7/29/2019 1_036_048

1/6

Surveying Ancient CitiesA ground-level search of abandoned settlements yields enough

artifacts to reconstruct urban history. It even turns upevidence that sharply focused excavation would miss

by Anthony M. Snodgrass and John L. Bintliff

I n August 1981, near the end of ourseason in Greece, we discoveredAskra, the home village of the earlyGreek poet Hesiod. Archaeologists hadsought its location intermittently fora century. Th e 15-hectare site in theBoeotian highlands of central Greecewas imp ortant for its literary associa-tions and for the window it offered onrural Greek life. It was also far beyondour resources for full excavation.The approac h we took to investigat-ing Askraa surface surveyis notmerely a low-budget alternative to tra-ditional archaeological methods basedon excavation. Surveys emb ody a fun-damentally different approach to thestudy of how communities are born,grow and eventually die. They offer abroad sample of life throughout a giv-en site rather than a statistically du bi-ous (albeit exhaustive) slice of a fewsmall plots. In some cases, surveys cangive evidence that contradicts histori-cal accounts. We found, for example,that rural Greece, commonly thoughtto have been deserted during the laterstages of the Roman Empire, was infact a thriving mix of diminished townsand intensively cultivated farmsteads.Survey archaeology was first con-ANTHONY M. SNODGRASS and JOHNL. BINTLIFF directed the survey investi-gations of the ancient Greek towns ofHaliartos and Thespiai and the villageof Askra. Snodgrass holds the LaurenceChair of Classical Archaeology at theUniversity of Cambridge; he graduatedfrom the University of Oxford with a de-gree in classics in 1959 and embarkedon a career in archaeology, earning hisdoctorate with a study of Greek armorand weapons. From 1961 to 1976 hetaught at the University of Edinburgh.Bintliff, Reader in Archaeology at Dur-ham University in England, has particu-la r interests in European social evolu-tion, spatial archaeology an d landscapeevolution. He received his Ph.D. fromCambridge and taught at the Universityof Bradford from 1977 to 1990.

ceived as a technique for rural areas,and so we had to modify it for applica-tion to an urban site containing poten-tially unmanageable numbers of arti-facts. Since investigating Askra, we haveapplied the methods we derived thereto two larger Boeotian cities, Haliar-tos and Thespiai. In the meantime thetechnique has also been adapted to asmall coastal townsh ip on the Cycladicisland of Keos, a major Minoan city inCrete, a town in Etruria and an inlandcity in the Ploponnse.The modern ground surface, as longas it is accessible and has not been al-tered, yields material representative ofevery period during which the site hasbeen occupied. Potsherds and roof tilesare most common, followed by stoneimplements and building fragments,bronze coins and terra-cotta objects.By systematically covering the entireaccessible area of a form er city, pickingup all distinctive material, recording itslocation and determining its identity,it is possible to construct a plot of thecity's periods of occupation, growth,shrinkage and shifts in location.To be sure, urban surveys are effec-tive only under certain prescribed con-ditions. The site must be largely free ofmod ern construction, and it must havebeen subjected to cultivation (the moreintensive the better) at least intermit-tently since the abandonment of thecity. The processes that bring potteryand other artifacts to the surface fromthe buried layers underneath are as yetunderstood only in a general way, butit is clear that they work. Cultivationplays a vital role in bringing materialto the surface, as does the gradual ero-sion of topsoil. Material from the mostrecent periods is generally overrepre-sented in the surface layer, whereas an-cient items are underrepresented.Although they may be highly effec-tive, surveys will never supplant exca-vation entirely. After all, the dating ofartifacts found on the su rface is possi-ble in large part only thanks to decadesof painstaking research by excavators,

