Archaeoseismology: Methodological issues and procedure

Fabrizio Galadini & Klaus-G. Hinzen &

Stathis Stiros

Received: 10 May 2005 /Accepted: 2 June 2006 / Published online: 7 November 2006# Springer Science + Business Media B.V. 2006

Abstract Archaeoseismic research contributes im-portant data on past earthquakes. A limitation of theusefulness of archaeoseismology is due to the lack ofcontinuous discussion about the methodology. Themethodological issues are particularly important be-cause archaeoseismological investigations of pastearthquakes make use of a large variety of methods.Typical in situ investigations include: (1) reconstruc-tion of the local archaeological stratigraphy aimed atdefining the correct position and chronology of adestruction layer, presumably related to an earthquake;(2) analysis of the deformations potentially due toseismic shaking or secondary earthquake effects,detectable on walls; (3) analysis of the depositional

characteristics of the collapsed material; (4) investiga-tions of the local geology and geomorphology to definepossible natural cause(s) of the destruction; (5) inves-tigations of the local factors affecting the groundmotion amplifications; and (6) estimation of thedynamic excitation, which affected the site underinvestigation. Subsequently, a ‘territorial’ approachtesting evidence of synchronous destruction in acertain region may delineate the extent of the areastruck by the earthquake. The most reliable results ofan archaeoseismological investigation are obtained byapplication of modern geoarchaeological practice(archaeological stratigraphy plus geological–geomor-phological data), with the addition of a geophysical-engineering quantitative approach and (if available)historical information. This gives a basic datasetnecessary to perform quantitative analyses which, inturn, corroborate the archaeoseismic hypothesis. Sincearchaeoseismological investigations can reveal thepossible natural causes of destruction at a site, theycontribute to the wider field of environmental archae-ology, that seeks to define the history of the relation-ship between humans and the environment. Finally,through the improvement of the knowledge on the pastseismicity, these studies can contribute to the regionalestimation of seismic hazard.

Key words archaeoseismology . historicalseismology . geoarchaeology . environmentalarchaeology . natural catastrophes

J Seismol (2006) 10:395–414DOI 10.1007/s10950-006-9027-x

F. Galadini (*)Istituto Nazionale di Geofisica e Vulcanologia,Via di Vigna Murata 605,00143 Rome, Italye-mail: galadini@ingv.it

K.-G. HinzenDepartment of Earthquake Geology,Geological Institute, University of Cologne,Vinzenz-Pallotti-Str. 26,51429 Bergisch Gladbach, Germanye-mail: hinzen@uni-koeln.de

S. StirosGeodesy & Geodetic Applications Laboratory,Department of Civil Engineering, Patras University,Patras 26500, Greecee-mail: stiros@upatras.gr

Introduction

The use of archaeological data to investigate unknownor poorly known historical earthquakes has continuous-ly increased since the first reports of earthquake effectsrecorded in the archaeological heritage (e.g., De Rossi1874; Lanciani 1918; Evans 1928; Agamennone1935). For the past few decades, the word ‘archae-oseismology’ has been used to define the investiga-tions related to the seismic effects on ancientstructures, uncovered by means of archaeologicalexcavations or pertaining to the monumental heritage.The increased interest is reflected in the publication ofspecial volumes and articles in seismological andgeological journals (e.g., Guidoboni 1989a; Stirosand Jones 1996; McGuire et al. 2000). Since 2002, aworking group ‘Archaeoseismology’ is active withinthe European Seismological Commission. However,though the methodological aspects have been investi-gated in a number of publications (e.g., Karcz andKafri 1978; Rapp 1986; Nikonov 1988; SantoroBianchi 1996; Stiros 1996; Korjenkov and Mazor1999a; Guidoboni 2000; Nur and Cline 2000; Noller2001), this interdisciplinary branch of the seismolog-ical science has so far not had a developmentcomparable to that of other related disciplines (e.g.,historical seismology and paleoseismology). This ismainly due to: (1) the absence of continuous,systematic and combined efforts for the refinementand extension of the methodological background ofthis research field; (2) the necessity to involve, in mostcases, a wide range of experts (archaeologists andhistorians specialized in specific fields, periods andareas, geologists, geophysicists, and engineers) dealingwith the range of topics usually included in thearchaeoseismological research; (3) the problem thatwhat may usually be regarded as signs of anearthquake (e.g., widespread collapses, sudden aban-donment of settlements, restorations) may have causesother than earthquakes or, inversely, (4) that variousnatural processes (e.g., landslides, floods, soil settle-ment) or even human activities (e.g., wars, revolutions)may have effects similar to those of earthquakes. Moregenerally, methodological problems arise from thenecessity to merge information collected with thedifferent methods pertaining to the earth and engineer-ing sciences and to the human-social sciences.

Being aware that these problems cannot be solvedwith simple remedies, or by a single paper, we believe

that methodological treatises can promote the causeand improve the discussion of the complex subjectmatter represented by archaeoseismology. For thisreason, in this paper we will discuss aspects related to(1) the displacement of archaeological remains alongshear planes; (2) the use of off-fault paleoseismolog-ical information in the archaeoseismological perspec-tive; (3) the traces of dynamic excitation on ancientbuildings including secondary earthquake effects; (4)the archaeological stratigraphic evidence of destruc-tion; (5) the use of architectural stratigraphy; (6) thequantitative in situ analyses; (7) the ‘territorial’archaeoseismological approach, summarising theproblems due to superposition of multiple earthquakeeffects and the contribution of historical information;(8) the quantitative characterization of an archaeo-seismologically detected earthquake. Each one ofthese points represents the object of a specific sectionor sub-section. The points and their succession partlydefine a theoretical line of procedure.

The paper addresses the methodological aspectsbased on the authors’ experience matured through thestudy of European cases. This regional perspective,however, is not considered as a limit since thearchaeological, geological, seismological and histori-cal practices are based on some methodologicalmilestones which are the same worldwide.

The role of archaeoseismology in the investigationsof past seismicity

The identification or characterization of past earth-quakes defines archaeoseismology as one of theprocedures suitable to enrich the knowledge of pastseismicity, together with paleoseismology, historicalseismology and instrumental recording of earth-quakes. These disciplines are the backbone of anyseismic hazard analysis.

Since archaeoseismological studies are related tothe ‘archaeological past’, a context of chronologicalinterest cannot be strictly defined. Indeed, the ‘archae-ological past’ is strongly conditioned by cultural as-pects and is therefore diachronous throughout theworld, being closely related to the long history ofcivilization. In theory, archaeoseismological data maybe derived from the analysis of archaeologicalfeatures of ages preceding the building of settlements(e.g., the coseismic collapse of frequented caves) or

396 J Seismol (2006) 10:395–414

from the archaeological excavation of coseismic ruinsfew decades after the occurrence of a recent earth-quake. It is evident, however, that in case of earth-quakes of modern or contemporary times, historicalinformation often provides more reliable data on thecoseismic effects than archaeological–architectural orpurely archaeological investigations. Moreover, con-sidering that the detection of seismic damage isconditioned by the presence of buildings or structureswhich may have been struck by the earthquake,

periods of history preceding the organization ofsettlements in towns, or the beginning of a life-styledependent on the building of edifices are lesssignificant in the archaeoseismological perspective.Therefore, since the historical information is sparserfor the antiquity and the Middle Ages (European area)and settlements and buildings potentially recordingearthquake effects already existed, the improvementof the seismological information from archaeoseis-mology is expected particularly for these periods.

Figure 1 Chronological intervals of application of differentresearches on past earthquakes (Italian case). The shade of thefilling colour towards the left in the rectangle defining archae-oseismology indicates that the application is much more limitedback in time since the fifth century B.C. Indeed, the archae-oseismological practice is generally in need of building remains.Remains suitable for archaeoseismology are, in the Italian case,almost always younger than the above reported age. The arrowon the right of the archaeoseismological rectangle indicates thatthese investigations may be potentially performed in order toinvestigate more recent historical events, but it may bemeaningless in facing more detailed historical information. Incontrast, paleoseismological techniques may be performed inorder to have a more detailed picture of coseismic geologicaleffects which, usually, are not or not fully described in historicalreports. The arrow on the left of the paleoseismological rectangleindicates that paleoseismology may be potentially performed in

order to find traces of pre-Late Pleistocene events. But this maybe meaningless if a list of Late Pleistocene-Holocene paleo-earthquakes suitable for hazard evaluations is available. Thelower chronological limit is conventionally constrained by theages produced in commercial radiocarbon analyses (the mostfrequently dating technique used in paleoseismology). The‘chronological definition’ of the event horizon has differentmeanings if applied to archaeoseismology or paleoseismology. Inthe former case, it indicates that for pre-historic/proto-historicevents the chronology may be defined in intervals of centuries,while the chronological definition may be in the order of decadesin case of historical events. In case of paleoseismology, thechronology is mainly based on numerical dating, often givingsigma values of centuries, and may be inadequate for historicalevents which necessitate a precise chronological definition. Thecompleteness of the historical information is based on recentstudies made on this issue (e.g., Stucchi and Albini 2000).

J Seismol (2006) 10:395–414 397

The diagram in Figure 1 illustrates the chronolog-ical application of the different analyses of pastearthquakes in Italy. It may be drawn in a similarmanner also for other regions. The intense gray colourof the archaeoseismological ‘line’ indicates that thebest application is related to earthquakes whichoccurred between the third and second century B.C.

and the Middle Ages. This is because: (1) largeamounts of archaeological data have been producedand are being produced, testifying that the territorywas densely inhabited and built; (2) historical sourcesare sparser; and (3) archaeological chronology, thoughusually not comparably precise as the historicalchronology, is often more precise than typical numer-ical ages produced in paleoseismology of historicaltimes. These are the reasons why most of thearchaeoseismological literature deal with events whichoccurred in the early first millennium A.D. in Europe.

As for archaeoseismological practice, Table 1 illus-trates the different kinds of analysis that may beapplied to a potential archaeoseismic event. In order todefine the procedural aspects, we divided the inves-tigations which are necessarily conditioned by in situactivities from those made in laboratories, archives andlibraries and which are mainly related to the historicalresearch. The list of analyses results from an assimi-lation, re-proposition and update of lists alreadyproduced in previous methodological works (e.g.,Karcz and Kafri 1978; Stiros 1996; Nur and Cline2000). The different sources of information summar-ised in Table 1 are discussed in the next sections.

