Dura-Europos: The Parchments and the Papyri Kilpatrick, George D Greek, Roman and Byzantine Studies; Fall 1964; 5, 3; ProQuest pg. 215

Dura-Europos: The Parchments 
and the Papyri 
George D. Kilpatrick 

D URA-EUROPOS is one of the most wonderful discoveries of the 
years between the two wars. Whatever interests us in the 
Roman Empire-Judaism, Christianity, the mixture of reli­

gions, of societies, of nations, commerce and industry, the Roman 
army and administration-all are illuminated by what the excavator 
has laid bare of this frontier city of the middle Euphrates. 

This is particularly true of the volume of the Final Reports contain­
ing the parchments and the papyri.! Among the many texts of great 
interest two small parchments are particularly noteworthy. One, a 
roll, consists of a Greek fragment of Tatian's Diatessaron, a harmony 
of the Four Gospels made about A.D. 170. The other consists of two 
Hebrew liturgical fragments fronl a sheet of parchment giving 
various benedictions. These two texts provide our starting point: one 
a Christian text in Greek, the other a Jewish text in Hebrew.2 To 
what kind of communities do they belong? 

Dura-Europos was a Seleucid foundation on the right bank of the 
Euphrates a little below the Khaber river. It became Parthian at the 
end of the second century B.C.; the Romans took it frOlTI the Parthians 
about A.D. 165 and held it for some ninety years before it was captured 
and destroyed by the Persians. 

It is these ninety years that concern us. During them Dura was a 
frontier fortress with a Roman garrison. It was no longer the important 
commercial centre that it had been under the Parthians, and for the 
last thirty years of its existence it was beset by the fear of Persian 
attack. The new Persian dynasty came to power about A.D. 224 with 
the support of strong national feeling. Against this threat the Roman 

1 C. Bradford Welles, Robert O. Fink, and J. Frank Gilliam, The Excavations at Dura­
Europos. Final Report V, Part J, The Parchments and Papyri (New Haven and London 1959). 
My study of the documents has been aided by two most instructive visits to the reconstruc­
tions of the Christian chapel, the Jewish synagogue and Mithraeum at Yale University. 

2 Nos. 10 and 11. 

215 



216 DURA-EUROPOS: THE PARCHMENTS AND THE PAPYRI 

garrison was increased, but this precaution did not prevent the city's 
destruction. In A.D. 256 it fell to the Persians, who left it in ruins. 

When these ruins were excavated between the two world Wars 
there came to light the remains of a Jewish synagogue and of a Christ­
ian chapel or house-church. It is with the Jewish synagogue that we 
first concern ourselves. 

Soon after the capture of the city by the Romans, perhaps about 
A.D. 170, the first synagogue came into being. It was a modest building, 
a converted house, and at the most could hold some sixty persons. 
About A.D. 245 it was replaced by a large and more imposing synagogue 
which suggested that the Jewish community was increasing and 
becoming more prosperous. About A.D. 250 it was given a series of 
wall decorations, most of which have survived. 

Let us now return to our Hebrew fragments. It is significant that 
this liturgical text is in Hebrew and not in Greek. There is no 
Latin text from the synagogue, and there are no other Jewish texts 
among the papyri. Apart from a few literary papyri we have mainly 
documents and military records which are most illuminating about 
the Roman army of this period. There are also a few private pieces. 

The range of languages is relevant to our inquiry. First, except our 
Hebrew liturgical text, all literary pieces are in Greek. Next, the 
official documents and military records are in Latin or Greek, or 
sometimes in both. Thirdly, the remaining private texts are in Greek, 
Latin, Aramaic, Parthian, Middle Persian, and one Syriac document, 
a deed of sale of A.D. 243. 

The linguistic picture of the area may help us to interpret this 
information. Before Alexander the lingua franca of the Middle East 
was Aramaic. The official language of Alexander and his successors 
was Greek. This was also the recognized language of their foundations. 
So we would expect Greek and Aramaic to be in use at Dura before 
the Parthian conquest. During the Parthian period we would expect 
Aramaic and Greek to remain in use with but little trace of Parthian. 
Under the Romans Latin would be the language of the army, Greek 
the language of the civil administration, and for the rest the same 
variety should be found as before. 