who have noted the sequences and as-sociations of each class of material intheir stratified deposits.Moreover, there are many questionsabout the history of a city that surfacesurveys cannot answer. Unless the sur-face archaeologist is lucky enough tofind the foundations of recognizablebuildingsfortification walls, monu-mental public structures and the like-there is little the method cancon-tribute to tracing changes in politicalpower and independ ence, for example.(The relative abundance of importedpottery and similar items can furnishonly indeterminate clues.)E ven when it was clear we had tosurvey Ask ra rather than trying toexcavate it, the site still posedproblems. Surface survey is a well-es-tablished archaeological practice, butsurveys of areas that include urbancenters had seldom been attempted be-cause of the huge range in the concen-tration of artifacts. In Mediterraneanlands the beirren mountain slopes yielda handful of pieces per hectare, where-as the densest urban areas display 20or even 20 0 artifacts per square meter.Anyone but a Mediterranean archae-ologist may find such figures hard tobelieve, but they are real enough. Fur-thermore, artifacts are not distributedin neat, discrete packets correspondingto ancient sites. Instead the nearly mil-lionfold variation in artifact densitytakes the form of a gradual distribu-tion extending hundreds or even thou-sands of meters from "primary" sites.The surveyor faces an extraordinarilyrich body of data.This embarrassment of riches posesproblems, however, in devising collec-tion strategies and determining whatconstitutes surface evidence for a site.Only after the highs and lows of the lo-cal distribution areknown is it possibleto place all the evidence in context. Aconcentration of artifacts that mightpass for a rural farmstead at one sitecould represent only part of thegen-

88 SciENTinc AMERICAN March 1991

7/29/2019 1_036_048

2/6

eral surface scatter at another. At thesame time, there are limits to theamount of material from a single sitethat archaeologists ca n fruitfully cata-logue and study. Early traverses of Ask-ra yielded 10 or more artifacts persquare meter, implying that a completesurvey might yield 1.5million artifacts.We devised a sampling method thathas since served well for larger sites insubsequent seasons. The work is donein two stages. The first stage covers theentire accessible surface area so thatthe site can reveal all its distinctiveartifacts and so that its phases ofoccupation can be dated. (Askra, forexample, proved to have been occu-pied intermittently for more than 4,000years.) The second stage is a moreprecise examination that provides anaccurate measure of the total den-sity of artifacts and helps substan-tiate dates derived from the initial pass.

We divided Askra into a series oftransects less than half a hectare inarea and surveyed each in its entirety.Walkers 15 meters apart counted visi-ble artifacts in a strip five meters wideand picked up any material they judgedlikely to provide a date. Then the teamscoured a 300-square-meter subsectionof each transect. Workers picked overevery square centimeter of ground byhand and counted all the artifacts,again picking up items that appeareduseful fo r dating.On average,we retained about 20 dis-tinctive pieces from the first walking ofeach transect, and we added a further15 from each intensive sample. Morethan half of the approximately 2,000artifacts collected from Askra turnedout to have chronological value, vin-dicating the on-the-spot judgments ofthe walking teams. As a result, we couldsubstantiate the mapping of times of

occupation of even such a relativelysmall site by hundreds of accuratelydated pieces for each period.F rom the very start of our survey,we had noted that virtually ev-ery site, however small, was sur-rounded by a halo of finds whose den-sity decreased as we moved away fromthe site itself. The most widely accept-ed explanation for such halos is thetime-honored ancient practice of fertil-izing the ground with the manure ofanimals kept in the vicinity of domesticpremises. Pieces of discarded potterybecame mixed with the dung; the den-sity of items in a given location is in-dicative of the density of cultivation.This hypothesis received striking con-firmation in 1986, when Brian E. Da -vies and Andrew Waters of the Univer-sity of Bradford found that concentra-tion patterns of heavy metals in the

SURFACE-SURVEY ARCHAEOLOGISTS comb a modern vine-yard in central Greece for artifacts left behind by millennia ofhuman activity. Sites of abandoned cities may yield more than100 objects per square meter. By dating the objects they find,archaeologists can determine the periods during which anarea was occupied and estimate its changing boundaries.

