3
Networks: Exile and Tourism
in the Roman Cyclades
Rebecca Sweetman
Attracted by the fame of the place, I was eager to come within sight of Delos, and
seemed to traverse the deep with languid pace. How oten did I chide the oars, as
slow in bearing us along? How oten complain that our sails were not stretched
by the stinted blasts? And now I had passed Mycone, Tenos, and Andros, and
bright Delos was within view; which I no sooner saw, than I cried out, Why does
the island seem to ly by me?
—Ovid, Heroides 21.77–84
he prevailing view of the Cyclades in the Roman period comes from literary sources that assert an
inaccessibility and insularity, as for example in Catullus, “Dare not deny her, insular Cyclades.”1 he
image of remoteness was further underscored by stories of piracy and Pompey’s eforts to eradicate
it.2 It is not surprising, then, that the islands were famed as places of exile, particularly in the imperial period.3 While the view of the islands from the ancient sources is overwhelmingly negative, the
modern perspective is also problematic. In practical terms, little is known about the administrative
history of the islands, and the islands are oten erroneously considered a single entity rather than
individual lands. he fact that it is diicult to ascertain to what province various islands belonged
at diferent times suggests that the degree to which the concept of a Roman province can enhance
analysis of the Cyclades is questionable;4 their shared administration did not necessarily create
bonds or a collective identity among the islands. A collective identity was lacking in other jurisdictions as well, for example in the joint province of Crete and Cyrene, where there appear to have
been few points of commonality and little surviving evidence of a shared visual culture.5 And the
labels that have been used to deine roles or relationships, terms such as province and colony, can
be misleading in that they prompt assumptions of how places should work or relate to Rome based
on categorization.6 A good example of this confusion occurs with Knossos, Crete’s only Roman
colony, where results of excavations have been interpreted through the lens of expectations for
Roman colonies.7 In the top-down approach of past scholarship, if there was little evidence that a
region conformed to the idea of a Roman province, that area was sometimes viewed as being of
little consequence—so, for example, Crete was seen as a backwater and the Cyclades as insular.8
46
he dearth of literary sources and the historical stereotyping of the Cyclades as marginal overshadowed the drive to understand the actual material evidence.
When examined without such preconceptions, however, the archaeology of the Cyclades
indicates that life on the islands during the Roman period was quite dynamic. Studies of the
agorai in Melos, hera, and Andros, for example, all indicate an energetic and escalating economy
with rebuilding and investment, particularly in the second century.9 here were at least eight
statue bases dedicated to Hadrian, revealing the positive attempt to gain imperial attention, and
the macellum of the Agora at Palaiopolis in Andros may even have been dedicated to Marcus
Aurelius.10 An erroneous concept of the unity of a province had encouraged the analysis of the
Cyclades as a single entity, a place of insularity and exile, thus neglecting the multiplicity of roles
and diversity among the islands and any real evidence for active participation in the empire.11 A
richer and more complex picture of the Cyclades comes into focus when individual islands and
their roles in the Roman network are analyzed.
I have argued elsewhere that a range of agents and motivations (local and central) were
needed in order for a place to become actively part of the Roman Empire, whether through trade,
the military, politics, administration, or personal connection.12 Analysis of the Cyclades makes it
clear that a range of diferent network connections, from personal to public, played a fundamental
role in afording opportunities to diverse islands, regardless of their provincial designation. he
islands were not a monolithic group; while some may have taken advantage of network connections, others may not have done so. hrough an analysis of personal interaction, exile, and tourism
and an examination of the archaeological data, this essay aims to assess the extent to which the
Cyclades were really isolated and to consider a range of roles individual islands may have played
over time in the Roman network. As Ovid wrote, once the individual islands are explored it is
possible to see how dynamic many really were.
A Brief history of the roman Cyclades
he insularity of the Roman Cyclades has become a self-sustaining trope along the lines of the
supposed austerity of Classical Sparta, and in both cases past scholars were deterred from undertaking systematic studies of the “poor” periods of these areas.13 For the Roman period, with the
exception of work on individual sites (Palaiopolis on Andros), surveys (Melos), and onomastic studies, few large-scale eforts have been attempted.14 Roman material from some individual
islands, such as Keos and Melos, is reasonably well known, but considerable quantities of Roman
material from other islands, such as Andros, hera, and Kythnos, remain underused in broader
studies.15 Unhelpfully, much of the work on the history and archaeology of the Cyclades in the
Roman period is typiied by issues of scholarly disagreement over which islands were part of which
province and at what period.
R. L. N. Barber has noted that signiicant Classical centers of the Cyclades—such as Klima
on Melos, Palaiopolis on Andros, and hera on Santorini—were continuously occupied from the
Geometric until the end of the Late Antique period (seventh century).16 he Cycladic islands were
almost always controlled by overseas powers, whether by Classical Athens and Rhodes, Hellenistic
Macedonia or Egypt, or inally Rome.17 During the Republican period, Sulla gave some of the
islands to Rhodes as a reward for its support during the First Mithridatic War. Marcus Antonius
gave Andros, Tenos, and Naxos to Rhodes, while Keos went to Athens.18 By the irst century Bce,
Networks
47
most of the islands seem to have been in the Roman province of Asia except Keos and Delos,
which belonged to Athens and were thus part of Achaea.19 Under Augustus, many of the islands
were given to Achaea; others seem to have been let in the province of Asia.20 he islands did
not become politically uniied as a single province until 294 ce, as Diocletian’s provincia insularium under a praeses insularum. But even then the arrangement excepted Keos, Delos, Kythnos,
Mykonos, Seriphos, and Syros, all of which remained part of Achaea. In terms of the management
of the islands, it is likely that most were relatively independent as long as dues were paid to the
Roman state.21 Onomastic data suggest that tax collectors and merchants were regular visitors to
the islands.22 While there are some political names, their presence on the islands seems to have
had inancial or religious purposes.23 It is known that Marcus Aurelius sent two imperial legates,
C. Vettius Sabinianus Iulius Hospes and L. Saevinius Proculus, to the Cyclades to take care of economic activities.24 here were other low-ranking oicials connected with cult foundations, as for
Mithras on Andros (by M. Aurelius Riinus, an evocatus Augusti) and for Isis on Syros (by Claudius
Secundus, a viator tribunicius).25 Similarly, the relatively higher-ranking oicials in evidence on
Tenos were associated primarily with the sanctuary of Poseidon and Amphitrite at Kionia.
