This paper stems from an hypothesis: the visible water, flowing in the Tiber river or appearing in the beautiful fountains of Rome, finished for hiding by the time a more complex hydraulic network. This happened not only for physiological reasons (pipes run usually underground), but also for cultural reasons (functional uses are often hidden), and for the urban development of the city (streams were covered and diverted).
Literature written in crucial periods of crisis in water supply or in hydraulics emergencies related to river floods is thought to have influenced also the historical interpretation of these aspects of the city life. In fact, this literature was often produced by erudite men and not by technicians, and had always dealt with the Tiber River. Is it really true that the city could not survive without the mills floating in the river and without restoring ancient aqueducts? Also, about the system created by the popes starting with 16th century, is that only the restoration of an ancient system or another one, different in structure and purposes?
On the contrary, this paper will try to show how the river represented mostly the final discharge of a more complex system of water circulation within the body of the city, and that the center of this system can be placed outside the river. The genesis of this system can be found in the modern age, in the period of more intense demographic growth of the 16th century. Also if the old structures of the ancient Rome had their importance, the characters of the hydraulic system of the modern Rome have a different function, that can be related to a model of water supply common to most part of the cities of Center Italy. Finally, an outline of a hypothesis will be attempted on the evolution of the system during a long period, disregarding the chronological limits of the archive research currently in progress.
It is difficult discuss Roman aqueducts without telling their often-repeated, long story, concerning both their ancient building and their modern restoring. But as it’s just upon this element that I will try to single out the differing aspects between the ancient and the modern system of water supply, it will be necessary to deal thoroughly with the subject. Another reason for this, is that several new studies have appeared in the last ten years, and they present many suggestions that exceed the terms of the problem as had been established by the classical works by Lanciani and Ashby. Anyway, it can be stated that one of the most important examples of emancipation of humanistic disciplines from dependencies with literary sources, is indeed the study of ancient Roman aqueducts. Distinguishing research based upon the mere literary reconstruction from the scientific topographic relief started with Lanciani’s studies and is one of the milestones of archeological science1. Also Ashby, understanding the relevance of leveling ruins, worked side by side with the engineers of the Scuola d’Ingegneria dell’Università di Roma: Reina, Corbellini, Ducci2.
A brief summary of the history of aqueducts: the first Roman aqueduct was the Appio (312 b. C.); it was followed by the first one coming from the Aniene river valley, the Aniolater named Vetus (272-269 b. C.), running underground for defense purposes. In 144 b. C., Q. Marcius Rex brought in the city and then up to the Capitolium some springs by the upper Aniene valley, with the Aqua Marcia aqueduct. In 125 b. C. the less important Aqua Tepula arrived, by the Colli Albani, and in 40 b. C. Agrippa, curator aquarum, added the Aqua Julia, restoring also the main conducts. In 19 b. C. Agrippa again joined near springs of Aqua Virgo near Salone in a conduct without raising parts, that turning beside the city hills entered the town by the Flaminia road bringing water in Campo Marzio. In the Empire period, the building of new public baths proceeded along with the building of new aqueducts to bring water to the thermae. After Agrippa’s death the 240 specialized slaves employed in hydraulic works went under the management of the emperor and the State, forming a familia publica that under Claudius the emperor reached 700 units. These workers were anyway employed in the usual management and supporting, while the new buildings were made under contract3.
Augustus the emperor carried out works of restorations and added new springs, and in 2 a. C. carried the Aqua Alsetina, a bad quality one, to supply a naumachia on the right side of the Tiber. In 52 a. C. Claudius had the two aqueducts built that Caligola had designed in 38 d. C.: the Aqua Claudia, supplied with the same springs of the Aqua Marcia; the Anio Novus, that drew its waters directly from the river. The subsequent emperors engaged themselves in important works of distribution of waters inside the city and of restoration. Only Traianus, in 109 a. C., built a new aqueduct on the right side of the river “to supply the industrial part of the town”, as Ashby says. Alexander Severus finally built the Aqua Alexandriana, later restored with the name of Aqua Felice by Pope Sixtus V.
Procopio’s evidence (14 aqueducts) can be justified either as an exaggeration, or as to include three secondary branches of the existing aqueducts4. Neglecting ancient aqueducts came about very gradually: several evidence affirm the continuation of restoration activity – in particular between the 7th and 9th century for the Aqua Traiana, supplying St. Peter’s basilica and the mills on the Janiculum5. Only after the 9th century the failure of water supply would have prompted to abandon the upper parts of the town for using wells – not only the river – situated in the lower parts near the Tiber.
