What actually kills after a Komodo dragon bites. The answer that took three decades to find.

The old story went like this. On the Indonesian island of Komodo, a dragon bites a water buffalo once and the buffalo gets away. The animal outweighs the lizard several times over, but the dragon does not need to kill it outright. Days later the buffalo weakens, goes down, and the dragon closes in, often with others. For most of the last century, biologists had a tidy explanation for that delayed death. They had the central part of it wrong.

The explanation came from Walter Auffenberg, an American biologist who spent a year in the 1970s living among the dragons and produced the first close study of how they hunt. He proposed that delayed killing was done by bacteria. A Komodo dragon's mouth, the thinking went, teemed with virulent microbes picked up from carrion and standing water. One bite seeded a wound with infection, and sepsis did the rest. The idea was vivid, it matched what he saw in the field, and it lasted. It became one of those facts everyone knew. A 2002 study that cultured genuinely nasty bacteria from dragon saliva seemed to settle it for good.

Then in 2009 a team led by the venom researcher Bryan Fry used MRI imaging and dissection to find something the bacteria story had no room for. Venom glands. They sat in the lower jaw, built from six compartments on each side, with ducts opening between the teeth. When the team analyzed what the glands produced, the proteins did two things that matter a great deal to a bleeding animal. They kept the blood from clotting, and they drove blood pressure down toward shock.

A Komodo dragon does not inject venom through fangs the way a viper does. Its teeth are curved and serrated, closer to a steak knife than a syringe. It bites deep and pulls, opening a ragged wound, and the venom works its way in. Later work on bite and pulling forces found it poorly built for a crushing bite and well built for exactly this pulling, slicing one. A deep wound, blood that will not clot, and falling pressure. In the revised account, the critical damage is not a cultivated infection. It is bleeding and shock, made worse by chemistry the bite delivers.

The bacteria, meanwhile, got a second look. In 2013 researchers cultured saliva and gingiva from sixteen captive dragons and found no evidence of a specialized septic flora. The pathogens earlier work had turned up looked less like a maintained biological weapon than ordinary passengers from food, water, and environment. The dragon's mouth, it turned out, is about as dangerous as its last meal.

Here is the part most retellings leave out. The venom is real and the glands are documented, and the dirty-mouth model has not survived. But scientists still disagree about how much the venom actually does in a real kill. Some, including the evolutionary biologist Kurt Schwenk, argue that a wound that deep, from an animal that size, would drop prey through blood loss and shock on its own, toxins or not. Calling the dragon "venomous," they say, oversells the chemistry's part in the field. Fry's group counters that the toxins measurably worsen the bleeding the bite begins. Both sides can point to evidence. What no longer has evidence behind it is the bacteria-as-venom story.