Most of the Cycladic islands were fairly self-suicient in terms of agricultural and wine
production, and some expanded their remit to other industries. For many of the islands the irst
century Bce was a watershed, with some islands emerging as more successful than others. During
the Hellenistic period, islands such as Delos were well integrated, and many of the nearby islands
(Tenos and Andros) beneited from their proximity to this network hub. Others, such as Paros
and Seriphos, did well economically because of their natural resources and were tied into the trade
network. Signiicant alterations to the islands’ established contacts came in the irst century Bce
with attacks like those by Mithridates on Delos and by the Romans on Seriphos (in 84 Bce). Delos
never recovered from this period of crisis, and aterward even some of the islands with natural
resources, such as Kythnos, Keos, Siphnos, and Seriphos, appear not to have been exploited as
fully in the Roman period.26
During the second century ce, the Cyclades, like other areas of Greece, seem to have
prospered. Strabo mentions Andros, Naxos, and Paros as important and implies the same for
Melos by comparison to its neighbors.27 Unsettling events in the third century may have contributed to the abandonment of some islands such as Delos.28 Recent data indicate that the Cyclades
were Christianized somewhat earlier in the Late Antique period (fourth century) than areas such
as the Peloponnese and Crete.29 Paros, Siphnos, and Amorgos were also episcopal seats, and the
islands appear to have been as prosperous as other areas around the eastern Mediterranean until
at least the seventh century, when many of the islands were raided by the Slavs.30
networks
Network analysis measures levels of diversity and connectivity over a diachronic period. Networks,
which can be static or dynamic,31 are deined as the means through which information is shared;
there are multiple types, including social, infrastructural, and biological. Networks can be created
through personal connections (such as exile, tourism, religious experience) and/or may be organizational (such as trade or information exchange). Networks can be controlled or mediated
by diferent groups or individuals, and they can be generated organically or manipulated.32 As
already indicated, networks may be scale-free, such as the Roman Empire, or small-world (local
48
Sweetman
ones).33 An examination of networks provides a better understanding of the individual places and
overall organization, and assesses diversity of relationships over time and space in accordance with
varying contacts created by diferent means and people.
Underpinning a network analysis is the idea of phase transition—the idea that wide-scale
communication between locations (hubs or nodes) will create further nodes.34 Hubs are highly
connected loci that are not necessarily signiicant cities, while nodes are locations that connect
small-world and scale-free networks. Evidence of phase transition clariies the level of participation in the wider Roman network.
Application of network analysis to archaeological evidence has developed signiicantly
in the last few years, particularly through the work of Tom Brughmans and Carl Knappett.35 Both
social networks of people (particularly cratspeople), and nonsocial/material networks (artifacts)
have been the subjects of analysis.36 In the case of Middle Bronze Age pottery in the Cyclades, for
example, the evolving networked relationships between artifacts, people, and space have been
examined with a focus on economic evidence for cultural transmission and the organization of
individuals.37 Evidence of economic exchange might be measurable by the existence of a product
in demand as well as by imported goods. he use of locations as stopover points is more diicult
to ascertain, although the presence of luxury or imported items in an otherwise subsistence-level
economy might indicate a transfer.
Deining the type of network allows for more complex interactions to be recognized (and
not merely as simple linear connections) in the Cyclades. As networks can be relational and not
necessarily categorical (that is, they do not have to be grouped according to common characteristics) , they allow for luctuation in interactions among island communities.38 Communication hubs
such as Melos may have connected the small-world network of the islands with the larger scalefree network of the Roman Empire. his nodal relationship would have enabled less-structured
communication links for islands that may not have exploited their connections to make them more
active or regular players in the larger Roman network.
Levels of isolation changed over time and degrees of insularity might have increased or
decreased even on a seasonal basis, as certain weather conditions may have made it necessary
to keep the sailing times shorter with more stops at islands.39 Travel across the Mediterranean
from west to east and from north to south was straightforward, but prevailing winds made the
return journeys far less predictable.40 hus a range of network connections might have been made
through casual visits, location on trade routes, or even nebulous weather. It is arguable that many
of the Cycladic islands were actually quite well connected to each other (small-world network)
and the center through personal interactions. Because only a small number of islands exploited
connections to become part of the larger Roman network society (scale-free network), news of
the majority of the islands stayed largely local or personal, thereby promulgating the idea, though
not the experience, of static seclusion.
Network analysis enables the recognition of connections on a macro (town) and micro
(individual) scale. he study of the Egyptian trade focuses on Red Sea coastal sites like Berenike,
where the evidence of being a hub and part of the scale-free network is seen in the variety of oferings in the sanctuaries within the towns, while the evidence for the small-world network is seen in
the spread of imported objects in satellite villages.41 On the other side of the empire, in the northwest, evidence for network connections can be found in imagery on coinage, although the exact
details of the functioning of the networks are opaque.42 In this essay I examine similar network
connections to rethink the nature of the Cyclades in the Roman period. he focus will be on the
Networks
49
personal networks and on whether or not they contributed to empire-wide structuring principles
that allowed access to a local rather than imperial perspective.43 Personal or emotional exchange
is not easy to measure in archaeological contexts, but in the case of the Cyclades, epigraphic data
and characterizations of exile can contribute to such analysis. his essay stands alongside others
in this volume that chart elements of material culture, with a focus here on the evidence for the
interconnectedness that impacted these islands, particularly in the contexts of political exile and
religious tourism.44
exile and tourism
According to Mary Braginton, 150 people are known to have been exiled in the imperial period,
but the actual number is likely to have been much higher.45 he nature of Roman exile changed
over time; in some cases remoteness was key (Augustus restricted those exiled to islands that
were “not less than 50 miles from the mainland,” and travel was forbidden); in others it was preferable to send an outcasts to well-populated places (like cities) so that they could be monitored—
for example, when Cicero went into exile to hessaloniki, or Helvidius Priscus to Nicopolis.
