Bailey Jaa 2007

7/28/2019 Bailey Jaa 2007

1/26

Time perspectives, palimpsests and the archaeology of time

Geoff Bailey *

Department of Archaeology, University of York, The Kings Manor, York YO1 7EP, UK

Received 25 June 2006; revision received 27 August 2006Available online 23 October 2006

Abstract

This paper explores the meaning of time perspectivism, its relationship to other theories of time used in archaeologicalinterpretation, and the ways in which it can be implemented through an analysis of the palimpsest nature of the materialworld we inhabit. Palimpsests are shown to be a universal phenomenon of the material world, and to form a series of over-lapping categories, which vary according to their geographical scale, temporal resolution and completeness of preservation.Archaeological examples are used to show how different types of palimpsest can be analyzed to address different sorts ofquestions about the time dimension of human experience, and the relationship between different types of processes anddifferent scales of phenomena. Objections to the apparently deterministic and asocial character of time perspectivism,and its apparent neglect of subjective experience and individual action and perception, are dealt with. The line of thinkingdeveloped here is used, in its turn, to critique other approaches to the archaeology of time, and conventional understand-ings of the relationship between past, present and future.

2006 Elsevier Inc. All rights reserved.

Keywords: Contemporaneity; Durational present; Epirus; Palimpsest; Time perspectivism; Timescale

Introduction

For many archaeologists, time depth is whatgives archaeology its distinctiveness as an intellectu-al discipline. For others, it is the emphasis on themateriality of human existence, once derided as a

second-hand method of studying human activities,but now turned into a virtue by the many studiesof material culture that emphasize the active roleof artifacts and material structures in human actionand interaction. These two themes are linked, for itis the durable properties of the material universe

that give to human awareness a sense of timeextending beyond individual lives and perceptions,and to archaeologists the opportunity for empiricalexploration of human activities beyond the reach ofpersonal observation, oral testimony or writtenrecords.

The past two decades have witnessed a prolifera-tion and diversification of theoretical discussionsabout time and its impact on archaeological inter-pretation, which have served to open up a far-reach-ing exploration of this link between time and thematerial world (see Bailey, 2005; Lucas, 2005; forsummaries). Discussion has followed severalintertwined though often divergent themes, drawingon a wide range of sources of inspiration includingthe intrinsic properties of archaeological data

Journal of Anthropological Archaeology 26 (2007) 198223

www.elsevier.com/locate/jaa

0278-4165/$ - see front matter 2006 Elsevier Inc. All rights reserved.

doi:10.1016/j.jaa.2006.08.002

* Fax: +44 1904433902.E-mail address: [email protected]
mailto:[email protected]:[email protected]

7/28/2019 Bailey Jaa 2007

2/26

themselves, and other disciplines such as socialanthropology, history, geology, paleontology, phi-losophy and mathematics.

One theme, which shows many points of conver-gence with the literature on site formation processes

(Schiffer, 1976, 1987), and with the concerns of geol-ogists and paleontologists (Behrensmeyer, 1982;OBrien and Lyman, 2000), is the examination ofthe temporal and spatial properties of archaeologi-cal data, how we measure these, more or less arbi-trarily, and how differences in temporal scale andresolution of archaeological datasets constrain orexpand the questions we can investigate empiricallyabout the past (Renfrew, 1981; Gamble, 1986;Ebert, 1992; Rossignol and Wandsnider, 1992;Stahl, 1993; Stern, 1993, 1994; Zvelebil, 1993;Ramenofsky and Steffen, 1998; Lock and Molyne-

aux, 2006). The latter theme, following Bailey(1981, 1983, 1987), is sometimes labeled as timeperspectivism (Fletcher, 1992; Murray, 1997,1999b; Holdaway and Wandsnider, 2006; Wandsn-ider, 2004; Hull, 2005). Important variants on thistheme that draw more heavily on other disciplinesto address differences of timescale, but which usual-ly eschew the time perspective label, are the appli-cation of ideas drawn from the Annales school ofhistory (Bintliff, 1991; Knapp, 1992), and the useof non-linear dynamic theory (Van der Leeuw and

McGlade, 1997).A second and readily comprehensible theme is

the examination of the temporal awareness of pastpeoples, their sense of past and future, how thatinfluenced their behaviour, and how it has variedor developed during the course of human history,whether for cultural or neuropsychological reasons(Clark, 1992; Murray, 1999a; Alcock, 2002;Bradley, 2002). A closely related theme is the dura-ble properties of the material record as an extradimension to human awareness and action, throughits capacity to symbolize the passage of time or toshape human activities, especially in the form ofthe built environment such as burial mounds anddwelling structures (Bradley, 1991, 1993; Bailey,1993; Fletcher, 1995).

Some discussions attempt to integrate elementsof all three themes, often with an emphasis on thesubjective element in temporal awareness (includingour own as archaeologists), and its cultural, politicalor philosophical referents, drawing on contempo-rary social theory and philosophy (Shanks andTilley, 1987; Gosden, 1994; Thomas, 1996; Harding,

2005; Lucas, 2005).

Throughout this literature there is a basic con-trast between the differential temporal patterns ofthe material world that past people may have con-sciously recognized and used in their social lifeand cosmology, and the differential temporal pat-

terns inherent in archaeological deposits that weas archaeologists seek to exploit to say more aboutthe past and our relationship to it.1

My emphasis in this paper is on the three percep-tions that inspired the original definition of timeperspectivism: the relatively coarse temporal resolu-tion and palimpsest nature of much of the archaeo-logical record; the possibility that the increased timedepth and varied time resolution of observationafforded by archaeological data might allow us toperceive phenomena and processes not visible atsmaller scales of observation; and the arbitrary nat-

ure of the boundary between past and present. Iconsider more carefully the definition of time per-spectivism and its theoretical and operational impli-cations, analyze the concept of palimpsest anddefine some of its variant properties, examine thesorts of processes that may become visible on longerand coarser timescales, and address the problem ofhow to reconcile such longer-term phenomena withthe emphasis on individual action and perceptionthat has dominated much recent archaeologicalinterpretation. For example, I draw on field data

from my own experience, in particular the Klithiproject, concerned with a 100,000-year record ofactivity in the Epirus region of northwest Greeceat the scale of archaeological site and region(Fig. 1), and at ethnographic and archaeologicalscales of observation, and more fully discussedand published elsewhere (Bailey, 1997; Baileyet al., 1998; Green et al., 1998; Green, 2005). Thistheme of time perspectivism has been slowest totake root, generated most criticism, and createdthe most puzzlement and resistance, the reasonsfor which I touch on later.

1 This corresponds to what I have previously described assubjective and objective approaches to time (Bailey, 1983),subjective meaning time concepts as experienced by prehistoricpeople, and objective meaning the temporal concepts as used byarchaeologists looking in from the outside. Objective here doesnot mean superior or neutral, nor does it deny that objectivestudies in this sense have their own varying subjectivities. Thedistinction has given particular problems to those who wish toblur the boundary between the perceiving mind and the perceivedobject, and I avoid the usage here in the interests of obviating

unnecessary misunderstandings.

G. Bailey / Journal of Anthropological Archaeology 26 (2007) 198223 199

7/28/2019 Bailey Jaa 2007

3/26

Time perspectivism defined

As originally stated, time perspectivism is thebelief that differing timescales bring into focus dif-ferent features of behaviour, requiring differentsorts of explanatory principles (Bailey, 1981, p.103), or the belief that different timescales bringinto focus different sorts of processes, requiring dif-ferent concepts and different sorts of explanatoryvariables (Bailey, 1987, p. 7). Expressed in thisway, the emphasis of these formal definitionsappears to be on the notion of how processes outthere operate, independently as it were, of thehuman observer, but with the added implicationthat what we observe of those processes dependson our timescale of observation or our timeperspective.

