Download file in PDF format: Prologue to a Study of Roman Urban Form (pp. 92–104)
PROLOGUE TO A STUDY OF ROMAN URBAN FORM
Simon P. Ellis
The archetypal Roman town was not the kind of urban form to which we
could easily relate today. Its monotonous rectangular street grid seems unfocused and cold, like the layout of many modern American cities. This
paper is intended as a first step in applying the ideas of modern town
planning to Roman cities. By studying the Roman city with the eyes of
those who seek to create urban fonns it may be possible to address a question that the Roman archaeologists have increasingly posed themselves in
moments of self-doubt. Was the Roman town a functioning community, or
was it an artificial product born out of the Roman mania for organising
In considering the modern town the planner looks for certain facilities.
These can be classified as buildings for community use, shopping centers,
housing, industry, and transport infrastructure. I see no reason why we
should not look at Roman towns the same way. Some of these topics may
seem inappropriate in studying Roman towns. On the other hand we all
must approach history with our modern preconceptions, and in adopting
an approach based on modern town planning my prejudices at least may
be more explicit.
Since the theories of modern town planning have hardly ever been used
in archaeology there are an enormous number of lines of enquiry to
pursue. Moreover to fully resolve many of the points I have raised would
require an exhaustive consideration of ancient texts. This paper can thus
only be seen as a prologue to the study of Roman urban form.
First I wish to explore the street grid. Roman cities can be divided into ROMAN URBAN FORM 93
two groups of fonns; the ‘planned’ and the ‘unplanned’. By ‘planned’ I do
not mean those that were pre-meditated, but rather those whose urban
design was made to follow a specific regular urban design. ‘Planned’ cities
range from new settlements like the relatively small Republican colonies of
northern Italy, to a large capital city like Carthage, which continued to expand using a regular street grid of equal size insulae from the first century
BC to the fifth century AD. ‘Unplanned’ cities, may have expanded following general principals of development laid down by the authorities, but
they do not have such a regular urban design. They tend to be pre-Roman
settlements, or those that grew up spontaneously at river crossings, road
junctions etc. In these the earlier settlement controls the fonn of the later
one, and the significance of a road or river crossing dominates the future
pattern of growth.
This distinction between ‘planned and unplanned’ is preferable to the
traditional assumption. This sees Roman urbanism as the Roman Republican colony and its strictly rectangular street grid. Following the republican
period it is then assumed that ‘true orthogonal planning dwindled as
erratic street patterns appeared ever more frequently’ (Macdonald 1986:
25). This theory does not account for cities such as Carthage where new
suburbs continued to follow the Augustan street grid and insula until at
least the early fifth century AD (fig. 7.1). Nor does it account for the continual construction of new formally planned towns throughout the Roman
period, including for example Caricin Grad in the sixth century AD.
The ideal Roman ‘planned’ town does not seem to have been designed
for ease of ‘navigation’. Whilst a rectangular street grid does lend itself to
easy directions of the ‘four blocks forward and one to the right’ type, it
does not provide direct routes from point A to point B, as anyone finding
their way around a North American city knows. Although it was theoretically easy to find one’s way to the forum, or capitol, by following the Decumanus Maximus, or Cardo Maximus, traveling to any other point in a regular Roman city was difficult. Macdonald (1986: 32-33) considered the
streets of a Roman city as ‘connective architecture’ or ‘continuous conduits’, in which the architecture of the street facades provided ‘few abrupt
changes of fonnal ambiance once the main gate has been passed’.
A number of planning theorists have stressed the importance of street
design in finding one’s way through a city. In the 1950s Lynch (1960) studied the mental maps of local inhabitants. He found that there were five
types of significant feature – landmarks, nodes (major junctions of confluences of traffic), paths (major roads), edges (the borders of familiar street 94 S. P.ELUS
/…..-r-r- r\ ( J t Ie
.. ” , ….. , ………
” ‘.: Ports ‘:.
“- .:.” -:’0 .’.
“- _. I… ..
— – \ . …..
• • ” 0
:. ‘:) : …
Figure 7.1. Carthage.
facades), and districts (neighbourhoods with a very strong character).
