Abstact
The focus of this paper is the amazing water works of Rome, those fabulous aqueducts which gave Rome clean, fresh drinking water for nearly a thousand years. This examination will focus on a single aqueduct, the Aqua Marcia, and detail how it provided water to the city of Rome. It will be shown that the Romans preferred to repair and rebuild existing aqueducts rather than construct brand new ones. In the process it will be shown that construction techniques evolved during the thousand year history of the water system.

A Brief History
Until the year 343 BC, the Romans were content to draw water from springs located around the city. The river Tiber was normally too muddy and turbid to serve as drinking water, and was used for refuse. In fact, the sewer, Cloaca Maxima, dumped directly into the river. The aqueducts were designed to bring water into the city, funnel it through fountains and baths and then flush the sewer into the Tiber (De Aquis, II, 111).

In the year 343 BC, the first rudimentary aqueduct, the Aqua Appia, was built to bring water into the city. It was a crude underground water channel built by censor Appius Claudius Caecus during the Saminite War (De Aquis, I, 5). Next, the Anio Vetus was built in 272 BC while Spurrius Carvilius and Luius Papirius Cursor were consuls. It was an improvement over the first aqueduct, but also traveled more than ninety-nine percent of its length completely underground (De Aquis, I, 6).

In 144 BC, praetor Q. Marcius Rex was commissioned by the Senate to construct a third aqueduct to bring water into the city. Using 180,000,000 sesterces,1 booty from the sacking of Carthage and Corinth (MacKendrick, p394), Marcius designed and completed the aqueduct in two years. It was not until 140 BC that water from the Upper Anio Valley actually flowed into the city through the aqueduct. The delay was caused by a conflict over the Sibylline books, which said that the water from the Anio Valley should not enter the Capitol. Marcius Rex was able to convince the Senate to allow the water into the Capitol, and the aqueduct was named Aqua Marcia after him (De Aquis, I, 7).

In 127 BC the censors Cn. Servillius Caepio and L. Cassius Longinus brought water from the region of the Alban Hills into the city of Rome. This water was called Tepula because of its warm nature (De Aquis, I, 8). Ninety-two years later, Marcus Agrippa built the Aqua Julia and caused it to mix with the waters of the Tepula in order to provide a cooler drinking water supply (De Aquis, I, 9). Herschel2 notes that the 63°F Tepula water dropped to 53°F when the 50°F Julia water was introduced (Herschel, p164).

Agrippa also built the Aqua Virgo in 20 BC, so named because a young girl is said to have led some thirsty soldiers to its source (De Aquis, I, 10). Agrippa, under direction of Augustus, built the Aqua Augusta in 33 BC to supplement the Aqua Marcia in times of drought (De Aquis, I, 12). Later, the Augusta would also supplement the Claudia when needed (De Aquis, I, 14).

In 2 BC, Augustus also built the Aqua Alsietina to serve a function that Frontinus believes was primarily to fill the Naumachia and for watering gardens and other uses that do not require quality water (De Aquis, I, 11).

Caligula was the first to actively destroy an aqueduct by razing the portion of the Aqua Virgo that passed through the Campus Martius to make room for his “monumental amphitheater” (Van Deman, p12). The Aqua Virgo was later restored by Claudius (Van Deman, p14).

Under the reign of Claudius, two major projects were completed that had actually been started by Caligula. Thus, it came to pass that in 47 AD, the Aqua Claudia and Anio Novus aqueducts brought water into Rome and brought the number of aqueducts serving the city to nine (De Aquis, I, 14). The Claudian water was often compared to that of the Marcia because of its clarity. The Anio Novus, however, was not always so reliable. Heavy rains caused its source, the Simbruinum River, to become “muddy and discolored” despite the efforts of the water men of the time (De Aquis, I, 15). At least they knew when a storm was approaching!

Nero damaged a branch of the Aqua Marcia by constructing an “imposing, but unstable” series of arches (Van Deman, p14). Trajan later repaired the amplum opus branch of the Marcia (De Aquis, II, 92).

In 109 AD, while Frontinus held the office of water commissioner, the Aqua Trajana was built to service the industrial section of the city (Ashby, p14). Caracalla created the Specus Antoniniana in 213 AD to help alleviate the strain on the water system from new public baths by tapping a spring of the same name (Ashby, p14). Alexander Severus built the Aqua Alexandriana in 226 AD to provide water for the rebuilt Thermae of Nero. (Ashby, p14).

