Skip to main content
SearchLoginLogin or Signup

Chapter 6: Roman Cities

Published onApr 22, 2021
Chapter 6: Roman Cities

Hippodamean Type

The Romans used more than one type of city plan, and it is thus not possible to contrast a Greek plan with a Roman one. We will consider first the type characterized by a rectangular grid derived directly from the Hippodamean city, Norba is the most ancient example of this type.


At the turn of the century Norba was thoroughly studied by L. Savignoni and R. Mengarelli.1 Their excavations along the perimeter walls and among the temples established these as dating from the fourth century B.C. The plan of the city which arose from this work clearly shows the location of the acropolis and the walls but disregards the rest. The perfect orthogonality of the system can be seen from the aerial photograph (Fig. 37).

Norba was built up of terraces buttressed by a system of walls that were almost always parallel and perpendicular to each other. The city is an interesting example of an urbanism that exploits the difficulties of the terrain to obtain striking results. From the lower portions of the city one viewed the rising terraces, culminating with the Great Acropolis at the central peak. This was a vast terrace perfectly incorporated in the general scheme. On the axis of this terrace, in the center of the city, is a larger, more recent building with a cistern. To the south of the acropolis some blocks still remain, one actus wide, separated by streets four meters wide. The task of future excavations, aside from ascertaining the system of the terraces, would be to explore the road pattern to determine the nature of the grid system. It is possible that a longitudinal road exists at right angles to the long axis of the blocks.

It is impossible to separate the chronology of the city plan from that of the walls and acropolis. Excluding the foundation of the colony in 499 B.C., such a complete rebuilding of the city could have occurred after the incursions of the Privernati in 342 B.C. (Livy VII, 42, 8) or during or after the Privernate war when the Monti Lepini were occupied.2

Alba Fucens

Notwithstanding the difficulties of planning and building on mountainous terrain, Alba Fucens3 is formed on an orthogonal pattern of streets (Fig. 38). This became apparent after recent excavations by a Belgian party. Founded as a colony in 303 B.C., it follows wide longitudinal axes cut by minor transverse streets4 that delimit blocks one actus wide and more than double that figure in length. The blocks are arranged with their long side parallel to the logitudinal axis (decumanus). This is a new expression of the Hippodamean plan, patterned by scamna and not per strigas (if the term scamna may be borrowed from the vocabulary of the military encampments and agrarian surveys). This system must have been preferred for its adaptability to the terrain. No precedents are known to exist, though later cities followed similar practice.

Figure 37 Aerial view of Norba (Istituto Geografico Mil.).

Figure 38 Alba Fucens (Mertens).


The work which the American Academy in Rome5 has carried out at Cosa in the last few years has contributed greatly to the study of regular planning systems. Cosa, a colony founded in 273 B.C. (Fig. 39), presents a strong orthogonal pattern notwithstanding the difficulties of the terrain. Thus Norba, Alba, and Cosa contradict the theory that regular patterning could be applied only to the cities in the plains.6 It is difficult to single out a main cardine and decumanus at Cosa, although the road system7 subdivides the city into rectangular blocks in the center of the city, 32.50 to 37 meters wide (1 actus) and usually more than twice as long (82 meters between K and O streets).


Detailed research could ascertain whether Benevento (268 B.C.) was also patterned by elongated rectangles such as we see in the modern plan, one actus wide and uniformly arranged on a north-south axis.

Faleri Novi

At Faleri Novi, too (apparently somewhat later than 241 B.C.) there is a regular pattern, as is apparent in Lugli’s reconstruction.8 Rectangular blocks can perhaps be discerned, one actus wide and placed on the principal axis, which is oriented east-west.


On the other hand, Haverfield is in error when he attributes to the Roman age that section of Modena which has very long and narrow blocks.9 The ancient nucleus is found instead to the east of this sector, as is recognized by both Lehmann–Hartleben10 and Corradi-Cervi.11

Figure 39 Cosa (Brown).

