Tracking the Samnites: Landscape and Communications Routes in the Sangro Valley, Italy

Author(s): Tyler Bell, Andrew Wilson and Andrew Wickham

Source: American Journal of Archaeology, Vol. 106, No. 2 (Apr., 2002), pp. 169-186
Published by: Archaeological Institute of America
Stable URL: http://www.jstor.org/stable/4126242 .
Accessed: 30/12/2014 01:01

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Archaeological Institute of America is collaborating with JSTOR to digitize, preserve and extend access to
American Journal of Archaeology.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
 Tracking the Samnites:

Landscape and Communications Routes

in the Sangro Valley, Italy
TYLER BELL, ANDREW WILSON, AND ANDREW WICKHAM

Abstract ward Bispham and Susan Kane, builds on the earli-

This paper is based on field surveywork carried out in er work through excavation on Monte Pallano and
the SangroValley,Abruzzo,Italy,between 1994 and 1998. at surrounding rural sites.2
While the preliminaryresults of the surveyare published The Sangro river valley (fig. 1) was chosen be-
elsewhere (Lloyd et al. 1997), this article presents the
results of a computer-based GISmethodology for recon- cause it forms a well-defined topographic unit, in a
structing routes of communication between ancient sites region where surprisingly little is known about the
in a landscape, designed to complement traditional sur- archaeology. The river itself runs from the Apen-
vey methodologies and explore the so-called off-site ar- nines into the Adriatic, a distance of some 120 km.
chaeology that exists between areas of intense antique It is located three river valleys north of the Biferno
occupation within the landscape. In doing so this re-
search not only demonstrates the theoretical value of Valley,where earlier field surveywork indicated the
such an approach,but also showshow its practicalapplica- existence of widespread rural settlement in antiq-
tion has helped the project understand in greater detail uity.' The area largely disappears from the literary
the ancient landscape of the Sangro river valley. The record after Sulla's devastation of Samnium in 80
principal results of this analysis applied to settlement B.C., and until recently little research or excava-
patterns in the middle Sangro Valley were to show that tion has been carried out in the Sangro Valley. De-
the fortified hilltop center of Monte Pallano was closely
integrated into the communications network of the sur- spite a recent increase in Samnite studies, we still
rounding settlements, that one of the largest Samnite know very little about Samnite settlement, rural or
andRomansettlementsdiscoveredbythe surveylayon a nucleated, and very little about the history of Sam-
junction of several well-frequented routes, and that a nium after its subjugation to Rome.4 The project
largeareaof off-sitematerialprobablyderivesfromIron therefore combined field survey, excavation of se-
Agecemeteriesoverlookinga heavilyusedroutearound
the southernflankof MontePallano.* lected sites, geomorphological study, and ethnoar-
chaeological enquiry to illuminate these issues.
The objective of the Sangro Valley Project has Data from all periods were collected, although the
been to investigate settlement processes and land- main research focus is from the Iron Age to the
scape history in an area of ancient Samnium now early Middle Ages. This paper presents some of
part of modern Abruzzo. The project, directed by the preliminary findings of the project, with partic-
John Lloyd and Gary Lock at Oxford, and Neil ular emphasis on the study of settlement and com-
Christie at Leicester, with the support of the two munications routes.
Italian superintendents responsible for the area, The Sangro river runs through a variety of differ-
Amalia Faustoferri and Cinzia Morelli, completed ent landscapes, starting in the mountainous heart
five full seasons (1994-1998), with a preliminary of the Apennines, a predominantly pastoral area,
season in 1993.1 A second phase, directed by Ed- and flowing through a wider but complex hilly land-

*Financialsupportfor the SangroValleyProjecthas been convertIdrisiimagesinto ArcViewgrids;to PaulMorris,David

provided by the CravenCommittee of the Universityof Ox- Thorpe, SarahRuth,and BrianWrightfor their discussionsof
ford, The Societyof Antiquaries,The Societyfor the Promo- the anisotropicalgorithm;and to AJA'sanonymousreviewers
tion of RomanStudies,and the BritishSchool at Rome,which for helpful suggestionson an earlierdraft.
also lent the projecta vehicle. We are gratefulto the Soprint- 'Two preliminaryreportshave been published:Lloydet al.
endenzaArcheologicadell'Abruzzo-Chieti fortheirsupportand 1997;Faustoferriand Lloyd 1998.
assistancethroughout the project.Alison Macdonald,who is 2Bisphamand Kane 2000;Bispham2001a;2001b.
studying the potteryfrom the project, kindly provided spot 3Barker1995a, 1995b.
dates on the sites analyzedhere. The authors are gratefulto SInaddition to Barker1995a,recent worksinclude Oakley
JustinMoatof the GISMadagascarBiodiversityProject,Royal 1995; Schneider-Herrmann1996; Tagliamonte 1996. These
BotanicGardens,Kew,U.K.,for his view-imp. avescript,used to are reviewedby Dench (1998).

169
American Journal of Archaeology 106 (2002) 169-86

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
170 TYLER BELL ET AL. [AJA 106

": .. . ..... .. . . .
" ?:?...??"............
??
•!"-?:?
.
.........
~ ~ .....
.....
~~........... i i i i i !:i
.........................
?

.~~~~ •:•i
.............. . ::
.......... ......:::
-. . i i i l i iii
:::: 0 .,..
:

::::::::::::::::::::::::::::::
. . .. ..
.. .

...............
.. .. .. .

...... . .. .... ...

" . . . "..
Ik. .......
.. . . . . .............
. . . . " ...
... . . ...... "" ":::.M P L A . •i
..........
J:•...... ...... ::::::::::::
..................
............============
....
:. .
. M
. .. .
........:::..::..
P :.:::::
Q.......
LA .....
.

... ..........
. . . .. . . .. . . . .
.
.

/
: ? {.i

5.
:•• ?... . . ., .. .. ..
. . . . . . . ..
• ::........
. .

"/ 0 ==========..
..... .:..
:::::::::::::::::::::::::::::::::::::::::::....:::::::::::::............:::::"........".•
...... =....==
...................•
ii "".'':::'.'...'..............., ".........
.~~~~~~~~~~~ "'""... ......
. ... .................
.

"'` .
2""" .. ..............
.........................: "? " '?? ?:?. '•.., :::
...........
... . . . . ..

