Professional Documents
Culture Documents
Andrew M. Riggsby
1
iv
1
Oxford University Press is a department of the University of Oxford. It furthers
the University’s objective of excellence in research, scholarship, and education
by publishing worldwide. Oxford is a registered trade mark of Oxford University
Press in the UK and certain other countries.
9 8 7 6 5 4 3 2 1
List of Figures ix
List of Tables xi
List of Plates xiii
Acknowledgments xv
A Brief Orientation 1
1. Lists 10
Ordered Lists 11
Indexed Lists 15
Tables of Contents 22
Nested Lists 29
viii { Contents
6. Conclusion 203
Where Are We Now? 203
Going Forward I: Power and Other Topics 210
Going Forward II: An IT Revolution in Late Antiquity? 216
References 223
Index 245
FIGURES
ix
x
TABLES
xi
xii
PLATES
xiii
xiv
ACKNOWLEDGMENTS
This project, at least some parts of it, dates back a long time. The seminar alluded to
at the beginning of chapter 2 was offered in the late 1990s, and I suspect that some
of the thoughts here probably first arose before I finished graduate school, while
I was reading Edward Tufte’s books from my mother’s book shelf. I have acquired
an unusually large number of scholarly debts over that time (and unfortunately
have doubtless forgotten others equally important). I got particularly extensive as-
sistance and commentary from Klaus Geus, Paul Keyser, Michael Koortbojian,
Rabun Taylor, and the readers for Oxford University Press. Tony Corbeill, Serafina
Cuomo, Tony Grafton, Joseph Howley, Nate Jones, Stephanie Frampton, John
Clarke, Eric Orlin, Liz Robinson, Philip Stinson, and Tyler Travillian all read
and commented on chapters in draft. I have also gotten other help, particularly
in the form of penetrating questions or advance access to work in progress from
Dorian Borbonus, Alan Cameron, C. Michael Chin, Megan Goldman-Petri, Julia
Hejduk, Alexander Jones, Duncan McRae, Reviel Netz, Carlos Noreña, Laura
Novick, Dan-el Padilla Peralta, Tim Parkin, J.-B. Piggin, Phil Resnik and Jiesi Shi,
Jane Sancinito, Josh Sosin, and a seminar which covered this material (Gabrielle
Bouzigard, Timothy Corcoran, Eli Fleming, Vera Leh, Will Shrout, and Alain
Zamarian). I would also like to thank audiences at Brown, Chicago, Columbia,
Duke, Johns Hopkins, Maryland- Baltimore County, Minnesota, NYU, North
Carolina, Penn, Princeton, Texas Tech, Yale, and the Finnish Institute in Rome for
subjecting various parts of the argument to friendly scrutiny. And, of course, I need
to thank Joe Farrell, the series editor, and Stefan Vranka, the sponsoring editor, for
their interest, encouragement, and assistance in transforming the “project” into an
actual book.
Finally, I would particularly like to signal the role in this project of my ongoing
interaction with two younger scholars. Seth Bernard and Sarah Bond in their dis-
tinct, inimitable ways provided a stream of questions, prods, prompts, and problems
and materials to work with. A project of this scope necessarily relies on the kind-
ness of strangers to have any hope of reaching the necessary breadth, but even be-
yond that the constant presence of these two kept me honest and on my toes.
While I have been working on parts of this project for many years, the core of
the research and writing took place over two academic years, and I am more than
happy to thank the funding entities that made that possible. In 2010–11, I held
the NEH/Roger A. Hornsby Rome Prize at the American Academy in Rome. In
addition to the scholars named earlier, I must thank the Academy for both the
xv
xvi
xvi { Acknowledgments
Plate 2 Riot in the Amphitheater (Pompeii, now in the Museo Archeologico Nazionale, Naples)
© Vanni Archive/Art Resource, New York
Plate 3 Landscape from the columbarium of the Villa Doria Pamphilj (A/XII)
Su concessione del Ministerio dei beni e delle attività culturali e del turismo—Museo Nazionale Romano
4
Plate 4 Landscape from the villa under the Farnesina, walkway F-G (inv. 1233)
Su concessione del Ministerio dei beni e delle attività culturali e del turismo—Museo Nazionale Romano
Plate 5 Landscape from the villa under the Farnesina, cubiculum D (inv. 1037)
Su concessione del Ministerio dei beni e delle attività culturali e del turismo—Museo Nazionale Romano
6
2 { Mosaics of Knowledge
have been more willing to see social and institutional supports for literacy, even
if they take different forms than modern ones and even if they only create larger
pockets of literacy rather than universalizing it. But the real differences lie not in
tweaking Harris’ numbers but in adding to his stock of questions and localizing
their answers. There has been an increasing analytic interest in what might be called,
in the plural of the title of Johnson and Parker’s 2009 book, Ancient Literacies.
That is, with the basic quantitative picture already in place, interest has shifted to
more qualitative questions of how reading and writing skills (and, to a lesser extent,
numeracy) were employed by individuals in particular contexts. What kind of infor-
mation, Woolf 2009 asks, would labels on commercial olive oil jars in their highly
stereotyped format have been able to convey to “readers” in the industry, who might
not be fully literate in general terms and might not even be Latin speakers? Or what,
Beard 1991 considers, is the motivating effect of rituals that mediated access to the
divine through writing?
I tell this story, which will already be familiar to many readers, because it has
multiple resonances with the unfamiliar story I will tell in the body of this book.
