Archaeological Survey

E. B. Banning
University oj Toronto
Toronto, Ontario, Canada

SPRINGER SCIENCE+BUSINESS MEDIA, LLC

Chapter I
Introduction

Archaeological survey is often the first stage of a long-term archaeological project.

At other times it is the principal method for studying some aspect of the past. Survey
allows archaeologists to discover sites they may wish to excavate, to assess potential
damage to archaeological resources from construction, road-building or other develop-
ment, and to assess aspects of past settlement systems and regional economies. Survey
can range from very informal exploration to detailed and explicit prospection or sam-
pling strategies designed to maximize the probability of detecting sites or artifacts over
a region, or to provide representative samples of cultural materials. It also ranges from
visual inspection of fairly obvious features and artifacts on the modern surface, some-
times called "fieldwalking," through dispersed excavations ("shovel testing"), to geo-
physical remote sensing of buried materials.
Survey is not simply a poor substitute for archaeological excavation, or meant only
to discover sites for us to excavate. In fact, it is uniquely able to address some research
questions that excavation alone will never answer. Only regional survey is capable of
producing the data we need to investigate prehistoric use oflandscapes, settlement hier-
archies, and human behaviors that were dispersed in space instead of concentrated within
the more obvious kinds of "sites."
This book will introduce concepts and methods relevant to investigating archaeo-
logical phenomena at the regional scale. It will deal with some of the common goals of
regional archaeological survey, the characteristics of archaeological remains that sur-
veys are meant to discover or document, and how these and the way we design a survey
affect the survey's results. It will not specifically address survey at the smaller, site-
specific, scale, although some of the same concepts are applicable to both regional and
intra-site survey.
We will begin with a brief history of archaeological survey, including some case
studies that will help to emphasize points in later chapters. A discussion of the role of
regional survey in archaeology will follow. The next section will deal briefly with mod-
els or assumptions of the ways cultural material can be distributed in space, and on
exposed surfaces or in buried deposits. Then a brief section on general aspects of re-
search design in archaeological surveys will set the stage for later chapters.
2 Archaeological Survey

1. A BRIEF HISTORY OF ARCHAEOLOGICAL

SURVEY

1.1 Early Archaeological Reconnaissance

European curiosity about the visible remnants of past civilizations led to documen-
tation of archaeological landscapes from at least the 16th century. In the United King-
dom, the Ordnance Survey began to include antiquities on its detailed maps as early as
1801, and the Gentlemen:~ Magazine published articles on discoveries of English ar-
chaeological sites in the 18 th century.
Leaders of some early archaeological surveys did not explicitly conceive of them as
such. Often archaeological observations were only adjuncts to geographical explora-
tion, as in early Russian expeditions to Siberia, or even military expeditions, such as
Napoleon Bonaparte's in Egypt. Even so, some of these explorers laid the ground-work
for methods that archaeologists have used ever since. At first, these were usually explo-
rations by Europeans in non-European parts of the world; indigenous people, after all,
already had long familiarity with their ruined monuments and had oral traditions about
their significance. In some cases, as in 18 th -century expeditions to relocate and study
ruined cities in Central America, the European antiquarians relied on guides and porters
to take them to sites that were already well known locally.
As the United States expanded its frontier westward, its settlers found considerable
prehistoric monuments. One ofthe major questions was the origin of the "mound build-
ers." Expeditions by Squier and Davis (1848) and later the Bureau of American Ethnol-
ogy (Thomas, 1894) devoted considerable time to exploration and survey of these mounds.
Exploration of the American Southwest also led to study of both contemporary Hopi
villages and the ruined pueblos that preceded them (e.g., Cushing, 1890).
Other Europeans and Americans turned to exploration of the antiquities of the Holy
Land. The earliest of these were mainly itineraries by pilgrims to the Middle East's holy
places, such as accounts by Theodericus (ca. lIn), Frescobaldi (ca. 1390), Gucci (ca.
1390), and Bertrandon de la Broquiere (1457). There were also similar early accounts
by Arab writers, such as Ibn Jubayr (1185). Later and more influential 19th -century sur-
veys of Ottoman Palestine, Syria, Transjordan, and Arabia included ones by Bucking-
ham (1821; 1825), Burckhardt (1822; 1829), Conder (1889), and Robinson (1856). These
documented mainly quite obtrusive sites, such as Roman ruins, and tried to identify
them with historical or biblical place-names, usually on the basis oftheir modem Arabic
names. At the time, there was no way for them to date sites except in the rare instances
when inscriptions were visible, and the sites they discovered were ones already quite
well known locally.
Introduction 3

