"Witch Swimming" is one of those pop-culture concepts that is often portrayed inaccurately and unfortunately has infected a lot of well-intentioned educators and books. Long story short, the idea that it was an automatic death sentence due to the accused being screwed either way is a myth.

My source I’m drawing on is “River Ordeal – Trial by Water” a paper by Péter Tóth G. (an ethnographer and historian in Hungary) which appeared in a publication for the Central European University.

A trial by water or "river ordeal" dates back to ancient times long before Christianity, and is mentioned in the Code of Hammurabi. It was not necessarily used for religious-based offenses, but rather for mundane crimes where the evidence wasn’t very strong. The accused was literally just thrown into a river, and if the current swept them away and they drowned, it meant they were guilty. Whereas if they were not swept away and instead made it back to shore, they were innocent. The logic was the gods knew they were innocent and would save them.

Versions of such “trials by ordeal” continued in one form or another into the Early Middle Ages where it was adapted to Christian beliefs (i.e. divine intervention), but the practice was eventually banned by the Fourth Lateran Council in 1215, because it was seen as “tempting Providence.” Contrary to popular portrayals, they were not actually a common practice during the High Middle Ages, as the “witch obsession” came later. Trial by water re-emerged in the 16th and 17th centuries with the new interest in “demons” and demonic possession. Possession was thought to make the person’s body weightless or at least very light, and thus buoyant. This aspect is believed to have come from old Germanic and Slavic folklore on “mara,” a type of malicious spirit. This was later reinterpreted as witches rejecting the sacrament of baptism, and so their bodies would resist being dunked in water. This lead to people reversing the earlier "river ordeal" outcomes: the person floating was considered evidence of guilt, and sinking, of innocence.

However, if the accused sank, they were pulled from the water before they could drown, which is why they were often lowered by a rope rather than merely being thrown in. There are several recorded case of people sinking, being pulled out and not only surviving, but being set free. It was still very dangerous, because there is always a risk of accidental drowning whether the person sank or floated (e.g. if they were face down with their hands tied).

One possible reason for the modern confusion is that throwing people accused of evil into water and intentionally letting them drown was sometimes done after their guilt had been determined by other means, and drowning was merely the method of execution.

I wanted to provide some further background to trials by water by cobbling together parts of older answers of mine.

Medieval historian Robert Bartlett has very useful information about ordeals, especially Trial by Fire and Water. Bartlett states that trials by ordeal seem to have originated in Frankish law in the 6th century, and began to be phased out after Pope Gregory banned clerical participation in such trials in the Fourth Lateran Council in 1215 (before this time, a priest was supposed to bless the hot iron in a trial by fire or bless the water - the Lateran Council wanted to discourage these practices). By 1300 trials by ordeal were rare and generally confined to remote parts of Europe.

Bartlett goes on to note that trials by ordeal tended to be replaced by trials by jury (mostly in England), or by inquisitorial trials using judicial torture as a means of last resort (in countries using Roman law).

There does seem to be a resurgence of "swimmings" for accused witches starting in southern Germany in the 1560s, and spreading to France and the British Isles (including ultimately the colonies of Virginia and New England) into the late 17th century and even beyond. But Bartlett notes that judicial authorities almost universally condemned the use of trials by ordeal for accused witches: the authors of the Malleus Maleficarium claimed that witches can influence the outcome, and later judicial scholars disliking the reliance on the supernatural (and the uncertain) for results. In many jurisdictions, the use of swimming was legally banned in this period, and took on aspects of mob rule. The use of swimmings could therefore at least be the subject of official oppobrium from superior authorities, or even possibly face penalties for violating prohibitions on the use of swimmings.

Bartlett also notes here that it seems to be mostly 19th century historians who conflated witch swimmings, which still occasionally happened at the time among more remote peasant communities, and the medieval trials by ordeal, as some sort of timeless superstition, even though they aren't necessarily connected traditions.

In the case of proper trials by water at least, drowning the person wasn't the point of the test. This seems to be something that has crept in to popular understandings of trials by ordeal as some kind of Medieval Catch-22 - "if you float, you're guilty and you die, if you sink, you're innocent but you drown."

But rather the idea of a trial by ordeal was that the court and the accused were relying on divine intervention in order to indicate the accused's guilt or innocence. The purpose wasn't to grievously harm the accused, but to create a situation where someone could be (in contemporaries' eyes) "objectively" examined for proof of guilt or innocence. The two types of trials were by fire (the accused would hold a heated metal rod, and then the burn marks would be examined - how they healed would determine guilt or innocence), or trial by water (the accused is submerged in cold water, and if they float they are guilty).

In 12th century England, for example, local Juries of twelve would indict individuals accused of robbery, murder or theft, or harboring people who committed these acts. Sheriffs would then seize the accused and bring them before a court. If the individuals were caught in the act (or with stolen goods), and were of bad repute, then their guilt was considered obvious. If not, they could undergo trial by water. If found guilty by this means, they would lose a foot and/or a hand. If they passed, but were still of bad repute, they would need to go into exile for life.

Bartlett examined the Pipe Rolls (English Exchequer rolls kept from the 12th through early 19th centuries), plea rolls, and Ranulf de Glanvill's The Treatise on the Laws and Customs of the Kingdom of England. Trials by fire tended to be reserved for women (for reasons of decorum) and free men, while male serfs tended to undergo trials by water. Breakdowns are roughly 17% trial by fire and 83% trial by water, with something like two thirds of those undergoing trial by water being acquitted.

As Barlett writes:

Since the majority of ordeals were of cold water, this statistic can be explained either by the way the human body behaves in water, or, more circuitously, by the way courts decided what it was to sink or to float. To place the case before God was also to put it into the hands of a small body of men with opinions and feelings of their own.

The main takeaways are: trials by ordeal were *not* originally applied to witchcraft (trials of witches were not common in this period; they are more a circumstance of the late Medieval and Early Modern period), but to criminal charges, and that the punishment for guilt was not usually death. On top of that, the acquittal rate from trials by water actually seems to be much better than even chances (and much higher than modern day trial acquittal rates, for that matter).