Many early objections to time perspectivismreflected the Keynesian dictum that critics of anew idea will tend to fluctuate between a belief that[it is] quite wrong and a belief that [it is] sayingnothing new (Keynes, 1973; [1936], p. xxi). In thequite wrong category is the reaction from thepost-processual wing of theoretical archaeology,who attempted to dismiss time perspectivism on

the grounds that it is simply a cloak for environ-

mental determinism or ecological functionalism,an attempt to justify the emphasis on environmental

or economic determinism in the study of the long-term and to dismiss social and cognitive variablesas short-term noise, and hence an attack on post-processualism (Moore, 1981; Tilley, 1981; Shanksand Tilley, 1987; Thomas, 1996). I have some sym-pathy with this view to the extent that it derivesfrom a general misunderstanding of the original for-mulation and in particular a misreading of my 1983paper, in which I equated the long-term with biolog-ical and environmental processes and the short-termwith social ones (Bailey, 1983, p. 180). That state-ment was a descriptive generalization about previ-ous views, particularly those advocated bypalaeoeconomists in the 1970s (e.g. Higgs andJarman, 1975), not a prescriptive advocacy for thefuture. As Hull (2005) has correctly identified,the intent of the passage in question was to movethe debate beyond such simple equivalences ratherthan to reinforce them. Instead, the post-processualreaction simply moved the debate from oneextreme, the palaeoeconomic one, to anotherextreme, and thus replaced one mono-scalarapproach with another, when what was needed

was an exploration of differences in scale and the

Fig. 1. Location map of Epirus, Greece, showing general topography, sites mentioned in the text and other Palaeolithic or Mesolithic sitesthat have produced excavated and dated material. Relief is shaded at 0600 m, 6001000 m and >1000 m. The 100 m contour shows theapproximate position of the coastline at the maximum lowering of sea level during the glacial maximum at about 20,000 years ago. Alarger number of Palaeolithic surface finds, not shown, are also scattered throughout the region. All sites, other than isolated occurrencesof individual artifacts, are cumulative palimpsests of varying resolution and time depth and in combination represent different episodes ina regional spatial palimpsest incorporating fragments of human activity extending over a time depth of at least 120,000 years. Data fromBailey et al. (1997).

200 G. Bailey / Journal of Anthropological Archaeology 26 (2007) 198223

7/28/2019 Bailey Jaa 2007

4/26

nature of the interactions between them. Time per-spectivism does not remove consideration of socialfactors from the long-term at all. Rather it questionsthe appropriateness of uncritically borrowing theo-ries and concepts from social anthropologists,

cultural geographers, historians and sociologists,or indeed ecologists, who work with very differentcontexts, and with very different scales and typesof observation and evidence.

In the nothing new category is the response ofsome of those archaeologists who have found inspi-ration in the Annales School of history. Bintliff(1991), for example, has asserted that time perspec-tivism is essentially a derivative of the Annalesprogram, while other authors (e.g. Harding, 2005)have described the Annales approach to prehistoryas an example of time perspectivism. This oversim-

plifies the varied intellectual genealogies of these dif-ferent approaches. There are undoubted points ofconvergence (cf. Fletcher, 1992; Smith, 1992a) butalso of considerable divergence, not least in thesources of empirical inspiration and the range oftime spans and time scales embraced by thesedifferent approaches (see Murray, 1997, 1999b).2

The Annales program is in fact an extremely hetero-geneous combination of ideas and approaches,which have changed and developed over a periodof more than 100 years (Knapp, 1992), and archae-

ologists who feel most comfortable identifying withthis approach are generally those who work onrecent millennia with a time depth and resolutionof data quite similar to historians, inclined to con-sign phenomena of greater time depth to Braudelssomewhat indeterminate longue duree. Time per-spectivism, in contrast, has been inspired by thechallenges of working with the much longer-timespans of the deeper archaeological past, and withmaterial evidence unaided by other sorts of records.It has also advocated a more critical and flexiblestance towards defining where the boundary is tobe drawn between event and structure, or betweencontinuity and change, in common with many oth-ers who have grappled with similar issues (Plog,1973; Dunnell, 1982; Schiffer, 1988; Smith, 1992a).In addition, time perspectivism highlights the rela-

tivity of knowledge that must result from observa-tions made by individuals located at differentpoints in a time continuum and working with differ-ent timescales of observation.

These reactions suggest that there is much that is

implicit in the formulation of time perspectivism,and a need for clarification. In particular there isan ambiguity in the use of the term time scale,which has often been used as short hand to describeone, or both, of two different concepts. The firstrefers to a simple question of relative size. Thus asmall-scale phenomenon in this sense is one thathas limited extent in time (and usually, but not nec-essarily, in space as well), let us say the actions of anindividual on a particular day, whereas a large-scalephenomenon occurs over a longer temporal (andgeographical) span, let us say the 100 years war or

the diffusion of prehistoric agriculture. The secondconcept of scale refers to the resolution of measure-ment available to describe different phenomena.Thus short-lived phenomena require highly resolvedmeasures of time for their observation and study,while larger and more extensive phenomena requireand permit a coarser scale of measurement. Ambi-guity can be avoided by using time span, longer orshorter, for the former, and time resolution, fineror coarser, for the latter, though I shall continueto use timescale when I wish to refer to both mean-

ings simultaneously.With these clarifications in mind, we can go on to

identify four different meanings implied by timeperspectivism:

(1) Different phenomena operate over differenttime spans and at different temporal resolutions,the meaning that is uppermost in the original defini-tions. This is the substantive definition of time per-spectivism, a definition in terms of how history issupposed actually to have happened. It is the sensein which other commentators have most oftenunderstood the term, but it is also the most difficultmeaning to pursue further because it raises the veryproblematic issue of how if at all different scales ofphenomena are supposed to interact.

(2) Different sorts of phenomena are best studiedat different time scales, that is to say using differenttime spans and different temporal resolutions. Thus,for example, the analysis of small-scale phenomenasuch as individual agency, inter-personal interac-tions and perception, which have become such adominant tendency in recent archaeological inter-pretation, is better focused on observations of,

say, present-day practices or recent historical peri-

2 Harding and Bintliff, like many other influential Europeancommentators (e.g. Bradley, 2002; Hodder, 1990), have archae-ological interests centered on the European Neolithic and BronzeAges, whereas my own interests cover a longer temporal rangeand a more diverse geographical one, hence one reason for the

differences of perspective between us.


7/28/2019 Bailey Jaa 2007

5/26

ods rather than the deeper prehistoric past. Theargument here is not that such small-scale phenom-ena did not exist in the deeper past and have a sim-ilar impact on past lives to what we observe in thecontext of our own, but that these phenomena are

much more difficult to investigate in earlier periodsbecause of the poorer resolution, quality and detailof the available data. This definition refers to howvarious periods of the past appear to us and aboutthe nature of the data available from different peri-ods and how we study them, rather than about howthe past really was, in and of itself, or how it wasexperienced by past people. This is a methodologicaldefinition, about how we study past phenomena,and how what we can observe of past phenomenais conditioned by our timescale of observation andthe data at our disposal. This leads on to consider-

ations about why archaeologists working in differ-ent time periods tend to prefer different sorts ofquestions with correspondingly different theoreticalorientationsor to put it the other way round,why archaeologists with different sorts of theoreticalinterests and research questions tend to gravitate todifferent periods of the past.

(3) A third meaning of time perspectivism isthe distorting effect that differential time perspec-tives have on our perception and understandingof the world, what we might call the strict defini-

tion of time perspectivism. We can grasp this thirdmeaning more clearly if we consider how we usethe term perspective when referring to spatialobservations. Spatial perspective actually conveystwo distinct notions. The first is that objectsbecome increasingly distorted with increasing dis-tance from the observers position in space. Theyappear to shrink until they literally disappear overthe visual horizon. Of course, we refer to thiseffect as a trick of perspective that we need tocorrect for, and are well aware that if we travelto the distant horizon, objects re-appear and growback to their normal size, and the horizon recedesagain into the far distance.

The second notion is that an awareness of per-spective allows us to see more clearly the spatialrelationships between different phenomena. Com-pare for example the way in which the layout of atown might appear from the vantage point of apedestrian observer, and its appearance in an aerialphotograph. The aerial photograph allows us toappreciate the correct spatial relationships betweenthe different parts of the town, indeed to see the

town more clearly in relation to its wider geograph-

ical setting, in a way that is far more difficult for theground observer. But, at the same time, the aerialview cannot reveal the details of small-scale rela-tionships apparent on the ground or the experienceof a three-dimensional world with buildings and

other features of different heights and visibility asviewed by the pedestrian observer.Perspectivism in this sense is a double concept,

conveying both the negative effect of distortion withincreasing distance that needs to be corrected, andthe positive effect of putting into their proper rela-tionship different scales of spatial patterning. It isprecisely this double aspect that I wished to conveyin the original use of time perspectivism. Phenom-ena that are more distant in time seem to shrinkuntil they disappear over a time horizon that we calla point of origin, thereby giving us a much distort-

ed conception of historical pattern. Equally, work-ing at larger timescales should enable us to seelarger-scale relationships that are obscured at asmaller scale. Those who want to reconstitutesmall-scale phenomena in the deeper past as theymight have been perceived by past individuals areresponding to the first aspect of the definition.Those who want to examine larger scale phenomenaare responding to the second.

(4) Finally, time perspectivism refers to the wayin which our observation of time is conditioned by

particular cognitive and symbolic representationsof time that are specific to particular cultures, statesof brain evolution, forms of social organization, orworld views. We could call this the subjective defini-tion of time perspectivism, an exploration of the dif-ferent ways in which different people, both past andpresent (including archaeologists), have thoughtabout the time dimension and their place within it.

