What the Romans Did to Prisoners Who Would Not Talk

A man is dragged into a Roman courtroom and everyone present already knows he will be tortured. He is not there because he is guilty of anything and he is not there because anyone has actually accused him of an offense. He is going to be tortured because of what he is and because the law says his words are worthless unless he produces them in pain.

He may be entirely innocent and he may know nothing whatsoever about the matter under investigation and neither of those facts makes the slightest difference to what is about to happen because the procedure is not designed to distinguish between the two. The procedure is the point and the procedure has no interest whatsoever in which of those two men is standing in front of it.

This was not a mob and it was not an emperor’s whim. It was written law administered by trained magistrates in a system that produced the legal thinking most of Europe still runs on. Rome had a formal, regulated, bureaucratic method for extracting information from people who would not speak set out in statute administered by officials who had been trained in it and applied continuously for something in the region of a thousand years.

So, how did the most sophisticated legal system in the ancient world end up building interrogation into its own rules of evidence? There were six parts to it and five of them follow a cold logic you will recognize even if you hate it. Who could be questioned? Who could not be touched? What happened when a master was killed? How a statement was checked? And what time in a cell was for? The sixth is the one I still find genuinely strange.

Rome’s most respected legal authority, a jurist whose opinions became the backbone of the entire Roman law code, wrote that this method was fragile, that it was dangerous, and that the truth was very often not obtained by it. That warning was not a pamphlet from a critic. It was placed inside the law itself, where every magistrate could read it.

And the practice went on for another thousand years afterward, across the fall of the empire and deep into medieval Europe. Start with the rule that shapes everything else, because without it none of the rest makes sense, and the rule is this.

In Roman law, the testimony of an enslaved person was not admissible, not weak, not doubtful, not treated with caution, inadmissible. A slave could stand in front of a magistrate and describe exactly what he had seen, in detail, accurately, and it counted for nothing at all in a legal proceeding.

Now, hold that against the reality of a Roman household. The people who knew everything about a Roman family were the enslaved staff. They served at the dinners where business was discussed and alliances were arranged. They carried the letters that went between houses, and they were present in bedrooms and offices and at private meetings, treated as furniture by people who spoke completely freely in front of them, and never once considered that anything was being overheard.

In any investigation into a wealthy Roman, the entire useful body of witnesses consisted of people whose word the law refused to hear. In any investigation touching a wealthy household, the entire useful body of witnesses therefore consisted of people whose word the law refused to hear at all.

So, Roman law built a mechanism to convert one category into the other. A statement made by an enslaved person under formal interrogation conducted by an authorized official under the supervision of a magistrate became admissible evidence. The same words from the same mouth about the same events. What changed in that process was not the reliability of the account, nor the honesty of the person giving it, nor anything at all about the events being described.

What changed was the legal category the account belonged to, and that was the entire function of the exercise. There was a further rule sitting on top of that one, and it is stranger still because an enslaved person could not, as a general principle, be compelled to give evidence against his own master. The person with the most complete knowledge of a household was specifically excluded from testifying about the one individual he knew best.

That looks like a protection for the enslaved until you examine who it actually served because it was not designed to spare anyone. It protected the owner, not the owned. It shielded a master from having his own property turned into a weapon against him. And it treated the enslaved as an extension of the household rather than as an independent witness with anything of his own to say.

The exceptions to that rule are revealing because they were carved out precisely where the state’s interest outweighed the master’s. Treason against the emperor, certain religious offenses, adultery, that was the list. In each of those, the protection dissolved, which tells you exactly what it had been protecting all along.

That distinction is the single most important thing, and hold on to it because it explains something at the end that otherwise looks insane. If the enslaved could always be questioned, the obvious next question is who could not be? And here, Rome drew a line that was, for a long time, taken extremely seriously.

A Roman citizen could not be subjected to this. That protection went back centuries before the empire to laws of the Republic, and it was one of the most concrete privileges attached to citizenship. Cicero built an entire prosecution around it, arguing to a Roman jury that to bind a Roman citizen was a crime, to flog one was an atrocity, and to kill one was close to killing your own father.

