This paper presents a study of the Aqueduct of Salona, capital of the Roman province Dalmatia, built in the 1st century bc. The aqueduct once transported water from the Jadro River spring, situated approximately 3 km east of the city. Even though it was built for a city of 15 ha in size, two centuries later it succeeded in managing the supply and demand of water for a city of 73 ha. In the 7th century Salona was destroyed by the Avars and Slavs, and consequently, the aqueduct ceased to function. Due to intensive exploitation of marlstone and uncontrolled 20th century urbanization, some of the aqueduct's sections have been destroyed. Research on Salona and its aqueduct started as early as the mid-19th century, however, the aqueduct and its route have never been systematically explored until 2014–2015. This paper provides the results and findings of the latest research including the following: the route of the aqueduct, its longitudinal profile, capacity and typical cross-sections, and the method of construction in different terrain conditions. The channel was built in the usual manner in accordance with the practice of Roman builders, using local materials.

HISTORICAL OVERVIEW

The ruins of the ancient city of Salona, the largest Roman city on the east Adriatic coast, are situated 5 km north of the city of Split, in the area of the town Solin (Figure 1). The city was built in a very convenient place: in the middle of the Adriatic coast, in a well-protected part of the Kaštela Bay, at the mouth of the river Jadro, and surrounded by large fertile fields. It is situated only 4 km from the passage to the inland and 3 km from the Jadro River spring, which is characterized by good water quality, large capacity and favorable elevation in relation to Salona.
Figure 1

Location of Salona on a map.

Figure 1

Location of Salona on a map.

The earliest traces of human presence in the Salona area date back to the late Bronze Age (Šuta 2012). During the Iron Age, the Jadro River mouth, with the isle Vranjic, was an important port and the point of contact of the indigenous Illyrian tribe the Delmatae with the Mediterranean world. Delmatian fortified settlements were situated on elevated positions around the port (Katić 2010).

The earliest mention of Salona, in written records, dates back 119 bc, when the consul Lucius Caecilius Metellus with his army spent the winter there. The oldest nucleus of the city, the so-called ‘Urbs Quadrata’, was located to the west of the mouth of the river Jadro and had a trapezium-shaped plan of 15 ha. As the capital of the Roman province Dalmatia, Salona underwent significant development after ad 9 when the Romans finally defeated the last Delmatian uprising, after which the pacification of Illyricum started. The city gradually spread to the east and west of the old nucleus, forming large suburbs which were surrounded by defensive walls and towers in the 2nd century, during the invasion of the Quadi and Marcomanni. The area of the city within the walls expanded to 73 ha. In subsequent centuries the size of the city did not change, but in the 6th century, due to new enemy penetration from the north, the city walls were reinforced and new towers were built (Figure 2). In the 7th century the city was destroyed by the Avars and Slavs, the population fled to the islands and some of them settled inside the safe walls of Diocletian's Palace nearby laying the foundations of the medieval town of Split. Salona was never reconstructed again.
Figure 2

Map of Roman city of Salona (Authors).

Figure 2

Map of Roman city of Salona (Authors).

As the capital, Salona was the administrative centre of great Dalmatia – in that time 7.5 times bigger than today. During the rule of Augustus and Tiberius, a regional road network was built, starting from Salona. The economy was based on agriculture and farming, but also on fishing, crafts and maritime affairs. The Dalmatian inland (present-day Bosnia and Herzegovina) was known for its iron, lead and silver ore wealth, which were exported through Salona port. Estimates have been made by various Salona researchers about its population, 40,000–60,000 inhabitants, on the basis of the size of the forum, the cavea of the amphitheatre, and the capacity of the aqueduct (Suić 1976; Zaninović 1980-81; Cambi 1991; Božić 1997).

The development of the city was followed by the development of infrastructure including water supply, surface water drainage and regulation of water inside and outside the city. Based on the inscriptions on lead pipes, C.JUL.AN.X. (Carrara 1991) ‘Iulius Eucarpus and C. Iulius Xantus’ (Gerber 1917), the Salona aqueduct can be dated to the time immediately after Salona gained the status of a colony in the 1st century bc (Abramić 1991; Kähler 1991). It is considered that the aqueduct ceased functioning in the 7th century, after it was damaged during the siege of the city (Novak 2005). According to Bulić, in the late 19th century, slabs from Salona aqueduct were taken to cover the channel of the Diocletian aqueduct which was under reconstruction (Cambi 1991). Over 1,000 m of the Salona aqueduct channel was destroyed in the 20th century due to intensive exploitation of marl for cement production in the local factories. Residents of this area also contributed to the destruction of the ancient aqueduct by demolishing the parts of the channel on their private plots.