...'. ......
..
...................
...........
. ... .
~.:::.:::.
lake? ?

landove 1 0m

0 50 kms'~ .f

Fig. 1. Central Italy,showing areas surveyed in the Sangro Valley. (Lloyd et al. 1997, fig. 1)

scape in the middle valley, where grain, olives, Settlement in the area surrounding Monte Pall-
wine, and fruit are grown, before the valley opens ano was investigated by an intensive fieldwalking
out into a much flatter floodplain as the river ap- survey, with walkers spaced at 10 m covering all walk-
proaches the sea. Clearly it was impracticable to able ground in a kilometer-wide transect from the
study in detail the entire valley, and survey there- treeline below the hillfort westward down to the
fore concentrated on three main areas: in the up- river, from the top of the hillfort eastward down to
per valley between Opi and Villetta Barrea, in the Tornareccio, and selected ground to the south in
middle valley around the hillfort of Monte Pall- the area around San Giovanni, and the plateau of
ano, and in the lower valley around a village called Fonte di Fontecampana (fig. 2). This survey re-
Fara, a distinctively Lombard toponym. This pa- vealed a large number of sites, the vast majority of
per concentrates on the middle valley, especially which lay on hilltops, ridges, or spurs. With the in-
the Samnite hillfort of Monte Pallano, with its tensive survey as a control, a smaller team then con-
massive polygonal walls, and on settlement in the ducted a more judgemental extensive survey, visit-
surrounding territory. Key questions were wheth- ing hilltops, spurs and ridges further afield, and
er the hillfort was occupied permanently, season- places with promising toponyms. Many more sites
ally, or only at times of crisis, and how far its terri- were discovered in this extensive survey. Most are
tory may have extended. Excavations on the pla- Iron Age/Classical, with a number continuing into
teau of the hillfort, conducted by both the Soprint- the early Middle Ages, but in general Medieval oc-
endenza under Amalia Faustoferri and by our cupation is less well represented in the field sur-
project under John Lloyd, Gary Lock, and Neil vey data.
Christie, have revealed substantial buildings, Sites discovered during field survey did not ex-
showing permanent occupation in the Roman ist in isolation, however, and the simple process of
period and certainly the presence of structures placing dots on maps does not do justice to the
from earlier periods.5 variety of ancient habitation; the importance of

5 Faustoferriand Lloyd 1998.

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 171

Sr La Crocetta
-. MPaI.ano -M.
.
i
..... '
Tornarecc
San AntonioI mbaN

1 1 /

Sambuceto
•
/W 't o.di
;~?':;
:>-, ; •.... • ...
?•... . ..
/ ..: I

i.-

eS Pietro
Cota
Lago di A a a
Bom ba
!:i:! -:
•
!:j•
• ..I.. .
>
....•!: \. .•
!, •.
,C
/ ... .........N
.
• ::!!:!•!:::,
i!!:!!:!i!!i• !!:!•
.v..-.vv ..v.

0 , ,
2kms areas
built-up (50mcontour
interval)
heavily wooded
heavily wooded terrain
terrain I::.:?:: areas of
=I] areas of intensive
intensive survey
survey

Fig. 2. Monte Pallano and surrounding area. North at top.

studying settlement within the landscape focuses able significance-are more ephemeral and leave
as much upon the relations between neighboring little, if any, indication of their former existence. It
or distant sites as well as their individual locations. is on these smaller-scale, local communication
Clearly, in such a hilly landscape as the middle San- routes that we focus here.
gro Valley, communication is difficult and slow, and Why are these smaller routeways important in our
the few communication routes that exist become understanding of the ancient landscape? If we were
correspondingly more significant. able to discover the courses of these minor ancient
Understanding the ancient communications roads, paths, and trackways, we might gain an im-
network on a local scale is essential to understand- pression of the relative importance of different sites,
ing the settlement process and the human exploi- and the analysis might also illuminate some of those
tation of the landscape. Larger, arterial communi- other concerns that persistently dog field surveys,
cation routes are still identifiable, many marked by in particular, the relationship between communi-
natural features such as river valleys and terrain that cations routes and indications of "off-site activity,"
will have changed little in the preceding centu- those high levels of "background noise" sherd dis-
ries. For interregional trade routes through the tribution that are nevertheless not dense enough
study area we may assume that the river valley re- to warrant treatment as a site scatter. But to address
mained the principal long-distance artery (al- these questions it is necessary to understand the
though the upper valley perhaps had closer links lines of ancient communications routes. How can
with the Fucine basin than with the Adriatic), but we trace them if they were merely tracks, and not
local communications between sites in the same paved or built with substructures or bridges?
general area of the valley are less obvious. These One approach was provided by the project's eth-
lines of minor communication routes-paths and noarchaeological work, which focused on pastoral
trackways linking long-disappeared sites of argu- transhumance.6 This research indicated that many

6 Conducted by Paul Beavittand MarinaAmbrosi;see Lloydet al. 1997.

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
172 TYLER BELL ET AL. [AJA 106

2014,
SMMARSICANO

'PI 0540
... . . ....
605* *re/f*606
.606

5701

0 2kms heavily wooded terrain site

areas of intensive survey (i0m contour interval)

Fig. 3. The upper valley survey area and sites discovered during survey. Site no. 620 is the
inscription recording the erection of the aedes;site no. 618 is the polygonal road revetment.

sites detected by the field survey in both the upper Marsica along the valley.' Some 2.5 km to the west,
and the middle valley lay on tratturi, traditional tran- an inscription of A.D. 14'4, recording the erection
shumance routes, some of which are marked on old of an aedes and marble statue to Jupiter,8 is cut into
maps or remembered in oral tradition. But the use the rock near the km 54 marker post on the same
of tratturi may have been interrupted by political and road and suggests that the modern road may be
economic changes; continuity across the early Medi- following the course of the ancient one for much of
eval period between Roman systems of long-distance this stretch of the valley.
transhumance and the Medieval tratturi cannot be In the constrained landscape of the upper valley
assumed. Nevertheless, although the permanence basin, the relationship between sites and commu-
of the longer, interregional routes may be doubted, nications routes is very easy to see. But for the more
they may have been built up initially from shorter complex terrain of the middle valley around Mon-
routes used earlier for local communications or, in te Pallano the situation is harder to understand
some cases, short-distance transhumance. (fig. 4). Do modern roads, with their hairpin bends
Another option is to read clues in the landscape to allow cars to climb steep hills, provide a reliable
itself. In the upper valley survey area, between Opi indication of routes viable in antiquity, or are farm
and Villetta Barrea, it was evident from the site dis- tracks, straighter and steeper for animals and cater-
tribution in the upper valley that that most sites lay pillar tractors, a better guide? Or might ancient
either on or overlooking the narrow corridor of routes have followed different courses altogether?
possible communication along the valley of the San- The essential problem, therefore, was to devel-
gro itself, or in the two valleys of its tributaries, the op a method of modeling likely human movement
Torrente Fondillo and the Torrente Scerzo (fig. 3). across the ancient landscape, based on known set-
A reassuring confirmation of the connection be- tlement locations and the topography of the region.
tween settlement and the existence of a main com-
THEORY AND METHODOLOGY
munications route along the main valley was the
discovery in 1995 of a massive substructure in po- Geographical information systems (GIS), de-
lygonal masonry, clearly the revetment of a road fol- signed for the analysis of spatial data, are an obvi-
lowing much the same line as the modern SS 83 ous tool to use when addressing problems of com-

7Lloyd et al. 1997, 29-32. 8 CIL10.5142.