1.2 Fieldwalking in Britain

One of the first authors to describe the methods of archaeological survey, and possi-
bly the first specifically to address methods for surveying artifact scatters, was W. G.
Clarke (1922:24-32), in a guide to amateurs interested in prehistoric lithics. Although
he does not discuss any explicit research design, he does give good advice on fieldwalking
for surface scatters, and notes the importance offtre-cracked rock as evidence of pre his-
toric cooking activity.
An examination of an arable field in a suitable district, a sufficient time after it has
been ploughed for the soil to have settled and a considerable proportion of the stones
to be lying on the surface, will disclose a perplexing quantity [oflithics] (Clarke,
1922: 25).
In describing one day's survey in southwestern Norfolk, he notes that, while walk-
ing over a heath, surveyors could only fmd artifacts where the burrowing of moles and
rabbits had thrown them up (Clarke 1922: 30). In a newly plowed field, by contrast,
their method was as follows.
Following our usual practice we walked across the middle in each direction, trusting
to get some indication of the best portion oftbe field. Scattered flakes occurred in all
parts, but were much thicker at one place than others, and to this spot we returned
and endeavoured to delimit the area which would best repay detailed searching.
This was soon revealed by a careful inspection and a systematic search was then
made, almost every yard being scrutinised (Clarke, 1922: 31).
In this he employs an assumption that archaeologists continued to use for decades:
that high-density clusters of artifacts on the surface are somehow more important or
more likely to correspond with the places where prehistoric activities were concen-
trated. He notes that in some areas lithic densities are so high that selective collection is
"inevitable" and provides advice on how to recognize culturally produced flakes and to
detect retouch quickly in the field. He also provides recommendations on recording
provenience:
On arrival home, surface implements such as these are washed and scrubbed, and
the locality in which each flint was found either indicated by a number, correspond-
ing to that in a register, a gummed label with parish and collectors's initials printed
thereon, or a written locality. A method by which the particular field in which the
implernent was found can be ascertained is to be recommended, as variations in the
industries of sites only a few hundred yards from each other, are thus made obvious
(Clarke, 1922: 32).
4 Archaeological Survey

1.3 Early Air Reconnaissance

One of the great boons to archaeological survey was the advent of aerial reconnais-
sance during the First World War. Archaeologists were quick to apply this new technol-
ogy to archaeological discovery in regions ranging from England (Crawford, 1929; 1953)
to Syria (Poidebard, 1934). Not only was it easier to discover earthworks and detect
patterns in their distribution by viewing landscapes from the air in raking light, some-
times buried ditches and building foundations were detectable in vegetation patterns
called "crop marks" (Bewley and Rackowski, 200 I; Dassie, 1978; Deuel, 1969; Kennedy,
1995a, 1995b; Kennedy and Riley, 1990).