When locals were asked for directions these features were mentioned
rather than other less recognisable elements of the city architecture. The
-ROMAN URBAN FoRt\f 95
degree to which a city was made up of these distinctive architectural elements made up its ‘legibility’. A ‘legible’ city was one which had a strong
distinctive character, to which its inhabitants could easily relate.
The description by Lynch (1960: 61) of the rectangular street grid of Los
Angeles could be applied to many Roman towns:
Almost every subject [questioned] could easily put down some
twenty major paths in correct relation to each other. At the same
time, this very regularity made it difficult for them to distinguish
one path from another.
Recent work by Hillier (Hillier and Hanson 1984) has considered the way
that streets, or ‘paths’ in Lynch’s terms, penetrate districts. Hillier noted
that legibility should be hierarchical. That it should be possible to distinguish a main through road, from a side road, and from a residential street.
There was thus a hierarchy of roads. If it was not possible to distinguish
between a main road and a side road then traffic tended to get lost in the
side streets, or through traffic could end up in residential neighbourhoods.
Main roads should be distinguishable by factors such as width, or lack of
pedestrian access. In residential neighbourhoods the homes should open
onto smaller streets, which should be narrower or winding to emphasise
that they were not designed for traffic. Hence in modern estates red brick
is used instead of tarmac, and flower boxes are placed in the road to slow
cars down after the German style of ‘traffic calming’.
In these circumstances it would have been easier to find the way across a
small ancient town than a large modern city. On the other hand ancient
Carthage consisted of some 200 regular insulae, and must have been confusing. Later in the sixth to seventh centuries AD many streets became
blocked and impassable (fig. 7.1). Though this may have caused traffic
problems it may have increased legibility, since only a few streets were now
major cross-town routes.
Hillier and Hanson (1984) also introduced the concept of ‘permeability’,
which indicates the degree to which streets enter an estate. They suggested
that it should be difficult enough to find one’s way into an estate, so discouraging intruders yet not too difficult for the residents to find their way
out. Hence a modern estate has a complex network of access roads, but
often includes some pedestrian short cuts to the shops.
Hillier (Hillier and Hanson 1984: 21-22) would regard the Roman town
plan as one in which the global needs of society dominate the local needs of
the settlement. ‘The global system is defined only by the relations between 96 S.P.ELUS
o n lr
Figure 7.2. Luna (after Frova).
the major ceremonial buildings linked as they are by “causeways”.’ The
local system is one like that of a medieval town, in which intricate street
networks give full play to street life, and neighbourhood formation.
The relative uniformity of Roman street grids and frontages, discouraged
‘legibility’, and the formation of neighbourhoods. In Hillier’s terms ‘perme-ROMAN URBAN FORM 97
ability’ was too great since it was easy to travel right across the city on most
roads. The Decumanus or Cardo Maximus might be somewhat larger than
other streets, but otherwise there was little to distinguish the frontages, or
colonnades. Instead the Roman would have had to use distinctive landmarks. In other words the principle was one of ‘turn left after the Baths of
Caracalla’ rather than ‘go down the street with all the bay windows’.
The lack of differentiation in the width of streets also suggests that the
Romans were not much concerned with traffic flow. It is quite common for
a wide main street of a Roman town to have steps across it. Provincial
settlements grew up around clearly distinguishable major through routes
between major towns, but in a planned town like the second century BC
colony of Luna (Fig. 7.2), the Via Aemilia, one of the grandest routes of the
empire,just looks like a rather wide street when it passes through the town.
There is no indication of measures to deal with any resulting traffic congestion.
When spontaneous ‘unplanned’ settlements, such as many of the small
rural towns of Roman Britain, grew up the settlement orientated itself
towards the major road which is as a consequence clearly identifiable.
Houses open onto the main thoroughfare: side streets are narrow and of
lesser importance. These points can be illustrated by the plans of two
important cities of the Roman period Ephesus and Dougga. Though both
were founded in pre-Roman times they saw a considerable amount of rebuilding in Roman times.