Despite the view that the remoteness of the Cyclades made them suitable places of exile, not all
islands used for deportation were small and remote. And there were diferent levels of severity
of exile. Dio Chrysostom, for example, who was exiled from Rome, was able to travel widely.
Others were restricted to one location. he evidence suggests that execution while in exile—as in
the case of Flaccus, sentenced to execution while in exile on Andros in 39 ce—was uncommon.
Although in most cases exile meant loss of citizenship, there were diferent levels of property
appropriation. In exile Ovid and Seneca kept their property, for example, and Lollia Pavlenia
had a large allowance.
James Scott notes that negative elements of being in exile can change (even while still
in exile) to a positive experience of diaspora.46 In some cases of Cycladic exile certain deportees
seem to have lourished. Glitius Gallus and Egnatia Maximilla, for example, ater being banished
to Andros by Nero, settled into their new residence with great lair, invested in their new homes,
and were held in high esteem by the local Andrians, who erected honoriic monuments to them.47
In contrast, Cassius Severus died in squalor ater having spent fourteen of his twenty-ive years
in exile on Seriphos, and when Vibius Serenus was recalled to Rome from exile in Amorgos he
arrived in rags.48 It is unclear whether there were multiple deportees occupying the same islands
at the same time and whether particular islands were favored at diferent periods.
he main islands used for exile were Amorgos, Andros, Gyaros, Kythnos, and Seriphos.
here are also occurrences of exile at Naxos and Syros—for example, in 65 ce Lucius Silanus
chose to be exiled to Naxos, and a false prophet was banished to Syros by Marcus Aurelius ater
the attempted usurpation of Avidius Cassius in 175 ce—but these are less frequent.49 Juvenal
likened imprisonment to being between Seriphos and Gyaros and described Gyaros as a vile
place,50 which indicates that certain islands were considered particularly dreadful places for banishment. Some families petitioned for more lenient sentences for individuals banished to Gyaros;51
for example, although they were originally sentenced to exile in Gyaros, Caius Iunius Silanus
was ultimately banished to Kythnos by Tiberius (23 ce) and Avillius Flaccus to Andros by Caius
(39 ce).52 Seriphos, considered poor and insigniicant, hosted the outcasts Vistilia (19 ce) and
Cassius Severus (24–33 ce). Kythnos, which was relatively close to the mainland and had pleasant
50
Sweetman
features such as hot springs and, according to Pliny, good cheese,53 may not have been so dreadful
for an exile. C. Iuncus Silanus, proconsul of Asia, was banished there in 23 ce.
Paul Allatson and Jo McCormack suggest that while exile is oten seen as “a disturbed
physical and psychic relation to space and home,” it is also now used iguratively to refer to
senses of dislocation or unease.54 Analyses of exile in the Roman period focus on the impact of
exile on an individual’s psyche and how that experience is articulated rather than on the place
of exile.55 Although elements of exile in terms of the patria have been studied, particularly when
there may be negative qualities associated with the homeland, the efects of exiled individuals on
the new place are not so commonly discussed.56 he phenomenon of mass exile in more recent
times has turned attention to the efects of the diaspora on the new place as well as the efects of
the loss of the exiles on the old place and cultures.57 Depending on the individuals exiled, their
presence on the island could contribute to a change in its welfare either through direct investment or through expenditure by visitors to the exiles, each circumstance providing diferent
network opportunities. his is particularly the case with those islands that harbored some of
the more renowned exiles, where not just family members but also supporters or clients may
have visited. An astrologer exiled to Seriphos did particularly well inancially ater his business
followed him to the island,58 and even an exile to the dreaded Gyaros, the teacher Musonius
Rufus, was visited by many while there.59 Although there is little to suggest that these islands
became signiicant players in the Roman network, they were not cut of from wider experiences
in the empire.
Epigraphic evidence indicates that there were opportunities to be had on fertile islands
such as Andros and Amorgos. On Andros, for example, two honoriic inscriptions were set up to
the exiles Glitius Gallus and Egnatia Maximilla (65 ce).60 Even on less promising islands, the exiles
made a diference; Philostratus indicated that the presence of Musonius on Gyaros improved the
island because people came to see him there. Apparently he uncovered a fountain and also worked
on digging through the isthmus.61
Although life cannot have been easy on many of the islands, it was by no means as remote
and barren as the generalizations of exile would imply. In De exilio 8 Plutarch discusses the happiness of life in exile, even on small islands, and says that the exiled are mistaken if they think
they are being punished. Every island has a “house, a walk, a bath, ish and hares for hunting and
sport,” he notes. he presence of exiles on some of the Cycladic islands would have changed the
connectedness of an island, in some cases temporarily and in other cases—through investment in
buildings or cult—more permanently.62
One prompt for more permanent investment was cult practice in connection with tourism.
