an ensemble of artificially cut rock chambers dated to the 4th or 3rd century BCE, in Kordestān Province, 20 km west of Takab. The site is of considerable importance because of its Greek inscription, one of the very few examples preserved in situ in Persia.


KARAFTO CAVES, an ensemble of artificially cut rock chambers with a Greek inscription, dated to the 4th or 3rd century BCE, in Kordestān Province.

History of Discovery. The chambers were carved into a rock ridge, 20 km west of Takab (see AZERBAIJAN), and are only accessible through a common entrance on the southside, ca. 9 m above the ground. Here the cliff precipitates almost vertically (FIGURE 1), whereas to the north the terrain slopes gradually. The site is of considerable importance because of its Greek inscription (FIGURE 2; see EPIGRAPHY ii), one of the very few examples preserved in situ in Persia.

The first European to have written about the site was the British army surgeon Dr. John Cormick (d. 1833), who served also as personal physician to the crown prince ʿAbbās Mirzā (1789-1833; cf. Bernard, pp. 301, 303 n. 3). Sir Robert Ker Porter (1777-1842) used Cormick’s unpublished diary when in 1819 he recorded the site and its inscription in his subsequently published travelogue (II, pp. 540-54, cf. 563). In 1838, Sir Henry Rawlinson (1810-95) visited the caves and presented an improved reading of the inscription (1840, I, pp. 44-45; II, pp. 100-101). He gave his copy to the well-known British topographer and antiquarian William Martin Leake (1777-1860), who, with remarkable precision and accuracy, recognized the inscription’s essential points: firstly, the place was under the protection of Herakles; secondly, the two lines may have been verses; and thirdly, the form of the characters belongs to the 4th or 3rd century BCE. On the advice of W. W. Tarn (1869-1957), an authority in the ancient history of the East, Sir Aurel Stein (1862-1943) explored Karafto on three days in the summer 1936 (Stein, pp. 326-45). He produced a plan of the chambers, which are arranged on two levels above each other (ibid., p. 329 fig. 22), and took a squeeze of the inscription (ibid., p. 326 figs. 98-99). The squeeze allowed Marcus N. Tod (1878-1974), at that time lecturer of Greek at Oriel College, Oxford, to confirm that the paleographic characteristics of the letters dated the inscription to the early Hellenistic period, between the end of the 4th and the beginning of the 3rd century BCE (Stein, pp. 332, 337-38). In 1975, Rudolf Naumann (1910-96) and the author visited the site, and the result was Professor Naumann’s improved plan of the rock chambers (FIGURE 3) and the first photographs of their interior, taken by the author (FIGURE 1, FIGURE 4, FIGURE 5, FIGURE 6; cf. von Gall, 1978, pls. 23-28).

Identification. The Greek inscription above the door to room k (FIGURE 2; cf. von Gall, 1978, p. 92 fig. 1, after Stein, fig. 98) reads:

ʿĒraklēs [ʾenth]ade katoikei   Herakles resides here

mēthen eiselthoi kakon      Nothing evil may enter

This distichon is a well-known Greek apotropaic formula known from many private houses in the Greek mother country, from the advanced Hellenistic until the Roman period (Weinreich, pp. 13-14; Bernard, pp. 304-305). In several Christian churches in the East (Weinreich, pp. 15-18), the pagan formula has survived in a Christian version which replaced Herakles with Christ, or even God.

Already Johannes Franz (1804-51; cf. Gutschmid, p. 89), and later W. W. Tarn (1951, p. 68), connected the site of Karafto with a passage in the Annals (12.13-14, tr. p. 344) of Tacitus (56-ca. 120 CE), where he mentioned a mountain called Sanbulos, probably located in the north of Media (Weissbach, p. 2232). At this place several local deities were venerated, the most outstanding of whom was Herakles. An oracle of horses relied on dreams to convey to Herakles’ priests the way he had taken. The historical context is an internal Parthian struggle for power between the ruling monarch Gotarzes II (r. ca. 38-51 CE; see GŌDARZ II) and his rival Meherdates who was backed by the Roman emperor. Any identification of Mt. Sanbulos has to be sought on an axis running from Arbela (the modern Arbil) to Ecbatana (the modern Hamadan), since the events of 49/50 CE must have taken place on this route: Arbela was the meeting point of Meherdates’ auxiliary force, while Gotarzes most probably had departed from the Arsacid summer residence at Ecbatana (Strabo,16.1.16) to intercept and finally defeat his adversary (Stein, pp. 340-41).

Some scholars locate this mountain in the vicinity of Kermānšāh (for the identification with Sonbola Kuh, see Rawlinson, 1839, pp. 41-42). Others even identify it as Mt. Bisotun (for a complete bibliography, see Tubach, pp. 245-50), arguing that the rock relief shows Gotarzes II. But the inscription “Gotarses Geopothros” above the Bisotun relief panel is certainly not a reference to a ruling Great King of the Arsacids and, rather, names a local dynast who is depicted as victor over an unidentified enemy in an unknown battle (so already, if hesitatingly, Rawlinson, 1839, p. 116; cf. Kawami, pp. 42-43; von Gall, 1990, pp. 11-12).

