The man-eating lions of Tsavo, the number that was four times too high, and the broken tooth that kept the receipts. · published 27 September 2026
Isotope and tooth-wear work on the preserved Tsavo lions shows the legend is wrong three times over: the death toll was roughly 35 rather than the 135 Patterson reported, only one of the two lions did most of the killing while the other ate mainly antelope, and their dental microwear resembles zoo lions eating soft food rather than starving animals cracking bone — suggesting people were the easiest option, not the last one.
These are the working notes, unedited. Everything the video
states as fact comes from here, and nothing else may be stated as fact.
The sections on limits and on what the finding does not show are
part of the brief, not an afterthought — see
the method.
The story. In 1898, during construction of a railway bridge over the Tsavo River in Kenya, two maneless male lions killed workers over roughly nine months. Lieutenant Colonel John Henry Patterson eventually shot both. He wrote a book, the story became famous, and it has been filmed. Patterson put the number of people killed at 135. The lions’ skins and skulls went to the Field Museum in Chicago, where they still are - which is why there is anything left to test.
The first correction: the count. Yeakel and colleagues, “Cooperation and individuality among man-eating lions”, PNAS, 2009. They measured stable isotopes in the lions’ bone collagen and hair - what an animal eats is written into the tissue it grows - and worked back to how much of the diet was human. Interpolated across the nine months of man-eating, the estimate came to about 10.5 humans for one lion and 24.2 for the other, roughly 35 in total, with a wide range of 4 to 72. The Uganda Railway’s own documented figure was 28. Patterson’s 135 is about four times the median estimate.
The second correction, which is stranger: they were not a pair of equals. In the final months, about half of one lion’s diet was human. The other ate very few people, living mostly on herbivores. “The man-eaters of Tsavo” were, in practice, one man-eater and one lion that mostly kept eating antelope.
The third correction: why. The assumption was desperation - injured or starving animals driven to easy prey. The teeth say otherwise. One lion had a broken canine with a root-tip abscess and three missing incisors, an injury that would have made a struggle with a buffalo agonising. But dental microwear analysis by Larisa DeSantis and colleagues found wear patterns resembling zoo lions that eat soft food - not the heavy bone-cracking of an animal consuming every scrap in desperation. Her conclusion was that people were not a last resort but the easiest available solution.
The part that is almost too neat. In 2024, “Compacted hair in broken teeth reveals dietary prey of historic lions” appeared in Current Biology. Researchers took hair that had packed into the cavities of those broken teeth and sequenced mitochondrial DNA from individual shafts. The prey came back as giraffe, human, oryx, waterbuck, wildebeest and zebra - including at least two Masai giraffe. The dental injury that may have pushed the lions toward people is the same injury that preserved the record of what they ate, for 126 years.
Limits
The isotope estimate has a range of 4 to 72, which is not precision; the median is the headline and the uncertainty is real. The motive is inference from tooth wear, not observation. And Patterson had every incentive to inflate a number he alone was in a position to report - which is an explanation, not a proven accusation.
What this does not show
Not that the killings did not happen. People died, the railway work stopped, and the fear was not invented. Not that Patterson was a fraud. The narrower claim is that a figure reported by a single interested party, repeated for a century, turns out on measurement to be about four times too large - and that the more interesting facts, that the two lions behaved completely differently and that neither was starving, were not in the story at all.