The Piscina Mirabilis – Water for Misenum

The Bay of Naples, Campania, Italy


By John Allen (University of Sheffield, Material Sciences, retired) and David Millar (University College, Dublin University, Computer Systems Engineering, retired)

The Piscina Mirabilis is a 1st Century BCE water cistern at the end of the Augusta Aqueduct, the longest ancient Roman aqueduct in Europe, which provided water to the port of Misenum, the headquarters of the Western Roman Fleet.

A map of the Bay of Naples area of Italy showing possible routes of Roman aqueducts


In 30-29 BCE, after the defeat of the forces of Antony and Cleopatra at the Battle of Actium, Octavian, soon to become the Emperor Augustus, moved the Western Roman Fleet from the silted up Portus Julius, a harbour near Baia, further south to Misenum.

However the area surrounding Misenum was low in fresh water and would have to be brought in. Roman culture was built around the need for large quantities of flowing water. Romans would use up to 500 litres per person per day whereas even in modern high consumption Western countries usage is rarely above 125 litres. Roman baths played a large part in this water usage.

A solution was needed. Work began on the Augusta Aqueduct under Octavian’s right-hand man, Marcus Vispanius Agrippa and much, if not all, of it planned by Lucius Cocceius Auctus.

The Augusta Aqueduct

There is no record of the construction dates but it had to start after the Battle of Actium (31 BCE) and it was in operation by 10 CE (a commemorative inscription of the opening of a new input channel has been discovered). It is reasonable to propose that planning and surveying took up to 10 years and construction a further 20.

Most of the route hugs the contours or is in tunnels. The land is volcanic, good for tunnelling but bad for building viaducts, and only a few viaducts existed with nothing like the Pont du Gard.

This was a major exercise. It was surveyed using simple tools like the groma and libra (see right). Other tools were the chorobates, a table with water levels, and the dioptra, an ancestor of the alidade. But the first was too cumbersome for rough land, and the second rarely used by the Romans at this time. To maintain the gradients there would have been years of iterative surveying, from source to end and back again, and by segment, again in both directions.

There then followed the construction, in all probability breaking the route into segments each assigned to a unit of the Roman Army, the main source of Roman construction expertise, supported by local labour and slaves. All these work teams along 145 km had to be managed, trained, fed, housed and checked. A massive project comparable to building the 19th Century railways!

diagram of a Roman Libra
clip art of a person holding a traffic sign
map of Misenum town, port and the Piscina Mirabilis

Misenum and the Piscina

The Western Roman fleet was now based in Misenum. 22,000 oarsmen and troops plus families, dockyard and support workers, and patricians lived here. They needed a secure supply of fresh water. The Augusta was a small aqueduct by Roman standards providing 3,000 cubic metres a day.

Also there was only a single aqueduct whereas most strategic Roman towns had at least two. So Misenum needed a fall-back in case of damage or repairs to the channel. This was provided by the Piscina Mirabilis, a cistern holding 14,5000 cubic meters .

Diagram of the Piscina Mirabilis cistern

The Structure of the Piscina

aerial photo of the remains of the Piscina Mirabilis

The Piscina stands on a hill opposite the town. Its 14,500 cubic metres of water are enough to keep the town going for nearly 6 days, more if rationing was introduced.

The floor is waterproof cement mixed with potsherds then polished. The walls are Roman herring-bone pattern of bricks, the side pillars are kiln-fired bricks and the central pillars are tuff ashlars – all covered by waterproof cement.

The Piscina has a brick-vaulted roof to prevent evaporation and there are a number of holes, now covered, in it. The aqueduct entrance is in the inside northwest corner by the entrance.

There is a ditch in the centre with a drainage hole at the south end, described in previous research as a “piscina limaria”, a settling pool for debris. There is a 1 m high ledge along the northeast wall, the function of which was unknown.

The Piscina is embedded in the surrounding volcanic tuff on three sides. On the fourth, northeast side there is a retaining wall supported externally by large buttresses.

The Reasons for the Study

When we looked at the Piscina, with our engineer’s hats on, two questions immediately came to mind:

• Where did the water go once the Piscina was full?
• How was water provided if the aqueduct stopped flowing?

Research in the 1930s, and repeated since, stated that the aqueduct ended at the Piscina, and the water was then extracted via the roof using “hydraulic engines”.

But Roman aqueducts flow continuously. You cannot turn one off. So the aqueduct had to run past the Piscina. An engineer does not build a tank for water collection without allowing for overflow – and the Romans were good, practical engineers! Regarding the extraction of water for the  town, we studied waterwheels and Archimedes Screws (both need fixed inputs and outputs and the sizes needed were beyond the Romans),water pumps (the Greeks invented one but the capacity was way below that needed), and buckets on pulley systems. This last seemed feasible.

Two men, John Allen and David Millar in the ruins

John Allen & David Millar

But mathematics showed that 10 systems on the roof operating 24 hours a day would still take more than a day to extract the daily amount of water required by the town. And these solutions also ignore the question of how did the water reach the population from the roof. To use pack animals would have required 250 such animals working 24 hours a day. Piping it was the obvious solution.