Recognizing forms potentially relatedto seismic effects

We divide the ‘direct’ evidence of seismic impact at anarchaeological site due to ancient or sub-recent eventsinto four main categories: (1) displacement along shearplanes; (2) coseismic geological effects due to shakingand related effects on building structures; (3) deforma-tion of building remains still in primary position; (4)evidence of destruction from the archaeologicallydefined site history.

These four points may define primary evidence ofan archaeoseismic event. Further evidence derivesfrom less conclusive archaeological information rep-resented by (1) the archaeologically detected aban-donment of a site and (2) the evidence of rebuilding

and restoration (archaeologically detected and/ordefined through the architectural stratigraphy ofmonuments still in primary position).

In the next sub-sections we will discuss the abovementioned points, and delineate potential problemsand procedures of analysis. This will include the insitu data gathering summarised in lines 1 to 8 ofTable 1. Since rarely a single form can be univocally

Table 1 Summary of the information of interest in thearchaeoseismological practice

Line In situ information/practice Extra situinformation/practice

1 �Local geology/geomorphology

�Archaeologicaldocumentationof pastexcavations

2 �Paleoseismology ofshear planes displacingarchaeological remains

�Relationshipbetweenarchaeologicaland historicalchronologicalconstraints

3 �Off-faultpaleoseismologicalinformation in thearchaeoseismologicalperspective

�Historicalframeworkof thearchaeoseismicevent

4 �Geotechnical informationon the foundation soils

�Numericalmodelling of‘fossile’ strongmotionseismograms

5 �‘On-building’ structuralanalysis of thedeformation

6 �Archaeologic–stratigraphicreconstruction

7 �Morphology anddepositional features ofthe collapsed material

8 �Architectural history ofmonuments(‘Architecturalstratigraphy’)

9 �Local seismic response10 �Dynamic behaviour of

the buildings11 �‘Territorial’ check of the

archaeoseismological information

398 J Seismol (2006) 10:395–414

related to an earthquake, we consider that the bestarchaeoseismic evidence derives from the summationof coseismic forms of different types. Each formpotentially related to seismic damage should bechecked in terms of alternative causes of deformation.The final attribution should result from the typicalprocedure of exclusion of other possible destructiveevents, natural or man-made. The attribution maystrongly benefit from quantitative modelling of thedeformation under dynamic conditions and from ananalysis of the local seismic response. Theseapproaches not only help to define the nature of thedestructive event at an archaeological site but mayalso give quantitative seismic parameters such as thepeak ground acceleration or the magnitude of theearthquake. For this reason the quantitative analyseswill be the object of a specific section.

Displacement along shear planes

Many cases of displaced archaeological remains havebeen reported in the available literature throughout theworld (e.g., Stiros 1988; Marco et al. 1997, 2003;Noller and Lightfoot 1997; Galli and Galadini 2001and references therein). This deformation processinvolved buildings, fortified walls, canals and otherkinds of manufacts. The case histories indicate thatthis type of event at an archaeological site is morefrequent than one might assume. The displacement isrelated to (1) the activation of a fault during an earth-quake or (2) of a shear plane in case of (a) differentialsettlement or (b) landsliding not necessarily inducedby a seismic event. The general tendency to settlevillages and towns in areas conditioned by recenttectonic activity, e.g., along the borders of tectonicintermontane basins (that is in areas affected by faultactivity and/or slope instability), suggests that thesekinds of phenomena may not be unusual. Particularlyvertical offset may result from different processes asmentioned above (convergence of forms). Indeed, thedisplacement of foundations due to coseismic activityof a fault or due to aseismic differential settlement orsliding (along newly formed shear planes or inheritedfractures and faults driving the gravitational displace-ment) may be completely similar (e.g., Karcz andKafri 1978).

The correct interpretation of factors conditioningthe displacement results from geological and geomor-phological investigations at the archaeological site.

Surveys should be made in order to define the role ofslope instability in the observed displacement. Thepossible presence of landslide scarps, tensional cracksor other features suggesting gravitational movementsshould be investigated. Reconstruction of the mor-phology during the proposed time of the damagemight be necessary, especially in places which werepopulated over long periods during which the mor-phology was anthropogenically altered. Geologicalinformation is necessary to understand whether thearchaeological settlement was located along an activefault. Moreover, the estimation of the amount of offsetmay allow discrimination between coseismic tectonicand aseismic gravitational displacements. In case offaulting, throw is a function of fault dimension and isrelated to earthquake magnitude (e.g., Wells andCoppersmith 1994), while the amount of motion insliding often largely exceeds that expected from faultactivation (Galadini 2006). Moreover, translationaland/or rotational sliding may produce deformationswith extension and tilting in such amount that thedisplacement can not be attributed to tectonics.

Paleoseismological techniques should be used inorder to define the history and the characteristics ofthe displacement along the shear plane (e.g., Nollerand Lightfoot 1997; Galadini and Galli 1999). Theclassification of foundation soils is also fundamentalto verify the possible occurrence of differentialsettlements due to laterally varying geotechnicalcharacteristics. Structures founded in an area wheresedimentary transitions between compressible anduncompressible sediments are present may experiencedifferential settlement, with or without dynamicforcing/excitation. It is evident that the absence ofcompressible layers prevents the possibility that thedisplacement was conditioned by differential verticalmovements. This kind of information is generallyobtained by means of boreholes reaching depths ofseveral tens of meters. Moreover, the geometry of theshear planes should be defined at depth. A shear planemay have a curvilinear shape in section or may detachin correspondence of a compressible layer in case oflandsliding or differential settlement, respectively.This geometry is not typical of faults. Informationon the shape of the shear plane is usually obtained bymeans of geophysical methods (e.g., Dolan and Pratt1997; Demanet et al. 2001; Improta et al. 2003).

Through these analyses, the impact of differentnatural potential causes of displacement can be

J Seismol (2006) 10:395–414 399

evaluated. The final hypothesis about the coseismicorigin derives from the exclusion of other naturalevents (e.g., Nikonov 1988; Stiros 1996). Moreover,in case of fault activity, the exclusion of fault creep asthe cause of displacement can result from the observa-tion of widespread building destruction contempora-neous to the displacement (i.e., with additionalarchaeological information, see below) or from paleo-seismological information on the fault behaviour.

Deformation/destruction of buildings due to shaking-related geological effects (off-fault paleoseismology)

The main coseismic geological effects which maydamage buildings are liquefaction, lateral spreading,and landsliding. Liquefaction, with or without lateralspreading, may occur many kilometres away from anepicentral area (Galli 2000), and may cause founda-tion displacement and building collapse (Seed 1967).The identification of liquefaction features, like soft-sediment deformation, sand dykes, remains of sandvolcanoes (e.g., Obermeier 1996), is generally amatter of the so called ‘off-fault paleoseismology’.In such cases, paleoseismological excavations focuson sedimentary structures suggesting the destructura-tion of sands resulting from the dynamically inducedincrease of pore pressure. Liquefaction phenomenacaused by seismic shaking are often widespread.Identification of liquefaction effects is a strongargument for coseismic nature of observed damage.

Archaeological excavations often uncover remainsinvolved in landslide deposits. In such cases wallremains can be mixed with blocks or debris of thelandslide body or buried by the landslide accumula-tion. Even if this destruction at an archaeological sitecan be attributed to landsliding, other pieces ofevidence (e.g., synchronous seismic damage else-where) are necessary to argue for a coseismic effect.Differently from liquefaction, landsliding cannot beunivocally correlated with the occurrence of anearthquake, although landslides are often triggeredby earthquakes (e.g., Keefer 1984; Jibson 1996).

Deformation of building remains stillin primary position

Only part of the monumental heritage has beenuncovered by modern or contemporary archaeologicalexcavations. In many cases (e.g., the Colosseum in

Rome or the Parthenon in Greece), the emergence ofmonuments has persisted throughout many centuries.In these cases, the monuments have been affected bychanges of the original use and form (e.g., restorations,superposition of more recent buildings, architecturalmodifications to permit a different use) due to variousreasons sometimes difficult to define. But certainly amonument persistent throughout the centuries repre-sents a structure potentially recording the effects oflong-term seismicity, once the researcher is able todecrypt traces of earthquake impact below the heavymantle of restorations and architectural modifications.A good example of this statement is represented by theColosseum in Rome, whose primary function as arenafor games ended in the sixth century (Rea 1999).Traces of earthquake damage due to events of theLate Antiquity, of the Middle Ages and also of theeighteenth century can be detected (Rea 1999).

Apart from the monumental heritage, modernarchaeological excavations may uncover significantlylarge portions of structures still standing in theirprimary position. The traces of seismic shaking mayhave been preserved on the walls after their burialwhich occurred during the past centuries. Uncoveringthese structures may therefore permit investigation andeventual interpretation of these traces of deformation.

Generally speaking, ancient structures can showdeformations related to seismic shaking similar tothose observed on masonry buildings due to modernearthquakes. Typical earthquake effects on manufacts(e.g., Stiros 1996; Korjenkov and Mazor 1999a, b;Nur and Cline 2000, for the archaeological features;e.g., Doglioni et al. 1994, for churches of variousages) are: (1) cross fissuring in the vertical plane(generally developing from the corners of windows ordoors) due to the action of shear forces; (2) cornerexpulsion due to the orthogonal motion of walls(Figure 2a); (3) horizontal and independent motion,lateral and rotational, of single blocks within a wall(generally well visible in walls made of similarly shapedblocks, with rectangular section); (4) height reductiondue to vertical crashing; (5) rupture or motion of the archpiers and internal collapse of the keystones; (6) walltilting and distortion; (7) rotation of pillars or elementsof pillars and drums of columns around vertical axes(Figure 2b). However, some of these deformationsusually observed after damaging earthquakes may alsooriginate without dynamic excitation (Karcz et al.1977) and finding one piece of evidence only (or at a

400 J Seismol (2006) 10:395–414

single edifice only) cannot be considered conclusive(Stiros 1996; Mazor and Korjenkov 2001). This aspectis by no means trivial if one considers the convergenceof forms deriving from different natural events poten-tially responsible for damage.