An Italian professor, Abbot Pirona, was first to note the remains of the Roman aqueduct of Salona in 1842. In his letter to Francesco Carrara, who was a director of excavations from 1842 onwards (Jeličić-Radonić & Pereža 2010), he stated that Salona ‘was covered with underbrush and grass and apart from the remains of the aqueduct almost no other remains indicate to the buried city’ (Carrara 1991). The same year a lead water pipe was found near the north gate (Carrara 1991). In 1846, Carrara began major research works in Salona and, among other important findings, he found parts of the Salona aqueduct (Carrara 1991). Frane Bulić, director of the Archaeological Museum from 1884 to 1926, recorded the existence of the Salona aqueduct channel in several places. He re-explored the channel within Salona, previously explored by Carrara, which has remained visible to this day (Bulić 1892, 1895, 1899, 1901, 1911) (Figure 3).
Figure 3

Channel of Salona aqueduct – probe 10 (Authors).

Figure 3

Channel of Salona aqueduct – probe 10 (Authors).

Salona city water supply, surface water drainage and sewerage were first described by Gerber in 1917 (Gerber 1917). He described the way the channel was built, its dimensions and observed calcification on its walls that showed the limit to which the channel was filled. Based on the section of water in the channel and the slope of 0.2%, he calculated, for the first time, the potential capacity of the aqueduct, which according to him was 0.140 m3 per second, i.e. 12,000 m3 per day.

Danish architect Ejnar Dyggve carried out research on the Salona aqueduct as a part of his work in Salona from 1922 onwards. From the late 20th century to the present-day, the most engaged archaeologists on the Salona aqueduct have been Jasna Jeličić-Radonić and Miroslav Katić, from the Conservation Department in Split (Katić 1999). New research conducted in 2014–2015 attempts to give, for the first time, a complete reconstruction of the Salona aqueduct from the spring to the city.

RESULTS OF AQUEDUCT RESEARCH 2014–2015

The research on the Salona aqueduct included a detailed and comprehensive analysis of archival data, drawings, old photographs and archaeological excavations. In order to plan new archaeological excavations, a map with all the previous findings on the aqueduct was made. In addition, based on old maps, photographs and drawings, the original terrain configuration, which has significantly changed over the time, was reconstructed. Afterwards, the field research started; cleaning and geodetic and architectural recording of the seven known and accessible parts of the aqueduct channel. Thanks to the new data it was possible to draw a reconstruction of the Salona aqueduct route, based on the assumption that the average slope of the channel is 0.25% (Vitruvius Pollio 1997), and therefore, the aqueduct conformed to the configuration of the terrain to avoid demanding constructions such as bridges, tunnels or siphons. During field reconnaissance, based on the presumed aqueduct route map and GPS measurements, the channel was found at four additional sites (Figure 4). Archaeological excavations on those sites and a topographic survey allowed a reconstruction of the channel route, altitude relations and determination of different building techniques. Limited funds did not allow probing in other places where the channel was expected to be found.
Figure 4

The route of the Salona aqueduct with circled numbers indicating positions of the probes (Authors).

Figure 4

The route of the Salona aqueduct with circled numbers indicating positions of the probes (Authors).

Characteristics of the aqueduct route

Thanks to the new research the aqueduct route from the Jadro River spring to the north-east corner of the oldest centre of Salona has been defined for the first time. The assumption that the route meanders following the configuration of the terrain proved to be correct. No significant structures were built on the route, such as bridges, tunnels or siphons to span the valleys to shorten the route. The route of the aqueduct conforming to the terrain was a common practice in Classical Antiquity, and was chosen because it was seen as the most economical construction technique (Samiotakis et al. 2004). The reason for this is the adequate elevation of water intake at the Jadro spring (33 m a.s.l.), which allowed gravitational water supply to the city of Salona whose highest point of terrain is at 10 m a.s.l.