We should remember that during this same era the Italian cities of Greek origin were completing their master plans, following the Hippodamean pattern (Pompeii, Soluntum, Agrigento). Even second-century Rome had an example of strigae. A block excavated in Via Nazionale near Villa Aldobrandini was 35 meters wide and 58.30 meters (about 200 feet) long.12

Cities Based on Intersection of Central Axes


Of exceptional importance in the context of Roman planning is the city of early Ostia, the plan of which is known today through special studies and through the publication of the excellent first volume reporting the excavations (Fig. 40).13

Ostia’s origin has been attributed to the last decade of the fourth century B.C. for archeological reasons (architectural terra cottas) and for historical congruity (the need to control the sea after the fall of Anzio).14 The city follows a regular plan (193.94 by 125.70 meters), divided into four equal sectors by two orthogonal axes which intersect at the center of the city. These axes can be termed cardine and decumanus15 by analogy to the techniques of surveying. The orientation is dictated by the axis of the river.

Thus begins a city type characteristically Roman, followed and developed through many colonies. This type is defined as castrum in comparison with the fortress cities of the Imperial era. We speak of the castrum of Ostia. However, the term is not exact if we consider the plan of contemporary military encampments, which have an entirely different scheme of internal division. Frontinus excludes any imitation of the encampments, Strat. IV, 1, 14, by his evidence that camps were not employed before the time of Pyrrhus.16


The plan of the oldest part of Minturno (Fig. 41) is very similar to that of Ostia, even in dimensions.17 It is a rectangle of about 182 by 155 meters. (Excavations of the walls on the west and the beginnings of those on the adjacent sides testify to these dimensions.) The layout was oriented not exactly to the north but on the axis of the Liri; the Via Appia cut the city in half, constituting the decumanus, and we can presume that there was a cardine intersecting at right angles at the center. This part of Minturno is considered the ancient Ausonian city, partly because of the polygonal work of the walls. The walls in opus quadratum extending out to the western sector, along the Forum, are attributed to the Roman colony of 295 B.C. The principal buildings of the Forum, the Capitolium and portico, are no earlier than the second century. Coins found below the portico date to 200–190 B.C. but the cobblestone canals and a temple identified with the aedes Iovis are considered to be earlier. According to Livy, the temple must have been standing in 207 and 191 B.C.18 He says that it was struck by lightning in those years and destroyed in 191. Architectural terracottas place the origin of the temple at 295 B.C.

Figure 40 Ostia (Scavi di Ostia).

Figure 41 Minturno (Aurigemma–De Santis).

A precise chronology of Minturno is lacking, however. The Ostia-type plan of the city may date from the colony of citizens in 295 B.C.; it is similar in size to the colonies at Ostia, Pyrgi, and Pozzuoli. The polygonal work of the walls may date from this time also, although the westward extension of these walls belongs to construction in the following years of the third century.


Pyrgi, a Roman colony established upon a former Etruscan city, is the third example of a town built on intersecting central axes. The date of its founding is unknown. It is first mentioned in 191 B.C. by Livy (XXXVI, 3, 6). Salmon19 maintains that the colony was first founded in 194 B.C., a year in which eight maritime colonies were established. Tibiletti20 thinks, however, that Pyrgi might date from before 194, and now Salmon too proposes a date preceding the first Punic war.21 Its geographical location and the polygonal work of the walls, comparable to that of Cosa, would suggest a very ancient settlement.

Unfortunately the plan of Pyrgi has been studied little since the early work of Canina and Dennis.22 An aerial photograph (Fig. 42) permits a simple reconstruction. The area covered by Pyrgi is larger than that of the preceding examples. The city walls, visible throughout, are well preserved, especially on the side of the castle. The southeast wall was short because it is broken by the coastline. Near this break there is a gate, probably open to a road from the center of the city. The missing wall opposite this would also have had a gate. It is possible to reconstruct the axis corresponding to a break in the northeast wall where a gate would have been (Fig. 43).

Figure 42 Pyrgi, aerial view (British School of Rome). British Crown rights reserved.

The street pattern of the three cities just described is not known. At Ostia there is a fragment of a street 31 meters southwest of the main cardine.

The pattern typical of the Roman city, as we know it from the later examples, is that of a regular network in which a net of streets of equal importance develops parallel to two basic axes that cross usually, but not always, at the center. The network is often enclosed by a square or rectangular perimeter of walls. These cities differ from the Greek and Hellenistic types which have a series of longitudinal axes crossed by narrow transverse streets forming rectangular blocks, and the whole enclosed by an irregular perimeter of walls.