1.4 Surveys in Northwest Europe

Northwest Europe has long had a tradition oflandscape archaeology, focussed largely
on the distribution of small farms, villages, burial monuments, pathways, field walls
and ditches. Since the 1920s, "fieldwalking" has benefited from aerial reconnaissance
and amateur documentation.
One of the interesting features of the European tradition of survey is its treatment of
whole landscapes, and especially evidence for agricultural land use, and not only of
settlement sites. Even before, but especially after aerial reconnaissance became avail-
able, European archaeologists noticed traces of old fields at different orientations than
the modern ones. Some of these old fields also had distinctive shapes, such as the "Celtic
fields," and sizes, corresponding, for example, to the traditional field unit of an "acre."
Some were bounded by parallel heaps of stone, called "reaves," or by "Iynchets," sedi-
ment that naturally accumulated on the downslope edges of fields plowed for many
years.
Some early research in British field archaeology dealt with the formation oflynchets
and "ways" or tracks (Clay, 1927), others with "Celtic fields" and other kinds of field
systems (e.g., Curwen, 1927), and still others with evidence from crop marks over bur-
ied roads and buildings (Crawford, 1929).
In Scandinavia, archaeology had close links with natural history even in the late 19111
century, and this sometimes led to a more ecological perspective on site distributions.
By the first quarter of the 20th century, regional settlement archaeology had become
more common (e.g., Almgren, 1914; la Cour, 1927), and later regional settlement sur-
veys by Therkel Mathiassen (1948; 1959) used large numbers of amateurs to help search
out sites in Denmark.

1.5 The Vir.. Valley Survey

Gordon Willey's classic study of the ancient Vini Valley in Peru took archaeogical
survey well beyond the simple prospecting for interesting sites. With some resemblance
to the European fieldwalking surveys, it focussed on "settlement pattern," which Willey
defined as
Introduction 5

... the way in which man disposed himself over the landscape on which he lived. It
refers to dwellings, to their arrangement, and to the nature and disposition of other
buildings pertaining to community life. These settlements reflect the natural
environment, the level of technology on which the builders operated, and the various
institutions of social interaction and control which the culture maintained (Willey,
1953:1).
This definition's emphasis on the distribution of sites and buildings on the land-
scape, without explicit reference to how it "reflects" human behaviour, does not do
justice to the actual research Willey and his colleagues conducted. Willey concerned
himself, not merely with the distribution of sites in space, but with site functions, popu-
lation sizes and sociopolitical organizations. This survey and other studies of settlement
patterns in the 1950s had already begun to address the identification of "community
patterns" with temporal, functional, ecological and social components in addition to the
spatial one (Willey, 1953).
This survey had a profound influence on Americanist archaeology, leading to the
kinds of spatial samples and environmental orientations that were particularly common
in surveys of the 1970s, and eventually even to the kinds oflandscape archaeology that
became popular in the late 1980s.

1.6 Diyala and Uruk Surveys, Iraq

Surveys in the Diyala and central Euphrates floodplains of Iraq inspired a whole
generation of archaeologists in the Near East (Adams, 1965; 1981; Adams and Nissen,
1972). From 1956 until 1975, these surveys' designs were based on
the premise that in a semiarid country like ancient Mesopotamia settlement would
have been possible only where water was available - along rivers and canals. Where
the settlements of a period showed linear patterns, it could be assumed that the lines
reflected the water-courses upon which the settlements depended (Jacobsen,
1981 :xiii).
In addition, traces of canal levees visible in aerial photographs provided a frame-
work for fmding less obtrusive sites, which could be assumed to occur along these canal
routes. In the aerial photographs, often "a pattern of linear discolorations emerged, gen-
erally consisting of the faint traces of ancient levees" (Adams, 1981 :28). Since a major
goal of these surveys was to document changes in agriCUltural land use in the region,
linking the survey so explicitly to the most limiting agriCUltural resource - water -
was a highly effective strategy. We will return to surveys of this type in discussion of
purposive survey, sampling frames, and the detection of spatial structure in chapters 4,
6 and 7.
6 Archaeological Survey

