See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/31740639

Roman Builders : A Study in Architectural Process. Cambridge University

Press, 2003.

Book · January 2003

Source: OAI

CITATIONS READS

24 781

1 author:

Rabun Taylor
University of Texas at Austin
53 PUBLICATIONS   231 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Rabun Taylor on 07 April 2021.

The user has requested enhancement of the downloaded file.

Roman Builders
A Study in Architectural
Process

Rabun Taylor
Harvard University
PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge, United Kingdom

CAMBRIDGE UNIVERSITY PRESS

The Edinburgh Building, Cambridge cb2 2ru, UK
40 West 20th Street, New York, ny 10011–4211, USA
477 Williamstown Road, Port Melbourne, vic 3207, Australia
Ruiz de Alarcón 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa

http://www.cambridge.org

© Cambridge University Press 2003

This book is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.

First published 2003

Printed in the United Kingdom at the University Press, Cambridge

Typefaces Sabon 10.5/14 pt. and Trajan System Quark XPress™ [mg]

A catalog record for this book is available from the British Library

Library of Congress Cataloging-in-Publication Data

Taylor, Rabun M.
Roman builders: a study in architectural process / Rabun M. Taylor.
p. cm.
Includes bibliographical references and index.
isbn 0-521-80334-9 (hb) – isbn 0-521-00583-3 (pbk.)
1. Building – Rome – History. 2. Architecture, Roman. 3. Rome – Antiquities. I. Title.
th16.t38 2003
690´.0937 – dc21
2002073306

isbn 0 521 80334 9 hardback

isbn 0 521 00583 3 paperback
Contents

List of Illustrations page ix

Acknowledgments xv

Introduction i

1 Planning and Design 21

2 Laying the Groundwork 59

3 Walls, Piers, and Columns 92

4 Complex Armatures 133

5 Roofing and Vaulting 174

6 Decoration and Finishing 212

Notes 257

Glossary 275

References 281

Index 293

vii
Illustrations

1 Baalbek: Temple of Dionysus before excavation page 3

2 Rome: Baths of Diocletian, frigidarium (now the
transept of Church of S. Maria degli Angeli) 5
3 Alcántara: Roman bridge of the Trajanic period 11
4 Perugia: funerary plaque found near Rome 29
5 Rome: plan, elevations, and reconstructed cutaway view
of Theater of Marcellus 30
6 Oplontis: megalographic fresco 31
7 Niha: marble model of temple 33
8 Baalbek: fragmentary stone model of Great Altar 33
9 Baalbek: reconstructed bird’s-eye view of Sanctuary of
Jupiter Heliopolitanus 34
10 Baalbek: cutaway perspective reconstruction of Great
Altar 34
11 Baalbek: analytical view of interior components of Great
Altar 35
12 Baalbek: a single block from Great Altar 35
13 A common method of establishing an equilateral
triangle in Greco-Roman architectural design 37
14 Tivoli: Hadrian’s villa, hypothetical design sequence for
Island Enclosure 38
15 Simple geometric methods to establish angles in ancient
planimetric design 39
16 Verona: design scheme of amphitheater 41

ix
x ILLUSTRATIONS

17 Rome: main vault and decoration of Arch of Septimius

Severus 44
18 Istanbul: interior of Hagia Sophia 45
19 Hypothetical reconstruction of Caesar’s wooden bridge
over the Rhine 47
20 Structural actions of an arch with and without a tie-rod 49
21 Rome: Basilica of Maxentius 50
22 Rome: reconstructed view of interior of Basilica of
Maxentius 51
23 Rome: reconstructed cutaway view of Baths of Diocletian 52
24 Tivoli: Hadrian’s villa, view of Island Enclosure 53
25 Rome: interior of church of S. Costanza 54
26 Baiae: engraving of interior of “Temple of Mercury” 55
27 Rome: interior of the Pantheon 57
28 Palestrina: model of Sanctuary of Fortuna 65
29 Dougga: hypothetical string layout scheme for the Baths
of Licinius 65
30 Rome, Via Appia: ideal and actual plans of mausoleum
at Villa of Maxentius 69
31 Rome: Baths of Caracalla, perspective plan of frigidarium
and natatio with adjusted column positions 71
32 Rome: elevation of natatio facade of Baths of Caracalla 73
33 Bologna: analytical illustration of the juncture of two
segments of the Roman aqueduct 75
34 Wooden shuttering for central-Italian Roman foundation
construction 77
35 Rome: Palace of Domitian, visible concrete foundations 79
36 Rome: hydraulic and service armature of Baths of
Caracalla, including downspouts and known drains 82
37 Pozzuoli: partial plan of Flavian amphitheater with
drainage system 83
38 Pozzuoli: section of Flavian amphitheater with drainage
system 83
39 Analytical cutaway view of a hypocaust system 85
40 Carthage: Antonine Baths, general view of subterranean
corridors 86
41 Trier: reconstructed cutaway view of imperial baths 87
42 Boscoreale: Villa Rustica, furnace and boiler serving
kitchen and baths 88
43 Trier: section through caldarium and external corridor
of imperial baths 89
44 Rome: Baths of Caracalla, reconstructed cutaway view 94
45 Rome: Baths of Caracalla, plan of central block 95
ILLUSTRATIONS xi