This is the case of the deformation mentioned atpoints 1, 3 and 6. Indeed shear forces may be inducedby differential settlements or aseismic gravitationalsliding (point 1). Motion of single blocks within awall may be induced by root penetration (point 3).The distortion of walls may be induced by landslid-

ing, soil creeping or to the poor characteristics of thebuilding (point 6; Karcz and Kafri 1978). Moreover,in case of archeological remains, the height reduction(point 4) may also originate from the burial or theload due to more recent buildings superimposed overthe older ones or due to natural deposition ofsediments. However, apart from the ambiguousinterpretation of the deformation, the identificationof different types of deformations, or the repetition ofa certain deformation on numerous edifices of asettlement can be considered as consistent with the

Figure 2 (a) Corner ex-pulsion detected on atypical Roman wall (ce-ment and covering ofcubilia), Amphitheatre ofMarruvium, San Benedettodei Marsi, central Italy; (b)rotation around the verticalaxis affecting a pillar (ce-ment and covering ofbricks), Roman Ostia, cen-tral Italy; (c) reworking ofcollapsed material for sub-sequent use suggested bythe exclusive presence oftiles in horizontal attitude,Avezzano, central Italy; (d)thick layer of abandonmentand deterioration made ofmixed small-sized (sandy)materials of various origin(also burnt wooden frag-ments) underlying chaoti-cally disposed coarserfragments ofwalls, theatre of Suessa,Sessa Aurunca, southernItaly; (e) wall still main-taining its internal originalorganization toppled over afew-ten-centimeter-thicklayer of abandonment, the-atre of Cales, CalviRisorta, southern Italy; (f)columns collapsed in thesame direction, Selinunte,Sicily, Italy. Photographs(d) and (e) are fromGaladini and Galli (2004).

J Seismol (2006) 10:395–414 401

occurrence of an earthquake, once other possiblenatural causes have been excluded, e.g., through thegathered geomorphological and geotechnical informa-tion (e.g., Ward-Perkins 1989; Stiros 1996).

Evidence of destruction from the archaeologicallydefined site history

Modern archaeological excavations performed fol-lowing the stratigraphic criteria define precise anddetailed successions of events, which conditioned thehistory of a site. Usually, the succession of events isdefined within a chronological framework obtainedby means of the age of the uncovered artifacts or, lessfrequently, by means of physical dating (generally14C, TTL or dendrochronological methods). Thechronological framework may be more or lessdetailed, depending on the quality (in terms of ‘datingpotential’) and quantity of the found materials. Thisaspect is topical in archaeoseismology, since one ofthe main goals of these studies is constraining theearthquake date. For this reason it will be discussed ina specific sub-section.

When dealing with buildings, the identified eventsare often represented by the architectural modifica-tions, the periods of frequentation (with definition ofthe typology of use), the phases of abandonment, thecollapse of the structure or parts of it. This informa-tion is arranged in a vertical stratigraphic successionmade of superimposed tamped earth floors, super-imposed walls, alternating layers of reworked materialdenouncing abandonment of the structure, ruinedblocks or material proving the collapse, etc. Thesedata are important in the archaeoseismological inves-tigation, provided that the different ‘archaeological’events can be related to the effects of earthquakes. Ifthe archaeological practice uncovers deformed wallsstill in primary position as described in the previoussection, a link between the occurrence of an earthquakeand the evidence of ‘crisis’ (abandonment, collapse,restorations or architectural modifications of theoriginal structure) may be hypothesised. However,not all archaeological excavations uncover portions ofwalls bearing such evidence. In many cases walls inprimary position are cut by subsequent layers and onlya few tens-of-centimetres high ancient walls areavailable, thus hindering the analysis of deformations.In such cases, the analysis of collapsed materialsbecomes topical, though rarely conclusive if not

corroborated by other pieces of evidence (Guidoboniand Santoro Bianchi 1995).

The analysis of the ‘collapse layers’

Building collapses may be sudden, progressive orboth. Materials suddenly collapsed which experiencedsubsequent reworking (e.g., for the recovery ofprecious furnishings or for subsequent land use,Figure 2c) may appear similar to that derived fromthe progressive collapse (Ward-Perkins 1989). Theslow deterioration is generally demonstrated by theaccumulation of wall, vault and ceiling fragmentswith different grain size, without internal organiza-tion, over one or more layers of reworked material ordebris denouncing the abandonment of the structure(Figure 2d). The deterioration of walls results inaccumulations having maximum thickness close to thewall feeding the material. The thickness decreases in ashort distance from the wall. Generally, ancientness,abandonment and lack of maintenance are the mainreasons for the deposition of the deterioration layers.

The sudden collapse of large portions of walls mayoccur together with the progressive deposition offragments. This generally happens due to the deteri-oration of parts of the structure having a bearingfunction. The result is a stratigraphic succession madeof (from bottom): (1) a tamped earth floor, (2) a layeror more layers of abandonment (reworked materialand accumulation of debris), (3) a deterioration layerwith various grain size made of abundant tiles ormaterials from the roof (abundance decreasing up-ward) mixed with and underlying chaotically dis-posed fragments of walls and vaults but generally notexceeding few tens of dm3, (4) large portions oftoppled walls and collapsed vaults (in case of vaultedbuildings) which still maintain their internal organi-zation, (5) as point 3, but tiles or material derivingfrom the ceiling are absent or almost lacking.

Since this kind of stratigraphy indicates a deterio-ration of a structure, external events (i.e., naturalcatastrophes), though possible, are not necessary toexplain the archaeological data. Therefore we consid-er that, in such cases, no conclusive archaeologicalevidence of an earthquake is available.

Unfortunately, abandonment and lack of mainte-nance is a leitmotif of the Late Antiquity (third to sixthcentury A.D.) in the Roman Empire. This means thatearthquake effects of this period are, in most cases

402 J Seismol (2006) 10:395–414

and for a large part of Europe, mixed with the effectsof the abandonments. Extrapolating the evidence ofthe former from a type of stratigraphy such as thatdescribed above is definitely problematic and onlypossible if additional evidence is available. Moreover,abandoned structures are also highly vulnerable. Thismeans that they may suffer damage also in case ofweak dynamic excitation. Consequently, even in casesof correct attribution of some collapses to earthquakeeffects, the earthquake size may be overestimated.Hence, archaeoseismological investigations related toearthquakes which occurred during the Late Antiquityin most parts of Europe are definitely more problem-atic than investigations related to preceding periods.

In contrast, the sudden collapse of a structure stillmaintained and functioning is indicated by largeportions of toppled walls (maintaining the originalinternal organization; Figure 2e), collapsed columns(sometimes with drums having a ‘domino’ attitude;Figure 2f), large patches of tiles (often buried by thetoppled walls) or huge fragments of the vaults overthe tamped earth floor. Particularly the collapse ofnumerous columns in a preferred direction is consid-ered as reliable evidence of seismic destruction (e.g.,Karcz and Kafri 1978; Stiros 1996; Guidoboni et al.2002; Bottari 2003). Sometimes, wide traces ofburning can be found, e.g., remains of burnt woodplanks, grey films of burnt material over the tiles, etc.Materials showing the use of the building at themoment of the destruction are preserved below thecollapsed walls (coins, large pottery shards, remains offurnishings, jewels) and the layer denouncing theabandonment is lacking below the traces of destruc-tion. The presence of these materials indicate that theinhabitants had no time to take them before thedestruction. Skeletons of killed people or animals havebeen rarely uncovered (e.g., Sakellarakis and Sapouna-Sakellaraki 1981; Soren and Leonard 1989; Stiros andDakoronia 1989; Kilian 1996; Nur and Ron 1996).

Layers of abandonment can be found over thecollapsed buildings (but their interpretation is prob-lematic, see next sub-section). Otherwise, a newoccupation is demonstrated by the presence of asuperimposed tamped earth floor and a completearchitectural re-organization of the area which expe-rienced destruction or pervasive restorations.

Once the occurrence of sudden destruction has beenascertained, the cause has to be defined. The suddencollapse of a structure still in use may have been

induced by man, e.g., due to wars or local instabilities.Such a possibility has to be checked on the basis ofhistorical information. In European countries it is quiteimprobable, however, that the reading of historicalsources of the Antiquity can cast light on theoccurrence of local struggles. In contrast, our histor-ical knowledge allows an understanding of whetherthe destructive event occurred in periods characterisedby political instability. For example, widespreaddestruction of inhabited buildings in the SulmonaPlain (central Italy) during the second century A.D. (aperiod of strong political stability in this part of theRoman Empire) cannot be attributed to wars or localstruggles (Galadini and Galli 2001).

Once the natural origin of the destruction has beenhypothesized based on the collected data, the type ofnatural catastrophe should be hypothesized by merg-ing the archaeological data with other in situinformation. Supplementary information, mainly pale-oseismological or from quantitative approaches (seebelow) may help in accepting or refusing thearchaeoseismic hypothesis.

Abandonment of sites

The role of the site abandonment as a potentialindicator of the occurrence of natural destruction isnot unequivocal (e.g., Santoro Bianchi 1996). Indeed,the tendency to persistent human occupation/habita-tion of a site, especially for social–strategic reasonsindependent from the occurrence of natural destruc-tive events can be derived from archaeological dataand from site histories (e.g., Ambraseys 1971, 2005;Ward-Perkins 1989). This evidence can be seen byconsidering both the time interval following acatastrophe and the longer historical period. In thecase of Egna, in the Adige Valley (northern Italy), thedestruction and vertical displacement of the founda-tions which occurred at about the half of the thirdcentury A.D. was followed by frequentation, modernland-use and recent building, exactly across the activeshear plane (Galadini and Galli 1999). On the otherhand, the consideration of natural catastrophes asstrong constraints of the civilization history has beenmade (e.g., La Rosa 1995; Driessen and Macdonald1997 for the Minoan Crete; Stiros and Dakoronia1989 for the overall problem).

However, the complexity of the relationship be-tween environmental effects and responses at the

J Seismol (2006) 10:395–414 403

scale of a settlement or of a civilization prevents theuse of abandonment as a single indicator of environ-mental solicitations, especially if other pieces ofevidence are not available.