Hence let’s consider in a more specific way what a Roman aqueduct was and what were its purposes: This analysis will be performed following the many suggestions that T. A. Hodge’s book contains6. He remarks that only the raising parts of aqueducts captured attention, and not the longer parts underground; not enough of this has been considered in Hodge’s opinion “the water supply system as a whole. In a word, the aqueducts have generally been studied rather as archeological monuments than as functioning machines”7. Roman aqueducts in fact were not built for drinking and washing: these basic needs were just satisfied, and continued to be, with the system of springs and private cisterns. They were instead a sign of luxury and urban status that satisfied civic pride, a matter the symbolic interest of which is underlined by a series of monumental and artistic uses of their water.
In other words, there were “two quite separate water systems operating in parallel and independently, fulfilling different purposes and observing different rules”, and the primary role of the aqueduct is still to demonstrate8. What is really important is the public and civic relevance of the aqueduct as an enterprise, compared to the private character of the cisterns. Public baths were the sign of civil living, and Romans often made this aspect in colonial cities perceivable: aqueducts were for their own needs, not for native inhabitants who continued to employ springs and cisterns9.
Let’s now examine the structural elements of the aqueduct as a system. When designing an aqueduct the quality of waters was firstly examined; afterwards difficulties in the road were appraised: it was neither possible to build very high arches nor very deep galleries, and often the choice was to border the hills. It was possible also to pass under the valleys by employing siphons, but in the Ashby’s opinion, Romans never realized the capacities of their own cement pipes: “resistance to compression of their hydraulic cement exceed the safety standard accepted today”10. In fact, they did not need pressure aqueducts as on the contrary the distribution system avoided high pressure carefully.
Hodge remarks that the aqueduct in Roman age was not a closed pipe: from a hydraulic point of view it must be considered an open channel, an artificial river. It is not true that lead pipelines could poison people: the internal surface of pipes was quickly covered with limestone to isolate them, and water never stayed inside for a long time11. It is important to underline this problem of aqueducts service: the “tartaro” in the Italian term, or “sinter” in the German one. Romans never solved this problem that produced a progressive reduction in supplying capacities. Deposits found in the archeological works are very impressive: the carefully worked calcareous stone was employed to replace travertine in churches – and in fact its chemical structure is very similar to the “natural” one in marble12.
For the distribution in town, waters were often filtered in multiple reservoirs; it was afterwards distributed by the means of principal castella (247 at the time when Frontinus writes) and still after to the secondary and private ones, the number of which is unknown. This cascade distribution allowed to control pressure in some critical points – not the pipes, but taps. The measurement of water, calculated in quinarie (40 m3 in a day), was practically reliable: the copper intakes, the pipe’s regular size and mostly the water level in the castellum, regularly kept at the same height, leaving exceeding water to flow away guaranteed good results.
Exception made for some few private houses with their own pipeline, domestic supply was assured by public fountains. In Pompei, where it is still possible to precisely reconstruct their locations, fountains were placed in a number and in places as no house had a fountain more than 50 meters distant; of course, the arrival of an aqueduct nearby modified everyday life: no more queues to take water and to irrigate gardens. Pipes, connections and other pipelines distributed waters: usually every user had his own pipe from the castellum to his house, as many parallel pipings have been found, and not a network distribution system13. Also a castella with a tank separated in three parts has been found, as Vitruvius recommended, in Rome a distribution of the water of the same aqueduct relating to the profile of users has never been made: different supplies were performed relating to different parts of the town. The division suggested by Vitruvius (public fountains, public baths, private users) was not necessary in Rome because there were many aqueducts and different employments could be performed – as will be shown further on – in a more sophisticated way, specializing an entire aqueduct to some users and not to others.
There were taps, but usually the system worked with a continuous flowing of water. From the hygienic point of view, the great amount of water arriving in Rome (Frontinus affirms it was almost 1 million m3 in a day) was not useless at all, even if many cities functioned well without sewers. In other words, the continuous flow could not be considered a waste of water: “the only guarantee for public health remained the continuous washing of drains”14. As concerns rain drainage, water was collected in the impluvium of the roman house, it was kept and used; also human dejection was often kept, but there were anyway drains along the roads, and sometimes open sewers.
Relating to water supply inside Rome, no reconstruction can be made without considering the evidence of Vitruvius and Frontinus. Hodge remarks that Vitruvius is often not the best teacher on this subject; moreover, he does not tell how ancient Romans built, but how they should build: for instance he was absolutely contrary to lead pipes, that we nevertheless find very frequently. As for Frontinus, he was not an engineer but a functionary, not very interested in technical subjects: he saw flowing water as an administrative object, a service supplied following a general map of the system; this mental map is not an engineer map, but rather the diagram of a manager.