46 Rome: concrete-and-brick construction method of the

imperial period 99
47 Roman scaffolding with continuous putlogs 101
48 Rome: Baths of Caracalla, modern restoration of a cross
section of a spur wall with visible bonding courses 103
49 Rome: Baths of Caracalla, niche in west oval room
containing bowed arch 108
50 Baiae: reconstructed interior view of annex of “Temple
of Venus” 109
51 Rome: perspective view of natatio, Baths of Diocletian 113
52 Stabiae: construction fresco from Villa S. Marco 114
53 Column drum with lifting bosses 116
54 Capua: funerary relief depicting treadmill crane lifting
column shaft 117
55 Rome: reconstructed view of frigidarium, Baths of
Caracalla 119
56 Hypothetical tilting mechanism for large column shafts 120
57 Baalbek: flank of Temple of Dionysus before
archaeological intervention 121
58 Baalbek: reconstructed view of same part of the temple 122
59 Baalbek: analytical view of roof construction method,
Temple of Dionysus 122
60 Baalbek: facade of Temple of Jupiter with estimated
weights of elements 123
61 Baalbek: corner block of pediment, Temple of Jupiter 124
62 Rome: perspective plan of frigidarium of Baths of
Caracalla with breaks of continuity in masonry 125
63 Baalbek: contact surfaces of monumental column base
and shaft 125
64 Rome: Baths of Caracalla frigidarium, socket left by
robbed entablature block 127
65 Rome: Basilica of Maxentius, southeast wall exterior 128
66 Rome: Pantheon, view of facade 129
67 Rome: Pantheon, elevations as intended and as built 130
68 Rome: plan of Pantheon porch with three hypothetical
phases of column positions 131
69 Rome: Colosseum, aerial view 135
70 Rome: analytical cutaway view of Colosseum 137
71 Rome: Colosseum, axonometric plan depicting raft
foundation and radii of curvature 139
72 Rome: Colosseum, analytical model of infrastructure,
Museo della Civiltà Romana 140
73 Rome: Colosseum, section view after Cozzo 141
xii ILLUSTRATIONS

74 Rome: Colosseum, section elevation with author’s

numbering system 142
75 Rome: Colosseum, podium of Level 3a and remains of
supporting structure seen from inside 143
76 Rome: Colosseum, Cozzo’s proposed construction
scheme for upper levels 145
77 Rome: Colosseum, view of the crowning arches and
infill of Piers 3–6 at Level 2 146
78 Rome: Colosseum, configuration of blocks between
Piers 3 and 6 147
79 Segovia: aqueduct design with cross-braced arches 148
80 Rome: Colosseum, exploded view of the travertine
courses forming piers 149
81 Analytical illustration of lewises and two-piece forceps
with a hypothetical reconstruction of their use on the
tomb of Caecilia Metella near Rome 151
82 Rome: Colosseum, construction of Level 1 153
83 Rome: Colosseum, the facade orders 155
84 Rome: Colosseum, hypothetical string guides for
establishing vault dimensions 157
85 Rome: Colosseum, outer ambulatory of Level 2, with
visible ribs in the vault 159
86 Rome: Colosseum, construction of Level 2 161
87 Rome: Colosseum, ambulatory of Level 2b 162
88 Rome: Colosseum, inner ambulatory of Level 2a 163
89 Rome: Colosseum, section of Levels 1–3 with typical
water-supply scheme 164
90 Rome: Colosseum, stairway of outer corridor of
Level 3a 164
91 Rome: Colosseum, interior surface of outer wall from
Level 3a to top 165
92 Rome: Colosseum, construction of Level 3 167
93 Rome: Colosseum, view of exterior from the north 168
94 Rome: Colosseum, “flying” staircases of Level 3b 169
95 Rome: Colosseum, inner elevation of Levels 3b and 4 169
96 Rome: Colosseum, construction of Level 4 171
97 Rome: Colosseum, inner surface of attic 173
98 Rome: Baths of Diocletian (now Church of S. Maria
degli Angeli), vault of frigidarium 175
99 Nîmes: Pont du Gard 180
100 Nîmes: Pont du Gard, mason’s marks designating
s[inistra], m[edia], d[extra] (“left, middle, right”) and
I, II, III, IIII, V 181
ILLUSTRATIONS xiii

101 Rome: relief discovered in the vicinity of the Palazzo

della Cancelleria depicting a structure with complex
timber roof 181
102 Nîmes: Pont du Gard, hypothetical centering scheme 182
103 Nîmes: Pont du Gard, hypothetical centering scheme 183
104 Rome: Baths of Caracalla, hypothetical centering scheme
for vaults with tile lining 184
105 Rome: Basilica of Maxentius, barrel vault preserving
imprint of centering 185
106 Rome: Tor de’ Schiavi mausoleum, pattern of lagging
boards impressed into dome 185
107 Semicircular centering rib that can be dismantled in stages 186
108 Rome: Baths of Caracalla, hypothetical centering scheme 187
109 Rome: Basilica of Maxentius, analytical view of coffered
vault 189
110 Rome: Basilica of Maxentius, fallen fragment of coffered
vault revealing construction materials and techniques 191
111 Rome: Tor de’ Schiavi mausoleum, hypothetical centering
scheme 192
112 Hypothetical tower centering scheme for large domed
rotundas 193
113 Rome: Pantheon, detail of hypothetical centering scheme 196
114 Rome: Pantheon, hypothetical centering scheme in plan
and elevation 197
115 Rome: Pantheon, author’s proposed centering scheme,
analytical view 198
116 Rome: Pantheon, section view of two phases of author’s
proposed centering scheme 198
117 Rome: Pantheon, base of dome stripped of plaster 199
118 Rome: Pantheon, first construction phase of dome 203
119 Rome: Pantheon, analytical cutaway view of vertical
construction 206
120 Rome: Pantheon, second phase of dome construction 207
121 Barcelona: nave of Antoni Gaudí’s church of the
Sagrada Familia 213
122 Miletos: reconstructed view of terminal aqueduct
distribution tank and nymphaeum 215
123 Boscoreale: decorated bedroom from the Roman luxury
villa 217
124 Rome: tomb of Arruntii near Porta Maggiore 220
125 Pozzuoli: stuccoed vault at Fondo Caiazzo 221
126 Pompeii: stuccoed vault in anteroom of apodyterium of
men’s baths, Stabian Baths 221
xiv ILLUSTRATIONS