1.7 The Basin of Mexico Project

Surveys by the Basin of Mexico Project were also quite influential. The project had
the ambitious goal of 100% coverage by pedestrian survey over a region some 600 km 2
in area to document changes in residential distributions and people's interactions with
their social and natural environments since 1000 BC. Methods changed over the period
from Sanders's and Parson's surveys of the Teotihuacan Valley of 1960-1966, through
Blanton's work on the Ixtapalapa region, Sanders's and Santley's survey of the Cuautitlan
region, and surveys by Parsons and Hrones and Parsons in the Texcoco, Chalco,
XochimiIco, and Temascalapa regions up to 1975, but shared some principles to protect
the data's consistency (Sanders et aI., 1979: 11-15, 20).
This was a site survey, with inferred settlement locations as its basic units of obser-
vation and analysis. The surveyors argued that settlements were places "where people
had spent enough time to leave some obvious, enduring physical traces on the present
ground surface" (Sanders et aI., 1979: 15). Their definition of "site" required that survey
teams recognize fairly discrete clusters of surface artifacts to which they could assign
boundaries, apparently where density dropped off to "empty space." They recognized
the possibility that some of these clusters were palimpsests of several overlapping occu-
pations, and strove to separate these. They thought they could estimate what type of
settlement some sites represented, and occasionally even used artifact densities to esti-
mate sites' numbers of residents. Importantly, they recognized the importance of ensur-
ing "that a blank on the settlement map for any particular time period was the product of
a lack of settlement rather than a lack of survey" (Sanders et aI., 1979: 16-17).
Their survey initially used modem fields, identified on air photos, as observation
units, rather than arbitrary geometric units. This presented problems in uncultivated
areas or where field boundaries had shifted since the air photos were taken. Later sur-
veys involved fast reconnaissance of broad transects, survey on and around known sites,
and systematic pedestrian transects with intervals typically around 50 m but varying
between 15 m and about 75 m. This variation depended on "common sense considera-
tions of topography, drainage, vegetation, modem occupation," and other factors (Sand-
ers et al. 1979:24). Except for small sites whose boundaries were fairly obvious, defini-
tion of site boundaries awaited analysis of the transect information, which showed where
survey teams had recorded "on-site" artifact densities.

1.8 Site Survey in the American Southwest

The American Southwest has a long tradition of archaeological exploration, and

work there has had a strong impact on theory and method in archaeological survey.
The earliest archaeological exploration, from about 1870, concentrated on docu-
menting very obtrusive pueblos and cliff-dwellings. Surveys by John and Richard
Wetherill (Pepper, 1902), Guernsey and Kidder (1921) expanded the coverage to cave
sites and earlier "basketmaker" remains.
Introduction 7

But the Southwest was also a testing ground for the "New Archaeology" in the
1960s, and this had an impact on archaeological surveys there. Fred Plog (1974) used
survey by the Southwest Archaeological Expeditions of the Field Museum of Natural
History of the Upper Little Colorado River region in 1967 and 1968, and especially the
Hay Hollow Valley, to test a number of hypotheses concerning demographic change,
site differentiation, social integration, and technological developments. More generally,
he used these in an attempt to demonstrate the power of positivist explanatory frame-
works that were beginning to become common in North American archaeology. Along
the way, Plog explicitly dealt with such problems as classifying sites on the basis of
activity sets, indirect measures of human population size (based on room numbers) and
ways to correct these for site longevity and rates of room abandonment.
Another influential survey in the Southwest was the Black Mesa Project (Gumerman
and Euler, 1976; Plog, 1978; Plog and Powell, eds., 1984). The project began with a
pipeline survey on the lease of a coal company in northeastern Arizona, but expanded in
scope to investigate many research hypotheses and to explore both theoretical and prac-
tical issues surrounding archaeological survey.
These and other surveys in the Southwest led to development of many of the theo-
retical concepts that will be discussed in later chapters.

1.9 Survey in North American Forests

In parts of North America where forest cover and leaf litter are major impediments
to the surface visibility of sites, archaeologists have devoted considerable effort to de-
veloping methods of subsurface survey. As we will see in later chapters, this work also
led to important theoretical advances in the detectability of archaeological materials
(e.g.,Alexander, 1983; Kintigh, 1988; Krakkeretal., 1983; Lightfoot, 1986; 1989; Lovis,
1976; Lynch, 1980; McManamon, 1980; Nance, 1983; Nance and Ball, 1986; Shott,
1985).