Figure 43 Schematic plan of Pyrgi.


A tentative reconstruction of Terracina,23 a colony founded in 329 B.C., is possible from a few elements. The Via Appia becomes the decumanus, and the principal cardine probably ran to the west from the Capitolium, between Porta Nova and the Forum. A pattern of cardines 70 meters apart can be deduced from the present layout. It seems as if another decumanus ran 50 meters south of the Via Appia. Terracina maintained an irregular perimeter.


The walls of Fondi24 describe an almost square perimeter. Four principal gates correspond to the lines of the major cardine and decumanus, which are known from the plan that has survived. Minor cardines are spaced 50 to 60 meters apart. The location of the lesser decumani is uncertain. The polygonal work of the walls may be attributable to the time of the municipium with partial rights (338 B.C.), to the time when full rights were granted (188 B.C.),25 or, as Lugli thinks, about the middle of the third century B.C.26


Sena Gallica, a colony from 283 B.C., was laid out on a grid of square blocks 50 meters on a side, according to the reconstruction by Ortolani and Alfieri.27 This reconstructed plan remains highly hypothetical, however, because it corresponds but little to the existing city.


The plan of Ariminum (Rimini), a colony of 268 B.C., is known.28 It was clearly established on the cardine and decumanus, each about 9 meters wide, and had blocks 74 by 110 meters (Fig. 44).


Placentia (Piacenza), a colony founded in 218 and reconstructed in 190 B.C., is patterned on a grid of square blocks 80 meters on a side.29


The blocks at Pozzuoli appear to have similar dimensions. The plan is probably Roman in origin and possibly goes back to the colony of 194, since it is divided into squares. With some 300 colonists, this was a modest town. Though it was neighbor to several Greek cities, Pozzuoli differs substantially from their type of plan. As has been noted, Pozzuoli is usually considered a Greek city, but there are no remains to prove such a time of origin.30 The Temple of Augustus followed the same orientation.


Bologna was established as a colony in 189 B.C.31 Within its square perimeter, the blocks measure 113,66 by 80 meters. Those adjacent to the decumanus are 70 meters wide.


Major axes that meet at the city center remain in Pesaro (184 B.C.). The grid cannot be reconstructed entirely.32


Part of the plan of Aquileia (181 B.C.) is known through excavations. A grid, not always uniformly patterned, extended throughout the city. Blocks of 62 by 83 meters have been found.33


The grid pattern of Parma is reflected in the present city and has been confirmed by various findings. The major axes intersect almost at the center, and the blocks are nearly square (45 by 55 meters).34 It is not known whether this plan corresponds to the original layout of the colony of 183 B.C. or to the rebuilding by Augustus after its destruction in 43 B.C.

Figure 44 Plan of Rimini (Mansuelli). Scale about 1 : 9000.

Figure 45 Sorrento (Mingazzini–Pfister).


Sorrento, reconstructed from the present layout, is thought to be Greek by von Gerkan,35 for example, and more recently by Mingazzini and Pfister (Fig. 45),36 who think that the gates and walls are definitely Greek. The disputable point is whether the plan itself should be attributed to the Oscans, Greeks, or the Etruscans, on the basis of research on the unit of measure used for the divisions into blocks (averaging 52 by 76.50 meters).37 Nonetheless, the town walls—the only elements considered to be of Greek origin—are not placeable at an early date. Rather, according to Lugli,38 they can be assigned to the era of Silla because of the wide arches. Most probably Sorrento was a colony at that time.

Beloch considered the urban pattern to be of the same type as that of other cities in Campania, except that Sorrento was laid out per scamna instead of per strigas. (Beloch wrongly supposes this to be true of Pompeii too.) There are other differences, however. Judging by the location of the gates, there is a clear pattern of two central axes. Thus the plan fully agrees with the type of Roman plan we are now discussing, and until contrary proof is offered it must be attributed to the Roman age.


Lucca, a colony of 180 or 177 B.C. (although some maintain that it was never granted that status)39 follows an irregular grid pattern (Fig. 46). The blocks are usually wide, about 75 to 120 meters.40

Figure 46 Aerial view and reconstructive scheme of Lucca (British School of Rome). British Crown rights reserved.