127 Workers incising lines in plaster 223

128 Rome: plastering divisions of a wall in House of Livia 224
129 Stabiae: Villa Ariana, close-up of decoration and incised
guidelines in diaeta 9 225
130 Sens: Roman relief depicting frescoists at work 225
131 Lepcis Magna: Hunting Baths, schematic drawing of
mosaic decoration 227
132 Ostia: plan with mosaics of Insula of the Muses 229
133 Vienne, Saint-Romain-en-Gal: hexagon mosaic from
House of the Five Mosaics 230
134 Verulamium: portion of a geometric mosaic illustrating
flaws in guilloche borders 231
135 Ostia: reconstruction of opus sectile wall decoration
of house outside Porta Marina 233
136 Method for applying revetments to walls 234
137 Ostia: House of Cupid and Psyche, opus sectile walls
and floor 235
138 Tivoli: Hadrian’s villa, prefabricated modules of opus
sectile 235
139 Rome: inscribed marble slab depicting assembly of an
opus sectile pattern unit on a temporary table 236
140 Petra: landscape with rock-cut tombs 237
141 Rome: Forum of Trajan, Basilica Ulpia facade; restored
elevation of bay of south facade 239
142 Baalbek: Sanctuary of Jupiter Heliopolitanus, restored
elevation of propylaea 241
143 Didyma: Sanctuary of Apollo, inscribed template of
column compressed vertically by a factor of 16 242
144 Tivoli: Hadrian’s villa, Water Court, plan of excavations
with planting pits and drains 246
145 Urbino: inscribed plan of a tomb and garden, with
hypothetical reconstruction of site 247
146 Nîmes: Tour Magne 249
147 Miletos: Baths of Faustina, reconstruction of thermal
window 250
148 Tivoli: Hadrian’s villa, view of Scenic Canal and
Triclinium from north end 251
149 Tivoli: Hadrian’s villa, view from Scenic Triclinium 253
150 Lepcis Magna: reconstructed view of Temple of Jupiter 255
1
Planning and Design

Let no one suppose that I selected an insignificant accom-

plishment, proposing to adorn it with my rhetoric. For I think
it is a sign of no small intelligence to conceive of new patterns
of beauty for common things; such is the accomplishment the
marvellous Hippias provided for us. It has all the virtues of a
bath: utility, convenience, good illumination, proportion, har-
mony with the site, provision for safe enjoyment; and further-
more, it is adorned with the other marks of careful planning:
two lavatories, numerous exits, and two devices for telling time,
one a water clock with a chime like a bellowing bull, the other
a sundial.1
– Lucian, Hippias or The Bath

All these works should be executed so that they exhibit the

principles of soundness, utility, and attractiveness. The princi-
ple of soundness will be observed if the foundations have been
laid firmly, and if, whatever the building materials may be, they
have been chosen with care but not with excessive frugality.
The principle of utility will be observed if the design allows
faultless, unimpeded use through the disposition of the spaces
and the allocation of each type of space is properly oriented,
appropriate, and comfortable. That of attractiveness will be up-
held when the appearance of the work is pleasing and elegant,
and the proportions of its elements have properly developed
principles of symmetry.
– Vitruvius 1.3.2

21
22 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

V itruvius’ principles of good design continue to resonate.

Even today his triad firmitas, utilitas, venustas adorns
the emblem of the Society of Architectural Historians.
In themselves they are strong and enduring principles (despite at-
tempted repudiations of the third); but they are not immune to the
vicissitudes of taste and cultural change. To varying degrees, all
three are subject to semantic drift. Suitably, soundness is the least
culturally encoded of the principles. We all obey the law of gravity
and design our buildings accordingly. But although every building
benefits from a good foundation and high-quality materials well
applied, these alone do not complete the idea of soundness. As I
suggest below, structure has a psychological as well as a physical
dimension. It must withstand the test not only of time and travail
but of human reception. Attractiveness is of course the criterion
most likely to change its rules over time and distance. Little has been
written on aesthetics of the Roman period, perhaps because our
own culture tends to view Roman architecture (and to a lesser ex-
tent Roman art) with an approving and sympathetic eye. The re-
action of a Roman like Lucian to a kind of architecture that we like
rather well – even if it is sadly fragmentary today – would seem to
approximate our own. Roman architecture, which for all its bursts
of genius and ferment was never a vessel of social revolt, was firmly
embedded in mainstream culture. Moralists like Cicero and Seneca
might rail at the built luxuria of the idle rich, but when it comes
right down to it, an archaeologist would be hard pressed to distin-
guish the villas of one breed from those of the other.

UTILITY AND SOCIAL FUNCTION

et us briefly discuss Vitruvius’ second criterion of successful ar-
L chitecture, utility. For the architect this was the primary concern
in the early phases of design. What use would a projected building
have, and how would the building maximize its usefulness? Build-
ings of importance were never useful in just one way. Beauty was
itself utility; an ugly building was likely to be avoided, especially if
it was eclipsed by competing structures. Martial’s epigrams are full
of plaudits and indictments of one bath or another according to ap-
pearance, clientele, or water quality. As I have already suggested,
signification was another form of utility. Whether or not an archi-
PLANNING AND DESIGN 23