Archaeological research has a habit of assuming earlier research without testing this out. Historians may not also be engineers.

Study Methods

Roman engineers and project managers have little visibility. The works of philosophy, politics and religion were preserved and copied to be passed down to us. Little is known of the creators of aqueducts and cisterns, and nothing of the managers, surveyors and builders. But existing remains demonstrate construction and hydraulic skills not matched for 1,800 years

Our study was based on ground work. John walked the land, studied remains, measured everything. With clipboard, camera and measuring instruments he was often mistaken for a tax inspector by the locals!

a collection of books, tools and maps related to the Piscina Mirabilis

David backed this up with research in web-based libraries and research bodies and, Covid epidemic permitting, university and library resources in Italy and the UK.

We built an extensive paper and electronic library; the works of 1st C CE Romans, etchings from the 16th and 17th centuries, works by 19th C academics, recent 21st C research, and our invaluable “Roman Aqueducts & Water Supply” by the late Professor Hodge.

Links were made with UK and Italian academics and we appreciate their advice and encouragement, especially the help from the Cocceius Association.

But foremost in our study were our engineering and planning skills. Everything was checked by observation and measurements, supported by the authors’ personal and practical experiences as engineers and project managers. At every point we asked “How would we have done this?”

Our Main Conclusions

  • The Augusta Aqueduct crossed Bacoli on a viaduct and entered the Piscina on the northwest corner at approx 45°.
  • The Aqueduct ran beyond the Piscina Mirabilis, through or alongside the Piscina, to a castellum divisorium where the water flow was divided between the users. Water was not extracted via the roof of the Piscina.
ruins of a cistern, photo © Carole Radata, Wikimedia

© Carole Radata, Wikimedia

When running normally, water was channelled to a castellum divisorium. When the Aqueduct was not providing water, water was received via pipes at the bottom of the Piscina.

Above shows a castellum divisorium in Nimes, France. Note the aqueduct entry at the back and the output pipe holes at the front. These would be controlled by sluice gates.

  • The function of the central basin was to be the exit point of the “emergency” low level output pipes and and the collection of debris was secondary.
  • The platform against the internal NE wall of the Piscina is part of the foundations of that wall. Having one edge unsupported by the surrounding tuff simplified the construction of the Piscina.
Diagram of buttress and foundation of the Piscina Mirabilis

Questions and Next Steps

  • Were the Aqueduct and the Piscina planned as an integrated system, or was the Piscina inserted after the Romans realised that they needed a back up and a provision of water pressure? We will never know unless an inscription is discovered.
  • Did the main aqueduct flow through or past the Piscina?
  • Where are the upper overflow channels and the lower extraction pipes in the Piscina?
An aerial view. The possible site of the castellumdivisoriumis in the orchard area to the SW of the Piscina © Google Earth

© Google Earth. An aerial view. The possible site of the castellum divisorium is in the orchard area to the SW of the Piscina

  • Where was the castellum divisiorum located?
  • What is the function of the two output pipes in the central basin?
  • Can we confirm that the elevated platform on the northeast wall is part of the foundations?

Publication and a Second Paper

We would like to return to the site with ground radar and thermal imaging equipment to examine the site where the castellum divisorium may be located, also to examine the internal walls of the Piscina and the central ditch looking for the remains of pipes.

An analysis of the cements inside the Piscina was carried out by Federico II University of Naples in 2015. We would like to carry out a chemical examination of the bricks and ashlars used.

Ground radar alongside the NE wall would help answer questions about its foundations.

Both the above areas outside the Piscina could be excavated depending on the ground radar results.

Our full 38-page version with references is now available. Contact us at if you would like an electronic copy.

We will publish a second paper, covering the route of the Augusta Aqueduct from Baia across Bacoli to the Piscina Mirabilis (approximately 3 km) in the winter of 2022/2023.

This will look at the aqueduct tunnel from Baia, the path round the two intervening volcanic craters, the location of the viaduct in Bacoli and a branch to the Castrum area.

Cover of the article 'Water Supply from the Piscina Mirabilis to Roman Misenum'
a person in a hard hat and mask crawling through a rocky tunnel, photo copyright Cocceius Association

© Cocceius Association

This second paper covers our research but also the exploration of the 1 km aqueduct tunnel from the baths at Baiae. This was done by the Cocceius Association, speleologists investigating man-made caves, tunnels and water conduits in the Naples area. David and John have been (non-caving) members of the Association since 2021. We are indebted to their support, especially their Director, Dott. Graziano Ferrari.

John Allen, University of Sheffield, Material Sciences. Initially technical advisor on metallic surface preparation and conditioning in UK and USA. Later Aerospace Ground test on complete aircraft in Cambridge, Belfast and Montreal. John has travelled extensively during his working career.

David Millar, University College, Dublin University, computer systems engineering, project management, originally in process control applications but latterly in the banking industry, also writer and lecturer in financial risk management. David worked extensively across six continents in his career

John and David are now both retired and resident in Cowes on the Isle of Wight, United Kingdom and regularly visit Italy and other European countries.