Greek and Roman Siphons

Overview:
Examples: Model of a siphon
HT = Header tank (left)
RT = Receiving tank (right)
VB = Venter bridge (bottom)
G = Geniculus (bend)
Headertank and ramp of the Yzeron siphon in the Gier aqueduct of Lyon (France) West part of the venter bridge in Beaunant of the Gier aquaduct of Lyon (France) The receiving tank and ramp of the siphon of the Brévenne aqueduct of Lyon (France) Male side of a stone cut block / pipe of the siphon in the Roman aqueduct in Cadiz (Spain)

HOME More literature on more aqueducts Last modified: April, 2014 - (W.D. Schram)

Introduction to siphons in aqueducts

Tunnels, bridges and arcades are means to adjust the trace of an aqueduct to the specific conditions of the landscape. In antiquity aqueducts operated by gravity flow and the engineers normally followed the countourlines around hills and mountains. This may have meant building very long masonry conduits, but it was in most cases cheaper than the alternative: tunnels to cross mountains and hills, substructions, bridges, arcades or inverted shiphons to cross valleys. Where valleys were too broad or too deep and where a detour was impossible or inpractical, one had recourse to a so-called siphon )1.
Principle and elements of siphons
In the pre-Greek eras one had discovered that water in an U-shaped tube in both legs rises to the same level: the law of the communicating vessels. This principle was applied in ancient siphons - as far as we know the first time in the royal palace in Knossos (Crete) - where the water in an open channel was poured into one or more closed pipes. These pipes descended to the bottom of a depression or valley and climbed up the other side to almost the original level. The water was then again transferred into an open conduit following its original course, see drawing above.
The most important elements of a siphon are the transition of an open channel into one or more pipes: the influx basin or header tank, the two bends in between hills and valleys: geniculi in Latin, the substructure to support the pipes in the valley: the so-called venter )2 (belly), and the transition of the pipe(s) into the open channel again: the outflux basin or receiving tank. Commonly the pipes in the bottom of a valley were laid down on a siphon-bridge so that the river in the valley could pass below without damaging the aqueduct.
The deeper the valley, the deeper was the (static) pressure in the pipes at the bottom. A siphon in a valley of 10m deep contains a water column of the same height which exerts 1 atmosphere extra pressure. The siphon in the Greek (!!) Madradag aqueduct crossing a valley of 200m deep caused a pressure of some 20 atmospheres in the lead pipe, a pressure which equals that in 'modern' steam engines.
Pipes
The Greeks often used cut-stone or terracotta pipes and the same applies for the Romans in Spain. In France in particular the Romans used lead pipes. These pipes had a small diameter which were easier to produce than the bigger ones: commonly a series of pipes were applied in siphons with lead pipes. The most striking example is the 8 - 10 pipes parallel in the nine (!) siphons in the four aqueducts of Lyon (France). Another interesting example is the Greek siphon of the Madradag aqueduct of Pergamon (Turkey) where only one lead pipe was applied encased in concrete (??). The Roman Angitia-siphon (Italy) was equipped with one pipe of masonry (!) which is quite exceptional.
Problems
The application of a siphon was not without problems. One had to be sure that the whole system was 100% watertight, since leaks and / or air-bubbles in the pipe(s) could harm the whole process and make it impossible to start the siphon.
A second problem was the high static (and dynamic) pressure on the pipe(s). A third problem could be caused by the accumulation of debris and calcareous deposits (sinter) inside the pipes at the bottom of the valley.
Classical literature
In the quotation given below, Vitruvius - a Roman architect from the 1 c BC who left a treatise called 'de architectura' - makes reference to the venter (the lowest part of the siphon) and the geniculus (vertical bend), but he also introduced the tern 'colliviaria' a term which has caused much discussion (see Hodge1983 and Kessener2001). One of its function could have been to let the air-bubbles escape from the pipes.
Frontinus is another classical author who wrote about aqueducts. Around 100 AD he was supervising the aqueducts of Rome and wrote the book 'de aquae ductu'. Remarkably he did not make any reference to siphons although in his time they were already applied all over the Greek and Roman world and even in and around Rome.
Choice and application
Under which circumstances were siphons applied? In some cases there was no other choice: a detour was impossible or far from realistic. On the other hand: we think that the Romans were reluctant to built aqueduct bridges of more that 50m high (the Pont du Gard in France is 49m high) because of possible stability problems. A siphon was a good alternative. Cost-effectiveness will have been an other major element in the decision process to apply a siphon. Particularly the Greek built their siphons in valleys which were less deep than 50m - maybe because they were already used to stone and terracotta pipes?
How often were siphons used? In ancient times people were used to fall back on or near well-defendable hilltops which structurally lacked sufficient fresh water sources. This automatically led to the need for an aqueduct. To overcome the valleys down the hill an aqueduct bridge was built or a siphon constructed, in order to get enough fresh water to the people on the acropolis. The same applies to Roman forts for example along Hadrians Wall, because of defensibility, were built on hill tops too. Therefore the determining factors in favor of a siphon will - among other - have been a combination of economy, landscape and location.
Conclusion?
The overview below lists some 60 Greek and Roman siphons. Elsewhere on this website literature was collected on more than 600 Roman and Greek aqueducts. This leads to the suggestion that 1 out of 10 aqueducts were equipped with a siphon.