tect was aware of the fact, every building’s form, function, and dec-
orative scheme were steeped in cultural meaning. Some structures,
such as honorary arches and triumphal monuments, functioned
principally as signifiers; their function was to engage one’s attention
and convey an ideological message. Another way in which architec-
ture made itself useful in Roman cities was to articulate a larger
urbanistic scheme. Thus the built environment could serve as a cat-
alyst of movement through the fabric of the city.2
Vitruvius is concerned primarily with utility as it is commonly
understood, the ability of a building to meet its preassigned role as
a place for enabling, sheltering, and organizing human or divine ac-
tivity. Certainly function is one of the most important determinants
of form. A structure must be envisioned in terms of the ways people
will use it. Will it be a multifunction facility, like a basilica? Or will
it serve a narrow range of functions, like a theater or odeum, or a
single function, like a latrine? What special uses has the building’s
sponsor stipulated? The architect, then, will start with a known plot
of land, a designated building type (temple, amphitheater, villa,
bath, etc.), the patron’s stated wishes, and probably a variety of
other factors that will condition his design (building codes, geolog-
ical and hydrological conditions, availability of labor and materi-
als, etc.).
In his studies of the divisions of private architectural space in
Pompeii and Herculaneum, Andrew Wallace-Hadrill has developed
a simple but practical matrix for evaluating the social uses of the
Roman house.3 His two scales of social functionality – public–
private and grand–humble – operate independently to form the axes
of a sort of Cartesian plane. This model can be adapted to Roman
architecture outside the private sphere, but with different axes of
differentiation. The grand–humble scale is still applicable, especially
in contrasting discrete parts of a building; however, depending on
the circumstances, one might choose instead a scale of social inte-
gration (segregated–integrated). Roman class-consciousness, which
was extreme, was not always manifested by absolute physical sep-
aration. Seating at spectacles was rigidly segregated according to
class or sex, but people of all classes (and occasionally both sexes)
bathed together, and the systems of slavery and patronage ensured
regular interaction between the grand and the humble in the same
space. The public–private scale is less useful outside of houses and
villas, and could profitably be replaced by a scale of specificity of
24 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

purpose (dedicated–multipurpose). Roman building types generate

a scatter in at least three quadrants of the new Cartesian plane: ded-
icated + segregated (amphitheaters, theaters, etc.); dedicated + in-
tegrated (baths, markets, etc.); and multipurpose + integrated (basil-
icas). The social reasons for this variability are too complex to
pursue here, but it is obvious that the architect had to be aware of
them and of the special wishes of his patron. In segregated build-
ings he would be responsible for establishing not only physical bar-
riers but also psychological ones, such as a change in decoration,
ceiling height, or corridor width to suit the social status of those in
spaces of especial privilege or ignominy.4 And he must account for
the positioning and distribution of human beings within his build-
ing, both visitors and those who controlled them. In the Colosseum,
one of the most segregated of all buildings, one can see the diminu-
tion of grandeur in decoration as one passes from the grand stuc-
coed senatorial entrance at ground level to the upper corridors,
though the architectural armature remains impressive throughout.
A network of gates, grilles, and guards would have helped to reduce
ambiguity for the spectator entering at a numbered gate and climb-
ing to his seating section. In equal measure a skilled Roman archi-
tect would have been expected to create integrated spaces that were
psychologically inclusive by emphasizing their public nature. Ro-
man baths are famously “democratic” in this respect; even emper-
ors were known to share an occasional bath with their subjects in
the great thermae of Rome. We must not forget, however, that these
establishments were operated by a huge service staff, which re-
mained largely unseen by the bathers.

VITRUVIUS AND DESIGN

itruvius identifies six components of architecture: ordering
V (ordinatio), design (dispositio), shapeliness (eurythmia), sym-
metry (symmetria), correctness (decor), and allocation (distributio).5
These terms have been the subject of endless debate.6 Whether they
merit such scrutiny is itself a point of contention. One thing is rel-
atively certain: Vitruvius is not interested in characterizing Roman
architecture as we have defined it. Except in domestic architecture,
his most important models were Greek; on the most “Roman” of
all building types, such as amphitheaters and triumphal arches, he
is silent, while he expounds at length about some of the great tem-
PLANNING AND DESIGN 25

ples of the Greek world. His adherence to Greek traditions carries

over to his intellectual system as well, which borrows terms and
ideas from Greek rhetorical and aesthetic theory.
Not that his six “components” are systematic. Shapeliness, sym-
metry, and correctness seem to be properties of structures; design
and allocation are processes involved in building them. Ordering
(ordinatio), depending on how it is interpreted, could fall into ei-
ther category. It pertains to the common use of a module, usually
a unit of measurement used for some iterative purpose such as the
spacing of columns. In the Vitruvian scheme this became the stan-
dard unit by which many other features were determined: width and
height of columns, height of capitals, and so on.7 As a principle by
which to design a whole building it was rarely used outside of tem-
ples. Whatever the precise meaning of ordinatio, one must remain
skeptical of attempts to marshal Vitruvius’ ragged sextet into two
symmetrical triads in which each property is the imagined result
of a congruent process in the other.8
Vitruvian symmetria is not symmetry in the modern sense but a
carefully proportioned relationship among elements. Since each el-
ement has its own inherent proportions (e.g., a column’s height-to-
width ratio), symmetry can be understood, at least in some passages,
as the proportioning of proportions.9 A pediment must not be too
large, too small, nor too steeply or shallowly pitched for the col-
umns on which it stands; thus the correct symmetry of a temple fa-
cade is a function of both variables (columns and pediment), among
others. Shapeliness (eurythmia), probably a notion borrowed from
Plato,10 seems to refer to the simple, inherent proportions of each
element, such as the aforementioned height-to-width ratio of a col-
umn. It has recently been argued that these Vitruvian qualities do
not inhere in buildings but are rather processes involved in their cre-
ation. They are physically evident in a number of Roman buildings
as refinements made upon established plans or rules.11
It is unclear how symmetry and shapeliness should be distin-
guished from ordering, or sometimes even from each other. Vitru-
vius himself, who often seems to be talking over his head, may have
been unequipped to explain himself.12 But plainly all three terms
presuppose an unmediated aesthetic process governed by psycholog-
ical principles. Correctness (decor) is something again, having more
to do with cultural convention or received common sense than to
first principles of cognition. It is the component of reception drawn
26 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