Major damage to the system was incurred during the Gothic siege of the city in 537 AD. Both Vitigis and Belesarius cut the aqueducts in an attempt to keep the other at bay (Ashby, p15). But time, more than any human force, proved to be the passive and limitless destroyer of the water works.

By the time the Empire began to crumble, Rome was served by eleven major aqueducts, with the Aqua Augusta, Specus Octavianus (off the Anio Vetus) and Specus Antoniniana (off the Marcia) being minor, branch aqueducts bringing the total to fourteen (Ashby, p15). During the reign of Severus, the aqueducts brought an estimated 250,000,000 gallons of water to Rome each day (MacKendrick, p389).

The Aqua Marcia
The Aqua Marcia travels much of its length underground, primarily for protection against invaders who would wish to cut the supply of water to Rome. It is interesting to note that later aqueducts, built during the Pax Romana, were carried on arches and other above ground structures for much greater distances than their earlier counterparts. This can not be contributed solely to safety, but also to a higher technological level. This is illustrated through increased usage of arches to cross valleys and other undesirably low areas with the Marcia as opposed to its other, more ancient siblings. The last of its length, as it approached the city, was above ground, either on arches, or on “substructures.” Within the city, water was distributed through lead pipes (Ashby, p149, Herschel, p224). A major distribution center is located at the Viminal Gate (Van Deman, p122). An inverted siphon was used to bring water up to the Capitol (Ashby, p152, Herschel, p193, Van Deman, p139). This is primarily conjecture, since no arch ruins or literary arch mention have been found. The water is filtered though piscinas to remove pebbles, like the kind that annoyingly find their way into shoes. The piscina for the Aqua Marcia was a large two storey structure, probably located underground, judging by the lack of opus reticulum on the outer walls (Ashby, p133). There were three chambers on the lower level, four on the upper. Water flows in the structure from the left, down into the lower chamber, left into the second chamber, and then up and out. The pebbles are trapped in the two lower chambers because while they can roll along the bottom of the aqueduct with the water current, they are too heavy to be lifted up and out of the lower chambers of the piscina (Herschel, p200).

The Aqua Marcia, with the Tepula and Julia piggybacking in their own channels, traveled in a single conduit from their respective sources to the piscina outside the city, from which it broke into six branches. An above ground branch served the Caelian and nearby hills and was shadowed by the Rivus Herculaneus which was an underground channel that primarily served the Caelian. A small branch served the Forum of Trajan, and another went to the reservoir of the baths of Diocletian. The Aqua Antoniniana broke from the Marcia inside the city and supplied the baths of Caracalla. Finally, the Marcia found its terminus at the Capitol (Van Deman, p69). The Marcia served the first, third through tenth and fourteenth wards of the city with its branches (De Aquis, II, 81).

The Marcia was continually shortened during its reconstructions. The reason for this is that initially much of the aqueduct was built underground for protection from invaders. Remember that it was originally built during the Republic, a time of continual warfare. Underground channels were more difficult to cut. But those threats did not exist during the Empire, and so the repair crews made use of this by eliminating unnecessary underground loops. An average length of the Marcia can be considered to be about ninety-one and a third kilometers (fifty-seven miles),3 with about eighty kilometers (fifty miles), or eighty-eight percent, underground and just over eleven kilometers (nearly seven miles) on arches or substructures (Van Deman p124). At its source, the Marcia was 274 meters (900 feet) above sea level (Herschel, p153) and delivered 61,088 liters of water every twenty-four hours (De Aquis, II, 67). More than twice this value entered at the source, but was lost to leaks along the way.4 The channel of the aqueduct was generally ninety centimeters wide (two feet, ten inches), but varied between 120 centimeters (three feet, ten inches) near its source to seventy centimeters (two feet, three inches) at the Porta Maggiore (Van Deman, p89, 124, & 128).