First, I hope also to make modern states of affairs seem less natural. One reason
(though not, of course, the only one) it was easy to accept a highly literate antiquity
was the ease with which reading comes to individuals today. It does not then feel
like a very strong claim to extend that across an entire society, though in fact it is.
The information technologies discussed in this book—things like numbered lists,
numerical tables, or mechanical weights and measures—offer a similar temptation.
Their use comes so naturally to anyone acculturated in the modern world that we
are likely to take it for granted that they were available in the ancient world as well.
Most insidiously, something merely similar to a modern device can readily be taken
for fully identical.
Additionally, however, I would like to try to borrow some of the qualitative focus
that characterizes much of the response to Ancient Literacy. That is, I ask not just
how often Romans used various technologies but also when, how, and why. Those
contextual questions are probably good historical practice in general, but they
strike me as particularly urgent in this specific area of technology. As it happens,
most of the information technologies discussed here were not huge successes that
spread through the Roman world in the way, say, blown glass or concrete construc-
tion did. But the reason I can talk about them at length is that they were at least
invented, unlike, say, stainess steel or stirrups. One need not (as most readers prob-
ably will not) believe in the necessary “progress” of technology to be puzzled by a
lack of eagerness to adopt what were at least technically useful devices. “Why did
the Romans use tables or scale representations differently than how we do?” for
instance, is interesting because it will likely tell us as much about the Romans in
general as about those tables or plans. Different people use technologies for par-
ticular purposes in particular circumstances. The question, thus, is almost never
whether a particular technology is “good” or “powerful” or “elegant,” or anything
else. The issue is whether particular people (or enough of them) find it worthwhile
A Brief Orientation } 3
to acquire that technology for some particular task before them. In a broad sense,
we could make this point about any adoption of any technology, but I will argue
that Roman cost-benefit calculations in this respect were particularly strict and
particularly local.
While I do not, I hope, adopt a teleological view of technological change,
I should probably also point out that I do not hold a purely culturalist view, ei-
ther (even though the previous paragraph might have been read that way). I don’t
think it is meaningful to describe any technology as “good” or “the best,” only
good or best for some particular end (which might itself be defined by a complex
of mechanical, social, and other aims). However, this does mean that to the extent
that we understand those aims, we can say that some technologies are objectively
better: they have a lower rate of false positives, they create less pollution, they are
cheaper, the hardware is less likely to malfunction, they require fewer (or, if this is
what circumstances demand, more numerous) human workers, they channel rev-
enue to a politically powerful class. My account is also imperfectly culturalist be-
cause I believe that the invention and diffusion of inventions are path-dependent.
Devices do not simply arise whenever and wherever cultural circumstance might
make them desirable. Various material preconditions and contingent discoveries are
required. I don’t claim to have proven the truth of this point of view. Rather, I have
sketched an approach which will stand or fall depending on how well it actually
works throughout the body of the study.
Though quite broad, the scope of this book will be limited in two important
ways. Chronologically, it will extend, at least in principle, from the earliest Roman
times to the year 300 (all dates will be ce unless otherwise noted). Any precise cutoff
is of course arbitrary to some extent, but I have chosen this one for several reasons.
As a practical matter, going substantially later would have expanded the available
evidence too much to be able to handle (the reader will have to decide whether I have
already bitten off more than I can chew). A cut-off around 300 ce also corresponds
to a fairly traditional sense of “classical” (that is, non-Christian) Roman culture.
That would not obviously be relevant in itself, but I suspect that it is connected
to another reason for the cut-off. It appears to me, on the basis of evidence I have
admittedly scrutinized less carefully, that there is an information technology revo-
lution in Late Antiquity. There seem to be significant changes to the technologies
described in most of my chapters (and the invention of at least one important new
one) during roughly the fourth and fifth centuries. I will say a little more about both
the shape and the possible reasons (some Christianity-based) for this revolution in
the last chapter, but the topic seems to me to require separate treatment. The end
date of 300 will not be applied mechanically. First, evidence from later periods can
sometimes be used to cast light on earlier ones. Second, I will be fairly generous in
allowing myself to use evidence (especially inscriptions) that is not definitively dat-
able to within my period.
The other constraint will also be partial, perhaps even more so. Though I have
been speaking (and will continue to do so) of “Roman” information technology, the
4
4 { Mosaics of Knowledge
focus of some of the chapters will be specifically on the Latin-speaking world. That
is hardly an obvious line to draw in a historical rather than, say, a literary context,
so let me say a few words about why a linguistic distinction might be appropriate
to the particular topic and in what ways in which this limitation will and will not
be observed over the course of the book. Many of the technologies discussed here,
especially in those first two chapters, can be seen as specialized forms of literacy.
This seems to me to be the central insight of Jack Goody’s famous 1977 book, The
Domestication of the Savage Mind, and its chapters on the cognitive operations
enabled or encouraged by tables, lists, formulas, and recipes. My subject matter is
deeper and narrower than Goody’s, but I also differ from him in one methodolog-
ical emphasis. He tends to look more at what written technologies can do than what
they actually do. The former was important for opening up a field of inquiry in a
theoretical kind of way, but the historical specificity that I am aiming for seems to
require the latter. Seeing these technologies as an aspect of literacy accounts for a
restriction on the scope of this book. Though I will generally speak of “Rome” and
“Romans,” my focus will be on the Latin world when the devices in question are
used and transmitted by writing, and so we would not necessarily expect them to be
constant across linguistic boundaries.