Let us look at the Roman period more closely. Our main sources of 
information are two, the inscriptions and the fragments on papyrus 
and parchment. The parchments and the papyri have been edited 
separately in the Final Report V, Part 1. The inscriptions are to be 



GEORGE D. KILPATRICK 217 

found in the ten fascicles of the Preliminary Reports CPR), in the Final 
Report VIII, Part I for the Jewish Synagogue, and in the Yale Classical 
Studies XIV.3 

In all there are about twelve hundred inscriptions from Dura, but 
not all of them belong to the Roman period. Approximate numbers 
of texts in the various languages dated to this period are as follows: 

Greek 439 
Latin 
Pahlavi and Middle Persian 
Safaitic 
Palmyrene 
Syriac 
Aramaic 

68 
14 
8 
7 

3 

1 

540 

While the figures for the other languages may be nearly exact (for 
example all the Latin inscriptions with one exception belong to this 
period), the number of Greek inscriptions would probably be con­
siderably increased if we knew the date of all the undated inscriptions. 
We may put together the number of the inscriptions in Syriac, 
Aramaic and Palmyrene as representing various forms of Aramaic. 
The Safaitic inscriptions point to Arab connexions. The Pahlavi or 
Middle Persian inscriptions remind us of the proximity of the 
Parthians and Persians. 

Aramaic inscriptions and numerous proper names sho\\' us that, 
though Aramaic was no longer the lingua franca of the area, it was 
submerged by Greek rather than eliminated by it. This opinion is 
supported by the fact that in two Greek inscriptions the participle is 
used where Greek would normally have a finite verb.4 From this 
example we see how Aramaic influence can show itself even in Greek. 

The Greek inscriptions predominate in this period even in the 
temples of gods who were Semitic in origin. Thus all the inscriptions 
in the temple of Atargatis,5 in those of Artemis-Nanaia, of Adonis, of 

3 Pp. 127-213 (1955): R. N. Frye, J. F. Gilliam, H. Ingholt, C. B. Welles, "Inscriptions from 
Dura-Europos." It is expected that when the volume of the Final Report containing the 
inscriptions is issued, some changes will have to be made in the figures now given, but it is 
not likely that they will be large enough to upset the present conclusions. 

, See PR VII-VIII, 129-131, nos. 867, 868. 
5 yeS XIV, 131--137, inscription 3, is in Palmyrene, but is probably of the first century 

A.D. The temple of Zeus Megistos has one Palmyrene inscription of the second half of the 
second century A.D., no. 5, two Iranian inscriptions, nos. 9 and 10, and five Greek, nos. 
6,7, 8, 11, 12. 



218 DURA-EUROPOS: THE PARCHMENTS AND THE PAPYRI 

Aphlad, of the Palmyrene Gods, of Azzanathkona, are in Greek, with 
one or two in Latin. The Mithraeum has one bilingual inscription in 
Palmyrene and Greek; the others are in Greek or Latin. A bas-relief 
of Herakles has a text in Greek and Palmyrene. A silver vase, found 
in a private house, with a Syriac inscription, may be a cult object. A 
Palmyrene inscription was found in Wall Street under a bas-relief of 
two figures which may be gods. A room in another private house has 
a Palmyrene text referring to four gods with Semitic names. 

The papyri support this evidence. From the Roman period there 
are 147 Latin texts, mainly military, 52 in Greek, two in Middle 
Persian, and one each in Parthian, Aramaic, and Syriac. The army 
administration is Latin, civil administration and private texts are 
almost entirely in Greek, but two or three texts bear witness to the 
proximity of Parthia and Persia, and the Aramaic texts remind us of 
the Aramaic substratum. 

When we turn to the synagogue all is different. There is one Hebrew 
text on parchment. Besides this we have only inscriptions. There are 
nineteen Greek, twenty-two Aramaic and fifteen Middle Persian and 
Pahlavi. It is not surprising that there is no Latin, but it is remarkable 
that the largest number of texts is in Aramaic. Striking too are the 
Middle Persian and Parthian inscriptions. 