from outside the object itself. In Vitruvius’ opinion, a correct design

must not mix Doric elements with Ionic, or match masculine gods
like Mars or Hercules to a “feminine” order like the Corinthian:
Such mismatches would offend the sensibilities of a culturally con-
ditioned Roman.13 And common sense or functionalism would
dictate “natural” correctness: The orientations of various rooms
should suit their functions. Winter quarters should face west to gain
the benefits of the afternoon sun in the winter, bedrooms should
face east to make full use of the morning sun, and so on.
There is nothing mysterious about allocation (distributio). It is,
in Vitruvius’ words, “the efficient management of resources and site
and the frugal, principled supervision of working expenses.” Like
Plato’s architektôn,14 Vitruvius’ architect oversees all aspects of the
job from start to finish. Such was common practice in the Roman
world, but not necessarily the norm.15 While keeping within a budg-
et may seem an obvious necessity today, it was not always obvious
in antiquity. Large building projects of earlier and later times were
often open-ended, either enjoying unlimited royal sponsorship or
occupying generations of master builders and patrons who could
each vouch for only a fraction of the overall budget. Many of the
grandest Greek temples were never completed at all.16 In the Ro-
man period it became possible as never before (and rarely thereafter
until the Renaissance) to complete a large project on time and on
budget. And for this, one needed a fairly comprehensive plan.
For our purposes, the most important of Vitruvius’ six elements
is design (dispositio). In a sense, it encompasses most of the others,
for one cannot envision the act of laying out a building without con-
stant attention to aesthetic principles. And it is only through design
that one can make meaningful projections of time, labor, and mate-
rials. Vitruvius tells us that design is manifested in three ways: ichno-
graphia (floor plan), orthographia (elevation), and scaenographia
(perspective drawing). These are still essential tools of the architect
today. A plan shows the disposition of walls and floors as seen from
directly above. It is the natural first step for an architect who is
dealing, after all, with a relatively horizontal plot of land enclosed
within definite boundaries. An elevation is a drawing of the upright
structure without perspectival distortion. It may take a frontal view,
as Vitruvius says, or in more complex projects, a view of another
exterior surface or of a section cutting through the proposed build-
ing on a flat plane. An elevation usually is derived from the plan;
PLANNING AND DESIGN 27

yet its proportions may cause one to rethink the plan, for a build-
ing’s vertical substance is as fundamental to good design as the treat-
ment of the horizontal plane. Perspective drawings have a more an-
alytical purpose; they strive to give the viewer a “feel” for a building
in space. They are uniquely valuable for the representation of in-
teriors, which are so important in Roman architecture.
I am not suggesting that Vitruvian drawings were the lingua fran-
ca of all Roman architects. Romans probably designed a wider rep-
ertory of buildings than any society before the nineteenth century.17
Quotidian structures could have been planned on site from experi-
ence and rules of thumb with the help of simple surveying and lev-
eling tools and perhaps a scale drawing of the ground plan. Even
some large-scale imperial structures show evidence of on-site plan-
ning in the form of optical or structural corrections.18 A certain
amount of improvisation was to be expected in every project; but
occasionally a builder’s failure to use drawings effectively got him
into serious trouble. The Sanctuary of the Deified Trajan at Perga-
mon, it would seem, was surrounded on three sides by colonnades,
the two lateral examples of which were adjusted in height after
construction had begun in order to harmonize them with the back
colonnade.19 Lapses of this kind, while not exactly rare at the sites
of imperially sponsored building projects, are not common either.
Most changes that leave traces in the remains – and there are many
– have other causes, such as shifts in resources or ideology.

DRAWINGS AND MODELS IN ROMAN DESIGN

ccording to a well-known story, the famous imperial architect
A Apollodorus of Damascus, annoyed at the architectural pre-
tensions of the future emperor Hadrian, told him, “Be off, and draw
your pumpkins.”20 The reference is to the pumpkin-shaped domes
of the sort that Hadrian later employed in his villa at Tivoli. Wheth-
er or not the story is true, its implications are clear: Hadrian and
the architects in his circle saw the value of thorough designs, not
just of floor plans but of elevations and vaulting structures too.
In recent years much attention has been given to the mechanics
of ancient architectural design in general, and specifically to the
techniques applied to architectural drawings and models. We now
have a much better understanding of the ways in which ancient ar-
chitects and builders communicated spatial ideas nonlinguistically.
28 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

