You are on page 1of 31

DIVINATION AT ARCHAIC BRANCHIDAI-DIDYMA: A Critical

Review
Author(s): Alan M. Greaves
Source: Hesperia: The Journal of the American School of Classical Studies at Athens ,
April–June 2012, Vol. 81, No. 2 (April–June 2012), pp. 177-206
Published by: The American School of Classical Studies at Athens

Stable URL: https://www.jstor.org/stable/10.2972/hesperia.81.2.0177

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/10.2972/hesperia.81.2.0177?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The American School of Classical Studies at Athens is collaborating with JSTOR to digitize, preserve
and extend access to Hesperia: The Journal of the American School of Classical Studies at Athens

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
he s per ia 81 (2012) DIVINATION AT ARCHAIC
Pa ges 17 7 – 2 06
BRANCHIDAI-DIDYMA
A C r i t i c al R e v i e w

ABS TRAC T

The oracle at Branchidai-Didyma was a major cult center in western Anatolia


during the Archaic period. This article assesses the evidence for different
aspects of the divinatory process used there, including divination by mantic
trance and cleromancy (using astragali), and the means by which responses
were delivered. The author argues that the assumption that mantic trance
was used is a retrojection of the Classical Delphic model of divination, and
that more direct evidence points to the use of astragali. The randomizing
function of astragali could be enhanced by modification, as supported here
by experimental data. Parallels are drawn to sites in Anatolia and elsewhere
across the Greek world.

INTRODU C T I ON

Prior to Delphi’s rise to Panhellenic status during the First Sacred War
(595–585 b.c.), the sanctuary of Branchidai-Didyma was the most impor-
tant and influential oracle of the Greeks of Ionia, yet virtually nothing is
known about the process of divination used at the site or how its oracular
pronouncements were delivered.1 While a veil of secrecy surrounded some
cults in the Classical period, such as the great cult at Eleusis, no such
mystery appears to have shielded the workings of contemporary oracles,
such as Delphi.2 Nonetheless, despite numerous tantalizing references in

1. The name “Branchidai-Didyma” advice and encouragement in the prep- Coulthard, Helga Donder, Yiğit Erbil,
is used throughout this article to differ- aration of this article. The staff of the Garth Gilmour, Salima Ikram, Alice
entiate the Archaic period of the oracle, British Museum were very helpful in Mouton, Simon Northwood, Robert
when it was operated by the Branchidai allowing me to study the astragali from Parker, Samantha Potts, Lucinda
family, from its post-revival period, the Artemision at Ephesos. Andrew J. Reeves, Anja Slawisch, Karen Wheat-
when the Branchidai were absent. This Scally of the University of Bradford ley, and to the patient, if somewhat
later incarnation is referred to here as performed the chi-square analysis and bemused, staff of the Esat Aile
“Didyma.” deserves very special thanks for giving Çiğercisi butcher shop in Ankara.
I am greatly indebted to Susanne his time and expertise. Special thanks 2. Holland 1933, p. 201.
Berndt-Ersöz and John Davies for their are also due to Roger Brock, Gina

© The American School of Classical Studies at Athens

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
178 a l a n m . g r e av e s

contemporary literature and history,3 we know very little about how even
this most famous of oracles operated. This general lack of evidence from
well-attested sites is compounded at Branchidai-Didyma, where there is
virtually no historical or literary evidence for divination practices from the
Archaic period.
In the absence of any direct evidence, scholars have often assumed that
the oracle here functioned by means of a mantic trance, along the lines of
the later Delphic oracle.4 In this article I review the evidence for how the
oracle may have operated at Branchidai-Didyma in the Archaic period,5
with particular reference to three aspects of ancient oracular practice:
mantic trance, cleromancy (divination by casting lots), and the delivery of
pronouncements. The evidence for each of these at Branchidai-Didyma
will be assessed and briefly compared to contemporary sites across Greece
and Anatolia, especially the great Apolline oracle at Delphi.

TH E H ISTORY OF BRANCH IDAI-DI DY MA

Together with Klaros, the oracle of Apollo at Branchidai-Didyma was one


of the most important and influential cult sites in western Anatolia during
the Archaic period.6 Its ruins are located in Aydın Province, in the town of
Didim, formally called Yenihisar. Until the start of the 20th century, the
place was known by the modern local Greek population as Ieronda, mean-
ing “Holy Place,” and for much of its history the settlement has derived its
name from the temple that dominates it (Fig. 1).7 These spectacular ruins
have long been a focus of scholarly, popular, and touristic interest. Histori-
cally, the Archaic temple lay within the territory of the powerful polis of
Miletos, and the fates of these two places were inextricably linked: Didyma,
as a cult site within Miletos’s sphere, lacked any status as an independent
community in its own right.8
The earliest artifact so far found at Didyma is a single fragment of
Mycenaean LH IIIA2 (14th century b.c.) pottery; there is no evidence of
a sanctuary here at this early date.9 It was only in the late 8th or early 7th
century b.c. that the first recognizable cult structure was erected at the
site. This took the form of a walled court (sekos), next to which a portico
was later built. A small building (the naiskos) was built within the walls of
the sekos, and this basic form of a walled structure surrounding an open
courtyard with a smaller inner temple10 was thought to have provided the
model for the temple’s form throughout the remainder of the site’s history.11

3. See, e.g., Euripides’ Ion. For a chidai-Didyma was probably destroyed 1988, p. 427.
survey of the oracle in myth, see Parke (see n. 14, below). 9. Schattner 1992.
and Wormell 1956, vol. 1, pp. 295–319. 6. For general works on the site in 10. Tuchelt 1973, p. 116.
4. See, e.g., Fontenrose 1978, p. 228; the Archaic period, see Parke 1985b, 11. Excavations in 2004, however, in
Burkert 1985, p. 115; Parke 1985b; pp. 1–111; Fontenrose 1988; Tuchelt a sounding on the south side of the
Höckmann 1996. 1991; Greaves 2002, pp. 109–129; Furt- Hellenistic temple may call into ques-
5. For the purposes of this article, wängler 2006a; Herda 2006; Johnston tion previous reconstructions of the
the Archaic period is defined as the 2008, pp. 82–90. Archaic building. Its precise form is
period between the late 8th century b.c. 7. Fontenrose 1988, p. 3. currently the subject of debate. See
and 494 b.c., the year in which Bran- 8. Ehrhardt 1998, contra Tuchelt Furtwängler 2006a.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 179

Figure 1. View of the Hellenistic


temple at Didyma, from the north- This early structure was built over in the second half of the 6th cen-
east corner. Photo courtesy F. Tronchin tury b.c. by a larger temple that, until recently, was generally thought to
have been of the same basic design. Whatever the precise size and configu-
ration of the Archaic temple, it was evidently built on a grand scale—an
impressive home for Apollo’s oracle. The building was decorated on the
outside with sculpted column bases and an ornate architrave, and it was
furnished with fabulous treasures by its pilgrims and sponsors.12 The temple
was probably operated by the Branchidai, about whom little is known, but
who are presumed to have been a family of priests. Their name became
synonymous with the oracle in the Archaic period: Herodotos referred to
it as “Branchidai in Milesia.”13
The main building appears to have continued in use throughout the
Archaic period, until it was sacked by the Persians in 494 b.c., following
the Battle of Lade and the subsequent fall of Miletos.14 The temple was de-
12. Tuchelt 1970, 1992; Schattner stroyed and robbed of its treasures, and according to Herodotos, its priests,
1996. See also Hdt. 1.92.
the Branchidai, were carried away to the East.15 The oracle was silenced.
13. Hdt. 1.46; Rubinstein and
Greaves 2004, p. 1059. The oracle regained its voice when Alexander the Great revived it after
14. There is some disagreement in taking Miletos in 334 b.c. In the meantime, there had been a break of over
the sources on the exact date of the sack one and a half centuries since the oracle last operated. Much information
of Didyma, but 494 b.c. is the most about the Archaic oracle can reasonably be expected to have been lost
reasonable and widely accepted date;
during this time, not least because the people who had been previously
see Tuchelt 1988, pp. 427–430.
15. Hdt. 6.19. responsible for its operation, the Branchidai, had been captured by the
16. Parke 1985a, pp. 62–65; Kallis- Persians. When Alexander discovered the descendants of the Branchidai
thenes FGrH 124 F14. living far to the east, he had them slaughtered.16 Their murder, whatever

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
180 a l a n m . g r e av e s

its motivation, together with the destruction of the site by the Persians,
marks an absolute and deliberate break between how the oracle had oper-
ated in the 6th and early 5th centuries b.c. and how it would function in
the Hellenistic period.17 As a part or a consequence of the revival of 334
b.c., there was building work done on a colossal scale at Didyma, and it
is these Hellenistic ruins, initially financed by Seleukos I, that the visitor
sees today (Fig. 1).18 The divination procedures of the revived oracle ap-
pear to have been modeled on those of Delphi and not of the Branchidai,
who were now in disfavor. This oracle attracted many pilgrims for several
centuries until, with the coming of Christianity, it fell silent permanently.

SOURCES FOR ARCHAIC BRANCH IDAI-DIDYMA

Very little evidence is available concerning the question of divination pro-


cedures at Archaic Branchidai-Didyma. The main types of source material
used by classical archaeologists—literary, epigraphic, and archaeological—
are all represented to a limited extent at the site.
The epigraphic and archaeological evidence is severely limited by two
factors. First, the Persian sack of the temple and the city of Miletos (where
inscribed responses from the oracle were apparently stored in the Delphi-
nion) destroyed many inscriptions and stripped the temple of the votives
and religious paraphernalia that might have provided information about
cult practice. Second, the much larger Hellenistic temple built over the
Archaic ruins almost completely obscures the earlier structure and hinders
archaeological investigation of the Archaic deposits. The few historical
sources that do exist are limited in their usefulness, inasmuch as they were
written after the destruction of the temple and none of them explicitly
describes the divination procedure.
Herodotos is the most important literary source on the Archaic oracle.19
His is the closest that we have to a contemporary account of the work-
ings of the oracle, but even so, he is writing a considerable time after its
destruction and alludes only vaguely to its operation. There are a number
of interesting later sources, such as Pausanias, Strabo, and several minor
writers, but all of these can be dismissed, both because they were writing
much later than Herodotos and because, in describing the oracle, they are
largely concerned with mythologizing.20
Only three Archaic inscriptions survive that record responses from
the oracle at Branchidai-Didyma, all dated to the 6th century b.c.21 They

17. A similar break in cult practice was revived in the Hellenistic period, [C 421]; 11.11.4 [C 517–518]; 14.1.5
at the site can be seen in another of its but had probably been dormant during [C 634]; and others (see Fontenrose
major religious ceremonies, the Molpoi the intervening century. 1988).
procession. This annual event con- 18. Curnow 2004, p. 133. 21. Kawerau and Rehm 1914,
nected the city of Miletos to the Bran- 19. Hdt. 1.46, 92, 157–159; 2.159; pp. 276–277, no. 132a; pp. 397–398,
chidai-Didyma sanctuary (Herda 5.39. It has also been suggested that, no. 178 (found at Miletos); Rehm and
2006). In a reassessment of the archae- based on its iambic meter, another pas- Harder 1958, p. 8, no. 11 (found at
ological and epigraphic evidence from sage in Herodotos (1.174) may be an Didyma); Jackson 1995; SEG XLV,
the Sacred Way, Anja Slawisch (2009) oracle given to Knidos by Branchidai- p. 1565; Fontenrose 1978, pp. 417–429;
has shown that this procession was a Didyma; see Parker 1985, p. 316. 1988, pp. 177–181; Eidinow 2007,
phenomenon of the Archaic period that 20. Paus. 5.7.5, 13.11; Strabo 9.3.9 pp. 53–55.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 181

are brief and fragmentary and are written boustrophedon, a method of


inscription that securely dates them to the Archaic period.22 One of these
three inscriptions is from the site itself, and the other two are from the
Delphinion at Miletos.23
Primary archaeological evidence from Branchidai-Didyma comes from
excavations dating back to the British expeditions in the 19th century.24
Excavations by Theodor Wiegand began in 1904, and the preliminary
reports and three-volume final publication remain essential.25 In the lat-
ter part of the 20th century, research excavations under the direction of
Klaus Tuchelt produced regular interim and final reports.26 An important
outcome of the investigations of the 1980s and 1990s was the discovery,
excavation, and publication of a sanctuary on the Sacred Way that connects
Branchidai-Didyma and Miletos.27 In the early 21st century excavations
are still ongoing, with new evidence coming to light every year.28

DIVINAT ION AN D DELI VERY OF RES P ON S ES

Despite the obvious prominence of Branchidai-Didyma in the Archaic


Greek world, we know little about the precise process of divination used
at the site. Divination in general can be understood as a two-stage process.
First, following various preliminaries, is the method of divination itself,
by which the message of the god is received or determined. Second, that
message is delivered to the inquirer. It is important to distinguish between
these two stages when considering how oracular sites operated. Two meth-
ods of divination widely thought to have been used in ancient Greece are
mantic trance (a form of spirit possession) and cleromancy (divination by
throwing lots). I consider both methods below in relation to the evidence
from Branchidai-Didyma, before going on to consider the possible modes
of delivery used there.