How are we to explain this? The presence of Greek texts needs no 
explaining but we want to know why proportionately they are so few. 
The simplest and most probable answer is that the Greek element in 
the congregation of the synagogue was so much the smaller and 
weaker. 

Why is the number of Aramaic texts larger? It is not enough to say 
that the Aramaic element was stronger than elsewhere. As we have 
seen, under the predominant use of Greek in Dura as a whole there 
are traces of Aramaic in occasional texts and in many proper names. 
We may recall that over the frontier in Parthian or Persian Mes­
opotamia were large Aramaic-speaking Jewish communities, and it is 
reasonable to suppose that it was the connexions with these com­
munities in lower Mesopotamia which kept Aramaic so much in use. 

With this we may connect the Hebrew liturgical fragment. Jewish 
liturgical texts in Greek of the fourth and fifth centuries A.D. from 
Egypt show that Greek was still a liturgical language, and possibly 
the liturgical language, for Jews in Egypt in this period. The third­
century Hebrew fragment from Dura is in contrast with this. It is 



GEORGE D. KILPATRICK 219 

dangerous to rely too much on so little evidence, but as far as it goes 
the fragment supports our conclusion. Hebrew was used liturgically 
by the Jewish communities of Persian Mesopotamia, and our Hebrew 
fragment shows the same practice present in Dura. 

We may now consider the Middle Persian and Pahlavi inscriptions. 
They too suggest a connexion with Parthia and Persia. What this 
connexion is we find it hard to determine, but the context of some of 
the Persian inscriptions is important. The synagogue contains inter­
esting and instructive paintings; on several of them are Persian texts. 
The interpretation of these texts is disputed, but the following 
translation is given from the Final Report VIII, Part 1, of one of the 
longest inscriptions (no. 43): 

In the month ML8r, in the year 14, and on the day 
Sa8revar when Yazdanpese, the scribe, and the scribe 
of the radak (building?) to this house came, [and by 
them] this picture [was looked at and] by them praise 
was made. 

There are ten similar inscriptions which relate the visits of Persian 
scribes to the various paintings. Among them no. 44 is particularly 
important. It reads: 

In the month MLOr in [the year] fourteen, and on the 
day [Pravar]tfn when Hormazd, the scribe, and Kantak, 
the zandak, and the scribe of the building, and this zandak 
of the Jews to his edifice of the God [of] the Gods of the 
Jews came and by them [ .. J this picture was beheld, and 
by them it was looked at and beheld [ ....... ] it was 
looked at [ ......... ] the picture [ ........ ]. 

What is the meaning of Zandak? Professor Lieberman of New York 
has related it to a Hebrew term in the Aggadath Esther zandfqfn 
which, he suggests, means «dignitaries, selected men, officials." 
They are in effect the representatives of the Jewish community before 
Pharaoh. I suggest that both the Hebrew term and the Middle 
Persian one are in fact the Greek word aVvoLKo~.6 There is no difficulty 
about the transliteration in Hebrew. One common meaning of 
aVV8LKOS in Greek is «public advocate, appointed to represent the 

6 Examples of Hebrew zayin representing Greek a are to be found in S. Krauss, Griechische 
und lateinische Lohnwiirter im Talmud, Midrasch und Targum II (1899) 239-248. For a instead of 
Greek v ,ve may compare sanhedrin for avv€op,ov. 



220 DURA-EUROPOS: THE PARCHMENTS AND THE PAPYRI 

interests of the state." There is an example of this use in an inscription 
from Palmyra of the second century A.D. 7 The word is also used at 
Athens of advocates chosen by the tribes to defend their interests. 
The word seems to describe a representative of the state or of a 
community. As we have seen this seems to be the meaning of the 
word in the Jewish text just mentioned, and it will suit the context 
in the Middle Persian inscription. "This zandak of the Jews" will be 
the representative of the Jewish community in Dura. Who then will 
Kantak the zandak be? Presumably the Persian representative, com­
parable perhaps to the JfJvapXT}r; of King Aretas at Damascus men­
tioned by St. PauLs Kantak, the Zandak, will accordingly be something 
like a consular representative nowadays. 

