LOCATIONAL MODELS AND THE STUDY OF ROMANO-BRITISH SMALL TOWNS
Simon Clarke

URBAN HlERARClllES IN ROMAN BRITAIN – —
When we attempt to consider the role played by urban settlements in
Roman Britain we usually work on the assumption that larger settlements
occupied a more senior position within the central place hierarchy. Hence
a large city like Corinium is usually seen as having served as a major economic as well as administrative and political central place. Small towns like
Worcester, which were clearly major centres for industry are often assumed
to have performed administrative duties, as capitals of tribal septs or in
some cases to have achieved civitas capital status. However it is clearly
dangerous to assume that central place functions existed just because a
large population is present. Clearly all men are not equal in the Roman
world, in political terms for example one noble man is more significant
than any number of landless farmers. When we consider the role that a
settlement played we must remember that size is not important, its what
you do with it that counts. This paper will reconsider the nature of the
central place hierarchies in Roman Britain placing much more emphasis on
the settlement’s character rather than simply its size.
CmusTALLER’S CENTRAL PlACE THEORY – —
Central place analysis in archaeology has borrowed heavily from the New
Geography of the 1960s. The most influential theory to be pressed into 82 S. CLARKE
K=3
1. satellite settlement
• 2. central place
3. road
4. central place territory
Figure 6.1. Christaller’s central place theory.
service in the prehistoric and early historic periods is without doubt
Christaller’s model which attempts to define an optimal, least cost organisational structure within a network of related sites (Fig. 6.1). This was first
outlined in the 1930s (Christaller 1935), but did not become widely known
until it was brought to the attention of the English speaking world by
Baskin in the mid 1960s (Baskin 1966). The theory is summarised by the
following five points:
i. Specialisation. Even in simple agrarian societies, certain sections of the
population demand products or services that they cannot provide for
themselves. Service centres are necessary for the circulation and exchange
of these services or products.
ii. Minimisation. To reduce the effort of obtaining services to a minimum LocATIONAL MODELS AND SMALL TOWNS 83
they are agglomerated within a single centre, which is loeated centrally
within a roughly circular territory. This also has the effect of maximising
the trade opportunities for those offering the services.
iii. Lattice packing theory. Assuming a featureless plain with even population density a triangular arrangement of centres each surrounded by an
hexagonal tributary area, has the most efficient geometric characteristics
(Haggett 1965: 49).
iv. Tiered hierarchy. The settlement hierarchy would be made up of distinct levels.
v. Variations oj the pattern. Within the above frame work a number of
different patterns are possible. The three simplest hierarchies are as follows.
K=3, Market maximising hierarchy, secondary settlements lie at the
boundary between three central places, to maximise choice of
market. A settlement system based on this pattern is part of a
highly commercialised society.
K=4, Transport maximising hierarchy, secondary settlements lie on
roads between the central places. This too is a highly commercialised system.
K=7, Administrative hierarchy, secondary settlement lie entirely
within the territory of the central place so that there is no scope for
competitive marketing.
(Note that K refers to the number of settlements that lie within a
central place field. Where a settlement lies on a boundary between
central places it counts only as a fraction.)
APPUCATION OF THE CENTRAL PLACE MODEL TO ROMAN BRITAIN —
Surprisingly little work has been directed towards loeational models in the
study of Romano British settlement since Ian Hodder’s study of larger
walled small towns (Hodder 1972) (Figs 6.2 and 6.3). Hodder saw urban
location as conforming closely to Christaller’s K=4 model. The model had
much to commend it. That the large public towns were central places for
the wider tribal area eannot be disputed. The choice of small towns plotted
as the next layer in the hierarchy also appears logical as they do represent
the largest walled areas. However there is no clear division in size or
characteristics between second and third tiers of settlements as suggested 84 S. CLARKE
• Civilils capitals ! colonia • principal small towns o 50
– – –
Figure 6.2. Small looms (after Hodder).
by Hodder. A number of walled small towns only slightly smaller than
those towns plotted by Hodder might reasonably be added to the map,
seriously disrupting his neat pattern of central places surrounded by six
satellites.
GENERAL OBJECTIONS TO CHRISTALLER’S CENTRAL PLACE THEORY –
One of the strongest objections to this use of central place theory is that
there is a tendency to assume that lower order settlements develop chronologically after and at the dictate of the larger settlements. For instance Got-LocAnONAL MODEI.S AND SMALL TOWNS
20 .— —
A
• 2. colonia
5. roads
\
.—- \
,
,
,
,
)0 •
3. principal small towns
6. small town territories
A 2
/”, • / ,
/ ,
\
\ 9
– – –.
/
/
I
I
,
\
\”, 25
—.
”
• 1. civitas capitals
4. less important small towns
7. cantonal capital territories
Figure 6.3. Schematic model of smaU toum distribution
in Roman Britain (after Hodder).