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 173

Fig. 4. The landscape of the middle Sangro Valley, near San Giovanni on the south side of Monte
Pallano. (Photo byAndrewWilson)

munications routes in a landscape. But computer point called the "target." The overall cost is deter-
modeling of ancient communications routes re- mined by an algorithm that incorporates both the
quires an effort to reconstruct those elements with- distance of that point from the target, as well as
in the ancient landscape that dictated movement additional, relative costs based on particular aspects
through the region, and this is no simple task, even of the landscape. These additional costs are user-
in a comparativelylittle-changed, rural environment defined and usually are derived from a classifica-
such as the Sangro Valley. Topographical and geo- tion system in which each feature of the landscape is
graphical factors will have had significant impact assigned a cost value that is relative to a base value.
upon the manner in which people moved across For example, to determine the least-cost route
the landscape, and some of these, such as the loca- through a region that is composed of swamp, scrub-
tion of settlement sites, groundcover, sources of land, grassland, and forest, cost values for vegeta-
fresh water, and to a certain extent land use, can be tion must be incorporated into the cost surface
reconstructed or intelligently estimated by the ar- because these would have significant impact upon
chaeologist. Other, non-environmental factors will any movement through the landscape: naturally
have also influenced movement across the ancient people would tend to avoid the swamp, traveling
landscape: political boundaries, social divisions, through the plains, scrub, and perhaps the forest,
cultural taboo or attraction associated with certain with greater difficulty. When building this model,
sites; these leave little if any mark in the archaeo- the GIS operator or fieldworker must ascribe a rel-
logical record, and therefore can be difficult or ative cost to each type of vegetation, thus the cost of
impossible to reconstruct. Even when these factors each class is relative to the class with the lowest cost.
are identified one is still faced with the formidable In this example the base unit, grassland, is given
task of attempting to quantify each influence: of the value of one because it is level, well-drained,
how much more influence was (for example) prox- and easy to traverse. Having established a base, ar-
imity to water than the repulsion of an ancient cem- eas of forest can be given a relative cost of six, scrub
etery or taboo area in determining one's chosen a relative cost of three, and swamp a relative cost of
route through the landscape? 12, which means that traversing land covered in
forest is six times as difficult as traversing grassland,
CostSurfaces scrub is three times as difficult, etc. These values
To date, GIS and cost surface generation have may be assigned arbitrarily or, where possible, de-
been the tools used in the attempt to model move- termined by actual tests in the field. The "cost" can
ment across a landscape and its inherent difficul- be conceived ofas time or energy; the actual units
ties. A cost surface is a computer-generated model of measurement are irrelevant because the cost is
of the landscape in which each part of the surface is relative, not absolute.
assigned a "cost,"representing the effort or energy The methodology was first effectively applied to
required to reach that point from a predetermined an archaeological project on the island of Hvar in

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
174 TYLER BELL ET AL. [AJA 106
the late 1980s, where Gaffney and Standic attempt- Iron Age, the presence of footpaths and communi-
ed to define the territories of ancient sites through cation routes that may have cut through it remain
the use of site catchment analysis.9 Traditionally site unknown. Social influences on landscape are more
catchments have consisted of a simple linear pro- difficult to detect and quantify, and although they
gression of time and distance, so that the further a are extremely important in our understanding of
given point is away from a site, the higher the "cost" human interaction with the landscape, they are ar-
in getting to that point and back within a day-the chaeologically ephemeral. Because geology and to-
resultant catchment resembles a circular "bullseye" pography are the most static and unchanging land-
in which each band marks the maximum distance scape elements (outside of the obvious exceptions
one can travel from the central point in a given time. in which erosion is a major factor, including coast-
Significantly advancing this approach, Gaffney and lines, river deltas, and some aeolian environments)
Stanc'ic employed GIS to develop a topographically- they therefore remain the most reliably reconstruct-
based cost surface, in which the cost is not simply ed aspects of the ancient landscape.
based on distance, but also on the slope that one Geology and topography are very much two sides
must cross to reach that point and return. This pro- of the same coin; the firmament of the landscape-
cess attempts to address the fact that surmounting a the rocks and their composition-affects the geo-
steep slope demands a greater cost (in time or en- morphology which in turn gives rise to the valleys
ergy) in relation to the comparatively low-cost pro- and hills with which we are familiar as topographic
cess of moving on the level. A site situated in moun- features. This article concentrates primarily on the
tainous terrain will therefore naturally have a small- shape and form of the surface, the topography, as
er catchment area than a similar site on the plains, this has particular impact upon how people move
because greater time and energy are required to from place to place, especially in hilly or mountain-
cross a varied terrain than a comparatively flat one. ous regions. Even in comparatively level areas the
By generating cost surfaces and using the base mea- slightest rise or dip in the landscape will force the
surement of the time taken to walk 5 km across the traveler into a different pattern of movement. The
Stari Grad plain on the island, Gaffney and Stan'ic topographic element, however, will never be the
could generate site catchments for the island's pre- sole influence upon movement across the land-
historic hillforts, and study the proportion of other scape, but in simulating ancient and prehistoric
sites, soils, and additional geographical factors that landscapes it remains the most reliable. Therefore,
fell within the catchments. The process results in a any study of movement through an ancient land-
better understanding of the hillforts and their rela- scape should use a topographic model as its foun-
ti6onship to the immediate hinterland. dation, upon which additional models can be con-
Although Gaffney and Stan'ic subsequently ad- structed, incorporating further aspects of environ-
dressed their over-reliance on environmental fac- mental and non-environmental data. The remain-
tors, their initial results were highly significant.'0 der of this section presents the authors' attempts to
Their topographically-based cost surface offers sig- advance further the theory and methodology of to-
nificant improvements upon the traditional mod- pographically-based cost surface analysis; the final
el, because the incorporation of slope data begins section describes the results of its practical applica-
to approach some of the more sophisticated fac- tion to the landscape of the Middle Sangro Valley
tors, both environmental and non-environmental, and its illumination of the archaeology discovered
which make up a landscape. in the field survey.
Of these factors, topography and geology are the In the Hvar study, Gaffney and Stan'ic developed
only two that cannot be readily manipulated by arti- a cost model based on the angle of slope. This is a
ficial means. Ground cover, the most frequently cit- particularly good basis of cost because it obviates
ed cost variable, can be altered in comparatively the need to generate a user-created scale of rela-
short time: swamps can be drained, forests can be tive costs such as the example of ground cover giv-
felled, and scrubland cleared; even streams and en above. However, there are two key elements of
rivers can be bridged, forded, or diverted. If archae- moving up and down a slope that must be included
obotanical evidence for a specific region suggests before any attempts are made to model human move-
that an area was covered by woodland during the ment across it. First, we cannot assume that the cost

9Gaffneyand Stan'ic 1991.The text providesa solid intro- StanEic1998).

duction to GIS,including basicaspectsof datacapture,hard- "0Gaffneyand Van Leusen 1995.
ware, and analysis;it is now availableonline (Gaffneyand