In the light of this, some of the Middle Persian inscriptions in the 
synagogue seem to be the approval given by the Persian repre­
sentatives, presumably to some of the paintings there. This 
explanation suggests that there was a connexion between the Jewish 
community and Persia before the capture and destruction of the city. 
It makes unnecessary the suggestion that a year or two before its fall 
the city was temporarily occupied by the Persians and then reoc­
cupied by the Romans, a suggestion for which there is no other 
evidence. 

There is one other piece of evidence for Persian influence in Dura 
before its fall. In the House of the Frescoes (House F, Block C7) in the 
main living room is a battle scene in which the Persians are victorious 
over the Romans or the Palmyrenes or both.9 This curious picture in 
a Roman frontier fortress could be explained if this were the house of 
the Persian representative in Dura. It has a number of Middle 
Persian or Pahlavi inscriptions. 

From all this evidence we can construct a clear picture of the Jewish 
community of Dura. It was one of the more obviously oriental 
sections of the population of Dura. We can associate with this char­
acteristic connexions both with the large Jewish community of lower 
Mesopotamia and with Persia. 

In considerable contrast to this is the Christian chapel. For our 
know ledge of Christianity in Dura we are dependent on a Greek parch­
ment fragment, a few brief inscriptions and the remains of the chapel. 

7 LS] 1703a s. v. 
82 Cor. 11:32. 
9 PR IV, 182-199. 



GEORGE D. KILPATRICK 221 

The parchment fragment has been published more than once. For the 
chapel and the inscriptions we have only the Preliminary Report V. 

Examination of the reconstruction of the chapel at Yale University 
encourages me to dispute one point in the reconstruction. It is assumed 
that the chapel contained a font, and it has actually been suggested 
that the chapel was a baptistery.1o I much doubt this and I am glad 
to find that others do also. A font was quite unnecessary. As wit­
nessed by the Didache, a second century text from Syria, it was the 
ancient custom to baptize in living water. This was available outside 
Dura in abundance in the Euphrates. 

Now let us turn to an attempt to determine the character of the 
church at Dura. First, all our texts are in Greek, both the parchment 
fragment and the five inscriptions. There is an apparent exception to 
this, a Syriac inscription which has been translated as follows: 

Khall~a son of Sennaq, of Qarha 0), disciple of Rama. 
May he be remembered before God.n 

It is held that as the god is not named, the phrase "before God" 
suggests that the man is a monotheist and that the use of the term 
"disciple" implies that he is a Christian. But in the three inscriptions 
of Epinicus and Alexander, the god is not named though he is 
probably one of the local divinities.12 If the man were a Christian, we 
would expect him to be described as a disciple of Jesus, not of Rama. 
Further the inscription is on an altar, and it is doubtful whether a 
Christian altar would be found outside the Christian chapel at this 
period. We may note that Sennaq is a Parthian name. As a whole the 
suggestion that the inscription is Christian seems unconvincing. On 
the evidence the Christian community is Greek-speaking. 

This indication is supported by that of costume. In the Mithraeum 
and the synagogue some of the costumes are Mediterranean and 
others oriental. When I observed that the Philistines in the synagogue 
paintings were dressed like Roman soldiers, I tried to apply to these 
paintings a principle according to which the sympathetic characters 
were portrayed in oriental dress and the unsympathetic ones in 
Mediterranean, but this distinction was not consistently observed. 

In general the fact that both kinds of costume are to be found in 
10 Reallexikon fur Antike und Cliristentum IV, 365; Religion in Geschichte und Gegenwart 3 II, 

291. 
11 PR III, 68-71. 
12 PR VII-VIII, 128-134 (nos. 867-869). 



222 DURA-EUROPOS: THE PARCHMENTS AND THE PAPYRI 

the synagogue and the Mithraeum need not surprise us. The syn­
agogue, as we have seen, had strong oriental as well as Greek con­
nexions. The cult of Mithras is of oriental origin but had much 
success in the West, particularly in the Roman army. 