SCIENTIFIC AMERICAN March 1991 89

7/29/2019 1_036_048

3/6

soil across our sites matched the densi-ty patterns of ancient pottery.It has long been known that heavymetals are deposited where people liveand work. Davies and Waters's findingssuggest that the refuse-deposition ac-tivities of 2,500 years ago have leftclearly graded and quantifiable tracesin both the form of visible potsherdsand invisible p ollutants. T his reinforcesour hypothesis that the rural sites werenuclei of intensive a gricultural activity.On a vastly larger scale, the cities thatwe investigated have their ow n muchbroader and denser pottery halos. Aseconomic historians have long main-tained, Boeotian towns were occupiedby cultivators.O ur technique had taken time toevolve, but it was ready by thetime our survey carried us upto the walls of Thespiai and Haliartos,the two main cities in who se territorywe were working. In particular, we wereprepared to test the results of our sur-

veys against written history. Whereasan obscu re village like Askra, for all itsliterary associations, had no connecteddocum entary history, these two townshad both been members of the Boeo-tian League and virtually independentpolitical entities.Thespiai earned a certain fam e for itslong opposition to Thebes, the mostpowerful city of the league. In 480 B.c.,when Thebes prudently sided with thePersians invading Greece, Thespiai en-joyed its greatest moment of glory asit sent 700 soldiers (from a total popu-lation of perhaps 10,000) to die withLeonidas of Sparta in the attempt tohold the pass of Thermopylae.In 424 B.C. Thespiai for once foughtside by side with Thebes, winning avictory over the Athenians at Delion.The city again suffered heavy losses,however, and the following year theThebans were able to exploit this weak-ness and compel Thespiai to dismantleits fortification walls. Not surprisingly,when T hebes achieved its military and

political zenith by an unprecedentedvictory over Sparta in 371 B.C.a battleactually fought on Thespian territoryThespiai supported Sparta. The lossput an end to the city's heyday.In the meantime Thespiai had pro-duced its most famous daughter, thebeautiful courtesan Phryne, who be-came the mistress of the sculptor Prax-iteles and posed for his most famousstatue, a nude Aphrodite. She dedicat-ed in her native city another of hisworks, a statue of Love that made Thes-piai a tourist attraction for the rest ofantiquity.Haliartos was by comparison a small-er and less famous town. Its popula-tion numbered no more than 5,000 atits peak. It came to notice in 395 B.c. asthe scene of a minor skirmish in whicha force of Thebans and Athenians am-bushed a group of Spartans. The Spar-tan General Lysander, then the mostpowerful individual in Greece, wasamong the few killed. The most memo-rable event in the city's official historywas also the last: in 1 71 B.c. Haliartosbacked the M acedonians and other en-emies of Rome in a war fought onGreek soil. The town was razed, its in-habitants killed or sold into slaveryand its land apportioned among thecitizens of Athens, which had takencare to join the Roman side. The terri-tory of both Thespiai and Haliartos isnow entirely free of permanent humanhabitation. The land is cultivated fromvillages some distances away.

T his official history furnished somecrude guidelines for our work,even though some episodes wereof the kind that a surface survey cannotexpect to trace or illuminate. T he storyalso left open many questions wehoped to illuminate. We hoped to es-tablish, for example, how changes intotal urban population affected thedistribution of occupied farmsteads inthe surrounding countrysideand howthe extreme case of city destructioninfluenced the exploitation of the sur-rounding land. One important issue,of course, was whether the rural andurban sectors grew simultaneously orat each other's expense. We also won-dered how the boundary between twostates might be reflected in the surfaceevidence.The picture that we were eventuallyable to construct was, in this and oth-er ways, much more detailed than anyavailable from documentary sources.For Askra, of course, there was hardlyany documentary information, so anynew knowledge was pure gain.Askra gave clear indications of a longbut interrupted occupation and a pro-

