I. CARRHAE
In Libanius' speech For the Temples (Or. 30), sometimes regarded as the crowning work of his career, he refers to an unnamed city in which a great pagan temple had recently been destroyed; the date of the speech is disputed, but must be in the 380 s or early 390 s, near the end of the speaker's life.Footnote 1 After deploring the actions of a governor appointed by Theodosius, often identified with the praetorian prefect Maternus Cynegius,Footnote 2 Libanius continues (30.44–5):
Let no-one think that all this is an accusation against you, Your Majesty. For on the frontier with Persia (πρὸς τοῖς ὁρίοις Περσῶν) there lies in ruins a temple which had no equal, as one may hear from all who saw it, so very large was it and so very large the blocks with which it was built, and it occupied as much space as the city itself. Why, amid the terrors of war, to the benefit of the city's inhabitants, those who took the city gained nothing because of their inability to take the temple as well (τοῖς ἑλοῦσι τὴν πόλιν οὐκ ἔχουσι κἀκεῖνον προσεξελεῖν), since the strength of the walls defied every siege-engine. Besides that, one could mount up to the roof and see a very great part of enemy territory, which gives no small advantage in time of war. I have heard some people disputing which of the two sanctuaries was the greater marvel, this one that has gone, or one that one hopes may never suffer in the same way, and contains Sarapis. But this sanctuary, of such a kind and size, not to mention the secret devices of the ceiling and all the sacred statues made of iron that were hidden in darkness, escaping the sun – it has vanished and is destroyed.
Jacques Godefroy (Gothofredus), best known for his edition of the Theodosian Code, also produced the editio princeps of the speech For the Temples, supplying a Latin translation and extensive notes. He hesitated whether to identify the city in question with Apamea in Syria or with Carrhae, ‘urbs superstitione Gentilicia tum referta’, but opted for a third choice: Edessa, the capital of Osrhoene. In doing so he took for granted that a law of the Theodosian Code (16.10.8), in which the emperors order that a pagan temple in Osrhoene remain open, referred to the same temple; I shall argue below that this is incorrect. Opinion continues to be divided, though with a majority favouring Edessa.Footnote 3 But this lay some ten or fifteen miles from the border with Persia, whereas Carrhae was directly on it, and is much more likely than Edessa to have had a temple from which one ‘could observe a vast area of enemy country’. The principal deity of Carrhae (Harran) was Sîn, the Moon God, said by some sources to be male, by others to be female. Describing how Caracalla was assassinated while on a pilgrimage to the god, Cassius Dio says that he had ‘set out from Edessa for Carrhae’, and was murdered on the way: according to Herodian, he was staying in Carrhae when he decided to go in advance of his army ‘and to reach the temple of the Moon, whom the local people greatly revere: the temple is a long way from the city [presumably Carrhae], so as to require a (special) journey’. Another emperor to visit the sanctuary was Julian on his march into Babylonia. Theodoret of Cyrrhus alleges that ‘he entered the sanctuary honoured by the impious’ and cut open a human victim, a woman suspended by the hair, in order to obtain an omen of his future victory.Footnote 4
In favour of Carrhae are some details in Libanius' description that have not received the attention they deserve, notably ‘the statues wrought in iron … hidden in darkness far from the sunlight’ (ὅσα ἀγαλματα σιδήρου πεποιημένα κέκρυπτο τῷ σκότῳ διαϕεύγοντα τὸν ἥλιον).Footnote 5 Iron statues are a great rarity. Pausanias singles out one at Messene (4.31.10) and comments on an image of Hercules and the hydra at Delphi. ‘Now to make images out of iron is a most difficult and laborious process. The work of Tisagoras (whoever he is) is therefore wonderful. And wonderful, too, in a high degree are the heads of a lion and wild boar at Pergamus, which are also of iron: they were made as offerings to Dionysus’ (10.18.6. tr. Frazer). Libanius appears to connect the fact that the statues were of iron with their ‘escaping the sun’, and that suits a temple dedicated to the Moon god, in which iron ‘hidden in darkness’ would have enhanced the nocturnal effect. The ‘secret devices’ recall the Serapeum of Alexandria, with which Libanius compares the temple in this same passage. According to Rufinus, this had a magnet set in the ceiling panels in order to give the impression that the Sun had arisen and was bidding farewell to Serapis, and Rufinus adds that ‘there were many other things as well built on the site by those of old for the purpose of deception’.Footnote 6 Lastly, Libanius might imply that the temple and its surrounding sanctuary were not within the city, but occupied an area as large as the city itself; the surrounding wall was impregnable, and the roof of the temple was so high that it gave a good view into enemy territory.