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 175
tween the two changes would therefore be mgy,:
mgy2. Since a relative cost surface assumes the mass
of the traveler to be the same, and gravity is a con-
Y2 stant, we can reduce the ratio to y, : y,. For the pur-
pose of GIS analysis, the landscape is divided into
squares or "pixels"-each pixel, in this case, repre-
senting an area 10 x 10 m. Slope quantification in a
GIS allows only for angles of slope, not absolute
heights within a pixel, but because y,=xtan(O,) and
x y2=xtan(02) we can obtain the same ratio expressed
as xtan(02). Furthermore, because x (which
Fig. 5. Two slopes in profile xtan(1O):
is the distance traveled, or the size of one pixel)
remains the same, we can finally reduce the ratio to
of climbing a slope is directly proportional to the an acceptable tan01: tanO2. Therefore, the relative
degree of slope: thus surmounting a 450 slope is cost of ascending slopes can correctly be expressed
not simply 45 times as difficult as moving on a level, as the ratio of the tangents of the slope angles, tanO,
0' slope. (Taking this to its conclusion would sug- : tanO2, rather than simply by using the angle of slope
gest that climbing a vertical slope of 90' is only 90 itself.
times as difficult as walking on the level.) A cost- In this case we use a 1o slope as our lowest cost to
surface based simply on straightforward slope val- avoid the division by zero that would result if we
ues will not incorporate proper relative costs. were to take the tangent of 0-a level surface. The
The actual relationship between slope and the new data set is produced by taking the tangents of
relative effort required is slightly more complex, all slopes within the landscape, which are then di-
and can be expressed mathematically: figure 5 vided by the tangent of V1, to ascertain the cost of
shows two slopes in profile, their angles represent- ascending a steeper slope relative to the cost of
ed by 0, and 02, and their heights represented by y, ascending a 10 slope. The result is a more accurate,
and y2respectively. X is the horizontal distance the nonlinear cost surface in which the cost of sur-
traveler must cross to ascend the slope. mounting a 60' slope is nearly 100 times as diffi-
The goal in this exercise is to discover the com- cult as traversing a nearly level surface (fig. 6).
parative cost of ascending the two slopes: the change The second point that must be considered when
in potential energy after ascending the slope is = using slope as a cost basis is that a slope does not
mass * gravity * height ascended, and the ratio be- exert a force uniformly in all directions, but in fact

Comparative Energy for Calculating Force Magnitude

tan(slope)/tan(l o)
700

600 -----------------------------------------------

500

0 400

.
300 -

200

100--

0
1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85
Angleof Slope (degrees)

Fig. 6. Graph showing the nonlinear relationship of cost based on the tangent of slope

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
176 TYLER BELL ET AL. [AJA 106
exerts 100% of its force downhill." In the field, a figures in this paper have been generated. While
slope exerts its full force against a traveler moving the combination of three programs may at first ap-
directly uphill. When an object moves perpendic- pear complex, their incorporation is not difficult
ular to the slope there is comparatively little im- and provides a very powerful combination that al-
pedance. The relative friction of a pixel based on lows maximum data flexibility, portability, and ma-
slope, therefore, is wholly dependent upon the nipulation.
direction of travel across it (although the applica-
tion of this model to the real world must of course Anisotropic Cost Surfaces
take into account the serious costs involved in de- The project's primary use of the Idrisi program
scending steep slopes; we address this below). was generating anisotropic (nonuniformly direc-
Earlier slope-based cost surfaces such as Gaffney tional) cost surfaces using its VARCOST module.
and Stan'ic's were based on uniform slope values An anisotropic algorithm incorporates both the
so that, in effect, the computer simulation assumed magnitude and direction of a force, so that the re-
that walking up, down, or across a slope demanded sultant cost surface is largely dependent on the
the same cost. While subsequent landscape analy- direction of travel across these various forces from
ses were valid in so far as they ascribed low costs to point A to B. When applied to a terrain model, the
level land and high costs to all slopes, they could module can use data on slope (magnitude) and
not, by their very nature, vary the cost of a pixel aspect (direction in which the slope faces) to pro-
based on the direction of travel across it. duce a directionally-based cost surface in which the
direction of travel is calculated against the steep-
GIS Software ness and aspect of the slope. The result applies the
A major focus of the Sangro Valley Project has full force against the traveler when traveling up-
been the concept of "off-site" archaeology and the hill, with the traveler when walking downhill, and
total-area coverage that a raster-based GIS can pro- in proportional incrementsrelative to direction of trav-
vide." Cost surfaces are usually constructed in a el when walking at acute angles to the direction of
raster environment in which the landscape is di- force (fig. 7).
vided into pixels-the smaller the pixel size, the This algorithm will clearly present problems with
greater the number that are needed to cover the very steep slopes because it would assume that mov-
landscape, and therefore greater the resolution. ing down a 500 slope is very easy (low-cost), when in
Thus if a pixel is assigned a value of 10 in a cost fact the descent of a 50' slope can be more of a
surface, it means that getting to that pixel from the tumble than a controlled perambulation. Even
target point "costs" as much as traversing 10 pixels slopes less steep can be almost as tiring to descend
with a base value of one, or five pixels with a value of as ascend. To address this issue we reclassified the
two, etc. The project's raster-based analysis, includ- slope data so that extremely steep slopes were giv-
ing surface coverage generation and more ad- en proportionally high costs reflecting the difficul-
vanced spatial functions, were performed using ty in traversing them from any direction. Further-
Idrisi, an inexpensive but powerful GIS well suit- more the module assumes that a direct traverse of
ed to this kind of analysis. However, the raster envi- the hillside-moving perpendicular to the force-
ronment in general, and the Idrisi program in par- would incur simply the base cost, no more than if
ticular, does not have the spatial precision and da- the walker were moving on level ground. This ap-
tabase connectivity that a project of this nature de- parent discrepancy can be accommodated by a user-
mands.'" Therefore the primary spatial model of defined function in the algorithm that incorporates
the project's GIS has been constructed within "corrections" for traversing a hillside at angles near-
ESRI's ArcView, and is tied to relational data in a perpendicular to the slope. Furthermore, the
Microsoft Access database. The results of the Idrisi VARCOST module can incorporate a traditional
raster analysis in turn could be imported directly isotropic cost surface (i.e., one with no weighting
into ArcView, in which we could use the greater according to direction of travel), which could be
functionality (and more friendly interface) of the used for incorporating additional geographic and
ArcView "Spatial Analyst" extension, with which the non-environmental costs, such as ground cover. The

"We are
dealing with frictions rather than forces, but to using Idrisifor DOS. There have been two majorrevisionsof
avoid confusion we will refer to the slope as exerting a force. the program;the latest (Idrisi32)offersgreaterdatabasecon-
2 Lock et al. 1999. nectivity and a friendlier user interface (http://www.idrisi.
'3The cost surface analysisfor this paper was undertaken clarku.edu).