It should be clear that the various ideas exposedby these different definitions overlap to some degree.For example, the subjective aspect of time perspec-tivism as experienced by past peoples must be partof the substantive record that we attempt to uncoverby archaeological study. At the same time the waywe view the past as archaeologists falls within thesubjective domain. And as we know from our ownexperience, different conceptions of time can be heldsimultaneously by different individuals or embodiedin different practices and beliefs within a given soci-ety. All these definitions pose quite formidable chal-lenges of empirical implementation, and I arguehere that an understanding of the palimpsest natureof the material universe is key to meeting that

challenge.


7/28/2019 Bailey Jaa 2007

6/26

Palimpsests and the structure of the material world

The term palimpsest has a long usage in archae-ology (Wandsnider and Holdaway, in press). It isalso used more widely as a powerful metaphor in

many other disciplines, for example in the study ofthe built heritage, in literary and theoretical dis-course, in discussions of memory, most famouslyin Freuds study of the unconscious, and in the per-forming arts (McDonagh, 1987; Roy, 1997; Cryder-man, 2002; Jones and Shaw, 2006). In this literature,the emphasis is on the interplay between erasure andinscription, often with cross reference to archaeo-logical data, between the text and the material medi-um through which it is expressed, and how thatinterplay creates complex layered and multi-tempo-ral entities that disrupt conventional views of tem-

poral sequence.The dictionary definition of a palimpsest (from

the Greek roots pakim meaning again, and wax,meaning rub or scrape) is paper, parchment,etc., prepared for writing on and wiping out again,like a slate, or a monumental brass turned andre-engraved on the reverse side (New Oxford Dic-tionary, 3rd ed., 1967). In common usage, a palimp-sest usually refers to a superimposition of successiveactivities, the material traces of which are partiallydestroyed or reworked because of the process of

superimposition, or the traces of multiple, overlap-ping activities over variable periods of time and thevariable erasing of earlier traces (Lucas, 2005, p.37). These definitions reveal a twin aspect to theconcept of a palimpsest. In its extreme form apalimpsest involves the total erasure of all informa-tion except the most recent. But palimpsests canalso involve the accumulation and transformationof successive and partially preserved activities, insuch a way that the resulting totality is differentfrom and greater than the sum of the individualconstituents.

In archaeology, palimpsests are typically viewedas a handicap, an unfortunate consequence of hav-ing to rely on a material record that is incomplete,and one that requires the application of complextechniques to reconstitute the individual episodesof activity, or alternatively a focus on the best pre-served and most highly resolved exemplars at theexpense of everything else, or the application of the-oretical or imaginary narratives to fill the gaps,which are in consequence immune to empirical chal-lenge. An alternative tradition of thought and one

made explicit in various papers published in the ear-

ly 1980s is to turn this limitation of the archaeolog-ical record into a virtue. Thus Bailey (1981, p. 110)refers to palimpsests as offering an opportunity tofocus on a different scale of behaviour, Binford(1981, p. 197) to a structured consequence of the

operation of a level of organization difficult, if notimpossible, for an ethnographer to observe directly,and Foley (1981, p. 14) to long term trends [that]may be of greater significance to the prehistorianthan the understanding of a few short events. Herethe emphasis is as much on the accumulative andtransforming properties of palimpsests as on theloss and destruction of evidence.

These preliminary considerations suggest that thenotion of palimpsest is a complex one that requiresfurther examination, and I suggest that we can dis-tinguish five categories.

True palimpsests

True palimpsests are palimpsests in the strictsense of the term in which all traces of earlier activ-ity have been removed except for the most recent.Imagine the floor of a Neolithic house that is regu-larly swept clean of all the various artifacts, materi-als, objects and debris that have accumulated theresince the last episode of cleaning. We might describethe floor as receiving successive depositions of mate-

rial, representing different layers of activity, eachone of which is wholly or largely removed beforethe accumulation of the next one. Only the very lastlayer of artifacts would remain in place immediatelybefore the abandonment or collapse of the house totell us something about the activities carried outthere. This process of cleaning and removal of mate-rial is of course exactly what is implied by the dis-tinction between primary and secondary refuse(Schiffer, 1976, 1987). Moreover, the fact that sec-ondary refuse appears to be very common in archae-ological contexts suggests that true palimpsests, orat any rate partial palimpsests, are likely to be awidespread occurrence (Schiffer, 1985).

If the cleaning process was less than complete,some material traces of the earlier episodes mightremain in place, for example smaller items more eas-ily trodden into the floor surface or missed by thenaked eye or the bristles of the broomresidualprimary refuse (Schiffer, 1987, p. 62). These earliertraces, most probably a biased selection of the origi-nal materials deposited in earlier episodes of activi-ty, would thus become incorporated into the final

layer. Moreover, that final layer might represent


7/28/2019 Bailey Jaa 2007

7/26

activities quite different from those that producedthe earlier layers. The floor area in question mighthave been used for a period as a general living area,then as a food preparation area or a storage area,and finally as a garbage dump. We would be fooled

if we thought that the final layer represented theonly activity that took placed inside the house. Wemight also be fooled if we thought that this finallayer was representative of how the house was usedthroughout its full history, since the formation ofmany archaeological deposits may be the result ofsuccessional use (Binford, 1981, p. 200). Rowley-Conwy (1994) provides a graphic illustration ofthe potential confounding effects that result fromsuch palimpsest effects in relation to the near-perfectpreservation and high stratigraphic resolution ofhouse floors at the Nubian site of Qasr Ibrim. If

we cannot untangle the detailed history of use ofparticular features at a site like Qasr Ibrim, we areunlikely to achieve better results with any otherarchaeological data. The successive demolition ofolder streets and buildings during urban renewaland their replacement by new ones represents amodern example of the true palimpsest (cf. Jonesand Shaw, 2006).

The definition of a true palimpsest, then, is asequence of depositional episodes in which succes-sive layers of activity are superimposed on preced-

ing ones in such a way as to remove all or most ofthe evidence of the preceding activity. Of coursethe true palimpsest may appear impossible to distin-guish archaeologically from a single episode,although various other sorts of material traces inthe immediate vicinity of the true palimpsest maygive us some clues about the likely existence of theearlier layers that were subsequently removed. Thesweepings from the house floor may reappear as sec-ondary refuse in adjacent ditches, for example.

Cumulative palimpsests

A cumulative palimpsest is one in which thesuccessive episodes of deposition, or layers ofactivity, remain superimposed one upon the otherwithout loss of evidence, but are so re-workedand mixed together that it is difficult or impossi-ble to separate them out into their original con-stituents. This is, I think, the sense in whichpalimpsest is most commonly used in archaeolo-gy. It is also a very common occurrence. Manyarchaeological deposits are very obvious palimps-

ests in this sense. The stone tools in a layer of a

Palaeolithic cave, for example, usually representthe aggregation of many different episodes ofknapping, use and discard that have becomecompressed into a single layer or surface, andcannot be resolved back into the individual epi-

sodes of activity. The same is true of shells inmany shell middens. We know that the manymillions of shells that make up a sizeable middenmust represent a multitude of separate shellgath-ering episodes, but resolving the mass of shellsinto these separate episodes except at a verygross level of stratigraphic or chronological reso-lution is rarely possible. The full range of vari-ability represented by the individual episodesburied within these palimpsests is usually inacces-sible and we can only observe the average ten-dencies represented by the palimpsest as a

whole. In similar vein, Stern (1993, p. 215) refersto time-averaged accumulations of materialremains, which corresponds to the concept of acumulative palimpsest proposed here, and hasfurther elaborated the implications for the studyof early hominid activities and land-use patterns(Stern, 1994).

Unlike the true palimpsest, where much of theevidence has been lost but the resolution of the finalepisode may be very high, the cumulative palimpsestis characterized not so much by loss of material but

by loss of resolution. The material traces from all ormany of the successive episodes of deposition arestill there, including the final one in the sequence,but so mixed together as to blur the patterningpeculiar to each individual episode. The mixing pro-cess may result from different causes, from churningand displacement of material by repeated humanactivity and foot traffic on the surface of the accu-mulating deposit or from low rates of deposition.But the end result is the same, mixing of materialsand blurring of the original pattern. This is not tosay that all detail is lost. Apart from the averagetendencies still detectable in the sequence as awhole, it is entirely possible that occasional highspots of patterning representing individual episodesmay be preserved at random, unaffected by the mix-ing process.

At the Klithi rockshelter, for example, a typicalPalaeolithic cave palimpsest, where loose sedi-ments accumulated quite slowly with considerablevertical and horizontal mixing of materials, weoccasionally found conjoined materials still inthe original position in which they had been dis-

carded, anatomically adjacent bones, or refitting


7/28/2019 Bailey Jaa 2007

8/26

flakes struck from the same flint nodule (Baileyand Woodward, 1997; Wenban-Smith, 1997).However, these are isolated occurrences amongstthe many hundreds of thousands of stone artifactsand bone fragments, no more than might be

expected to have survived by chance in the mixingprocess, and certainly too small a sample to sus-tain any useful generalizations about the patternand integrity of all the individual episodes thatmake up the palimpsest as a whole.