Drawings, used at least as early as the Egyptian and Mesopotamian

periods, appear also in the Roman era. They survive in various
scales and with varying degrees of care, etched in stone or drawn
on papyrus, even in one instance appearing on a mosaic.21 Relative-
ly crude area maps, showing buildings without wall thicknesses, are
known from Rome and other sites. A much finer effort appears on
a funerary plaque from Rome, now in Perugia (Fig. 4). Plans had
varying purposes and functions. Some were used in building con-
tracts.22 Some were maps; that is, they represented an actual, not
a planned, state. The Perugia inscription, however, may be a direct
transcription of architectural plans, for the three buildings repre-
sented on it are all on different scales and show many of the wall
measurements. There seems to be no good explanation for this un-
less the inscriber was copying three separate plans directly. This
plaque and the fragment from Rome, showing the floor plan of the
Temple of Castor and Pollux near the Tiber River, reveal that even
in the cumbersome medium of stone plans could be rendered with
apparent accuracy.23
It has been suggested that Roman elevations were drawn on a
larger scale and in greater numbers than plans, just as they are
today, to account for the many more details of vertical surfaces –
column capitals, moldings, friezes, and so on.24 Certainly elevation
drawings with some degree of detail were used in ancient Egypt; a
beautiful, and rather meticulous, elevation of two faces of a shrine
has come down to us on carefully gridded papyrus from the eigh-
teenth dynasty.25 On the other hand, matters of detail in stonework
may have been left to the master craftsmen, just like the later phases
of decoration such as mosaics and plasterwork. Greek craftsmen
seem to have relied on templates (anagraphês) or full-scale models
(paradeigmata) for detailed stone carving, not on drawings.26 At
both Greek and Roman sites full-scale drawings of architectural fea-
tures in elevation have been found etched onto stone or plaster sur-
faces, where they served as templates for preassembly of building
features.27 It is much more likely that elevations, rather than plans,
were resolved in medias res by means of such templates or draw-
ings. But in certain situations, such as the hillside construction of
the Esquiline Wing of the Golden House (Domus Aurea) of Nero,
which was built on the ruins left by the great fire of 64 a.d., even
some of the plan may have been improvised on site.28 We have no
evidence that drawings in antiquity were ever granted the kind of
PLANNING AND DESIGN 29

4. Perugia: funerary
plaque found near
Rome. Corpus
Inscriptionum
Latinarum.

autonomy they acquired in the Italian and Spanish Renaissance and

still hold today.29 The tyranny of design at Sydney would have held
feeble court in Sardis or Syracuse. The Roman way was not to gen-
erate a design of perfect refinement and supreme authority but to
establish a finite set of objectives pictorially. The means to accom-
plish the ends, sometimes utterly unpredictable, were a matter of
daily dialogue and improvisation on the site; it is these means that
occupy most of this study. Drawings alone can never plot a path
to completion; they can only set the destination.
It is possible to envision how great stone temples, stoas, arsenals,
and other rectilinear structures were built in antiquity without plans.
One may even conceive of a Gothic cathedral or a middle-Byzantine
church being raised without them, bay by bay, overseen by a master
builder of great skill and experience.30 But grand bath buildings,
theaters, and amphitheaters of the Roman period exceed these in
volumetric complexity, and the sheer speed of their realization
would not have allowed the careful cross-checking or methodical
pace of some Greek or Western medieval construction. It is incon-
ceivable that such complex buildings as freestanding theaters or am-
phitheaters – veritable warrens of tubular voids wending through
the building’s swooping fabric, wrapped about one another, care-
fully penetrating or bypassing their neighbors – were given form
without detailed plans and elevations (Fig. 5). While published
handbooks may have existed to aid in design and planning, ulti-
30 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

mately every project was unique. Any structure with so many ris-
ing diagonals or conic surfaces – seating banks, ramps, stairways –
presents a supreme challenge to the draftsman. A few sections in
elevation may suffice, but planimetric (horizontal) sections must be
made in quantity, for none resembles the next. Each drawing must
to some extent have generated a logistical plan for construction;
many issues of building material, method, and sequence would only
have occurred to the architect during this expository phase of the
creative process.
Exposition – the elaboration and communication of ideas – comes
only after those ideas are in place. Plans, elevations, and sections

5. Rome: plan,
elevations, and recon-
structed cutaway
view of Theater of
Marcellus. Ward-
Perkins 1981. By
permission of Yale
University Press.
PLANNING AND DESIGN 31

6. Oplontis: megalo-
graphic fresco.
Deutsches Archäo-
logisches Institut,
neg. 74.2689. Photo:
Sichtermann.

carrying the authority of blueprints tend to be generated toward the

end of the design phase, when a fairly complete general concept is
well in hand, even if logistical and constructional details are still un-
certain. For the development of ideas and envisioning the whole one
needs a more dynamic medium in which the creator can experience
the emerging building as a living environment. Vitruvius’ term scae-
nographia, often translated as perspective drawing, is generally
thought to designate a technique that renders buildings as they
would appear optically, replicating the entire cone of vision. Ap-
proximations of the latter technique appear in many Pompeian wall
paintings (Fig. 6).31 Scaenographia probably had two functions, as
an aid in the design process and as a means of presenting a concept
to the client. Drawings for clients are mentioned several times in the
literary sources,32 and Pierre Gros has made a good case for the use
of scaenographia as a perceptual aid in the design process.33 Roman
architects were fond of establishing horizontal corridors of vision
32 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

(“enfilades”) through multiple spaces by the artful alignment of

doorways, intercolumniations, and windows. Such effects, along
with sensations of verticality, could have been tested and “experi-
enced” with perspective drawings, even if the principles of perspec-
tive were not fully realized.
Scale models are the most direct, paradigmatic means of develop-
ing and conveying architectural ideas. Plutarch’s bidders for public
works may have used presentation models, though his term para-
deigmata could simply refer to drawings.34 It is only natural that the
semieducated artisans who spent their lives working in three dimen-
sions would have been able to read and comprehend models far
more directly and completely than the comparatively schematic
shorthand of drawings or diagrams. Michelangelo evidently built
elaborate limewood (i.e., linden) and poplar models for the Medici
Chapel for similar reasons.35 Indeed Plato seems to suggest that a
builder (oikodomounta) of his own time should be trained by play-
ing with “toy houses” (paideia oikodomêmata).36
The preferred material for working architectural models has al-
ways been wood,37 which rarely survives from antiquity. Fortunate-
ly we possess a number of partial scale models in stone which, even
if simple in conception, probably served as design tools and perhaps
as blueprints for the craftsmen as well.38 The finest example, found
adjacent to the building it represents, is a carefully carved 1:24 stone
replica of the podium and stairs of a temple at Niha, Lebanon (Fig.
7). The superstructure, now lost, was probably made of wood. It
would appear at first sight to be a presentation piece, but a closer
look reveals otherwise. Several steps on the model are inscribed with
dimensions in feet. These refer not to the model itself but to the ac-
tual building. Evidently the model was a conceptual aid onto which
the builder inscribed his modifications for the final design. A polyg-
onal design is inscribed on the model’s cella floor, and this too was
realized in the temple itself (after yet another phase of modification)
as a sort of columned baldachin to shelter the cult statue. Another
marble model, in the museum at Ostia, also represents a temple po-
dium and stairs. It too was a working model: Two variant positions
of the column bases are represented. Dowel holes in the preserved
bases indicate that the superstructure was separate. Again, the latter
was probably of wood, and most likely was detachable from the
podium so that the builder or client could examine its ground plan
and interior.
PLANNING AND DESIGN 33