90 SCIENTIFIC AMERICAN March 1991

7/29/2019 1_036_048

4/6

gressive shift of several hundred me -ters in the nucleus of the settlementover time. A small area of the site wasfirst inhabited for a period around2500 B.c. After a very long break, thesame locale was first reoccupied andthen enlarged into a substantial villageof 1,000 people or more between about900 B.c.and A.D. 100. A second, muchshorter break seems to have intervenedbefore the final period of settlementbetween about A.D. 300 and 1600;since then the site has reverted to openfarmland.The two breaks are inferred onlyfrom negative evidence: we found nomaterial whose date fell within thosetime spans. The first interruption, how-ever, is so long that there can be nodoubt of it; furthermore, for part of thetime, a neighboring hilltop site was in-habited, suggesting an alternative fo-cus of local settlement. And the secondbreak is corroborated by the Greektravel writer Pausanias, who visited thearea around A.D. 170 and described itas deserted.There is little to say about the first pe-riod of occupation. The second, 1,000years long, embraces the life of thepoet Hesiod (circa 700 B.c.) in the vil-lage's earlier stages, when Askra was nomore than a few scattered dwellings.Ensuing centuries saw first a steadygrowth and filling in of the settlement,then a marked shrinkage of populationand inhabited area.Throughout the period, the nucleusof the settlement shifted to the south.When the site was reoccupied in thefourth century A.D., the shift resumeduntil, by the final phases of Askra's oc-cupation in Late Byzantine and Turkishtimes, the settlement had no overlap atall with the original nucleus of prehis-toric and early historical times.Lke Askra, Haliartos had a pre-historic forerunner, sited on thehighest ground in the area. Anda phase of apparent abandonment, al-though much shorter than that ofAsk-ra, preceded the establishment of thecore of the historical city. ClassicalHa-liartos spread progressively down theslopes to the south and north until itattained a population of 5,000 or moreand an area approaching 30 hectares.The easternmost extremity is lost un-der the houses of the modern townthat shares the same name.One interesting discovery was thatalthough by about 400 B.c. the territo-ry of the more populous Thespiai wasmore densely covered with permanentstructures than that of Haliartos, thereverse had been true three centuriesearlier at the beginning of the histori-

cal period. Apparently the smaller townwas faster to colonize its rural territorywith independent farmsteads.During the classical period, the an-cient city was surrounded by wallswhose foundations are still partiallyvisible. The original nucleus on highground became the city's acropolis andhad an inner fortification of its own.(All these walls were most probablytorn down in 171 B.C.)As predicted from the official histo-ry, there is an abrupt break in the se-quence of dated pottery in the secondcentury B.c., and for the rest of antiqui-ty the greater part of the site was en-tirely deserted. A short distance to theeast, however, on the outer fringesof the classical city, a new settlementgrew up in Byzantine and early Turkishtimes. From that point onward, in anunbroken cycle, the decline and deser-tion of one location has been accom-panied by the simultaneous growth ofan alternative center of population notfar away. Tw o such shifts are trace-able between the 17th and 19th cen-turies A.D., before the foundation ofthe modern town and the revival of thename "Haliartos" around 1900.Th e unique feature of the story ofHaliartos is the ruthlessness and thor-oughness of the Roman sack of 171B.c. More gradual shifts, in contrast,are typical of all the major settlementsin the area we study. These shifts ex-plain why the modern villages with theofficial names of Askraia, Thespiai andHaliartos all lie at a certain removefrom their ancient forerunners.T hespiai proved the most daunt-ing of the urban surveys weundertook. We had expected itwould be larger than Haliartos, but wewere hardly prepared for a city thathad exceeded 120 hectaresm ore thana square kilometerin area during theperiod of its greatest extent in the fifthand fourth centuries B.C.Th e two-stage procedure of observa-J.

