THE WESTERN PART OF THE VIA EGNATIA
By N. G. L. HAMMOND
(Plates IX-X)
This article arises from a visit to Albania in September 1972, when I travelled up and
down the valley of the Shkumbi river, and from recent discoveries by Albanian archaeol-
ogists which have been reported in Albanian periodicals, Monumentet 1971, i, 43-59, and
ii, 25-35, and Studia Albanica 1972, i, 85-106. The new evidence has confirmed some views
and overthrown others which I put forward in A History of Macedonia i (1972), 19 f., based
as they inevitably were on the reports of Pouqueville and other travellers down to Praschniker
and Schober. I have now seen part of the scene for myself and have had the advantage of
conversations with the discoverers themselves, Hasan Ceka, his son Neritan Ceka and
Llazar Papajani, to whom I am most grateful. The new evidence is in Section A 1-3 of the
article, and some conclusions are given in Section B.
A. I. ORAKE TO MIRAKE
In this Section the argument evolves more coherently if we move from east to west.
Having climbed from the shore of Lake Ochrid to the head of the pass, we come almost
immediately to Orake which was the mansio Claudan6n of the Itinerarium Burdigalense
(hereafter It. Burd.).1 We now enter the marshy basin known as Fusha e Kododeshit; here
Pouqueville 2 followed a raised way, 12 ft wide, which was paved, and Heuzey and Daumet 3
saw a ruined bridge which corresponds in distance with the Pons Servili of the Peutinger
Table (hereafter Tab. Peut.). The descent through the plateau of Domousova to the valley
of the Shkumbi (the ancient Genusus) near Qukis is gradual and easy; here L. Vidman 4
noted a small Roman bridge and a piece of ancient road by Perrenje. At Qukes itself the
river flows in a narrow bed between rocky outcrops of limestone, which offer excellent
footings for a bridge. L. Vidman noted the remains of two Roman bridges here, which were
linked on the left bank of the river by an ancient road.5 At this point we may place the
mutatio In Tabernas (It. Burd.) which was evidently the Tres Tabernae of the Antonine
Itinerary (hereafter It. Ant.).6
Below this point the river valley changes its nature. The river cuts a tortuous course
through the high, steep and eroded slopes of a flysch formation. The slopes are very friable
and offer no footings for a bridge. In the stretch from Qukes to Cotaj, where the river makes
an almost right-angled turn, I noticed only one place where a limestone formation was
visible beside the river. Below Cotaj the terrain changes. At first the geological formation
is what Nowack 7 calls 'mainly Peridotite and Serpentine', and at Mirake it becomes
limestone. The bed is narrow still below Miraki, and then opens up into a wide flat area
where the river pursues a frequently changing course through a maze of limestone boulders.
Nowack showed two bridges over the river, one at Mirake itself, and the other downstream
where a tributary enters the river from the south. These two bridges were the first below
Qukes at that time. One was called the bridge of Mirake, the other the bridge of Hadji
Bektari. An Italian map of the late I930's shows two bridges by Mirake and also that at
Hadji Bektari.
As Pouqueville and other travellers realized and as the Albanian scholars have
confirmed, the Via Egnatia below Qukes climbed high above the left bank of the river and
ran along the rocky mountainside which towers above the steep slopes of the flysch forma-
tion. Here the going is much better under foot, but the slopes are steep, and very many
long zig-zags were needed for wheeled vehicles. Neritan Ceka and Llazar Papajani 8 have
traced the general line of the ancient road between Qukes and Babie, and at certain points
1The edition of 0. Cuntz, Itineraria Romana i 6 As published in 0. Cuntz, op. cit.
(Leipzig, 1929) is used in this article. 7 E.
Nowack, Geolog. Karte von Albanien :20o,ooo
2Voyage dans la Grece (Paris, I820) iii, 63 f. (Berlin, 1928), based on material collected in 1922-24.
'Mission archdologique de Macddoine (Paris, 1876) 8 ne luginen e Shkumbinit ne kohen
'Rruga
346.
4' antike,' Monumentet 197I, I, 43-59, with a summary
Voyage epigraphique a Elbasan ', Listy Filo- in French; and 'La route de la vallie du Shkumbin
logick lxxxv (x962), 62 with fig. 7. dans l'antiquite,' Studia Albanica 1972, i, 85-1o6.
' ibid. with figs. 4 and 5. The latter periodical is referred to as SA hereafter.
i86 N. G. L. HAMMOND
it: ANCHORAGE
II, XXIV ec:. M/ILI./
PASSUUM
532, 1740 etc.: HEIGHTS
I/ /ME7ERE?S
FIG. 6: MAP OF THE VIA EGNATIA
The map shows the western part of the Via Egnatia (from Dyrrachium to Claudianon); the coastal road (from
Aulon to Dyrrachium); the side roads from the south (Aulon-Genusus, Apollonia-Ad Novas-Clodiana,
Apollonia-Stefanaphana-Absos-Clodiana); and an earlier road (Apollonia-Semen-Belsh-Top9ias).