The length of the aqueduct from the Jadro spring to the last preserved point north of the ‘Porta Cesarea’ city gate is 4,879 m, while the air distance between these two points is 3,238 m. These data show that the average slope of the channel bottom was about 0.36%, which is slightly steeper than the minimum slope recommended by Vitruvius. The measurements show that the longitudinal slope of the channel varies from 0.18% to 0.27%. However, the average longitudinal slope of the section between probes 7 and 9 is 0.69%, which significantly differs from the rest of the route, therefore, this section should be additionally explored and the longitudinal drop of the channel should be checked, with possible correction of the route (Figure 5). Likewise, the last part of the channel in the length of 250 m between probes 10 and 12 also has an unusual longitudinal slope of 0.89% (from probe 10 to 11) and even 1.77% (from probe 11 to 12), which should be further explored and explained.
Figure 5

Longitudinal section of the Salona aqueduct, circled numbers indicating the positions of the probes (Authors).

Figure 5

Longitudinal section of the Salona aqueduct, circled numbers indicating the positions of the probes (Authors).

Structures on the route of the aqueduct

Every aqueduct, i.e. the main supply channel, usually has two main structures: water intake at the spring, river or lake and distribution tank at the entrance to the city (castellum divisorium), i.e. at the beginning of the city's water supply network (Hodge 2002; Monteleone et al. 2007). There are other structures along the channel route, necessary for stabilization and protection of the channel and its maintenance.

The water was captured and conveyed from the Jadro River spring. The Jadro spring is a permanent karst water source of large and variable capacity (4–60 m3/s) characterized by good quality water, which is still used today, only with a water disinfection process. The water temperature is around 15 °C, varying slightly throughout the year. The concentration of suspended matter and aquatic organic matter is very low. Shorter periods of intensive rainfall and high flows when the turbidity of water is significant are exceptional. The Jadro spring is a cave-spring where the water flows from a limestone cave, i.e. groundwater emerges through several openings in a rock. At the Jadro spring neither the ancient intake structures (castellum fontis) of the Salona aqueduct nor the aqueduct of Diocletian's Palace are preserved. Due to the reconstruction of Diocletian's aqueduct in the late 19th century and the construction of a hydropower plant channel in 1908, the Roman remains were covered or destroyed (Figure 6). It can only be assumed what the intake structure of the Salona and Diocletian's Palace water supply looked like. Therefore, the elevation of the starting point of the reconstructed Diocletian's aqueduct, which is at an absolute elevation of 33.00 m (Katanić & Gojković 1972), was taken for the assumed elevation of the starting point of the Salona aqueduct.
Figure 6

The existing structures at the Jadro spring (M. Maslov).

Figure 6

The existing structures at the Jadro spring (M. Maslov).

The channel fully conformed to the configuration of the terrain, conveying water from the spring to the city of Salona. On its route, at six sites it crossed small intermittent streams, of relatively narrow gorge, which flow into the Jadro. In these sites there were probably smaller bridges or other structures (siphons, incomplete siphons, barriers) and reinforcement which protected the channel from water and stabilized it. However, these structures have not been found or recorded and should be explored. The last section of the aqueduct channel in front of the city, in the length of about 25 m, was laid on a bridge (Kähler 1991; Rendić Miočević 1991) because at that point the terrain drops abruptly (Figure 7). The bridge was built of stone masonry piers 185 × 205 cm in size on which 35-cm-thick massive stone beams/plates rested. The channel was built on the top of them.
Figure 7

Aqueduct bridge south of Porta Suburbia II (Authors).

Figure 7

Aqueduct bridge south of Porta Suburbia II (Authors).

According to the Roman building practice, the aqueduct channel ended in the water distribution structure inside the city (castellum divisorium, castellum aquae) from which lead pipes transferred water to the consumer or to the tanks situated on the lower ground levels. The tanks had the function of maintaining a constant water level or pressure in the distribution system (fountains, bathrooms and the like) (Monteleone et al. 2007). One water tank, of approximate dimensions 4 × 4 m, is situated directly north of ‘Porta Suburbia II’ (Bulić 1911; Carrara 1991) before the water supply channel entered the old city (Figure 8). It can be assumed that this water distribution structure supplied the eastern extension of the city. The main city water distribution structure has not yet been found, but it is assumed to be located at the site of the channel entrance into the old city, ‘Urbs Quadrata’, as is the case in Pompeii (Adam 2008). The channel continued further south, in the thickness of the eastern city wall, passing above ‘Porta Cesarea’ to another water tank which was located in the thickness of the eastern city wall next to the city gate. There are the remains of a public fountain, the Nympheum (Kähler 1991), on the western side of the water tank, which should be explored in more detail. Other water management structures such as reservoirs, water towers or similar have not yet been found.
Figure 8

Water distribution structure north of Porta Suburbia II (Authors).