Div inat i on by Man t i c Tran c e


Mantic trance is a state in which spirits are believed to enter the bodies
of human beings, animals, or objects.29 In this altered state of conscious-
ness the individual displays different behaviors that are interpreted as the
influence of an alien spirit.30 A variety of methods can be used to induce
such a state; once it is achieved, according to different cultural beliefs, the
spirit may reveal insights into the past, present, or future.
Mantic trance is the method of divination commonly thought to have
been used at that most important and influential of all Greek oracles—
Delphi.31 Its use is not restricted to Classical Greece, however, and it has

22. Parke 1985b, p. 28. pp. 29–52, for an account of these 28. Furtwängler 2006a, 2006b, 2007.
23. See n. 21, above. An inscribed excavations. 29. Bourguignon 1973, pp. 3–4;
bone tablet from Olbia may also be 25. Wiegand 1908, 1911, 1924; Maurizio 1995, p. 73.
some version of an oracular response, Knackfuss 1941. 30. Maurizio 1995, p. 76.
or ritual text, from Didyma; see Bur- 26. Tuchelt 1996; Tuchelt et al. 31. Parke and Wormell 1956, vol. 1,
kert 1994. 1996. pp. 17–41; Parke 1967a, pp. 72–89;
24. See Society of Dilettanti 1821, 27. Tuchelt et al. 1996. Price 1985.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
182 a l a n m . g r e av e s

been documented across many different periods and cultures, and in several
parts of the world, including Africa, China, and Tibet.32
At Delphi, the divination procedure involved a woman, the Pythia.
How the Pythia entered her altered state of spirit possession is unclear,
although it has been suggested that it was induced by chewing laurel (i.e.,
bay) leaves or inhaling vapor.33 Delphic responses were probably delivered
directly to the inquirer in verse form, most likely by the Pythia herself,
and not, as was once thought, after being translated into hexameters by
the prophêtes.34
At Didyma, mantic trance along the lines of the Delphic model was
the method of divination used in the post-Archaic period, following the
revival of the oracle in 334 b.c., as we know from the format of surviving
inscriptions and from descriptions in historical sources.35 There have been
various suggestions about how the altered state may have been induced
at Didyma, including sitting on an axle, being immersed in water, inhal-
ing vapor,36 or chewing a laurel leaf.37 Also in keeping with the Delphic
procedure, the responses from post-revival Didyma were recorded in verse
form, in dactylic hexameter.38
It has been assumed by some scholars that mantic trance was the
method of divination used throughout history by the oracle at Didyma.39
Parke said about the Archaic period at the site that “one pictures [the re-
plies] as being produced by some form of trance and possession like that
attributed to the Pythia.”40 Fontenrose, who has written extensively about
Didyma,41 argued that all Apolline oracles operated as Delphi did; on the
subject of Archaic Branchidai-Didyma in particular he wrote that “it is un-
certain whether a man or woman spoke for Apollo at Didyma. Presumably
this person received the god’s message as at Delphi.”42 Höckmann refers to
the “kontemplativer Haltung” (“contemplative posture”) of the statues of
“prophets” along the Sacred Way,43 although it has now been convincingly
argued that these statues were originally arranged in family sanctuaries or
groupings such as that identified on the Sacred Way between Miletos and
Didyma and are probably votive offerings to the god that are unconnected
with prophecy.44 There is, in fact, no specific evidence from Branchidai-
Didyma to suggest that mantic trance was in use in the Archaic period.
If the Delphic model was already in place at this time, then we might
expect that the inscriptions from the site would be presented in verse form.
Yet none of the three known Archaic inscriptions from the site is in verse.
Even at Delphi it cannot be proven that mantic trance in its fully developed
form operated in the Archaic period, because we know little about how the

32. On Africa, see Whittaker 1965; Maurizio 1995, p. 79; Bowden 2005, Osborne 2003, p. 272, n. 58), but
Peek 1991; Maurizio 1995, pp. 72–76; pp. 21–22. Fontenrose (1988, p. 82) questions
on China, see Park 1963, p. 198; Mau- 35. Parke 1985b, pp. 210–219; Fon- Lucian’s serious intent in this passage.
rizio 1995, p. 80; Graf 2005a, pp. 61–62; tenrose 1988, pp. 78–85. 38. Fontenrose 1988, p. 85.
on Tibet, see Fontenrose 1978, pp. 231– 36. Iambl. Myst. 3.11. 39. E.g., Burkert 1985, p. 115.
232; Tsering 2000, pp. 111–119. 37. Lucian, Bis accusatus sive tribu- 40. Parke 1985b, p. 32.
33. Holland 1933; Parke and nalia 1. Laurel trees are known to have 41. Fontenrose 1944, 1988.
Wormell 1956, vol. 1, pp. 17–18; Price played some part in the revived cult 42. Fontenrose 1978, p. 228.
1985, pp. 137–141. at Didyma (Rehm and Harder 1958, 43. Höckmann 1996, p. 93.
34. Fontenrose 1978, pp. 223–224; pp. 295–296, no. 493; Rhodes and 44. Herda 2006, pp. 349–350.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 183

early oracle worked.45 Our main source for Delphi’s operations is Pausanias,
who was writing at a time when the oracle was already in decline.46 Even
when Delphi was at its height, from the 6th century b.c. onward, other
methods of divination were used alongside mantic trance,47 perhaps owing
to high demand for the Pythia’s services. Prophecy may have been of only
secondary importance compared to the site’s primary religious functions.48
A cleft in the bedrock at Delphi appears to have played a central role in
its rituals. This idea is believed by some to be a purely mythical invention,
but the geology of the region is such that there may indeed have been a
chasm there at the time the oracle operated, possibly even one that issued
smoke.49 Such a feature would be specific to that particular geological
location and could have had no parallel at Didyma, which is in a region
with a different geology.
It would be hard to prove or disprove the use of mantic trance at Ar-
chaic Branchidai-Didyma (or indeed at any site for which there is no direct
literary or epigraphic evidence), because the process requires no materials or
paraphernalia that could reasonably be expected to survive archaeologically.
There is therefore no reason to assume, as Parke and others have, that this
was the method used at Branchidai-Didyma. Such an assumption does
not recognize the significance of the ca. 160-year gap between the Persian
sack of 494 b.c. and the oracle’s revival. As noted above, this gap appears
to have been almost complete, and the massacre of the Branchidai made
the break in cult tradition absolute. Whatever divination procedures were
used in the post-revival period, it is unlikely that they were the same as
those in use prior to the destruction of the site. The assumption that mantic
trance prevailed is a red herring that has drawn attention away from the
consideration of other possibilities.

Cle r oman c y by Mean s of Ast ragal i


Astragali—Latin tali, or “knucklebones” in English—are a common find on
archaeological sites around the world.50 The casting and reading of astragali
is thought to have been a popular method of divination in antiquity.51 As-
45. Parke and Wormell 1956, vol. 1, tragali have been widely used throughout history for cleromancy—that is,
pp. 17–18. divination by casting lots—but precisely how they were used for divination
46. Walker 1977, pp. 4–9. in Archaic Greece is unclear. In this section, I consider the physical nature
47. Amandry 1950, p. 33; Parke and of astragali and how they lend themselves to use in divination and gambling,
Wormell 1956, vol. 1, pp. 18–19; Price
1985, pp. 132–133.
before turning to the evidence for their possible use at Branchidai-Didyma.
48. Whittaker 1965, pp. 28–30;
Price 1985, pp. 144–145. The Phy sic a l Nat u re of A st rag a l i
49. Parke and Wormell 1956, vol. 1,
pp. 19–24; Fontenrose 1978, p. 476; Casting lots with astragali works on the principle that because these bones
Piccardi 2000. are roughly square in section, they can be cast like dice and one of their
50. See, e.g., Lewis 1988; Gilmour four faces will always land upright. It is impossible for them to land on the
1995; Dandoy 2006. proximal or distal ends because these ends are rounded. It is known from
51. Gilmour 1997.
inscriptions that in antiquity each of these four sides was given a name—
52. Schädler 1996. In some periods
and cultures, different names or social Πρανής, Ὕπτιος, Κῷος, and Χῖος—and each had a number associated
roles are associated with the different with it (Table 1; Fig. 2); different combinations of these numbers were
sides; see Dandoy 2006, pp. 132–133. interpreted as having a particular significance for the inquirer.52

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
184 a l a n m . g r e av e s

TABLE 1. SIDES OF AN ASTRAGALUS


Associated Figure
Ancient Name Side Description Number Reference

Pranes Dorsal Broad convex side 4 2:a


Hyptios Plantar Broad concave side 3 2:b
Koos Medial Narrow flat edge 6 2:c
Chios Lateral Narrow concave edge 1 2:d

Source: Graf 2005a, p. 60.

Figure 2. The four sides of an astra-


a b galus, comparing in each pair a mod-
ern unmodified example (left) with a
modified example (right): (a) pranes;
(b) hyptios; (c) koos; (d) chios. The
modified example is from a mixed
archaeological context at Kinet
Höyük in eastern Cilicia. Photo
T. Çakar, courtesy Marie-Henriette Gates

c d
The side that will land uppermost when the bone is thrown cannot be
predicted, but because of the relative sizes of the different sides, some sides
land upright more frequently than others. Finds of astragali from archaeo-
logical contexts show that the sides of these bones were often planed down
or filled in with metals such as lead, presumably to affect the way in which
they fell.53
To demonstrate the natural random distribution of astragali when
thrown, before modification, an experiment was conducted using unmodi-
fied bones from modern animals. The results from throwing these bones
were then compared to those using a modified astragalus from an archaeo-
logical context in order to see how the modification affected the distribution
pattern of the fall.54
Epigraphic and historical evidence from the Hellenistic and Roman
periods suggests that only the astragali of ungulates such as sheep, goats,
and pigs were used for divination in antiquity.55 For my experiment, the 53. Gilmour 1997, pp. 168–171;
bones of a modern sheep (Ovis aries), goat (Capra hircus), and wild boar Dandoy 2006, p. 133, fig. 1 (illustrating
(Sus scrofa) were used. These three species form Group 1: animals that are ground and unground astragali).
known to have been used for cleromantic divination in antiquity. 54. Experiments similar to this,
using unmodified astragali, have previ-
It is not known why only sheep, goat, and pig astragali were considered
ously been published by Lewis (1988)
suitable for cleromancy. Did it have something to do with the shape and and Schädler (1996, p. 63, n. 13, citing
weight of the astragali of these particular animals? Were their bones better Deubner 1929).
suited in some way than the bones of other species? To test the proposition 55. Nollé 1987, p. 42.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 185

TABLE 2. RESULTS OF CAST ING


ASTRAGALI OF DIFFERENT SP ECIES
Side Uppermost after Casting

Species Pranes Hyptios Koos Chios

Gr oup 1

Goat 77 74 19 30

Sheep 73 76 24 27

Pig 74 80 20 26

Gr oup 2
Cow 81 83 23 13
Buffalo 91 89 15 5
Ass 74 105 21 0
Horse 92 89 19 0

Figure 3. Graph showing the results


of casting astragali of different
species
that the way in which the astragali of these species fall influenced the deci-
sion to use them in divination, I experimented with a second group of
bones. Cow (Bos taurus), water buffalo (Bubalus bubalus), horse (Equus
caballus), and ass (Equus asinus) were selected because all of these species
lived in the region and are archaeologically attested at this time, but they
are not known to have been used for divination. These species form Group
2: animals whose bones are not known to have been used in antiquity. The
present experiment uses a wider range of species than any previous study,
which allows interspecies comparisons to be made. For the experiment,
unmodified bones of each of these seven species were cast 200 times.56 The
results are presented here in tabular and graphic form (Table 2; Fig. 3).
When the results of this experiment are reviewed, it is evident that the
outcomes from casting the astragali of Group 1 species are very similar to
one another. Those from Group 2 vary greatly from those of Group 1 and
from each other (with the exception of cow, which yields a pattern similar
to Group 1). Group 1 therefore produces a more even, and less predictable,
spread of results than the Group 2 species, with the exception of cow. The
bones of Group 2 species evidently lacked a broad enough range of results
56. The bones used were from the
study collection at the British Institute to make them usable as dice as they are very predictable in the way in
at Ankara. They were cast by hand onto which they fall. By contrast, the Group 1 species produced a good range of
a firm, flat surface. random results with no noticeable difference between the three species.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
186 a l a n m . g r e av e s