Rebuilding, restoration

Archaeological reconstructions of the site history, mayprovide evidence for structural modifications, recon-structions or restorations induced by the earthquakedamage (e.g., the Byzantine houses of Pergamoninvestigated by Rheidt 1996). While the procedure tochronologically constrain these events is the same asfor the destruction or abandonment layers (see thefollowing sub-sections) and based on the archaeolog-ical stratigraphy (plus a chronological input whichmay derive from the architectural characteristics ofthe new structures), their correlation with seismiceffects is not unequivocal. For the ancient Saepinumin southern Italy, a town definitely affected byearthquake damage at about the half of the fourthcentury A.D. (Galadini and Galli 2004 and referencestherein), the extent of restorations in that periodprobably also had an ‘evergetic’ origin beyond thenecessity of earthquake repairs (Gaggiotti 1991).Therefore, the relationship between repairs, structuralmodifications or complete rebuilding and the earth-quake effect has to be proven by merging differentpieces of evidence (e.g., Caputo and Helly 2000, forthe Larissa, Thessaly – Greece, theatre in the third tofirst century B.C.).

In some cases, however, the relationship betweenearthquake and repair seems more evident, i.e. whenthe adoption of ‘anti-seismic’ solutions in the restora-tion or the building of reinforcing structures, such asbuttresses or significant increase of foundation sizecan be detected (e.g., Hodges 1995; Stiros 1995;Korjenkov and Mazor 1999a, b, Hinzen and Schütte2003).

The age of the event

The chronology of an archaeoseismic event is afundamental aspect. It is not only important in orderto characterise the specific earthquake, but also for thecorrelation of presumed earthquake effects at differentarcheological sites or the attribution of archaeoseismo-logical evidence to a chronologically well definedhistorical event (e.g., Stiros 1996). Empirical earth-

quake frequency relations, the basis of probabilisticseismic hazard analysis, also depend on earthquakechronology. A chronologically constrained archaeo-logical stratigraphy can define the occurrence of acertain event within chronological intervals. When thechronological information is particularly abundant, theoccurrence of an archaeologically detected event canbe defined within a few decades. Rarely the archaeo-logical chronology can be more precise. Once thearchaeological analysis is completed, reliable historicalaccounts may help to better constrain the date of theevent (e.g., Ellenblum et al. 1998; Marco et al. 2003),although the correlation between historical andarchaeological data may be problematic (see below).

Defining a date for the destruction is probably themost complicated issue (see also Bottari 2003, on thisproblem). The age of an event horizon is defined onthe basis of the dating remains underlying thecollapsed materials and setting a terminus ante quem.The best dating remains are generally pottery shardsand coins. Since the materials are related to humanactivities, they may have been characterised by aprolonged use. For this reason, the most reliable lowerchronological limit of the event horizon derives fromthe youngest remains of a significant amount ofmaterials buried by the collapsed structure. In usualstratigraphic frameworks, the upper chronologicallimit (terminus post quem) should be defined by thematerials contained in the unit subsequent to thedestruction. However, in archaeology these units maycontain older remains, preceding the destruction. Thisis due to the already mentioned persistent use offurnishings and coins (coins, for example, may beused for many decades after the coinage; e.g., Cesano1913; Morrisson 1980; Reece 1984; Molinari 1994).For this reason the upper chronological limit shouldrepresent a sort of mean derived from the age of thematerials contained in the unit subsequent thedestruction, but only obtained from those remainsyounger than – or similarly aged to – the youngestpiece buried by the collapse. However, archaeologicalexcavations do not always uncover dating pottery orcoins; sometimes the gathered remains are too sparseto define a narrow chronological interval. In mostcases, undatable materials are uncovered and theinformation is not precise; e.g., an event occurredafter the building of an edifice whose age is definedon the basis of the architectural characteristics andbefore the building of another edifice sealing the

404 J Seismol (2006) 10:395–414

ruins, whose age is also architecturally defined. Thesechronological limits may constrain a time span ofcenturies. Similar problems arise with physicallydatable materials (e.g., by using the 14C method).Especially after a destructive earthquake older mate-rial might get placed on top of the destruction layerduring repair works and new constructions.

In these archaeological sub-sections we presentedfundamental characteristics of archaeological findingspossibly related to seismic damage. It is evident that(1) the sedimentary features of a unit marking thecollapse, and (2) a reliable stratigraphic reconstruction(including the correct attribution of the datingmaterials to the phase of collapse or to the unitssubsequent to the destruction) are topical aspects (alsosuggested by Guidoboni and Santoro Bianchi 1995).Particularly point 1 is not exactly the object of thearchaeological investigation and the archaeologicalreports rarely describe the sedimentary characteristicsof the collapse units in enough detail. Nor are thestratigraphic descriptions formulated in a perspectiveconducive to understanding the origin of catastrophicevents. As observed by Guidoboni (1989b), archeol-ogists sometimes underestimate the amount of infor-mation which can be derived from an archaeologicalexcavation. This ‘avoided’ information cannot berecovered, considering the ‘destructive’ character ofthe archaeological analysis. This suggests that peopleinvolved in archaeoseismology should participate inthe archaeological excavations, since the origin of thedestructive events which struck an ancient building ora village are fundamental parts of the site history andof the seismic history of a region. This may be theonly way to optimize the archaeological data. The

practice of reading and interpreting archaeologicalreports in the archaeoseismological perspective yearsor decades after excavations (line 1 column 2 ofTable 1) can only be considered a makeshift solutionin the face of total absence of information directlycollected in the field.

Based on the previous sections and as summarizedin Table 1, a complete archaeoseismological surveymust integrate aspects from different scientific dis-ciplines. Conclusions are complicated by multifacetedobservations, ranging from rotated column drums,shifted or tumbled parts of walls to the abandonmentof sites. In practice it will not be possible to presentproof that each of the observations must necessarilyhave a seismogenic cause. In order to structure theinterpretation of results, Hinzen (2005a, b) proposed afeasibility matrix in a case study of the excavation ofa Roman villa near Kerkrade (the Netherlands).Observations form the columns of the matrix, whilepossible causes form the rows. The matrix is thenfilled with a simple three-degree rating (feasible,questionable, and unfeasible) of the proposed causeas an explanation of the specific observation Figure 3).In the above-mentioned example the matrix was usedfor a small study; however, in principle it may also beextended to large surveys. The matrix concept helpsstructure the necessary discussion and allows a moregradual presentation of the results than the oftenrequested ‘decision’ about whether the uncoveredfindings were caused by an earthquake. This conceptalso illustrates which questions are still open or mightrequire further investigation. Therefore, for a pro-posed earthquake a certain level of probability ofoccurrence can be quantified: the fields in the matrix

DA I DA II DA III DA IV

static settlement

natural decay feasibleslow tectonic movements

antropogenic (war etc.) questionableseismogenic during occupationseismogenic after abandoning unfeasible

mining

Figure 3 Example of a feasibility matrix as a summary of theresults of an archaeoseismological investigation. The columnsDA I to DA IV represent four distinctive damages and the seven

rows give possible causes. Rows and columns can be extendedin larger surveys to cover the whole scenario and the result canbe quantified as outlined in the text.

J Seismol (2006) 10:395–414 405

which show a dynamic cause as feasible are quanti-fied by ‘1’, those showing non seismic causes asfeasible as ‘−1’, and the questional cases as ‘0’. Halfof the sum of the matrix values divided by the numberof matrix fields plus 0.5 gives a number between 0 and1 ranging from completely unfeasible to very probablecoseismic effects, respectively. This number, which is0.64 for the example in Figure 3, can directly be usedas weighting factor for the branches in a logic treeapproach of a probabilistic seismic hazard analysis.

Architectural stratigraphy

Architectural stratigraphy is the analysis of monu-ments from the architectural point of view, in order toreconstruct the building history. This field of studygenerally identifies restorations, modifications, spoli-ation and puts them in a stratigraphic order, possiblywith chronological constraints.

Evidently, this type of analysis is applied toidentify traces of past earthquakes or to clarify theamount of damage due to historical events. In this light,some works have used the architectural history of thebuildings (usually churches) in order to improve theknowledge of the effects of earthquakes sparsely known(e.g., ENEL 1986; Galadini et al. 2001; Stiros et al.2006). In case of the 1117 earthquake in northern Italy,an evident clustering of interventions to romanesquechurches has been identified in the zones which, basedon the sparse historical information, should represent themost damaged areas (ENEL 1986; Guidoboni et al.2000; Galadini et al. 2001).

The main problem of architectural stratigraphy isthe definition of the reasons for the architecturalinterventions (Guidoboni et al. 2000). For example, incase of the 1117 earthquake, political reasons may haveconditioned the modifications in the same period oftime, in order to create a new architectural style innorthern Italy, different from that of the rest of Europe(Suitner 1991). In this case, we have a superposition ofeffects of processes (impossible to discriminate)affecting the architectural modifications in a time spanof a few decades.

Quantitative in situ analysis

‘Quantitative analysis’ refers to investigations of thecharacteristics of the seismic source, ground motion,

and of the building response in order to test thearchaeoseismic hypothesis. It can be done provided thatreliable archaeological data and sufficient geophysicalinput parameters are available. These aspects aresummarised in lines 9 and 10 of Table 1.

Modelling of strong ground motion can be a fruitfulprocedure especially if the archaeological informationis derived from various excavations or sites within asingle ancient town. In such cases, a distribution of thepresumably coseismic damage may be detected. If arelationship exists between the damage distributionand the geological characteristics of the substratum (interms of potential amplification of the ground motionand of secondary effects), the coseismic hypothesis iscorroborated.

The first step in in situ analysis is a 3D reconstruc-tion of the local geology (e.g., Funiciello et al. 1995for the Colosseum in Rome). Geotechnical parametersare defined for the different stratigraphic units in thefoundation area of a settlement. Subsequently thefundamental resonance frequencies of the differentrock units are defined, e.g., by means of instrumentalrecordings of ambient vibrations and model calcula-tions (e.g., Hinzen and Schütte 2003; Fäh et al. 2006,this volume). When a model for the distribution of theengineering geophysical parameters (shear wavevelocity, density, damping) for the substratum isdefined, the frequency and possibly amplitude depen-dent ground amplification at the site can be calculat-ed. Following this procedure, Fäh et al. observed thatthe collapsed parts of the ancient town of AugustaRaurica in Switzerland correlate with areas wherethe amplification of the ground motion occurs in thefrequency band of the main building eigenfrequen-cies. The destruction was limited to the parts of thetown built on soils prone to amplification of theground motion and this evidently corroborates thearchaeoseismic hypothesis. Through a comparableprocedure, Hinzen and Schütte (2003) showed thatsecondary effects due to lateral spreading is a likelydamage scenario for the probable collapse of thePraetorium in Cologne.