EPHESUS (FIG. 7.3) —
Ephesus was a great classical city with a long history. Although its plan was
partly determined by two hills on the east of the site, it can in many ways
be seen as a prototypical example of Greek and Roman urban design.
Foss (1978: 56-57) notes the difference between the two main late
antique streets of Ephesus, the Arcadiane and the Embolos. The Arcadiane
ran from the harbour to the centre of the town, at the theatre. At lO.7m in
width plus colonnades on either side it was the grandest street in the town,
yet to Foss it is ‘cold and lifeless’ compared to the Embolos. The Embolos
was a narrower street that curved uphill from the theatre to the municipal
buildings. It was lined with large numbers of statues. There were numerous
late antique graffiti on the facades and the pavement.
Both streets were fronted by shops but as Foss concludes the Arcadiane
‘was a monument and vital artery, but not a centre of urban life’. The nar-98
o 100 JOO ..
flIff 8ATHS OR tU;,;l PAL AC E
Figure 7.3. Plan of Ephesus (after Foss).
rower width and steep slope of the Embolos meant that it was less suited to
wheeled, or four legged traffic. At the same time its position outside the
municipal buildings lined with statues of past politicians made it an ideal
centre for political discussion and abuse. In the case of the Arcadiane there ROMAN URBAN FORM 99
seems to be a contradiction between its monumental aspect and its use as a
‘main artery’. I am inclined to think that it was the monumental aspect that
was most important, and that it was designed for ceremonial processions of
dignitaries who had arrived by sea. Though its great width made it eminently suitable for wheeled traffic I think its ceremonial function was more
important than its commercial one.
In Hillier’s terms the Arcadiane was the ’causeway’ built by the central
administration to link the main public buildings. Although the Embolos
was lined with statues of civic dignities it had a great deal of ‘character’.
This character was created by its lively street life, its narrow width and
winding route, the statues along it, and a great variety of building facades
(e.g. Temple of Hadrian). To Lynch it would be a very ‘legible’ street. The
facades would create two clear edges to the districts on either side. To the
west there were civic buildings set against the steep hill of the acropolis. To
the east there were rich houses, set attractively on a gentler slope overlooking the street.
DOUGGA (FIG. 7.4) —-
By contrast to Ephesus, there is very little of the ‘planned’ in Dougga
(Poinssot 1983). For much of its history Dougga remained part of the territory, or perlica of Carthage. There was a community of Roman citizens in
the town from 46 BC, but the town did not gain full municipal status until
Nevertheless it would be wrong to imagine that the city grew without
guiding principles. It is commonly stated that the plan of Dougga derives
from that of the earlier Numidian settlement. In truth there is little clear
archaeological evidence for the plan of the Numidian settlement. Dougga
is a typical hill top town and like such settlements allover the world, of all
periods, its streets simply follow the topography of the ground.
Most of the public buildings in Dougga date from the later second
century AD. Many (like the theatre, the Temple of Caelestis, and the
Temple of Saturn) are sited to take advantage of views across the rich agricultural valley that provided the city’s wealth. There were virtually no
straight streets anywhere in the town. At first sight this would seem to
make it very hard to find one’s way across town, but I would suggest the
opposite is true. Public buildings are spread evenly across the town, and
wherever a pedestrian was there would always be one nearby. The public
buildings would be useful landmarks in directing people. Their various 100 S.P.ELLlS
< ……………………. .
Figure 7.4. Plan of Dougga, Tunisia (after Poinssot). ROMAN URBAN FORM 101
facades, as at the Embolos in Ephesus, would combine with the winding
routes of the streets to create a strong neighbourhood identity.
Dougga was thus a very ‘legible’, easily understood city. Town planners
have found that people today generally prefer a town with strong character
and ‘legibility’. Given the choice between living in a planned Roman colony
and Dougga I expect most people today would choose the latter.
In the law of 45 BC, the Lex Heraclea (GIL I: 593) Caesar decreed that
after the tenth hour in the evening, only vehicles were allowed to travel the
streets of Rome. The streets, he said, were too crowded with pedestrians in
the day time, and only in this way could passage be secured for wheeled
traffic (Carcopino 1940: 62-63). Nevertheless the law was not extended to
the rest of the empire until the second century AD.