Scale: : 800,000
Drawn from details supplied by author. Copyright reserved
they have found pieces of pavee which they attribute to three different periods of construction
as follows:
I. What is in their opinion the earliest paved track is only i 20 m wide. This track
corresponds with Pouqueville's note of a chaussee by Qukis 'four feet wide and three-
quarters of a league long'. The section near Dardhe which is illustrated by Ceka and
Papajani is paved with limestone. It is suitable for cavalry or pack-horses but not for
vehicles.9
2. The next in time is a paved track 4 30 m wide. A piece between Dzhure and Dardhe
is illustrated.10 This corresponds with Pouqueville's paved way 12 ft wide in the Fusha e
Kododeshit, except that it was raised there because of the marshy ground, whereas our
piece was flat.
3. The next in time is not covered with a paving of flat slabs of limestone, but is built
with large stones and has a top dressing of gravel. It varies in width between 4 m and 5 m.
A stretch between Dzhure and Dardhe is illustrated.11
9 Monumentet, loc. cit. fig. and SA, loc. cit. fig. 4; 10 Monlumentetfig. 5, and SA fig 5.
4, " SA fig. 6.
Pouqueville, loc. cit.
THE WESTERN PART OF THE VIA EGNATIA I87
4. Last in time is a route 6*70 m wide. Where it is on level ground, it was made of
large stones and had a top dressing of gravel; but on sloping ground it was paved with
limestone slabs and had gravel also. This route avoided steep slopes by pursuing a zig-zag
course. At a later date some steps were added to the route, at the zig-zags especially. The
pieces which are illustrated are again between Dzhure and Dardhe. One photograph shows
a number of zig-zags which are attributed variously to the Roman period, the Byzantine
period and the Turkish period.l1
We need to bear in mind that merchandise travelled along this route as the only
practicable one between Orake and Babi6 from the sixth century B.C. at least down to
Turkish times. It was taken by pack-horses, except during the Roman and Byzantine
periods when the route was possible for wheeled vehicles. The forms of construction in
2, 3 and 4 are suitable for carriage by vehicles and belong presumably to the Roman and
Byzantine periods. Ceka and Papajani have come independently to the conclusion which
I reached in my book that this was the line of the Roman road and that the stations known
as In Candavia, mansio Grandavia and Ad Dianam are to be located between Qukes and
Babie.l3
The next stage is the descent from Babie to the bed of the Shkumbi (the ancient
Genusus) and the crossing from the left bank to the right bank at Genesis flumen (Tab. Peut.)
or mutatio Treiecto (It. Burd.), which are both VIIII m.p. from the next station to the
west, Scampis or Hiscampis. The problem is where the road crossed the river. When
Pouqueville followed the Turkish road, he descended ' by a steep and tortuous path ' to the
bridge of Hadji Bektari, which had three arches (one is still standing) and crossed part of the
wide flat bed of the river. The surviving arch and the flat bed are shown in P1. IX, i. It
seemed to me when I wrote my account in Macedonia that this path was not suitable for
the Roman road,14 and I therefore advanced the hypothesis that the Romans bridged the
river higher up by Cotaj; but now that I have visited the valley by Cotaj, I do not think
my hypothesis acceptable. However, the reason which led me to look for a bridge higher
upstream still holds. Moreover, Malchus 257, i (Corp. Script. Hist. Byz. i) described an
attack which was made upon the Goths who were travelling with wagons on this route
westwards. The vanguard of the Goths fled from somewhere near Babie, descended to the
river, destroyed the bridge over ' a deep gorge ' (papacyyti P3aOia), and so escaped onto
level ground (EiS TOTr68iov). The ' deep gorge ' cannot be at the bridge of the Hadji
Bektari, where the valley bed is wide and level.
Before we try to solve the problem, we should note the discoveries made by Ceka and
Papajani. In their report they say nothing of any sign of the Roman road between Babie
and Hadji Bektari. The bridge itself shows work of three stages which they have noted for
the first time. Where the bridge begins on the right bank the abutment was made of
masonry with a core of rubble and mortar of a special kind, ' mortier-chamotte,' and an inset
of bricks which measure 35 x 28 x 4 *5 cm. Onto this abutment a piece of medieval bridge
was built, and then finally the Turkish bridge ran on from there across the river in three
spans. Furthermore, some 500 m downstream from Hadji Bektari, Ceka and Papajani
found the remains of a bridge, called locally the bridge of Kegi. Seven piles survive. They
are built with a facing of rectangular stone masonry and with a core of rubble and ' mortier-
chamotte '; and there are lines of bricks in the facing and in the core. These are of the
same size as those at Hadji Bektari. Remains of the arches of masonry which spanned the
piles are visible.15 The river has changed its course since this bridge was in use and now
flows farther north in the wide expanse of boulders.