Similar effects can be found in rock art, for exam-ple in some of the classic sites of southwest France,where some rock surfaces and engraved stone slabshave been repeatedly engraved with outlines of ani-mals superimposed on previous engravings. Theresult is a maze of uninterpretable lines, but withoccasional outlines of individual animals, or

identifiable heads and limbs, standing out fromthe background noise (Bahn and Vertut, 1988; Figs.25 and 93).

The boundary between a true palimpsest and acumulative palimpsest is not a sharp one and thetwo types may grade into each other. In fact agreat many archaeological palimpsests probablyshare elements of both, being characterized bothby mixing of material of different ages and age-re-lated loss of material resulting from successive epi-sodes of clearance and removal or progressive loss

of in situ material by physical and chemical decay.The key trait they share in common is that bothresult from the repetition of activities and thedeposition of material in the same location, orin similar locations with considerable overlap.The key difference is that cumulative palimpsestsmay acquire a significance that is greater thanthe sum of the individual constituent episodes,both for the people who used them and for thearchaeologists who study them. Consider, forexample the large shell mounds that result fromthe slow incremental deposition over many hun-dreds of years of thin layers of shells to formmajor features of the landscape. Such cumulativepalimpsests are prominent in the archaeologicalrecord precisely because they are formed by therepeated accumulation of materials in the sameplace, from which derives their archaeological vis-ibility and relative ease of discovery and analysis,and also their symbolic significance for the peoplewho used them (cf. Luby and Gruber, 1999). AsBinford (1981, p. 197) puts it the greater theapparent disorganization, the more intense the

use of the place in the past.

Spatial palimpsests

At this point we might be tempted to argue that ifthe different activities represented in a cumulativepalimpsest and their resulting material traces had

been carried out in different locations, each episodewould have preserved its original pattern intact andwe would be able to recover the original pattern ofinter-episode variability without loss of resolution.That hope has been a powerful incentive to a reori-entation of Palaeolithic excavation and survey strat-egies away from deeply stratified cave sites to open-air locationsthat and the recognition that thestratified sites must represent a tiny fraction of thetotal material output of Palaeolithic societies(Foley, 1981; Bailey et al., 1997). However, the hopethat spatial segregation might result in greater

resolution is almost certainly illusory.If the individual episodes of shellgathering that

make up a large mound, or the individual assem-blages of stone tools that make up a layer in a strat-ified cave, had been dispersed across the landscape,many would now be lost to view. They would havebeen degraded by weathering, chemical attack orother destructive processes, obscured by later over-burden of sediments and vegetation, or displacedand disaggregated by soil erosion. Whichever waywe look at this distribution of spatially distinct

activities, we are still likely to be confronted withloss of material or loss of resolution. In other words,we are still dealing with a palimpsest, except that weare dealing with a palimpsest at a larger spatialscale. All that happens when activities become spa-tially segregated in this way is that they merge into amuch larger-scale palimpsest, the sedimentarypalimpsest that characterizes the surface of the wid-er landscape. Some episodes are buried or obscuredfrom view, some are destroyed, some are disturbed,some retain high integrity and resolution of pattern-ing and some are accessible to archaeological dis-covery and analysis. The reality is that mostsettlement patterns reconstructed from archaeologi-cal site distributions are remnant settlement pat-terns (Dewar and McBryde, 1992), in which sitesrepresenting cumulative palimpsests achieve promi-nence and visibility for reasons that have less to dowith their significance to their original occupantsthan with the frequency of re-visiting and re-use ofspecific locations in a wider landscape.

Open-air sites like Pincevent (Leroi-Gourhan andBrezillon, 1972) in France, the Meer site in Belgium

(Cahen et al., 1979), Boker Tachtit in the Negev


7/28/2019 Bailey Jaa 2007

9/26

(Marks, 1977), the Dunefields site in the South Afri-can Cape (Parkington et al., 1992), or the horsebutchery site at Boxgrove (Roberts and Parfitt,1999), have attracted considerable attention andinterest in Palaeolithic studies precisely because they

appear to represent individual episodes of activityand high resolution events, distinct from the moreusual cave palimpsest. Whether or not these sitesare truly individual episodes or palimpsests of somesort is open to discussion. But they certainly pre-serve a higher resolution and integrity of patterningthan the typical cave deposit. However, the impor-tant point is that these sites are absolutely rare.They are no more common in relation to the gener-ality of Palaeolithic sites in the wider landscape thanthe occasional pockets of high resolution data thatwe have noted above as occasionally standing out

in the cumulative palimpsest of a cave deposit. Inseparating out the individual episodes of ourpalimpsest into a scatter of spatially distinct loca-tions, we have not escaped the palimpsest problemat all. All we have done is translate it to a larger spa-tial scale, that of the geomorphological palimpsest,a mixture of episodes of soil formation and move-ment, erosion and sediment accumulation.

The palimpsest nature of these spatially discreteepisodes of activity is further reinforced by theproblem of chronological correlation. The generali-

ty of archaeologically detectable finds, for examplestone tools or sherds in the plough-soil or on theground surface, are often very difficult to dateexcept at a coarse scale. However, many of thesesites may well represent precisely the sort of spatial-ly discrete body of data sought after in the attemptto disentangle cumulative palimpsests. Usually,what is gained in spatial resolution is lost in termsof chronological control. Yet the most commontype of archaeological find in many regions isopen-air surface finds lacking any stratigraphicintegrity at all or any means of dating except thecrudest, and often disregarded as a consequence.However, they have considerable informationpotential because of their precise location in space,whatever they may lack in temporal resolution.They thus form an important component of spatialpalimpsests.

The lack of dating control highlights anotherfeature of all palimpsests, the problem of contem-poraneity. Objects in a layer, or more preciselythe deposition of those objects, may be said tobe contemporaneous, but if the layer is a palimp-

sest, this can only true within certain margins of

error. True contemporaneity of two or more suchevents might be said to occur if they are all linkedas part of the same sequence of operations. Con-

joining flakes from the same stone-tool reductionsequence, and anatomically adjacent bones from

the same animal skeleton, are obvious examples,and it is often the high frequency of such conjoinsthat identifies high-resolution, short-lived episodesof activity such as those recorded in sites likeMeer and Pincevent. But these are absolutely rareboth in cumulative palimpsests and spatialpalimpsests. Similarly, we might refer to two dif-ferent settlements within the wider landscape ascontemporaneous if they were inhabited simulta-neously over the same span of months or years.However, methods of dating control and correla-tion that enable us to achieve such precision with

a large sample of sites are as yet unavailable, andmay be a physical impossibility.

It is theoretically conceivable that such methodsof high-precision correlation might be devised inthe future. However, all materials decay, some ofcourse more slowly than others, and the furtherback in time one goes, the more that is lost. Decayof radioactive isotopes is, of course, the basis formany dating methods, but the radioactive isotopeswith short half-lives that give us higher resolutiondating also only work over shorter and more recent

time spans. Whether cosmogenic or other datingmethods based on the accumulation of physical orchemical properties can defy this loss of resolutionwith the passage of time remains unclear. But itseems likely that the holy grail of a high-resolutiondating framework that can be extended to every cor-ner of the archaeological record is an unattainablegoal that defies the physical laws on which our uni-verse is based. In archaeological interpretation, thereality is that in order to combine sufficient datatogether to make a large enough sample for analy-sis, we inevitably end up aggregating data from tem-porally distinct episodes of activity. Thus, incomparing different episodes of activity, we haveto make certain assumptions about the time depthwithin which we are willing to accept as contempo-raneous the various events or materials to be com-pared and this is as true of intra-site spatial analysis(Galanidou, 1997) as it is of inter-site analysis(Bailey et al., 1997; Papaconstantinou and Vassilop-oulou, 1997; see also Papagianni, 2000). Contem-poraneity is thus an arbitrary concept with noabsolute measure, and the resolution that we can

achieve in making chronological correlations


7/28/2019 Bailey Jaa 2007

10/26

depends both on the dating methods at our disposaland the questions we are trying to investigate (Papa-constantinou, 1986). This is not a peculiarity or lim-itation of dealing with archaeological data, but anatural consequence of working with palimpsests

and the physical laws of our universe.I suggest that we call these large-scale distribu-tions spatial palimpsests, a variant of the cumulativepalimpsest but distinct from it and defined as a mix-ture of episodes that are spatially segregated butwhose temporal relationships have become blurredand difficult to disentangle. As with true palimps-ests, the boundary between cumulative and spatialpalimpsests is not a sharp one. Both may be charac-terized by a variety of locations of activity and bydifferent degrees of spatial and temporal integrity.The key difference is rather one of geographical

scale.One other variant of the spatial palimpsest worth

noting before we move on is the spatial disaggrega-tion of materials that were once accumulated in thesame place. A characteristic example of this phe-nomenon is the one described earlier in relation toour hypothetical Neolithic house floor, the truepalimpsest, where previous layers of material havebeen cleared away and re-deposited elsewhere. Theerosion of soil from a hill slope and its re-depositionin a sedimentary basin is another example of this

type of spatial palimpsest. The hope in such circum-stances is that the combination of spatially distinctassemblages of material might allow us to reconsti-tute the sequence of episodes originally carried outin the same place. This hope too may be illusory,because the spatial separation of materials, especial-ly if it results from cleaning and clearance activitiesin the archaeological context, is almost certain toresult in loss of resolution or blurring of patterningand the creation of other sorts of palimpsests. Thustrue, cumulative and spatial palimpsests not onlyform a graded series of types with considerableoverlap, but they may be combined as the productsof an interlinked series of actions, activities or pro-cesses to create a sort of composite of palimpsests ata larger scalea palimpsest of palimpsests!