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 177

Upslope

StraightUphill:100% of force against walker

450 uphill:50% of force against walker 45' uphill:50% of force against walker

Traverse:no force against walker Direction of Traverse: no force against walker

Travel Across Slope

45" downhill:50% of force withwalker 45" downhill:50% of force withwalker

StraightDownhill:100% of force withwalker

Downslope
Fig. 7. Anisotropiccost surfaces:a diagramdemonstrating the algorithm'sapplication of force
vectors upon movement across a slope determined by the walker's direction of travel. The
algorithm is documented in the Idrisi Reference Manual, section 15.
end result is a combined cost image that more close- ther insight into the causes behind the positioning
ly reflects the relative difficulties of moving through of some sites within landscape and even contribute
a natural landscape. to an understanding of existing landscape features
such as trackways and field boundaries.
Cumulative Pathway Analysis The overlapping of pathways between multiple
The cost surface is actually an intermediary data sites can be quantified in a GIS to determine where
set used to generate optimal paths in which the GIS and how often pathways between different sites cor-
calculates the least-cost route between a target site respond. The fictitious example in figure 8 shows
and a point (or series of points) within the land- optimal paths that have been generated from point
scape. When this analysis is applied to data from a B to points A and C. The two pathways are then
regional archaeological survey, possible routes of spatially overlain to produce a cumulative pathway,
communication between contemporary sites can be in which cells common between the two receive a
mapped. Because cost surfaces calculate the cost value of two; overlapping cells of a third pathway
of getting to or from a specific point in the land- would assign the cell a value of three, and so on. In
scape they are strictly site-specific, and no single a multisite environment this quantification of path-
cost surface can be generated for the purpose of way overlap can help point to those paths that may
modeling bidirectional movement among multiple have been used most frequently in antiquity.
points in an entire region. As such, the analysis of
APPLICATION: THE SANGRO VALLEY PROJECT
any multisite landscape requires the generation of
a single cost surface for each site. The following section illustrates the pathway
The route that is generated between sites can be networks that have been generated between sites
very informative (and surprising) by itself, but to- around the slopes of Monte Pallano in four very
gether with other pathways can contribute greatly to broad periods: the Bronze Age; the Iron Age/Sam-
understanding avenues of communication through nite period (Impasto and Black Gloss pottery); the
a region. A network of these pathways can add fur- Roman period, and the Medieval period, starting

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
178 TYLER BELL ET AL. [AJA 106

Fig. 8. Cumulative pathway analysis to site B from sites A and C: where the two
pathwaysoverlap the corresponding cells (pixels) are given a value of 2, shown here
as a darker cell (the cell size has been exaggerated for clarity).

about the 6th century A.D. Pottery analysis is ongo- di Fontecampana (ca. 1 x 0.5 km) and the arable
ing, so the dating of sites in the middle valley is land below the wooded slopes in the southeast cor-
provisional yet acceptable for this preliminary anal- ner of the map grid (fig. 2). Most sites within these
ysis. Cost surfaces as described in the previous sec- areas were found on hilltops, ridges, and spurs-
tion were generated in Idrisi for each site, using a indeed, very few topographical eminences lacked
pixel size of 10 m. Cumulative pathway analysis was sites-and many similar sites were discovered dur-
then performed for all sites in occupation within ing the later extensive survey.
each of four broad periods for which the pottery With the exception of the summit of Monte Pal-
typology allows recognition. lano, where stretches of massive polygonal walling
The landscape was walked in teams of four to survive and excavations have revealed buildings of
seven people spaced 10 m apart, collecting all arti- Roman date, and of scatter 18 (Fonte di Fontecam-
factual material within a meter on either side of pana), where there is an enigmatic rock-cut fea-
them, thus giving a notional 20% coverage. Al- ture, none of the sites in this area is signalled by
though in the first season we plotted our position visible structures, and all were located as scatters of
from maps and imposed rectangular transects on artifacts. No attempt was made to define sites by the
the landscape with tape and compass, this proved rigid application of a threshold density of sherds
overly time-consuming in hilly terrain and from the per unit area, as we felt that would give a false sense
second season onward we used air photographs tak- of precision to very imprecise surface data; more-
en from 10,000 ft. and enlarged to 1:5,000 to locate over, we were conscious that the background scat-
our position, treating each modern field as a sepa- ter of sherds varied considerably within the survey
rate unit for recording purposes. Within the 10 x region, from nothing or almost nothing to an ap-
10 km IGM map grid area around Monte Pallano, a preciable quantity of off-site material, which was
transect 1 km wide from Monte Pallano down to the nevertheless too diffuse to represent a discrete con-
river was surveyed, with the exception of the steep- centration of artifacts. The important criterion for
ly wooded upper slopes of Monte Pallano itself; we site definition, we felt, was that a concentration of
also walked the unwooded plateau around Fonte sherds should stand out as noticeably more dense

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 179

CumdlatiePaths
HighOverlap

ULowOverlap

r•
[Sltlilm
S•t,]

Fig. 9. The Bronze Age sites and pathwaynetwork in the immediate area around Monte Pallano.
The Sangro Riverruns from Lago di Bomba (artificiallydammed in modern times) along the valley
to the left of scatters 61 and 62. The area shown here and in following figures is the area shown
on the map in figure 2, covering 5.7 x 7 km. For location of toponyms not marked on this figure,
see figure 2.

than the background scatter in the immediate vicin- 65 (on a hilltop) and two sites (58 and 18) along
ity. This judgement necessarily could only be made the springline at the foot of the heavily wooded
in the field by the team leader. In many cases the section of the slopes of Monte Pallano. A seventh
edges of the site were clear and could be defined site (91) was exposed in a track cutting south of
with some certainty, but the exception was the re- San Giovanni.
gion around the southern flank of Monte Pallano, Figure 9 shows the paths of least resistance cal-
where levels of off-site material (especially impasto) culated between the sites producing Bronze Age
were high between much denser concentrations of pottery. Paths between scatter 91 by San Giovanni
scatter material, and the edges of sites were blurred. in the southeast take one of two routes to the sites
The overall settlement patterns show a fairly dense discovered in the transect below Bomba-a lower
occupation of the rural landscape, especially in the route, running past Colle S. Pietro (which becomes
Iron Age to Roman periods. We were unable to in- important in later periods), and an upper route,
vestigate in detail the relationship of settlement to running through a plateau where we found our larg-
the river Sangro itself, as the modern Lago di Bomba est Classical site, scatter 18 at Fonte di Fontecampa-
(1961) had flooded the river valley below (west of) na. The slight plateau at the Fontecampana site in
Colle S. Pietro and Colle Butino. Below the modern fact becomes the key to communications routes
dam, however, some sites were found on the river between the south slopes of Monte Pallano and the
terraces by the railway station at San Antonio. Al- eastern slopes. The site scatter here includes Bronze
though the Lago di Bomba has clearly altered this Age and Iron Age sherds, and seems to be richest
part of the landscape since 1961, it does not affect in sherds and wasters of the Classical/Roman peri-
our current analysis of the communications between od, when it covered all the ground visible in figure
the sites that we did find, as all our survey work lay to 10 up to and beyond the farmhouse. In the Bronze
the east of the river, and none of the calculated routes Age the outline of later route networks begins to
involves an attempt to cross the river. develop, as does the existence of particular sites
that were to remain important on a local scale.
Bronze Age Settlement and Communications
Seven sites in the Monte Pallano area yielded Iron Age Settlement and Communications
Bronze Age material (low-fired, handmade ceram- Many more sites produced Iron Age and Sam-
ics, often associated with late and crude struck nite material, of which handmade impasto sherds
flints): two small sites on river terraces on the right were the dominant indicator, with some Black Glaze
bank of the Sangro (scatters 61 and 62), a small present for the later part of this period. The distri-
spur site (67) below modern Bomba, and scatters at bution of Iron Age impasto and Samnite finewares