Clive Gamble and Carl Knappett argue that connections are not limited to those with “face-to-face
interaction.”63 Reasons of religious or cultural interest also come into consideration in the form
of pilgrimage and tourism practices. Evidence for tourism to the islands in the Roman period is
limited to a small number of elite references from Cicero, Mucianus, and the like. hese sources
do at least make it known to us that there were several island tourist attractions: for example, on
Naxos, the Tomb of the nurse of Ariadne and Ariadne herself; on Ios, the Tomb of Homer; and
on Paros, the Tomb of Archilochos.64 In 51 Bce Cicero toured the Cycladic islands on his way to
take up his governorship in Cilicia, visiting Keos, Gyaros, Syros, and Delos before heading over to
Samos.65 Some suggest that the visits were due to weather conditions rather than being planned
events.66 Hadrian certainly sailed through the Cycladic islands on the way from Ephesus to the
Peloponnese in 124 ce and seems to have stopped of at Astypalaia.67 Sextus Pompeius holidayed
Networks
51
on Keos on his way to take up the governorship of Asia in about 27 ce, and the historian Florus
(ca. 70–140 ce) also visited the Cyclades.68
he Cycladic islands also had a number of prominent sanctuaries and sites of miracles.
Especially famed were the temple of Apollo and Artemis on Keos, which was excavated in the early
nineteenth century, and the sanctuary of Delian Apollo at Paros.69 he sanctuary of Dionysos at
Andros was known for its miracles.70 Anna Collar, Hugh Bowden, and Ian Rutherford have successfully shown how network analysis can be applied to the spread of religious ideas and practices
of diferent periods in the Graeco-Roman world,71 but more remains to be done in this vein for the
Cyclades in the Roman period. Although it is diicult to assess the extent of everyday occurrences
of religious tourism, it is known, for example, that Mucianus visited the Cycladic sanctuaries of
Dionysos on Andros and Apollo on Delos.72 Andros was primarily an agricultural island, as well as
being used for exile, but there was a high volume of traic because of the sanctuary.73 As discussed
above, there was external investment in cult sites such as Mithras on Andros, Isis on Syros, and
Poseidon and Amphitrite on Tenos. Although the temples of Mithras at Palaiopolis on Andros
and of Isis on Syros have been identiied only through epigraphic evidence, foundations of the
temple of Poseidon and Amphitrite on Tenos were revealed in the twentieth century by excavations undertaken by the French School of Archaeology.74 Founded in the fourth century Bce, the
sanctuary was lavishly constructed of a variety of marbles. It reached its zenith in the Hellenistic
period; both Augustus and Tiberius took an interest in it, particularly Tiberius, who gave it asylia.
Later investment is also evidenced through epigraphic data as well as Roman sculpture including
Hadrian and Trajan.75 Elsewhere, personal connections of other kinds may be shown through
investment and patronage. Islands such as Paros, hera, and Andros beneited from donations
from wealthy individuals.76
he Cycladic Islands and networks
One of the hindrances to understanding the Roman Cyclades was a relative lack of archaeological data from this period. In the last few decades, however, this lack has been addressed, and in
addition to the evidence from the larger islands of Andros, Melos, and Tenos more attention has
been paid to the Roman material from smaller islands such as Amorgos and Kythnos.77 Although
a full analysis is not possible here, a brief look at the archaeology of the islands in relation to economic issues, exile, and tourism will make clear the impact of networks on a reconsideration of
the Cyclades in the Roman period.
Melos went out of its way to attract trade by exploiting its location and its resources,
which included hot springs, and it seems to have done exceptionally well in the Roman period.78
Recent work has been undertaken on the Roman minerals used for industry and medicine and on
mining in the area of Agia Kyriaki; it appears that the mines were well exploited.79 Additionally,
the Melians engineered good connections to the networks through the establishment of a number
of small ports around the island to maximize trade in many kinds of weather, although not all
ports were in constant use—for example, the main harbor silted up in the early Roman period.80
Harbor installations were excavated at Klima, and it is likely that substantial data will be recovered at Palaiochori (ancient Zephiri). 81 he signiicant amounts of archaeological material already
excavated on Melos (at Klima, Agios Elias, and Tramythia)—from public places such as gymnasia,
baths, and theaters (ig. 3.1) to private buildings, industry, and mortuary evidence—have provided
52
Sweetman
figure 3.1 he theater on Melos
a broad view of the Roman occupation.82 Close to the catacombs near Trypiti is a small Christian
church, and a Late Antique church is located in the modern town of Plaka. Melos’s comparatively
extensive networks are relected in onomastic data that show a Melian, Ti. Claudius Frontonianus,
becoming a Roman eques.83 Furthermore, it is clear from an inscription thanking the Romans
for their investment that there was external spending on the island.84 he extensive architectural
remains, mosaics, and sculpture hint at the return the Melians got on their investment in being
part of the network society. here may also have been a minor ripple efect on Kimolos, enabling
the exploitation of Kimolian chalk. More recently, a Hellenistic and Roman settlement has been
identiied in the south of the island.85 he sites of Ellenika and Palaeokastro were occupied until
the Late Antique and Byzantine periods, respectively, and it is hoped that the recent survey will
provide more data.86
While Melos may have had a positive impact on nearby islands in the imperial period, it
is likely that only in the Hellenistic period did Tenos beneit from being in Delos’s wake. here is
evidence of Roman bankers, proxenoi, and a Roman community on Tenos in the irst century Bce.