Two examples might help show how this linguistic division clearly can work in
practice (one is from outside the realm of this study and the other will be treated at
length in c hapter 2). For the former, I offer the following observation on differences
in substance between contracts surviving from Roman Egypt correlated with the
language in which they are written (Alonso 2016, 65):
The contracts in question, even the ones in Greek, were concluded between Roman
citizens. The difference is thus not one of culture in general or even legal culture
but, rather, of language. Alonso suggests, plausibly enough, that the key factor was
reliance on notaries among the less privileged classes to produce these kinds of
documents. The second example has to do with tables (the organizational device,
not the piece of furniture). These are not, as it turns out, particularly common in
either the Latin-or Greek-speaking worlds, though they do appear in certain, very
limited contexts. One of those contexts in the Greek world was the display of as-
tronomical data of various sorts. Yet these do not appear in Latin. Roman authors
either give (some of) the same information in continuous prose or in lists or, in the
extreme case of Vettius Valens (second century), write in Greek themselves.
These examples are not meant to show that Latin and Greek information
technologies are entirely cut off from each other, a claim that would clearly be false.
Rather, I simply wish to illustrate that there are both theoretical and empirical
A Brief Orientation } 5
6 { Mosaics of Knowledge
offered in this area by others, they have typically been framed in terms of form.
What, for instance, does a “table of contents” look like? In the last instance, I am
more interested in function instead, but that cannot normally be observed directly.
Fortunately, as a practical matter there are nonarbitrary connections between form
and function, and in many cases the broader context can help us out. Different
ways of organizing data facilitate different operations on those data, and it might
be more or less plausible that anyone would need to consult a document in the
ways that are theoretically possible. Thus I will integrate ideas of function into the
“definitions.” That, however, will mean offering more elaborate specifications of
form and context to stand in as proxies for direct observation of function. This is
not, as I say, a way of giving a truer definition, nor do I suggest that people who use
the words differently in other contexts thereby fall into error. Rather, it is a way of
specifying the precise phenomena that I intend to investigate. I am interested, for in-
stance, in which table-like objects (in a broad, formal sense) serve the functions that
are most typically associated with tables. (Those “associations” will generally be
established by modern comparative and/or experimental research.) Which formally
map-like objects can serve map-related purposes? And so on for other key terms. In
most cases, as it happens, the result is that I end up focusing on a narrower range of
objects than might be expected, but occasionally (as with the case of maps) I will
be arguing on the same basis for taking a broader view than is sometimes taken.
For the reader’s convenience I conclude this introduction with a brief survey of
the individual chapters.
1. Lists. While simple lists are ubiquitous in the Roman world, this
chapter treats only specialized types: tables of contents, alphabetized
lists, indexed lists, and nested lists. Each of these is rare and is largely
or entirely restricted to quite specific contexts. For instance, tables of
contents are used only for miscellanies with little inherent structure,
nested lists for large and continuously expanding public records. (In
passing, I demonstrate that a number of proposed listing schemes,
such as the supposed numbering of theater seats, are mistaken.) These
sophisticated list-forms are used overwhelmingly to confirm or authorize
information rather than provide it in the first instance. Moreover, they
tend to arise in tandem with some physical process of collecting or
sorting data, not merely contemplation of data in the abstract.
2. Tables and Tabular Organization. Tables (in the narrow sense of
matrices with meaningful rows and columns) are also vanishingly rare.
The chapter contrasts a variety of areas in which they might have been
expected but are not in fact found (e.g., grammatical, arithmetical, or
calendrical tables) to the few where they are attested (e.g., centuriation
formae, military duty rosters). Tables tend to arise only in the context of
a combination of circumscribed expert communities of users/producers;
constitution rather than recording of data; contexts in which priority
A Brief Orientation } 7
8 { Mosaics of Knowledge
and thus to collect evidence for their limited (but indisputable) patterns
of development and use. (The vast majority of the objects discussed in
this chapter operate in the titular two dimensions, but there are a few
exceptions.)
6. Conclusion. The first section attempts to tie together the content of the
main chapters, particularly in terms of the themes listed immediately
following. Then I sketch out several areas of related interest that I think
are ripe for further inquiry in light of the conclusions of this book.
My subject matter is large and complex, so it should not be surprising that there
is no monocausal explanation for the state of Roman information technology. As
a result none of the following themes plays a role in every chapter. Still, all recur
frequently, and the ensemble, I hope, binds the work together:
1}
Lists
Perhaps the most basic information technology made possible by writing is the list.
Lists were ubiquitous in the Roman world, and I do not intend anything like a full
account of them here. The scope of this chapter will be restricted in two ways. First,
in keeping with the idea of “high” technology, I will be talking not about simple
lists but about versions that have been augmented in various ways: ordered lists,
indexed lists, tables of contents, and nested lists. Second, even within most of those
categories, I will not attempt an exhaustive survey of either themes or materials but,
rather, make selected points.