The type of brick which is used in the abutment of the Hadji Bektari bridge and in the
piles of the bridge of Ke9i is known also in the citadel at Elbasan where Gj. Karaiskaj has
12
Monumentet fig. 6, 7 and 8; SA figs. 7, 8 and 9. Babie near a high point ' 757 ' on the Italian map
I am less confident about the attribution of forms of I:5o,ooo, reproduced by the British GS in I944;
construction and types of zigzag than the authors, but mansio Grandavia at Spathar; In Candavia near
the broad distinction between packhorse tracks and Qukis (cf. Macedonia i, 28, map 4).
carriageable gradients can be made firmly; see my 14So too L. Vidman, loc. cit., describing the route
comments on the main road in antiquity through the as ' tout a fait impraticable ', even on foot. He
Megarid in BSA xlix (I954) I13 if. noticed the traces of an early road on the left bank
13 The distances given in Tab. Peut. and It. Burd. coming up to the bridge.
608 enabled me to place Ad Dianam southeast of 15 SA I972, i, 102, fig. Io.
I88 N. G. L. HAMMOND
dated its first use within the period after 378 and before 450.16 The sizes vary only slightly,
and this type of brick was no doubt used for several centuries afterwards. I am inclined to
think that we should regard the bridge of Ke9i as earlier than the bridge of Hadji Bektari,
which seems to have had a continuous history from Byzantine to Turkish times. But the
'
bridge of Ke9i itself is later than the bridge over the deep gorge' which Malchus mentions,
because a seven-piled bridge is designed only for a wide, shallow bed. It seems probable
then that we should look for the bridge of Malchus in the gorge of Mirake where Nowack
showed an actual bridge in his map of the situation as it was in i922-24. He shows a
separate path leading from that bridge to Babie and this path may be more or less where the
Roman road ran. On crossing the bridge in the gorge of Mirake to the right bank one soon
enters 'the plain ' which contains the wide bed of the river. As we shall see later (A. 2,
below), such a position for the Roman bridge is more appropriate to the distances which
are in the Itineraries.
A. 2. MIRAKE TO DURRES
The discovery which has thrown most light on the Via Egnatia was made in i968, when
a number of buildings of the early second century A.D., including Baths and a Nymphaeum,
were excavated at Bradashesh above the right bank of the river. No report of the excavations
has been published, but we were taken to see the site in September I972. The buildings
are shown in Plate IX, 2. They stand in a line on a terrace some way back from the river
and parallel to its course; they looked out on the Roman road which ran presumably just
below them. There is a remarkably strong head-spring or 'kephalovrysi' above the
buildings, and its water pours down the hillside into the Nymphaeum. The buildings
certainly formed a station on the Via Egnatia.17
In this sector of the Via Egnatia the site at Bradashesh can be identified only as either
the mansio Clodiana or the mutatio Ad Quintum which are mentioned in the Itineraries.
It so happens that the numbers of Roman miles in them are consistent for the stretch from
Clodiana to Scampis, that is from west to east; namely Clodiana XX Scampis (Tab. Peut.),
Clodiana XX Scampis (It. Ant. 3I8), Clodiana XXII Scampis (It. Ant. 329), mansio
Clodiana XV mutatio Ad Quintum VI mansio Hiscampis (It. Burd. 6o8).18 Thus there is
every reason to regard the numbers as correct and not corrupt for this stretch. If we identify
Bradashesh with Clodiana and apply the distances, we have to place Ad Quintum XV m.p.
= 22 km away at Miraki and Scampis VI m.p. = 9 km away at a point upstream from
Cotaj. However this is unacceptable, because the crossing of the river at Treiecto has then
to be too high up the river. It follows that Bradashesh is to be identified with the mutatio
Ad Quintum.
From Ad Quintum as a fixed point we can apply the number of Roman miles for this
stretch of road.19 Clodiana being XV m.p. == 22 km distant, was at Mafmutaga, some 4 km
east of Peqin. Scampis, being VI m.p. = 9 km distant, was at a point some 3 km east of
Elbasan at Teqin Madhe where one enters a small pass. Genesis flumen (Tab. Peut.) or
mutatio Treiecto (It. Burd. 6o8), being VIIII m.p.= 13 km from Scampis, was at Mirake.
Here the road crossed from the right bank to the left bank, and the ascent was made to
Ad Dianam between Babie and Dardhe. As we have seen (p. i85 above), the crossing of the
river is likely on other grounds to have been at Mirake.
Proceeding now westwards from Clodiana = Mafmutaga to Dyrrachium, we are in no
doubt about the route. It proceeded along and above the right bank of the river as far as
Rogozhine and then turned north below the foothills and followed the line of the coast to
16 MonumentetI97I, i, 6I f.; the Elbasan bricks 19 The Roman mile is calculated here at 1,482 m,
x x
measure 36 24 4 5 cm. C. Praschniker and and I have taken my measurements on Nowack's map
A. Schober, Archaologische Forschungen in Albanien i: 200,000. I have also made checks on the Italian map
und Montenegro (Vienna, 1919) 48 f. put the first 1: 50,000. One cannot assume that the kilometres of a
fortification of the site at any time between 300 and modern road correspond to the distance on a Roman
530. road, because the latter was much more direct. It
17 Monumentet 1971, i, 52, fig. 9, and SA 1972, i, is interesting that the distance on the road from
fig. II. Durres to Elbasan was 78 km on the Italian map of
18 Differences of one or two miles may be due to the 1930S, and 82 km on the modem interational
inclusive or exclusive reckoning, as I have argued in road map of the Istituto Geografico di Agostini.