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
180 TYLER BELL ET AL. [AJA 106

Fig. 10. Fonte di Fontecampana (scatter 18), apparentlya key site in the communications of the region
from the Bronze Age onward. A spread of pottery and tile extended from the camera position to
beyond the farmhouse in the distance, across the full width of the natural terracebetween the treeline
to the left and a point where the ground falls awaybeyond the right hand edge of the picture. (Photo
byJennie Lowe)

(and probable coarsewares) was very similar, and nite period; impasto sherds of this date cover a
although it would be possible to separate these larger area than the Classical and Medieval nucle-
two periods, they have here been combined be- us. Scatter 18 at Fontecampana produced impasto
cause the pathway analysis is virtually identical for over much of its area; that some of this material
both. All the Bronze Age sites continued in occu- may derive from burials is suggested by an earlier
pation. Excavations on Monte Pallano indicate find of an Iron Age bronze chatelaine from the Fon-
occupation from the fourth or third centuries B.C. tecampana area, typical of female elite burials of
onward, and although the megalithic walls on the fifth century B.C. An enigmatic rock-cut fea-
Monte Pallano are not closely dated, they evident- ture has been seen as a "donario" (receptacle for
ly belong to the Samnite period, indicating an ritual offerings) but is probably better interpret-
important fortified center.14 A temenos wall and ed as a rock-cut sarcophagus with lid seating; its
quantities of architectural terracottas discovered date is uncertain but it probably belongs to the
in recent excavations seem to indicate a substan- Samnite rather than to the Roman period.16
tial Samnite sanctuary of the third to late second New hilltop sites appear at scatter 88 and 84 (Col-
or early first centuries B.C.1" To the west of Monte le S. Pietro)-at the latter site Iron Age grave goods,
Pallano, several large sites are interpreted as ham- including a spearhead, were found several years
lets or villages: scatters 58, 65, and 18, at Fonte di ago."7 On the southern flanks of Monte Pallano a
Fontecampana. Scatter 65, on a low hill below Bom- rash of new small sites appears, mainly concentrated
ba, may be truncated by the modern highway, but around the break of slope at the springline where
measured at least 450 x 200 m at its maximum the steep wooded slopes give way to flatter arable
extent, which appears to be in the Iron Age/Sam- land between San Giovanni and Colle Butino.

4Faustoferriand Lloyd 1998, 11-2. 16Chatelaine:

Cuomo 1985,43; Faustoferriand Lloyd1998,
1 Bispham2001a;report on the 2001 season availableon- 15;"donario":Cuomo 1985,26.
line in Bispham(2001b,http://www.sangro.org/reports/2001/ 7Faustoferriand Lloyd 1998, 18.
contents 2001.html).

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 181

CunulativePaths
HighOverlap

LowOverlap

Fig. 11. Iron Age/Samnite sites and pathwaysaround Monte Pallano

The resulting path network is more complex (fig. tain might well have used routes across the sum-
11). Nevertheless, the high and low routes are still mit, rather than going around the south of the
apparent, and the importance of the Fontecampa- mountain via San Giovanni. Settlement on the sum-
na plateau is clearly evident as a junction of a mul- mit of Monte Pallano thus needs to be considered
titude of routes. Fontecampana also provides the as part of the regular communications network, rath-
jumping-off point for almost the only viable route er than as isolated and ill-frequented.
up to the top of Monte Pallano from its western The second point to note is the frequency of path-
side. The modern road also passes through this ways between the numerous sites on the southern
point to ascend Monte Pallano but winds a much slopes of Monte Pallano, in a belt along and below
more circuitous route with hairpins up to the sum- the spring line between San Giovanni (scatters 14
mit, which takes about 15 minutes in a vehicle. Con- and 15) and Colle Butino toward Fontecampana.
sequently the authors had initially believed the top This is the area shown in figure 12, looking north,
of Monte Pallano to be a difficult place to reach, and the most frequented pathways run approxi-
and one that may have been somewhat remote from mately along the line of the modern road visible as
the settlements around its base. However, after test- a hedgerow in the middle ground.
ing routes on foot it was found that the excavations Perhaps significantly, this zone is the reported
of the Soprintendenza on the summit could be location of the find of the Atessa torso, a Samnite
reached in 25 minutes from Fontecampana (using funerary sculpture of the same generic type as the
a route close to that predicted by the computer), Capestrano Warrior (ca. 550 B.C.), and now in Chi-
with a further 35 minutes down the other side to eti Museum (fig. 13)."8 Furthermore, in this area a
the village of Tornareccio, east of Monte Pallano, scatter of Iron Age impasto (50-52) below the
where Iron Age graves have been found-a total of treeline coincides with a group of trees growing
only one hour on foot from Fontecampana to Tor- out of stone cairns (fig. 14). The regular spacing of
nareccio. Alternative routes from Fontecampana to these just below the treeline, and the scatter of im-
Tornareccio around the mountain take about half pasto around each one, raises the suspicion that
an hour by car; on foot they would take much long- these may not be simple clearance cairns but rath-
er as the distance is several miles. Field-testing the er funerary markers in a line overlooking a well-
computer-generated pathways therefore demon- traveled road. The nearby site of Acquachiara (scat-
strated that the summit of Monte Pallano is readily ter 21) has also produced an archaic statue-stele
accessible, and that communications between set- fragment from a funerary marker and a spread of
tlements on the east and west sides of the moun- impasto wares.

' Faustoferriand Lloyd 1998, 13, 15.

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
182 TYLER BELL ET AL. [AJA 106

Fig. 12. Terrain around the south flank of Monte Pallano. Predicted ancient communications routes
roughly follow the line of hedgerow and trees running from left to right in the middle distance. (Photo
by AndrewWilson)