That argument worked because the men on that jury already believed it. Citizenship was not a piece of paper. It was a physical shield around a body guaranteed by a chain of laws going back to the early Republic and reinforced repeatedly across the following centuries, each new statute adding penalties for the magistrate who ignored it.

There is a well-known moment that shows exactly how the shield worked in practice. A man is stretched out and about to be flogged by a Roman officer in Jerusalem, and he asks a simple question:

“Is it lawful to do this to a Roman citizen who has not been tried?”

The officer stops immediately. He goes to his commander.

The commander comes back alarmed and asks whether the claim is true and admits that he himself paid a great deal of money for his own citizenship. The man replies:

“I was born with it.”

Everything stops the moment the claim is made before anyone has produced a shred of proof that it is true. Nobody holding that status needed to argue his case or produce a document because the officer standing there with the whip understood perfectly well what a wrongful flogging of a citizen would cost him personally in career terms, and the risk was entirely his. The shield was real, and it worked precisely because the penalty for ignoring it fell on the official who did the ignoring. It was also, and this matters enormously, a shield of status rather than a shield of principle because nothing anywhere in it said that this was a wrong thing to do to a human being.

It said it was unlawful to do to a particular category of human being, and everyone outside that category remained available. That framework held as long as ordinary investigations stayed ordinary. Then there is a specific situation where Roman law abandoned all restraint and it produced one of the most disturbing episodes recorded from the entire period.

Under a decree of the Senate, if a master was murdered in his own home, every enslaved person who had been under that roof could be interrogated. All of them. And if the killer was not identified, all of them could be executed regardless of individual guilt on the reasoning that any household containing a murderer had collectively failed in its duty to protect the master.

In the year 61, the city prefect of Rome, a senior magistrate named Pedanius Secundus, who held responsibility for order in the capital itself, was killed inside his own house by a member of his household. That household numbered around 400 people, which was not unusual for a man of his rank in a city where the size of a retinue was a public statement about status.

Men, women, children, the elderly. People who had never been near him. Under the law, every one of them was to be put to death. What happened next is the part worth knowing. Rome objected loudly and physically in a way that the ancient sources rarely record about anything happening to enslaved people. Ordinary Romans filled the streets around the Senate house and the crowd was large enough and angry enough that the sentence could not simply be carried out quietly, which forced the Senate into an open debate about whether to enforce its own law.

The argument that won was not that the punishment was just. It was that abandoning the rule would leave every Roman master unsafe in his own home. Troops lined the route. The sentence was carried out in full. The historian who recorded it writing decades later does not hide his own discomfort. That episode tells you something no legal text can.

The system was not unaware of what it was doing. It knew exactly. Exactly. It was argued about openly and it proceeded anyway because the alternative threatened the structure of Roman property. So Rome had a legally sanctioned method and it applied that method at scale, which is where the story becomes considerably more interesting than most people expect.

Roman magistrates did not simply accept whatever came out of an interrogation. There were procedural rules designed to test it. And those rules exist because everyone involved understood that a person under enough pressure will say anything. Statements were expected to be checked against physical facts that could be independently verified.

Investigators wanted to know where the money had gone, whether the letter existed and could be produced, and whether the sequence of events described could actually have happened in the time available. A story has to hold. A statement that could not survive contact with the rest of the evidence carried very little weight with an experienced magistrate.

And the surviving legal commentary is quite clear that a magistrate who convicted on an uncorroborated account had done his job badly. Questioning was often repeated and separated in time on the reasoning that a fabricated account is harder to reproduce identically than a remembered one. Multiple people from the same household were questioned separately rather than together, and their accounts were then compared against one another for the specific inconsistencies that reveal a fabrication, which any modern investigator would recognize. And a confession standing on its own was generally not treated as sufficient to close a case, which is a remarkable position for a legal system to hold given the trouble it had gone to in order to obtain that confession. It had to be supported by something external before a magistrate would rely on it.

If any of this sounds familiar, it should. These are recognizably the concerns of modern investigation, and Roman jurists arrived at them by exactly the route you would expect, which is by watching the method fail repeatedly and writing down what went wrong. We have one case where a Roman governor describes the process in his own words in a letter that survives.

