In 1901, working as an assistant in the University of Vienna's Department of Pathological Anatomy, Karl Landsteiner ran an experiment nobody had thought worth trying: he mixed blood serum from one healthy colleague with red blood cells from another and watched what happened. Sometimes nothing happened. Sometimes the cells clumped together into visible masses — agglutination — exactly as they did when human blood met an animal's. Sorting the results, Landsteiner found that human blood fell into four distinct types, later named O, A, B, and AB. Until then, physicians performing transfusions had treated one person's blood as interchangeable with another's.
That assumption had been killing people. Blood transfusion had been attempted since a scientist named Lower first transfused blood between two dogs in England in 1666; results over the following two centuries ranged from lifesaving to lethal, and the practice was largely abandoned as too often fatal. Landsteiner later recalled that doctors had simply regarded it as obvious that one person's blood serum was harmless to another person's cells — a belief nobody had tested, because it had never occurred to anyone that blood might come in types at all. His discovery showed the belief was false, and that the wrong pairing could trigger the same violent reaction that made animal blood dangerous.
The fix was not new medicine; it was a check. Mix a drop of the donor's blood with the recipient's serum before the transfusion, and the four-letter answer tells you whether it is safe. Adoption was slow — Landsteiner noted it took the mass casualties of the First World War to make cross-matching routine practice — but by the time he described it in his 1930 Nobel lecture, the numbers argued for themselves. At Kiel, 2,300 transfusions over roughly five years produced not one fatal accident. Where the check was skipped or botched, patients died from it directly: one of two deaths among 1,467 transfusions at Bellevue Hospital in New York was traced to an incorrect blood-group determination, and three of the deaths among 1,036 cases at the Mayo Clinic came from the same error.
The difference between those hospitals was never the blood itself. It was whether someone had actually checked whether two specific bloods matched, instead of assuming they would. That is the pattern behind every place truth is load-bearing: the danger is rarely the substance in the bag. It is treating an unverified guess about it as a settled fact.
A blood type is not an opinion. It is a fact with four possible answers, and for two and a half centuries doctors transfused patients without asking the question. Once they asked it, the dying mostly stopped.