Temporal palimpsests

A temporal palimpsest is an assemblage of mate-rials and objects that form part of the same depositbut are of different ages and life spans. On firstdescription this sounds like a cumulative palimpsest

by another name. However, in the cumulative

palimpsest, the association of objects of differentages is really an aggregation due to the effect of mix-ing together what were originally distinct episodesof activity or deposition. The temporal palimpsestcomprises what, from the point of view of a cumu-

lative palimpsest, might be viewed as a single epi-sode, a so-called closed find such as a shipwreck,a burial chamber or the room of a house, whereall the materials are found together because theyare constituents of the same episode of activity ordeposition.

Oliviers (1999) discussion of the Late Hallstattprincely grave of Hochdorf in southern Germanyprovides an illuminating example. This is one of anumber of similar mounds associated with fortifiedIron Age hill settlements attributed to the 6th centu-ry BC. Inside the chamber was a range of grave

goods, including clothing, jewelry, drinking cups,and other fittings and furnishings. As Olivier showsin an elegant analysis, there are a number of differ-ent periods in this process of deposition, which haveoccurred between the lifetime of the objects andtheir final placement in the grave (Olivier, 1999, p.126). The archaeological funerary assemblage thusrepresents a series of different temporalities andincorporates the accumulation of a whole series ofdifferent events between the death of the man andthe final occupation of the monument.

It might be objected that the interment of theprincely chief was an event that must have occurredon a single occasion, but the point of Oliviers anal-ysis is to demonstrate that we cannot know what thedate of that event was, except within a very widetemporal envelope of several hundred years (Olivier,2001). This takes us back to the problem of contem-poraneity identified above. In contrast to cumula-tive and spatial palimpsests, where temporalresolution is limited by the mixing process itselfand by the available methods of chronological cor-relation, the temporal palimpsest is the result ofdeliberate combinations of materials before theirentry into the archaeological record, and our inabil-ity to date them with greater precision is due moreto differences in the age of the various objects them-selves, rather than to post-depositional loss of reso-lution or the imperfection of available datingtechniques.

Palimpsests of meaning

Study of the life history or cultural biography

of objects has proved to be a productive line of


7/28/2019 Bailey Jaa 2007

11/26

inquiry (Schiffer, 1987; Gosden and Marshall,1999), and opens up investigation of another typeof palimpsest effect.3 A palimpsest of meaning canbe defined as the succession of meanings acquiredby a particular object, or group of objects, as a

result of the different uses, contexts of use andassociations to which they have been exposedfrom the original moment of manufacture to theircurrent resting place, whether in the ground, amuseum, a textbook, an intellectual discourse, orindeed as objects still in circulation and use. Itis distinct from all the other types of palimpsestsso far discussed in that it can apply to an individ-ual object, and because it brings us more obvious-ly into the domain of subjective time experience.Lucas (2005) provides some good examples of thiseffect. Stonehenge, for example, is not only a Neo-

lithic and Bronze Age monument, but an IronAge one, a Medieval one and a modern one, withdifferent significance for different people in succes-sive periods, and perhaps a different significancefor different people within the same society,including, in modern society, archaeologists, heri-tage managers, Druids, New Age enthusiasts andforeign tourists. Similarly, Olivier (1999, p. 127),in discussing the Hochdorf grave, refers to astratification of meanings developed throughoutthe archaeological life of the various objects both

before and after deposition, some of which areidentifiable by physical modifications, includingthe ongoing significance of these objects asarchaeological remains or museum exhibits.

It may be objected that the sequence of differ-ent meanings that an object acquires during its

journey through time is not strictly a palimpsest,because we can resolve the accumulation of mean-ings associated with the object into its originalcomponent meanings. I think this is highly ques-tionable. Let us take as an example a flint toolthat started out as a knife with a sharp cuttingedge, and was then converted into a blunt-edgedscraping tool, which required the removal bymicro-flaking of the original sharp edge. We canidentify a change of meaning in this case preciselybecause the flint flake has undergone physicalmodification. But the physical modification hasitself removed some of the characteristics that

would have allowed us to identify the originaluse of the implement. Modification in this way,by definition, must remove or obscure some partof the evidence by which the earlier meaningcould be identified. In this example, the working

edge of the flint tool is a good example of a truepalimpsest. We can sometimes infer the earlier his-tory of a specific artifact by looking at the flakesremoved in the process of modification, or otherartifacts that represent earlier stages in this pro-cess. That of course is the basis for analyses oftechnological reduction sequences in the produc-tion and modification of stone artifacts (e.g. Dib-ble, 1987). For any given artifact, however, itsplace in the reduction sequence is, to greater orlesser extent, a matter of inference.

Other changes of meaning not accompanied by

such physical modification might be much harderto reconstitute. We can sometimes say somethingabout the meaning of an object by looking at thecontext in which it was found and its associationwith other objects. But, of course, the context of dis-card may be different from the context of use, andalmost certainly a palimpsest of some sort. Or itmay be part of a votive offering, a burial hoard ora funerary offering, in which case its significanceat the time of deposition may be quite different fromits previous meaning.

Thus in trying to identify the meaning of anobject, whether we study it in isolation or in the con-text of other materials, we cannot escape thepalimpsest effect, which must of necessity makethe task of disentangling the successive meaningsdifficult if not impossible to achieve with any cer-tainty. Arguably we might be able to disentanglethe full history of meanings if we could talk to theindividuals who used or owned the object duringits career. But that would necessarily limit the ageof the objects which could be treated in that way,even if it were possible to track down everyonewho might have made use of the object or had someother association with it.

Moments in time

As we have worked through the different sorts ofpalimpsests, it has become increasingly difficult toidentify any situation or location, whether fromthe archaeological past, or in the contemporaryworld, whether it is in the built environment of amodern city or an archaeological context, in an

institutional building or outdoors, that does not

3 Life history and biography are the terms most widely usedin this context, but are slightly odd usages for inorganic materialsthat were never alive in the biological sense. History or career

trajectory are alternative terms that avoid this connotation.


7/28/2019 Bailey Jaa 2007

12/26

constitute some sort of palimpsest. Even individualobjects do not escape the palimpsest phenomenon.What about the simple stone flake removed by a sin-gle blow to provide a cutting tool? Surely that rep-resents a single action, a moment in time? But let

us consider the matter further. Simple reflection tellsus that such a stone artifact from, say, the Palaeo-lithic period must represent at least three momentsin time, the moment when the raw material was firstacquired, the moment when it was first shaped intoan artifact, and the moment when it was finallythrown away. In fact, we must extend this to afourth moment, the moment when it was recoveredby an archaeologist, a fifth moment, when it wasillustrated in an archaeological publication orexhibited in a museum case, and all the subsequentmoments when it was referred to in the course of

discussions such as this, to say nothing of all theprevious moments of use and storage between itsfirst manufacture and its eventual disposal. It is, inshort, a palimpsest of meanings with a duration thatreaches from the very distant past to the presentday.

The material world is, of necessity, a compositeof objects of differential duration, which representat the very least either temporal palimpsests orpalimpsests of meaning. Material objects are bydefinition durable. If they did not have such prop-

erties they would not exist, or not in a way thatanyone could know about. In fact the notion thata material object can represent a moment in timeis self-contradictory. Material objects by definitionhave duration, a duration that extends from atleast as early as the time when they were first cre-ated to the current moment of observation or dis-cussion, and indeed will most likely extend farinto the future. Moments in time that leave nomaterial traces are unknowable, at least from thearchaeological past.

In short, palimpsests are neither exceptions, norinconveniences, nor oddities that need to betransformed into something else before they canbe interpreted and understood. On the contrary,palimpsests are universal, an inherent feature ofthe material world we inhabit. They are not somedistorted or degraded version of a message thatneeds to be restored to its original state before itcan be interpreted. To a large extent they are themessage. In so far as the palimpsests we study asarchaeologists differ from those we encounter inour day-to-day environment, they differ only in their

scale and resolution.