Engineering seismological models may describe thecoseismic behaviour of a structure, once the architec-tural features of the ancient building are defined (e.g.,Croci et al. 1995; Papastamatiou and Psycharis 1996).This procedure reveals the characteristics of theseismic motion necessary for the collapse or signifi-cant damage of the building. Generally, the first step

406 J Seismol (2006) 10:395–414

of an analysis is the evaluation of the seismic input. Aseismogenic source is defined on the basis ofavailable seismotectonic information. The energeticparameters may be empirically and physically relatedto source dimensions (e.g., Kanamori and Anderson1975; Wells and Coppersmith 1994).

Based on the seismotectonic model, strong groundmotion seismograms are simulated and the seismicresponse at the investigated site is estimated frommodels based on the available geological/geotechnicalinformation. Finally, based on the seismic input, thedynamic behaviour of a building is studied, generallywith finite element models.

This procedure has been recently adopted forfostering the archaeoseismic hypothesis of the originof damage observed along the Late Antique city wallsof the locality of Tolbiacum, the present city ofZülpich (Germany; Hinzen 2005a, b). The authorhypothesized a possible coseismic damage due to theactivation of one of the known normal faults in theLower Rhine Embayment. As for the investigatedmanufact, the frequency band of the building reso-nance was identified and compared with the charac-teristics of the ground motion obtained from thesimulated earthquakes. The author discovered thatsome of the simulated earthquakes were strongenough to trigger the ground motion necessary todamage the city walls. In this way further evidence infavour of the coseismic origin of the damage wasproduced.

‘Territorial’ archaeoseismology

This term has been used in the methodological andresearch works by Guidoboni (2000) and Guidoboniet al. (2000) to indicate the correlation of archaeo-seismic effects throughout a region in order to definethe extension of the mesoseismal area. The extensionis crucial because the main difference betweenearthquake damage and damage related to othernatural causes is the distribution of the effects. Effectsof landslides, collapse of caves, and floods generallyaffect areas smaller than those damaged by strongearthquakes. Only volcanic eruptions may be respon-sible for extended effects throughout a large region.However, the origin of volcanic catastrophes can beeasily defined through the burial of the archaeologicalremains by thick volcanic deposits (e.g., Sigurdsson

et al. 1982; Livadie 1999; Mastrolorenzo et al. 2002for three Vesuvius eruptions in southern Italy).Moreover, only periods of war may cause suddendamage extended throughout a region. Extendedhuman-induced destruction in periods of politicalstability, however, is improbable.

For these reasons, the identification of buildingswhich have been simultaneously damaged in severallocations in periods of political stability (line 11 ofTable 1) or reviewing published or unpublishedarchaeological material on this aspect (line 1, column2 of Table 1) may be a logical method to test thearchaeoseismic hypothesis (e.g., Galadini and Galli2004; Guidoboni et al. 2000; Nur and Cline 2000;Jones and Stiros 2000; Stiros 2001).

However, one limitation of this procedure is thereliability of the archaeological chronology (Ward-Perkins 1989). In the sub-section dedicated to thechronological definition, we stated that a certain eventcan be defined sometimes only with a very largechronological uncertainty. A chronological resolutionwithin a time span of a few decades (which isdefinitely a good archaeo-chronological result), maylead to the correlation of different seismic eventsthroughout a region, especially in cases of areasaffected with frequent damaging earthquakes (Stirosand Dakoronia 1989; Guidoboni 2000). For example,the uncertainty associated with archaeoseismic dam-age attributed to the 346 A.D. earthquake in southernItaly is caused by another earthquake in 375 A.D.

nearby (Galadini and Galli 2004). Furthermore, in theCalabria region (southernmost portion of peninsularItaly), archaeoseismic evidence at different places hasbeen attributed in different works (Guidoboni et al.2000; Galli and Bosi 2002) to the same earthquake(historically known through an epigraph), whichoccurred in 374 A.D. This implied that the earthquakewas attributed to different seismogenic sources in thementioned works, based on the archaeoseismicevidence. A critical conclusion is that if the coseismicinterpretation of the archaeological features is reliablein both works, the archaeoseismic observations areassociated with more than one earthquake.

The chronological uncertainty limits the ‘territorial’approach as a tool for defining the archaeoseismicorigin of certain damage. If the effects of differentearthquakes can not be unequivocally ascertained, thesynchroneity of the seismic damage throughout aterritory due to a single event cannot be invoked as a

J Seismol (2006) 10:395–414 407

reliable tool to define the occurrence of an archae-oseimic event (see also Ambraseys 2005, on thisaspect).

Moreover, if the archaeological information can-not be verified in the field (e.g., excavations ofdecades ago, with limited availability of publisheddata) in order to corroborate the coseismic cause, thelarge time interval which defines the chronology ofthe destruction at the different sites cannot excludethe possibility that events with different origin (e.g.,due to seismic shaking, ancientness, landsliding,etc.) are being correlated. In the next sections wewill see how the chronological problem affects thedefinition of the characteristics of an archaeoseismicevent.

The historical information

In many cases archaeoseismological investigations arerelated to earthquakes already included in seismiccatalogues and known through historical information(see for example the archaeoseismological worksdedicated to the 365 A.D. earthquake in the Mediter-ranean area; Stiros 2001 and references therein).Generally, in case of earthquakes of the Antiquity,the information is very sparse, sometimes limited to asingle source (e.g., Guidoboni 1989a). Even in laterperiods up to the late Middle Ages original writtensources might also be very sparse. It is evident that insuch cases archaeoseismological data may giveadditional information about a poorly known event.

However, merging historical and archaeologicaldata always implies the comparison of informationwith different chronological resolution (line 2, column2 of Table 1; e.g., Santoro Bianchi 1996; Stiros 1996;Guidoboni 2000; Ambraseys 2005). Also in case ofpoor information, an historical event is usuallydefined in terms of the year of occurrence or of atime span of few years compared to decades inarchaeology. This implies that relating observeddestruction to an historical event, especially in regionsof frequent destructive earthquakes, cannot be con-sidered reliable (Guidoboni 2000; Guidoboni et al.2000). Since the list of historical earthquakes cannotbe complete, archaeoseismic evidence due to differentevents may be attributed only to the known historicalearthquake. This procedure leads to the enlargement ofthe perceived destruction and consequently to an

overestimation of the event size (Guidoboni 2000;Guidoboni et al. 2000; Ambraseys 2005).

In addition to the chronological resolution ofarchaeoseismology and historical seismology, it isalso necessary to define the historical framework ofthe event (see for example the importance of thisaspect in the case of the 365 A.D. earthquake; Traina1989; Stiros 2001 and references therein). The mainpurpose of this operation (line 3 column 2 in Table 1)is the political, social and economic characterizationof the period (e.g., Guidoboni 1996) in order tounderstand the degree of maintenance of the buildingsor their ‘quality’ in response to the economic situation(e.g., Molin and Guidoboni 1989), or to exclude thathuman factors (wars, decadence and abandonment)conditioned the destruction. For example, interferencebetween coseismic damage and war effects can befound in the archaeological dataset of the easternMediterranean area for the Late Bronze Age (Nur andCline 2000). Furthermore, widespread evidence ofarchitectural modifications and evidence of destruc-tion at about the half of the third century A.D. innorthern Italy has been traditionally attributed to theAleman invasions (e.g., Buchi 2000). The definitionof the areal extent of an earthquake which occurred inthe Adige Valley in that period is, therefore, definitelyproblematic (Galadini and Galli 1999).

Reconstructing the characteristicsof an archaeoseismic event

Important goals of the investigations on past earth-quakes are the evaluation of epicenter locations,amount and direction of slip, and magnitudes. On-fault paleoseismology and historical seismology ob-tain these results in different ways: the former byrelating high-magnitude events (generally responsiblefor surface faulting) to specific fault-sources and byestimating the magnitude from the observed offset perevent, from the length of the surficial ruptures, orfrom the extension of the zone with secondary effects(e.g., liquefaction). Historical seismology analyzes thedamage distribution and assigns intensities thatcharacterize the different localities. An intensity valuemay be attributed to the archaeoseismic damageobserved at a certain locality (e.g., Hinzen 2005a, bin the already mentioned case of Tolbiacum inGermany). Once the intensity has been defined, the

408 J Seismol (2006) 10:395–414

magnitude may be derived by using one of theavailable empirical relations (e.g., Ambraseys 1985)linking the two parameters; however, a magnitudebased on only one or very few intensity observationscan be biased. If the damage is defined for morelocalities, the definition of the magnitude may bemore precise and based on equations taking intoaccount the extension of the damaged area (e.g.,Gasperini and Ferrari 1997). Moreover, in such cases,magnitudes and epicenters might be inverted withprocedures such as those described by Bakun andWentworth (1997, 1999) and adopted by Hinzen andOemisch (2001) for earthquakes in the NorthernRhine area or those described by Sirovich andPettenati (2001) and Sirovich et al. (2002). Inconclusion, archaeoseismological data in their finalform may be close to the data expressed by historicalseismology, i.e., localities bearing evidence of acertain coseismic damage may be plotted on a mapand used to define seismic parameters.

The reliability of the areal distribution of archaeo-seismic damage is strongly conditioned by thechronological problems discussed in the previoussection. Damage at localities struck by differentearthquakes which occurred in a time span of somedecades may be considered as the effect of a singleevent. This might lead to an overestimation of themaximum observed magnitude of earthquakes whichstruck the region in the period under investigation andhas an obviously misleading effect in the search forthe epicenter. Also the effects of aftershocks, whichalways follow strong earthquakes, can add to thedegree of damage and lead to an overestimation of themagnitude of the main event. The case of the 346 A.D.

earthquake in southern Italy (Galadini and Galli 2004)may be used as an example of this kind of problem.The archaeologists have attributed presumed coseis-mic damage detected at numerous sites in central andsouthern Italy to this event. The result is a perceiveddamage distribution, which cannot be physicallyattributed to a single earthquake. It may result froma highly destructive seismic sequence, similar to theone which occurred in 1456 A.D. in the same territoryand attributable to the progressive activation of three(or four) seismogenic sources in a time span of a fewdays. However, considering the scarce chronologicaldefinition of the archaeoseismic evidence at somesites, this hypothesis cannot be considered as conclu-sive. For this reason, a different and as much

‘extreme’ view has been proposed, by taking intoaccount only the available information conclusivelyrelated to the 346 A.D. event (Galadini and Galli2004). Thus, the mentioned authors proposed twocompletely different pictures of the earthquake dam-age, showing the uncertainty related to the archaeo-seismic interpretation of the 346 A.D. earthquake.