This trait may account for the choice of these particular species for the
casting of lots, although such choices are also likely to have been dictated
by religious tradition and taboo. Another factor that is likely to have in-
fluenced the choice of animals is their use in sacrifice. All of the species in
Group 1, together with cattle, were common sacrificial animals in Greek
cult. Having already been dedicated to the god once (as sacrifices), these
animals may have seemed a logical choice for use in divination within a
cult context.
Although the shape of cow astragali makes them suitable for use as
dice, their size makes them unwieldy, especially when multiple sets of bones
are being cast. The reason why cow astragali were apparently not used in
antiquity is not clear, but their considerable size may be a factor, especially
if, as has been suggested, astragali were cast using a small pouch, basket,
gourd, phormiskos, or other form of jar.57
In Group 1, the broader pranes and hyptios sides of the astragali have
the greatest probability of coming up, but the koos and chios sides still turn
up often enough to provide a relatively even spread of results. There is
no strong difference between the results for pranes and hyptios. A slight
difference exists, however, between koos and chios, with chios being more
likely to land facing up. The koos side of the astragalus is flatter than the
more pointed chios, and, when thrown, the astragalus is more likely to land
on the koos side, resulting in the chios side being uppermost more often.
To test how the pattern of distribution is affected by modifying the
bones, I conducted a second experiment. As noted above, astragali from
various archaeological contexts are frequently found to have been modi-
fied.58 Such modification can take the form of planing or flattening the 57. E.g., an astragalus-shaped jar in
edges or, in some cases, filling in the concave sides with lead. For this the British Museum (E804); see Neils
second experiment, the results of casting modified and unmodified bones 1992, pp. 230–234.
58. See n. 53, above. In addition to
were compared using sheep astragali, one of which had been modified in cult deposits, contexts include burials
antiquity by having its shorter edges (the koos and chios) smoothed off.59 A and backfilled refuse. For an example
similar experiment has been carried out with deer bones by Lewis,60 but of a burial context, see Özyiğit 1990,
this is the first time that a modified bone from an archaeological site has pp. 136–137, fig. 16. For a Hellenistic
been compared to an unmodified example of one of the species attested in grave at Erythrai and an example of
a rubbish fill, see Bar-Oz 2001. For a
classical sources. The results of the second experiment are presented here
general survey, see Gilmour 1995.
as a table and graph (Table 3; Fig. 4). 59. I am grateful to Marie-
The effect of this form of modification is to create a more even spread Henriette Gates for allowing me
of results when the bone is thrown. As Dandoy notes, “the manipulation to conduct this experiment while at
only changes the odds.”61 The shorter sides (koos and chios) have a greater Kinet Höyük. The modified bone
chance of landing faceup after modification, and correspondingly, the fre- used (no. KT.9346) was from a mixed
archaeological context, and the experi-
quency of the longer sides (pranes and hyptios) landing upright is reduced. ment was conducted on a flat, padded
Although the chance of each side landing faceup is not increased by 25% surface under the close supervision of
(or a frequency of 50 in this experiment), modification certainly produces a conservator.
a more even spread of outcomes. Crucially, this gives the astragali the 60. Lewis 1988.
appearance of creating a more random result, but how these results were 61. Dandoy 2006, p. 132.
62. For example, as many as five
interpreted, or how they were combined with the results of other astragali
astragali may have been cast at once;
thrown at the same time, is not known.62 see Paus. 7.25.10; Nollé 1987; Neils
Chi-square analysis is useful in evaluating the significance of the 1992, p. 234, n. 41; Graf 2005a,
outcomes from casting modified versus unmodified astragali. Such analysis pp. 59–60.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 187

TABLE 3. RESULTS OF CAST ING A


MODIFIED AND AN UNMODIFIED
SH EEP ASTRAGALUS
Side Uppermost after Casting

Condition Pranes Hyptios Koos Chios

Modified 58 60 38 44
Unmodified 73 76 24 27

Figure 4. Graph showing the results


of casting a modified and an unmodi-
fied sheep astragalus confirms that the act of modification did indeed change the probability
of the astragali landing on one side or another. The chi-square test also
indicates that there is a high degree of consistency in the probabilities of
throws made using bones from goat, sheep, and pig, and that there is not
a statistically significant difference in this respect between these species
and cow (although a greater number of throws would be desirable to con-
firm this). The results from casting astragali from buffalo, ass, and horse
do, however, differ systematically from casting astragali taken from the
other four species.63
To sum up, astragali of certain species of animals are shaped in a way
that makes them natural randomizing devices and ideal for use as dice. This
function can be enhanced by modifying their form and weight by shaping
and filling them, in order to create a more even, and therefore a seemingly
more random, distribution of outcomes when thrown.

63. Andrew J. Scally of the Univer- be firmly rejected. A test of the hy- three animals. A comparison of the
sity of Bradford kindly provided the pothesis that there is no change in the relative probabilities of the four sides
following statistical analysis: “The test probability of landing on any particular for cow compared to the aggregate data
of the null hypothesis that all four sides side following modification gives chi- for goat, sheep, and pig gives chi-square
of the modified bones have an equal square = 10.84 (3 degrees of freedom); = 7.64 (3 degrees of freedom); p = 0.054.
chance (25%) of landing upright yields p = 0.013. Therefore, one would con- The difference is not statistically sig-
a chi-square value of 6.88 (with 3 clude that the probabilities are changed nificant, but again a larger series may
degrees of freedom) and a significance in some way following modification. identify a systematic difference. A com-
level (p) of 0.076. Therefore, with these A test of the null hypothesis that the parison of each of buffalo, ass, and
data you would not reject the null probability of landing on any particular horse with the aggregate data for goat,
hypothesis, but a larger series of throws side is the same for goat, sheep, and pig sheep, and pig all give a Fisher’s exact
would be required to identify any resid- (i.e., Group 1) gives chi-square = 1.34 p-value of <0.001, indicating clearly
ual bias, following modification. For the (6 degrees of freedom); p = 0.97. There- that they all differ systematically from
unmodified bones, chi-square = 48.2 fore, there is a very high consistency in goat, sheep, and pig.” The size of the
(3 degrees of freedom); p < 0.001. In the probability of each side landing up- series was limited to 200 throws by
this case, the same null hypothesis can right for the unmodified bones of these conservation considerations.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
188 a l a n m . g r e av e s

The Us e of Ast ragali for Di vi nat ion


The natural form of astragali, modified or otherwise, thus make them well
suited for dice, and as noted above, they are a common find on archaeo-
logical sites across the Old and New Worlds.64 Their morphology would
have encouraged their use as divinatory randomizing devices, but their
use was by no means restricted to cult purposes. Numerous studies have
demonstrated that in many contexts the principal function of astragali ap-
pears to have been for gaming.65 Bronze astragali from the art market have
also been interpreted by some scholars as having had a practical function
as weights with no cultic significance,66 but the lack of secure provenance
complicates this interpretation. Pierced, drilled, and inlaid astragali are also
widely documented, suggesting that they were worn as amulets.67 Dandoy
has argued, however, that their use as amulets and ornaments and ultimately
for gaming derives from their primary function as a tool of divination: “in-
dividuals could have used the astragalus as a form of decoration or talisman
stemming from ritualistic uses and continuing as the bone travelled to its
more mundane place as a gaming piece.”68
Like mantic trance, divination by cleromancy is a form of spirit pos-
session. In mantic trance, the diviner believes that he or she is possessed by
a spirit, or god, who channels the divinatory response, while in cleromancy,
it is the dice, not the diviner, that are possessed by the spirit. When used in
cleromancy, astragali function as randomizing devices because, in Maurizio’s
words, “they insure that the human diviner or client cannot control the
outcome of the divination, which appears instead as the spirit’s message.”69
As Parker notes, “the oracular mechanism must be such as to provide what
has been called ‘resistance.’ ”70 That is, the use of such a device makes it ap-
pear that the inquirer has no control over the outcome.71 This may account
for why astragali were used, because of their natural propensity to create an
apparently random outcome, a feature further enhanced by modification.
The way in which the results of the casting of the astragali were con-
verted into an oracular response is less clear. The numbers allocated to
each side (see Table 1) may have had significance for the client, or could

64. See n. 50, above. have handlelike attachments, similar to (IG V ii 125) lists astragali of various
65. Amandry 1984, p. 376; Lewis that on a colossal knucklebone pre- sizes being dedicated in a temple in the
1988; Gilmour 1995; 1997, pp. 171– sumed to have come from Didyma 2nd century a.d. Given that astragali
173; Kurke 1999. On the other hand, (discussed below, p. 194). These arti- have strong cultic associations, even in
Dandoy (2006, pp. 132–133) asserts facts are all quite small, varying in size the Near East (Gilmour 1997), the pre-
that their original primary purpose was from 1.6 to 8.2 cm in length, and in sumed source for the Elayis’ “weights,”
for use in cult. weight from 7.11 to 455 g. and that Didyma provides a parallel
66. For example, Elayi and Elayi Three of these bronze astragali for astragali being cast in metal, in-
(1997, pp. 74–76, nos. 123, 126, 133, carry inscriptions: one in Phoenician scribed, and dedicated, it is possible
135) have published a number of (no. 123); one bilingual, in Greek and that the artifacts catalogued by Elayi
bronze astragali that they interpret as Phoenician (no. 126); and one whose and Elayi as “poids phéniciens?” are in
possible Phoenician weights. Because inscription is illegible (no. 133). The fact votives. Their lack of firm prove-
these items were identified from the inscriptions are just a few letters long. nance, however, inevitably leaves open
antiquities market, they have no archae- The fact that astragali are sometimes the question of their function.
ological provenance and it cannot be inscribed with the names of gods and 67. Amandry 1984, p. 377.
proven that they are weights and not heroes (see, e.g., Roller 1987, p. 125; 68. Dandoy 2006, p. 134.
cult votives, similar to those found at Dandoy 2006, p. 133) supports the 69. Maurizio 1995, p. 81.
Branchidai-Didyma. Three of the illus- suggestion that these objects were also 70. Parker 1985, p. 300.
trated examples (nos. 123, 126, and 133) cult items. An inscription from Tegea 71. Park 1963, p. 198.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 189

be interpreted as significant by the diviner. Such formulas appear to have


been presented on a table,72 but could also have been directly interpreted
by the diviner without recourse to a written formula. The astragali may
also have been inscribed with words or symbols that had significance to
the diviner. It would be wrong to think that such methods of divination
can supply only yes or no answers because, especially when used in com-
bination, cleromantic devices such as inscribed rune stones are capable of
creating complex and nuanced readings with meanings for the client that
go far beyond a simple toss of the dice.
At Delphi, there is evidence to suggest that cleromancy may have
been used alongside mantic trance throughout much of the lifetime of
the sanctuary.73 This evidence includes an inscription that records a treaty
between Skiathos and Delphi (ca. 350–340 b.c.), which has been restored
to mention a method that used two φρυκτοί, or “beans”;74 use of the verb
ἀναιρεῖν (“to take up,” presumably referring to the diviner picking up and
throwing lots);75 various images of the Pythia or Apollo holding a phiale,
which may have been used for the casting or reading of lots;76 and the
existence of a legend that depicts the Pythia choosing a Thessalian king
by lot.77 Scholarly opinion is divided as to whether cleromancy was indeed
used at Delphi and, if it was, how important a role it played. Fontenrose,
for example, ruled out the existence of any kind of oracle other than the
mantic rites of the Pythia.78 Amandry, on the other hand, argued that the
lot-oracle was of secondary importance to the prophetic oracle, and this
opinion has been maintained by others.79
The assumption that the lot-oracle was inferior to the Pythia is based
on the fact that the Pythia’s time was limited and it was less expensive
and simpler to consult the lot-oracle than to go through the lengthy full
consultation procedure.80 The preliminaries to the full procedure involved
the sacrifice of sheep or goats; purchase of a sacred cake that had a fixed,
and probably very high, price; and payment of other taxes.81 For all that,
however, the method by which an oracular response was achieved had no
bearing on its validity to the inquirer. Rather, the importance of an oracle
derived from how it was perceived,82 and, as ethnographic comparisons
have shown, the regard in which it is held by society.83 Oracles, as op-
posed to unsolicited omens such as birds, also have authority by virtue
of the fact that they are “besought.”84 Other factors that may have added

72. Paus. 7.25.10; Nollé 1987; Neils 75. Amandry 1950, pp. 25–36. no. 316; Parker 1985, p. 312.
1992, p. 234; Graf 2005a, pp. 82–94; Maurizio (1995, p. 80, n. 70) observes 78. Fontenrose 1978, pp. 222–223.
2005b, p. 37. The inscribed versions of that 10% of Delphic responses cata- 79. Amandry 1950, pp. 25–36;
such tables date to the 2nd century a.d. logued by Parke and Wormell (1956) Parke and Wormell 1956, vol. 1,
73. For full bibliography, see Mau- use this verb, but there is no pattern pp. 18–19; Walker 1977, pp. 6–7;
rizio 1995, p. 80, n. 70. evident in its usage across the surviv- Maass 1993, p. 6; 1996, p. 15.
74. Amandry 1939, p. 184; Aman- ing corpus. 80. Holland 1933; Parke and
dry 1950, p. 245; Parke and Wormell 76. Amandry 1950, pp. 66–70. See, Wormell 1956, vol. 1, pp. 30–34.
1956, vol. 1, pp. 18–19; Whittaker e.g., an Athenian red-figure cup by the 81. Parke and Wormell 1956, vol. 1,
1965, pp. 35–36. It is interesting to Codrus Painter from Vulci in Etruria, p. 32; Price 1985, pp. 134–136.
note that fortune-telling using beans, as now in Berlin, Antikensammlung 82. Price 1985, p. 148.
well as cards, coins, and other small F2538, dated to ca. 450–400 b.c. 83. Park 1963; Whittaker 1965,
objects, was still practiced in Turkey by (Beazley Archive no. 217214). pp. 43–44.
máyissas (“witches”) until the early 20th 77. Ephoros FGrH 70 F56; Parke 84. Parker 1985, p. 298.
century (Garnett 1890, pp. 142–143). and Wormell 1956, vol. 2, pp. 126–127,