The grid of Florentia (Florence) is variously dated between the second century B.C. and the era of the triumvirs. It presents astronomical orientation, a trapezoidal perimeter, an almost central intersection of two major axes, and blocks of varying width, usually square and 60 meters on a side.41


The town of Alife retains a nearly square perimeter of walls (540 by 405 meters) with rounded corners (Fig. 47).42 The four gates opening near the center of the sides correspond to the major axes which are still part of the present city. A series of secondary streets has also been preserved. These form a network of rectangles 50 meters wide and of varying lengths. Perhaps the plan belongs to the colony of Silla, which Mommsen believes to have existed,43 since the walls in opus incertum date from that period.44


Ascoli Piceno appears to have been divided into square blocks 75 meters on a side. I would say, therefore, that it was probably a Roman city founded after the Social War45 and not, as Lehmann–Hartleben supposes, a Picene city.46 He reaches this conclusion on the basis of its relation to the regular plan of Belmonte of the sixth century B.C., which in turn he believes to have been under Etruscan influence.

Como and Pavia

Square blocks from 70 to 80 meters on a side and the central intersection of major axes form the urban pattern of Como47 (possibly dating from the days of Caesar) and of Pavia48 (either the first century B.C. or A.D.).


Verona, built during the first century B.C., perhaps between 75 and 50, is laid out on a uniform grid of almost-square blocks, 75 to 80 meters on a side.49


Rectangular blocks are found at Concordia, a colony probably founded during the triumvirate.50

Figure 47 Alife.


The grid plan of Libarna51 dates either from the last days of the Republic or the beginning of the reign of Augustus. Its square blocks are 60 meters on a side.

Augusta Bagiennorum

A grid of varying block size is found at Augusta Bagiennorum.52 Some of the blocks measure about 80 by 100 meters.


A square perimeter, a central axis, and a plan oriented to the coastline, apparently with square blocks between 70 and 80 meters wide, characterize the Augustan colony of Fanum Fortunae.

Cities of the Imperial Period

The axial scheme with a grid of square or nearly square blocks was frequently used in the Empire. Among the better-known examples are Emona (34 B.C.), Lincoln, Autun, Trier, Silchester, Caerwent, and Timgad.53 The blocks are very wide, often more than 100 meters. Timgad is the only exception, with blocks barely 70 feet on a side. An astronomical orientation of the grid was employed frequently.54

The Axial Plan as Influenced by the Encampment Plan

Among the varieties of grid plans, a particular Roman type emerges in which the intersection of the major axes is shifted to one side; this is clearly inspired by the military encampments. Instead of the axis conventionally called the cardine, there are two parallel axes which we can call the via principalis and the via quintana, borrowing the terms of the military camps. Turin and Aosta are the best examples.


The Augustan colony55 of Turin56 is a classic example of Roman city planning (Fig. 48) both because of the excellent preservation of the layout in the modern plan and because it was one of the first examples to bring into focus the general problems of delimitation. Turin presents an almost square perimeter57 which appears to have been 669 by 720 meters or 20 actus.58 A grid of streets 5 to 8 meters wide59 defines blocks some 2 actus to a side. There are 7 streets in one direction, 8 in the other. The intersection of the major axes does not occur at the center. In fact, although the northwest-to-southeast axis (Corso Garibaldi) is in the center, the axis orthogonal to this that leads to the Porta Palatina is far to the eastern side of the city (Via dell’Arsenale—San Tommaso —Porta Palatina). This corresponds to the via principalis of the camps. It is not by chance that one of the parallel streets in the eastern sector was called via quintana.

Figure 48 Turin (Barocelli).


Aosta’s plan is very similar to that of Turin (Fig. 49). Its walls form a rectangle 724 by 572 meters. Apparently the intention was to establish a rectangle of 20 actus by 16. The 6 or 7 meters on each side which were in excess of this measure perhaps represent the thickness of the walls or of the pomerium. The pattern of the sewage system indicates a grid of seven cardines and seven decumani,60 subdividing blocks of about 70 to 80 meters on a side, as at Turin. Furthermore, two of the cardines appear particularly important in that they lead to the city gates. We can call them via principalis and via quintana, and in fact medieval documents of Aosta also refer to a street called via quintana.61

Axial Plan with Subdivision per scamna

Among the plans characterized by a network of square or nearly square blocks, a type can be distinguished in which the central division by main decumanus and cardine is accompanied by blocks whose long side runs parallel to the decumanus. These blocks are termed scamna.