The Course of the Marcia
The Aqua Marcia drew its primary source from quiet, deep green hued pools called the second and third Serena that were collection basins for a series of springs (Van Deman, p69). In the highlands of Tivoli, near the source, the aqueduct was generally cut from rock or in shallow trenches and hugs the hillsides. It is this upper section of the aqueduct that exhibits a curious inclusion of both the primitive pointed roof structure and the more advanced “vault” style (Van Deman, p124). It appears that the more advanced flat style of roof was used where the terrain did not permit usage of the pointed style. The new flat roofed style of channel also required a better mortar than previously used,5 and so the builders of the original channel developed a “pseudo-concrete,” which was mostly river-bed sand mixed with good quality lime. So necessity was the mother of invention when it came to the highland stretches of the aqueduct (Van Deman, p125). A section of the original channel is of better design, sporting a flat roof. This channel has been cut from the hillside rock, and the channel made whole using limestone blocks.

As the aqueduct began its descent into the uplands below Gericomio, the conduit changed to accommodate the new surroundings. Instead of hugging hills and generally following the contour of the countryside, the aqueduct was buried in deep channels, tunneled through ridges, and crossed valleys and ravines on arches (Van Deman, p126). This change was to maintain level for water transport since the aqueduct was at its lowest point, and no longer in hill country (Van Deman, p91).

At Capannelle, the aqueduct approached the city above ground, either on substructures, or when closer to the city, on arches. A stretch of magnificent arches topped only by those of the Claudia, was nine and a half kilometers long (nearly six miles), with a height of nearly fifteen meters (forty-eight feet) (Van Deman, p126). The arches generally had a span of five and a half meters (more than seventeen and a half feet) (Van Deman, p127). The stone blocks that make up the arches were well cut and dressed, with pure lime as a mortar between them. Common to the time was to place poor tufta in the core of the arches and have the better stone on the outside. This provided a measure of economy in construction (Ashby, p135).

Much of the original channel was made from cut blocks of stone which secured with out the benefit of clamps. A groove was cut in the end of each block, and when placed next to another block, the grooves together formed a circular cavity. Mortar was pored in, and when it hardened, formed a dowel pin of six centimeter (three inch) diameter, that resisted shifting of the blocks (Van Deman, p106,115). The channel was made fairly water tight by a thin layer of lime mortar (Van Deman, p105).

Inside the City
The aqueduct, upon reaching the city, joined the city wall and traveled through the Porta Maggiorre and Porta Tiburtina and others. It then broke off and moved into the city for distribution to the various wards. There were fifty-one local reservoirs of the Marcia around the city (Van Deman, p69). Aqueducts stacked upon each other can clearly be seen, with the Marcia on the bottom, the Tepula in the middle and then the Julia top (Van Deman, p119).

The Marcia’s official terminal fountain in the Capitol was marked by a statue of Q. Marcius Rex, its original designer behind the temple of Jupiter Capitolinus (Van Deman, p139). The Rivus Herculaneus broke from the main channel near the Porta Tiburtina (De Aquis, I, 19) and remains of the water tower, which sent water underground, have been found in the Aurelian Wall (Van Deman, p140). This branch, built by Augustus met its terminal reservoir near the Porta Capena, and gave the arch the nickname arcum stillantem, the dripping arch (Martial, Epig. III, 47). The above ground branch that carried water to the Caelian, the amplum opus, was possibly a component of the original construction (Van Deman, p142), but this is difficult to tell, since Nero knocked it down for his Arcus Neroniani (De Aquis, II, 76). The reservoirs that belonged to that branch were given to the Claudia until Trajan restored the amplum opus (De Aquis, II, 76).

Lead pipes have been found bearing the names of Petronius Sura, procurator aquarium, and the Emperor Hadrian that led to the Forum of Trajan (Van Deman, p143). Van Deman believes that Hadrian completed construction on the forum and the water supply leading there (Van Deman, p144). On its way, the branch supplied several fountains, including one near a house of Martial, who applied for a direct connection to the system (Martial, Epig, IX, 18). In 212 AD, Caracalla constructed his extensive baths and used the Marcia to supply water to the huge reservoir of the baths. Remains of the reservoir can be found behind the ruins of the baths (Van Deman, p146). To counter the volume of water diverted for the baths, Caracalla built a secondary supply for the Marcia, called the Specus Antoniniana instead of constructing a whole new aqueduct (Van Deman, p144).