1.10 "Non-site Survey," and "Landscape Archaeology"

Prior to about 1980, most surveys in North America and the Old World concentrated
on the documentation of "sites." There was not much consistency in the definition of
sites, but generally these were concentrations of material culture, sometimes involving
traces of architecture, that archaeologists tended to assume marked the locations of an-
cient settlements, workshops, and so on.
Over the last 25 years, however, archaeologists have shown growing interest in the
residues of dispersed activities that did not necessarily occur in settlements or other
kinds of "sites." This trend began with surveys that used very minimal definitions of
"site;" sometimes discovery of a single artifact was sufficient. Increasing recognition
that hunter-gatherers conceived oflandscapes as consisting, not of sites and empty space,
but of fishing holes, hunting grounds, pathways and portages, accelerated the pace of
this development.
8 Archaeological Survey

In the late 1960s, David Hurst Thomas began a project that was to be very influen-
tial in the rise of "non-site" surveys. He worked in the Reese River Valley of central
Nevada in an attempt to test Julian Steward's ethnographic model of Great Basin hunter-
gatherer adaptation on archaeological evidence (Thomas, 1972; 1975). Rather than sur-
vey for sites in the environmental zones associated with different economic activities in
Steward's model, Thomas measured attributes of artifacts found in 140 spatial units,
each 500 m x 500 m in size, distributed across those zones. Thomas assumes that indi-
vidual artifacts or cultural items, rather than sites, are his minimal units, but actually
uses spatial units.
In the Amboseli Basin of Kenya, Robert Foley's research also led him toward what
he calls an "offsite" approach (Foley, 1977; 1981a; 1981b). Foley noted that, like plant
ecologists, archaeologists were interested in the distributions of small items, usually
artifacts rather than plants, and that, in the Amboseli Basin, these artifacts were distrib-
uted in a virtually continuous manner. Unlike most of the archaeologists who had con-
ducted surveys up to that time, he explicitly considered the site-formation processes that
would have affected those distributions, and began with models that would guide his
field research and data interpretation. With these models he conceived of sets of over-
lapping distributions, created by a variety of processes (see below, pp. 18-19). Conse-
quently "sites," as defined by elevated densities, are only anomalies in the continuous
spatial distribution of artifacts, and not necessarily the locations where habitation or
other activity was concentrated at any particular time in the past. Although this conclu-
sion has often led to the assumption that artifacts, rather than sites, are our basic units of
analysis, Foley's approach, like Thomas's, actually leads us to a method based on the
analysis of spatial units, such as quadrats (see chap. 4).
Some people object to some kinds of non-site survey because of the relatively high
cost per unit area. However, Ebert (1992:170-171) argues that they may nonetheless
offer lower costs per unit of information, as evidenced by comparison of his Seedskadee
Survey results with the data found in the site-based records ofthe Intermountain Antiq-
uities Computer System. He suggests that the Seedskadee data is equivalent to the infor-
mation from a site-based survey with 528 sitesJkm2•

1.11 Intertidal and Shallow-lake Survey

One type of survey that has only recently begun to receive the attention it deserves is
the survey of intertidal zones and lake-bottoms revealed by falling water levels. Changes
in sea levels, erosion of coasts, the excellent preservation of organic remains, and the
human tendency to make use of coastal areas all combine to make it extremely impor-
tant to survey the surfaces that receding waters briefly expose.
The archaeology of foreshores or intertidal zones, which lie between the high- and
low-tide levels of sea coasts, has a much longer history than one might expect. William
Borlase (1753; 1756) was a natural philosopher who, on a visit to the Isles ofScilly west
of Cornwall in 1752, saw boulder walls forming enclosures in the islands' broad inter-
tidal zone and recognized them as the remains of field walls and houses. He concluded
that the islands had sunk over time, finding the alternative hypothesis - that sea levels
Introduction 9