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
190 a l a n m . g r e av e s

to the status of such oracles are the relative infrequency with which the
Delphic oracle spoke85 and the circulation of stories about the accuracy
of its predictions.86
In recent years, there has been increasing support for the idea that
the lot-oracle may have held a position of some significance at Delphi.
Price writes that some scholars think the lot-oracle was more important
than the Pythia, but he does not express his own opinion on the matter.87
Maurizio notes that “it is possible . . . that even the Pythia at Delphi used
cleromancy,” but she goes on to say that there is insufficient evidence to
draw a firm conclusion.88 This suggestion is certainly feasible, as it appears
that the kind of questions the Pythia was often approached with did not
require anything more than a yes or no response.89 Even if Delphi did
operate by means of cleromancy, it is not known whether astragali were
part of that procedure. Suidas, a considerably later source, expressly states
that Delphi operated by using lots,90 and Holland has connected this to the
mention of bones kept in a tripod at Delphi, although there is so far secure
evidence only for the use of beans (see above).91 Just 4 km away on Mount
Parnassos lies the Korykeion Cave of the Nymphs, a high cave sanctuary,
where 22,771 astragali were found, deposited over several centuries.92 Of
these, 3,866 were modified, 158 were inscribed, 36 were inset with lead,
and two were gilded. Treatments such as gilding with a high-status metal
like gold suggest that at least some had cult significance, possibly having
been used for divination.
Delphi was not the only oracle in Greece, and elsewhere the oracular
tradition may date back even to Homer, as there are mentions of Delphi
(Od. 8.80), Dodona (Il. 16.233–235), and an oracle that is now identified
with Cape Phokas on Lesbos (Od. 3.130).93 At Zeus’s oracle at Dodona,
which was supposedly the oldest oracle in Greece, the priestess is known
to have drawn lots, as well as to have used other methods of divination.94
Inscriptions at Tegea and Athens also mention dedications of astragali.95
From Cyprus, modified, drilled, inscribed, and weighted astragali were
found at an altar dedicated to Aphrodite at Kourion,96 and astragali were
also found at Enkomi, Tamassos, and Kition.97
In Anatolia, astragali appear to have had significance from as early as
the Chalcolithic period,98 and modified astragali are first known from a
Chalcolithic context at Alişar.99 Closer to Didyma, a modified astragalus,
sawn and polished on both sides, was found in a Middle Bronze Age
context at Aphrodisias,100 and 77 astragali were found in a single Late

85. Bowden 2005, p. 17. 1978, p. 229; Parker 1986, pp. 266–267;
86. Whittaker 1965, p. 42. Johnston 2008, pp. 63–72.
87. Price 1985, pp. 132–133. 95. Tegea: IG V ii 125; Athens: IG
88. Maurizio 1995, p. 80, esp. n. 70. II2 1533, line 32.
89. Price 1985, p. 133; Parker 1985, 96. Foster 1984, pp. 79–80.
p. 299; Maurizio 1995, p. 71. 97. Dandoy 2006, p. 132.
90. Suda, s.v. πυθώ. 98. Watson 1979, p. 199.
91. Holland 1933, pp. 202–203. 99. Osten 1937, p. 91.
92. Amandry 1984; Poplin 1984. 100. Joukowsky 1986, p. 672,
93. Quinn 1961. no. 648.13; p. 675, fig. 483:26.
94. Parke 1967b, p. 46; Fontenrose

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 191

Bronze Age context at Beycesultan.101 Similar finds are common across


Anatolia and the Near East.102 Divination by lot was known in the Hit-
tite texts of the 13th and 14th centuries b.c. as a KIN-oracle,103 and the
Hittites may have been the first to use oracles in this way.104 Although it
would be tempting to make a connection between the Bronze Age use of
astragali and lot-oracles and their use in the Archaic period, the treatment
of modified astragali from Gordion and Kaman-Kalehöyük shows a clear
change in practice between the Bronze Age and Iron Age strata of these
sites.105 The nature of their modification, and presumably also usage, is
evidently mutable, and it is therefore preferable to focus on evidence only
from the Archaic period when considering how astragali may have been
used at Branchidai-Didyma.
From Archaic Anatolia, there is plentiful evidence for astragali at the
Artemision at Ephesos. The excavators discovered “many” natural astragali
as well as complete and half specimens of astragali carved from ivory, a
number of which were modified, pierced, or incised in some way.106 These
astragali come from cultic contexts at the site and were probably used for
divination, which appears to have been performed at the temple up until
Classical times.107 Further north, there is also thought to have been an
oracular shrine in the Uyuzdere Grotto, near Metropolis. This cave was
sacred to the Mother Goddess and contained many glass models of astragali,
together with actual bones, some of which were inscribed.108 The site ap-
pears to have been a cult center from the Archaic to the Roman period,
and the astragali are cited by the excavator as evidence that divinations
were conducted there. A cleromantic oracle of Apollo Kareios has been
identified at Hierapolis,109 to the east of Didyma, and astragali were also
associated with the Anatolian cult of the Kabeiroi.110

Ev iden c e f or Ast ragal oman c y at Bran c h i dai -


Didy ma
The presence of astragali in archaeological contexts is evidently a wide-
spread cultural and historical phenomenon. Although knucklebones are
commonly associated with divination, they should not be taken as evidence
for divination per se as they also have other uses. Archaeologically, there are
three criteria relating to astragali that can be taken as suggestive of a cultic
function: the context of their discovery, their modification or production
in unusual materials, and an inscription on the astragalus itself. Examples

101. In room 13 of level II; see apparently active and others passive, 105. Dandoy 2006, pp. 135–
Lloyd 1972, pp. 13–14. and they may have been moved on a 136.
102. Amandry 1984, p. 350, n. 8; board. The specialist who performed 106. Hogarth et al. 1908, pp. 190–
Gilmour 1995, 1997. such divinations was referred to as the 192, pl. 36.
103. This appears to have been a “Old Woman” (MUNUS SHU.GI), 107. Hogarth et al. 1908, p. 191.
divinatory technique, attested only in who also performed snake-, or more 108. Meriç 1982, pp. 38–40, 101–
Anatolia, the details of which are properly, eel-oracles. See Archi 1974, 103; 1996, pp. 36–37.
uncertain, but which most likely 1982; Orlamünde 2001; Lefevre- 109. See d’Andria 2006, pp. 121–
involved the manipulation of objects Novaro and Mouton 2008. 122.
that represented concepts such as 104. Ünal and Kammenhuber 1974; 110. Amandry 1984, p. 376.
health and death. Some objects were Kammenhuber 1976, p. 10.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
192 a l a n m . g r e av e s

of each of these three criteria will be briefly examined before they are ap-
plied to the evidence from Branchidai-Didyma.
The archaeological context in which astragali are found is essential for
differentiating between their various functions. As Dandoy has commented,
“the practice of divination can be put into the same category as gaming, but
in a ritualistic context.”111 In a detailed study of the incidence of astragali
in the Iron Age Levant and eastern Mediterranean, Gilmour observed that
they frequently occur in ritual or funerary contexts.112 This would certainly
appear to be the case for the carved ivory astragali from the Artemision at
Ephesos, a number of which came from a large deposit of votive objects
found in the Artemision beneath the Central Basis, a structure of unknown
function.113 Meanwhile, astragali are also found at sites not known to be
associated with divination. For example, close to Branchidai-Didyma at
the Archaic temple of Aphrodite at Miletos, modified astragali have been
found set with lead, together with lead models of astragali.114
As discussed above, astragali are often found in a modified form. Such
modification does not in itself indicate that they were used for divination,
but it does suggest that they were modified in order to be thrown as dice
and are not merely butcher’s waste.115 Astragali can also be given treatments
such as gilding,116 or inlaid with precious materials, such as amber. They
can also be imitated in other materials.117 Such skeuomorphic models of
astragalus bones have been found cast in lead and bronze, or carved in
ivory,118 rock crystal, agate, and onyx.119 Treatment in prestigious materials
such as these may be taken as indicative of cult significance because of the
high status it affords an otherwise mundane item.
Astragali can also be inscribed, and the nature of the inscription may be
expected to give an indication that the artifact had a cult function. Various
sites in the Mediterranean have produced astragali inscribed with the names
of gods and heroes, including Nike, Achilles, and Hektor at Hellenistic
Gordion;120 Hermes at Sha’ar-Ha’amakim in Israel;121 and Herakles, Thetis,
Achilles, and Ajax at the Korykeion Cave of the Nymphs at Delphi.122 In
the context of this discussion, it is interesting to note that there are depic-
tions in vase painting of Homeric heroes gambling,123 but there are no
known myths that make an explicit connection between these particular
mythical figures and gambling. Therefore, even when astragali are inscribed
with the names of heroes or deities, they can still be interpreted as having
a noncultic function for gaming, or as being used as props for storytelling.
Inscription alone should not be taken as proof of cult.

111. Dandoy 2006, p. 132. tions, at least 18 of the astragalus- in the form of a dwarf ); GR 1814.7-
112. Gilmour 1995, 1997. shaped ivories from Archaic Ephesos 4.1134 (rock crystal); GR 1814.7-
113. Hogarth et al. 1908, pp. 190– are inlaid with amber, and one appears 4.1091 (onyx); and GR 1772.3-11.206
192. to have been gilded. (agate).
114. H. Donder (pers. comm.). On 118. Hogarth et al. 1908, p. 190. 120. Young 1962, p. 154; Roller
the site, see Senff 2003. 119. See Gilmour 1997, p. 170; 1987, p. 125; Dandoy 2006, p. 133.
115. On the butcher marks typically Dandoy 2006, p. 133. Examples of un- 121. Bar-Oz 2001.
found on astragali, see Foster 1984, certain provenance and date currently 122. Amandry 1984, pp. 370–375.
p. 80. on display in the British Museum 123. See, e.g., Paris, Musée du
116. Amandry 1984, p. 363. include GR 1864.10-7.1883 (lead); Louvre F291 (Beazley Archive
117. Judging from personal observa- GR 1772.3-11.179 (bronze astragalus no. 11295); Moignard 2006, pp. 83–84.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 193

Figure 5. View of the Archaic altar in


front of the temple at Branchidai-
Didyma. Photo A. M. Greaves

As we turn to the archaeological evidence from Archaic Branchidai-


Didyma itself, a key problem with understanding the operation of the
site in this period is that the remains of the Archaic temple building at
Branchidai-Didyma are obscured by the massive Hellenistic structure
built over it (Fig. 1). Consequently, the precise style and dimensions of
the Archaic temple remain the subject of much speculation; various re-
constructions have been proposed, only to be revisited and redrawn from
time to time in the light of new evidence.124 Despite the attention given
to the architecture of the Archaic temple building, it is often overlooked
that the single most important part of the temple complex stands clear of
the Hellenistic building, and it was here that astragali were found.
In ancient Greek cult, the altar was the most sacred part of any temple
complex.125 Altars were built up from the accumulated ash of sacrifices and
incorporated votives that can provide insights into the character of the god
worshipped at the site.126 At Branchidai-Didyma this Archaic structure is well
preserved because it stood in front of the Archaic temple and was therefore
not subsequently covered by the Hellenistic building. The altar is a large round
structure made of finely worked, hard limestone blocks (Fig. 5). Its purpose
was to hold the burned bones of animal sacrifices. During excavation it was
also found to contain pottery of the 7th and 6th centuries b.c. and, in the
excavator’s words, “eine Anzahl von bleiernen Votivastragalen.”127
This observation from Wiegand’s preliminary report is of crucial im-
portance, but it is rarely referenced by scholars writing about cult activity
124. Tuchelt 1970, 1973; Naumann
1987; Schattner 1996; Schneider 1996; at the site. It tells us not only that a number of astragali (now lost) were
Furtwängler 2006a. found at the site, but that they were cast in (or possibly modified with) lead,
125. Whitley 2001, p. 134. which indicates that these were not just detritus from the animal sacrifices
126. See, e.g., Voyatzis 1990, that the altar was primarily used for. This altar is one of the few surviv-
pp. 269–273. ing closed archaeological contexts at the site that date from the Archaic
127. Wiegand 1911, pp. 41–43, for
period, and it stood in pride of place directly in front of the temple itself.
the mention of lead votive astragali.
128. Paus. 5.13.11. According to Pausanias,128 this altar was founded by Herakles, who was
129. Paus. 7.25.10; Amandry 1984, also associated with another oracular cult using bones in a cave at Boura
p. 377; Fontenrose 1988, p. 110. in Achaia.129 There can be no clearer evidence for the cultic significance

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
194 a l a n m . g r e av e s

Figure 6. Bronze astragalus from Susa.