Roman Carthage is the best example of this type of plan (Fig. 50). Its layout has been reconstructed by C. Saumagne and P. Davin62 from the ruins of the sewer system (besides several monuments such as the cisterns of Bordj Djedid which occupy an entire block). They were able to reconstruct a rigorous uniform grid of streets which delienate very long rectangular blocks. Calculations show the roads to have been 24 feet (7.06 meters) wide, except for the major axes which were 40 feet (11.76 meters) wide. The blocks measure 35.28 by 141.12 meters (1 by 4 actus). Thus the area enclosed is 2 square actus or 2 jugers or 1 heredium. The rectangular blocks should be considered scamna rather than strigae because the road that has been termed the cardine maximus was more likely the decumanus maximus, being parallel to the coast line.

Figure 49 Aosta

Figure 50 Carthage (Davin).

Figure 51 Zara.

That the city is Roman and not Punic is attested by the numerous Punic tombs, dating at least from the third century B.C.63 Whether this Carthage was the colony of Gracchus from 122 B.C. or belonged to the Augustan era, 35–15 B.C. remains to be decided. The latter date is generally accepted because the grid layout of the city appears to derive from the land assignment and survey of 122 B.C.


The same general scheme and dimensions found at Carthage were very likely employed in the Augustan colony of Zara (Iader) (Fig. 51); the scamna measure 1 by 4 actus (an area of 2 jugers). The use of subdivision by rectangles continues into the Imperial period. It is found in the Severian city of Leptis,64 where the blocks are unusually small, and at Sufetula,65 where the blocks measure 50 by 100 meters.

Encampment Plan

The Hippodamean plan has inspired not only the cities here discussed but also the form of the Roman military encampment. As already noted, especially in regard to the plan of Ostia, the encampment does not entirely follow the so-called castrum scheme with its central intersection of the major axes. Instead, as found in Polybius (VI, 26–42) and throughout much archeological documentation, the encampment is patterned along the via principalis (100 feet wide according to Polybius 28, 1) and the via quintana. The blocks, whose long dimension is at right angles to the major roads, are subdivided by secondary streets. So arrayed, they are termed strigae by Iginus Gromaticus (De munit. castrorum 1). Rectangles arranged parallel to the major axes are termed scamna.

Streets perpendicular to the via principalis and via quintana are of lesser importance. Aside from a basic Hippodamean layout, the plans show a tendency toward centralization in that there is a gate at the center of each short side and that the perimeter walls are square or almost square. Other than the reports of Polybius, Schulten’s work has been chiefly responsible for bringing to light a series of Republican encampments near Numantia (see Figs. 52 to 54 for the camp of the consul Fulvius Nobilior of 153 B.C., the camp of Marcellus at Castillejo in 151 B.C., and camp V in the same zone, later than 90 B.C.66).

The same scheme is found during the Imperial era, at Carnuntum (Fig. 55) and at Lambesi, for example. Also at the Castra Praetoria of Rome, dwelling units arranged per strigas have been found, each half an actus wide.67 The direction of these strigae is a decisive element in determining the principal axis and thus the orientation of the Castra Praetoria (Fig. 56). Then the porta principalis sinistra would face toward Rome.68

Another type of encampment is based on the central intersection of a cardine and decumanus, as at Ostia and other cities. Examples of this type are found in England, Germany, and Syria.69

Yet the basic plan remains similar to the type assigned to Hippodamus of Miletus. With this in mind, Polybius’s comparison between encampment and city can be better understood (VI, 31, 10):

τὸ μὲν σύμπαν σχῆμα γíνεται τῆς στρατοπεδείας τετράγωνον ἰσόπλευρον, τὰ δὲ κατὰ μέρος ἢδη τῆς τε ῥυμοτομίας ἐν αὐτῇ καὶ τῆς ἂλλης οἰκονομίας πόλει παραπλησίαν ἔχει τὴν διάθησιν. (The whole camp forms a square, and the way in which the streets are laid out and its general arrangement give it the appearance of a town.)

