymphaeum, and a similar wall was built against the original south­ern wall of the taberna on the north side of Museum Street (Figs 3.42a-3.42d). The importance of the work can be deduced by the fact that the second phase walls reduced the width of Museum Street from 5.3 m to 3.8 m. The wall abutting the original southern wall of the Nymphaeum may also have been built in the sec­ond stage (Figs 3.43a-3.43c). The timing of the build­ing of the buttress walls is currently uncertain: it seems likely that the first support wall was added after a seismic event had impacted the city and demonstrated the dam­age risk to the monument, and on present evidence, it appears that the first strong earthquake impacts were in the fourth/fifth centuries. The inscribed blocks included in the wall as spolia provide a second/third-century ter­minus ante quem. Brenk, Bowden, and Martin suggested that the first stage wall on the northern elevation of the Nymphaeum was part of a reconstruction of the area west of the Nymphaeum carried out in the first half of the fifth century ¹⁶⁶. From the available evidence, it is specu­lated that the building of the first and second phase but­tress walls was triggered by damage to the Nymphaeum caused by the seismic events of the fourth/ fifth centuries.