Wilke Schram, March 2006

Notes:

)1 Stricktly speaking the term 'inverted' siphon should be used since a siphon brings water over a hill or other obstacle, while an inverted siphon leads it through a depression.
)2 Sometimes the word venter is also used to indicate the complete siphon, see Kessener2001.

For those who want to know more:

Top

Quotations

Vitruvius book 8.6
5. An aqueduct which is made of lead, should be thus constructed; if there be a proper fall from the spring head to the city, and hills high enough to cause an impediment do not intervene, the low intervals must be brought to a level by means of substructions preserving the fall directed for channel aqueducts, or by means of a circuitous course, provided it be not too much about; but if there be long valleys, and when it arrives at the bottom, let it be carried level by means of a low substruction as great a distance as possible; this is the part called the venter, by the Greeks koilia; when it arrives at the opposite acclivity, the water therein being but slightly swelled on account of the length of the venter, it may be directed upwards.

6. If the venter were not made use of in valleys, nor the level substruction, but instead of that the aqueduct were brought to an elbow, the water would burst and destroy the joints of the pipes. Over the venter long stand pipes should be placed, by means of which, the violence of the air may escape. Thus, those who have to conduct water through leaden pipes, may by these rules, excellently regulate its descent, its circuit, the venter, and the compression of the air.

Top

Literary sources

Reference)* Short title Page(s) Remarks
Amit2002 Aqueducts in Israel ....  
Bennett2003 The Remnants of Roman Ancyra np Bilkent University Newsletter nr 2
Bestue2006 Roman siphons in Spain 311 In: Cura Aquarum in Ephesus 2006
Biernacka1973 Waterworks 280  
Bol1990 Gadara in der Dekapolis pag 201 AA 1990
Bozic2004 "Talijanova Buza" ..Pag 28-38 SdFG vol 26 (2004)
Casado1983 Hidraulica 487 + 527 table without lit. ref.
Crouch1984 Morgantina 356 plus note 22 in: AJA vol 88 (1984)
Esperou2009 Aqueduc de Béziers 37  
Fahlbusch1982 Vergleich 65  
Garbrecht1982 Wasserversorgung im antiken Rom 181-183 Frontinus series part 1
Garbrecht1994 Desert Fortresses 168  
Hodge1983 Siphons 185 n34 only Roman siphons
Hodge1992 Roman aqueducts 33 + n49 / 428 n43 Greek / Roman
Ilakovac 2008 Roman aqueducts on Pag 130  
Mazar2001 Wasserleitungen nach Jerusalem pag 165 ff in: Dierx2001
Mazloum1936 L'ancienne canalisation d'eau d'Alep Fig 1 Roman?
Ozis1996 Historical water schemes 362 In: Water Resources Devel vol 12-3
Smith1976 Attitudes 52 / 54 Greek / Roman
Stenton1986 Oenoanda 23+24 n33  
Tanriöver2006 Water Supply in Caria 127 in: Cura Aquarum in Ephesus 2006
Tolle-Kastenbein1990 Wasserkultur 75 ff  
Van Buren1955 Druckrohrleitungen 472ff In: Wissowa RE
Weber1898, 1899,1904/5 Kleinasiatischen Stadte   incl. Smyrna and Laodicea
Wilson2012 The aqueduct of Butreint pag 77 - 96 In: Butrint 4