7. Niha: marble
model of temple.
Illustration: R. Taylor
after Will 1985.

Perhaps the most precious and least remarked of all known Ro-
man models is the 1 :30 fragment of the elaborate Great Altar of
Baalbek. It appears to be one of several stacked sections that could
be dismantled to reveal the staircases inside (Fig. 8).39 Its form is
schematic but unmistakable. The two tower-altars opposite the
Temple of Jupiter at Baalbek are themselves tours de force of stere-
otomy, each among the most complex organizations of space ever
realized in solid stone (Fig. 9, 10).40 Their components are not, like
most building materials, small modular units assembled around a
void. Within the structure solid and void compete as volumetric

8. Baalbek:
fragmentary stone
model of Great Altar.
Illustration: R. Taylor
after Kalayan 1971.
34 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

9. Baalbek: reconstructed bird’s-eye view

of Sanctuary of Jupiter Heliopolitanus.
Great Altar shown in center of courtyard.
Ward-Perkins 1981. By permission of
Yale University Press.

10. Baalbek: cutaway

perspective recon-
struction of Great
Altar. Collart and
Coupel 1951. By
permission of Société
Nouvelle Librairie
Orientaliste Paul
Geuthner.

equals, each interlocking with the other. Joints and seams cease to
correspond consistently to edges; angles are incorporated into the
solids themselves. A component block of the altars might comprise
dozens of curved and planar surfaces defining both figures and voids
and cut to a perfectly conceived analytical plan, their multiplanar
PLANNING AND DESIGN 35

11. Baalbek: analyt-

ical view of interior
components of Great
Altar. Collart and
Coupel 1951. By
permission of Société
Nouvelle Librairie
Orientaliste Paul
Geuthner.

12. Baalbek: a single

block from Great
Altar. Wiegand
1921–5.

faces commingling as snugly as organs in an anatomical model (Figs.

11, 12). Taken separately, each part conveys a distinctive, identi-
fiable fragment of the larger idea. Without models, indeed several
generations of models, the Great Altar was unbuildable. The surviv-
ing model fragment reflects the principal formal ideas of the final
product, but considerably simplified. It is a prototype, the first or
second in a series that evolved from a medium of exploration into
a tool of communication.
36 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

The obvious benefit of models should not obscure the likelihood

that for Roman architects themselves, as opposed to the stonema-
sons and bricklayers, perspective drawings were the most important
creative aids. Models are invaluable tools but they are expressively
opaque, even misleading in the oblique bird’s-eye perspective they
force upon the beholder. Especially in such an interior-dominated
architecture as the Roman, human perspective and scale, the inter-
action of body and building (even if both must be virtual) are para-
mount. You cannot get inside a model to experience it, to intervene
in its volumes, to probe its voids. But you can read the script of a
picture and imagine yourself in the action. No doubt this is why, in
a well-known passage from Aulus Gellius, Fronto’s builders present-
ed rival plans and “specimens” for a proposed bath building in the
form of paintings on parchment (depictas in membranulis varias
species balnearum).41

GENERATING DESIGNS
new building design emerges from such disparate concerns as
A site specifications, the patron’s needs, traditional form, inno-
vation, and available methods for building the mental construct.
This final variable comprises a sort of visual phonetics of architec-
tural language deeply embedded in Greco-Roman intellectual cul-
ture and tradition. The basic linear elements of planar design were
the straight line, the simple curve, and the complex curve (such as
a three-point oval or an ellipse), each with the capacity to project
any of the others into the third dimension. The principles binding
these elements together into coherent forms were the module (i.e.,
a relative measurement specific to a building), absolute measure-
ments in standard units, and pure proportions derived from geomet-
ric and mathematical theory. All were used in Roman design, some-
times in combination.42 Classic modules based on the distance of an
intercolumniation, so essential to the Greeks and to Vitruvius, con-
tinued to govern many building designs, as did a host of other units
and relationships in a process Mark Wilson-Jones calls aggregative
composition.43 Modules emerged in guises entirely apart from col-
onnades or arcades, such as a simple square or circle from which
all other geometric designs of the ground plan emerged. Standard
units of length based on the Roman foot (the pes, 0.296 m) were
instrumental in beginning any design and were commonly used to
PLANNING AND DESIGN 37

13. A common
method of establish-
ing an equilateral
triangle in Greco-
Roman architectural
design. Illustration:
R. Taylor.