The impression is that Roman towns were not designed for wheeled or
hoofed traffic. At both Ephesus and Dougga it is possible to find many
streets which have steps across them or which are very narrow. Ease of
access never appears to have been considered when locating important
buildings. At Ephesus, it was the social context which determined the form
of the streets rather than the need for access. If a visiting dignitary arrived
at the city by sea expecting to give an important speech at the civic assembly, his triumphant procession up the Arcadiane would have been
followed by a more ignominious climb up the narrower, steeper, bustling
The Roman town was inturned and closed like the Roman house, which
presented a blank exterior to the outside world (see further below). If, as I
have been implying, it was often difficult to find one’s way round a
‘planned’ Roman town then it is legitimate to ask who the towns were designed for. Did the Roman towns function more for residents or for outsiders and regular visitors? Residents would have been more familiar with
the landmarks that would have been used for finding one’s way according
to Lynch’s theory. Visitors may have found it very difficult, and the whole
experience may have seemed rather daunting if they came from the
countryside and had not seen cities before. Simon Clarke has shown in this
volume how the location of Romano-British towns can be explained by an
aim to obtain the best possible location for an administrative, or market
centre. The arguments in his paper and mine bring out the use of towns to
‘control’ a society. This does not necessarily mean military or political
control. It may rather be control over the movement of goods and people,
guiding them along certain routes, which can be observed or from which
they can be directed to certain key locations. 102 S.P.ELUS
The tendency in the western provinces to place all the public buildings in
a single central location, could be seen as an attempt to create legibility of
the town plan. It can also be seen as a way of making sure that most visitors
to the town are directed towards the political and economic centre of the
settlement. Two major factors can be found to influence the location of
major public buildings, politics, and available land, but not access.
In modern town planning preparing for major new developments, involves what is known as the ‘land assemblage’ problem. This involves buying up the plots of many smaller owners to assemble the site for a very
large building. As more and more small owners sell up the developer
becomes more and more financially committed to the project, because of
the purchase of so much land, and pragmatically because he is increasingly
the owner of a large part of the site. The last pieces of land to be bought,
or ‘assembled’ are known as the ‘ransom strip’. This is because the developer is so committed that he can be held to ransom by the last of the
small owners who can demand an exorbitant price. If the developer does
not pay, he is left owning three quarters of the land, but unable to build.
At Carthage it is notable that all the major public buildings of the second
century AD – theatre, baths, circus, amphitheatre, lie outside the kernel of
the ancient city. Recent excavations suggest that these major buildings were
built at the edge of the urban area of that time. It is not possible to tell
whether this was an attempt to avoid site assemblage problems, but it does
indicate a desire, at the very least, to avoid the political problems of demolishing the property of, and dispossessing, a large number of citizens.
In Rome a more radical solution to the site assemblage problem, as suggested by ancient authors, was to start a fire in the area (Plutarch Crassus
2.5; Suetonius Nero 16). Though there could be large open areas in Roman
towns these would still have been under private ownership and site assemblage problems would have resulted. Arson is said to have been a successful strategy as it made the price of the land crash.
The public baths is perhaps the institution of the Roman world which
most closely resembled a modern community centre, and thus had the
potential to ‘control’ the lives of the townspeople. In the Mediterranean
there were normally several distributed throughout the town. Thus the
Baths can be said to be a neighbourhood facility. On the other hand there
were not really enough baths to cover every neighbourhood. Baths were
normally financed by private munificence rather than by systematic state
provision, and were maintained by an entrance charge (Juvenal Satires II).
The impression is that people had their favourite baths and would travel ROMAN URBAN FORM 103
across town to VISit it. Perhaps the best analogy is the modern leisure
centre, of which their can be two or three in larger towns located in ‘main
centres’. If we accept this view then it seems inescapable that in Roman
Britain the systematic construction of public baths as part of the forum
complex was a deliberate attempt to encourage Roman behavior in the
natives, as Agricola is said to have intended (Tacitus Agricola 21). The
Romans were deliberately using urban design to influence local people.