Frontinus was born at the end of Tiberius kingdom and died in 103-104 a. C.; he was appointed to the cura aquarum in 97 a. C., and he is considered a sympathizer of the senatorial party. His text does not deal with the course of single aqueducts, but focuses on problems of measuring waters and the needs for supply.
From his own survey, the water furnished was more than the disposable amount, and moreover measured at the springs it seemed to be almost double the quantity that the water really supplied in town. As he used a mere geometrical method of measuring the final pipes (fistulae), his evaluation of flow was not reliable; a modern evaluation15 assigned to the roman quinaria a value of 0,48 liters per second for an amount of a million m3 in a day, supplied but the 8 aqueducts functioning at the time of Frontinus16. This value is surprisingly similar to the modern roman oncia. Probably measuring at the beginning and at the end of the aqueduct was not comparable, but anyway Frontinus tried to eliminate many abuses and irregularities in the previous administration of the cura aquarum. They concerned irregular diversions and the respect of public property of the terrain where an aqueduct was placed; he carefully assured a different use for waters relating to their characteristics, separating domestic, industrial, hygienic and irrigation uses.
In spite of the declared purposes of his Commentari,presenting guidelines for the administration of roman waters, political purposes of the pro-senatorial Frontinus appear by the repeated criticism to the previous management, by often comparing himself with Agrippa and by the preparation of maps: “Frontinus therefore presents in his treatise an image of himself as a loyal lieutenant of an enlightened princeps determined to correct the abuses of the past”17. Also if his evidence is the source for all studies on roman waters, his text is “a document presented to celebrate its author and the policies of the emperor who appointed him”18.
A study by Evans remarks how few are the things we know about water supply inside ancient Rome, and tries to answer some questions on the basis of the evidence of Frontinus himself, as it remains the only statistic we have. There has been a long discussion about the amount of the waters flowing in the ancient aqueducts, calculated either in a million m3 in a day, or only in 5 or 600 thousands m3. However his text gives also some information upon the water supply of suburban region, as he claims that almost one third of the water was distributed before getting in the city walls.
As stated, water arrived flowing to castella, whence it was distributed. Vitruvius’ theory, the main tank should be divided in three parts to supply separately fountains, public baths (he considered that they brought in relevant taxes to the state treasure) and private houses. It must be noted that in his classification there are no mills or industrial users. As for the public users, Frontinus indicates public buildings (thermae, circus, etc,), the castra, the munera (that were both monumental fountains and emergency basins), and the lacus (where exceeding water flowed and could be drawn). Let ‘s review again every aqueduct, considering the specific uses.
The Aqua Appia was brought in the lower parts of the town that were only supplied by springs, also for supplying the river harbor (an example was the port of the roman-Etruscan site of Cosa) and commercial activities of the Campo Boario. It followed a low course and brought in a medium amount of water, supplying many lacus. Its maintenance up to medieval period in fact is not clearly distinct by the Marrana stream, that arrived under Aventine hill and near St. Maria in Cosmedin, as will be seen further on. The Anio Vetus flowed higher (it was provided with 35 castella), but not so much to supply the highest parts of the town; its function seems to be limited to the new expanding eastern districts, Esquiline hill. The modest quality of its waters convinced Frontinus to suggest a general use of it, to keep the better drinking water of other aqueducts19.
The Aqua Marcia, for its quantity and higher level of distribution (51 castella) succeeded in supplying with several branches almost every district of the city, including the highest hills. Its excellent quality supported its utilization as drinking water for private users and public lacus, and kept its fame up to the restoration of Marcia-Pia aqueduct in 1870. The Aqua Tepula, an adding stream coming from the Castelli Romani using the rising arches of the Marcia with its own pipeline, was worse in quality and was distributed to private users mixed with the Julia and with some Marcia for general purposes. The Aqua Julia, in spite of its similar quality to the Tepula, was mostly employed for public buildings.
The Aqua Virgo was instead an important aqueduct for distribution; it didn’t worked for the high parts of the city but just for the lowest ones of the Campo Marzio: a relevant part of it supplied the suburban districts. In Rome it supplied mostly public buildings.
Aqua Claudia and Anio Novus flowed parallely in separate pipelines, and integrated one another in a some complex way furnishing both the emperor’s palace (Domus Aurea and Palatinus hill) and a spreading private network, arriving in Transtiberimto increase the modest supply of other aqueducts. Their remarkable branches were the Arcus Celimontanus and a branch flowing high to supply the Marius Trophy on the Esquiline20. Claudia’s quality was excellent, and the amount of water of the two aqueducts together was a special reason of pride for Frontinus.