Despite the decline of Delos in the Roman period, however, Tenos continued to do well. here was
some patronage on the island, but it was by families unconnected with the imperial elite,87 and in
turn a number of patrons and Roman oicers are known to have received honors on the island.88
Strabo noted that the famous Poseidia festival attracted signiicant numbers of visitors, and he also
noted that the island did not have a major city. 89
he main town on the island, Xobourgo, seems to have been replaced by the town closer
to the modern capital sometime in the ith or fourth century Bce.90 Although the sanctuary of
Poseidon and Amphritrite at Kioni has been excavated,91 little of the Hellenistic and Roman occupation of the town of Tenos has been investigated, and there are hints of Roman occupation at
Networks
53
Sparto and Akroterio Ourion. Tenos was largely self-suicient, but it had few signiicant resources
other than some marble quarries. In spite of the small urban space and limited trade, epigraphic
evidence suggests that Tenos did well throughout the imperial period, which may have been in part
through its agriculture and religious tourism and perhaps even through its use as a place of exile.92
Tenos may actually have beneited from the decline of its neighbor Delos, as might
Andros. At Andros, epigraphic data indicate trade and banking activity, but the level of patronage found in Tenos has not been shown here.93 Since the island has seen some concentrated
archaeological work, however, this assessment may change in the future. As the second largest
of the Cycladic islands, Andros had well-watered and fertile land in addition to wood and some
marble resources. he renowned irst-century Bce copies of the Matron of Herculaneum and
Hermes Psychopompos were found at Palaiopolis (ig. 3.2). Zagora was deserted in favor of a
new foundation at Palaiopolis, and excavations have revealed vast quantities of Hellenistic and
Roman material including the walls and the Agora, mosaics and a cistern and a Late Antique
church with mosaics.94 Furthermore, recent excavations at sites such as Virokastro, Peristerionas,
and Kolymbos have revealed Roman occupation, and at Sotera (near Palaiopolis) a Late Antique
building was excavated.95 Quarries at Kato Phello and the nearby Roman settlement have been
investigated.96 At Tourlos a Roman grave and sarcophagus have been excavated.97 here is also
ongoing work at the harbor.98 A Mithraeum was established on the island and attracted a great
deal of external investment.99 Pliny discussed the miracles of the river lowing wine at the sanctuary of Dionysos; it may be that this was another reason to visit the island.100 Andros clearly made
use of its varied connections—those of economic resources, of religious tourism, and of banishment to exile—and as a result, the island attracted a wide range of visitors, from the evocatus
Augusti M. Aurelius Riinus and three praetorian soldiers to the exiled Glitius Gallus and Egnatia
Maximilla and, consequently, various members of their family.101
Evidence for personal connections like tourism and exile is found primarily in epigraphic
data; economic connections may be more readily seen in archaeological remains, from pottery to
harbor installations. But one cannot assume that personal connections always made an impact.
Keos, for example, had longstanding connections through its associations with the family of
Herodes Atticus, but there is little evidence as yet to indicate any personal investment on the island,
which in turn encourages theories about lack of Roman involvement there.102 Neither the celebrated sanctuary nor the eminent tourists seem to have had any long-term impact on the island,
perhaps because of the one-directional nature of traic. Keos did not have good clays, so nearly
all its pottery was imported. G. A. Zachos’s study of the ceramic data from Late Antique Keos suggests that the normal African red slip and local table wares (Attiko pottery) were imported and
that specialist wares such as beehives were imported along with oil and wine amphorae.103 Zachos
suggests that ships oten stopped in Keos before going to Piraeus from Constantinople and Asia
Minor. In this case we see evidence for local networks (with the use of Attiko pottery, produced in
the Cyclades) as well as one-way traic from the east.
Assessing participation in the broader network of the empire brings a further tension to
the fore. Because the Cyclades today form part of the modern Greek state, the default perspective
is to look for connections with Greece rather than Asia Minor.104 Some of the islands, however, had
figure 3.2 Hermes Psychopompos from
Palaiopolis, Andros, 1st century bce. Marble.
Andros, Archaeological Museum
Networks
55
closer connections with Asia Minor than with mainland Greece. Epigraphic data show that families from hera, Santorini, were active in Ephesus, a connection supported by Enora Le Quéré’s
suggestion that the agora in hera is stylistically similar to that of Ephesus; hera’s basilica hall,
too, shares architectural features with the one at Ephesus.105 On Santorini, the focus of work in
more recent years has been on Bronze Age material, but the acropolis city of hera was extensively
occupied in the Hellenistic and Roman periods. It was in the Roman period that baths were constructed and the theater rebuilt (ig. 3.3).
To date, most of the evidence for Roman-period material on the larger islands of Mykonos,
Naxos, and Paros has been identiied through rescue excavations.106 Some long-term occupied
settlements have Roman levels at Divouni (Mykonos), Rizokastelia (Naxos), and Paroikia (Paros).
Given their size and resources, it is surprising that these islands show little evidence for signiicant
exploitation of their assets.107 In the case of Paros, however, new work is changing the picture, and
the same may be true for the other islands. On Paros, recent studies by Sophia Detoratou indicate
that, contrary to earlier assessments, marble continued to be quarried, but it stayed a little more
local than it had in the past.108 he recent survey undertaken on the island has also identiied some
new quarries, including one at Karavos.109 Otherwise, although many of the islands would have
been on key routes, particularly the northern ones, eforts to exploit the position were not made.