When I say that lists were “made possible by writing,” a reasonable objection
might immediately be raised. Human beings have of course always talked about one
thing, then another, then another. That is not just a matter of, say, the catalogs that
figure prominently in the earliest poetry of Homer and Hesiod; it is the linear nature
of human speech. But I want to insist that there is nonetheless something distinc-
tive about writing down sequences of information. Here, I follow Goody’s classic
discussion of lists, tables, and formulas as specialized technologies of literacy. He
argues that writing transforms this oral discourse, giving the proper written list a
variety of distinctive features:
These physical features in turn give rise to various practical consequences such as
increased abstraction, increased attention to categorization, and an invitation to al-
ternative arrangement or processing of the items listed. Not every written list takes
advantage of all, or even any, of these opportunities, and—as often—the distinc-
tion could be framed as a spectrum rather than as a binary opposition. Still, Goody
has identified a distinctive set of written forms, and convincingly connected them to
important differences in use and power. Some of these advantages derive from liter-
10 ally having the kinds of written forms specified; others could perhaps attach to an
Lists } 11
oral list within a more broadly literate culture. My interest lies in texts that are not
merely sequential but which also exhibit at least some of Goody’s “written” charac-
teristics, whatever other specific additions (e.g., indexing, ordering) are the focus of
each individual section of the chapter. (This same set of observations about writing
down will be important to the next chapter, as well.)
Ordered Lists
I begin with what I will call “ordered lists.” Of course, all lists have an order; that
might even be considered their principal definitional feature (even if, as I mentioned
earlier, readers do not have to respect that order). What I mean here is much more
specific than that. “Ordered lists” will be shorthand for ones in which the order of
the text itself is meant to inform the reader. This can happen in two rather different
ways. On the one hand, a list can be organized by reference to properties of the
words (or other individual entries) themselves (“word order”). Typically this helps
a reader find out whether an item is on the list at all (sometimes to be used as an
index). On the other hand, a list can be ordered in terms of some feature of the real-
world entities that are referred to by its entries (“topic order”). This can in theory
be used in the same way, but more often it is meant to convey something about that
feature—for example, “what is the fourth most populous city in Italy?”
In principle, word order could appeal to the length of the words or some other
feature, but in practice it normally means alphabetical order. Unlike the topics of
the rest of this chapter and the next, alphabetization in the ancient world has al-
ready been the subject of an extensive and valuable basic study. In addition to var-
ious specific points which I will note individually, the basic narrative of the next
two paragraphs will rely heavily on Daly 1967. With this framework in place it will
be easy to move quickly to my own analytical remarks and in particular to some
observations that will foreshadow many of the important themes of the rest of
the book.
Latin lists in deliberate alphabetical order appear to go back at least as far as a
joke in a play by Plautus—that is, the earliest extended Latin texts we have today
(c. 200 bce)—and continue through to the end of our period. (There are somewhat
earlier Greek examples, and Plautus may well be adapting one such in his text.) The
earliest examples, and perhaps the bulk of all those of our period, are alphabetized
only with respect to the first letter of each word. Occasionally we see alphabetiza-
tion through the second or later letters. (Full alphabetization appears occasionally
in Greek, but seemingly not in Latin until after our period.) There is not, however,
an overall trend over time toward more detailed arrangement.
There seems to be something ordinary or matter of fact about Roman use of
alphabetization. The joke in Plautus’ play does not call explicit attention to the
ordering. Rather, a character starts reeling off a list of (otherwise unremark-
able) names in alphabetical order. The humor seems to lie in having the audience
12
12 { Mosaics of Knowledge
recognize on the basis of form that the list threatens to extend to the end of
the alphabet (Asin. 864–6). It is unclear what fraction of the audience could
have gotten the joke—we do not really know the fraction of the population that
would have had the low requisite level of literacy—but the fact that Plautus took
the trouble to make a somewhat elaborate joke anyway suggests that many would
have. It’s also true that authors who use alphabetical order either do not bother
to explain what they are doing or use only a summary phrase to do so (“ar-
ranged by letters,” “in the order of letters,” etc.). This is in contrast to the use of
some other devices we will see later. Alphabetization is also a common feature
of school exercises, so it is not hard to see why it might be familiar to the general
(literate) public.1
At the same time, as Daly has noted, it appears to be used principally in contexts
which are anything but general or ordinary. He points out a number of schol-
arly uses; I would add only the case of Flavius Caper (Verb. Dub. 107–12K)—no
later than Daly’s earliest example—whose brief list of morphologically “doubtful
words” is transparently in alphabetical order. Daly also notes administrative uses.
He finds their number less than impressive, and that is a fair judgment, but I would
note that subsequent scholarship (not focusing on alphabetization as such) has ex-
tended the roster of examples at least a bit.2 It is still not common in either realm,
particularly in “outward-facing” texts—that is, those aimed at a reading public dif-
ferent from the producer(s).
I am inclined to explain the gap between high accessibility and low actual use
by another phenomenon that Daly 1967.52 had already noted: alphabetization
was inelegant. The contexts in which it typically appears (education, handbooks,
internal state documents) are all relentlessly utilitarian. The two texts of some pre-
tension which nonetheless use alphabetical order resort to it only in residual cases;
it is how “the other” items in various categories are listed once more substantive
categorization has been exhausted (Var. RR 1.1.9; Plin. HN 3.46, 26.164, 37.138).
And at one point Pliny even apologizes as he begins a shorter group by noting
that he is only alphabetizing peoples “who do not warrant closer attention.”3 Daly
notes in this connection that the agricultural writer Columella, in borrowing two
alphabetized lists of authors from Varro, re-sorts one by birthplace and appears to
invert two persons in the other simply to avoid alphabetical order.4 There is also a
well-known passage in the Aeneid in which Italian heroes are listed in alphabetical
1
Cribiore 1996.161–7. Frampton 2018.62–70.