Around the year 112, Pliny the Younger was governing a province and trying to work out what a new religious group was actually doing because the rumors circulating about them were lurid, and he could not confirm any of it. He followed the procedure exactly as the law prescribed. He identified two enslaved women who held a position within the group, and he had them formally interrogated in order to obtain admissible testimony about what really happened at their meetings.

His conclusion, written to the Emperor Trajan, is almost comic in its flatness. He found nothing. Nothing beyond what he called a debased and extravagant superstition. The most powerful legal instrument available to a Roman governor, applied correctly by a conscientious administrator, produced no useful information whatsoever.

And he wrote to his emperor to say so and to ask for guidance. Which raises an obvious question that Rome itself asked, and we will get to the answer. Before that, there is a fifth element, and it is the one that leaves the least trace in the records, precisely because it left no marks at all. Time.

Roman imprisonment was not a punishment in the modern sense. Rome did not sentence people to years in a cell, because the empire had no interest in feeding and guarding people for decades. Prison was a holding place, rather than a sentence, used for those awaiting trial, awaiting execution, or awaiting a decision from somebody important enough to be slow about making it. But holding a person in a cell is itself a form of pressure, and everyone administering the system knew it.

Conditions in the old state prison at the foot of the Capitoline were, according to every surviving description we have of the place, genuinely appalling, even by the standards of a society that was not squeamish. It sat underground, without light or sanitation of any kind, cold and cramped, and reached only through a hole cut in the floor of the chamber above it.

A person left there for weeks with no information about what would happen next, no contact with family, and no idea whether an interrogation was scheduled for tomorrow or never, is being worked on continuously without anyone laying a hand on them. Isolation does work. So does uncertainty.

So does time, entirely on its own, and they leave no evidence behind that anyone did anything at all. There is one detail here worth pausing on. The Roman state generally did not feed prisoners adequately and expected families to bring food, which meant that a prisoner with money, status, and relatives on the outside had a very different experience from a prisoner without them in the same building on the same charge.

A prisoner with money, relatives, and standing therefore ate, and a prisoner without them did not in the same building on the same charge. Status followed you all the way down into the dark.

And now the sixth thing, which is the one that has never stopped bothering me. Return to the warning mentioned at the start. The jurist was Ulpian, and he is not a marginal figure. He was one of the most influential legal minds Rome ever produced, serving at the very top of imperial administration in the early 3rd century.

And his written opinions were later queried in enormous quantity for the great compilation of Roman law assembled under the Emperor Justinian in the 6th century. That compilation is why we have any of this. All of it. It preserved Roman legal thinking through the collapse of the Western Empire and delivered it intact and organized by subject to the European scholars who rediscovered it 700 years later.

The section dealing with interrogation is a discrete title within it, and Ulpian’s assessment sits near the opening of that title, which means it was among the first things any student of the subject would read. In the section of that compilation dealing with interrogation, his assessment is preserved in plain language.

The method is fragile. It is dangerous. It frequently deceives. Some people are tough enough to endure it and reveal nothing, and others are so unable to endure it that they will say absolutely anything to make it stop, which means the outcome depends on the constitution of the person rather than on what they know.

He is describing the mechanism of false confession accurately and in plain terms in the early 3rd century, working purely from what he had seen happen in courtrooms across a long administrative career. And then Roman law kept the procedure anyway, unchanged in its essentials. So, why? Here is where the rule from the very beginning comes back.

The purpose of the procedure was never primarily to discover truth. It was to produce admissible evidence. In a system where the statement of an enslaved person had no legal standing whatsoever, this was the only mechanism that converted an unusable account into a usable one. It was a legal gateway, not an investigative technique.

The two look identical from a distance because they used the same room and the same officials. Once you see it that way, Ulpian’s warning stops being a contradiction. He is not saying the system should be abolished. He is telling magistrates that the thing they are legally required to do produces unreliable material, so they had better check it against everything else.

Which is exactly what the fourth part described. Put the six together and look at what Rome actually built. A rule that made most witnesses legally silent. A procedure that converted their silence into admissible testimony. A shield around citizens that was really a shield around status. A collective punishment rule that everyone knew was monstrous and kept anyway.

A set of corroboration requirements that show the system did not trust its own output. And a written warning at the center of the law code admitting the whole thing was unreliable. It is not the work of people who had never thought about the problem. It is the work of people who had thought about it extremely carefully and built the machine anyway.