Lina Mendoni and Sophia Zoumbaki indicate a dearth of Roman names found in
the Cyclades (certainly in comparison to the mainland), and the number of complete Roman
figure 3.3 he agora on hera
56
Sweetman
names indicating citizenship is even more limited.110 Andros, Melos, Syros, and Tenos have the
richest onomastic evidence, and all except Melos have famous religious sanctuaries. It is likely
that Andros, Melos, and Tenos were well connected in part because of their resources and the
eforts they made to be connected. here is an archaeological bias in that all but Syros, which
has been developed for industry, have seen extensive archaeological work. Still, there are indications of occupation at locations such as Ermoupolis, as well as tantalizing remains such as
theater seats and a statue base with a dedication to Hadrian.111 It is worth noting that a dearth of
resources or agriculture—as in Syros—did not automatically mean an island was less well connected. he epigraphic data from Syros indicates signiicant conspicuous consumption (in the
form of public banquets) along with an Italian and foreign presence.112 Syros was on a number of
routes, and it may be that the island was well connected, but with the lack of archaeological investigation it has remained on the sidelines. he lack of information afects other islands too. Tiny
islands such as Tsimindiri and Rheneia seem to have been occupied in some way, perhaps just as
cemeteries,113 yet many of the small islands, including Anai,114 Antimilos, Danousa, Despotiko,
Foleganros, and Gyaros Heraklea, along with some of the larger ones, such as Ios, Seriphos,
Siphnos, and Sikinos,115 are still relatively unknown in terms of archaeological evidence pertaining to the Roman period.116 he bias is being corrected, and islands such as Keros, Kouphonisi,
and Schniousa have been the focus of recent work, with Roman material coming to light through
survey and excavation.117 Furthermore, increasingly systematized underwater surveys are shedding light on the nature of trade—at Polyaigos, for example, where Carystos marble and amphorae were found of the coast.118
Connections and networks
Positive evidence for Roman investment and connections in the Cycladic islands has been muted
by the tropes of exile and isolation. Network analysis of the archaeological, epigraphic, and literary
data, however, indicates a more vibrant view, even of the very smallest of the islands. Mendoni
and Zoumbaki found that the Cyclades did not make a signiicant impact on high imperial policy
in terms of individuals, particularly as signiicant amounts of land were required in order to join
the senatorial order.119 he islands were not the focus of imperial attention except to fulill speciic
aims; for example, Tiberius took some artwork by Skopas, and Nero took material from Andros
and Mykonos. he islands’ onomastic evidence may not indicate much of an impact on the political life of the empire, but there is good evidence for sustained relationships through merchants,
exiles, and tourists.
In this discussion I have focused on the personal network connections that survive in the
contexts of exile and tourism, and highlighted some of the parallel economic conditions revealed
by the archaeological data. here is a great deal of work still to be done on the Cyclades and other
network connections between the islands and between the center and individual islands. Work
focusing on groups of speciic artifacts such as mosaics and on sites such as sanctuaries might
help to identify speciic small-world networks that may have existed between the islands without
necessarily having direct connections to Rome. Although Mendoni and Zoumbaki argue that
there is little evidence for movement of people between the islands and certainly no evidence
for seeing the islands as a political unit, 120 there may have been movement of artifacts and cratspeople, which may have created levels of “glocalization,” which can be understood in terms of the
Networks
57
local reworking of global products.121 Melos seems to have been involved in the Roman network,
but there is little evidence of phase transition, as few of the nearby islands appear to have become
nodes as a consequence.
he islands of the Cyclades had diferent types of contact with the main centers of the
Roman Empire—personal or business, direct or indirect, planned or accidental. In earlier scholarship, the top-down approach to an analysis of the islands resulted in negative views of isolation.
However, the islands do not have to be viewed in relation to Rome.122 hey had their own decisionmaking processes irrespective of Rome or indeed other regions.123 Although life on the islands
may have been in parallel to what was going on in the broader empire, there was still individual
choice. For example, some islands or simply cities or individuals may have openly courted imperial attention, as can be seen in Palaiopolis on Andros with the eight statue bases dedicated to
Hadrian. In other cases, involvement in the empire may have been imposed on certain islands for
inancial or strategic reasons, or as a result of an island being selected as an internment site, each
circumstance producing diferent dynamics. For many islands, day-to-day life may have continued
unchanged, particularly if people had little contact with other areas of the Roman Empire.124 In
other cases, such as Paros, it is diicult to see why the island had little success in becoming part of
the imperial network, particularly since it had resources to exploit and had already been central
to regional communication links. he evidence indicates the luctuating nature of the networks in
the Cyclades. he islands that had the means and motivation to participate, such as Melos, seem
to have been more successfully incorporated into the Roman network. It had resources, and it
used its multiple ports to make those resources accessible. It is suggested here that, rather than
being a negative, the use of the islands as places of exile provided certain levels of integration and
opportunities to be part of a network, even if those opportunities were not commonly exploited.
From the local perspective it seems that those living on the islands may have had a choice to participate in the Roman network. If they did not participate, it relects not passive segregation but
active judgment.
he analysis of the Cyclades in the Roman period highlights the importance of local
networks that may not have included Rome as well as those that did.125 While being part of a
province may have made it easier to take advantage of imperial network connections, it did not
automatically mean that connections would be sustainable or successful. Ironically, being considered peripheral to the empire and used as a place of exile may have stimulated temporary network
connections that let lasting efects of investment. Few of the islands were truly insular—there
was plenty of contact with diferent parts of the empire, and on a range of levels—but few had
multiple connections. Furthermore, while Melos may have been a network hub, the current data
do not allow identiication of other islands as nodes or even bridges in the scale-free network.
As Brughmans has noted, it wasn’t enough just to create links through the presence of people or
patronage of the island; something had to be actively engineered to make the connections work.126
In this respect, although an island like Gyaros had large numbers of visitors because of its noteworthy exiles, there was no motivation for the island to be plugged into the network, nor were
there resources signiicant enough for the center to be interested.127 his is hardly surprising, given
the island’s small size of 23 square kilometers. According to Knappett, nodes of a network may be
identiiable in relational terms, and the situation on Melos in comparison to many of the other
Cycladic islands is a good case in point.128 Unlike those on Gyaros, Melians did all they could to
exploit their island’s contacts and ports.
58
Sweetman
It is clear from the archaeological and epigraphic evidence that many of the Cycladic
islands were far from isolated places suitable only for exile. he application of network theory to
these data allows the diversity of relationships to be highlighted. It gives us a means for looking
at individual islands as well as groups without preconceptions about how they should relate to
each other or to Rome based on provincial categorization. While individual islands may have had
sustained network links with Rome, others may have had only temporary links with Rome but
prolonged links with other islands within a small-world network. Preliminary analysis of material
such as marble and amphorae suggests small-world networks were in operation involving Paros,
Naxos, Andros, and Tenos. he multitude of visitors Strabo mentioned adds weight to this possibility. It is also possible to see Melos as a key link between the Cycladic small-world network and
the larger Roman scale-free network. Furthermore, some islands may have had connections with
other provinces (as in the case of Santorini and Asia Minor), which have been overshadowed by
research approaches. he increasing archaeological data from survey, research, and rescue excavations in the Cyclades is helping to provide a better understanding of life on individual islands in
the Roman period. Network connections help to explain diverse relationships and reveal active
engagements otherwise not seen. he study of this material is only deepened by new questioning
of our old ideas of how provinces should behave.
notes
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Catullus, Carmina, translated by Sir Richard Francis
Burton (London, 1894), poem 4.
