In 1898, two male lions attacked people around the Uganda Railway construction camps at Tsavo in Kenya. Their story became a famous hunting narrative and an enduring problem for historical and biological research. The animals themselves survive as museum specimens. Researchers have used their damaged teeth, microscopic tooth wear, tissue chemistry and trapped hairs to ask what they ate and why they changed prey. These methods illuminate different parts of the story; none supplies a complete list of the people killed.
Patterson’s account and its viewpoint
Engineer John Henry Patterson’s 1907 memoir describes scattered camps along the railway, repeated night attacks and efforts to protect workers with thorn enclosures, traps and raised shelters. His account places a major work stoppage at the beginning of December 1898. He eventually killed the first lion on 9 December and the second on 29 December.
Patterson also attributed supernatural interpretations to the workmen, including a belief that the lions embodied hostile spirits. This is his retrospective account of other people’s beliefs, written in the language and hierarchy of a colonial hunting memoir. It should not be mistaken for separately recorded testimony from each worker. The attacks themselves were attacks by lions, not evidence establishing the spirit explanation.
At the end of the lion chapters, Patterson gave twenty-eight Indian labourers and an unspecified additional number of African victims. Reading his actual wording matters: the familiar precise total of 135 belongs to the story’s later development. The original memoir is a participant’s source, but its confident storytelling does not remove the need to compare it with records and specimens. Read Patterson’s original account.
The 2001 reconstruction
Julian Kerbis Peterhans and Thomas Patrick Gnoske reviewed historical literature, Patterson’s original journals, game records and the museum specimens in a 2001 Journal of East African Natural History paper. They traced how the recorded and claimed totals changed: Patterson’s journals discussed fourteen individuals, the 1907 book referred to twenty-eight Indian workers plus other Africans, and his 1925 account specified 107 Africans to produce the total of 135. The researchers found no published official tally supporting that larger number.
They also examined the animals as individuals. One skull showed substantial remodelling associated with a badly broken canine. Their interpretation combined such injury with environmental circumstances, including the effects of rinderpest on potential prey and dense thorn scrub favourable to ambush. The paper also documented reports of attacks in the wider area before and after the famous pair.
Its argument was consequently more complex than a single exceptional pair appearing in an otherwise safe landscape. It also noted counterexamples: injured lions were not necessarily human predators, and healthy animals could attack people. A broken tooth was a possible contributing condition, not a universal rule. Read the original 2001 reconstruction.
What the isotope estimate measures
Justin Yeakel and colleagues’ 2009 PNAS paper analysed carbon and nitrogen isotope ratios in the lions’ tissues and compared them with potential prey sources. Bone collagen represented a relatively long dietary history; tail hair provided a shorter window near the end of life. Their model suggested that the two lions maintained different diets, with one relying more heavily on humans during its final months.
The widely repeated estimate of about thirty-five people came from a calculation, not a newly discovered death register. The authors assumed an average food requirement of roughly six kilograms per day, an amount of edible tissue consumed per person, and a time window represented by the hair. Interpolating the results across nine months produced a median estimate of 34.7 people consumed, with a broad possible range of 4–72 under the stated model.
Those assumptions and uncertainties belong beside the headline number. Amount eaten and number killed are different quantities; a lion need not consume an entire victim. The model also found substantial non-human prey in the diets. Cooperation between the pair did not mean that they ate identical proportions of the same foods. Read the original isotope study and its assumptions.
Microscopic wear and damaged teeth
Larisa DeSantis and Bruce Patterson investigated dental microwear textures in a 2017 Scientific Reports paper. They compared the Tsavo specimens, another historical human-eating lion from Zambia, wild lions, captive lions and other carnivores. Their question concerned the mechanical properties of foods consumed shortly before death, particularly whether the lions had been heavily processing bone.
The microscopic textures did not support extreme bone crushing. They were compatible with softer foods and incomplete consumption of carcasses during the recent dietary interval represented by tooth wear. This challenged a simple reading of the bone-crunching sounds described in Patterson’s memoir, without establishing exactly what produced every sound he reported.
The researchers considered severe dental disease a plausible reason for selecting more easily subdued prey. They distinguished serious disease from ordinary tooth damage, which is common in lions and does not automatically cause attacks on people. They also considered recovery of human bodies as one possible limit on the animals’ access to bones. Microwear therefore tests a particular feeding hypothesis; it cannot directly reveal an animal’s intentions or reconstruct all nine months of behaviour. Read the 2017 analysis.
Hair preserved inside the teeth
A 2024 Current Biology study by Alida de Flamingh and colleagues used DNA from hairs compacted inside the lions’ broken teeth. The published findings identified human, giraffe, oryx, waterbuck, wildebeest and zebra material, as well as lion hair. The giraffe evidence included at least two individuals belonging to the Masai giraffe subspecies associated with south-eastern Kenya.
The authors’ accompanying data documentation describes four individual hairs and three hair clumps used for sequencing, with comparisons against reference sequences. An individual hair and a mixed clump are different sampling units; detecting a species is not a measure of its total contribution to the lions’ diet. Lion hair may reflect grooming rather than an additional prey animal.
These results strengthen the picture of dietary variety and demonstrate what museum material can preserve. They do not count every human victim or show that the lions had abandoned ordinary wildlife. This account uses the accessible published summary and authors’ methods documentation; the full publisher-hosted article was unavailable during checking. See the 2024 paper and the authors’ accompanying documentation.
The specimens and the remaining questions
The Field Museum’s own account records that Patterson sold the lions’ hides to the museum in 1924 after using them as rugs. Taxidermists restored them as the mounts shown in the photograph above. Their present poses are museum reconstructions, while the preserved tissues and skulls have continued to support research. The museum’s research account explicitly distinguishes estimates of people eaten from the number killed. Read the museum’s collection and research history.
Tsavo remains a case in which an established tragedy has generated questions that can be investigated rather than simply retold. Historical records address particular people and events; isotope modelling estimates dietary proportions; microwear addresses recent food texture; hair DNA identifies sampled species. Agreement between methods strengthens some conclusions, especially that these lions consumed both humans and other animals. Differences in what each method measures explain why a single confident total or a single-cause explanation remains harder to defend. The case belongs to Kenya, where the attacks occurred, even though the specimens were later taken to Chicago.
Map location: Approximate area: Tsavo locality reference, rounded from a Harvard Museum of Comparative Zoology locality record georeferenced with a stated 2,553-metre uncertainty radius. This represents the railway-camp area, not a particular attack spot; the camps were scattered, and the museum specimens are now in Chicago. Coordinate reference.