When we examine the Christian chapel, we find a different practice. 
So far as they can be discerned, the figures in the Christian paintings, 
and there are a number of them, all wore Mediterranean costume. 
Some are very obscure from the damage they have suffered, but there 
is no clear example of oriental costume among them. 

These pieces of evidence suggest that Christianity came to Dura 
from the Greek West. This agrees with what we know of Christianity 
in Syria at this time. About A.D. 200 Serapion, the active bishop of 
Antioch, consecrated Palut the first historical bishop of Edessa.13 The 
origins of the Christian community at Dura seem to belong to the 
same period. Syriac Christianity at Arbela, for example, appears to 
be independent and older in origin. 

One group of facts may seem not to agree with the suggested origin 
of Christianity at Dura. The evidence of literary texts discovered in 
Egypt is that before the fourth century, Jewish and pagan texts were 
almost without exception written on rolls, while Christian texts 
equally regularly were written in books, codices. At Dura eight of the 
nine pagan literary texts are on rolls, but the ninth, apparently a 
fragment of Appian, is written on a codex. On the other hand the 
Christian fragment of the Diatessaron is written on a roll. 

How is this to be explained? We do not know, but two points may 
be made. First, we have only one Christian text from Dura, and it 
would be imprudent to generalise from this. Secondly, the practice 
of Egypt may not be true for other parts of the Roman world. We 
have a fragment of a Jewish liturgical roll in Greek from Egypt of the 
fourth century.!4 It may be that in Syria Christians followed longer 
the Jewish practice of using rolls rather than codices. In this case our 
Christian fragment would be evidence for a Jewish background to 
Christian practice but not for any origin in a Semitic language. 

An even more serious objection to the view that Christianity at 
Dura was derived from the Greek West is to be found in the text of 
the fragment of the Diatessaron. To explain this objection we must 

18 Lietzmann, Geschichte der alten Kirche II, 270-272. 
a POxy. 17.2068. I plan a new edition of this text with commentary to demonstrate my 

identification of it as a Jewish text. 



GEORGE D. KILPATRICK 223 

say a little more about the Diatessaron itself. It is agreed that the 
Diatessaron was a harmony of the Four Gospels compiled by Tatian 
about A.D. 170. It is disputed whether he compiled this harmony in 
Greek or Syriac. At first our fragment was regarded as strong evidence 
for the view that the Diatessaron was originally composed in Greek. 

Soon after its publication scholars suggested that it was part of a 
retroversion into Greek from Syriac.15 They infer that first there was 
a Syriac text and that our Greek fragment comes from a Greek trans­
lation of this Syriac. They support their suggestion by arguing that 
several features in the Greek are due either to misinterpretations of 
the Syriac or to the misreading of Syriac characters. 

Before we consider these features let us look at two points. First, 
the wording of the fragment is, with little exception, that of the 
Greek Gospels. To achieve this effect the translator from Syriac had 
either to have a Greek manuscript of the Gospels beside him when 
he made his translation from Syriac, or else to have the wording of 
the Greek so fixed in his mind that he could reproduce it at will. In 
either case we may wonder why he troubled himself to translate from 
the Syriac at all. In effect the Greek wording of our fragment as a 
whole is in favour of the view that we are dealing with a text that is 
derived directly from the Greek Gospels without any Syriac inter­
mediary. 

Secondly, as we have seen, all other evidence about the Christian 
community at Dura points to a Greek background. To counteract 
this evidence we need concrete instances of a contrary nature; so far 
these have not been produced. These two points put the burden of 
proof on those who argue for a Syriac text behind our Greek fragment, 
and suggest that we give the benefit of the doubt, where doubt exists, 
to the Greek evidence. 