The Aqua Alsietina of the Martignano Lake was driven to town especially for the naumachia in Transtiberim, and was forgotten after the arrival of the Aqua Traiana. The latter finally supplied good drinking water directly to the districts of the right side of the Tiber; the discovery of channels and mill stones confirm the reconstruction made by Lanciani in his Forma Urbis21. Its quantity was enough to distribute it on the other side of the Tiber, in the opposite direction than before. Without Frontinus’ evidence, we cannot exactly affirm the aqueduct purposes, although the demographic growth of Transtiberim created of course needs in water supply.
Aqua Claudia/Anio Novus are the main water suppliers both for private and for public users: this system appears to be the backbone of water supply of the imperial city. Exception made for many springs inside the town that continued to play their role, other aqueducts showed particular specialization. This specialization is organized for functions and for geographic areas: some waters had a wide spreading (the Marcia as drinking water, the Tepula and the Julia for more general purposes); others had concentrated supply for a less number of users, especially public ones (the Virgo, the Anio Vetus). To private uses prevailing in water supply and irrigation of suburban districts corresponds inside the city a slight prevailing of public uses (44%) on private ones (40%), and an important presence of imperial uses (20% in nomine Coesaris). The older republican aqueducts continued to play their role supplying public basins, the lacus.
A separate problem concerns special uses. Irrigation first of all, as it is known to have been relevant in the suburban area: the amount of water necessary for irrigation is an enormous quantity compared to the basic needs of drinking and washing, also considering that in ancient times it was used for continuous flowing. Mechanical irrigation was not practiced, for the high costs and because it was not possible to raise waters very high. There is evidence about the Aqua Cabra coming down by the Tusculum hill, which was employed in an irrigation channel: maybe it was utilized with a rotation system among users, as it had become common practice in the middle age. But Roman agriculture was in fact dry farming; it seems that the Romans diverted water employed by Berber farmers to supply thermae in their colonial cities22.
The most interesting subject anyway concerns industry: it will give us the essential point to appreciate the differences between ancient and medieval-modern situation23. In Hodge’s opinion – but Evans too seems to be very interested in this matter – water was employed in mines to wash minerals and in town by fullers, but especially to move the hydraulic wheels of mills. The role of hydraulic mills was not as secondary as historical tradition believes, and it is also uncertain if the vertical “roman” wheel is an evolution of the “greek” horizontal one: a simple mechanism could be more important than power, in those times. Anyway, a vertical wheel is particularly efficient if moved by an upper flow, not if moved only by a flowing stream. Hodge is not convinced by arguments to justify a rare utilization of mills in the roman world (the only discoveries in Rome are at the Janiculum and at the Thermae of Caracalla): the Romans knew the structure and function of hydraulic mills very well24. Moreover, aqueducts were an ideal supply for mills. In other words: “the only reason that we think water-mills were rare is that we have not found many, and this may reflect simply a failure either of archeological evidence, or of our interpretation of it”25. However, how many among the hundred of mills cited in medieval documents have archeologists discovered? As the water used by the mill could be introduced in the aqueduct, this utilization could be have been practiced at the birth of the aqueduct and Frontinus would never have talked about it; so, there could be a wide hydraulic network for mills, completely unknown to us. In fact, ruins of a serial mills construction have been found in Barbegal, supplied by their own aqueduct, different by the parallel one used for the water supply of the city of Arles. In the archeological reconstruction, it seems to be a system of eight buildings with two wheels each, a truly industrial system: “how many Barbegals are there awaiting discovery?” Hodge asks, affirming “this may well be the largest and most important question raised by this book”26.
His thesis must be discussed carefully, not only for the importance that author himself place in it: is it only a provocation, or a fantasy? In fact the discovery of millstones in 1990-1991 under the American Academy on Janiculum confirms the existence and the importance of those mills27. Anyway, the fact that not a lot of mills have been found does not prove, and does not exclude, that they really existed; even if wooden parts can disappear in time, mill stones are a durable part, and it’s easy to discover it. While it is usual to find marble parts from ancient buildings employed again in modern ones, no ancient mill stone has yet been found, let’s say, on the ground floor of a medieval church.
Hodge’s opinion seems to be restricted too much to the technological aspects of the problem: adopting a technology is not only a technical matter, but also an economical one. The main argument of economic historians on this subject is that the presence of so many slaves made the labor saving use of mills not necessary, and that the different conditions made it an essential element of development in the urban medieval economy. In fact, the same number of mills would have a quite different relevance in imperial Rome with one million inhabitants and in the pope’s city, with a modest population.