had risen - too implausible. Naturalists and amateur archaeologists of the 18th and
19th centuries took considerable interest in traces of submerged forests, and sometimes
archaeological sites, that coastal construction or storms periodically exposed along Eng-
lish coasts (e.g., Boyd Dawkins, 1872; Codrington, 1869; Cox, 1894). In the early 20th
century, archaeologists' attention turned to intertidal remains related to the submerged
"Lyonesse surface" along the Essex coast (Crawford, 1927; Smith, 1955; Warren et aI.,
1936).
It was only in the late 1970s, however, that intertidal archaeology began to depend
regularly on survey, rather than chance exposures. Philip Hobler (1978) demonstrated
the importance of the intertidal zone to the early prehistory of British Columbia, when
sea levels were as much as 5 m lower than they are today, in a survey of Moresby Island.
Wilkinson and Murphy (1986) discussed the problems and potential of intertidal survey
on the basis of work along the Essex coast ofEnaJ.and that represented the resurgence of
this approach in the United Kingdom. Shortly thereafter, the Severn Estuary Levels
Research Committee began extensive research in the intertidal zones of the Severn River
(Allen and Fulford, 1986; Bell, 1993; Bell et ai., 2000). More recently, English Heritage
and Historic Scotland have devoted considerable attention to intertidal archaeology (e.g.,
Ashmore, 1994; Fulford et aI., 1997; Historic Scotland, 1996). Intertidal survey is now
a growing component of archaeology and heritage assessment in the United Kingdom
(Aberg and Lewis, 2000; Wilkinson and Murphy, 1995), Ireland (O'Sullivan, 2001),
northwestern North America (Fedje and Christensen, 1999; Langdon, 1987); and the
South Pacific (e.g., Felgate, 2002; Reeve, 1989; Sheppard and Aswani, 1997; Wickler,
2001).
Intertidal surveys pose unusual practical problems, notably in the timing of field-
work (see pp. 47, 65, 165), but also have great rewards, particularly in the excellent
preservation of wooden and other organic remains that often pertains. They also yield
unparalled evidence for watercraft, bridges, piers, fish weirs, salt pans, duck-decoy ponds,
and other material culture associated with coastal and wetland economies.
Surveys of surfaces that emerge when lake levels are low are very similar in some
respects to intertidal surveys. The most famous early examples involved discovery of
the "Lake Villages" of Switzerland and parts of France, Italy, and Germany during the
dry spell of 1853-54 (e.g., Keller, 1866), but some reports of wooden structures in shal-
low lake waters are even earlier in Ireland (e.g., Lewis, 1837), and further work on
crannogs (lake dwellings) in Ireland, Scotland and the Continent closely followed (e.g.,
Munro, 1882; 1890; Wood-Martin, 1886). As with intertidal survey, there has been a
recent resurgence of work on shallow and receding lakes, including survey along the
shores of Lake Tiberias in Israel (Nadel, 1993), and survey of crannogs and other sites
in Irish lakes (Buckley and Sweetman, 1991; O'Sullivan, 1998).
Chapter VI
Purposive Survey: Prospection

We may be looking/or an object, and a/though we do not know where it is, we can
reason out the chances o/its being in various possible places. With a limited time to
find it, where shall we look first and how long shall we look in each place? (Koopman,
1980:139).

The goal of some surveys is to discover particular kinds of archaeological materials

or sites, including rare sites and evidence that can help to evaluate a specific hypothesis,
rather than to estimate parameters of a more general archaeological population. Some
archaeologists use the term "purposive survey" simply to describe haphazard surveys
that lack careful design, but here I use the term in a more restricted way for surveys
designed to optimize the probability of discovering particular kinds of archaeological
materials with a given amount of search effort, or "prospecting."
Archaeologists have given far less explicit attention to the theory behind purposive
survey than to surveys of the sampling type. Probably this is due to the widespread
perception that surveys that do not employ statistical sampling are somehow "unscien-
tific" (but see Asch, 1975). This has led, in turn, to the use of sampling as a prospecting
tool when, in fact, it is usually quite poorly suited to the discovery of at least some kinds
of archaeological "targets," and specifically sites. In fact, purposive search for particu-
lar kinds of archaeological evidence is a perfectly legitimate research goal that is im-
plicit in much of the work that archaeologists do (Banning, 2002 ).
Prospectors use their experience, theoretical models, or hypotheses based on em-
pirical data to predict where targets, such as settlement sites or shipwrecks, might be
located or to restrict the amount of territory a survey must cover to provide satisfactory
results. Sometimes this involves intentional bias in favor of locations most likely to
exhibit the material culture of interest, or omitting areas where deposits of the appropri-
ate age do not occur. Where prospecting must be combined with estimation, surveyors
may favor disproportionate stratified sampling or adaptive sampling (chapter 6). Often,
they use a suite of remote sensing methods, especially aerial ones, as well. Modem
surveyors are also able to use Geographic Information Systems (GIS) and information
from previous surveys to model the distribution of material culture in space and thus