Working with palimpsests

How then should we deal with a palimpsest, letus say a cumulative palimpsest at the scale of anindividual archaeological site? In the archaeologi-

cal context the natural tendency is to try and unrav-el the palimpsest into its constituent parts. This is aperfectly legitimate strategy if the techniques andmaterials are available to facilitate it. What is more,we cannot really know what the limits of resolutionof any palimpsest are until we try to find out. Avariety of techniques including improved methodsof dating, taphonomic analyses that attempt toidentify the post-depositional histories of differentobjects, conjoining studies, and many others, canbe harnessed to such a strategy. Estevez et al. (inpress) provide a good example of the process of

peeling away the individual constituents of shellmidden sites in Tierra del Fuego, where, unusuallyfor shell deposits, individual episodes of activityare separated by sterile layers of sediment. Wemight call this the microscopic tendency, the ten-dency to seek understanding by working downthrough successively smaller scalesultimately tothe small-scale of individual actions, beliefs, andsocial interactions that we find most familiar interms of our everyday expectations. However, weshould not delude ourselves into thinking that by

doing this we are disposing of palimpsests andthereby making the data more easily interpretable,or that we will ultimately reach some pristine andirreducible core of meaning or moment in time.What we are doing is changing the scale and per-haps the form of the palimpsest, and making it ame-nable to the interrogation of a different set ofquestions.

If we try to take the process of disentanglementtoo far, we may end up with individual episodestoo small or limited in number to sustain any gener-alization, as noted earlier, or worse with pseudo-ep-isodes that cannot bear the weight of interpretationput upon them. Sometimes this type of approachmay serve to demonstrate the limits of resolution.Holdaway et al. (1998), for example, have applieda detailed program of geomorphological and radio-metric dating to surface finds of stone tools in NewSouth Wales and have shown that seemingly con-temporaneous sites and materials that one mightbe tempted to treat as components of a coherent set-tlement or social system refer, in fact, to quite dispa-rate temporal episodes, with sharp temporal

discontinuities and long hiatuses within a time


7/28/2019 Bailey Jaa 2007

13/26

envelope of at least 6000 years. Each temporal epi-sode appears to indicate only one spatial fragmentof the original regional pattern with which it wasassociated, and this is exactly what might be expect-ed of a large-scale spatial palimpsest.

Another strategy is to go the other way, what wemight call the macroscopic tendency, which seeksunderstanding by placing phenomena in a wideningperspective of large-scale comparison. In this casewe accept the palimpsest for what it is as a generaltendency, and in effect shrink the palimpsest to asingle episode, a single dataset, in order to examineit in the wider comparative context of other suchdata sets. Again, we should not imagine that byshrinking the cumulative palimpsest in this wayand expanding our scale of comparison we areremoving the palimpsest problem. As should be

obvious from earlier discussion, changing the scaleof observation in this way is simply to move froma cumulative palimpsest to a spatial palimpsest orfrom a small-scale spatial palimpsest to a larger-scale one.

Whichever direction we move in, whether alongthe microscopic or the macroscopic pathway, weare going to be confronted with a palimpsest ofsome sort. What is different about these differentsorts of palimpsests is their spatio-temporal scaleand their resolution. No one type of palimpsest is

the best, except in relation to some a priori set ofexpectations. Each is appropriate to the examina-tion of different sorts of questions. Nor can we real-ly find out what the appropriate level ofinterpretation is until we try to find out.

Timescales and processes

What then are these different sorts of questionsand substantive issues that we should be focusingon at different scales of observation? What are thedifferent sorts of processes that are revealed at thesedifferent scales? What can we do with large-scalepalimpsests, be they cumulative, spatial or tempo-ral? This is perhaps the most difficult aspect of timeperspectivism.

An example: the eroded landscapes of Epirus

Consider the eroded hill and mountain slopes inthe European countries that border the northernMediterranean, a phenomenon familiar to archaeol-ogists who conduct survey and excavation in these

regions. Such features of course are fairly prominent

in the archaeologists perception of landscapebecause they are potent factors in variously buryingand preserving archaeological material or exposingit to view and perhaps destroying or reburying itin the process. In many areas the effects of erosion

are clear to see, sometimes resulting in barren bad-lands landscapes bare of vegetation, which appearto have lost whatever productive potential for cropagriculture, animal husbandry or arboriculture theymight once have had. Much of this erosion can bedated in broad terms to the postglacial period inassociation with the expansion, establishment andintensification of agricultural practices, and thecauses and consequences of this recent erosion,and the extent to which it may be considered as det-rimental, are much debated (Van der Leeuw, 1998;Hordern and Purcell, 2000; Grove and Rackham,

2001).The usual culprit is supposed to be human inter-

vention, the cutting down of trees for timber, the useof deep ploughs or tractors for cultivating fragilehill soils, or the over-grazing of domestic goats,capable of eating almost anything but the toughestthorns, including the plastic bags and labels usedin archaeological excavation and the freshly washedunderwear of ones field crew left out to dry in thesun. Little wonder that goats are credited with suchpowers of destruction! We might say that goats (or

tractors or tree-fellers) are the cause of erosion.Moreover the progressive impact of erosion hasaggravated worries in our conservation-minded eraabout the loss or irreversible degradation of ourcapital reserves of soil, and stimulated policiesaimed at removing goats, tractors or tree fellingfrom areas at risk, so that the protective cover ofshrubs and trees can be allowed to regenerate.

This pattern is especially clear in the Epirusregion. This is a region of complex topographyand changes of elevation extending from sea levelto the heights of the Pindus Mountains over2600 m across a distance of less than 100 km(Fig. 1). The deeper history of human settlementreaches back over a time span of at least 100,000years, informed by a series of cave sequences andopen air artifact scatters that represent a typicalcombination of mostly cumulative and spatialpalimpsests of varying resolution (Bailey et al.,1997). Here extensive erosion with tracts of barehillside is particularly prevalent on the flysch geolo-gy, a type of metamorphic sandstone that producessoils especially susceptible to erosion. But equally

spectacular areas of bare ground can be found in


7/28/2019 Bailey Jaa 2007

14/26

limestone country, notably the curious red beds,famously associated with Palaeolithic implements,extensive areas of heavily gullied bare soil, as atKokkinopilos, whose origins and causes of erosionare the source of notorious controversy. Much of

this erosion is most probably of postglacial dateand has occurred within the past 10,000 years andsome of it perhaps very recently.

But if we extend our timescale back to the past100,000 years of landscape history, as we wererequired to do by the demands of a Palaeolithicarchaeological sequence, or indeed the past 1 mil-lion years, it becomes apparent that erosion toohas a much longer history. Many of these erosionepisodes cannot be dated, but some can be by look-ing for artifacts or other dateable material in thedown-stream sediments produced by the erosion,

and it is clear that there is a succession of such epi-sodes extending well back into the Pleistocene(Macklin et al., 1997). During the last glacial period,and almost certainly during earlier climatic cyclestoo, erosion took place on a massive scale, resultingin the accumulation of huge fans of sediment at thefoot of the higher mountains, and the accumulationof thick accumulations of alluvial deposits in riverand lake basins. Indeed the scale of erosion quitedwarfs the supposed impact of human interventionin the postglacial period. This erosion of course

took place long before the appearance of domesticgoats or the invention of tractors and some of itlong before the appearance of human communities,and is largely attributed to the impact of cold anddry glacial climates that removed most of the treecover and further accelerated the breakdown of soiland bedrock through freezethaw effects.

On the longest time spans of all, it is apparentthat the Epirus landscape has been subjected to pro-gressive tectonic compression and uplift over tens ofmillions of years, in which offshore sediments creat-ed by earlier cycles of erosion have been compactedand uplifted to produce the hard-rock geology wesee today. Thus the underlying tectonic instabilityhas made the land surface especially susceptible todisturbance, whether triggered by earthquakes, cli-matic effects or human intervention (King et al.,1997).

The erosion of the postglacial, far from appear-ing to be an exceptional effect of recent millennia,turns out to be quite normal in relation to the long-er-term history of the physical landscape, andindeed perhaps less dramatic than in earlier periods.

Since erosion has a much longer history than the

domestic goat, it becomes hard to pin all the blameon the latter. On the contrary, from this longer-termperspective, it seems more likely that goat husband-ry, so far from being the cause of erosion, representsa peculiarly successful adaptation to a degraded

landscape that was in existence long before humansettlement and cannot be made productive forhuman benefit in any other way. Whereas on theshorter timescale of the postglacial, it appears thatgoats cause erosion, on the longer-timescale of thePleistocene the roles of cause and effect appearreversed, such that it would more appropriate tosay that erosion causes or, better, selects forgoats.