Because it served a purpose that had nothing to do with finding out what happened. But the entire arrangement rested on one assumption. And it was an assumption nobody thought to defend because it seemed permanent. That the line between citizen and non-citizen would hold. It did not, and the way it collapsed is one of the great ironies in legal history.

The first cracks came from treason. In cases touching the safety of the emperor, the citizen protection was set aside on the argument that no privilege can shield a man plotting against the state itself. That exception looks narrow written down. In practice, it was elastic because the definition of an offense against the emperor could be stretched by any prosecutor with ambition.

And under several emperors, it was stretched a very long way indeed. Rome had no public prosecutor in the modern sense, so criminal cases were brought by private individuals who stood to gain materially from a conviction, frequently taking a share of the condemned man’s confiscated estate. Loose charge, cash reward.

That machine finds treason wherever there is money. Tacitus describes the result in detail across the reign of Tiberius, and what emerges is not a state hunting genuine conspirators, but a profession of professional accusers working a system that paid them. Then came the change that finished it, and it arrived disguised as the most generous act in Roman history.

In the year 212, the emperor Caracalla granted Roman citizenship to virtually every free inhabitant of the empire. Millions of people in a single edict received the status that Cicero had described as a shield around the body, and the protection immediately became worthless. A privilege that belongs to a minority is enforceable because the state can afford to exempt a small group without losing the ability to prosecute anyone.

A privilege that belongs to everyone protects nobody at all because no state can operate a legal system in which not a single person may ever be compelled to answer a question. So, the line did not disappear when the old category dissolved. It was simply redrawn somewhere else along a boundary that had always existed socially and now acquired a formal legal name.

Roman jurists of the following century stopped writing about citizens and non-citizens and began writing about the more honorable and the more humble, a division based on wealth, rank, and office, rather than on legal status. The upper category kept the old protections more or less intact. The lower category, which took in the overwhelming majority of the population of the empire, including a great many people who had just been made citizens, lost those protections entirely within a couple of generations of receiving them.

And that is the real ending of this story. The protection was never about what may be done to a human being. It was about which humans were exempt, and when the exemptions were extended to everyone, the category was simply rebuilt on a different foundation within a generation. Nothing about the underlying principle was reformed at any point in this process.

The line simply moved, and it moved wherever it had to so that the machinery kept running against the same people as before. What survived was the machinery, and it survived far longer than Rome did. When the Roman law code was rediscovered and studied in Europe from the 11th century onward, beginning with the scholars at Bologna, who worked line by line through the recovered text, the interrogation procedure came with it.

And it entered continental legal practice as a formal part of proving a case, complete with the same rules about corroboration and the same requirement for supporting evidence. There is a reason it took hold so quickly. And it is the same structural reason that produced it in Rome. Medieval courts had abandoned trial by ordeal and replaced it with a strict standard of proof requiring either two eyewitnesses to the act itself or a confession from the accused.

Two eyewitnesses to a crime committed behind a closed door almost never exist for the obvious reason that nobody commits one in front of an audience. So, confession became the only route. In a very large proportion of serious cases, and once a legal system makes confession structurally necessary, it will find a way to obtain confessions.

The procedure did not return because Europe became crueler. It returned because Europe adopted a rule of evidence that could not function without it. Ulpian’s warning came, too. It sat in the law books that European jurists were studying in plain language saying that the method frequently deceives.

That warning was quoted by opponents of the practice for the better part of 600 years, appearing again and again in the arguments of jurists and reformers who had noticed that the most authoritative source in their own legal tradition agreed with them right up until the wave of reforms that finally removed the procedure from European law across the 1700s.

And the modern research is unambiguous about who was right. Decades of research into interrogation and false confession have confirmed, in considerable detail and across many jurisdictions, the thing a Roman lawyer wrote down 1800 years ago from observation alone, which is that extreme pressure produces statements shaped by a person’s capacity to endure it, rather than by anything they actually know.

It is the reason coerced confessions are inadmissible in courts today. Rome worked out that the method did not reliably produce truth, wrote it into the law, and then used the method for another thousand years because truth had never been the reason it was there.

They understood the tool was broken. They kept it because breaking it was never the point.