Macedonia i, 2I.
THE WESTERN PART OF THE VIA EGNATIA I89
Durres. The actual distance is some 45 km = between XXX and XXXI m.p. At this
point the data in the Itineraries are chaotic: Tab. Peut. gives XV and XXVI m.p. (totalling
41), It. Ant. 329.6 XXIIII, It. Ant. 329.7 XXV, It. Ant. 318.1 XXXIII,20 and It. Burd. 608
a total of 57. As we know the correct figure to be XXX or XXXI m.p. we should emend the
XXVI of the Peutinger Table to XVI; 21 its total then becomes 3' m.p., and the point of
division between XV and XVI is to be put at Shtodher, where there was evidently a
'mutatio '. Its name was probably Genusus, as we shall see (A. 3, below).
A. 3. THE ROADS FROM APOLLONIA
Having completed the run from Oraki to Durres, we turn now to consider the roads
which ran into the Via Egnatia from the south. In Macedonia i, p. 26 and map 3, I suggested
that the discontinued line from Aulona with the figure XVI on the Peutinger Table repre-
sented the coastal road from Aulona to Apollonia, that is from the site at Cape Treporti to
the site at Poyan.22 The distance XVI m.p. = 23} km is correct for a road from the seaward
end of the ' channel ' which exists from the lagoon of Arte, via Poro to Poyan. Praschniker
saw traces of this road as a raised way (to avoid flood-water) by Feras and Frakulla, and a
Roman milestone has been found at Levani (CIL iii, 7365).23 The coastal road continues
on the Peutinger Table from Apollonia to Dyrrachium, crossing the rivers Semeni and
Shkumbi, which are recorded on the Table as Hapsum fl. and Genesis fl. Taking the
distances on the Table to be correct, I suggested that this road went via Fier, where
Praschniker noted footings of an ancient bridge over the Gjanice, onto the western edge of a
low range of hills which runs northwards, parallel to the coast, as far as Barderoll.24
The figures XVIII m.p. = 26? km and XX m.p. = 291 km put the crossings of the Apsus
(Semeni) by Guri just north of Babunje and the crossing of the Genusus (Shkumbi) at the
latitude of Shtodher, where we have argued (p. i88 above), that there was a mutatio on the
Via Egnatia XV m.p. from Dyrrachium. The name of this mutatio is preserved in the lists
for the coastal road in the Ravenna Geographer 379.2 and Guido 540.3 as' Genesis '. Thus
the mutatio was named after the river, as the modern village Semen is named after the River
Semeni.
It is on the line which I had suggested for this coastal road that three Roman milestones
have been found. The report of them has been published by Hasan Ceka in Monumentet
I971, ii, 25-35 (with a summary in French). Two were discoveredI962 in in a district of
Brostar called ' the bridge '. It is here that Nowack showed the crossing of the Semeni river
by ferry-boat; a bridge has probably been built since then. The third was discovered in
1933, but as far as I am aware it was not published until I971; it came from Sulzotaj, a small
village near Nov6, where Nowack showed a crossing of the Shkumbi by a bridge; this was
the only bridge over the Shkumbi below Elbasan in I922-24. The inscriptions on the three
milestones are as follows. Two of them are shown in Plate X.
A. From Brostar;P1. X, a, b.
Imp(erator) Caes(ar) / M(arcus) Aurelius Antoninus / Pius Felix Aug(ustus) /
sacerdos amplissimus / dei Solis invicti Elagabali / Pontif(ex) Max(imus) /
trib(unicia) pot(estate) III / co(n)s(ul) III / P(ater) P(atriae) / restituit.
'
This inscription records the restoration ' of the road to good condition by the
Emperor Elagabalus in the year 220.
Two later inscriptions were added to this milestone;P1. X, c.
(i) Domino no(stro) Galerio Valerio Maximino nobilissimo Caesari
This inscription was made between 305 and 308 or at the latest 310.
20 With variant readings of XXIII and XLIII. 23C. Praschniker, loc. cit. 6o. A piece of paved
21C. Praschniker, 'Muzakhia und Malakastra ', road 6 m wide has been reported just outside the walls
JOAI xxi-ii (1922-24) Beiblatt, I26 came to the of Apollonia (Bul. Univ. Shtet. Tirane s 196o, i, 95).
same conclusion: ' ausserdem ist die Zahl XXVI auf 24 The modem track keeps to the western side of
dem nachI Clodiana
fuihrenden Strich in XVI zu the range of hills and that is why I have put the
iindem.' ancient road there; but the eastern side is also
22 For the site of
Aulon, see N. G. L. Hammond, possible and there are two places on this side called
Epirus (Oxford, 967) I32 f. and 689. Gradishte, which implies some ancient remains.