tion and counting had to be repeated598 times in all. It turned up well over10,000 datable artifacts. Wewere ableto draw up seven successive plans ofthe town's occupation and desertionover a span of some 7,000 years.Unlike Haliartos, Thespiai did notgrow from a single compact nucleus sit-ed on top of a prehistoric settlement.The material of the eighth, seventh andsixth centuries B.c. is grouped in halfa dozen separate clusters, suggesting ascatter of hamlets only welded into asingle urban complex by the growthofthe high classical era in the fifth andfourth centuries.Th e inevitable sequel was a declinein size and, once again, a shift in thenucleus of occupation. On e later phasestands out with special clarity: in themiddle of the Roman Imperial period(between A.D. 30 and 300), Thespiai'sinhabitants found it expedient to builda new fortification, enclosing a drasti-cally reduced area of 12 hectares.This polygonal circuit incorporatedmany blocks of classical stone; it stooduntil the late 19th century, when, iron-ically, it was destroyed by a classicalepigraphist eager to get at its inscribedstones. Its outline is still faintly visibletoday. Not surprisingly, we found thatmaterial of the later Roman period(from about A.D. 300 to 600) was veryheavily concentrated within and aroundthis enclosure. Of the 32 transects thatproduced more than six later Romanpieces, 25 lay inside the fortified poly-gon and the rest immediately to theeast and northeast.Although late Roman Thespiai wasmuch reduced in size, the large numberof farmsteads we found in the sur-rounding countryside indicate that itlay at the center of a thriving rural econ-omy. After the classical period, the lateRoman is the most intensive period ofrural settlement in the entire history ofthe central Grecian landscapeit wasalso the time of the reoccupation ofAskra. This late Roman recovery is now

MODERN SURFACE

i/

ANCIENT ARTIFACTS are brought to the surface by a continuous process of cultiva-tion, erosion of topsoil and other geologic effects that are not yet fully understood.Older objects, from lower strata, are typically underrepresented at the surface.SCIENTIFIC AMERICAN March 1991 91

7/29/2019 1_036_048

5/6

LATE ROMAN THESPIAI

| ] UNSURVEYED AREAS- ROADSAND PATHS

ANCIENT CITY of Thespiai was explored by surface survey.Dates of items found show that the settlement was occupiedfo r more than 7,000 years. Archaeologists marked the site off

ARTIFACT DENSITY

me 5 10 15

CLASSICALTHESPIAI

POTTERY FINDS GEOMETRIC

into transects (areas shown by black lines), which were thenexamined. A specified part of each transect was painstaking-ly searched to verify the conclusions of the broad survey.92 SCIENTIFIC AMERICAN March 1991

7/29/2019 1_036_048

6/6

being revealed by surveys elsewherein Greece. It is a good example of adevelopment for which the documen-tary sources, with their gloomy pictureof the age, had done nothing to pre-pare us.Thespiai also provides a correctiveto widely held views of postclassicaldecline for later periods. A settlementsurvived there in Byzantine times, en-tirely to the east of the polygonal en-closure; the ruins of several churchesbear witness to continued vitality.In the end, although the city outlast-ed all its contemporaries, large andsmall, it died. When the British travelerColonel Leake visited Thespiai in 1802,he found a few inhabited houses stillstanding on the eastern part of the site.A few years later the last inhabitantsmoved to the hilltop village just tothe north, where their descendants stillflourish. The ancient city on the plainwas given over entirely to cultivationfor the first time in nearly 3,000 yearsa near-perfect condition for archaeolog-ical survey.T he results of our surveys inBoeotia cast light not only on theconditions there but also on thenature of the questions that archaeolo-gists and historians attempt to answer.Official history presents events in aform that is memorable and, as far asthe facts allow, gratifying to those whoread it. Individuals and organizationsappear to formulate consistent policiesand carry them through either more orless successfullyas was once said ofthe Times of London, official history de-scribes what ought to have happenedrather than what did happen.Additional constraints affect the caseof ancient cities: the written sourcesthat survived were chosen mostly fortheir literary quality rather than theirfaithfulness to events. Some ancienthistorians did share the modern desirefor objective truth, and other kindsof documentary sources such as coinsand inscriptions also survive. But allsuch records are vulnerable to distor-tions, and all too often the sum totalof historical evidence yields a portraitof a city that begins with a founda-tion legend and ends with a visitationof the punishment of the gods. It issignificant that some modern authori-ties have turned as often to poets andphilosophers such as Homer, Plato orAristotle as they have relied on theprose historians.Conventional archaeology offers onlylimited help in augmenting the histori-cal picture. Financial and political exi-gencies usually combine to prevent theexcavation of more than a limited part