Paris, Musée du Louvre Sb 2719. Photo
courtesy Musée du Louvre, Paris

of astragali at Branchidai-Didyma, especially given the limited nature of


the surviving Archaic deposits at the site.
More equivocal supporting evidence comes in the form of a colossal
bronze inscribed knucklebone discovered during excavations on the acropo-
lis at Susa (Fig. 6). This knucklebone measures 27.5 x 39.0 x 24.5 cm and has
looped attachments, possibly for connecting it to another knucklebone; the
inscription on it implies that it was one of a pair.130 According to museum
records, the astragalus weighs 93.07 kg, and thus the original dedication
of the pair would have contained ca. 186 kg of bronze.131 The inscription
tells us that it was dedicated by Aristolochus and Thrason to Apollo, and
the epigraphic style suggests that the dedication took place somewhere in
Ionia, ca. 550–525 b.c.132 Many scholars believe that it came from Didyma
as an ἄγαλμα (a pleasing gift for the god) after a fruitful response from the
oracle.133 Parke writes that it has generally been accepted that “the offering
was made by two Milesians, and that it was looted by the army of Darius
in 494 b.c. and never returned to its proper home.”134 The editors of SEG
also accept this provenance,135 although there must always be a question
about the object’s connection to Branchidai-Didyma. The sheer size and
prestigious material of the artifact, however, together with its inscription,
suggest that this item had considerable significance for the site from which
it came, presumably Branchidai-Didyma.
Before considering whether these discoveries provide any evidence for
the divinatory procedures at the site, we should mention another possible
interpretation. As noted above, astragali were commonly used for gambling,
and given that Branchidai-Didyma was an oracle, it is possible that these
astragali had been dedicated in gratitude for, or anticipation of, success in
gambling about which the oracle had been consulted.136 If this were so,
then their ritual context, prestigious material, and votive inscription would
not necessarily shed light on divination procedures. In common with altar

130. Rehm and Harder 1958, device, that may have existed to join the 2007, pp. 54–55.
p. 7. original two artifacts together. The total 133. See, e.g., Perdrizet 1921,
131. For the weight, I thank Marie- weight of the original dedication can- pp. 64–68; Picard 1929, pp. 132–135;
Jo Caster, Musée du Louvre (pers. not, therefore, be determined. See also Boardman 1999, p. 108.
comm.). This is equivalent to 220 Parke 1985b, p. 31. 134. Parke 1985b, p. 31.
Milesian minas; see Hitzl 1996, 132. SEG XXX 1290; Rehm and 135. SEG XXX 1290.
pp. 151–153. This calculation does not Harder 1958, pp. 1–11, esp. 6–7; 136. Eidinow 2007, p. 54.
allow for any chain, or other linking Syll.3 3g; Jeffery 1990, p. 334; Eidinow

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 195

deposits in other Greek sanctuaries, however, the burned deposits at the altar
of Herakles probably accumulated over a period of time,137 and it is quite
possible that the lead astragali deposited there represent numerous acts of
dedication, and not a single thank offering. The discovery of carved bone
dice at the sanctuary of Aphrodite at Miletos would appear to constitute
less equivocal evidence for this form of one-off dedication.138 One cannot
be certain about the details of the dedication of the lead astragali, but the
high-profile context (the altar), the valuable materials (lead and bronze),
and the monumental size of the Susa astragalus suggest a ritual significance
beyond that required of grateful gamblers or generic cult dedications.
As Amandry noted, astragali may also have functioned as amulets,139
which could account for their presence in cult contexts at sites not otherwise
associated with divination. But because the votive astragali at Branchidai-
Didyma were made of lead and were therefore heavy, it seems unlikely that
they would have been worn as amulets.
The monumental astragalus found at Susa bears an inscription saying
that it came ἀπὸ λείο δεκάτη (“as a tithe from the booty”).140 Although the
precise meaning of ἀπὸ λείο in this context has been questioned, no one
suggests that it refers to winnings or the proceeds of gambling.141 If we ac-
cept the most popular interpretation, that this was booty, we can speculate
along with Jackson that one of the three surviving Archaic responses from
the oracle can be read as sanctioning a raiding party.142 According to this
view, then, the artifact is best understood as a tithe of the spoils from one
such successful raid that was dedicated to the god who had sanctioned it.
In such a martial context, the astragalus form of the dedication makes no
sense unless it is connected to the oracle, without whose approval the raid
would not have taken place.
In summary, a review of the available archaeological evidence from
Branchidai-Didyma has revealed one key fact: the lead astragali found
together with Archaic pottery at the altar in front of the temple provide
the only direct archaeological evidence for any kind of activity that may
be connected to divination. The proposition that divination at Archaic
Branchidai-Didyma was carried out by means of cleromancy using astragali
is therefore a reasonable one. It may even have been the core form of divina-
tion and should not be considered merely a kind of sideshow, secondary to
the “main” oracle, usually presumed to have involved mantic trance. This
137. Cf. Voyatzis 1990, p. 42; Whit- proposition is strengthened if we accept the common interpretation that
ley 2001, pp. 134–136. the monumental bronze astragalus from Susa was originally taken from
138. Miletos Museum Z94.43.5 and Branchidai-Didyma. The argument does not hinge on acceptance of this
Z94.132: Greaves 1999, p. 184, n. 72.
view, however, because the smaller lead astragali are from a more secure
139. Amandry 1984, p. 377.
140. Trans. Eidinow 2007, p. 54. archaeological context at the site itself.
141. Liddell and Scott translate λεία Such an interpretation does not contradict the epigraphic evidence
as “booty” or “plunder”; see also SEG from the site. Cleromancy is capable of generating more complex replies
XXX 1290; Parke 1985b, p. 229. than a simple yes or no, and it is feasible that the responses recorded in the
142. Didyma DI.11: Rehm and surviving inscriptions were achieved by this means. Oracles were often ap-
Harder 1958, p. 8, no. 11; Jackson
proached with questions that were worded in such a way as to require only
1995. See also n. 177, below.
143. Parker 1985, p. 301. See also a simple yes or no response, and they often gave “simple answers to simple
Whittaker 1965, pp. 27–28; Fontenrose questions.”143 It was the function of oracles to provide guidance on matters
1978, pp. 36–38; 1988, pp. 90–91. in which a decision was required, not to predict the future, and oracles were

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
196 a l a n m . g r e av e s

therefore consulted in order to approve or disapprove a particular course


of action.144 As Whittaker has stressed, it is also possible to create a longer
response by means of cleromancy,145 for example, by casting a number of
dice, runes, or similar items and interpreting the results; entering a trancelike
state of “spirit-possession”; or inverting the original question.146
In her discussion of the astragalus from Susa, which she attributes to
Branchidai-Didyma, Eidinow writes:
There were other such lot oracles in Asia Minor—in which a throw
of dice provided the answer to a ‘yes or no’ or ‘x or y’ question. . . .
Here, the dedication of the bones may suggest that a similar process
was used at Didyma, at least during the Archaic period. The two
other 6th-century oracular responses, both of which are in prose
and can be easily fitted to questions constructed around alternative
options, would support this.147
This statement reflects a common misunderstanding of the nature of the
responses that can be achieved by the casting of astragali. It is not the
case that if one wanted to achieve a response to “a ‘yes or no’ or ‘x or y’
question,” one would simply toss a coin, or a similar two-sided random-
izing device. Throwing multiple sets of four-sided astragali, which may
have been modified to increase the likelihood of a random outcome or
inscribed or otherwise endowed with cultic meaning, is unlikely to produce
anything as simple as a “yes or no” answer. On the contrary, the use of
astragali suggests a sophisticated mantic procedure in which faith is put
in the reader’s ability to interpret the signs accurately. Mantic trance was
therefore not the only means of divination by which a narrative response
could be achieved.
What is clear from an archaeological perspective is that astragali did
have cult significance at the site of Archaic Branchidai-Didyma. The sug-
gestion that they were connected to the site’s operation as an oracle is en-
tirely consistent with known oracular practices and responses at cult sites
in Anatolia and Greece and is also consistent with the form and content
of the epigraphic evidence from the site itself.

De li v e ry of the Re s p ons e
The method by which an oracle arrived at its responses and how these were
then communicated to the inquirer are two separate processes, and it is
important to consider them separately. We have only limited evidence
from Archaic Branchidai-Didyma, but the matter of how the results of
divination were delivered to the inquirer is nonetheless worth discussing.
As noted above, it was once thought that the incoherent mutterings
of the Pythia at Delphi were transposed into verse by the prophêtes before
being delivered to the inquirer. Parke and Wormell wrote that the Pythia
144. Whittaker 1965, p. 27; Price 146. For inverting questions, see
1985, pp. 132–133; Maurizio 1995, also Fontenrose 1988, p. 180; and
p. 71. n. 177, below.
145. Whittaker 1965, pp. 35–36. 147. Eidinow 2007, pp. 54–55.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 197

“was regarded as the unconscious instrument of a divine revelation and


as such she counted for little.”148 Maurizio has argued, however, that the
Pythia herself delivered her pronouncements directly in verse form, and
cites evidence that, in Greek literature, poetry was often depicted as coming
from a divine source and that to deny this is to deny the Pythia her agency
as a woman and a poet.149
As at Delphi, the oracle at post-revival Didyma appears to have been
a woman, the prophêtis or promantis.150 Fontenrose assumed that her re-
sponses were then versified by a priest because he could not envisage that
the promantis could be under Apollo’s divine inspiration and still compose
complex dactylic hexameter verse.151 By contrast, the recorded oracular
responses from Archaic Branchidai-Didyma are all in prose, not verse, and
the delivery procedure must have been different from the Delphic model.152
In the post-revival period at Didyma, the oracle’s pronouncements appear to
have been delivered from a raised platform above the pronaos and addressed
to those standing below.153 This feature was probably an architectural in-
novation of the Hellenistic period under Ptolemaic Egyptian influence,154
and cannot therefore be assumed to have influenced the appearance or
function of the building in the Archaic period.
The Archaic inscriptions from Branchidai-Didyma raise a particular
question about how the responses were delivered. In Greek tradition the
gods did not generally speak directly to mortals, even at oracular sites;
instead, they always spoke through a medium. Yet at Branchidai-Didyma,
one of the three surviving Archaic inscriptions records the voice of the god
in the first person: μυθέομαι “I said . . .”;155 the two other surviving Archaic
responses are given in the third person: θεὸς πεν “the god said. . . .”156
Herodotos’s account of a consultation at Branchidai-Didyma also
seems to imply that the god himself spoke directly to a consultant.157 In
this passage, Herodotos describes how the men of Kyme had asked the
oracle whether they should give up a suppliant of theirs to the Persians. The
oracle had replied that they should. But handing over a suppliant to one’s
enemies was considered an impious act, and so Aristodikos, a citizen of
Kyme, took exception to this response. He traveled to Branchidai-Didyma
148. Parke and Wormell 1956,
and posed the same question, and once again the oracle gave the same
vol. 1, pp. 34–35.
149. Maurizio 1995, esp. pp. 79, 86; reply. Outraged, Aristodikos began pulling birds’ nests from the walls of
see also Flower 2008. the temple until the voice of the god from the inner shrine (the adyton)
150. Fontenrose 1988, pp. 55–56; spoke to him, chastising him for his impiety in dislodging the birds that
Morgan 1989, p. 27. had taken refuge in his temple. Aristodikos replied that it was also impious
151. Fontenrose 1988, pp. 84–85. for the men of Kyme to give over a man who had come to them in search
152. Parke 1985b, p. 29; Fontenrose
1988, pp. 179–181.
of refuge. The voice spoke again, telling him never to return to ask such
153. Parke 1986, pp. 127–128. questions of his oracle again.
154. Parke 1986, p. 123. In this passage we are led to believe that the god spoke directly to
155. Kawerau and Rehm 1914, Aristodikos and responded to him, thereby entering into a dialogue, albeit
pp. 397–398, no. 178; Parke 1985b, a short one. The phrasing of the command indicates that we are supposed
pp. 28–32.
to believe that this was the voice of Apollo himself. The historical truth
156. Rehm and Harder 1958, p. 8,
no. 11; Kawerau and Rehm 1914, of the passage is obviously questionable, and it is in no way an accurate
pp. 276–277, no. 132a. description of a typical consultation at this, or any other, oracle. Herodotos
157. Hdt. 1.158–159. himself inserts the phrase λέγεται (“it is said”) when the god first speaks,

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
198 a l a n m . g r e av e s

implying a degree of skepticism about the part of the story he is retelling.