The Restorations
During the restoration of Agrippa and Augustus, the portions of the original channel that had pointed roofs were not always replaced, but covered by a coarse layer of “friable” concrete. This restoration was characterized by an early coarse concrete made from low quality tufta and thin mortar of poor lime quality. The facing was of irregularly shaped reticulate blocks (Van Deman, p84). The limestone blocks makes up the arc of the arch itself. The rest of the structure exhibits a slightly better concrete with better aggregate, suggesting the builders were slightly more careful. The wall on the left of the road is a reinforcement of Titus, judging by the higher grade concrete (Van Deman, p85). The Augustan phase of restoration was actually a two part process, first Agrippa worked on the system during 33 BC, followed after his death by actual supervision by the Emperor himself, Augustus, between 11 and 4 BC. It would appear that Augustus fixed what Agrippa had not the time for. In some places, where the channel was in such disrepair as to cause a nearly complete reconstruction, the workers of Agrippa and Augustus altered the line of the aqueduct. This was not a major course change, merely an adjustment, with the general line unchanged (Van Deman, p129). Beginning with the Augustan restoration, the course of the aqueduct was marked by cippi to help enforce the law requiring a four and a half meter (fifteen foot) buffer zone on either side of the aqueduct that must be cleared of trees, roads and buildings (De Aquis, II, 127). The cippi were white stones placed every seventy-five meters (240 feet) often had marks indicating which emperor restored that particular stretch (Ashby, p128). Putei (Ashby, p128) were placed at irregular intervals during construction and were used for maintenance, such as the cleansing of “deposit” of calcium carbonate left from the extremely hard water (MacKendrick, p393).

The restoration of Titus is characterized by a mix of medium and good concrete with large to medium aggregate and faced on the inside with five centimeter (two inch) thick bricks and thick mortar (Van Deman, p83). Concrete was often laid in irregular rows. Bricks used for facing were medium sized triangles also irregularly laid (Van Deman, p89). Titus added a down shaft to the Anio Vetus from the Marcia using curiously course concrete for the time (Van Deman, p89). The course of the aqueduct was unchanged during the Titian restoration (Van Deman, p131). The restored aqueduct was given as a gift of the gods after being broken from old age (Pliny, XXXI 3, 24).

The Hadrian style of restoration preserved the general design of the aqueduct, for it can be identified as the Marcia (Ashby, p123). The Hadrian restoration is characterized by a use of concrete that is close in texture and homogenous in composition. The accompanying mortar is fine-grained and hard and has a dark gray color. The construction is faced with dark red triangular bricks (Van Deman, p79). The Hadrian restoration also marks the first time local materials were preempted in favor of quality supplies from the city (Van Deman, p134).

The Severan restoration is characterized by firm and compact concrete made with small aggregate laid in straight rows with fine grained mortar. The concrete of this phase is faced with sixty-two and a half centimeter (two foot) tiles called bipedales. During the Severan period, the “fashionable” construction method was to attempt to break apart massive structures by adding pilasters that made the structure appear less heavy, while offering added strength and support (Van Deman, p93).

The restoration of Diocletian was characterized by good concrete with medium aggregate laid in regular rows with coarse, reddish mortar. The concrete is faced with narrow rectangular bricks similar to bipedales (Van Deman, p100). Diocletian liked to add heavy buttressing walls to add support and strength to bridges and arches (Van Deman, p137).

Conclusion
The materials used for the aqueducts feeding the city of Rome began with stone slabs and primitive packed earth-lime mixtures and moved through various stages of sophistication as the Roman engineers learned about how to effectively use concrete. Except for a trend of moving water channels above ground, major techniques of aqueduct building changed little, as can be seen by the few course changes during the various reconstructions. Most emperors preferred to update an aqueduct rather than build new ones when the system was taxed. Diocletian and his Specus Antoniniana branch of the Marcia is an excellent example of that. While the aqueducts provided quality drinking water and helped keep down disease by cleansing the city through baths and the flushing of sewers, they were not maintained terribly well, and as a result, required major reconstructive restorations. Nevertheless, their magnificence and steadfast service to the city prompted many to comment on the amazing system. Frontinus once asked if anyone “will ... compare the idle Pyramids, or those other useless though much renowned works of the Greeks with these aqueducts, with these many indispensable structures?”6 And Pliny remarked “if one takes careful account of the abundant supply of water for public purposes, for baths, pools, channels, houses, gardens, suburban villas; the length of the aqueduct courses - arches reared, mountains tunnels valleys crossed on the level - he will confess than there has never been a greater marvel in the whole world.”7