)* Detailed literature references



List of over 60 siphons

Name Country Material Length (m) Height (m) Date Literature source Remarks
France
Arles France lead pipes       Genier 1960 p85
Hodge1983 p189
crossing the river Rhone
Bavay France   1000-1700 15-25   Jolin1955  
Lyon-Mont d'Or-Limonest France lead pipes 420 30 mid 1 c BC Burdy1987 See our own website
Lyon-Mont d'Or-des Plances France lead pipes 3500 70 mid 1 c BC Burdy1987  
Lyon-l'Yzeron-Craponne France lead pipes 2200+3600 33+91 end 1 c BC or early 1 c AD Burdy1991 double siphon
Lyon-Brévenne-Ecully/Tassin France lead pipes 3500 90 early 1 c AD Burdy1993  
Lyon-Brévenne-Lyon? France lead pipes 500 20 early 1 c AD Burdy1993  
Lyon-Gier-Genilac France lead pipes 700 79 mid 1 / early 2nd c AD Burdy1996  
Lyon-Gier-Le Garon France lead pipes 1210 93,5 mid 1 / early 2nd c AD Burdy1996  
Lyon-Gier-L'Yzeron France lead pipes 2660 122,3 mid 1 / early 2nd c AD Burdy1996  
Lyon-Gier-Trion France lead pipes 575 38 mid 1 / early 2nd c AD Burdy1996  
Rodez France lead pipes?   100   Grenier1960 p153  
Saintes France   2500 10-12 early 1c AD Triou1968 pag 125-127
Hodge1983 p192
mainly based on topographical evidence
Vienne France lead pipe     2 / early 3 c AD Pelletier2001 crosing the river Rhone
Greece
Akragas Greece       Greek Crouch1984 pag 356 n22 probably
Methymna on Lesbos Greece stone or stone/clay     Hellenistic Buchholz1975 p57/58
Merckel1899 p506
Koldewey1890??
 
Morgantina Greece       Greek Crouch1984 pag 356  
Nicopolis Greece         Tolle1990 based on Leake 1835  
Olympia Greece lead pipes       Merckel1899 p496
Tolle citing Graber1892
 
Olynthos Greece clay pipes   10 Hellenistic Tolle1990
Crough1993 p71-75 (from Wik p647)
 
Samos Greece stone pipes     Hellenistic Tolle1990 based on Tolle: Samos XIV (1974) p53ff  
Syracuse Greece       Greek Crouch1984 pag 356 n22 probably
Israel
Sartaba / ALEXANDREION Israel masonry
+ slabs (!)
190 110 1 c BC Garbrecht 1994  
Tiberias - Beth Yerah Israel basalt pipes   50   Amit2002 side branche of the
aqueduct to Tiberias
Jerusalem Israel       Amit2002
Samaria / SEBASTE Israel stone pipe       Amit2002  
Susita / HIPPOS Israel stone pipe       Fahlbusch1997
Amit2002
 
Italy
Angitia Italy masonry
+ slabs (!)
    1rst c AD Blake1959 p82
Giovannoni1935 p70
 
Alatri Italy moulded or casted lead pipes 3000 101 2nd c BC Montauzan1908 pag 194-197
Tolle1990 from Bassel1881
Hodge1983 p192 + n54
Bassel1919??
Catania Italy         Wilson2000b pag 25 San Venera al Pozzo near Catania
Rome-Anio Vetus Italy         Van Deman1934 pag 61 just below Ponte Lupo
Rome-Aqua Marcia Italy lead pipes       Status Silv 1,3 66-67 [H92]
Van Deman1934 pag 61??
at Tibur under the river Anio to the villa of Manlius Vopiscus
Rome-Aqua Marcia Italy lead pipes??     140 BC Ashby1935 p152
Van Deman p139
to the Capitol
Rome-Aqua Claudia Italy lead pipes??       Ashby1935 p249-251
Van Deman pag 267
Blake1959 p123n95
Winslow ??
to the Palatine (with doubts)
Termini Imerese Italy lead pipes 1300 40   Belvedère1986,2000 Barratina and Tre Pietre
Spain
Muruzabal de
Andion / ANDELOS
Spain            
Almuñécar
SEXI
Spain terracotta pipes 1100 38 Roman Casado1972 (p149)
Casado1983 p488, p508
Hodge1983 p190 n50+51
Bestue2006
See our own website
Alcanadre
Calahorra
Calagurris
Spain terracotta pipes       Casado1972 (p193)
Casado1983 p510+516
Hodge1983 p191
 