Significantly, this passage tells us nothing about how the original responses
had been delivered, whether by direct speech or some other means. Parke
suggested that this episode was probably an invention or elaboration of
a story that is historically accurate in other respects.158 There are clearly
many unusual circumstances surrounding this particular episode that call its
historicity into question. For example, the people of Kyme were breaking
the rules of the oracle in seeking exemption from a religious obligation.159
Whatever the truth of the incident may have been, Herodotos’s telling of
it remains an exceptional case in which a god supposedly spoke directly to
a mortal at a Greek oracular temple.
Although exceptional, such direct speech is not unprecedented. The
oracle of Trophonios at Lebadeia also spoke directly to its inquirers.160 It
was not unheard of for consultants at the Delphic oracle to receive a spon-
taneous response from the Pythia;161 a post-revival inscription from Didyma
records that the god spoke to the prophet Ulpius Athenagoras αὐτομα-
τισμός (i.e., speaking spontaneously).162
Comparative ethnographic studies of cultures that practice divination
indicate that gods and spirits are sometimes thought to speak directly to
mortals. This process can take place via a medium or in other ways whereby
the god’s voice is projected, seemingly from nowhere. At the oracle of Mwari
in southern Rhodesia (now Zimbabwe), for example, a woman is hidden in
the back of a cave.163 In such cults it seems to be important for the person
who is speaking with the voice of the god to be out of sight of the inquirer.
Similar pronouncements may have been given at sites in Greece, where
the architecture suggests that oracles were delivered from a hidden location.
At Corinth, for example, oracles may have been delivered from a tunnel
under a shrine.164 Although no architecture exists in the cave at the oracle
of Trophonion at Lebadeia,165 Clark has speculated that “the voice of the
unseen δαίμων was probably that of a concealed priest.”166
The rock-cut Phrygian shaft monuments of central western Anatolia
are near contemporaries of the Archaic oracle at Branchidai-Didyma, and
have been interpreted by Berndt-Ersöz as possible oracular cult sites.167
As a significant form of Anatolian religious monument found close to
Branchidai-Didyma itself, these shaft monuments deserve further con-
sideration. The sites include Mal Taş, Bahşış, Değirmen Yeri, Fındık, and
Delikli Taş. These monuments consist of vertical shafts cut directly into
the bedrock and sometimes situated behind a rock-cut facade or niche,
with a hole cut through the rock connecting the shaft to the facade. In her
interpretation of the monuments, Berndt-Ersöz proposed that a priest, or
similar person, could be hidden from sight in the shaft behind the facade
or niche and pronounce prophecies through the connecting hole, giving

158. Parke 1985b, pp. 15–17. p. 192, no. 278; Fontenrose 1988, 1994, pp. 66–89.
159. Parker 1985, p. 313. p. 103. Fontenrose, however, classifies 166. Clark 1968, p. 73.
160. Hdt. 1.46, 8.134; Paus. 4.32.5– this as a fictional response (p. 232). 167. Berndt-Ersöz 1998, pp. 96–98;
6. See Clark 1968; Schachter 1994, 163. Fontenrose 1978, p. 230. 2003, pp. 171–177; 2006, pp. 89–142,
p. 80. 164. Mee and Spawforth 2001, on date; 2003, pp. 239–242; 2006,
161. Parke 1962; Parker 1985, p. 154. pp. 143–205, on function; see esp. 2006,
pp. 318–319. 165. Paus. 9.39.14. For further pp. 191–193, on divination.
162. Rehm and Harder 1958, discussion of this oracle, see Schachter

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 199

the impression that it was the voice of the mountain, or the goddess Kybele
herself, that was speaking.
As Berndt-Ersöz herself notes, however, there are limitations to this
interpretation of the Phrygian shaft monuments. For example, the shaft at
Fındık is too small to accommodate a person, and even if the shafts were
oracular sites, there are other ways in which they might have been used.
Divination by lot may have been conducted, for instance, and the stones or
lots (or indeed astragali) may then have been delivered through the holes.168
Anatolian parallels such as these will always present us with a problem of
identification and interpretation because of the absence of contemporary
written sources.
Since Berndt-Ersöz’s reevaluation of the Phrygian shaft monuments
was published, a similar monument has been published from Ephesos,
to the north of Branchidai-Didyma in Ionia.169 Cut into the bedrock of
the hill known as Bülbüldağ on the edge of the Classical city of Ephesos,
this new monument has all the appearance of a Phrygian cult site, with
steps and a facade, behind which was a circular shaft sunk into the rock.
Although, as with the Fındık monument, this shaft appears to be too small
to accommodate a person, the similarity to the Phrygian shaft-monuments
group in general is remarkable.
Although the neighboring contemporary Anatolian cultures of Phry-
gia, Lydia, and Caria played a major role in defining the identity of the
Ionian Greeks,170 including those at Branchidai-Didyma, they have left
very few written sources about their divinatory practices that might provide
a historical context for the interpretation of the archaeological remains of
their cult sites.171 It is possible that the peoples of contemporary Anatolia
practiced divination,172 as their better-documented Hittite and Luwian
predecessors had done before them.173 Without textual evidence, however,
it is difficult to argue convincingly that divination took place. If, for ex-
ample, there had been no written evidence for the existence of an oracle at
Trophonion in Greece, there would have been no archaeological evidence
to distinguish it from any other cave and its existence as an oracle would
never have been suspected. This lack of textual evidence for the operation,
or even existence, of Anatolian oracular sites therefore makes it difficult to
balance this region’s influence on the cult at Branchidai-Didyma with that
from better-documented and excavated sites in Greece.174
In summary, the delivery of oracles directly to the inquirer by a priest,
168. Corsten 1997, p. 42; 2002, or similar individual, who was hidden from sight appears to be a practice
p. 127; Berndt-Ersöz 2006, p. 193. common across various divinatory traditions, both ancient and modern,
169. Bammer and Muss 2006. although such a separation would have made it harder for the priest to
170. Greaves 2010, pp. 219–232. base judgments on the reactions of the inquirer—an important element
171. The inscribed clay tablets
found at the Carian oracular site of
even in seemingly objective divinatory practices such as the mathematical
Labraunda are, however, a tantalizing practice of astrology.175 As we have seen, a convincing argument has been
clue that such texts may have once put forward by Berndt-Ersöz that this may have been how the shaft monu-
existed. See Meier-Brügger 1983; ments of Phrygia (and now Ephesos) were used, which would provide a
Greaves 2010, p. 17. regional parallel for such a practice at Branchidai-Didyma. Although there
172. Lefevre-Novaro and Mouton
is no archaeological evidence that the speaker was hidden from sight at
2008.
173. See n. 103, above. Branchidai-Didyma, because of the lack of architectural evidence from
174. Greaves 2010, pp. 1–26. the east end of the temple building, the epigraphic and literary evidence
175. Thomas 1971, p. 287. suggests that responses were delivered by direct speech.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
200 a l a n m . g r e av e s

DISCU SSI ON

Although no great air of secrecy surrounded the workings of ancient Greek


oracles, we have little firm evidence about how they operated. It is the nature
of oracles that even when their existence is historically well attested, they
leave very little archaeological evidence of their activities.176 That said, it is
still worthwhile to consider in detail the methods and processes of divination
used at different oracular sites, especially in the period before Delphi rose to
preeminence and became the model for oracles across the Greek world. If
we can gain a better understanding of how the central mantic procedure at
an oracular site such as Branchidai-Didyma worked, we will gain a deeper
understanding of the nature and character of divination in the pre-Delphi
era, and the local and regional basis for the relationship between the “ra-
tional” Greeks and their seemingly irrational devotion to these sacred sites.
At Archaic Branchidai-Didyma, the available evidence is insufficient to
determine conclusively the central divination process. We should not allow
ourselves, however, to assume that the oracle operated by mantic trance
simply because this was the method used following its Hellenistic revival.
There is no aspect of the site’s archaeology, epigraphy, or history that indi-
cates that mantic trance was used in the Archaic period. There are, however,
indications that astragali, which are known to have been widely used for
divination in the ancient world, played an important cultic role at the site,
and may well have been used for cleromantic divination. Furthermore,
the epigraphic and historical evidence, as well as parallels in Anatolia and
Greece, supports the suggestion that responses from Branchidai-Didyma
were delivered to the inquirer by direct speech, with the speaker possibly
being concealed from sight.
Such an interpretation, based on archaeological remains, does not con-
tradict the surviving epigraphic evidence: cleromancy is capable of producing
answers that are more substantial than a simple yes or no response, and we
do not know what questions were posed to the oracle, only its responses.177
Nor does this interpretation conflict with the closest surviving historical
account of Herodotos (1.158–159), which may give us some indication of
the mode of delivery, albeit under exceptional circumstances, but tells us
nothing about the mantic process itself.
Although the most persuasive pieces of evidence in support of such a
176. Morgan 1990, pp. 149–150.
conclusion—the lead astragali from the altar in front of the temple build- 177. Fontenrose (1988, p. 180) has
ing—have been known since the publication of Wiegand’s preliminary conjectured what one of the questions
report in 1911, scholars have continued to argue that divination here was was by inverting the answer: “plunder-
by mantic trance. One reason why the astragali from Branchidai-Didyma able? It is right to do as your fathers”
have been overlooked may be the fact that they are mentioned in an ar- (Didyma DI.11: Rehm and Harder
1958, p. 8, no. 11) may have been the
chaeological publication, specifically a preliminary excavation report.178
response to the question “Is it right to
The starting point for studies such as those of Parke and Fontenrose has do as our fathers have done to keep
generally been the literary and epigraphic record, not archaeology. In com- from being the subject of plunder?”
mon with other branches of historical archaeology, classical archaeology has 178. See Greaves 2010, p. 23, on
often taken written sources as its starting point, valuing them over material differing attitudes toward the use of
evidence.179 This orientation has resulted in scholars making connections preliminary reports by different archae-
ological communities of practice.
between the archaeological evidence that they find and a preexisting 179. Shanks 1996, pp. 5–6; Johnson
“metanarrative,” or structure of knowledge, that they have developed based 1999, p. 154.
on their readings of classical literature.180 180. Greaves 2010, pp. 27–44.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 201

Such a text-based approach will also inevitably favor Greek cultural


parallels over the contemporary cultures of Anatolia, whose languages are
less well understood and produced no extant body of literature comparable
to that of ancient Greek history. There are scant ethnohistorical sources
that allow the contemporary Anatolian cultures to speak for themselves. Yet
Anatolian archaeology can provide important parallels for the Branchidai-
Didyma material and it is essential to recognize this if we are to form a
balanced picture of the site. Anatolian cultures are now recognized as having
had a significant influence on the development of Branchidai-Didyma and
the site; perhaps even the cult figure of Apollo himself had pre-Hellenic
Anatolian origins.181 Some features of the site are common to both Greek
and Anatolian cultures, such as the fact that it was built around a water
source.182 There are other elements, however, that are more identifiably
Anatolian in character, such as the Carian name of the site183 and possible
Urartian elements in its architectural design.184 In the absence of contem-
porary textual evidence from Anatolia, and in light of new discoveries at
Ephesos of Anatolian-style cult installations185 and recognized Anatolian
influences on the site of Branchidai-Didyma itself, it is therefore entirely
reasonable to bring the Phrygian shaft monuments into consideration
when speculating about how the oracle may have delivered its responses.
Contemporary scholarship of Greek religion has been greatly influ-
enced by the work of Walter Burkert.186 Burkert placed great emphasis on
altered states and the “prophetic madness of Apollo” and dismissed lot-
oracles as just a “decision-making mechanism.”187 In doing so, however,
181. Parke 1985b, pp. 2–3; Burkert he failed to recognize that possession and cleromancy are not mutually
1985, p. 117; Fontenrose 1988, p. 117;
Brown 2004. See also Graf 2009,
exclusive, as discussed above, and that the cleromantic diviner can practice
pp. 130–142, esp. pp. 132, 136–137, her (or his) art while in a state of spirit possession. He also misrepresented
where he questions attempts to place cleromancy as a simplistic device, one not afforded high status, which is
the origins of Apollo in Anatolia on contrary to the ethnographic record and historical instances of its use in
etymological grounds. ancient Greece. It is now understood that it was often not the oracle that
182. Berndt-Ersöz 1998, p. 97;
was simplistic, but the questions that were posed to it.188
2003, pp. 11–13; Whitley 2001,
pp. 134–136. It is important to take into account the chronological variance in prac-
183. Gorman 2001, p. 187, n. 37. tice at oracular sites. For example, it has long been recognized that mantic
184. Greaves 2002, pp. 111–114. trance was a later innovation at Delphi, probably introduced from the
185. Summarized in Greaves 2010, Near East and then superimposed on preexisting cults.189 Prior to the
p. 195. introduction of mantic trance, Delphi appears to have worked by means
186. See, e.g., Parker 1986, p. 273.
187. Burkert 1985, p. 111.
of hypnomanteia (the interpretation of dreams, also referred to as “incuba-
188. Parker 1985, p. 301. tion”)190 or cleromancy,191 and it is conceivable that the nearby Korykeion
189. Burkert 1985, pp. 116–117. Cave may have been a precursor to the Pythian oracle farther down the
190. Eur. IT 1259–1260; Hec. mountain.192 Delphic influence can then be seen to have extended over
70–71. See Clark 1968, p. 74. the operations of a number of other oracles and cult sites, including the
191. Suda, s.v. πυθώ; Holland 1933,
Boiotian cults of Apollo Ptoion, Apollo Tegyra, and Apollo Telphousa.193
p. 203.
192. Amandry 1984, pp. 377–378, Recognizing the possibility of variance in practice over time is even more
citing Hom. Hymn Merc., lines 550–558. relevant at Didyma, where there was a break in cult practice caused by
193. Schachter 1981, pp. 54, 75, and the Persian sack and the slaughter of the Branchidai. To suggest that
76–77, respectively. Branchidai-Didyma may have operated by means of mantic trance in the
194. See Greaves 2010, p. 156, on Archaic period because this is how it is known to have operated in the
teleological argumentation in the inter-
pretation of destruction horizons in
Hellenistic period is a teleological argument; that is, it seeks to move the
Ionia and their presumed links to the argument toward a predetermined outcome, in this case the fact that the
Ionian Revolt. oracle operated by mantic trance in the Hellenistic and Roman periods.194