2
Nicolet 1991.135, 173–7; Salway 2012.200–2. Tim Parkin points out to me that the alphabetical list
of centenarians from Italy’s regio VIII preserved in Phlegon, Makrobioi (taken with Pliny the Elder’s
access to a similar data set; HN 7.162–4) could reflect some government document ordered on the
same principle, though (as he also points out) alphabetization by praenomina seems very odd in such a
context.
3
Plin. HN 3.130, quos scrupulosius dicere non attineat.
4
Perhaps also relevant here is the frequently made observation that Latin authors who receive a list
in Greek alphabetical order often fail to rearrange it to maintain that order in Latin.
Lists } 13
order except for the intrusion of Messapus between Caeculus and Clausus. O’Hara
1989, however, has shown that Vergil tells Messapus’ story in such a way as to as-
similate him to another well-known hero, Cycnus. Cycnus’ name would be in cor-
rect first-letter position here. Vergil, too, seems to be going out of his way to avoid
alphabetization.
There is also an exception which may prove the rule. The Augustan-era scholar
Verrius Flaccus wrote an extensive lexicon, perhaps closer to a modern encyclo-
pedia than a dictionary. Our preserved versions have gone through some kind of ed-
iting at the hands of Festus (perhaps second century; very fragmentarily preserved),
and large parts survive only via the subsequent epitome of Paul the Deacon (eighth
century), but it seems likely that the basic structure is Verrius’. The work is organ-
ized by first-letter alphabetization of the head-words, and within each letter group
there is a first section that is further alphabetized and a second one which is grouped
more or less thematically. Glinister (2007.23–4, 29–32) has made a strong argu-
ment that alphabetization is not a pragmatic choice here. The work has no cross-
references, and many discussions are attached to lemmata that are not necessarily
obvious. The innocent reader would find it very hard to track down a particular bit
of information, and even the expert might be daunted. Verrius, Glinister argues,
uses alphabetization’s blandness to assert intellectual authority, the authority to
distribute knowledge according to a system that is superficially objective, but in fact
quite idiosyncratic.
There are several patterns here that we will see throughout the book. First, al-
phabetization demonstrably exists in the Roman world, but its use and diffusion is
perhaps surprisingly limited. This quantitative phenomenon also has a qualitative
parallel. Romans do develop more “sophisticated” or powerful or (to put it less
prejudicially) intensive alphabetization, but the more complex forms do little or
nothing to displace the simpler ones. One should never cite technological progress
as an explanation for anything, but from a modern perspective it is probably useful
to point out that progress does not work even as a description here. In particular,
in the variation between one-letter, two-letter, and so on alphabetization, we see
a Roman tendency not to apply any more technology than is required by a par-
ticular situation. Second, the development is generally sparse, but also clustered.
Alphabetization is relatively common in a few contexts and unknown in others
where it could have been used. Additionally, I have mentioned an apparently aes-
thetic objection to alphabetization, and we will see other cases of resistance to for-
malism in data organization.
The last pattern is slightly more complicated. Small 1997.63– 5 has
hypothesized that the rarity of alphabetization is due (in part) to the instability
at the time of the notion of the alphabet itself. Following her, Glinister 2007.22
has gone so far as to say alphabetization “is not an obviously useful means of
arranging data in a world where ‘alphabetical order’ may fluctuate, where there
appears to be no concept of the alphabet as a distinct entity.” Subsequent work
by Frampton 2018.67–70 has shown that the premise is false—the alphabet was
14
14 { Mosaics of Knowledge
very much reified by the period we are talking about—but in any case, the evi-
dence cited here for the ordinariness of alphabetization tells against such a deep
cognitive explanation of the weakness of the technology.5 Cognitive factors
will be significant throughout the book, but will rarely give simple, categorical
answers to our questions. Here we should favor shallower and perhaps more di-
verse explanations.
The other approach to ordered lists is by sorting the topics, not the words. In
principle, the potential variety of topic orders is infinite; one could imagine, for
instance, lists based on the date, location, or size of the items listed. There are
many such lists surviving, at least at first glance, but it is often difficult to know
that the objective order is meant to have (or in fact had) any informational value
for the reader rather than being an artifact of composition.6 For instance, the list
of canonical authors recommended by the rhetorician Quintilian (late first cen-
tury) is at least roughly in chronological order within each genre (10.1.46–131), but
there are no absolute dates or even (for the most part) indications of gaps in time.7
Moreover, chronology applies only within genres, and those generic categories are
in turn subordinated to a division between Greek and Latin texts. Chronology may
have helped Quintilian pick an order, but it is so far down the organizational hier-
archy that it could only help a reader find an author that she is nearly on top of
already. Or take the lists of cities on the pocket sundials recently studied by Talbert
2017.147–52, which instruct the user how to calibrate the timepiece for use at the
local latitude. Many show rough clumping of nearby places, and some may reflect
itineraries, but do not appear to follow these tendencies precisely, which means that
the order is not in fact predictable. Nor is it clear what value the sundial reader
would get even from knowing a correct order. All she really needs to know is her
current position. For an author, though, ordering even a fairly simple list might
make sense as a check for completeness.
In the two cases just mentioned, there is good reason to think of the lists as not
being ordered in my restricted sense, but more often there is little clear reason to
guess one way or the other. However, I can suggest two sets of circumstances in
which order is more likely than not to be important. On the one hand, we have a
number of kinds of practical records such as financial accounts, rosters of mili-
tary units (which are arranged by rank, then by year of enlistment), and various
5
More plausible is the notion (Glinister 2007.22–3) that the syllable was more prominent in the
Roman mind, and so may have made limited alphabetization more natural (cf. Frampton 2018.63–4 on
the status of syllables). However, the order of syllables seems still to depend on the order of the compo-
nent letters.