I90 N. G. L. HAMMOND
(ii) D(omino) n(ostro) Fl(avio) Val(erio) Constantino nob(ilissimo) Caes(ari)
This inscription was made between 306 and 3IO. One may compare these inscriptions
with that on a milestone of 305-6, found on the shore of Lake Ostrovo in upper Macedonia
on the line of the Via Egnatia (BCH xvii (1893) 635).
B. From Brostar. Hasan Ceka gave the inscription as follows; P1. X, d.
D(omino) n(ostro), Iuliano a GBRN III m(ilia) p(assuum)
This inscription is to be dated within the period 36I-3. As Brostar is not very far from
Apollonia, it is most unlikely that there was a station between Brostar and Apollonia.
Accordingly we may assume that if a place was signified by the letters GBRN, it was III
m.p. = 4- km to the north of Brostar; it was presumably a mutatio and should be placed
near Nicas on the seaward side of the range of low hills. Hasan Ceka suggested that the
letters GBRN stood for G(a)br(aio)n, such a site being named as a castle in Epirus, restored
by Justinian (Procop., de aedif. iv, 4, p. 75). However, I realized that if the vowels were
omitted, one would expect 'a Gabraio' and so only 'a GBR', or one could emend the
Gabraion of Procopius to Gabranon.
At this point I consulted Dr. A. R. Birley. He suggested that the inscription should
be read as follows:
D(omino) n(ostro) Iuliano A(u)g(usto) B(ono) R(eipublicae) N(ato) III MP.
He pointed out that the phrase Bono Reipublicae Nato is found on milestones of Julian
and that, although the usual abbreviation is BRPN, this abbreviation BRN is found on
CIL v, 7988, a milestone on the Tergeste-Pola road. When I looked at the photograph of the
stone in Monumentet I971, ii, z6, d, I noticed that a small v is visible between the capital
letters A and G. It is clear that Dr. Birley is correct. There is thus no indication of the
name of the mansio from which the three Roman miles is measured.
C. From Sulzotaj
D(ominis) n(ostris) Crispo et Constantino Caes(aribus).
The inscription was made in the period 317-326.
The road from Aulon via Apollonia to Dyrrachium was a continuation of the coastal
road which came up the coast of Greece. It was not in itself a part of the Via Egnatia,
although it gave access to it. The Peutinger Table and the milestones show that the road
was maintained from the late second century to the late fourth century. When we apply the
distances on the Peutinger Table, we see that the Apsus (Semeni) flowed farther north
C. 21I-I7 than it does to-day and was using a channel which is marked on modern maps as
'the old channel '. This channel may have been used during more than one period of
time; 25 for Nowack shows water in the bed and a bridge over it at Babunj6. Similarly the
Genusus (Shkumbi) flowed northwards between Barderoll and Shtodher and not as now
past Sulzotaj. The same was true of the Aous (Vijose) and the Drilon (Drin). The former
used to flow past Apollonia, making the city accessible by water, but it changed its course
at some time after the beginning of the Empire and before c. 286 when the Antonine
Itinerary 329,1 gives Aulon as the terminal for the Via Egnatia.26 The Drilon used to flow
close to or into Lake Skutari and not beside the ancient city of Lissus; it too was navigable.27
Similar changes in the courses of rivers occurred during antiquity in the coastal plain of
Macedonia.28
The explanation of such changes in the course of a river is well known. When the river
is in an exceptional spate, it deposits such masses of rubble and silt that its channel becomes
blocked and it cuts a new channel at a point of less resistance.29 This comes about only
when the river is left to its own devices and not contained in an artificial channel, as it is
to-day. When the Apsus and the Genusus were uncontrolled, the whole plain of the Myzeqije
25 C. Praschniker, loc. cit. I5, gives an interesting was navigable according to Strabo, C 3 6, and Anna
account of changes in the course of the Semeni Comnena, Alex. I2, 9.
between I804 and I922. 28 See Macedonia i, 142. f.
26 See C. Praschniker, loc. cit.
55; there has been 29 Anna Comnena, Alex. i, 7, 3, gave this
a change again between I868 and 1922. explanation.
27 See
J. M. F. May in JRS xxxvi (I946) 54 f.; it
THE WESTERN PART OF THE VIA EGNATIA I9I
was liable to floods. Indeed, when I travelled along the inland side of the plain in April 1932,
I saw floodwater stretching as far as the eye could see. On the other hand, in September
1972, we were shown a photograph of such extensive flooding in the Myzeqij6 as a sign of
the primitive conditions which the present regime has abolished. The reason for the course
of the Roman road near the coast is now clear. It had to keep as much as possible to the
higher ground offered by the range of low hills, in order to avoid the floods. Moreover the
crossing of the rivers where they entered the plain at Ku9 and Thanaj was done by ferry, for
instance by Julius Caesar in 48 B.C. (Bell. Civ. iii, 75 and 77) and by travellers in I922-24,
as we see from Nowack's map.30 On the other hand on the line of the Roman road it may
have been possible to build bridges near Babunje over the Apsus and between Sulzotaj and
Shtodher over the Genusus.
These conditions in the Myzeqije help us to understand the lines taken by three other
roads which joined the Via Egnatia from Apollonia. I list them individually.