Seton Lloyd and William Brice made a three-week survey of Carrhae and its environs in 1950, and were able to trace the original city wall. Searching for the Moon temple, they noted that one Arabic source identifies it with the castle in the south-east quadrant of the city, while ‘others refer to shrines clearly situated some distance from the city’. After discussing one of these, at a site called ’Ain-al-’Arus, they observed: ‘Another outlying village, having similar associations with the moon-cult, proved of even greater interest. This was Aşağı Yarımca, a modern village beside a high mound and an open pool of water, situated about four miles from Harran, on what may have been the old road to Edessa. It would thus correspond to the situation of the moon-temple visited by the Emperor Caracalla shortly before his assassination; and indeed it was here that in 1949, the discovery of a stele, bearing the emblem of the god Sin and a cuneiform inscription, led to the recognition of a very large stone building, apparently of the Assyrian period, just beneath the surface.’Footnote 7 Lloyd and Brice were perhaps influenced in favour of a site outside the city by Herodian's phrase, ‘the temple is a long way from the city’, though the city in question is probably Edessa rather than Carrhae. The castle would have the advantage of being closer to the Persian border, though it is incorporated in the circuit of the present walls and not outside them, and could hardly be said to be as large as the entire city (though Libanius may well be exaggerating). A third possibility is the conspicuous hill within the city circuit.Footnote 8 Without a more modern investigation, a decision does not seem possible.
If the temple referred to by Libanius was indeed the principal one of the Moon at Carrhae, its destruction did not signify the end of paganism there: it is only necessary to qualify the view that ‘the old order continued [in Carrhae] as if nothing had happened’.Footnote 9 In the fifth century Theodoret of Cyrrhus several times mentions the paganism of Carrhae, while asserting that under its bishop Abraham it ‘remained free of its previous thorns, and now flourishes with the harvests of the Spirit’. If this was ever true, the city relapsed into its old ways: when the Persian king Chosrhoes was ravaging Roman Mesopotamia in 540, and had plundered Syrian Antioch and other Christian cities, ‘the people of Carrhae met him offering much money, but he said it did not belong to him, since most of the inhabitants are not Christians, but are of the old belief’.Footnote 10 According to a celebrated hypothesis of Michel Tardieu, a continuous strain of Neoplatonic teaching can be traced from sixth-century Harran down to the ninth century, though this theory must probably be rejected.Footnote 11
II. BATNAE
The above-mentioned law of Theodosius and his co-rulers, dated to November 382, is addressed to Palladius,Footnote 12 the dux of Osrhoene, and orders that a previously pagan temple be allowed to remain open and unharmed, though not used for sacrifice. I give the text of Mommsen and Krueger (to which I have added some punctuation) and the translation of Clyde Pharr:Footnote 13
IDEM AAA. Palladio duci Osdroenae. Aedem olim frequentiae dedicatam coetui et iam populo quoque communem, in qua simulacra feruntur posita artis pretio quam diuinitate metienda, iugiter patere publici consilii auctoritate decernimus, neque hic rei obreptiuum officere sinimus oraculum. Ut conuentu urbis et frequenti coetu uideatur, experientia tua omni uotorum celebritati seruata auctoritate nostri ita patere templum permittat oraculi, ne illic prohibitorum sacrificiorum huius occasione aditus permissus esse credatur. DAT. PRID. KAL. DEC. CONSTANTINOP(OLI) ANTONIO ET SYAGRIO CONSS.
The same Augustuses to Palladius, Duke of Osrhoene. By the authority of the public councilFootnote 14 We decree that the temple shall continually be open that was formerly dedicated to the assemblage of throngs of people and now also is for the common use of the people, and in which images are reported to have been placed which must be measured by the value of their art rather than by their divinity; We do not permit any divine imperial response that was surreptitiously obtained to prejudice this situation. In order that this temple may be seen by the assemblages of the city and by frequent crowds, Your Experience shall preserve all celebrations of festivities, and by the authority of Our divine imperial response, you shall permit the temple to remain open, but in such a way that the performance of sacrifices forbidden therein may not be supposed to be permitted under the pretext of such access to the temple. Given on the day before the kalends of December at Constantinople in the year of the consulship of Antonius and Syagrius.