According to Tacitus, Gotarzes halted at Mt. Sanbulos to offer vows to the local deities and thereafter encamped at a defensive line formed by a river Corma, probably the Lesser Zab in Iran (Zab-e Kuček; cf. Stein, p. 341 n. 6). Since the cliffs of Karafto perfectly match the described landscape Stein considered it proven that this place was Tacitus’s Mt. Sanbulos. Consequently, he interpreted the rock chambers an oracle sanctuary “dedicated to a local god, whom the Greeks called Herakles” (Stein, 1940, p. 339).

Interpretation. The author (1978, pp. 99-112) has supported Stein’s view, while adding further comparative material drawn from Greek sanctuaries. Of particular importance are the accurately carved barrel vaults in chambers b and k (FIGURE 5) and the devices to bar the doors from the inside in chambers k and l (FIGURE 4; von Gall, 1978, pp. 95-96). The lintel of a window in room k shows elements of wood architecture (FIGURE 6).

The very limited lookout on the southside, whose view is destructed by the surrounding rocks (FIGURE 5; cf. von Gall, 1978, pl. 27 no. 3), speaks strictly against a military interpretation, for example, as frontier post between Greater Media and Media Atropatene (Bernard, pp. 307-308). But also a purely residential character (Boyce and Grenet, p. 85) can be excluded due to the lack of any provisions for such a purpose.

Remnants of a broad staircase in the corridor (i in FIGURE 1), which leads from the lower to the upper level, suggest the performance of solemn ceremonies. The traces of a meḥrāb in room g are important, since they indicate that even in Islamic times the site was used for prayer.

It seems advisable, though, to place the chambers of Karafto neither in a strictly religious, nor in a purely secular context. Strabo (64/63 BCE.-23 CE ?) cites a passage by Eudoxus of Cnidus (first half of the 4th century BCE) when he describes Hyrcania (modern Gorgān) in the southeast of the Caspian Sea: “there are some cliffs facing the sea with caverns underneath [....] and rivers flowing from the precipices above” (11.7.5, tr.,  p. 257). Amidst a marvelous nature, native people gathered for “feasting and sacrifice, and sometimes they recline in the caverns down below and and sometimes they enjoy themselves” (ibid.) in the waterfalls. Franz Cumont (1868-1947) overemphasized the religious aspect in his view that the Hyrcanian caves were archetypes of the Mithraic spelaea (p. 56; see MITHRAISM), since Eudoxus’s passage leaves no doubt that the caves’ visitors were also looking for pleasure. The rock-hewn chambers at Karafto with its scenic splendor may have had a different function too: Some may have served as assembly halls for feasting and meetings, while others could have been used for religious purposes.


L. W. Adamec, Historical Gazetteer of Iran I: Tehran and Northwestern Iran, Graz, 1976.

P. Bernard, “Héraclès, les grottes de Karafto et le sanctuaire du mont Sambulos en Iran,” Stud. Ir. 9, 1980, pp. 301-324.

M. Boyce and F. Grenet, A History of Zoroastrianism: III – Zoroastrianism under Macedonian and Roman Rule, Leiden, 1991.

F. Cumont, Textes et monuments figurés relativs au mystères de Mithra, vol. I, Brussels, 1899.

J. Franz, Corpus Inscriptionum Graecarum, vol. III, Berlin 1853, p. 270, no. 4673.

H. von Gall, “Die Kulträume in den Felsen von Karaftu bei Takab (West-Azarbaidjan),” AMI, NF 11, 1978, pp. 91-112.

Idem, Das Reiterkampfbild in der iranischen und iranisch beeinflußten Kunst parthischer und sasanidischer Zeit, Teheraner Forschungen 6, Berlin, 1990.

A. von Gutschmid, Kleine Schriften, ed. F. Rühl, vol. III, Leipzig, 1892.

T. Kawami, Monumental Art of the Parthian Period in Iran, Acta Iranica 26, Teheran,1987.

R. K. Porter, Travels in Georgia, Persia, Armenia, Ancient Babylonia, 2 vols., London 1821-22.

H. Rawlinson, “Notes on a March from Zohab to Khúzistán (Susiana),” JRGS 9, 1839, pp. 26-116.

Idem, “Notes on a Journey from Tabriz through Persian Kurdistán, to the Ruins of Takhti- Soleíman,“ JRGS 10, 1840, I, pp. 45-64.

Idem, “Memoir on the Site of the Atropatenian Ecbatana,” JRGS 10, 1840, II, pp. 65-158.

Tacitus, Annals, tr. J. Jackson, vol. IV, Loeb Classical Library 322, London, 1937.

A. Stein, Old Routes of Western Iran, London, 1940.

Strabo, The Geography, tr. H. L. Jones, vol. V, Loeb Classical Library 211, London, 1928.

W. W. Tarn, The Greeks in Bactria and India, 2nd ed., Cambridge, 1951; please note: 1st ed., Cambridge, 1938, and rev. and enlarged 3rd ed., Chicago, 1985.

J. Tubach, “Herakles vom Berge Sanbulos,” Ancient Society 26, 1995, pp. 241-71.

O. Weinreich, “De dis ignotis questiones selectae: B – De Diogenis quae fertur epistula XXXVI et de epigrammatis graeci historia,” ARW 18, 1915, pp. 8-18.

F. H. Weissbach, “Sanbulos,” Pauly – Wissowa, I A 2, Berlin, 1920, p. 2232.

(Hubertus von Gall)

Originally Published: December 15, 2010

Last Updated: April 24, 2012

This article is available in print.
Vol. XV, Fasc. 5, pp. 533-536