Another aspect which has strong implications forthe estimation of earthquake size is the vulnerabilityof ancient edifices during periods of decadence orpolitical instability. For example, the abandonment ofbuildings such as theatres or pagan temples was quitecommon in the Roman Empire during the LateAntiquity (e.g., Liebeschuetz 2001). The edifices,without maintenance, experienced rapid decadence. Insome cases they even suffered from the practice ofspoliation. Huge monuments were considered asquarries for the extraction of materials which wereused for new buildings (e.g., the Colosseum in Rome;Rea 1999). It is evident that the vulnerability of someof these buildings during the Late Antiquity wassignificantly high. Therefore some archaeoseismolog-ical investigations on seismic events of the LateAntiquity may lead to systematic overestimation ofthe earthquake size.

Quantitative analysis of the ground motion at anarchaeological site may help in estimating the energyassociated with an earthquake. For example, high-level damage may result from strong amplifications ofthe seismic waves or secondary effects. If such effectsof wave spreading and ground motion are neglected,the obvious consequence is an overestimation of theearthquake magnitude. Ground motion amplificationmodelling at the ancient town of Augusta Raurica hasled Fäh et al. (2006; this volume) to consider thepossible archaeoseismic damage as strongly condi-tioned by site effects. For this reason, the authors haveproposed that the previously estimated magnitude(MW 6.9; Fäh et al. 2003), based on the amount ofdamage at Augusta Raurica, was overestimated.Moreover, the lower magnitude, and the destructiveeffect suggest that the earthquake probably originatedin an area not far from the investigated site. In thisway, the analysis of the amplification effects at thearchaeological site also produced information aboutthe possible epicentral location of the earthquake.

In the case of Tolbiacum the modelling of thedynamic response of the damaged structure, based ona seismic input due to the activation of known

J Seismol (2006) 10:395–414 409

seismogenic sources, has led Hinzen (2005a, b) torestrict the hypotheses about the seismogenic sourcespotentially responsible for the observed archaeoseis-mic damage. The two closest sources (known as theStockheimer and Kirspenicher faults) may haveproduced the damaging earthquake. The associatedmagnitude is MW 6.5.

Archaeoseismological analysis: Steps and themerger of practices

In the previous sections we have illustrated singleaspects of the archaeoseismological investigation.Each data source is characterized by specific prob-lems and caveats. However, reliable archaeoseismolo-gical analyses derive from the merger of various field

and archive/laboratory practices (Table 1). Figure 4presents a flow chart linking data sources, methodsand results, based on the discussion of the previoussections. Due to the complexity of the issue, only themain aspects of the archaeoseismological researchhave been outlined. Since every field case is different,additions and alternative modelling might becomenecessary or steps can be or have to be dropped.

Concluding remarks

In this paper we discussed some aspects related to theinvestigation of the archaeological traces of past earth-quakes. As stated in previousworks, archaeoseismologyneeds contributions from many different pieces ofinformation, derived from different disciplinary

DATA

DATA CATEGORY

EVALUATION OF ALTERNATIVE CAUSES (e.g. application of the feasibility matrix)

DATING AND MERGE R OF RESULTS

VALIDATION OF SEISMIC HYPOTHESIS WITH RESULTS FROM OTHERSITES

Geomorphological setting

Paleoseismological data

Subsurficial geological-structural setting

Morphological setting

Geotechnical data

Description of the deformation (wall

tilting/rotation, rupture of arch piers, vertical collapse

of the keystone, ...)

Archaeological stratigraphic succession

with facies attribution

Archaeological and "sedimentological" evidence of

sudden damage/collapse

Architectural stratigraphy

Numerical dating

Building model

Site amplification

Attenuation model

Seismicity model

Seismotectonic model

1) Displacement along shear planes 2) Off-fault geological effects

Deformation of building remains still in primary

position

Evidence of destruction from

archaeological site history

Seismotectonic/ Engineering-seismological

1) Fault activity 2) Landsliding 3) Differential

settlement

1) Deformation in the static field

2) Coseismic deformation 3) Non seismic natural

deformation

1) Ancientness 2) Coseismic destruction 3) Man-induced destruction

Modelling of the "fossile" strong motion seismograms (within thebounds of the seismotectonic model or not)

Coseismic forms of different types 1) Architectural stratigraphy 2) Historical sources3) Archaeochronological constraints 4) Numerical dating

Damage scenario and attribution of intensity

Territorial approach

Territorial extent of damages (within a locality or a region)

Synchroneity of the event and evaluation of the time resolution

Review of the published or unpublished archaeological material

Review of the seismological framework and evaluation of the alternatives

In situ investigations for each site

DATA

DATA CATEGORY

EVALUATION OF ALTERNATIVE CAUSES (e.g. application of the feasibility matrix)

DATING AND MERGE R OF RESULTS

VALIDATION OF SEISMIC HYPOTHESIS WITH RESULTS FROM OTHERSITES

Geomorphological setting

Paleoseismological data

Subsurficial geological-structural setting

Morphological setting

Geotechnical data

Description of the deformation (wall

tilting/rotation, rupture of arch piers, vertical collapse

of the keystone, ...)

Archaeological stratigraphic succession

with facies attribution

Archaeological and "sedimentological" evidence of

sudden damage/collapse

Architectural stratigraphy

Numerical dating

Building model

Site amplification

Attenuation model

Seismicity model

Seismotectonic model

1) Displacement along shear planes 2) Off-fault geological effects

Deformation of building remains still in primary

position

Evidence of destruction from

archaeological site history

Seismotectonic/ Engineering-seismological

1) Fault activity 2) Landsliding 3) Differential

settlement

1) Deformation in the static field

2) Coseismic deformation 3) Non seismic natural

deformation

1) Ancientness 2) Coseismic destruction 3) Man-induced destruction

Modelling of the "fossile" strong motion seismograms (within thebounds of the seismotectonic model or not)

Coseismic forms of different types 1) Architectural stratigraphy 2) Historical sources3) Archaeochronological constraints 4) Numerical dating

Damage scenario and attribution of intensity

Territorial approach

Territorial extent of damages (within a locality or a region)

Synchroneity of the event and evaluation of the time resolution

Review of the published or unpublished archaeological material

Review of the seismological framework and evaluation of the alternatives

In situ investigations for each site

Figure 4 Schematized flow chart for archaeoseismological investigations.

410 J Seismol (2006) 10:395–414

approaches. In contrast to previous discussions, westress the importance of quantitative analyses in orderto corroborate the archaeoseismic hypothesis. More-over, the points discussed here indicate the centralityof the stratigraphic perspective (in agreement withGuidoboni and Santoro Bianchi 1995) and theneccessity of data collection sensitive to the archae-oseismological perspective. This implies that archae-oseismological information should be collected duringan archaeological excavation in collaboration with thearchaeologists.

The complicated procedure to investigate a pre-sumed archaeoseismic event, the necessity of passingthe evidence through numerous ‘filters’ before con-sidering it reliable imply that only few cases ofdestruction recorded in the archaeological heritagecan be considered as archaeoseismological evidence.Therefore even in regions with high seismicity and along history of settlements the number of reliablearchaeoseismologically detected earthquakes might besmaller than one would expect.

Both the qualitative interpretation of the origin ofthe presumed traces of past earthquakes and quanti-tative analyses need significant contributions fromgeological (surficial and sub-surficial) geomorpholog-ical and geophysical investigations. The merger ofarchaeo-stratigraphy, geological and geomorphologi-cal information implies that archaeoseismology large-ly results from field geoarchaeological practices,thereby classifying geoarchaeology as archaeologicalresearch, based on geological concepts and methods,sharing with archaeology part of the aims (e.g. Rappand Hill 1998). The concept of ‘aim’ is central. It isevident that the aim of an archaeoseimologist is thecharacterization of past earthquakes, but it is alsoevident that a seismic destruction at an archaeologicalsite is a particularly important event within the sitehistory. With these goals in mind archeoseismologicalpractice has a double function: to enrich the knowl-edge of past seismicity of the territory and to helpreconstruction of the site history. Due to this secondfunction, archaeoseismological information contrib-utes to the archaeological environmental picture of acertain site, considering ‘environmental archaeology’as the study of the long-term relationship betweenhumans and the natural environment (e.g., Dincauze2000).

We believe this perspective places archaeoseismol-ogy as an integral part of the necessary investigations

at an archaeological site, rather than just a sub-branchof paleoseismology (as indicated by McCalpin, 1996)or a multidisciplinary procedure strictly related tohistorical seismological investigations of sub-recentseismicity.

While historic earthquakes in addition to instru-mentally recorded events have always been used inboth deterministic and probabilistic seismic hazardanalysis, the importance of palaeoseismic informationhas become obvious only in the last few decades. Webelieve that with increasing systematic archaeoseismo-logical investigations and wider acceptance of themethodology, the more reliable archaeoearthquakeswill be considered in the procedures of hazardestimations. Within this perspective, we have toconsider that archaeoseismology has a use which ispotentially wider than that of paleoseismology. Indeed,buildings and other man made structures of archaeo-logical importance might have suffered damage fromsmaller earthquakes than those causing significantgeological effects, detectable by means of paleoseis-mological techniques.

The different aspects discussed in this paper andthe perspective we present are the result of experiencegathered over the last decade, derived from archaeo-logical practice in the field and modelling results, andmore recently from activities of the WG Archae-oseismology of the European Seismological Commis-sion. We consider this paper as a positive step towardthe definition of a standard procedure in archaeoseis-mology. Though the various aspects discussed hereillustrate the unfortunately frequently inexact andundefined character of archaeoseismology, we areconfident that the parallel practices of field inves-tigations to uncover the archaeological traces of pastearthquakes and the contemporaneous methodologicalelaboration will provide the necessary and increas-ingly more reliable information on poorly known orunknown historical seismicity of a region.