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
202 a l a n m . g r e av e s

This is also contrary to the evidence from Delphi and elsewhere that shows
that mantic trance was a later development.
Another reason why some scholars may have interpreted the oracle at
Archaic Branchidai-Didyma as operating by mantic trance is the inclina-
tion to produce reductionist models to explain ancient Greek culture. Such
models serve only to homogenize our image of the Greeks and deny them
their individual agency and localized character. As Burkert recognized, “the
methods of imparting oracles are almost as varied as the cult forms.”195 Fon-
tenrose tried to demonstrate that all oracles of Apollo, regardless of where
they were in the Greek world, operated in the same manner. He wrote:
“Other Apolline Oracles appear to have been similar to Delphi in methods
and operations,”196 and “all non-Apolline oracles in Hellas make use of
divinatory devices or of incubation. . . . These various devices yielded yes/
no answers or made choices.”197 In this way Fontenrose seeks to make sense
of a diverse, limited, and often confusing data set by imposing uniformity
onto it, but in so doing he misrepresents the nature of the different forms
of divination and the possibility of differences in practice between cults of
Apollo. It would be more appropriate to recognize the pre-Hellenic origins
of Branchidai-Didyma and those characteristics of its cult practice that
differed from oracles elsewhere in the Greek world. Such an approach also
recognizes that the people who developed the cult expressed their agency
by choosing to actively combine elements of Greek and Anatolian cult at
the site, and that they were not part of a “homogenized” Greek identity.
Likewise, it was their choice to defer decisions of great importance to the
will of Apollo, as expressed by the fall of astragalus bones, or other such ran-
domizing devices. There is no reason to consider cleromancy an inherently
lesser form of divination than mantic trance, which is, after all, just another
randomizing device, but one in which the diviner’s body is the tool of choice.
Fontenrose and others have assumed that cleromancy cannot have
been used at Branchidai-Didyma because they misunderstand the nature
of the responses that this method of divination is capable of producing. For
example, Parke wrote: “For the first time among our literary or epigraphic
examples Apollo speaks in the first person. So it would be unreasonable to
conjecture that the response was obtained by drawing lots.”198 Yet we must
remind ourselves again that we underestimate cleromancy if we think it is
only capable of producing simple answers. The responses that it produces
can be complex, and the diviner who uses astragali or similar devices often
does so while in a possessed, or otherwise altered, state and would be quite
capable of speaking with the authority, or even the voice, of the god.199 It
is also evident that the questions posed to the oracles were formulaic and
often would not have required more than a simple response, or an inversion.
There is consistency in the way in which many Greek oracles oper-
ated, but not in the way that Fontenrose proposed. The connection is that
many of them operated by means of some form of spirit possession. As 195. Burkert 1985, p. 114.
Maurizio demonstrated through examination of ethnographic comparisons, 196. Fontenrose 1978, p. 228.
197. Fontenrose 1978, p. 229.
all divination occurs by means of spirit possession, because the spirits work
198. Parke 1985b, pp. 29–30.
to affect randomizing devices that reveal the answer to the question posed 199. Holland 1933, p. 203; Maurizio
by the inquirer.200 Such devices can be objects, including astragali, but they 1995, p. 80.
can also be a person, such as the Pythia. It is spirit possession that is the 200. Maurizio 1995.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 203

common element of the divinatory practices at both pre- and post-revival


Didyma, Delphi, and many other Greek and non-Greek oracles. It could
be conjectured that the same was also true of how the oracle arrived at her
responses. That is, that each priestess arrived at her responses using different
randomizing devices (e.g., beans, astragali, her dreams, herself ), but it was
her ability to be possessed by Apollo while divining that was important,
not the means by which she achieved her responses. With this in mind,
it is worth remembering that the only firm evidence for any conceivable
form of divination at Branchidai-Didyma remains the discovery of the
astragali reported in 1911.

CON C LU S I ON S

As Whittaker pointed out some time ago, “oracles occur because man is
cursed with knowledge that the future exists and at the same time that he
is unable to know how to control it.”201 This being so, it is understandable
that oracles and fortune-telling are a feature of many seemingly rational
societies, including our own.
We may be uncomfortable joining our popular notion of the “rational”
Ionians—the founders of pre-Socratic philosophy—with the idea that
their most sacred and influential oracle operated on something as simple
as the toss of knucklebones.202 But such a notion of rationality is a modern
construct, and it has been argued here that the Ionians who consulted the
oracle at Branchidai-Didyma did indeed submit themselves, and their
states, to decisions made in this way. As with all divination, the success of
the oracle at Branchidai-Didyma will have depended on the power of belief
and context,203 and there are clear examples in Greek literature of the lot
being used to decide issues of momentous social or political importance,
such as the sending out of colonists, the selection of officials, or the division
of land.204 There is, therefore, no reason to assume that divination by lot
could not have had a significant, possibly even central, role at this impor-
tant temple. The oracle’s apparent lack of independence from Miletos did
not affect its popularity with Greeks and non-Greeks from across western
Anatolia,205 and neither should the way in which it operated.
As I have stressed repeatedly, divination by cleromancy using astragali
is strongly indicated by the archaeological evidence from, and associated
with, the site of Branchidai-Didyma, and would be consistent with the
epigraphic evidence, but we have no proof of how the oracle operated.
Similarly, the separate suggestion, discussed here, that responses were
delivered by direct speech (and possibly from a hidden location) cannot be
proven without unequivocal archaeological, textual, or epigraphic evidence.
This interpretation is, however, in keeping with our current understanding
201. Whittaker 1965, p. 43. of contemporary practices in Anatolia and Greece. Oracles undoubtedly
202. See Greaves 2010, p. 94, on changed their mode of operation over time and may also have employed
Ionian philosophy and economics.
several methods of divination at once. It is thus necessary to question the
203. Whittaker 1965; Price 1985.
204. See, e.g., Hdt. 4.153; Taylor assumption that all ancient Apolline oracles worked by means of mantic
2007; Eidinow 2007, p. 71. trance if we are to gain a more accurate picture of them as independent
205. Greaves 2002, pp. 124–127. regional cult entities.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
204 a l a n m . g r e av e s

REF ERENCES

Amandry, P. 1939. “Convention reli- Brown, E. L. 2004. “In Search of Ana- ———. 1988. Didyma: Apollo’s Oracle,
gieuse conclue entre Delphes et tolian Apollo,” in ΧΑΡΙΣ: Essays Cult, and Companions, Berkeley.
Skiathos,” BCH 63, pp. 183–219. in Honor of Sara A. Immerwahr Foster, G. V. 1984. “The Bones from
———. 1950. La mantique apollinienne (Hesperia Suppl. 33), ed. A. P. the Altar West of the Painted Stoa,”
à Delphes: Essai sur le fonctionnement Chapin, Princeton, pp. 243–257. Hesperia 53, pp. 73–82.
de l’oracle, Paris. Burkert, W. 1985. Greek Religion: Ar- Furtwängler, A. 2006a. “Didyma,” in
———. 1984. “Os et coquilles,” in chaic and Classical, trans. J. Raffan, Stadtgrabungen und Stadtforschung
L’antre corycien 2 (BCH Suppl. 9), Oxford. im westlichen Kleinasien: Geplantes
Athens, pp. 347–380. ———. 1994. “Olbia and Apollo of und Erreichtes. Internationales Sym-
Archi, A. 1974. “Il sistema KIN della Didyma: A New Oracle Text,” posion 6./7. August 2004 in Bergama
divinazione ittita,” OA 13, pp. 113– in Apollo: Origins and Influences, (Türkei) (Byzas 3), ed. W. Radt,
144. ed. J. Solomon, Tucson, pp. 49– Istanbul, pp. 73–80.
———. 1982. “Hethitische Mantik 60. ———. 2006b. “Didyma 2004,” Kazı
und ihre Beziehungen zur mesopo- Clark, R. J. 1968. “Trophonion: The Sonuçları Toplantısı 27, pp. 205–
tamischen Mantik,” in Mesopota- Manner of His Revelation,” TAPA 212.
mien und seine Nachbarn: Politische 93, pp. 63–75. ———. 2007. “Didyma 2005,” Kazı
und kulturelle Wechselbeziehungen im Corsten, T. 1997. “Ein neues Buchsta- Sonuçları Toplantısı 28, pp. 405–418.
alten Vorderasien vom 4. bis 1. Jahr- benorakel aus Kibyra,” EpigAnat 28, Garnett, L. M. J. 1890. The Women
tausend v. Chr. XXV. Rencontre assy- pp. 41–49. of Turkey and Their Folk-Lore,
riologique internationale Berlin, 3. bis ———. 2002. Die Inschriften von Ki- London; http://archive.org/stream/
7. Juli 1978 (Berliner Beiträge zum byra (Inschriften Griechischer womenturkeyandt00garngoog#page/
Vorderen Orient 1), ed. H.-J. Nissen Städte aus Kleinasien 60), Bonn. n4/mode/2up (accessed March 15,
and J. Renger, Berlin, pp. 279–293. Curnow, T. 2004. The Oracles of the An- 2012).
Bammer, A., and U. Muss. 2006. “Ein cient World, London. Gilmour, G. 1995. “The Archaeology
Felsdenkmal auf dem Bülbüldağ Dandoy, J. R. 2006. “Astragaloi through of Cult in the Southern Levant in
von Ephesos,” Anatolia Antiqua 14, Time,” in Integrating Zooarchaeology. the Early Iron Age: An Analytical
pp. 65–69. Proceedings of the 9th Conference of and Comparative Approach” (diss.
Bar-Oz, G. 2001. “An Inscribed the International Council of Archaeo- Oxford Univ.).
Astragalos with a Dedication to zoology, Durham, August 2002, ed. ———. 1997. “The Nature and Func-
Hermes,” NEA 64, pp. 215–217. M. Maltby, Oxford, pp. 131–137. tion of Astragalos Bones from Ar-
Berndt-Ersöz, S. 1998. “Phrygian d’Andria, F. 2006. “Hierapolis in Phry- chaeological Contexts in the Levant
Rock-Cut Cult Façades: A Study gia,” in Stadtgrabungen und Stadt- and Eastern Mediterranean,” OJA
of the Function of the So-Called forschung im westlichen Kleinasien: 16, pp. 167–175.
Shaft Monuments,” AnatSt 48, Geplantes und Erreichtes. Internatio- Gorman, V. B. 2001. Miletos: The Orna-
pp. 87–112. nales Symposion 6./7. August 2004 ment of Ionia: A History of the City
———. 2003. “Phrygian Rock-Cut in Bergama (Türkei) (Byzas 3), to 400 B.C.E., Ann Arbor.
Shrines and Other Religious Mon- ed. W. Radt, Istanbul, pp. 113– Graf, F. 2005a. “Rolling the Dice for
uments: A Study of Structure, 124. an Answer,” in Mantikê: Studies in
Function, and Cult Practice” (diss. Deubner, L. 1929. “Zum Astragalspiel,” Ancient Divination (Religions in the
Stockholm Univ.). AA 1929, pp. 272–282. Graeco-Roman World 155), ed. S. I.
———. 2006. Phrygian Rock-Cut Ehrhardt, N. 1998. “Didyma und Milet Johnston and P. T. Struck, Leiden,
Shrines: Structure, Function, and in archaischer Zeit,” Chiron 28, pp. 51–97.
Cult Practice (Culture and History pp. 11–20. ———. 2005b. “Würfel- und Los-
of the Ancient Near East 25), Eidinow, E. 2007. Oracles, Curses, and orakel,” in Thesaurus cultus et rituum
Leiden. Risk among the Ancient Greeks, antiquorum (ThesCRA) III: Divina-
Boardman, J. 1999. The Greeks Overseas, Oxford. tion, Prayer, Veneration, Hikesia,
4th ed., London. Elayi, J., and A. G. Elayi. 1997. Asylia, Oath, Malediction, Profana-
Bourguignon, E. 1973. “Introduction: Recherches sur les poids phéniciens tion, Magic Rituals and Addendum to
A Framework for the Comparative (Transeuphratène Suppl. 5), Paris. Vol. II Consecration, ed. J. Paul Getty
Study of Altered States of Con- Flower, M. A. 2008. The Seer in Ancient Museum, Los Angeles, pp. 37–39.
sciousness,” in Religion, Altered States Greece, Berkeley. ———. 2009. Apollo, New York.
of Consciousness, and Social Change, Fontenrose, J. E. 1944. The Festival Greaves, A. M. 1999. “A Socio-
ed. E. Bourguignon, Columbus, Called Boegia at Didyma (CPCA 1, Economic History of Miletos
pp. 3–35. no. 11), Berkeley. to the End of the Archaic Period”
Bowden, H. 2005. Classical Athens and ———. 1978. The Delphic Oracle: Its (diss. Univ. of Leeds).
the Delphic Oracle: Divination and Responses and Operations, with a ———. 2002. Miletos: A History,
Democracy, Cambridge. Catalogue of Responses, Berkeley. London.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
d i v i n at i o n at a r c h a i c b r a n c h i d a i - d i d y m a 205