ODYSSEY OFFALIARTOS shows how towns migrate as well as growing or shrink-ing. The city encompassed nearly 30 hectares before it was razed by Romantroops. Subsequent settlements skirted the edges of the old city for two millennia;a new Haliartos was built to the east of the ancient site just under a century ago.of an ancient urban complex (in con-trast with the potentially complete un-earthing of small prehistoric settle-ments). Digs must be sited on land freeof existing buildings and available forpurchase or expropriation. Only by therarest good fortune does excavation re-sult in a valid cross section of urbanlife. Indeed, many investigators wouldopenly disavow such a mundane objec-tive. Better by far to locate the civiccenter or the main sanctuaries, wherethey may be able to make some spec-tacular finds and have a virtual certain-ty of forging some kind of link with theofficial history.Furthermore, even with the best ofintentions, the small samples ofmateri-al and the peculiar nature of preserveddeposits makes "commonsense" infer-ences from excavation notoriously un-reliable. The contents of graves, for ex-ample, may present a poor picture ofthe structure of a society and its atti-tudes toward the living. The goods in-terred with a corpse may or may notcorrelate with social or economic stand-ing. Implements found in a grave orscenes depicted in it may or may notmatch those in everyday life.Conventional archaeology, then, runsthe risk of tautology if excavationsproceed only in areas designated byhistorical accounts. And it can yield athoroughly skewed sample of ancientlife if researchers rely on the contentsof rare caches of well-preserved ma-terial. Surface survey, in contrast, ex-tracts a limited but valid picture fromthe detritus and other evidence thathuman habitation cannot help leavingon the ground.Those inside and outside archaeol-ogy have often asked us whether theresults of surface investigation should

not be put to the test by excavation.Ideally, perhaps they should. But evenexcavation yields findings that are of-ten inconclusive, and they are only val-id for the area actually dug. Further-more, resources seldom extend to bothsurvey and excavation, so the choice ofone means excluding the other.Our surface survey of Thespiai cov-ered 99 percent of the city's maximumarea in the initial transects and about13 percent in the regularly spaced in-tensive samples. For the same cost, wecould have excavated perhaps two orthree trenches five meters square downto virgin soil: a total of about 0.005percent of the city's area. Even if the ex-cavation uncovered an equivalent bodyof dated material, there is no questionwhich method produces a better statis-tical sample of the physical traces of acity's past.

FURTHER READINGARCHAEOLOGICAL SURVEY IN THE MED-ITERRANEAN AREA. Edited by Donald R.Keller and David W. Rupp. Oxford, Brit-ish Archaeological Reports, Internation-al Series S155, 1983.THE CAMBRIDGE/BRADFORD BOEOTIANEXPEDITION: THE FIRST FOUR YEARS.John L. Bintliff an d Anthony M. Snod-grass in Journal of Field Archaeology,Vol. 12, No. 2, pages 123-161; Summer1985.BEYOND THE ACROPOLIS: A RURAL GREEKPAST. Tjeerd H. van Andel and Cur-tis Runnels. Stanford University Press,1987.OFF-SITE POTTERY DISTRIBUTIONS: A RE-GIONAL AND INTERREGIONAL PERSPEC-TIVE. John L. Bintliff and Anthony M.Snodgrass in Current Anthropology,Vol. 29, No. 3, pages 506-513; June1988.

SCIENTIFIC AMERICAN March 1991 93