Acknowledgments We are grateful to many colleagues forthe useful discussions during meetings, especially during theworkshop on “Archeoseismology at the beginning of the 21stcentury” held in Messina, June 9–11, 2004 and in Taormina,Selinunte and Palermo on June 12–13, 2004. The review by S.Marco (Tel Aviv University, Israel) and A. Marturano (IstitutoNazionale di Geofisica e Vulcanologia, Napoli, Italy) hasimproved the original text. We thank Sharon Reamer (CologneUniversity) for helpful comments and careful reading of themanuscript.

J Seismol (2006) 10:395–414 411

References

Agamennone G (1935) Come e quando rovinò il tempio diVenere Genitrice. Boll Soc Sism Ital 33:89–91

Ambraseys NN (1971) Value of historical records of earth-quakes. Nature 232:375–379

Ambraseys NN (1985) Intensity – attenuation and magnitude.Intensity relationships for northwest European earth-quakes. Earthq Eng Struct Dyn 13:733–778

Ambraseys NN (2005) Archaeoseismology and neocatastro-phism. Seismol Res Lett 76:560–564

Bakun WH, Wentworth CM (1997) Estimating earthquakelocation and magnitude from seismic intensity data. BullSeismol Soc Am 87:1502–1521

Bakun WH, Wentworth CM (1999) Erratum to estimatingearthquake location and magnitude from seismic intensitydata. Bull Seismol Soc Am 89:557

Bottari C (2003) Unknown earthquakes in Sicily and Calabriabefore 1000 AD. PhD Thesis. University of Messina, pp185

Buchi E (2000) Dalla colonizzazione della Cisalpina allacolonia di “Tridentum”. In: Buchi E (ed.) Storia delTrentino, II, L’età romana. Bologna, pp 47–131

Caputo R, Helly B (2000) Archéosismicité de l’Égée: étude desfailles actives de la Thessalie. Bull Corresp Hell 124:560–588

Cesano L (1913) Della moneta Enea corrente in Italianell’ultima età imperiale romana e sotto i re ostrogoti.Riv Ital Numis 26:511–551

Croci G, D’Ayala D, Liburdi R (1995) History, observation andmathematical models in the seismic analysis of theValadier abutment area in the Colosseum. Ann Geofis 38(5–6):957–970

Demanet D, Evers LG, Teerlynck H, Dost B, Jongmans D(2001) Geophysical investigation across the Peel boundaryfault (The Netherlands) for a paleoseismological study.Neth J Geosci 80:119–127

De Rossi MS (1874) La basilica di Santa Petronilla pressoRoma, testé discoperta, caduta per terremoto Bull. VulcItal 1:62–65

Dincauze DF (2000) Environmental archaeology. CambridgeUniversity Press, Cambridge, UK, p 587

Doglioni F, Moretti A, Petrini V (eds.) (1994) Le chiese e ilterremoto. Trieste, Italy, p 320

Dolan JF, Pratt TL (1997) High-resolution seismic reflectionprofiling of the Santa Monica fault zone, west LosAngeles, California. Geophys Res Lett 24:2051–2054

Driessen J, Macdonald CF (1997) The troubled island. MinoanCrete before and after the Santorini eruption, Annalesd’archéologie égéenne de l’Université de Liège et UT-PASP. Aegeum 17, p 284

Ellenblum R, Marco S, Agnon A, Rockwell, T, Boas, A (1998)Crusader castle torn apart by earthquake at dawn, 20 May1202. Geology 26:303–306

ENEL (1986) Studi ed indagini per l’accertamento dellaidoneità tecnica delle aree suscettibili di insediamento diimpianti nucleari, Regione Lombardia, Area Viadana,indagini di sismicità storica, rapporto finale, il terremotodel 3 gennaio 1117, ISMES unpublished report RAT-DGF-0012, p 907

Evans A (1928) The Palace of Minos, part II. London (844 pp)

Fäh D, Giardini D, Bay F, Bernardi F, Braunmiller J,Deichmann N, Furrer M, Gantner L, Gisler M, IseneggerD, Jimenez MJ, Kästli P, Koglin R, Masciadri V, Rutz, M,Scheidegger C, Schibler R, Schorlemmer D, Schwarz-Zanetti G, Steimen S, Sellami S, Wiener S, Wössner J(2003) Earthquake catalogue of Switzerland (ECOS) andthe related macroseismic database. Eclogae Geol Helv96:219–236

Fäh D, Steimen S, Oprsal I, Ripperger J, Wössner J,Schatzmann R, Furger A, Spottke I, Huggenberger P(2006) The earthquake of 250 A.D. in Augusta Raurica, areal event with a 3D site-effect? J Seismol (this volume)

Funiciello R, Lombardi L, Marra F, Parotto M (1995) Seismicdamage and geological heterogeneity in Rome’s Colosse-um area: are they related? Ann Geofis 38:927–937

Gaggiotti M (1991) Saepinum. In: Capini S, Di Niro A (eds.)Samnium Archeologia del Molise. Roma, pp 243–246

Galadini F (2006) Quaternary tectonics and large-scale gravita-tional deformations with evidence of rock-slide displace-ments in the Central Apennines (central Italy).Geomorphology (in press)

Galadini F, Galli P (1999) Paleoseismology related to thedisplaced Roman archaeological remains at Egna (Adigevalley, northern Italy). Tectonophysics 308:171–191

Galadini F, Galli P (2001) Archaeoseismology in Italy: casestudies and implications on long-term seismicity. J EarthquEng 5:35–68

Galadini F, Galli P (2004) The 346 A.D. earthquake (Central–Southern Italy): an archaeoseismological approach. AnnGeophys 47:885–905

Galadini F, Galli P, Molin D, Ciurletti G (2001) Searching forthe seismogenic source of the 1117 earthquake in northernItaly: a multidisciplinary approach. In: Glade T, Albini P,Frances F (eds.) The use of historical data in naturalhazards assessment. Kluwer, Dordrecht, The Netherlands,pp 3–27

Galli P (2000) New empirical relationships between magnitudeand distance for liquefaction. Tectonophysics 324:169–187

Galli P, Bosi V (2002) Paleoseismology along the Cittanovafault: implications for seismotectonics and earthquakerecurrence in Calabria (southern Italy). J Geophys Res107:2044 (no. 0, doi 10.1029/2001JB000234)

Galli P, Galadini F (2001) Surface faulting on archaeologicalrelics: A review of case histories from Dead sea to Alps.Tectonophysics 335:291–312

Gasperini P, Ferrari G (1997) Stima dei parametri sintetici:nuove elaborazioni. In: Boschi E, Guidoboni E, Ferrari G,Valensise G, Gasperini P (eds.), Catalogo dei fortiterremoti in Italia dal 461 a.C. al 1990, vol 2. ING-SGA,Bologna, Italy, pp 56–64

Guidoboni E (ed.) 1989a. I terremoti prima del Mille in Italia enell’area mediterranea. Storia, Archeologia, Sismologia.ING-SGA, Bologna, Italy, p 765

Guidoboni E (1989b) Archeologia e sismicità storica: I luoghidel terremoto e le domande evitate. In: Guidoboni E (ed.) Iterremoti prima del Mille in Italia e nell’area mediterranea.Storia, Archeologia, Sismologia. ING-SGA, Bologna,Italy, pp 413–416

Guidoboni E (1996) Archaeology and historical seismology:the need for collaboration in the Mediterranean area. In:Stiros S, Jones R E (eds.) Archaeoseismology. British

412 J Seismol (2006) 10:395–414

School at Athens, Fitch Laboratory Occasional Paper 7, pp7–13

Guidoboni E (2000) Archeosismologia. In: Francovich R, Mana-corda D (eds.) Dizionario di archeologia. Bari, Italy, pp 29–34

Guidoboni E, Santoro Bianchi S (1995) Collapses and seismiccollapses in archaeology: Proposal for a thematic atlas.Ann Geofis 38:1013–1017

Guidoboni E, Muggia A, Valensise G (2000) Aims and methodsin territorial archaeology: possible clues to a strong fourth-century A.D. earthquake in the Straits of Messsina(southern Italy). In: McGuire WJ, Griffith DR, HancockPL, Stewart IS (eds.) The archaeology of geologicalcatastrophe, Geological Society Spec. Publ. no. 171,London, 45–70

Guidoboni E, Muggia A, Marconi C, Boschi E (2002) A casestudy in archaeoseismology. The collapses of the Selinuntetemples (southwestern Sicily): Two earthquakes identified.Bull Seismol Soc Am 92:2961–2982

Hinzen K-G (2005a) The use of engineering seismologicalmodels to interpret archaeoseismological findings inTolbiacum, Germany: A case study. Bull Seismol SocAm 95:521–539

Hinzen K-G (2005b) Archaeoseismological case study of a villain Kerkrade Holzkuil, The Netherlands. In: Tichelman Get al (ed) Het villacomplex Kerkrade-Holzkuil. ADC-rapport 155, Amersfoort

Hinzen K-G, Oemisch M (2001) Location and magnitude fromseismic intensity data of recent and historic earthquakes inthe Norther Rhine Area, Germany. Bull Seismol Soc Am91:40–56

Hinzen K-G, Schütte S (2003) Evidence for earthquake damageon Roman buildings in Cologne, Germany. Seismol ResLett 74:124–140

Hodges R (1995) Gli scavi archeologici a San Vincenzo alVolturno. In: G. De Benedittis (ed.) San Vincenzo al Volturnodal chronicon alla storia. Isernia, Italy, pp 29–54

Improta L, Zollo A, Bruno PP, Herrero A, Villani F (2003)High-resolution seismic tomography across the 1980 (Ms6.9) Southern Italy earthquake fault scarp. Geophys ResLett 30:494 (doi 10.1029/2003GL017077)

Jibson RW (1996) Use of landslides for paleoseismic analysis.Eng Geol 43:291–323

Jones RE, Stiros S (2000) The advent of archaeoseismology inthe Mediterranean. In: McGuire WJ, Griffith DR, HancockPL, Stewart IS (eds.) The archaeology of geologicalcatastrophe. Geological Society Spec. Publ. no. 171,London, pp 25–32

