A recent study of urban infrastructure in the Roman Empire reveals a consistent mathematical relationship between the population size of ancient settlements and the carrying capacity of their aqueducts. The findings indicate that as Roman cities grew, their engineered water supply increased at a slower, predictable rate compared to the total number of residents. This research was published in the journal Antiquity.
The Roman Empire spanned the Mediterranean basin, Europe, and parts of North Africa and the Near East between 400 years before the Common Era and 400 years into the Common Era. The inhabitants of this vast territory lived in varied environments ranging from arid deserts to temperate coastal zones. To support urban life in these growing settlements, Roman architects and civil engineers constructed elaborate water systems. They built channels, tunnels, and elevated bridges to transport fresh water via gravity from distant springs into urban centers, supplying monumental fountains, public bathhouses, industrial workshops, and elite homes.
Urban scientists use a framework called settlement scaling theory to observe how modern cities grow and function over time. According to this theory, as contemporary cities increase in population, their physical infrastructure tends to expand at a slower, predictable rate known as sublinear scaling. This pattern means larger modern cities generally require fewer structural resources per person than smaller towns. Modern water consumption remains a notable exception to this rule, typically growing at a directly proportional rate to the number of residents, prompting archaeologists to question how ancient water systems fit into these models.
Historically, scientists have doubted whether the physical dimensions of ancient aqueducts could provide reliable clues about the number of people living in a specific Roman city. Determining a baseline for ancient daily water usage is difficult because consumption varied widely among different social classes, with wealthy individuals using significantly more water than average residents. Scholars have also questioned whether Roman engineers actually possessed the demographic data required to plan infrastructure according to specific population demands. This widespread skepticism led many experts to assume that aqueduct construction was driven more by political display than by precise civic planning.
The investigation was led by J.W. Hanson of the University of Oxford, working alongside Duncan Keenan-Jones of the University of Manchester and Davide Motta of University College London. The research team analyzed records from the Atlas Project of Roman Aqueducts, a comprehensive database containing information on roughly 1,300 aqueducts across nearly 1,000 ancient sites. Because physical flow estimates only exist for a fraction of these historic ruins, the researchers narrowed their focus to a specific and well-documented sample. They isolated 57 different Roman settlements that possessed both estimated aqueduct carrying capacities and established population estimates.
To determine the populations of these 57 cities, the researchers relied on a mathematical model that calculates inhabitant numbers based on the density and total count of residential units found in the archaeological record. They then estimated the total maximum volume of water each city’s aqueduct system could deliver in a single day. For the capital city of Rome, which featured 11 distinct aqueducts, the team combined modern uniform flow estimates to determine a total daily supply of nearly one million cubic meters of water. The researchers applied regression analysis to compare the estimated population sizes across all 57 settlements with the maximum water capacity of their respective aqueduct networks.
The statistical analysis reveals a systematic, predictable relationship between the delivery capacity of Roman aqueducts and the population size of the cities they served. The physical dimensions and flow rates of the engineered channels indicate that water infrastructure scaled with urban growth in a precise, sublinear way. The maximum capacity of the aqueducts increased at a slower rate than the overall population of the settlement, with a scaling exponent of about two-thirds. In practical terms, this suggests that larger Roman cities provided less aqueduct-supplied water per capita than smaller towns, diverging from the proportional water consumption seen in modern urban centers.
The study indicates that Roman planners deliberately engineered aqueduct systems to meet functional civic demand, providing a structured approach to urban water management depending on the size of the settlement. By calculating the base ratio of this mathematical relationship, the researchers suggest a maximum daily supply baseline of about 33 cubic meters, or 33,000 liters, per person across the surveyed sites. Roman aqueducts flowed constantly, and the continuous overflow was used to power mills and flush away street waste or sewage. This high baseline volume reflects the total civic water cycle powering the broader urban environment, factoring in massive public bathhouses and municipal cleaning operations.
The dataset captures a wide range of settlement sizes, from the massive urban center of Rome with nearly a million residents to the small city of Hippus with just under 1,000 inhabitants. While the average capacity across the 57 studied sites was about 44,000 cubic meters of water per day, some settlements deviated slightly from the underlying mathematical trend. The capital city of Rome represents a positive outlier in the data, receiving a greater volume of water than its population size alone would dictate under the scaling model. The cumulative capacity of Rome’s aqueducts rose sharply across the survey period, tracking an extraordinary era of demographic expansion in the empire’s center.
Estimating the capacity of ruined ancient infrastructure carries inherent challenges due to centuries of structural degradation and erosion. The water flow rates used in the study represent maximum carrying capacities derived from channel geometry and mineral deposits left behind by hard water, which may overstate the actual daily volume of water delivered during antiquity. The researchers frequently relied on the mid-point of estimated value ranges to run their regressions. The study authors suggest that applying probabilistic statistical models to these architectural ranges might yield more precise historical flow figures in the future.
The study also leaves open the question of how much water Roman city dwellers acquired from unmeasured alternate sources. Residents likely supplemented aqueduct deliveries with rainwater collection, groundwater wells, and local rivers, but the exact proportions of these localized supplies remain unknown. Exploring how natural water availability varied by region, climate, and local geology will provide a deeper understanding of classical urban living conditions. Integrating these alternative water sources into future archaeological studies will help complete the picture of how ancient civilizations sustained their rapidly growing populations.
The paper, “Settlement scaling theory, aqueducts and the Roman Empire,” was authored by J.W. Hanson, Duncan Keenan-Jones, and Davide Motta.