(I) From Apollonia to near Elbasan. This road is known mainly from its remains.
Praschniker saw the traces of a raised way (to avoid flood-water) on the inland side of the
plain near Verbas, at Novosel; and near Kurian. The road went on to Ku9 where the
Semeni enters the plain. Here Julius Caesar ferried his army across the Apsus. At Ku9
Praschniker saw the piers of an early bridge, not necessarily an ancient one.31 The road
went probably through a district of low hills to Belsh, where there are remains of the Greek
and Roman periods at a site called Gradishte and also at a small lake fed by hot springs.32
Some 2 km to the southwest of Muroqan Praschniker saw the remains of a one-arched
bridge, which he thought was of the Roman period, and a raised road extending for several
hundred metres from either side of it. The piece of road to the north-east ran towards the
remains of an ancient bridge over the Shkumbi near Shenjan at a point called Top9ias,
which Praschniker has described.33 I have argued elsewhere that the sixteen piers of this
bridge, some 300 m long, were made in the Hellenistic period.34 On crossing the Shkumbi
this road joined the route taken by the Via Egnatia in the Shkumbi valley a little to the west
of Elbasan.
This road served the interests of trade particularly well. For it ran close to the ancient
cities at Margellec 35and Krotine (Dimallum) 36 and Gradishte by Belsh, and on a hill near
Shenjan where a ramp leads up to what I take to have been the ' oppidum Parthinorum ' of
Caesar Bell. Civ. iii, 41, I.37 There is no doubt that the road existed in one form or another
long before the Romans appeared on the scene, because it was vital to the trade of Apollonia
and its satellites and neighbours.
(2) Apollonia to Clodiana via Stefanaphana. This road is recorded by only one
itinerary, It. Burd. 6o8, 2-8. As I have argued in Macedonia i, 23 f., the entry mansio
Marusio should be deleted as a corruption of the text, because if it stands, we are given
three ' mansiones' in a row, which is quite unparalleled, and the name Marusio occurs
nowhere else.38 The distances fit the first part of the road we have just described: civitas
Apollonia XVIII mutatio Stefanaphana XII mansio Absos bringing us in XVIII m.p.
= 261 km to a point between Kurian and Dronofrose and in XII m.p. = I8 km to Semen
village on the right bank of the river Semeni. It seems then that the mansio Absos took its
name from the river Apsus, just as the modern village Semen takes its name from the
river Semeni. The next entry, if we delete mansio Marusio, is XIIII Clodiana, which
distance being 21 km brings us via Rrasse and Pekishti to Mafmutaga = Clodiana.
30 Also in April Pfeiler Ansiitze von einer Gewolbekonstruction '.
I932 when I travelled on this route.
The Austrian army in the First World War built a There were some later repairs, and these may be the
bridge 900 m long over the Shkumbi at Thanaj, but reason why N. Ceka and L. Papajani in SA 1972, i, 98
the river soon destroyed it. think the bridge was of the Roman period.
31C. Praschniker, loc. cit. I 8 f., with figs 45 and 35
Praschniker and Schober, op. cit. 75 f., and
46. L. Derveni, ' Gjurme te reja ne kalane e Margellicit,'
32 YOAI xxi-ii (I922-24), Beiblatt 218 f., with fig. Monumentet 1971, ii, 147 f.
128; SA 1972, i, 89. I visited the lake in 1932. 36 JRS lviii (I968) I2 f.
33 C. Praschniker and A. Schober, op. cit. 37 It was evidently the counterpart of Elbasan
(n. i6
above) 59 f., with figs. 69, 70 and 71. which is now the chief city of the Shkumbi valley;
34 N.
G. L. Hammond, Epirus 235. Praschniker see Epirus 737.
and Schober, op. cit. 60, noted that the superstructure 38Praschniker, Ioc. cit. I23, did not notice the
on the piers may have been made of wood, because anomaly and took the road the long way round by
there is no sign of any arch: ' es sind an keinem der Lushnj6 and Thanaj.
I92 N. G. L. HAMMOND
(3) Apollonia to Clodianavia Ad Novas. This road is recorded by only one itinerary,
It. Ant., 329, 5-6: Apollonia XXIIII Ad Novas XXV Clodianis. The most likely route
for this road, if we assume the figures of distance to be correct, is as follows. It used the
coastal road from Apollonia to near Babunjeand then branchedeastwardsacross the plain
via Sopez to Lushnje on the edge of the plain. If we measure the distance of XXIIII
m.p. = 35-1km we place Ad Novas at Shenepremte. In antiquity when the Apsus flowed
past Babunje, the road followed the slightly higher ground by Shenepremte to the north
of the river. Nowadays there is a track from Shenepremte to Lushnje; it was shown on
Nowack's map as existing in 1922-24. From Lushnje the road went along the edge of the
to
plain Thanaj, as the modern road does, and one then crossed the Shkumbiprobablyby
ferry to join the Via Egnatia and reach Mafmutaga = Clodiana. The distance XXV
m.p. = 37 km is correct for the journey from Shenepremte to Mafmutaga. It was on the
stretch Ad Novas to Clodiana that Praschnikernoted traces of a Roman road between
Lushnje and Thanaj.39
B. A SUMMARY OF CONCLUSIONS
It will be convenient to summarizeour conclusions chronologicallyand from west to
east. The first point to note is that the roads from Dyrrachium and Apollonia which ran
into and up the Shkumbi valley were of primaryimportance in the earlier period for the
Balkanarea only. Then they served the interests primarilyof Dyrrachiumand Apollonia,
the two Greek cities, which traded far into the interior especially in the Hellenistic period.