6
Cf. Riggsby 2007.96–8 for structure in HN that seems to be a residue of Pliny’s methods of compo-
sition, likely invisible to his original readers, who had neither the modern paratextual apparatus we do
nor (I suspect) professional interest in reverse-engineering Pliny’s working methods.
7
It might fairly be objected that what we have in Quintilian’s text is not really a list, since it fails
in most respects to show the Goody features described a few pages ago. I imagine that Quintilian was
working from an outline in the form of a much rawer list. If the end product was not really a list itself,
that just suggests all the more strongly that the ordering was meant for the author not the readers.
Lists } 15
government registers. The registers record, for instance, births in the order received
and note both the date and the position of the record within the sequential register.
We will see all these document types again soon, but let me note here one feature
they share (which itself will also be important later). In all of them the chrono-
logical order seems to be generated automatically, as the records in question are
compiled over time. That is, for instance, a small business owner might list pay for
his employees for a given day, then repeat the process on the same piece of papyrus
the next day and the next, and so on.8 The fact that the individual records already
in order in these cases are tagged with the date in question suggests that the order
of composition was deliberately being harnessed to provide the user with chrono-
logical information.
On a grander scale, the various state annals seem to have been constructed in the
same way, with various kind of information (e.g., lists of prodigies) recorded in order
at least yearly.9 I separate these out because they point, at least by association and
perhaps also causally, to the second set of contexts in which pointedly ordered lists
are used. These are works of what might broadly be called scholarship. Most obvi-
ously there is Rome’s entire tradition of annalistic history, but we might also look at
Cicero’s Brutus, which writes the history of Roman oratory mainly through a series of
capsule biographies arranged in generations of orators. Unlike Quintilian, Cicero ex-
plicitly points to his chronological arrangement, and he uses it to make a point about
what he sees as progress over time. In a less literary vein, we can point to the lists
of officeholders, military triumphs, and calendrical information Romans carved into
stone and which go under the cover term Fasti.10 We do not generally know who com-
posed these, but the one exception (Fasti inscribed at Palestrina and composed by the
first-century bce scholar Verrius Flaccus11) is likely representative. The people who use
topic-ordered lists, then, look like roughly the same people as who use alphabetical
lists. They are scholars and bureaucrats who frequently deal with information as such
(without necessarily intervening in the management of whatever reality the data de-
scribe) and who interact intensively with each other more than speaking to a broader
public.
Indexed Lists
“Indexed lists” might include in principle any whose members are matched one to
one with a set of indices—numbers, letters, colors, and so on—to help find material
8
For a specific example, see chapter 2, note 46.
9
For the Annales Maximi, see Cic. de Or. 2.52 and DServ. Aen. 1.373, with Frier 1979. For other
magisterial (and thus annual) records, see the references cited at Riggsby 2006.134. Most or all of these
will have been internal rather than published records (Riggsby 2006.140, 149–50).
10
On Fasti, see further at chapter 2 “Appearance: Fasti”.
11
Suet. G&R 17.
16
16 { Mosaics of Knowledge
within the list. As with the ordered lists, however, this type is dominated by a single
sub-type: the numbered list.12
The fact that labels are almost always based on numerical order creates a poten-
tial evidentiary problem when we do not have the entire list in hand. If we have a
passing mention of, say, the fifth item on the list, does that mean that the whole list
was fitted with numerical references (and so is a truly indexed list) or did someone
just happen to count forward from the beginning on a particular occasion?13 This
distinction will often not be a real problem, but it will be significant in at least a few
cases that will come up in what follows. Thus I want to stipulate that for the rest of
this section I mean only lists with built-in indices, not ones generated on the fly and
partially by a later user.
Compared with most of the other technologies discussed in this book, numbered
lists are fairly common. Here are some contexts in which they are used (most of
these will be discussed at more length at various points in the book):
• sections of archives
• segments of buildings
• administrative regions of Rome and of Italy
• military units of various sizes
• milestones along roads
• lots
• units of time
• segments of texts
In what follows I want to argue for the existence of two limitations on num-
bered lists. One limitation is (like the items in the list) essentially a matter of genre
or use-context. The other is more functional, a way that numbering of lists is not
exploited, even when it exists.
12
Two (or three) exceptions: the Roman market (nundinae) cycle divided an eight-day “week” into
days labeled A–H; a work by Casellius Vindex (cited by Charisius 254.18, 312.24B) was arranged not
in books 1, 2, etc., but A, B, though here the letters appear to be literal; each book discusses words
that begin with the letter in question. Years were often identified by the names of (at Rome) the year’s
consuls or (elsewhere) various other “eponymous” magistrates; Julius Obsequens’ book of prodigies is
essentially a list structured in this way. But the unpredictability of these names, however, means that
they cannot be used as an index in the same way as numbers or letters.
13
There are also at least a few cases that turn out to be neither. Roman legions were conventionally
designated by numbers. At various points in the early empire there were two fifth legions, two sixth,
two tenth, and (it appears) three third legions (generally distinguished by further epithets); the numbers
are clearly not generated by counting, but because they are often repeated, they do not really work as
indices at all. Also, probably, the tokens known today as spintriae. These feature large numerals on one
side and erotic scenes or imperial portraits on the other. It has been hypothesized that they correspond
in various ways to a catalog of sexual positions or the like, but Campana 2009 makes a strong case
for regarding them as game tokens, with the numbers perhaps representing the value of the individual
piece. The same is likely the case for similar tokens which match numerals on one side with counting
gestures on the other (Williams and Williams 1995.590–1).