Moreover what applies in the temporal dimen-sion also applies in the spatial dimension too. Ifwe expand the spatial scale of observation we imme-

diately observe that erosion in one place results inthe accumulation of sediment somewhere else. Themassive Pleistocene fans of sediment that in them-selves form low hills at the foot of the more promi-nent mountain ridges are often the focus of modernvillage settlements because of their attractive soilsand water supplies. In the complex topography ofthe Epirus landscape, erosion actually has a benefi-cial effect in bringing together soil that is thinly dis-tributed over hill and mountain slopes, andconcentrating it in intermontane basins and lowland

river valleys and coastal plains, where it providessome of the most important agricultural land forthe modern economy. Thus erosion, which seemedat a local scale to be largely negative, turns out ata larger spatial scale to be positively beneficial,observations which seriously challenge some cher-ished assumptions of modern conservation policiesand reinforce the need for a long-term perspective(Van der Leeuw, 1998; Van der Leeuw and Red-man, 2002).

The picture becomes even more interesting if wego into the more spectacular badlands landscapesand ask the local people who live there what theythink about their eroded hillsides. When we firstdid this their initial reaction was to look at us asif we had landed from another planet or at the veryleast were lacking more than a few brain cells, andask us with that inimitable combination of Greeklanguage and gesture: What erosion? After muchfurther discussion it emerged that the eroded slopesthat have such a dramatic impact on visiting archae-ologists and geomorphologists were largely unrec-ognized by the local people because they were of

no significance in their day-to-day life. In a pastoral


7/28/2019 Bailey Jaa 2007

15/26

economy largely devoted to sheep and goat hus-bandry, it mattered little if one part of the landscapedegraded, the animals simply moved on (Green,1997, 2005; Bailey et al., 1998).

If houses or even whole villages occasionally got

damaged or buried by an earthquake-induced land-slide, that did not matter too much either becausethe agricultural land, which in any case contributedrelatively little to the local economy, was mostlyowned by absentee landlords, by the ruling Otto-mans before Greek independence or by the Church.Many of these traditional mountain villages hadalso been moved on more than one occasion, eitherbecause of forcible relocation under the Ottomans,or to more secure locations to avoid raids by taxcollectors, bandits or the disruptions occasionedby Civil War and the invasion of foreign armies that

followed the end of Ottoman rule. Even today,many villages comprise two settlements, one onthe hill and one in the valley below, which are usedat different times of year, or by different members ofthe community in response to the changing seasonalneeds of livestock and the ebb and flow of social life.Moreover most of the adult men were absent foryears at a time, carrying on commercial activitieselsewhere in the Balkans or Turkey, providingincome less accessible to Ottoman tax collectors.More recently they have taken well paid jobs in Ath-

ens, Munich or New York, returning money to theirageing relatives in the home village, where theybuild new homes for summer visits and ultimatelyfor their retirement. Both historically and in themodern era the whole way of life of these mountainvillages has been organized around different formsof individual and collective mobility, which can beseen as a highly successful response not only tothe physical instabilities of a tectonic landscapebut to the social and political instabilities for whichthe Balkans have become such a byword in histori-cal and recent times.

Reflecting on this sequence of encounters andinterpretations, it is clear that we have been movingthrough a series of palimpsests at a succession ofdifferent geographical and temporal scales, rangingfrom the very large-scale spatial palimpsests associ-ated with the geological history of the region to thepalimpsests of meaning associated with particularparts of the local landscape. The eroded hill slopesthat formed the starting point for exposition turnout to be both true palimpsests in a geological sense,parts of wider spatial palimpsests in which eroded

soil or sediment have been re-deposited elsewhere

on many different timescales, and palimpsests of dif-ferent meanings associated with different observers.For the people who made their living in such a land-scape, these eroded surfaces were of little signifi-canceindeed in a sense invisible, for the visiting

geologist a highly visible symptom of underlyinggeophysical dynamics, and for the visiting archaeol-ogist a magnet for the search and discovery of pre-historic artifacts exposed by the removal of thesedimentary overburden.

In viewing different scales and types of palimps-ests, it is as if we have been uncovering differentlayers of meaning, referring to different scales andtypes of activities and processes, and even to differ-ent patterns of cause and effect. Insofar as there is acommon narrative thread for the human history ofthe region it is one of what Green (1997) calls inter-

weaving, and what we might call co-evolution, inwhich changes in the physical landscape, theeconomic exploitation of it, and the perceptions ofit by the people who live there have created a closelyinterwoven set of mutually adjusted processesthat have resulted in structures of great durabilityand flexibility in an environment otherwise sub-

ject to considerable physical, social and politicalinstability.

Relationships between scales and the problem of

determinism

The general emphasis of the above discussion hasbeen on ecological relationships resulting from envi-ronmental changes, although a social and politicalelement is apparent at the more recent end of thetemporal spectrum. That may reflect no more thanthe fact that environmental processes tend to revealtheir full effects only over quite long time spans andare therefore particularly useful for illustrating theinfluence of the larger-scale in human affairs. Ormore simply it may reflect the fact that I have cho-sen a large-scale geological palimpsest as a startingpoint for discussion, which necessarily leads inter-pretation in an environmental and ecological direc-tion. Whether an ecological and environmentalapproach is a necessary consequence of looking atlarge-scale palimpsests is another matter, and whatlonger-term social or cognitive patterns and pro-cesses might look like through the analysis oflarge-scale palimpsests requires further exploration.

Some clues as to how to frame such an investiga-tion are suggested by Benjamin (1985), a practic-

ing social anthropologist and ethnographer. In


7/28/2019 Bailey Jaa 2007

16/26

reflecting on the relationship between human valuesand motives and their environmental consequences,he suggested that on long timescales we need to usedifferent units of cultural analysis, not identifiablesocio-political groupings or the decision-making

units normally studied on ethnographical time-scales, but shadowy organizational themes or clus-ters of ideas (Benjamin, 1985, p. 223). In his view,there is no particular reason necessarily to expecta very close fit between human actions and environ-mental constraints over short time spans.

. . ..peoples actions are not reactions to someexternal force but the active constructs of theirown individual and collective choices. Humanchoices could until recently be variously good, bador indifferent from an adaptive point of view with-out coming under any marked environmental con-

straint during the life times of the people whomade those choices or during the life times of theirknown ancestors and descendants. These consider-ations change, however, if we shift our attentionto the much longer-timescales of prehistoric andethnohistorical anthropology. While system andfunction are to my mind quite misleading whenapplied to ordinary ethnographic field-data, some-thing like these concepts seems. . .to become morerelevant as the timescale is increased. . .it is in thelong time-spans considered by prehistoric anthro-

pologists that the environmental consequences of,and constraints upon, different ways of life showthemselves (Benjamin, 1985, p. 223).

He went on to emphasize that historical factors,that is unpredictable extra-systemic accidentsmight become more apparent after long periodsand the evidence for active choices might actuallybecome clearer on longer-timescales, especially if itcould be shown that several more or less equally via-ble alternative ways of life had persisted for longperiods of time.

This would seem to reinforce the point that if wewant to look at social phenomena in the longer-term, as with any other feature of human existence,we will have to work out our own concepts, prob-lems and tools of investigation, rather than relyingon ideas from other disciplines that deal with quitedifferent phenomena, scales of enquiry and methodsof observation, and in general much shorter timespans.

One of the most vexed issues in relation to thissubstantive aspect of time perspectivism is how weare to conceive of the relationships between process-

es that operate on different timescales. Some of the

most successful examples of such analysis are withphenomena or processes that have closely overlap-ping temporal properties. Hull (2005), for example,has analyzed the interaction between processesoccurring within the lifespan of the individual and

demographic trends that extend over multiple timespans in the 6000-year sequence of eastern Califor-nia. Cobb (1991) has demonstrated an interactionbetween relatively short-term and cyclical patternsof trade extending over centuries in the later prehis-toric record of Midwestern North America andcumulative increases in agricultural production overseveral millennia. In discussion of Neolithic andBronze Age Europe, with a similar range of timespans and timescales, there is an active debate aboutthe definition of event and structure and the natureof their inter-relationship, (Harding, 2005). Smith

(1992b) working with a shorter time span of 600years in the Postclassic of Mexico identifies a con-trast between short-term cycles of urban and impe-rial growth and decay and progressive demographicand agricultural expansion, a contrast which reflectsboth differences in scale and resolution of observa-tion and differences in the inherent dynamic ofdemographic and socio-political processes, respec-tively. Similar examples have been elaborated bythose working within the Annales framework(Bintliff, 1991; Knapp, 1992).