Cadiz - GADEZ Spain stone pipes       Casado1972 (p162)
Hodge1983 p191
Bestue2006
See our own website
Segóbriga / Saelices Spain     20   Casado1983 p515-520
Bestue2006
 
Toledo Spain         Casado1983 p514
Alonso1997
Bestue2006
uncertain, little remains. (venter?) bridge with 1,2 or 3 orders with or without siphon
Zaragoza / CAESARAUGUSTA Spain         Bestue2006  
Turkey
Akmonia Turkey stone pipes   15-25 Hellenistic / early Roman Weber1904  
Alinda Turkey clay pipes       Tanriover2006  
Ankara Turkey stone pipes     Flavian date? Bennett2003  
Antiochia on the Meander Turkey clay pipes   15 Hellenistic Weber1904  
Antiochia in Pisidia Turkey stone pipes   28 early 1c AD Weber1904
Burdy1997
 
Apamea Kitobos Turkey stone pipes   28 early Roman? Weber1904  
Aspendos Turkey stone pipes 1670 45 Roman 2c AD Ward-Perkins1955 p115-123
Kessener1998a, 1998b, 2000, 2001b, 2004
double siphon
Blaundos Turkey clay pipes   15 early Roman? Weber1904  
Cremna Turkey terracotta pipes 300 ?   Owens1991  
Ephesus Turkey lead pipes in stone sockets     archaic / classical Bammer1972 p714-728
Ozis1996
 
Gerga Turkey stone pipes     unknown Ozis1996  
Hierapolis Turkey clay pipes       Tanriover2006  
Laodicea ad Lycum Turkey stone pipes 800-1000 25-50 early Roman Weber1898, Ozis1996  
Magnesia ad Sipylum Turkey clay pipes   30 early Roman Weber1904  
Mylasa?? Turkey         S76 dus Merckel/vBuren??  
Milete Turkey clay pipes       Tuttahs  
Oenoanda Turkey stone pipes 500 22 1c AD Stenton1986  
Patara Turkey stone pipes 260 20 early Roman Hodge 1983 n34
Hodge1992 p33+n49
Ozis1996
Merckel1899 p504-6
 
Pergamon-Madradag Turkey lead pipes 3250 201 1rst half 2c AD Garbrecht1992
Ozis1996
Greek and Roman
Pergamon-Selinus / Atallos Turkey clay pipes   25-30 200 BC Garbrecht1992
Ozis1996
 
Pergamon-Selinus / Demophon Turkey clay pipes   25-30 200 BC Garbrecht1992
Ozis1996
 
Philadelphia Turkey clay pipes   20 Hellenistic Weber1904  
Prymnessos Turkey stone pipes   40 Hellenistic Weber1904 with doubts
Smyrna Turkey stone pipes 3800 158 Hellenistic + Roman Weber1899 + 1904 p95  
Tralleis Turkey stone pipes   75 early Roman? Weber1904  
Trapezopolis Turkey stone pipes ?   40 early Roman? Weber1904  
Tripolis ad Meandrum Turkey ??       Tanriover2006  
United Kingdom
Caerwent United Kingdom wooden pipes       Burgers2001 based on Hansen1970 p85  
Chester United Kingdom earthenware pipes       Stephens 1985 p60  
Colchester United Kingdom wood       Wilson1996 p22 based on Essex Archaeology 12 (1995) piii Gosbeck cultural center
Lincoln United Kingdom ceramic pipes       Thompson1955 pag 122
Smith 67 n37
with doubts
water lifted + siphon?
Other
Butrint Albania Wilson 2012
Constantina / Cirta Algeria         Gsell1901 pag 252
Wilson1997
2x, possibly 3x
Hippo Regius Algeria         Wilson 1997 n 278 probably
Plovdiv Bulgaria         Biernacka1973 p280  
Svishtov Bulgaria         Biernacka1973 p280  
Jader/Zadar Croatia         Bozic2004
Caska (on Pag island) Croatia lead     unknown Ilakovac 2008
Aleppo Syria         Mazloum1936 fig 1  
Palmyra Syria         Garbrecht1982 Mazar2001 refers to Stenton 1986
Uncertain
Gadara Jordan         Bol1990  
Apamea (2x) Syria         Mazar2001 refers to Stenton 1986
Béziers France         Esperou2009 in the Puech Noyé branch
Top
HOME More literature on more aqueducts Last modified: April, 2014 - (W.D. Schram)