Lists } 17
The last item on the foregoing list was “segments of text.” (I mean here indi-
vidual texts; archives will come up later.) By the time we have longer surviving lit-
erary texts in Latin, they are divided by their authors into “books”—that is to say,
units that will notionally fit on a single papyrus roll—which are then numbered.14
(I use “literary” here in a broad sense common in the Classics—texts which were
meant to be copied and circulated to an indefinite audience, not just those of some
artistic ambition.) Given this nearly universal convention, it is then striking that
smaller units of Latin texts are seldom if ever numbered. In principle, several such
units were available. Today we refer to bits of Latin poetry by line number and (in
the case of collections of short poems) the number of a given poem within its book.
There seems to be no evidence for either practice in antiquity, whether directly in
manuscripts or in external reference (say, in commentaries). This is despite the fact
that lines were often counted in individual manuscript copies, apparently as part of
the pricing process.15
Prose works were often thought of as being divided into capita, a word
often translated as “chapters,” but perhaps closer to English “sections” or even
“paragraphs” (there is a related, perhaps original, sense of “high points”).16 Not
only do we have frequent reference to the general idea, but there is even some man-
uscript evidence for the marking out of such sections with ekthesis (reverse inden-
tation) or even the letter k (for kaput, as the word was often spelled).17 Crucially,
however, none of these divisions appears to come with reference numbers, not even
in prose texts explicitly divided into units by rubrics. Latin also has a word pagina
which can mean the same thing as the English word we derive from it: page. Before
the rise of the codex, however, it typically refers to a field within a larger support
medium—say, a column of writing on an inscription or roll of papyrus. In no sense,
however, are paginae used in our period as the basis for standard numeration of lit-
erary texts (we will see later that the case of archives is somewhat different).18
The lack of numerical references (other than book numbers) in commentary is
particularly striking.19 The genre was a flourishing one already in antiquity, with
subject texts including Vergil, Cicero, Terence, Homer, and others. In its modern
form, the commentary is typically organized by numeration first (line numbers
for verse, section/chapter numbers for prose), then by rubric within any given nu-
merical range. Ancient commentaries are generally organized by rubric alone. In
14
Moatti 1997.223.
15
Birt 1882.159–209; Hall 1913.9, 13.
16
On the terminology, see Butler 2009.16–7, 20.
17
Butler 2009.10–5, 18–9, 21–3.
18
Grafton 1997.30 cites references to page numbers in an apparently standardized edition of certain
legal texts, but this dates from the fifth century. Shafer 2017 has argued that certain effects in Vergil’s
Georgics presuppose that certain words and phrases will fall at predictable points on the “page,” though
for poetry this would only have required an early edition with a set number of lines to the page.
19
For possible (but still extremely scanty) Greek examples, see Barnes 2015.348–50, Mansfeld and
Runia 2009.199.
18
18 { Mosaics of Knowledge
continuous reading, there is little difference, but if you pick up a literary text in
the middle, it is much harder to find your place in an ancient commentary than
in a modern one. For comparison, it may be useful to look at one very partial
exception to the usual ancient practice. In the manuscripts of Asconius’ histor-
ical commentaries on several speeches of Cicero (first century), we have frequent
(though not systematic) allusion to the position of his lemmata within Cicero’s text:
This seems to be unique among commentaries. In fact, it has even been suggested
that these headings were not part of the original text but, rather, were introduced by
a later reader for his own benefit, then absorbed into the manuscript tradition. I will
return to the reasons for this proposal in a minute. Whatever their source, these
references obviously do not refer to standardized numerical labels. They are vague,
they vary choice of units (lines, “parts,” none at all), and they count in whichever
direction is locally convenient.
The same approach seems to be visible in a text which is not literary even in the
broad sense used here but also which bears on the more general issues of both this
section and the last section of this chapter. In the early second century, a man named
Vesbinus erected an inscription at Caere which recorded acts of the municipal gov-
ernment and subsequent correspondence with an imperial official authorizing him
to set up a shrine. His inscription quotes extensively and with citations from the
records (commentarius) of the municipality. The sections with the reference (as they
are usually translated) are as follows:
in which [commentarius] had been written this which is written below: In the
consulship of L. Publilius Celsus and C. Clodius Cripinus [i.e., 113], on the
Ides of April, in the [local] dictatorship of M. Pontius Celsus and aedile-
ship of C. Suetonius Claudianus, the commentarius of the municipality of
the people of Caere. From there (inde) page 27, sixth chapter. [provisional
approval of the project subject to higher authority] From there page two, first
chapter. [Letter to the relevant imperial official] From there page eight, first
chapter. [Reply from that official] (CIL 11.3614.6–18)
In all three cases, the reference is to page and chapter. Birt 1882.158 takes the page
and chapter numerations as running independently, but in parallel, so pages 2 and
8 would fall within c hapter 1, while page 26 lay somewhere in c hapter 6. This hy-
pothesis ignores the fact that the three documents—the initial decree calling for
Lists } 19
(inter alia) the approval from the curator, the letter making this request, and the
curator’s reply—are clearly in chronological order, as they would have been in the
commentarius which is a running record.20 Birt’s scheme makes the initial decree
the last of the three to appear in the commentarius. In fact, the same objection still
applies even if we take pages to be individually broken down into chapters. I sug-
gest that the page references have wrongly been translated as “page 27/2/8.” They
should instead be read as “the 27th/next/eighth page” from the previous citation
(an equally possible reading of the Latin). That is, inde (“from there”) doesn’t refer
generally (and thus redundantly) to “from the commentary” but, more specifically,
“from the last page mentioned.” The locations of the second and third documents
are given by counting forward from the first and second documents, respectively.