Strabo 7.7.5.
A number of inscriptions also refer to piracy in
the Cyclades, for example, one from Tenos (IG XII
5.860), which mentions that the island was constantly being attacked by pirates, and another from
Syros (IG XII 5.653), honoring Onesandros from
Siphnos, who helped a victim of a pirate raid (de
Souza 1999, 63, 85).
As Jiménez in the previous essay.
Sweetman 2013, 8–9.
Revell (in this volume) notes the dichotomy created
in contemporary scholarship between what is
Roman and what is native.
Sweetman 2007.
he negative view of provinces is discussed in detail
by Jiménez in this volume.
Le Quéré 2011.
Palaiokrassa-Kopitsa 2012, 34; Le Quéré 2011, n. 9.
For the Bronze Age and Classical periods, counteranalysis for the broad-brush approach has been provided by Broodbank 2000 and Constantakopoulou
2007, respectively.
Sweetman 2007; see also Papaioannou in this
volume.
See Papaioannou in this volume.
For an onomastic study see Mendoni and Zoumbaki
2008.
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Sneeringer 2011 is a promising master’s thesis on
the economy of the Roman Cyclades with potential
for future development; Leekley and Noyes 1975
focuses on recent archaeological work, while Freely
2006 provides an overview of Cycladic history and
archaeology.
Barber 2005; Leekley and Noyes 1975; Freely 2006.
Livy 31.15.5–11.
Mendoni and Zoumbaki 2008, 24.
Mendoni and Zoumbaki 2008, 24.
Mendoni and Zoumbaki 2008, E350.
Papageorgiadou-Banis 1997, 8.
For example, M. Cosconius M. F. Poll. Fronto,
procurator ad vectig(al) XX her(editatium) per
Asiam, Lyciam, Phrygiam, Galatiam, insulas
Cycladas, a second-century inheritance tax collector
(Mendoni and Zoumbaki 2008, 27–28).
Mendoni and Zoumbaki 2008, 27.
Mendoni and Zoumbaki 2008, 27.
Mendoni and Zoumbaki 2008, 27.
Mendoni and Zoumbaki, 2008, 32–33.
Strabo 10.5.7, 10.5.1.
For example, Gothic raids in 268 ce had an impact
on a number of islands like Paros (Kulikowski 2007,
19–20). See also Freely 2006, 14–15.
It is known that there was a bishopric established on
Santorini in 342 ce (Freely 2006, 15), but there were of
course Christians present on the islands much earlier
than the fourth century, as indicated most prominently by the irst-century catacombs on Melos.
Networks
59
30 Curta 2004, 520–21. Curta notes (n. 23) that the
evidence for Slavic raids on the Cyclades has been
disputed.
31 Brughmans 2010.
32 Van Dijk 2005; D’Souza et al. 2007.
33 Brughmans 2010, 278–80.
34 Chen and D’Souza 2011, 1.
35 Brughmans 2010; Knappett 2011, 2013.
36 Brughmans 2010, 282.
37 Nikolakopoulou and Knappett 2005; Knappett 2011,
7–9.
38 Knappett 2011, 57.
39 Constantakopoulou 2007, 3.
40 Rauh 2003.
41 See Gates-Foster in this volume. For comparable
examples, see also Moser and Revell in this volume.
42 Cassibry and Wicker in this volume.
43 Knappett 2011, 9. Many studies of Cycladic
networks focus on spatial distribution of material
and on human connections rather than on social
space.
44 he contributions by Cassibry, Gates-Foster, Revell,
and Wicker in this volume all discuss diferent
elements of material culture to analyze network connections and their ramiications for understanding
the provinces.
45 All details (and the quotation) in this paragraph are
from Braginton 1944, 391–97.
46 J. M. Scott 1997a, 173.
47 Mendoni and Zoumbaki 2008, 30.
48 Braginton 1944, 399.
49 Mendoni and Zoumbaki 2008, 30.
50 Juvenal 10.170, 1.73.
51 Braginton 1944, 401.
52 Tacitus, Annales 3.68–69; Mendoni and Zoumbaki
2008, 29.
53 Naturalis Historia 13.134.
54 Allatson and McCormack 2008, 9.
55 Said 1984; Rossbach 2007. Neil and Allen (2011)
examine evidence for responses to exile in the Late
Antique.
56 Cohen 2006.
57 Scott 1997b; Sanders 1997; Kilduf and Corley 1999.
58 Juvenal 6.562–64; Braginton 1944, 394. Another
astrologer, Pammenes, was exiled under Nero. While
the place of exile is not known according to Tacitus
(Annales 16.14), he had a steady number of visitors
continuing to visit him for consultations (Rudich
1993, 145).
59 Braginton 1944, 401.
60 Other examples include Avillus Flaccus (39 ce) on
Andros and Vibius Serenus on Amorgos (23 ce):
Mendoni and Zoumbaki 2008, 29–30.
61 Philostratus, Vita Apollonii 7.16. Mendoni and
Zoumbaki 2008, 29.
60
Sweetman
62 his is particularly pertinent in terms of temporal
connections (as discussed by Knappett 2011 and
Gamble 2007 and discussed in the conclusions
here).
63 Gamble 2007; Knappett 2011, 11.
64 Stumpf 2003, 438.
65 Stark 2006, 90.
66 Mendoni and Zoumbaki 2008, 26.
67 Stumpf 2003, 453.
68 Mendoni and Zoumbaki 2008, 26.
69 Strabo mentions the Keos temple (10.5.6). At Paros,
various religious sites have been identiied, such as
those to Asklepius and Delian Apollo near Paroikia
(Freely 2006, 147–54).