This area produced not only many sites but also a nouncod ridge overlooking the natural depression
high level of off-site material. This is in sharp con- through which a modern north-south road still
trast to other areas of the survey where the sites are runs.
often very well defined and there is almost no off-
site material visible in plowed fields. While some Roman Settlement and Communications
of the background noise in this area may be manur- Excavations above the Fonte Benedetti on the top
ing scatter or spread around sites themselves, it of Monte Pallano have revealed a porticoed struc-
appears to be very diffuse and not related to neigh- ture, probably a public building, erected probably
boring sites; much of this off-site material is impas- between the mid first century B.C. and the early
to, concentrated along the line of the predicted first century A.D. Subsequent alterations are asso-
communications routes. The high levels of off-site ciated with ironworking activity, and the complex
material might be a result of colluviation or of the seems to have been abandoned in the late second
collapse of ancient agricultural terraces at the top century A.D. Surface finds on the plateau around
of the arable slopes, but either explanation seems the excavations suggest a large area of Roman peri-
to imply a concentration of sites along the lower od occupation, and small excavations at the south-
edge of the woods at the break of slope. The possi- erly peak of the summit, La Torretta, indicate Ro-
ble cairns at sites 50-52 and the evidence for fu- man structures and material there too.19 It is not
nerary sculpture from the area suggest that the cor- clear whether the former Samnite sanctuary re-
relation between the communications routes and tained its importance after the Social War, but the
the high levels of off-site material may be explained extent of surface material and the relative richness
if the break of slope here was lined with tombs over- (for the region) of the excavated remains suggest a
looking major routes. sizeable early Roman settlement on the top of Mon-
A final point to note is that the lower route be- te Pallano. Clearly by this period the defensive util-
tween the eastern and the southern sites is com- ity of the hilltop had ceased to be important, and
manded by scatter 84 on Colle San Pietro, a pro- the site here should be seen as closely linked into

' Faustoferriand Lloyd 1998, 8-12.

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 183
Scatter 58, at the springline on the break of slope
below the trees on the flank of Monte Pallano, mea-
sured at least 140 x 200 m and certainly extended
further than could be seen. Visibility of the site,
however, was obscured by vegetation, modern quar-
rying, and colluviation on the uphill (east) side,
where landslip exposures revealed an archaeolog-
ical sequence with a quern fragment and other
material in the section.
Scatter 18, at Fontecampana, is also situated on
the springline at the break of slope between the
steep wooded ground and the gentler arable ter-
rain. The plateau here contains the largest scatter
discovered in the immediate surroundings of Monte
Pallano, and indeed the largest in the entire sur-
vey, measuring some 500 x 300 m. Unlike scatter
65, Fontecampana seems to have peaked in the
Roman period, to judge from the quantities and
spread of Roman material. It is also one of the most
varied assemblages, yielding finds of quernstones,
tile and dolium wasters, coins, and a piece of prob-
able hypocaust pila, hinting at the presence of baths.
Three inscriptions (a dedication and two tomb-
stones: CIL 9.2949, 2972, and 2978), now in the
adjacent village of Sambuceto also probably derive
originally from Fontecampana. The pathway net-
Fig. 13. The so-called Atessa Torso, part of an over-lifesize works show the continued importance of the Fon-
archaic funerary sculpture found near Colle Archiano on
the south flank of Monte Pallano. (Courtesy of the tecampana site, and its location at a junction of
SoprintendenzaArcheologica dell'Abruzzo-Chieti) heavily frequented routes helps to explain its pre-
eminent size.
the surrounding settlement pattern as a communi- To the south of Monte Pallano, between Colle
cations hub rather than a fortified retreat. Butino and San Giovanni, are a large number of
Around Monte Pallano many of the Iron Age/ smaller sites (50-100 m across), which probably
Samnite sites remained in occupation into the Ro- represent farmsteads exploiting the arable in this
man period, and consequently the predicted com- area. One of the richest of these (in terms of fin-
munications network looks very similar (fig. 15). ewares) is scatter 21 at the spring of Acquachiara.
None of the sites could be termed a villa-there But here too there are hamlets or villages-scatter
are no architectural moldings, worked stones, mo- 15 to the west of San Giovanni is evidently one of
saic tesserae, or other material suggesting architec- these, a large spread of material partially obscured
tural pretension, and only one small piece of red by modern cultivation but producing a rich set of
painted wall plaster was found during the entire Roman finds, including a stamped tile reading
survey of the middle valley area. To the west of Mon- [S]EXPONT, identical to one from Monte Pall-
te Pallano the dominant settlement type seems to ano.20 It is very likely that this scatter represents
be the hamlet or small village: scatters 65, 58, and part of the same site as scatter 14, to the east of San
18 (Fontecampana) all produced material of this Giovanni, and that the present village obscures con-
period from areas over 200 m across. The hilltop tinuity between them.
site, scatter 65, produced a sizeable scatter of early
Roman material (ITS sherds), though from a more Medieval Settlement and Communications
restricted area than had yielded impasto sherds; Medieval occupation discovered by the survey is
but further contraction of the site was evident with- much reduced in quantity from the Iron Age and
in the Roman period, as ARS sherds came only from Classical periods. It may be that the majority of Me-
a limited area in the center of the site. dieval sites have in fact survived to the present day

20Cuomo 1985, 56.

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
184 TYLER BELL ET AL. [AJA 106

Fig. 14. The site of scatters 50-52, on the south flank of Monte Pallano. A scatter of pottery and tile
across the area in the foreground represents a Roman site, while large quantities of impasto found
immediately below the treeline where trees are growing out of stone cairns may indicate an Iron Age
cemetery. (Photo by AndrewWilson)

as the villages of Bomba, Tornareccio, San Giovan- period, evidence from zoned collection within the
ni, and others, and so remain invisible to fieldwalk- area of a site indicates that these sites were shrink-
ing techniques. But in each of those cases it is topo- ing in occupied area (or at least in the area produc-
graphically likely that there was prehistoric and clas- ing sherds), with the Medieval sherds coming from
sical settlement at those sites too, and in fact this a reduced nucleus in the core of the Classical sites.
has been demonstrated by finds at Tornareccio and At scatter 65 the late ARS and Medieval wares come
San Giovanni. The Medieval period therefore wit- from a restricted area in the center of the overall
nessed either an overall population decline or an scatter. The apparent reduction in dispersed set-
increased nucleation of settlement, or both. Where tlement leads to a fall in the number of predicted
Roman sites do continue into the early Medieval communications routes and a consequent simplifi-

Curnlative Paths
[] Hgh Overlap

ELawOdap

oc5
SC6 0 xcvaton
esc8

. **4

e.'Ge
[@[Rle1' _]

Fig. 15. The Classicalsites and pathwaynetwork

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
2002] LANDSCAPE AND COMMUNICATIONS ROUTES 185