The first instance appears to count from a more public reference point marked out
by the list of eponymous officials (perhaps some important date in the civic year).
Note also that this material is placed after “what is written below” and so looks to
be itself a quotation from the record. Moreover, the first date quoted (13 April) is
very early if we take it to be the time of the original council decree, since the letter
they ordered dispatched was not sent until 13 August (line 15). Rather, it is a ref-
erence point considerably before that decree, as represented by the twenty-six-page
gap. It does not look as if this record had fixed numbers attached to its paginae, just
ad hoc counting. In principle the capita could have been numbered, but that seems
much less likely if there was no indexing at the higher level. This is made possible by
the fact that there was, seemingly, only a single exemplar of the text.
All that said, I know of two circumstances in which texts do have numbered
subsections: one in a particular text and the other an entire category. First, in the
mid-first century, Scribonius Largus wrote a collection of recipes for 271 medicines.
The individual recipes are numbered, and the opening table of contents uses this
numeration to direct the reader from symptoms to their proper remedy. I will say
more about why this might be at the end of this section, after the other example
and some of its implications. Second, though arguably not literary texts even in
the broad sense I have offered, Roman statutes were often divided into numbered
capita.21 For instance, Cicero takes for granted that he can refer to these capita by
number even when speaking of a legislative proposal not yet passed:
Why does it matter in chapter 3 you require the ratification of officials by the
passage of a law by the curiae, when in chapter 4 the elected are given the
same powers even without such a law. (Leg. agr. 2.29)
20
Riggsby 2006.138.
21
The tablets on which a few statutes are inscribed seem to be numbered, but this is almost certain
to be for local use in placing the tablets, not a general reference system. The Flavian municipal law does
not even show consistent formatting at lower levels, so it is unlikely the system assumed uniform carving
(González 1986). In the enabling law for Gallia Cisalpina (CIL 11.1146) the (apparent) tablet numbers
do not respect breaks in the chapter numbering. Cf. Butler 2009.16–21.
20
20 { Mosaics of Knowledge
Similar references appear sporadically in inscriptions and in the later juristic lit-
erature, and there is plausible reason to think that such numeration had at least
appeared centuries earlier as illustrated by the three numbered capita of the lex
Aquilia.22
Note, however, that in the Cicero passage I have just quoted, the reference to
chapter numbers is not strictly necessary, since the relevant material from the cited
source is paraphrased anyway. Looking up the proposal Cicero refers to could con-
firm what he says, but that appears to offer authority rather than information. This
illustrates the second, functional limitation I alluded to earlier. There is a strong
Roman norm against what I’ll call “obligatory cross-reference.” That is, they do not
like to force readers to look up a second text to be able to understand what they are
reading in the first. Here is a modern example of an obligatory cross-reference of
the sort I contend the Romans avoid:
This means nothing without consulting other sections of both the Texas penal code
and the U.S. federal code. This is entirely normal legislative drafting. Here is a
Roman example, which I suggest is extraordinary:
There are several points to be made about the Roman example. First, it is the
only obligatory cross-reference to a numbered chapter in the surviving remains
of Roman statute law. And I have not yet found another example in the ancient
legal literature more generally. Again, a reference to provide authority is common
enough, but not ones that must be tracked down to figure out what the immediate
passage is saying. Second, the very text in which this reference appears would itself
be difficult to refer to in the same way. It is an instance of the so-called Flavian mu-
nicipal law. The latter is a template for town charters in late first century ce Spain.23
We have fragmentary instances of laws based on the same template from other
towns in which the chapters are numbered. At Irni, however, the only place where
this quoted passage is preserved, the chapters of the law are identified by rubric, not
by number.24 In fact, as a general matter, the chapters of inscribed laws are more
often marked off by summary rubrics, by spacing, or by nothing at all rather than
22
Gaius 3.210, 215, 217; Rotondi 1966[1912].241–2.
23
González 1986.
24
González 1986.148.
Another random document with
no related content on Scribd:
SOUTHWORTH LIBRARY
⚜ ⚜ ⚜ ⚜ ⚜ ⚜ THE EDEN SERIES ⚜ ⚜ ⚜ ⚜ ⚜ ⚜
Only Complete Edition of Mrs. Southworth’s Novels
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside
the United States, check the laws of your country in addition to
the terms of this agreement before downloading, copying,
displaying, performing, distributing or creating derivative works
based on this work or any other Project Gutenberg™ work. The
Foundation makes no representations concerning the copyright
status of any work in any country other than the United States.
1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if
you provide access to or distribute copies of a Project
Gutenberg™ work in a format other than “Plain Vanilla ASCII” or
other format used in the official version posted on the official
Project Gutenberg™ website (www.gutenberg.org), you must, at
no additional cost, fee or expense to the user, provide a copy, a
means of exporting a copy, or a means of obtaining a copy upon
request, of the work in its original “Plain Vanilla ASCII” or other
form. Any alternate format must include the full Project
Gutenberg™ License as specified in paragraph 1.E.1.
• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”
• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.
1.F.