Here Polybius is not drawing a parallel between the square castrum and the urbs quadrata such as Ostia, as a hasty reading would suggest; only in the second part of the passage does he make a comparison between the castrum and the city (τὰ δέ…), that is the ῥυμοτομία and the οἰκονομία noted by von Gerkan.70 The comparison between Republican encampment and Hippodamean city is quite exact, especially because the subdivision per strigas and the pattern of two major east-west axes were employed in both.

Figure 52 Camp of Nobilior.

Figure 53 Camp V of Castillejo near Numantia (Schulten).

Figure 54 Camp of Marcellus near Numantia (Schulten).

Figure 55 Carnuntum, camp (Der röm. Limes in Österreich).

Figure 56 Castra Praetoria, showing the strigae (Lanciani).

According to Frontinus71 the Romans did not use encampments before the time of Pyrrhus. This would seem to rule out the consideration of Ostia, which is older than that, as an encampment plan. However, as was observed earlier in this chapter, the two types of plans are independent. In considering whether the city derived from the military camp, or vice versa, the answer is that the encampment’s form seems to have come from the Hippodamean city.

Characteristics of the Uniform Roman City

The uniform Roman city as known from the fourth century onward has assumed a large variety of forms when compared to the Greek city type. These forms are:

  1. The city derived from the Hippodamean type, subdivided per strigas (except Alba, which is patterned on scamna).

  2. The central crossing of axes scheme, with a system of parallel streets almost always forming a square or nearly-square grid.

  3. A variation of 2, based on the layout of military encampments.

  4. A system of central axes with subdivision per scamna.

These variations paralleled the different agrarian subdivisions. The first system is directly comparable to the agrarian division per strigas; the second is like the traditional centuriation. The fourth scheme corresponds to the ager per centurias et scamna adsignatus. It should be noted that the various systems of land division can be traced back just as far as the systems for patterning cities (the centuriation of Terracina in 329, the strigatio of Suessa Aurunca in 313 B.C.).

There is no overlap among the various city systems. Cosa, belonging to type 1, comes close to being of the fourth type because some of its axes appear to be more important than others. Among the cities having a central intersection of axes there is a wide variety in the proportions of the blocks. In fact there is so much variety that cities such as Sorrento and especially Alife are considered to be close to the fourth type.

A wide variety is found also in the dimensions of the city blocks. Hippodamean cities usually have blocks 1 actus wide and usually more than twice as long. In the plans with almost square blocks the most common dimensions are between 70 and 80 meters. An area of two jugers seems to have been preferred.

At times the grid followed a celestial orientation (in Florence and perhaps Lucca) but more often conditions were set by geographical configurations such as coast lines (Pyrgi, Carthage, Fano, Pesaro), rivers (Ostia, Minturno), major communications arteries (Terracina, Fondi), or mountains (Norba, Alba, Cosa). The Forum is normally found at the intersection of the major axes.

The chief interest in these regular Roman cities lies naturally in those types, such as Ostia, which were inspired by the principle of axial symmetry, developed without imitation of the encampments. As has been said earlier, this principle is not typical of Italy, and its origins are traceable to the Greeks rather than to the Etruscans and Italics. However, it was in Roman colonies from the fourth century on that axial symmetry found its most rigorous application, extending even to the perimeter of the quadrangular or square city, and it was here that it was most extensively used. Such a rich production of cities and widespread use of agrarian surveying was quite naturally the object of much theorizing by scholars of the late Republic. Yet the Roman urban planner was little moved by celestial speculation and adopted the principles of axial symmetry because they corresponded to Roman taste. For example, this symmetry appears in the atrium-type house as well as in the Imperial Forum. Furthermore, axial symmetry embodied the concept of military discipline and centralized political power, focusing the city upon a single point,72 where the magistrate exercised his authority. The same was true of the military camp. This idea of a central focus becomes more evident when the Roman plan is contrasted to the layout of the Hippodamean city, in which the uniformity of the pattern is accompanied by the concept of decentralization. This is characteristic of the Greek city, because it corresponds to the looser political plan.73

No comments here
Why not start the discussion?
Read Next