Some Roman cities had formal markets. Sometimes these were stalls in
the forum. In other places for example at Wroxeter in Britain, there were
purpose-built market buildings. However the majority of shops seem to
have been spread throughout the town. In large Roman cities it is common
to see shops flanking the entrances to the big houses. The shopkeepers
were presumably tenants, and had the added advantage of providing more
security to the big house.
Shop-keepers also found themselves under obligation to the community.
In late antique Antioch shopkeepers were obliged to provide lamps to light
the porticoes outside their premises (Amm Marc 14.1.9). Street lighting
was also provided in late antique Alexandria and Constantinople, but not
in early imperial Rome (Foss 1979: 56-57, Bean 1979: 140). In RomanoBritish towns, shop-keepers and traders often lived in the traditional ‘strip
houses’ with a shop on the street, living quarters behind, and a yard, or
workshop to the rear again. There is much debate amongst classical historians, as to whether trades were concentrated in particular streets or neighbourhoods (MacMullen 1974: 66-70). The evidence does not seem conclusive. Suffice to say there was no form of ‘shopping centre’, or anything
to compare with the medieval ‘souq’.
The Roman house has more in common with the modern oriental types
of house than with those of present day Europe. The Roman house was inturned as regards both its architecture and its life. It did not present many
openings onto the street. There was often an entrance corridor, and reception suite, but these formed a part of the house. This is different to the
‘intermediate zone’ between house and street, that is formed by the
modern front garden, or the verandah. This intermediate zone, studied by
Classie (1975: 137), and other analysts of urban form, creates an area,
which is half in the house, and half in the street. Residents can feel they are
still at home, passers-by can feel they are not intruding, and interaction can
take place freely.
The shops and stalls in the colonnades of streets such as the Embolus at
Ephesus, appear to create such an intermediate zone, but this is deceptive. 104 S.P.ELUS
Most interaction with the Roman householder took place inside the house
in an area clearly under the control of the owner (Wallace-Hadrill 1988;
Ellis 1991). The fact that much of Roman political and economic negotiation took place behind closed doors helpd to explain the conspiratorial
nature of Roman politics. Though many Roman cities, especially those in
the Mediterranean clearly had a sense of ‘community’, at least as expressed
in the munificence of their leading citizens, the closed nature of the
Roman house did not allow for open socialising in a local neighbourhood.
The conclusion from this brief overview of planning in Roman towns is
that, in terms of the overall plan, and in terms of the location of the main
public buildings within that plan, the Roman town was an environment
which exerted a large amount of control over its inhabitants movements.
There is also evidence that control was exerted through particular types of
buildings. Social pressure was placed on the inhabitants in the public baths,
and the residences of the rich were used by their owners to enhance their
power over others.
Bean, G. 1979. Aegean Turkey. London: Benn.
Carcopino,]. 1940. Daily Life in Ancient Rome. Harmondsworth: Penguin.
Ellis, Simon P. 1991. Power, Architecture and Decor: How the Late Roman
Aristocrat Appeared to His Guests. In Gaza (ed.), Roman Art in the
Private Sphere, 117-134. Ann Arbor: University of Michigan.
Foss, C. 1979. Ephesus after Antiquity. Cambridge: Cambridge University
GIassie, Henry 1975. Folk Housing in Middle Virginia. University of
Hillier, Bill and Julienne Hanson 1984. The Social Logic of Space. Cambridge:
Cambridge University Press.
Lynch, K. 1960. The Image of a City. Cambridge, Mass. Massachusetts Institute of Technology Press.
Macdonald, W. 1986. The Arcitecture of the Roman Empire II: an Urban
Appraisal. New Haven: Yale University Press.
MacMullen, Ramsay 1974. Roman Social Relations 50 BC to AD 284. New
Haven: Yale University Press.
Poinssot, C. 1983. Les mines de Dougga. Tunis: I.N.A.A.
Wallace-HadriIl, Andrew 1988. The Social Structure of the Roman House.
Proceedings of the British School at Rome 56:43-97.