Seismic memory in stone: Acropolis wall and Areopagus rockfall point to 0.21–0.23 g shaking in the 1705 earthquake

A forensic back-analysis combines staged pseudostatic soil–structure interaction models of the Acropolis Circuit Wall with a limit equilibrium study of a limestone block detachment at the Areopagus to reconstruct the shaking of the 1705 and 1805 earthquakes.

Historical earthquakes leave traces in monuments – collapses, permanent tilts, cracks, displaced blocks and rockfalls – but reading those traces quantitatively is difficult, because the structure and its ground change over centuries. In a paper for the 4th International Symposium on Geotechnical Engineering for the Preservation of Monuments and Historic Sites (ISSMGE TC301) in Athens, Dimitrios Egglezos (GEO.PER. S.A. & DOMOS S.A. and the Hellenic Open University), archaeologist Dr D. Moullou (Ephorate of Antiquities of East Attica and the Hellenic Open University) and surveying engineer V. Evgenikou describe a forensic workflow that uses two independent “mechanical archives” at the Acropolis of Athens. The first is the Acropolis Circuit Wall (CW), a composite gravity retaining system whose sectors record either collapse or permanent deformation. The second is a joint-controlled limestone block detachment at the nearby Areopagus hill. By expressing both in terms of a common effective acceleration, the authors narrow down the shaking produced by the earthquakes of 1705 and 1805.

Three sectors, three outcomes

The reconstruction targets three observed signatures on the Circuit Wall. The eastern sector (section S11, Area A5) collapsed, and the authors attribute this to 1705; it serves as the collapse benchmark. The southern sector (S18, Area A6) preserves permanent deformation and cracking without collapse, a cumulative signature of repeated loading. The cornice-bearing sector of the north wall (SP1) collapsed, primarily during the 1805 earthquake, after progressive deterioration.

This leads to the guiding hypothesis: the 1705 earthquake must explain both the east-wall collapse and the start of permanent deformation in the south wall, while the 1805 event must be decisive for the north cornice sector.

Defining historical seismic demand

The analyses focus on events between 1694 and 1805. For each event the authors tabulate the source region, epicentral distance, magnitude, estimated PGA range on rock and effective accelerations for each wall sector. The 1705 North Attica earthquake (M 6.3, about 11.5 km away) has an estimated rock PGA of 0.229–0.485 g and an effective EW acceleration of 0.210 g for the east wall. The 1805 event north of Athens (M 6, 24.2 km) gives 0.166 g. The Chalkida events of 1694, 1726 and 1785 are treated as secondary for the sectors considered.

Three relations define the seismic input. Effective acceleration is taken as aeff ≈ 2/3 PGA, supported by dynamic–pseudostatic calibration within the Circuit Wall study; it represents a demand level that produces comparable permanent deformation and damage-state transitions, not transient response. Soil class A is used for the east wall, the north cornice sector and the Areopagus, and class B for the backfilled south wall, with aeff,B ≈ 1.2 aeff,A following EC8-based differentiation. Directionality is represented by aNS ≈ (0.4–0.5) aEW, with a central value of 0.5.

What the 2019 recordings showed

The directionality relation comes from site recordings. Four accelerographs on the Acropolis Hill recorded the 19 July 2019 NW Attica (Parnitha) earthquake, of magnitude about 5.1–5.3. Stations ACRA, ACRE and ACRJ gave EW/NS PGA ratios of about 2.44, 2.18 and 1.97 – NS amplitudes about 41–51% of EW – with peak EW accelerations of 85–111 cm/s². A fourth sensor, ACRK, mounted at a crown position on the south wall, recorded 251 cm/s² EW and a lower ratio of about 1.41. The authors interpret this as crest-level amplification, possibly with a local structural contribution, rather than free-field input, and use it as an indicator of amplification rather than to define directionality.

Staged soil–structure interaction and a drift-based collapse criterion

Representative Circuit Wall sections were analysed with staged nonlinear pseudostatic SSI to follow the evolving state of the wall–backfill–foundation system. The stages cover construction and modification phases, changes in backfill compaction and state, ageing and degradation of interfaces, rainfall and pore-pressure episodes, excavation, unloading and refilling, and the historical earthquakes applied in sequence. For the south wall, the staged history runs from 1694 to 1999.

Performance is judged by crest displacement normalised by wall height (criterion K1). Relative roof-to-base drift below 2% corresponds to damage level D1 (almost full functionality), 2–5% to D2 (limited damage), 5–10% to D3 (significant damage) and 10–15% to D4 (near collapse). Collapse is defined as crossing the K1 threshold.

The Areopagus block was analysed by limit equilibrium along its mapped failure plane to find the acceleration giving a factor of safety of 1, and then checked with SSI to confirm that slip localises on the same discontinuity with large displacements. Inputs include a limestone rock mass with GSI of 100 (intact) and 55 (jointed), compressive strength of 60 and 4.8 MPa and modulus of 15.6 and 12.2 GPa. The schist rock mass has GSI of 100 and 15. Joint properties are normal and shear stiffness of 60 and 12 GPa/m, peak cohesion of 96 kPa, peak friction angle of 59° and residual strength of c = 0 and φ = 30°.

Two independent thresholds that agree

The Areopagus analysis gives a critical acceleration of 0.23 g for the onset of joint sliding. Independently, the eastern Circuit Wall section S11 reaches the K1 collapse threshold at aEW = 0.21 g. The authors consider this agreement meaningful because it comes from different systems and mechanisms. They attribute the small difference to the modelling approach: the wall analyses use characteristic, conservative structural parameters suited to intervention design, so 0.21 g is a lower-bound estimate rather than a best estimate.

For the south wall (S18), the staged history with 0.13 g for 1705 and 0.09 g for 1805 (soil class B) produces permanent displacements consistent with the observed deformation and cracking without collapse. Converting these to soil class A gives 0.108 g and 0.075 g for the north cornice sector. There, the K1 threshold is crossed during the 1805 stage, which the authors see as the decisive increment on a structure already weakened by earlier damage. The same family of events thus produces collapse, permanent deformation or delayed failure depending on local resistance and history.

Why it matters for practice

The study offers a transferable workflow for heritage sites where structural damage and nearby geological failures act as parallel records of past earthquakes. It combines a structured evidence inventory, explicit target mechanisms for each sector, a common demand measure, staged-history modelling that makes later responses depend on earlier damage and interventions, and independent cross-checks of threshold accelerations. It also states the direction of modelling bias openly, rather than forcing an exact numerical match. The authors conclude that 1705 and 1805 were the decisive earthquakes for the Circuit Wall and for Athens more generally, and that their reconstructed demand levels give engineering constraints to complement historical catalogues for the 18th and early 19th centuries.

Reference

Egglezos, D., Moullou, D., & Evgenikou, V. (2026). Seismic Memory in Stone: Structural and Geological Evidence of Historical Earthquakes at the Acropolis of Athens. Proceedings of the 4th International Symposium on Geotechnical Engineering for the Preservation of Monuments and Historic Sites (ISSMGE TC301), Athens, Greece. Paper TC301-2026-60.