Pths
CTmdative
i FighOverlap

Low
Ovedap

Fig. 16. The Medieval sites and pathwaynetwork

cation of the pathways network. The main arteries We can conclude that many of the existing routes,
from previous periods remain in use, especially the including the tratturi, in the Monte Pallano region
upper route along the springline, on the south- appear to have been dictated by topographical fea-
western flank of the mountain (fig. 16). tures (which is not surprising in such a mountain-
ous area), and that the predicted trackways between
CONCLUSION ancient sites appear to follow much the same pat-
Evidence that the computer model is predicting tern of communications. Although this does not in
sensible and perhaps long-lived routes is provided itself prove that the tratturi must go back to remote
by the two calculated "classical period" crossroads antiquity, it is very possible that some of them-or
north of Fontecampana. Although survey did not at least segments of longer-distance routes-may
reveal sites at these locations, we did find junctions do so. Barker has noted that in a neighboring re-
of modern farm tracks, which appeared to take sim- gion of Abruzzo the Bronze Age seems to mark the
ilar routes along gullies in the area to the routes start of limited exploitation of the Apennine pas-
predicted by the computer. The modern metalled tures for seasonal pastoralism, although modalities
road, by contrast, takes a winding route with several of seasonal transhumance varied considerably be-
hairpin bends in this region, and is much slower tween then and the Medieval period.22
for a traveler on foot than are the farm tracks run- Future research will extend the same techniques
ning along the routes preferred by the computer's to a wider area, once we have obtained elevation
algorithm. Furthermore, the route down from La data for more of the region around Monte Pallano.
Crocetta (scatter 93), a site below the north end of This should allow us to calculate routes between
Monte Pallano, to scatter 58 and the sites below sites found by the extensive survey, and to set Mon-
Bomba to the west of the mountain, broadly follows te Pallano in a wider framework of longer-distance
an ancient tratturo. The frequent coincidence be- communications. Such a framework might help to
tween actual trackways and the communications address the question of territorial limits; the ap-
routes predicted by the GIS therefore suggests that plication of time analysis to the pathways, for ex-
our assumptions about the effects of slope and as- ample, could provide a model of the areas that
pect on human travel are broadly valid and are ap- could be reached from Monte Pallano in a given
plicable to this region.21 amount of time.

2A similarmethodologywasrecentlyappliedin Englandto tween hillfortscorrespondscloselywith the route of the exist-

a series of Iron Age hillfortsconnected by an extant, prehis- ing trackway.
toric trackwayin the BerkshireDowns (Bell and Lock 2000). 22Barker1989; 1995a, 157-8.
The analysisshows that the course predicted by the GIS be-

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions
186 T. BELL ET AL., LANDSCAPE AND COMMUNICATIONS ROUTES
This analysis has focused on very local routes in a logical and Geomorphological Record.London: Leicester
spatially limited area around the slopes of Monte University Press.
Pallano, restricted by the zone for which detailed, Bell, T., and G. Lock. 2000. "Topographic and Cultural
Influences on Walking the Ridgewayin Later Prehistor-
digitized elevation data was available. Nevertheless, ic Times."In BeyondtheMap:Archaeology and SpatialTech-
it has shown the power of computerized pathway nologies,edited by G. Lock, 85-100. Amsterdam: IOS.
analysis techniques for understanding settlement Bispham, E. 2001a. "Sangro Valley Project 2000." Papers
patterns within a landscape, and in particular for of theBritishSchoolat Rome69:409.
-.2001b. TheSangroValleyProject.http://www.sangro.
assessing the relative importance of different sites.
The role of the Fontecampana site as a communi- org (4 March 2001).
Bispham, E., and S. Kane. 2000. "Excavations at Monte
cations hub is brought out very clearly; but path- Pallano (CH) in Abruzzo." Papersof theBritishSchoolat
ways may in some cases serve also to suggest possi- Rome68:397-8.
ble explanations for anomalously high levels of off- Cuomo, L. 1985. Mostrafotograficaarcheologicasu Monte
site material. Comparison between predicted an- Pallano e dintorni.Lanciano: Editrice Rocco Carabba.
Dench, E. 1998. Review of TheHill-Fortsof theSamnites,by
cient routes and actual modern roads or farm tracks S.P. Oakley; The Samnites of the Fourth CenturyBC as
showed a reasonably high level of correspondence, Depictedon Campanian Vasesand in OtherSources,by G.
with tracks perhaps offering a better guide for the Schneider-Herrmann; and I Sanniti: Caudini, Irpini,
quickest routes for a traveler on foot. This may sug- Pentri, Carricini, Frentani, by G. Tagliamonte. AJA
102:441-3.
gest that many of these existing tracks-and not Faustoferri, A., andJ.A. Lloyd. 1998. "Monte Pallano: A
least the tratturi-could perpetuate routes that Samnite Fortified Centre and Its Hinterland." JRA
predate recorded history. 11:5-22.
Gaffney, V., and P. Van Leusen. 1995. "Postscript - GIS,
OXFORD ARCHDIGITAL Environmental Determinism and Archaeology." In
27 PARK END STREET Archaeologyand GeographicInformationSystems:A Euro-
pean Perspective, edited by G. Lock and Z. 367-
OXFORD OX 1 1HU Stan.ic,
82. London: Taylor and Francis.
ENGLAND
Gaffney, V., and Z. Stan'ic. 1991. GISApproachesto Re-
TYLER.BELL@OXARCHDIGITAL.COM gional Analysis:A CaseStudyof theIsland ofHvar. Ljubl-
jana: Znanstveni Inititut Filozofske Fakultete.
INSTITUTE OF ARCHAEOLOGY
1998. GISApproachesto RegionalAnalysis:A Case
•-.
Study of the Island of Hvar. http://www.archaeology.
36 BEAUMONT
STREET
usyd.edu.au/VISTA/gaffney_stancic (4 March 2002).
OXFORD OX 1 2PG
Lloyd,J.A., N. Christie, G. Lock, S. Agostini, M. Ambro-
ENGLAND si, P. Beavitt, T. Bell, L. Dallai, C. Doherty, A. Faustof-
ANDREW.WILSON@ARCH.OX.AC.UK erri, A. MacDonald, C. Morelli, L. Shepherd, L. Tuli-
pani, A. Wickham, and A. Wilson. 1997. "From the
Mountain to the Plain: Landscape Evolution in the
29, THE GREENWAY Abruzzo. An Interim Report on the Sangro Valley
UXBRIDGE Project (1994-5)." PBSR65:1-57.
MIDDLESEX UB8 2PJ Lock, G., T. Bell, andJ.A. Lloyd. 1999. "Towardsa Meth-
ENGLAND odology for Modelling Surface Survey Data: The San-
ANDREW.J.WICKHAM(UK.PWCGLOBAL.COM gro Valley Project." In GeographicalInformationSystems
and LandscapeArchaeology,edited by D. Mattingly and
M. Gillings, 55-64. Oxford: Oxbow.
Works Cited Oakley, S.P. 1995. TheHill-Fortsof theSamnites.London:
British School at Rome.
Barker, G.W.W. 1989. "The Archaeology of the Italian Schneider-Herrmann, G. 1996. TheSamnitesof theFourth
Shepherd." Proceedingsof theCambridgePhilologicalSoci- CenturyB. C. as Depictedon CampanianVasesand in Oth-
ety215:1-19. erSources.BICSSuppl. 61. London: Institute of Classi-
- 1995a. A MediterraneanValley.LandscapeArchae- cal Studies University of London, and the Accordia
.
ologyand Annales Historyin theBifernoValley.London: Research Centre University of London.
Leicester University Press. Tagliamonte, G. 1996. I Sanniti: Caudini, Irpini, Pentri,
, ed. 1995b. TheBifernoValleySurvey.TheArchaeo- Carricini,Frentani.Milan: Longanesi 8&C.

This content downloaded from 128.235.251.160 on Tue, 30 Dec 2014 01:01:27 AM

All use subject to JSTOR Terms and Conditions