In many other cases, the temporal gulf betweendifferent phenomena may be so wide that we maynot be able to imagine any relationship at all, letus say between tectonic plate motions at oneextreme, and the decisions of an individual at theother. Does our recognition that different processesmay be apparent at different timescales condemn usto accept that there are parallel universes, each vary-ing in its own terms, but between which there is noconnection? Do the larger scale phenomena merelyset very broad boundary conditions, within whichthe small-scale variability of the every day can varywidely within its own terms? In what ways and atwhat points do slow-moving, large-scale processesintrude into the world of small-scale events andinteractions, and what examples can we find inwhich the direction of causation goes the otherway? Non-linear dynamic theory offers some power-ful precedents for how small-scale events can havevery large-scale and enduring consequences, andcomputer simulations can be used to model theinteraction of different variables over longtime spans, with interesting and often counter-intu-

itive results (Flannery et al., 1989; Mithen, 1997;


7/28/2019 Bailey Jaa 2007

17/26

Van der Leeuw and McGlade, 1997; Winder, 1997).However, there is as yet little clear guidance as tohow far the outcomes of such simulations can beinfluenced by or evaluated against the empiricalrecord of large-scale palimpsests.

Perhaps the biggest obstacle to investigatingquestions such as these is the necessary implicationthat they imply the existence of some processes thatare beyond the conscious awareness or control ofthe individual actors caught up in them. This canquickly turn into the belief that even to accept thepossible existence of large-scale processes, whetherenvironmental or social, is to subscribe to a sortof determinism that is an affront and even a threatto our sense of identity and individual free will. Ifthe actions and beliefs of individuals are lost inthe palimpsests of the longer-term, and all that

seems to matter in the long run is the operation oflarge-scale processes with a momentum of theirown, this seems to condemn us to the fate of grainsof sand swept along by a tidal wave, perhaps withthe illusion of independent will, but ultimately atthe mercy of much vaster forces over which we haveno control. However, as the student of non-lineardynamics or of geological processes would quicklypoint out, even grains of sand can in their cumula-tive effect create new areas of land that ultimatelytame and irrevocably change the tidal flow.

If behaviour in any given context represents theintersection and interweaving of many differentsorts of processes with different sorts of temporalrhythmsoperating over different time spans andwith different frequencies and amplitudes of varia-tionperhaps the best place to bring them all with-in the scope of one enquiry and observe theircombined effects in interaction is at the point wherethey intersect in our own lives in the world which wenow inhabit. This is where we can simultaneouslyobserve and integrate long-term processes withsmall-scale perceptions and individual actions. Forexample, the idea that tectonic plate motions mighthave affected individual decisions and choices in thePalaeolithic past, or been affected by them, doesindeed seem unlikely, but large-scale tectonic pro-cesses certainly affect individual decisions today,most notably those of the geophysicists who studythem. And while we cannot yet control platemotions, we are well on the way to predicting someof their more dramatic consequences and identify-ing where and approximately when the next majorearthquakes are likely to occur (Nalbant et al.,

1988). We thus live in a world where such seemingly

slow, large-scale and almost invisible processes havebeen made known to us through scientific investiga-tion and brought within our conscious understand-ing and control of the world that we presentlyinhabit. Such considerations suggest both a solution

to the problem of how to analyze interactionsbetween widely differing scales of activity, that wecan best do it in the context of our present-dayworld, which brings within reach the widest possiblerange of different scales of activity, and a solution tothe problem of determinism, that we cannot bedetermined by factors of which we have some con-scious knowledge, and over which in principle wecan exercise some control.

The past, the present and the future

If palimpsests representing variable and differen-tial temporalities are a universal feature of the worldwe live in, where exactly do we locate the present?And where do we draw the boundary between thepresent and the past or for that matter betweenpresent and future?

It is part of our conventional understanding ofthe world, and of our western intellectual inheri-tance, that the present is what we observe aroundus and can know about by direct experience. It iswhat we think we know best, and we use that

knowledge to interpret the past, which we believewe can know only indirectly and imperfectlythrough the experiences and actions of others, orto extrapolate forward into the future, which webelieve is unknowable except by inherently unreli-able methods of prediction.

In archaeology this conventional view has a fur-ther consequence. If we believe that we can knowthe present better than the past, it follows that weshould defer to the authority of the present and tothose who study present phenomena. Thus it is thatlarge areas of theory in archaeology are introducedby reference to authorities on modern and short-term phenomena, social anthropologists, sociolo-gists, ecologists, biologists, occasionally historians,and philosophers. The results of archaeologicalinvestigation are thus often variously evaluated,sanctioned or attacked, not according to how farthey correspond to observations of the empiricalevidence, but according to judgments about whichauthority is to be preferred. There is a double logicto this. Because we believe that the present is knownor knowable better than the past, we must seek our

inspiration in studies of present phenomena and our


7/28/2019 Bailey Jaa 2007

18/26

concepts and theories from authorities on the pres-ent. Because the past is knowable only imperfectlyand less well than the present, the evidence fromthe past cannot be relied on to provide an empiricalchallenge to pre-existing preference or authority.

Such logic, of course, only tends to reinforce theopinion of social anthropologists, sociologists, his-torians and perhaps also philosophers, that archae-ology is a derivative discipline that attempts tostudy with inherently imperfect data the past tenseof phenomena that are better studied in the present.Archaeologists who go down this route of enquiryalso inevitably end up chastising themselves andtheir colleagues for always being one step behindthe chosen authority discipline.

This view of the world, that we know the presentbetter than the past, is also powerfully reinforced by

our common-sense understanding, and it is one thatwe are very reluctant to abandon because it wouldcall into question our sense of identity, our beliefin our autonomy, and our sense of authority overthe world in which we live. It is also sanctioned inthe world of science by the principle of uniformitar-ianism, in principle a belief that the past was like thepresent, but in practice a far more complicated com-bination of concepts which actually provide us withthe tools to investigate empirically a past that couldhave been very different from the present (Bailey,

1983, pp. 174180).But if the world we can know about is a material

world whose main characteristic is duration, thecommon-sense understanding of the present mustbe misleading, or at any rate arbitrary, since materi-ality reaches forward from the past into the presentand extends into the future.

The notion that we know the present better thanthe past or the future can be examined from anotherangle. Consider how much knowledge I can reallyacquire by direct observation of the world aroundme. At this moment it is confined to what is happen-ing in roughly a 50 m radius around my presentlocation, an impossibly limited knowledge of theworld that cannot represent more than an infinites-imally small fraction of the totality of things thatare happening elsewhere at this particular moment.Of course I can switch on the television or the com-puter and learn a little of what is happening in otherparts of the world at this same moment from thesatellite news channels, but I am doubtful that thiswould give me anything other than a rather limitedview of what is happening in just a few of the

worlds trouble spots. If I want to know more I

can read tomorrows newspapers or the weeklymagazines, but these will be discussing events thathappened yesterday or last week or last month,and it is doubtful that one should believe everythingone reads in the newspapers. For greater depth I

might read books on the history of particularregions, and perhaps even their archaeology. Thereis a real sense in which we can only acquire a fullerknowledge of what is going on in the present bylooking back at it retrospectively from some futurevantage point, from which the events of moreenduring significance can be disentangled from thebackground noise and the pattern of relationshipsmore clearly understood in terms of contemporane-ous events elsewhere and their various antecedentsand consequences (Evans-Pritchard, 1961).

The durational present

For the archaeologist, where exactly is theboundary between the past and the present? Forsocial anthropologists the ethnographic present isthe past 100 years or so, the period within whichobserver participation, the main technique ofanthropological enquiry, has been practiced, essen-tially the lifetime of practicing anthropologists andtheir predecessors, who have left records of theirobservations. Before that period, evidence amenable

to anthropological investigation recedes from viewand becomes a blur, the shadowy organizationalthemes quoted earlier from Benjamin (1985), whoalso notes that the cultures and societies studiedby anthropologists have rarely been shown to havea time depth of more than about 6 human genera-tions. For many historians and sociologists, thispresent era has a time depth of about 300 years,separated by the Industrial Revolution from whatcame before, and characterized by detailed docu-ments and bureaucratic records that allow explora-tion of social and political movements. Palaeolithicarchaeologists in more enthusiastic moments havebeen known to refer to anything after 10,000 yearsago as modern (and therefore beyond the rangeof interest), the upper limit being marked by theNeolithic Revolution, its lower limit marked bythe Human Revolution 2 million or more yearsago. A highly topical field of research during thepast decade or so has been the appearance of ana-tomically modern humans, a modern phenome-non that has a time depth of as much as 100,000years, and whose appearance is closely associated

with a so-called Upper Palaeolithic Revolution.


7/28/2019 Bailey Jaa 2007

19/26

Structural geologists are inclined to refer to any-thing within the past 13 million years or so asmodern, the superficial overburden of soft sedi-ments that hides the hard rocks that are the objectsof real interest.

In all these examples, what counts as past orpre as in pre-Industrial or pre-history or pre-Neolithic or pre-human refers to something con-sidered too ea

Bailey Jaa 2007

Documents

Transcript of Bailey Jaa 2007