round off lengths and distances that may have been established by
proportional means.44
Let us then briefly examine proportions themselves, understood
as the relative lengths of two separate elements or the relation of
two dimensions of a single element. Arithmetic proportions, among
which modules are counted, come in simple numeric relationships:
1 : 2, 2 : 3, 5 : 4, 9 : 1, and so on. Vitruvius works with arithmetic
proportions, as when he prescribes the inner proportions of domes-
tic atria or the sizes of subsidiary rooms in relation to them.45 These
continued to be important in columnar orders and in overall design;
for example, the height of the colossal columnar order of the frigi-
darium at the Baths of Caracalla was equal to a third of its overall
width.46 Geometric proportions, usually manifested in the ratio of
width and length of a room or building, or either of these measures
in proportion to its height, were often based on popular irrational
relationships in pure geometry – for example, the ratio of the side
of a square to its diagonal, 1 : √2. Many proportions were facilitat-
ed by the use of the compass, which could quickly transfer lengths
of diagonals to the sides of rectangles, or turn side lengths inward
to intersect with each other. A common proportion in Greek ar-
chitecture is the length of a base of an equilateral triangle to its
height (1 : √3).47 The proportion could be easily drawn by turning
the compass inward from one side of a square to the midline of the
square (Fig. 13).
The extent to which such methods actually were used by the
Greeks, especially in the design of the elevations of buildings, is still
hotly debated.48 There is no such disagreement about Roman archi-
tecture, even if there will always be uncertainty about the exact pro-
38 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

14. Tivoli: Hadrian’s

villa, hypothetical
design sequence for
Island Enclosure.
Jacobson 1986.
By permission of
D. Jacobson.

cedures used in specific circumstances.49 New architectural forms

were devised on the basis of elaborate and even abstruse inner rela-
tionships comprising tangents and intersections of lines, circles, and
polygons, as surely was the case with the fanciful buildings at Ha-
drian’s villa (Fig. 14). Plans were drawn exclusively with a compass
and a ruler50 and were later transferred onto the ground, floor, or
slope with larger versions of the same tools. Protractors seem to
have been avoided; instead, angles were established by geometric
tricks of the trade. Simple right angles could be produced with set
squares forming a 3–4–5 triangle, a method that Vitruvius inverted
for establishing the profiles of stairways.51 A more precise method
is to draw two intersecting circles centered on a baseline and then
to connect their two points of intersection (Fig. 15, a).52 If those two
circles are made to share a radius along the baseline, then sixty-
degree angles (and equilateral triangles) can be formed by running
lines from the centers to the intersections (Fig. 15, b). Any angle can
be bisected by swinging a cord of a uniform length from equal dis-
PLANNING AND DESIGN 39

15. Simple geometric

methods to establish
angles in ancient
planimetric design.
Illustration: R. Taylor.

tances along its sides from the vertex and running a line from the
vertex to the point where the two arcs intersect (Fig. 15, c). Many
other similar procedures allowed architects to lay out buildings with
very few actual measurements, either of angle or of distance. It has
been proposed, for example, that the great bath block of the Baths
of Caracalla was designed geometrically around a few initial linear
measurements of a hundred-foot module.53
Although proportions are widely observed in Greek architecture,
many temples of the Ionic order and some of the Doric seem to have
been laid out in a sequential fashion, allowing them (at least in part)
to be designed and modified as they went up.54 Such may have also
been the case for some conventional Roman stone temples. But con-
crete, and the operational problems and opportunities it created,
fundamentally changed the old ways. First, its malleable nature and
stonelike integrity enabled buildings to be increased in complexity.
Second, the speed with which it was laid demanded efficiency. Con-
crete and worked stone, both of which continued to be used togeth-
er in buildings of many types, took shape at different speeds in a
nonlinear process. Stone entablature A had to be fully envisioned
and its surfaces of contact dressed before its place in concrete wall
B was realized. Otherwise it would not be ready to be positioned
at the critical moment, in turn delaying concrete vault C, which
would rest upon the entablature, and so on. Concrete demanded
speed, and it inspired complexity. The corollary of speed and com-
plexity is high design.
 40 ROMAN BUILDERS: A STUDY IN ARCHITECTURAL PROCESS

By high design I mean a reasonably comprehensive process in

which numerous visual approaches to the building are played out
at the drawing board, and by which multiple drawings are produced
to be used as blueprints on site. The procedure is far from sponta-
neous: “designing is a process that proceeds from the simple to the
more complex, in which an initial scheme may inevitably become
compromised.”55 To the extent that complexity could be minimized
up front, it was. Roman architects were fond of simple internal pro-
portions and round numbers. But Vitruvius allows that a judicious
designer will know how to compromise on principles of proportion
for visual effect. The use of principles other than pure geometry was
widespread. As buildings grew more complicated, merging straight
lines and flat planes with curved surfaces, perfect inner logic retreat-
ed out of reach. A classic example of this is the mensural tension of
circular and oval buildings such as rotundas, theaters, and amphi-
theaters. It is impossible to design a circular building with both di-
ameter and circumference in round numbers of feet, for example.
Usually one of the two core generative processes – geometry and
arithmetic – will yield to the other.56 As a general rule Roman plans
seem to use a minimum of calculation, perhaps because a profusion
of numbers invites error. If a circular colonnade has an irrational
circumference, the intercolumniations can be determined either by
division or simply by bisecting angles from the circle’s center. The
second method requires no units of measurement at all, and is far
less prone to error.
Recent studies on amphitheaters have investigated how Roman
designers tried to reconcile regularity of measurement and practical-
ity with the elegance of pure geometry.57 As so often happened, the
very first design decision was a geometric one. How to generate the
shape? Though amphitheaters look like ellipses in plan, very few of
them – and none of importance – are true ellipses, because of the
difficulty of producing a continuous grandstand with a perfectly
uniform width around an elliptical arena. Almost all amphitheater
plans comprise segments of circles with different radii, joined at
carefully predetermined points where they share tangents. Typically
there are four segments, two with longer radii forming the long sides
and two with shorter radii forming the “ends.” One of the most
straightforward schemes is used for the amphitheater at Verona (Fig.
16). The basic units are two equilateral triangles sharing a side on
the main axis. The four corners of these joined triangles are the cen-
ters of the four circles whose segments merge to form the oval. All

View publication stats