When Philip II of Macedon extended his authority to the coast of the Adriatic Sea, it is
possible that he and his successors on the Macedonian throne developed the road which
ran from Ochrid via Orake into the valley of the Shkumbi. However, when the Roman
conquest of the East was completed, this road became of wider importance. Dyrrachium
and Apollonia (or later Aulon) were now, as it were, the Channel ports for Italy and the
West, and the eastern end of the Via Egnatia was at the entry into Asia Minor. In the
same way Calais and Dieppe were only of local importance in early times but became
important ports for cross-Channel trade when maritime commerce developed. The
establishment of the Roman Empire thus imposed a new pattern upon the road-system
which had existed in Hellenistic (and earlier) times. The influence of Rome first became
effective perhaps in the period from 120 to ioo B.C.; 40 its full impact was felt after the
establishmentof the Principateof Augustus.
We owe our informationabout the Hellenistic road-systemto Polybius. When Strabo
reached Epidamnus (the Greek name of the city which the Romans called Dyrrachium)
and Apolloniain his descriptionof the Ionian Gulf, he describedwhat was to him (but not
to Polybius) the ' Via Egnatialeading to the east' (C 322 fin.) He then cited Polybius as his
authority for the measurementof the road by stades in the Hellenistic or Macedonian41
mannerand for the conversionof stades into Roman miles. Three further statementscome
from Polybius: that the road started from Apollonia and Epidamnus; that those who set
out from Apollonia and those who set out from Epidamnustravelledequal distances before
they joined ' the same road '; and that the first part of the road (i.e. ' the same road ') was
called ' the road of Candavia', an Illyrianmountain(C 323 init.; cf. C 327 init. and C 329,
fr. 56). Thus in the Hellenistic road-systemone must look for the junction of the Apollonia
and Epidamnus branches at a spot equidistant from each city and in a region where 'the
road of Candavia' begins. The latterrequirementis easily met. At Elbasanone sees for the
first time the towering ranges of Mt. Candavia; this was well expressed by J. G. von
Hahn 42 ' diese Gebirgsansichtbildet den westlichen Eingang zur alten Candavia, durch
welche die r6mische Via Egnatia fiuhrt.' The equidistance is fortunately provided by the
two roads which we have considered,namely that from Dyrrachiumto just west of Elbasan
and that from Apolloniato near Elbasan(p. I89 above, road (i)), the latter crossingthe river
Shkumbiat Top9ias near Shenjan. For the former,if it was less straightthan a Romanroad,
39ibid. 122. 'The Location of Cellae and the route of the Via
40 See Macedonia Egnatia in Western Macedonia ', CP xlvi (I951) II f.
i, 56, n. 2.
41 For a stadion-stone and road-markers in The stone was published in AM xviii (I893), 419.
Macedonian sections of the route see C. F. Edson, 42 Denkschriften (I867) 2, 50, n. 3.
JRS vol. LXIV (1974) PLATE IX
THE VIA EGNATIA: (I) THE SITE OF THE HADJI BEKTARI BRIDGE (MID-PICTURE) OVER THE SHKUMBI, LOOKING
UPSTREAM FROM THE ROAD ON THE RIGHT BANK OF THE RIVER (see p. I87). (2) THE STATION AT BRADASHESH, SHOWING
ENTRANCE TO NYMPHAEUM ON THE RIGHT (see p. I88)
Photographs (I) by R. A. Crossland, (2) by A. F. Harding. Copyright reserved
JRS vol. LXIV (1974) PLATE X
(a) (b)
(c) (d)
THE VIA EGNATIA: TWO MILESTONES FROM BROSTAR (see p. i89)
Photographs taken from ' Monumentet ' I971, 2, 25
THE WESTERN PART OF THE VIA EGNATIA I93
takes some 75 km, and the latter takes some 44 km to Semen and some 32 km from Semen
to the same point near Elbasan,totalling some 76 km. Any other pair of branch-roadsgives
very unequal distances: Clodiana-Ad Novas-Apollonia is 73 km, while Clodiana-
Dyrrachiumis only 45 km, and Genusus-Apolloniais 66 km while Genusus-Dyrrachium
is only 22 km. We conclude then that the Hellenistic roadsfrom Apolloniaand Dyrrachium
met just west of Elbasan.
The Roman road-system began to develop towardsthe end of the second century B.C.
During the Republic it was used primarilyfor military purposes; Cicero regardedit as a
military road in 56 B.C. (Prov. Cons. 2), and Pompey's officer Bibulus travelled along the
road from east to west by day and night with changes of horses in the Civil War in 49 B.C.