Let us now look at the principal examples where we are expected 
to find traces of a Syriac original behind the Greek. First, Luke 23: 49, 

describing the \vomen who followed Jesus from Galilee, has the 
phrase aL yvvaLKES' aL uvvaKoAovOovuat avr0 "the women who 
together followed him." The equivalent in our fragment is aL 

YVVULK€S' [T£VV UV]VUKOAov81}uavT£uv u[v]rcfJ "the wives of the men who 
together followed him." It is suggested that this depends on a mis­
interpretation of the Syriac. But such a hypotheSiS is quite unnecessary: 
all we need imagine is a simple desire to preserve the proprieties. A 

15 See the reference in Final Report V, Part 1. 



224 DURA-EUROPOS: THE PARCHMENTS AND THE PAPYRI 

picture of Jesus followed by unaccompanied women could easily give 
rise to calumny; introduce their husbands and all is well. 

Secondly, Joseph of Arimathea occurs in all the Gospels, but our 
fragment gives an unparalleled form of the name of his village, not 
APEIMABAIA but EPINMABAIA. The appearance of the N is held 
to be due to the lengthening of a yod into a nun in Syriac script. In 
principle this is possible, but no example of such a mistake with 
Arimathaea is produced from Syriac manuscripts. Let us overlook 
this for the moment. We note that this suggestion does not explain 
the E at the beginning of the name in our fragment. 

An alternative explanation is possible within an entirely Greek 
tradition. Imagine that a scribe writes APEMMABAIA with double 
M for 1M by mistake. This is resolved to APEINMABAIA. Con­
currently APE at the beginning becomes EP. These two changes 
give us the form of the name as it occurs in our fragment without 
any need of a Syriac hypothesis. 

Thirdly Luke 23 :49 has opwactL'TCt.v'TCt. "seeing these things," des­
cribing the women who were witnesses of the crucifixion. This 
becomes in our papyrus opwaCt.t 'TOY CTA (=a'TCt.vpwf)Ey'TCt.) "seeing the 
crucified." It is suggested that the Syriac had translated 'TOY a'TCt.vpoy 
"the cross," and that this was wrongly translated back into Greek as 
'TOY a'TCt.vpw8ly'TCt.. It is pointed out that the words for cross and crucified 
are indistinguishable in Syriac. We may remark that 'TOY CT A and 
'TOY CTN are easily confused in Greek and that this reference to the 
Syriac does nothing to explain the change of 'TCt.V'TCt. to 'TOY a'TCt.vpoy or 
'TOY a'TCt.vpw8ly'TCt. in the first place. 

We may conclude from this that the suggestion that our Greek 
fragment is the product of retranslation into Greek from the Syriac 
is unconvincing. If we accept this conclusion there remains no evidence 
to controvert the view that Christianity came to Dura from the Greek 
West. This view we may regard as holding the field until much 
stronger evidence is produced on the other side. We must, however, 
remember that the volume of the Final Report dealing with the Christ­
ian chapel has not yet appeared. 

We may notice one further point. The Decian persecution broke 
out in A.D. 251, four or five years before the fall of Dura. We have 
from the Fayyum and elsewhere in Egypt a number of certificates of 
sacrifice issued in connexion with this persecution. The Christian 



GEORGE D. KILPATRICK 225 

chapel was found in what seemed to be a kind of church house. It is 
unlikely that this Christian centre was unknown to the Roman 
authorities. Yet \ve find no trace of persecution in the remains.16 In 
view of the great dangers this frontier fortress faced, did the Roman 
authorities decide not to add to their difficulties by alienating the 
Christian community by persecution? They may not have been sure 
of the loyalty of the Je,vish community and have preferred to let the 
edict of persecution remain a dead letter. 

Much has been left unsaid about the church and the synagogue 
of Dura, the papyri and the parchments. Nonetheless something, 
perhaps, of the varied character of the two communities comes out 
clearly, the Christian derived from the Greek West and the Jewish 
with its strong connexions with the Persian Orient.17 

THE QUEEN'S COLLEGE, OXFORD 

June, 1964 

16 In particular no libelli have been found. 
17 The substance of this paper ... vas given as lectures at the Centre d'Erudes Orientales in 

the University of Geneva on 2 and 6 March 1964 and the University of Basel on 5 March. 
To the editors and printers of the Yale Dura-Europos texts I wish to record my apprecia­
tion for the high quality of the volume: they have combined to publish the texts in a 
form which enables us to use them with the greatest ease and confidence.