Figure 8

Water distribution structure north of Porta Suburbia II (Authors).

Characteristics of the channel

The channel of the Salona water supply is of a rectangular cross-section in the whole length, built of massive slabs of local limestone (floor board, sides, cover) mainly on a slope/cut (Figure 9). Solid stone slabs form the basic flow section of the channel. The dimensions of the bottom slabs and the coverings vary from 60 to 130 cm, measuring ca. 25 cm in thickness and 200 to 250 cm in length, and therefore their weight is between 1,000 and 2,000 kg. The thickness of the side slabs varies from 13 to 20 cm, the height from 90 to 120 cm, and the length from 60 to 150 cm. On the outer side, in relation to the slope of the terrain, the channel is walled by the smaller-format rough-cut stone in lime mortar, thus creating a wall of a total thickness 60 to 80 cm. On the other side, which leans against a cut, between the lateral slab of the channel and the cut there is a fill of crushed stone in lime mortar. In some places crushed brick is added to the lime mortar. In order to prevent leakage in the most sensitive place, at the junction of the floor and side slabs, a triangular gasket of waterproof mortar was made, approximately 10 × 20 cm in size. On particularly steep terrain the channel is placed on a retaining wall made of massive stone blocks (Figure 10). The construction of the channel with the stone slabs was not common practice in constructing Roman aqueducts. This would have meant an increase in the demand for heavy labor; extracting blocks of stone, material handling, and placing of the slabs. The precision of the design was high, and the gaps between stone slabs were small, without the application of surface treatment. Only in probe 3 does the channel structure entirely deviate from the usual. Here the channel is made with slabs of smaller thickness, of different quality in the way that the lateral slabs do not lean on the bottom slab, but against it. Only there is the whole interior of the channel (bottom and sides) abundantly plastered with waterproof mortar for reinforcing water tightness (Figure 11).
Figure 9

Typical cross-section of the channel – probe 9 (Authors).

Figure 9

Typical cross-section of the channel – probe 9 (Authors).

Figure 10

Probe 4 (Authors).

Figure 10

Probe 4 (Authors).

Figure 11

Cross-sections of the channel at the probes (circled numbers) marked in Figure 4 (Authors).

Figure 11

Cross-sections of the channel at the probes (circled numbers) marked in Figure 4 (Authors).

Figure 12

Probe 6 (Z. Sunko).

Figure 12

Probe 6 (Z. Sunko).

Only two openings of rectangular shape have been found so far on the whole aqueduct route, which were used for revision, cleaning and repairing of the channel, one in probe 6 (Figure 12) and the other in Salona west of the Episcopal centre. Considering the length of the channel, there should have been more shafts. Until now, no floodgates have been found. The settling tanks or settling basin, which was set at the bottom of the aqueduct channel, have not been found. Natural spring water has a small amount of suspended solids, therefore, there was no need for sedimentation.

On the channel walls, in a number of places, calcification-sinter is preserved (calcite bound to the walls and channel bottom), which clearly shows the level of the water in the channel and the quantity of water that constantly flowed through it. Where the channel is more distant from the Jadro spring there is more calcification. This is a normal feature of karst water. The small accumulation of sinter incrustation was found in the conduit due to the fact that the aqueduct was functioning for six centuries. Sinter-accumulation roughened the channel and increased the contact area for a given amount of water, and it is, therefore, presumed that there were no significant retardations of flow.