85
18
•
lund’s model (Gotlund 1956), as utilised by Hodder, which suggested that
secondary centres had developed at the intersection of two central place
territories, where commercial competition was weakest. However Marshall
argued that normally lower tiers of settlement hierarchy should come into
existence before central places. Higher order settlements should develop as
a result of stimulation from the preceding tier, not vica versa (Marshall
1964). In studies of Roman Britain this tendency has been encouraged by
the obvious desire to work from known to unknown; from the better 86 S. CLARKE
understood civitas capitals, to smaller urban sites, not all of which have
been identified and which are generally much less extensively excavated.
A second and more fundamental objection to the application of
Christaller’s model to Roman Britain was its assumption that all central
place functions were concentrated in a single location. The principle of
minimisation of effort by agglomeration of functions at a single site cannot
be taken for granted. The large numbers of major religious central places
(for example Chedworth and Lydney Park) and production centres (for
example the highly dispersed Oxfordshire pottery industry) located within
the country side strongly suggest that centrifugal as well as centripetal
forces were in existence. Furthermore there is a growing realisation that
urban settlements did not take administrative responsibility proportional to
their importance as commercial centres. Conversely industry did not
necessarily take off at the important administrative and cultural centres.
Corinium though a civitas and possibly later a provincial capital has so far
produced very little evidence for industrial production. Sjoberg has noted
that societies’ administrative elite and merchant and artisan classes were
drawn to the same locations by their mutual interdependence. Merchants
and artisans needed a market for their goods while the aristocracy needed
the material symbols of power and to keep a tight grip on what was an
alternative power base, which might have become a threat to their traditional position (Sjoberg 1960). However there is some reason to believe
that these elements in society repelled as well as attracted each other.
Although this is a factor which has frequently been overlooked it should
come as no shock to the student of urbanism. It has long been noted that
towns, while acting as centres dedicated to the perpetuation of the ruling
class and their value system (Sjoberg 1960) on one hand, also acted as a
source of innovation, social change and modernity on the other. Of course
what one element of society saw as progress could be seen by another as
social pathology, moral disorder and destruction of the community
(Holton 1986: I). Although the landed elite needed the services of merchants and artisans they might have wanted to limit their contact with such
a polluting influence. Artisans and merchants might similarly have withdrawn from those elements of society which sought to depreciate manual
and commercial activity as degrading. In short the models proposed by
Chris taller and used by Hodder were two simplistic. The relationship
between settlements cannot be treated as a single dominant force. It must
be broken down into its constituent parts: political, military, ideological
and economic. LoCATlONAL MODElS AND SMALL TOWNS
• Acts as a central place in the K-3, 4 and 7 systems
• Acts as a central place in the K-3 and 7 systems
87
• Acts as a central place in the K =4 and 7 systems
, Acts as a central place in the K- 3 and 4 systems
note that only sites that act as central places to more than one system have been plotted
Figure 6.4. Losch ‘5 modifuation to central place theory.
LOSCH’S MODIFlCATION —-
Chris taller’s model assumed fIxed-K hierarchies in which the relationship
between settlements at one level could also be applied to higher levels.
Christaller himself noted that under ideal conditions administrative, trans-88 S. CLARKE
• Acts as a central place In the K – 3, 4 and 7 systems
• Acts as a central place In the K – 4 and 7 systems
• Acts as a central place in the K – 3 and 7 systems
• Acts as a central place In the K – 3 and 4 systems
Rich sector
D Poor sector
Figure 6.5. Rich and poor sectors in the Loschian landscape.
port and market central place systems had different requirements. A modification to the model was proposed by Losch in which the requirement that
central place functions be concentrated in a single location was removed
allowing the K”‘3, 4, 7 and higher order systems to diverge in their selection of central place locations (Losch 1954) (Fig. 6.4).
If the different order hierarchies are rotated about a common hub until
the maximum number of central places overlie each other a predictable
pattern emerges. Figure 4 demonstrates the principle with just K”‘3, 4 and
7. The Loschian landscape created by the overlaying of many hierarchies
(the nine simplest K systems) with a city centre at the hub, was one of
twelve alternating rich and poor sectors (Haggett 1965: 122, 124, figure
5.9) in other words, sectors which contained many central places and those
with relatively few (see Figure 6.5).
The theory has been used to explain rich and poor city sectors in
modem cities, including Chicago (Haggett 1983: 391), but its relevance
here is in explaining the distribution of minor urban sites around the
colonia and civitas capitals of Roman Britain. For the sake of simplicity the
landscape has not been built up past the K”‘7 hierarchy. As a result a
pattern of contiguous hexagons has been created. This consists of a single
settlement which operated as a central place to all three hierarchy systems
and eighteen settlements which possessed various combinations of two
central place functions (see Figure 6.6).