An oddity of this ruling is that the emperors seem to refer to a particular temple, and yet do not locate it. The reason is doubtless that for the compilers of the Codex what mattered was the ruling, not the particular place to which it had originally applied, still less the name of the particular god.Footnote 15 Mommsen and Krueger in their edition of the Codex appear to have sensed this difficulty, since they suggested that the name of some pagan god (nomen numinis cuiusdam gentilium) had been excised, and almost casually they cited a passage of Ammianus (14.3.3). This concerns a large trading fair held at the city of Batnae in ‘Anthemusia’, a region of Osrhoene near Edessa:
Batnae municipium in Anthemusia conditum Macedonum manu priscorum ab Euphrate flumine breui spatio disparatur, refertum mercatoribus opulentis, ubi annua sollemnitate prope Septembris initium mensis ad nundinas magna promiscuae fortunae conuenit multitudo ad commercanda quae Indi mittunt et Seres aliaque plurima uehi terra marique consueta.
The town of Batnae in Anthemusia, built by the hand of the Macedonians, is separated by a short distance from the river Euphrates, filled with rich merchants, where by an annual custom about the beginning of September a large throng of persons of differing degrees of wealth meet to trade in goods sent by the Indians and the Seres and in many other items usually conveyed on land and sea.
On his march into the Persian realm in 363, Julian avoided Edessa because of its devotion to Christianity and instead stayed in Batnae, ‘which had a large and ancient temple of Zeus’ (ἔχουσαν μέγα Διὸς ἱερὸν ἀρχαῖον). There he met with an ill omen when fifty calones died in a fire.Footnote 16
Some have identified the temple mentioned by Theodosius in his ruling of 382 with the one mentioned by Libanius, and have placed this composite temple in Edessa.Footnote 17 But it was argued above that Libanius' temple is more likely to have been in Carrhae than in Edessa; and it seems implausible that Theodosius should order the governor of Osrhoene to protect a temple in 382, only to see it destroyed by another governor a few years later. Moreover, though Edessa had a very mixed population of Christians, Jews, Manichaeans and pagans, in 363 it had become such a stronghold of paganism that Julian preferred to bypass it. In 382 it can hardly have had a pagan temple serving as a meeting place for the entire citizenry.Footnote 18
III. EMESA
As well as deploring the recent destruction of a temple in Osrhoene, Libanius in two speeches close in date to the speech For the Temples mentions a city that had escaped punishment from Constantius II after overturning one of his statues. The first of these is On the Riot (Or. 19). As the text is usually printed, Libanius says: ‘The city was Edessa, (and) the citizens, resenting something done to them, took down a bronze image of him [Constantius], then hoisting it up began to beat it face-down with a whip, as is the rule with boys in school’. Gothofredus took the reading Ἔδεσσα from a manuscript in Paris, and was followed by Förster and after him by Norman; but it is now known that all the other witnesses read Ἔμεσα.Footnote 19 Moreover, Libanius mentions the same incident in the speech After the Reconciliation (Or. 20): ‘With whom then can one compare you [Theodosius] except with one who also suffered insult through his image on the borders of Syria (ἐν τοῖς ὁμόροις τῇ Συρίᾳ)? Yet if you consider Emesa, its festivals, and the rules pertaining to the festivals … you will find that there is a great difference in the way both (cities) treated their images.’ This time it was Richard Förster who read Ἔδεσσαν in order to bring this passage into conformity with the earlier one.Footnote 20 Here Edessa has the attraction that it was never part of the province of Syria, whereas Emesa had been. Advantageously situated in the upper reaches of the Orontes and on an important trade route, Emesa was a vassal state ruled by an Arabian dynasty until being annexed to Syria under the Flavians. None the less, when Septimius Severus divided Syria into two parts, Emesa was included in the southern part, Syria Phoenice. The novelist Heliodorus, writing in the third or fourth century, calls himself a ‘Phoenician of Emesa’ (10.41.4). By Libanius' day Emesa was no longer in Syria at all but in Phoenicia Libanensis.Footnote 21 The reading in both passages of Libanius should therefore be Ἔμεσα(ν), and it was Emesa, not the largely Christianized Edessa, that so insolently mistreated the statue of Constantius.
To summarize: (I) the temple mentioned by Libanius in Or. 30, 44–45 is that of the Moon (Sîn) at Carrhae, not in Edessa; (II) the temple in Osrhoene that Theodosius ordered to remain open was in Batnae of Osrhoene; (III) the city that overthrew the statue of Constantius II is Emesa, not Edessa.Footnote 22