Capacity of the channel

Based on the collected data a hydraulic calculation of channel capacity was made for different drops, i.e. for different sections and filling heights using the Colebrook–White formula (Figure 13). Assuming that the average filling of the channel was about 50 cm (about 50% of the channel) and absolute roughness was 5 mm, about 400 l/s flowed through the channel at a speed of 1.0 m/s if the slope was 0.177%, to about 650 l/s at a speed of 1.5 m/s if the slope was 0.69%. According to this, around 34,560 m3/day of water flowed into Salona, which is significantly higher than the previous calculation (Gerber 1917). Assuming that 300 l/capita/day were spent in the city (the quantity which is assumed to have been spent in Rome), this capacity could supply up to 115,200 inhabitants. Salona never had so many inhabitants. Further research will explore these issues in more detail. The rate of flow was higher than 0.8 m/s, therefore, the fast current would keep the channel cleaner, i.e. prevent the concentration of very fine suspended solids.
Figure 13

Flow curves of the channel at two typical sections with different bottom slope (Authors).

Figure 13

Flow curves of the channel at two typical sections with different bottom slope (Authors).

CONCLUSION

The Salona aqueduct was built in the 1st century bc. At the beginning it supplied a city of about 15 ha in size and at the end a city of about 73 ha. It was completely built as a gravity supplier with free water table.

The dimensions of the Salona aqueduct channel range from 62 to 100 cm in width and from 72 to 121 cm in height. These dimensions are similar to the dimensions of the channel flow section of the Diocletian Aqueduct: 60 cm in width, and 120 cm in height (Marasović et al. 2014). Changes in size and construction techniques were conditioned by local features of the terrain. The channel follows the contour lines in the entire length and is built on the terrain or is shallowly buried beneath the ground surface. Where the channel is laid under the ground it is about 1.2 m high and about 60 cm wide. Where it is built on the ground it is wider and its height is slightly less. The channel does not have any significant structures such as bridges or siphons, except for a small bridge before the entrance to the old city. It appears that the route and the construction of the channel without major facilities was an optimal solution hydraulically and economically. The average longitudinal channel slope is about 0.36%, however the average slope at the longest watercourse is 0.25%. The rate of flow was greater than 0.8 m/s due to the channel slope, the settling of suspended solids was prevented, and water was conveyed from the spring to the city in a short period of time of ca. 2 h. Fresh water, at a water temperature of ca. 15 °C, was conveyed to the city.

This aqueduct was built entirely from stone slabs, which was not a common practice. The only known similar aqueduct is the one in Toda (Riera 2006), nevertheless, similar aqueduct construction can be seen at the channel of the Roman aqueduct Anio Vetus (Ashby 1991). More than 20,000 stone slabs, 300 kg to 3,000 kg, were required for its construction. The stone slabs had to be cut, transported, and implemented. Good organizational skills, skilled workers, and a large amount of time were necessary for the execution of the construction works.

The Salona aqueduct illustrates good engineering practice. The channel or the water supply system was in function for more than six centuries, which clearly demonstrates the quality of construction and even more the quality of maintenance and management of this important urban water system. The quality of water was good, partially due to effective water pollution prevention. It is once again shown that the Romans made sure to provide sufficient quantities of quality water to their cities in order to strengthen the sustainability of life in the cities. This paper presents the current, preliminary results of a 1-year study. Analysis of available data and archaeological research will continue in order to study in more detail this valuable historical hydro-technical facility, as well as the Roman practice of construction of urban water systems on the east coast of the Adriatic Sea.

ACKNOWLEDGEMENTS

This paper is based upon work primarily supported by the Croatian Scientific Foundation under HRZZ Research Project IP-11-2013.