———. 2010. The Land of Ionia: Society Lloyd, S. 1972. Beycesultan III.1: Late Social Contexts,” JRAI 93, pp. 195–
and Economy in the Archaic Period, Bronze Age Architecture, London. 209.
Chichester. Maass, M. 1993. Das antike Delphi: Parke, H. W. 1962. “A Note on αὐτο-
Herda, A. 2006. Der Apollon-Delphinios- Orakel, Schätze, und Monumente, ματίζω in Connection with Proph-
Kult in Milet und die Neujahrspro- Darmstadt. ecy,” JHS 32, pp. 145–146.
zession nach Didyma: Ein neuer ———. 1996. Delphi: Orakel am Nabel ———. 1967a. Greek Oracles, London.
Kommentar der sog. Molpoi-Satzung der Welt (Archäologische Veröffent- ———. 1967b. The Oracles of Zeus:
(Milesische Forschungen 4), Mainz. lichungen des Badischen Landes- Dodona, Olympia, Ammon, Oxford.
Hitzl, K. 1996. Die Gewichte griechischer museums 1), Karlsruhe. ———. 1985a. “The Massacre of
Zeit aus Olympia (OlForsch 25), Maurizio, L. 1995. “Anthropology and the Branchidae,” JHS 105, pp. 59–
Berlin. Spirit Possession: A Reconsidera- 68.
Höckmann, U. 1996. “Die Sitzstatue tion of the Pythia’s Role at Delphi,” ———. 1985b. The Oracles of Apollo in
des ‘Propheten’ aus Didyma,” JHS 65, pp. 69–86. Asia Minor, London.
IstMitt 46, pp. 93–102. Mee, C. B., and T. Spawforth. 2001. ———. 1986. “The Temple of Apollo
Hogarth, D. G., C. Harcourt-Smith, Greece: An Oxford Archaeological at Didyma: The Building and Its
A. H. Smith, B. V. Head, and A. E. Guide (Oxford Archaeological Function,” JHS 106, pp. 121–131.
Henderson. 1908. Excavations at Guides), Oxford. Parke, H. W., and D. E. W. Wormell.
Ephesus: The Archaic Artemisia, Meier-Brügger, M. 1983. Die karischen 1956. The Delphic Oracle, 2 vols.,
London. Inschriften (Labraunda II.4), Stock- Oxford.
Holland, L. B. 1933. “The Mantic holm. Parker, R. 1985. “Greek States and
Mechanism at Delphi,” AJA 37, Meriç, R. 1982. Metropolis in Ionien: Greek Oracles,” in Crux: Essays
pp. 201–214. Ergebnisse einer Survey-Unternehm- Presented to G. E. M. de Ste. Croix on
Jackson, A. H. 1995. “An Oracle for ung in den Jahren 1972–1975 (Bei- His 75th Birthday (History of Politi-
Raiders?” ZPE 108, pp. 95–99. träge zur klassischen Philologie 142), cal Thought 6), ed. P. A. Cartledge
Jeffery, L. H. 1990. The Local Scripts of Königstein. and F. D. Harvey, London, pp. 298–
Archaic Greece: A Study of the Origin ———. 1996. Metropolis. Excavations: 326.
of the Greek Alphabet and Its Devel- The First Five Years, 1990–1995, ———. 1986. “Greek Religion,” in
opment from the Eighth to Fifth Cen- Izmir. The Oxford History of the Classical
turies B.C., rev. ed., with a supplement Moignard, E. 2006. Greek Vases: An World, ed. J. Boardman, J. Griffin,
by A. W. Johnston, Oxford. Introduction (Classical World and O. Murray, Oxford, pp. 254–
Johnson, M. 1999. Archaeological Series), Bristol. 274.
Theory: An Introduction, Oxford. Morgan, C. 1989. “Divination and Peek, P. M., ed. 1991. African Divina-
Johnston, S. I. 2008. Ancient Greek Div- Society at Delphi and Didyma,” tion Systems: Ways of Knowing,
ination, Chichester. Hermathena 147, pp. 17–42. Bloomington.
Joukowsky, M. S. 1986. Prehistoric ———. 1990. Athletes and Oracles: Perdrizet, P. 1921. “Le témoignage
Aphrodisias: An Account of the Exca- The Transformation of Olympia and d’Eschyle sur le sac d’Athènes par
vations and Artifact Studies (Archae- Delphi in the Eighth Century B.C., les Perses,” RÉG 39, pp. 57–79.
ologia transatlantica 3), Providence. Cambridge. Picard, C. 1929. “Les antécédents des
Kammenhuber, A. 1976. Orakelpraxis, Naumann, R. 1987. Didyma Führer, ‘astragalizontes’ Polyclétéens, et la
Träume, und Vorzeichenschau bei den Istanbul. consultation par les dés,” RÉG 42,
Hethitern (Texte der Hethiter 7), Neils, J. 1992. Goddess and Polis: The pp. 121–136.
Heidelberg. Panathenaic Festival in Ancient Piccardi, L. 2000. “Seismotectonic
Kawerau, G., and A. Rehm. 1914. Athens, Princeton. Remarks and a Hypothesis for the
Das Delphinion in Milet (Milet I.3), Nollé, J. 1987. “Südkleinasiatische Geologic Environment of a Myth,”
Berlin. Losorakel in der römischen Kaiser- Geology 28, pp. 651–654.
Knackfuss, H. 1941. Didyma I: Die zeit,” AntW 18, pp. 41–49. Poplin, F. 1984. “Contribution ostéo-
Baubeschreibung, 3 vols., Berlin. Orlamünde, J. 2001. “Überlegungen archéologique à la connaissance des
Kurke, L. 1999. “Ancient Greek Board zum hethitischen KIN-Orakel,” in astragales de l’antre corycien,” in
Games and How To Play Them,” Kulturgeschichten: Altorientalistische L’antre corcycien 2 (BCH Suppl. 9),
CP 94, pp. 247–267. Studien für Volkert Haas zum 65. Athens, pp. 381–393.
Lefevre-Novaro, D., and A. Mouton. Geburtstag, ed. T. Richter, D. Pre- Price, S. 1985. “Delphi and Divina-
2008. “Aux origins de l’ichthyo- chel, and J. Klinger, Saarbrücken, tion,” in Greek Religion and Society,
mancie en Anatolie ancienne: pp. 295–311. ed. P. E. Easterling and J. V. Muir,
Sources textuelles et données Osten, H. H. von der. 1937. The Alishar Cambridge, pp. 128–154.
archéologique,” Anatolica 34, Hüyük Seasons of 1930–32 (OIP 30), Quinn, J. D. 1961. “Cape Phokas,
pp. 7–52. Chicago. Lesbos: Site of an Archaic Sanctu-
Lewis, R. B. 1988. “Old World Dice in Özyiğit, Ö. 1990. “1988 Yılı Erythrai ary for Zeus, Hera, and Dionysus?”
the Protohistoric Southern United Sondaj Çalışmaları,” Kazı Sonuçları AJA 65, pp. 391–393.
States,” CurrAnthr 29, pp. 759– Toplantası 11, pp. 125–150. Rehm, A., and R. Harder. 1958.
768. Park, S. K. 1963. “Divination and Its Didyma II: Die Inschriften, Berlin.

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms
206 a l a n m . g r e av e s

Rhodes, P. J., and R. Osborne, eds. ed. C. Gates, J. Morin, and T. Zim- Kultbezirk an der Heiligen Strasse von
2003. Greek Historical Inscriptions: mermann, Oxford, pp. 29–34. Milet nach Didyma, Mainz.
404–323 B.C., Oxford. Society of Dilettanti. 1821. Antiquities Ünal, A., and A. Kammenhuber. 1974.
Roller, L. E. 1987. “Hellenistic Epi- of Ionia I, London. “Das althethitische Losorakel Kbo
graphic Texts from Gordion,” Taylor, C. 2007. “From the Whole XVIII 151,” Zeitschrift für verglei-
AnatSt 37, pp. 103–133. Citizen Body? The Sociology of chende Sprachforschung 88, pp. 157–
Rubinstein, L., and A. M. Greaves. Election and Lot in the Athenian 180.
2004. “Ionia,” in An Inventory Democracy,” Hesperia 76, pp. 323– Voyatzis, M. E. 1990. The Early Sanctu-
of Archaic and Classical Poleis, 345. ary of Athena Alea at Tegea and the
ed. M. H. Hansen and T. Heine Thomas, K. 1971. Religion and the Other Archaic Sanctuaries in Arcadia
Nielsen, Oxford, pp. 1053–1107. Decline of Magic: Studies in Popular (SIMA-PB 97), Göteborg.
Schachter, A. 1981. Cults of Boiotia: Beliefs in Sixteenth and Seventeenth Walker, A. 1977. Delphi, Athens.
Acheloos to Hera (BICS 38.1), Century England, London. Watson, P. J. 1979. Archaeological Eth-
London. Tsering, D. 2000. Dalai Lama, My Son: nography in Western Iran (Viking
———. 1994. Cults of Boiotia: Potnia A Mother’s Story, London. Fund Publications in Anthropology
to Zeus (BICS 38.3), London. Tuchelt, K. 1970. Die archaischen Skulp- 57), Tucson.
Schädler, U. 1996. “Spielen mit Astra- turen von Didyma: Beiträge zur früh- Whitley, J. 2001. The Archaeology of
galen,” AA 1996, pp. 61–73. griechischen Plastik in Kleinasien Ancient Greece, Cambridge.
Schattner, T. G. 1992. “Didyma, ein (IstForsch 27), Berlin. Whittaker, C. R. 1965. “The Delphic
minoisch-mykenischer Fundplatz?” ———. 1973. Vorarbeiten zu einer Oracle: Belief and Behavior in
AA 1992, pp. 369–372. Topographie von Didyma: Eine Ancient Greece and Africa,” HTR
———. 1996. “Architrav und Fries des Untersuchung der inschriftlichen 57, pp. 21–48.
archaischen Apollontempels von und archäologischen Zeugnisse Wiegand, T. 1908. Sechster vorläufiger
Didyma,” JdI 111, pp. 1–23. (IstMitt-BH 9), Tübingen. Bericht über die von den Königlichen
Schneider, P. 1996. “Zum alten Sekos ———. 1988. “Die Perserzerstörung Museen in Milet und Didyma unter-
von Didyma,” IstMitt 46, pp. 147– von Branchidai-Didyma und ihre nommenen Ausgrabungen, Berlin.
152. Folgen: Archäologisch betrachtet,” ———. 1911. Siebenter vorläufiger
Senff, R., 2003. “Das Aphroditeheilig- AA 1988, pp. 427–438. Bericht über die von den König-
tum von Milet,” Asia Minor Studien ———. 1991. Branchidai-Didyma: lichen Museen in Milet und Didyma
49, pp. 11–25. Geschichte und Ausgrabung eines unternommenen Ausgrabungen,
Shanks, M. 1996. Classical Archaeology antiken Heilgtums (Zaberns Bild- Berlin.
of Greece: Experiences of the Disci- bände zur Archäologie 3), Mainz. ———. 1924. Achter vorläufiger Bericht
pline, London. ———. 1992. “Tieropfer in Didyma: über die von den Königlichen Museen
Slawisch, A. 2009. “Epigraphy versus Ein Nachtrag,” AA 1992, pp. 143– in Milet und Didyma unternommenen
Archaeology: Conflicting Evidence 217. Ausgrabungen, Berlin.
for Cult Continuity in Ionia during ———. 1996. “Schriftenverzeichnis bis Young, R. S. 1962. “The 1961 Cam-
the Fifth Century b.c.,” in Sacred 1996,” IstMitt 46, pp. 13–16. paign at Gordion,” AJA 66, pp. 153–
Landscapes in Anatolia and Neigh- Tuchelt, K., P. Schneider, T. G. Schattner, 168.
boring Regions (BAR-IS 2034), and H. R. Baldus. 1996. Didyma: Ein

Alan M. Greaves
Univ ersit y of Liv er p o ol
sc ho ol of ar c haeol o g y, c l assics, and eg y p tol o g y
12–1 4 aber cr omby sq uare
liv er p o ol l69 7wz
united kingd om
greav es@liv.ac.uk

This content downloaded from


138.236.79.137 on Sun, 30 Oct 2022 15:18:21 UTC
All use subject to https://about.jstor.org/terms

You might also like