VARIATIONS IN THE SMALL TOWNS —-
In the Roman period these ‘two function’ central places might have been
represented by the so called small towns. This settlement class was perhaps LoCATIONAL MODELS AND SMALL TOWNS 89
•
•
• Acts as a central place in the K-3, 4 and 7 systems • Acts as a central place in the and 7 systems
• Acts as a central place in the K=3 and 7 systems , Acts as a central place in the K=3 and 4 systems
– Boundary of the repeating group of settlements (not a central place territory)
Figure 6.6. Repeating pattern of multiple function central places.
the most heterogeneous that has been applied to Roman Britain. Some
possessed populations that were as large as those in the smaller colonia and
civitas capitals, defensive circuits and large scale industry. The prime
example of such a site was Water Newton. Others were without defences
and would be better thought of as villages or ‘local centres’ as Hingley
(1989) has described them. The Losch modification to Christaller’s central
place model may therefore be useful, not in its predicative ability, which
Hodder thought so helpful (Hodder 1972: 887), but in helping to explain
the variety of forms which urbanism could take.
THE EMERGENCE OF CIVITAS CAPITALS —-
It is important to stress that the settlement which is central to the hexagon
in figure 6.6 is not a higher order central place serving the surrounding 90 s. CLARKE
settlements. However it is unique amongst the settlements of its hexagon in
bringing all the central place functions together to one location. With this
advantage that site might be the obvious choice as central place should a
higher tier develop in the settlement hierarchy. This might explain why
some Iron Age central places develop into urban centres in the Roman
period, while others do not. In the Dobunni region there are at least four
sites which have some claim to be oppida, but only one of these, Bagendon
seems to have become fully urban, being succeeded by Cirencester a short
distance away. The other sites were not wholly abandoned as central places
however. Minchinhampton oppidum was succeeded by the major villa at
nearby Woodchester (Clarke 1982), while another oppidum, Grim’s Ditch
became the location of a group of major villas. Such sites are frequently
overlooked as central places even though Woodchester in particular is built
on a scale similar to that of the largest public buildings. Villas of this order
were certainly the homes of important ‘central people’ and as such occupy
positions at the centre of extensive patronage systems. Similarly the large
hillfort or oppidum at Salmonsbury (Bourton-on-the-Water) took on some
but not all major central place roles in the Roman period. It became a
communications node and important commercial centre, but failed to
develop an important cultural or administrative role, although it may have
been the site of a mansio.
CONCLUSION —-
In conclusion it can be said that a civitas region closely resembled a network of central places on the model of Losch rather than Chris taller. The
city represented the hub of the system, upon which all central place fields,
from the highest to the lowest order focused. This created a fully urban site
with a large non-agricultural population, planned form, defences and legal
autonomy. The settlements of the civitas hinterland however presented a
contrasting picture. Central place fields were out of phase due to the variations in their sizes. A number of central place functions could gather
together in a single location by chance, but never with the frequency that
they did at the systems hub, the civitas capital. This led to the creation of a
range of rural and only semi-urban centres, some of which are commonly
called small towns. Losch’s work has often been overlooked as too complex
a model to transfer from the uniform plain to the highly variable landscape
of the real world. However I am not suggesting that his model can explain
the distribution of central places so much as their character. Civitas capitals LoCATIONAL MODELS AND SMALL TOWNS 91
were all pretty much of a much, their form easily described. Characterising
small towns has proved much more difficult, requiring almost as many
categories as known examples. I believe Losch’s model helps us to understand this diversity and explains why there can never be a satisfactory
definition for the minimum requirements of an urban site.
References
Baskin, C. W. 1966. Central Places in Southern Germany. New Jersey.
Chris taller, W. 1935. Die Zentral Orte in Suddeutschlarui. Jena.
Clarke, Giles 1982. The Roman Villa at Woodchester. Britannia 13:197-228.
Haggett, P. 1983. Geography: a Modern Synthesis. 3rd edition. Cambridge.
Godlund, S. 1956. The Function and Growth of Bus Traffu within the Spheres of
Urban Influence.
Haggett, P. 1965. Locational Analysis in Human Geography. London.
Hingley, Richard 1989. Rural Settlement in Roman Britain. London: Seaby.
Hodder, Ian R. 1972. Locational Models and the Study of Romano-British
Settlement. In David L. Clarke (ed.), Models in Archaeology, 887-909.
London.
Holton, H.J. 1986. Cities, Capitalism and Civilization. Boston.
Losch, A. 1954. The Economics of Location. New Haven, Connecticut: Yale
University Press.
Marshall, J. V. 1964. Model and Reality in Central Place Studies. Professional
Geographer 16:5-8.
Sjoberg, G. 1960. The Pre-Industrial City. New York.