REFERENCES

REFERENCES
Abramić
M.
1991
O povijesti Salone.
In:
Antička Salona
(
Cambi
N.
, ed.),
Književni krug Split
.
Split
,
Croatia
, pp.
39
46
.
Adam
J.-P.
2008
Le castellum aquae de Pompéi, étude architecturale. Revue archéologique 2008 (1, n° 45)
,
37
72
.
Ashby
T.
1991
Gli acquedotti dell'antica Roma
.
Quasar, Rome, Italy
.
Božić
N.
1997
Solin – Pregled povijesnoga i urbanističkog razvoja grada. Prostor 5 (2, no. 14) 357–378
.
Bulić
F.
1892
Ritrovamenti risguardanti la topografia urbana dell’ antica Salona
.
Bullettino di archeologia e storia dalmata
XV
,
53
54
.
Bulić
F.
1895
Ritrovamenti risguardanti la topografia urbana dell’ antica Salona
.
Bullettino di archeologia e storia dalmata
XVIII
,
78
.
Bulić
F.
1899
Ritrovamenti risguardanti la topografia urbana dell’ antica Salona. Bullettino di archeologia e storia dalmata XXII
,
244
.
Bulić
F.
1901
Ritrovamenti antici nella campagna di Clissa
.
Bullettino di archeologia e storia dalmata
XXIV
,
176
.
Bulić
F.
1911
Trovamenti riguardanti l'acquedotto urbano di Salona
.
Bullettino di archeologia e storia dalmata
XXXIV
,
66
68
.
Cambi
N.
1991
Predgovor
. In:
Antička Salona
(
Cambi
N.
, ed.),
Književni krug Split
,
Split, Croatia
, pp.
7
36
.
Carrara
F.
1991
Topografija i iskapanja Salone.
In:
Antička Salona
(
Cambi
N.
, ed.),
Književni krug Split
,
Split
,
Croatia
, pp.
99
192
.
Gerber
E. W.
1917
Forschungen in SalonaI. Druck und Verlag der Österreichischen Staatsdruckerei, Vienna, Austria
.
Hodge
A. T.
2002
Roman Aqueducts and Water Supply
.
Duckworth
,
London
,
UK
.
Jeličić-Radonić
J.
Pereža
D.
2010
Topografija antičke Salone (II), Istraživači Salone u XIX. stoljeću. Tusculum 3 (1)
,
167
203
.
Kähler
H.
1991
Porta Cesarea u Saloni
. In:
Antička Salona
(
Cambi
N.
, ed.), Književni krug Split,
Split, Croatia
, pp.
205
240
.
Katanić
N.
Gojković
M.
1972
Građa za proučavanje starih kamenih mostova i akvedukata u Hrvatskoj. Jugoslovenski Institut za Zaštitu Spomeniku Kulture, Belgrade, Serbia and Zagreb, Croatia
.
Katić
M.
1999
Salonitanski vodovod
. In:
Dioklecijanov akvedukt
(
Belamarić
J.
, ed.),
Ministarstvo Kulture Republike Hrvatske – Konzervatorski odjel
,
Split, Croatia
, pp.
58
63
.
Katić
M.
2010
Prethistorijske gradine solinskoga i kliškoga područja
.
Tusculum
3
,
7
19
.
Marasović
K.
Perojević
S.
Margeta
J.
2014
Water supply system of Diocletian's Palace in Split – Croatia
. In:
IWA Regional Symposium on Water, Wastewater and Environment: Traditions and Culture
,
Hellenic Open University
,
Patras
, pp.
163
173
.
Monteleone
M. C.
Yeung
H.
Smith
R.
2007
A review of Ancient Roman water supply exploring techniques of pressure reduction
.
Water Science & Technology: Water Supply
7
(
1
),
113
120
.
Novak
G.
2005
Povijest Splita
.
Split, Croatia
.
Rendić Miočević
D.
1991
Salona ‘Quadrata’, Salonitanski oppidum (Caes., B.C. III 9) u svjetlu novih istraživanja. In: Antička Salona (N. Cambi, ed.), Književni krug Split
,
Split, Croatia
,
47
63
.
Riera
I.
2006
I cunicoli idraulici. Geologia dell'ambiente. Supplemento 4 (Technica di Idraulica antica)
,
15
44
.
Samiotakis
A.
Palogos
Y.
Christodoulakos
Y.
Angelakis
A. N.
Gikas
P.
2004
Engineering aspects of the aqueduct in Roman Minoa Akrotiri, Chania, Hellas
. In:
IWA Regional Symposium on Water, Wastewater and Environment: Traditions and Culture
,
Hellenic Open University
,
Patras
, pp.
569
578
.
Suić
M.
1976
Antički grad na istočnom Jadranu
.
Liber
,
Zagreb, Croatia
, pp.
200
201
.
Šuta
I.
2012
Topografija kasnorepublikanske Salone u kontekstu novih nalaza iz Bencunuša
.
Tusculum
5
,
7–19
.
Vitruvius Pollio
M.
1997
De architectura libri decem
.
Preveo Matija Lopac
,
Zagreb, Croatia
.
Zaninović
M.
1980–81 Greek land division at Pharos. Archaeologia Iugoslavica 20–21
,
91
95
.