70 Pliny, Naturalis Historia 2.106. he sanctuary of
Dionysos has not yet been identiied on Andros;
however, Freely (2006, 86) suggests that it may be
located at the monastery of Panagia Zoodochos Pigi
near Agios Petros, which has a holy well.
71 Collar 2009; Bowden 2009; Rutherford 2009.
72 Stark 2006, 95–-96.
73 Mendoni and Zoumbaki 2008, 35.
74 Dumoulin 1902.
75 Dumoulin 1902, 411.
76 Mendoni and Zoumbaki 2008, 35.
77 On Amorgos, the site of Minoa and buildings
such as the temple of Serapis and the gymnasium
were in continuous use until the third century ce
(AR 2001–2, 100; AR 1989–90, 69–70; AR 1999,
46). Roman imperial seals have also been found,
in addition to houses endowed with mosaics
(AR 2006–7, 90), which were also excavated at the
site of Katapola (AR 2009–10, 164; AR ID 555). An
ongoing project at Kythnos by the University of
Volos has helped highlight the extensive archaeological evidence particularly at the sanctuary to
Apollo and Artemis (AR 2007–8, 93). Furthermore,
the harbor at Mandraki with its various extensions
has also been studied both on land and under water
(AR ID 840; AR 2008–9, 55; AR ID 1307).
78 Mendoni and Zoumbaki 2008, 40.
79 Photos-Jones et al. 1999.
80 Sneeringer 2011, 44.
81 C. Smith 1895–96.
82 Renfrew and Wagstaf 1982.
83 Mendoni and Zoumbaki 2008, 3. Only one other
example is known from the Cyclades, T. Flavius
Cleitosthenes from hera.
84 IG XII 3.1097.
85 AR 2001–2, 99.
86 AR 2003–4, 71.
87 For example, P. Serveilius P. f. Isauricus, proconsul
of Asia 46–44 Bce, restored some sculpture that had
been dedicated earlier by C. Pandusinus Cn. f. in
about 100 Bce (Mendoni and Zoumbaki 2008, 36).
88 Praefectus classis Q. Calpurnius Q. f. along with his
family was honored with proxeny and local citizenship. P. Quinctilius Varus was a quaestor and was
honored on Tenos as a patron.
89 Strabo 10.5.11.
90 AR 1955, 28.
91 AR 1973–74, 30; AR 1974–75, 22.
92 Palaiokrassa-Kopitsa has published extensive accounts of the archaeology of Tenos; see
Palaiokrassa-Kopitsa 2007.
93 Mendoni and Zoumbaki (2008, 35) point out various
oicial roles (such as stephanephoros) that had to
be repeated by the same citizens, as there were not
enough to hold the public functions.
94 Respectively, AR 2003–4, 68; AR 2003–4, 68;
AR 2006–7, 82.
95 AR 2006–7, 82, and AR 1999–2000, 114 (for the Late
Antique church).
96 AR 1999–2000, 115.
97 AR 2001–2, 95, 116.
98 AR 2007–8, 88.
99 Mendoni and Zoumbaki 2008, 31.
100 Naturalis Historia 2.106.
101 Mendoni and Zoumbaki 2008, 28–31; Braginton
1944, 398.
102 Papageordiou-Dias 1997, 8–9. Strabo mentions the
two major towns, Ioulis and Karthaia, and also
the temples of Apollo and Artemis, which were
excavated in the early nineteenth century (Strabo
10.5.6). Work undertaken on the site and pottery
at Karathaia (AR ID 2046) and on the harbour of
Poiessa on the west side of the island indicate more
activity during the Roman period than previously
assumed (AR ID 2071).
103 G. A. Zachos 2010.
104 See Jiménez in this volume for further details.
105 Le Quéré 2011, 341; Stinson 2007, n. 21. For another
case of a basilica’s design suggesting overseas links,
see Morton in this volume.
106 For example, the Roman kilns at Chora on Naxos,
AR 1999–2000.
107 he marble quarries were taken out of the hands of
the locals and run by the Romans.
108 http://phdtheses.ekt.gr/eadd/handle/10442/28675.
109 Bruno 2010; Bruno, Lazzarini, et al. 2010.
110 Mendoni and Zoumbaki 2008, 43.
111 AR 2003–4, 68, and Freely 2006, 141.
112 Mendoni and Zoumbaki 2008, 37–39.
113 AR 2006–7, 85.
114 here is likely to have been murex production at
Anai (AR ID 560), however.
115 On Ios, Hellenistic and Roman remains have been
identiied at Pounta, Managari, and Mylopotamos
(AR 1981–82, 46). On Seriphos, a Roman bathhouse
with mosaics has been excavated on the eastern
slope below the modern village of Chora (Pantou
2001–4, 13). Sikinos may have evidence for a tomb of
an exile at Episkopi (Frantz, hompson, and Travlos
1969).
116 Many of these—Danousa, Foleangaros, Sikinos,
Gyaros, Seriphos—are known as places of exile.
117 AR 2007–08, 92; AR ID 1978.
118 AR ID 544 and AR ID 1350.
119 Mendoni and Zoumbaki 2008, 24.
120 Mendoni and Zoumbaki 2008, 24.
121 Knappett 2011, 9.
122 See Jiménez in this volume.
123 he concept of free choice is also highlighted by
Papaioannou in this volume.
124 See Sweetman 2007 on how Knossos seems to have
remained largely unchanged despite the foundation
of the Roman colony there.
125 Revell in this volume.
126 Brughmans 2010, 282.
127 In the case of Gyaros, it was murex, as noted by
Larsen (1938, 485).
128 Knappett 2011, 38.
Networks
61