133
134 Archaeological Survey

predict where certain kinds of materials are likely to be found. Not only can modelling
improve the ability of surveys to find archaeological materials, purposive survey is also
a way for us to test those models.
In prospecting for specific targets, such as particular shipwrecks or settlements
known only from historical documents, some of the methods of Operations
Research, including Bayesian analysis and the search patterns employed in
search-and-rescue, can be helpful.

1. PROSPECTING

A few examples will suffice to illustrate some typical prospecting problems. Gener-
ally, the prospecting problem involves either targets that were previously known but
whose location is uncertain, or ones that are suspected, but not known to exist. In addi-
tion, prospecting can be for a single target, or for multiple targets whose locations are
assumed to be either independent or interrelated. The wreck of a single ship known to
have sunk in a storm is a classic example of a single target known to exist, but whose
location is uncertain. A fleet of such ships, however, would involve multiple, interre-
lated targets. In the case ofthe Chesapeake flotilla, for example, even the pattern (bow-
to-stem) in which they were scuttled in the Patuxent River 1814 is known (Shomette
and Eshelman, 1981). Unspecified archaeological sites, features, or artifacts accumu-
lated over a long period in a small region might be effectively independent occurrences,
meaning that we have no prior knowledge to suggest that the location of one target had
an influence on the location of others. In many archaeological instances, however, there
is reason to suspect that targets of approximately the same age were not independent.
For example, artifact distributions are often clustered or sites show spatial autocorrelation,
while settlements sometimes show fairly regular spacing. In addition, the locations of
sites from one period of time sometimes influence the locations of succeeding settle-
ments. In places like the Near East, this temporal autocorrelation has led to the creation
of deeply stratified tells. Keeping these variations on the search problem in mind is
important because the number of expected targets and whether or not their locations are
independent can affect the strategy for finding them.

1.1 Surveying for Historically Documented Sites

Sometimes even a small amount of historical information can be a tremendous help

in directing the search for a particular site, the precise location of which has been forgot-
ten. For many years archaeologists have used historical information to narrow such a
search, as in Schliemann's search for Troy. Yet, although recognizing the need for "a
systematic approach to prospection, reducing the area of search and thereby incremen-
tally increasing the probability of success" (Garrison et al. 1985: 302), searchers have
rarely selected near-optimal strategies for this purpose, and sometimes even begin with
a random sample that does not take advantage of prior information.
Purposive Survey: Prospection 135

Today, historical sources are often useful in the search for medieval and historic
sites, including shipwrecks and specific buildings. Marine survey for wrecked ships
typically relies heavily on historical evidence for the likely location of the wrecks. In
some cases historic maps, historical documents, and ethnohistoric sources can provide
important clues to the location of specific terrestrial sites as well, including medieval
settlements in Europe and villages encountered by explorers and early settlers in North
and South America.
In general, surveys of this type involve beginning the search in the area where prior
information suggests the probability of finding the target (or targets) is highest, and then
widening or intensifying the search in the light of information gained as survey progresses.