Kanamori H., Anderson DL (1975) Theoretical basis of someempirical relations in seismology. Bull Seismol Soc Am65:1073–1095

Karcz I, Kafri U (1978) Evaluation of supposed archaeoseismicdamage in Israel. J Archaeol Sci 5:237–253

Karcz I, Kafri U, Meshel Z (1977) Archaeological evidence forsubrecent seismic activity along the Dead Sea–Jordan Rift.Nature 269:234–235

Keefer DK (1984) Landslides caused by earthquakes. Geol SocAmer Bull 95:406–421

Kilian K (1996) Earthquakes and archaeological context at the13th century BC Tyrins. In: Stiros S, Jones R E (eds.)Archaeoseismology. British School at Athens, Fitch Labo-ratory Occasional Paper 7, pp 63–68

Korjenkov AM, Mazor E, (1999a) Earthquake characteristicsreconstructed from archaeological damage patterns: Shivta,the Negev Desert, Israel. Isr J Earth-Sci 48:265–282

Korjenkov AM, Mazor E (1999b) Seismogenic origin of theancient Avdat ruins, Negev Desert, Israel. Nat Hazards18:193–226

Lanciani R (1918) Segni di terremoti negli edifizi di Romaantica. Bull Comm Archeol Comunale Roma 45:1–28

La Rosa V (1995) A hypothesis on earthquakes and politicalpower in Minoan Crete. Ann Geofis 38:881–891

Liebeschuetz JHWG (2001) The decline and fall of the Romancity. Oxford University Press, London, UK, p 479

Livadie CA (ed.) (1999)L’eruzione vesuviana delle “Pomici diAvellino” e la facies di Palma Campania. Proc of the IntWorkshop, Ravello July 15–17 1994. Bari, Italy, p 416

Marco S, Agnon A, Ellenblum R, Eidelman A, Basson U, BoasA (1997) 817-year-old walls offset sinistrally 2.1 m by theDead Sea transform, Israel. J Geodyn 24:11–20

Marco S, Hartal M, Hazan N, Lev L, Stein M (2003)Archaeology, history, and geology of the A.D. 749earthquake, Dead Sea transform. Geology 31:665–668

Mastrolorenzo G, Palladino DM, Vecchio G, Taddeucci J(2002) The 472 A.D. Pollena eruption of Somma-Vesuvius(Italy) and its environmental impact at the end of theRoman Empire. J Volcanol Geotherm Res 113:19–36

Mazor E, Korjenkov A (2001) Applied archaeoseismology:Decoding earthquake parameters recorded in archaeolog-ical ruins. In: Krasnov B, Mazor E (eds.) The MakhteshimCountry: laboratory of nature. Pensoft, Sofia, Moscow, pp123–153

McCalpin J (1996) Paleoseismology. Academic Press, SanDiego, 588 pp

McGuire WJ, Griffith DR, Hancock PL, Stewart IS (eds.)(2000) The archaeology of geological catastrophes. Geo-logical Society Spec. Publ. no. 171, London, p 417

Molin D, Guidoboni E (1989) Effetto fonti effetto monumenti aRoma: i terremoti dall’antichità a oggi. In: Guidoboni E(ed.) I terremoti prima del Mille in Italia e nell’areamediterranea. Istituto Nazionale di Geofisica, Bologna,Italy, pp 194–223

Molinari MC (1994) Discussione. In: Francovich R, Noyé G(eds.) La Storia dell’alto Medioevo italiano (VI-X secolo)alla luce dell’archeologia. Firenze, Italy, pp 540–541

Morrisson C (1980) La Trouvaille d’Aïn Kelba et la circulationdes minimi en afrique au début du VI° siècle. Melangés deNumismatique d’Archéologie et d’Histoire offerts àJeanne Lafourie. Paris, pp 239–247

Nikonov AA (1988) On the methodology of archaeoseimicresearch into historical monuments. In: Marinos PG,Koukis GC (eds.) The engineering geology of ancientworks, monuments and historical sites, preservation andprotection. Proc of an Int Symp organized by the GreekNational Group of IAEG, Athens Sept. 19–23 1988.Rotterdam, The Netherlands, pp 1315–1320

Noller JS (2001) Archaeoseismology. Shaking out the historyof humans and earthquakes. In: Goldberg P, Holliday VT,Ferring CR (eds.) Earth sciences and archaeology. Plenum,New York, pp 143–170

Noller JS, Lightfoot KG (1997) An archaeoseismic approachand method for the study of active strike–slip faults.Geoarchaeology 12:117–135

J Seismol (2006) 10:395–414 413

Nur A, Ron H (1996) And the walls came tumbling down:earthquake history in the Holyland. In: Stiros S, Jones R E(eds.) Archaeoseismology. British School at Athens, FitchLaboratory Occasional Paper 7, pp 75–85

Nur A, Cline EH (2000) Poseidon’s horses: plate tectonics andearthquake storms in the Late Bronze Age Aegean andeastern Mediterranean. J Archaeol Sci 27:43–63

Obermeier SF (1996) Using liquefaction-induced features forpaleoseismic analysis. In: McCalpin J (ed.) Paleoseimol-ogy. Academic, San Diego, California, pp 331–396

Papastamatiou D, Psycharis I (1996) Numerical simulation of theseismic response of megalithic monuments: preliminaryinvestigations related to the Apollo temple at Vassai. In:Stiros S, Jones R E (eds.) Archaeoseismology. BritishSchool at Athens, Fitch Laboratory Occasional Paper 7,pp 225–236

Rapp G (1986) Assessing archaeological evidence for seismiccatastrophes. Geoarchaeology 1:365–379

Rapp G, Hill CL (1998) Geoarchaeology, the earth-scienceapproach to archaeological interpretation. Yale UniversityPress, New Haven, Connecticut, p 274

Rea R (1999) Il Colosseo attraverso i secoli. In: Gabucci A(ed.) Il Colosseo. Milano, Italy, pp 161–227

Reece R (1984) Coins. In: Hurst HR, Roskams SP (eds.).Excavations at Carthage: The British Mission. 1. Sheffield,pp 171–181

Rheidt K (1996) The 1296 earthquake and its consequences forPergamon and Chliara. In: Stiros S, Jones R E (eds.)Archaeoseismology. British School at Athens, Fitch Labo-ratory Occasional Paper 7, pp 93–103

Sakellarakis Y, Sapouna-Sakellaraki E (1981) Drama of deathin a Minoan temple. Natl Geogr 159, pp 205–222

Santoro Bianchi S (1996) Dalla sismologia storica all’archeo-sismologia. OCNUS 4:239–248

Seed B (1967) Soil stability problems caused by earthquakes,Soil Mechanics and Bituminous Materials ResearchesLaboratory. University of California Press, Berkeley,California, p 20

Sigurdsson H, Cashdollar S, Sparks SRJ (1982). The eruptionof Vesuvius in A.D. 79: Reconstruction from historical andvolcanological evidence. Am J Archaeol 86:39–51

Sirovich L, Pettenati F (2001). Test of source-parameterinversion of the intensities of a 54,000-death shock ofthe seventeenth century in Southeast Sicily. Bull SeismolSoc Am 91:792–811

Sirovich L, Pettenati F, Cavallini F, Bobbio M (2002). Natural-neighbor isoseismals. Bull Seismol Soc Am 92:1933–1940

Soren D, Leonard JR (1989) Archeologia sismica a Kourion: unapproccio multidisciplinare in azione per un terremoto del

IV secolo d.C.. In: Guidoboni E (ed.) I terremoti prima delMille in Italia e nell’area mediterranea. Istituto Nazionaledi Geofisica, Bologna, Italy, pp 438–449

Stiros SC (1988) Archaeology – A tool to study activetectonics. EOS 69:1633–1639

Stiros SC (1995) Archaeological evidence of antiseismicconstructions in antiquity. Ann Geofis 35:725–736

Stiros SC (1996) Identification of earthquakes from archaeo-logical data: methodology, criteria and limitations. In:Stiros S, Jones RE (eds.) Archaeoseismology. BritishSchool at Athens, Fitch Laboratory Occasional Paper 7,pp 129–152

Stiros SC (2001) The A.D. 365 Crete earthquake and possibleseismic clustering during the fourth to sixth centuries ADin the Eastern Mediterranean: a review of historical andarchaeological data. J Struct Geol 23:545–562

Stiros SC, Dakoronia P (1989) Ruolo storico e identificazionedi antichi terremoti nei siti della Grecia. In: Guidoboni E(ed.) I terremoti prima del Mille in Italia e nell’areamediterranea. Istituto Nazionale di Geofisica, Bologna,Italy, pp 422–438

Stiros SC, Jones RE (eds.) (1996) Archaeoseismology. BritishSchool at Athens, Fitch Laboratory Occasional Paper 7, p 268

Stiros SC, Papageorgiou S, Kontogianni V, Psimoulis P (2006)Church repair swarms and earthquakes in Rhodes Island,Greece. J Seismol (this volume)

Stucchi M, Albini P (2000) Quanti terremoti distruttiviabbiamo perso nell’ultimo millennio? Spunti per ladefinizione di un approccio storico alla valutazionedella completezza. In: Galadini F, Meletti C, Rebez A(eds.) Le ricerche del GNDT nel campo dellapericolosità sismica (1996–1999). GNDT, GruppoNazionale per la Difesa dai Terremoti, Roma, Italy,pp 333–343

Suitner G (1991) Le Venezie. In: Italia Romanica. Milano, Italy,p 436

Traina G (1989) Fra archeologia, storia e sismologia: il casoemblematico del 21 luglio 365 d.C. In: Guidoboni E (ed.)I terremoti prima del Mille in Italia e nell’area medi-terranea. Istituto Nazionale di Geofisica, Bologna, Italy,pp 449–451

Ward-Perkins B (1989) Archeologia e terremoti: le tracceperdute. In: Guidoboni E (ed.) I terremoti prima del Millein Italia e nell’area mediterranea. Istituto Nazionale diGeofisica, Bologna, Italy, pp 409–413

Wells D L, Coppersmith KJ (1994) New empirical relationshipsamong magnitude, rupture length, rupture width, rupturearea, and surface displacement. Bull Seismol Soc Am84:974–1002

414 J Seismol (2006) 10:395–414