The road-terminalon the Adriatic coast was now Dyrrachium. For when Roman armies,
and merchantsalso, were travellingfrom Italy to the East, the best route was from Bari or
Brindisito Dyrrachiumand thence by the Via Egnatia; for this did not involve the crossing
of any largeriveruntil one cameto the crossingof the Shkumbifar inland at Mirake. On the
other hand the voyage from Brindisi to Apollonia involved contending with the current
of the Vijose, and the journey from Apollonia to the Via Egnatiameant that one had first to
be ferried over the Semeni (Apsus), as we know from Caesar,Bell. Civ. iii, 77, and then to
cross the Shkumbi probably near Shenjan, perhaps by the Hellenistic bridge. Strabo, for
instance, when he was not quoting the views of Polybius but writing for himself, referredto
the Via Egnatiaas startingfrom Dyrrachium(C 329, fr io.). The Via Egnatiawas described
as startingfrom Dyrrachiumin two inscriptionswhich recorda repairof the road by Trajan
(BCHlix (I935) 403 ' Viam a Dyrrachi(o) usque Acontisma ' and 408). This is the case also
on the Peutinger Table, which may be based on a Romanroad map of the Antonine period,
and in the Antonine Itinerary 317,7 which was compiled in 211-17.
At a later stage in the EmpireAulon was the road-terminal. This is so in the Antonine
Itinerary 329, I, which was composed shortly after 286, and again in the Itinerarium
Burdigalense,which was written by a travellerwho used the Via Egnatiain 333. Aulon, not
Apollonia, was now the port of the Via Egnatia. It is probable that the Vijose (Aous) had
changed its course and flowed far away from Apollonia. In addition Aulon was an all-
weatherharbour,since it had two havens.43But it is not easy to see why Aulon was preferred
to Dyrrachiumat this period. It is possible that the inner harbourat Dyrrachium, which
was presumablyto the east of the town where there was a lagoon until recently, had silted
up and become less serviceable.
When Aulon becamethe terminal,it was more than ever desirableto have a good route
into the Shkumbi valley. The road Aulon-Apollonia-AdNovas-Clodianais recorded only
in'that part of the Antonine Itinerarywhich was composed shortly after 286. It must have
been constructedat a time when the Semeni (Apsus)was flowingto the south of Shinepremte
and when either the plain between Babunje and Lushnji was not subject to flooding, or
when floods were controlled by channels or dams.44 These conditions no longer obtained,
we may assume, when the travellerof the ItinerariumBurdigalensepassed this way some
50 years later; for he alone recorded a differentroute from Apollonia via Stefanaphanato
Clodiana,the Semeni (Apsus) being crossed by ferry or by a bridge at the point of entry
into the plain. The route from Dyrrachiumwas of course also in use at this period, and the
fact that both the roads from Apollonia aimed for Clodianaon the route from Dyrrachium
shows that the northernside of the Shkumbi was still preferred. There was apparentlyno
attempt to develop the Hellenistic road from Apollonia to near Elbasan as a substitute.
One reason for keeping to the northern side of the Shkumbi was that the road was
alreadyfully developed and had been maintainedwith regularstations for changing horses
and for staying the night. The excavationsat Bradasheshhave shown us for the first time
what fine facilitieswere providedat a mere ' mutatio ' on this part of the route from the early
second centuryA.D.onwards. It was desirablealso to be on the north side of the river from
Elbasanto Mirake,becausethe terrainthere was much more suitablethan the terrainon the
south side. From Miraketo Orakethere was only one route in use as far as we know, and this
was due primarilyto the configurationof the mountainswhich were called ' Candavia'.
48
Described in Epirus I32 f. across the plain for the first time at the end of the
44 There is an interesting parallel in the coastal third century A.D. or thereabouts; see C. F. Edson,
plain of Macedonia, where a Roman road was built CP 1 (1955) i8o, and Macedonia i, I6o f.
194 THE WESTERN PART OF THE VIA EGNATIA
The coastal road was a part of the Roman road-system in Greece and Illyria. For
instance, in Epirus the road ran along the coast, keeping as close to the sea as possible and
touching on the main ports of the period. Side-roads ran inland from the ports, for instance
from Ambracia (Arta) to Larissa and from Onchesmus (Sarande) to Hadrianopolis (near
Gjorgucat).45 The road from Apollonia to Dyrrachium was just such a coastal road. As long
as the Aous (Vijose) flowed close at hand and was navigable, Apollonia was an important
port from which one road ran inland to Hadrianopolis and another road went to join the Via
Egnatia. It is possible that the Apsus (Semeni) and the Genusus (Shkumbi) were navigable
at some seasons of the year; if so there were probably subsidiary ports near the mouths of
these rivers. However, during most of the Roman period Dyrrachium was the most
important of the harbours; from it ran the main road which connected the western provinces
and Italy with Macedonia and the eastern provinces of the Empire.
The University of Bristol
4 See Epirus 690 f. and
map 18. A Roman existence of a branch-road from Onchesmus to
milestone found recently near Murzine has shown the Hadrianopolis.