1.1.1 Location of Upper Creek Villages in Alabama

Lolley (\996) used a range of accounts by early travellers in Alabama and historic
maps to predict the locations of many settlements that the Upper Creek Muskogees in
the Coosa-Tal1apoosa region of central Alabama reportedly occupied in the 17th, 18th
and early 19th centuries. By focussing on the most useful maps, and especial1y those
that show both villages and topographic features, and comparing the reported locations
of villages on a large number of maps of varying date, Lolley was able to plot circles
marking the predicted locations of those villages. Large circles indicate some uncer-
tainty of location, while small circles indicate more precise locations. Lolley tested the
predicted locations against site distributions known from previous archaeological in-
vestigations and found that many corresponded.
The obvious next step is to use the model to search predicted locations for sites that
have not previously been documented.

1.1.2 Search for La Navidad

Historical information was useful to narrow the search for Columbus's settlement at
La Navidad (Deagan, 1989). The sources limited the possible locations of this settle-
ment to a particular area, and mentioned a number of geographical features and the
settlement's relationship to a native village, which helped to narrow the possibilities
further. Although this work did not make explicit reference to prospection theory, it
fol1owed many of the principles that that theory involves, and provides a good example
of the search for a single terrestrial target.

1.1.3 Nautical Survey for The Monitor

The Monitor provides a good example of a single, specific target in survey at sea
(Newton, 1975; Sheridan, 1979). This turreted ironclad of the Union Navy was lost off
Cape Hatteras in December 1862, its exact location remaining unknown until a purposive
survey employing side-searching sonar rediscovered it in 1973.
136 Archaeological Survey

1.1.4 Search for the Submerged City of Helike

The discovery of Helike, a Greek city that disappeared into the Gulf of Corinth
during an earthquake in 373 BC (Pausanias 7.24.3-7), is another case that calls for the
principles of prospection. Schwartz and Tziavos (1979) report on their use of geological
evidence to try and limit the area within which they must search for this lost city. They
concluded that the site lay offshore somewhere between their boreholes 2 and 4, reduc-
ing the possible location of the site to a one-kilometer-wide band along the coast.
However, this is also a good example of how such reasoning often only identifies
high-probability regions for further prospection, and how new information may alter
those probabilities. Survey of the band that Schwartz and Tziavos identified was, in
fact, unsuccessful because some of the assumptions on which it was based proved faulty.
Eventual discovery of the settlement inland, on the alluvial fan of the Helike River,
was due to the realization that there had been uplifting of the fan that could have taken
place after the city's destruction and burial. New boreholes on the delta showed a buried
layer containing marine fauna as well as artifacts, and later test excavations finally re-
vealed traces of the ancient city (Soter, 1998; Soter and Katsonopoulou, 1999; Soter et
al.,2001).

1.2 Using Aerial Photography or Satellite Imagery

In some situations, certain kinds of archaeological phenomena, such as buried build-

ings, ditches, canals and roads, are much more obtrusive and visible from the air than
they are to an observer on the ground. When aerial photographs or satellite images show
traces that are suspected of being phenomena of interest, survey can then consist of
ground checks, walking over the ground or conducting small test excavations at loca-
tions tagged during inspection of the photographs.
For example, during the first World War, O. G. S. Crawford and others noticed that
aerial photographs from military reconnaissance missions in western Europe often showed
unusual linear patterns that seemed to result from buried archaeological features
(Crawford, 1929). In the areas where these features were observed, ditches and stone
walls were not buried very deeply, and sometimes lay in shallow soil over chalky bed-
rock. During some periods, vegetation growing over these sites varied in color and height
as a result of variations in soil depth and water availability. Grass growing over stone
walls, for example, was stunted or yellowed, while grass growing over buried ditches
was tall and green. When viewed from above, the color variations sometimes showed
linear patterns. In addition, viewing the site in early morning or evening when sunlight
struck the ground at a raking angle also caused taller vegetation to cast shadows.
North American archaeologists have less often relied on aerial reconaissance in their
archaeological surveys, but in the Old World this method has had a great impact on
archaeology. Over the past 80 years, aerial reconaissance has allowed archaeologists in
western Europe to identify thousands of archaeological settlements and buildings, as
well as whole field systems. Once a suspected site is detected in an aerial photograph,