Andrewsarchus

Andrewsarchus mongoliensis was a large middle Eocene mammal known from one skull without a lower jaw or securely associated body bones. The reconstructed skull is about 83.4 centimetres long and directly establishes an enormous head with powerful teeth. A wolf-like torso, two-metre shoulder height, one-tonne mass and attacks on giant prey are scaling results, not fossil discoveries.

The animal was originally interpreted as a gigantic mesonychian. Later phylogenetic work often placed it closer to entelodonts within the broad even-toed mammal and whale branch. Its exact position remains unstable because limbs, pelvis, vertebrae and the lower jaw cannot be checked.

Quick facts

Kan Chuen Pao's discovery

In spring 1923 the American Museum of Natural History's Central Asiatic Expedition worked at Irdin Manha, the “Valley of Jewels”, in Inner Mongolia. The young preparator and field assistant Kan Chuen Pao found the great skull soon after camp was established. Walter Granger, who led the palaeontological work, recognised its importance, while expedition leader Roy Chapman Andrews brought it to wide public attention.

Henry Fairfield Osborn named the genus and species in 1924. The genus honours Andrews and the species refers to Mongolia. Holotype AMNH 20135 remains at the American Museum of Natural History in New York and is still the only specimen that can be assigned to the genus without dispute.

“Known from one skull” does not mean a flawless complete head. The upper and rear portions were distorted, some regions were restored and the mandible is absent. Even so, the upper tooth row and much of the snout preserve enough real anatomy to separate fact from the dramatic reconstructions built around it.

Geology, age and environment

The skull came from the lower deposits of the Irdin Manha Formation. These rocks belong to the middle Eocene, roughly 46–43 million years ago. The age of the formation is better constrained than an exact date for the individual skull, so a narrower number would imply precision the specimen does not possess.

Sediments formed on river and floodplain systems under a seasonally dry climate. Brontotheres, uintatheres, early even-toed ungulates, rodents and several carnivorous mammal lineages inhabited the wider community. The animal lived during the Palaeogene Period, but one skull cannot define its full geographic range or population density.

Rare preservation does not necessarily mean the living animal was rare. Its body bones may have failed to fossilise, may remain unrecognised or may not yet have been found. The single individual also cannot reveal sex differences, growth stages, social organisation or the duration of the lineage.

Skull and teeth

The snout was long and broad, the zygomatic arches were powerful and the back of the skull offered substantial attachment for neck muscles. Large canines, premolars and molars are preserved. Cheek teeth combine shearing crests with more rounded crushing areas.

This was neither the carnassial system of a cat nor a simple copy of a hyena's dentition. The teeth could tear flesh and break relatively hard material while remaining compatible with a broader diet. Large premolars might have crushed bone, shell or tough plant matter; choosing among them requires wear traces, fractures and the missing lower teeth.

The robust skull implies strong musculature, but bite force cannot be calculated precisely. The depth of the mandible, arrangement of lower teeth, complete jaw joint and muscle geometry are unknown. Declaring a record bite from skull length alone replaces missing measurements with assumptions.

From mesonychian to a possible entelodont relative

Osborn compared the skull with mesonychians, hoofed carnivorous or omnivorous mammals of the Palaeogene. Early artists enlarged the body of a mesonychian to match the head and produced the familiar long-legged “wolf on hooves”. At the time, that was a reasonable comparative model, not a discovered skeleton.

Later character matrices placed Andrewsarchus closer to entelodonts and other cetartiodactyls. In that model the trunk may have been deeper, the limbs more robust and the feet built on an even-toed pattern. Yet the result depends almost entirely on cranial and dental features, so the animal should not simply be called a giant pig. Entelodonts themselves were not true pigs.

New limb bones with a securely associated skull could transform the reconstruction and perhaps the family tree. Until then, the classical mesonychian and newer entelodont-like bodies are alternatives of unequal current probability, not two observed forms of Andrewsarchus.

Size without an invented mass

AMNH 20135 is genuinely one of the largest skulls known among terrestrial mammals with carnivorous or mixed diets. Osborn compared it with Mesonyx and suggested an animal about twice as large. Popular restorations later reached 3.5–4 metres in length and roughly one tonne in mass.

The critical unknown is the ratio between head and body. Entelodonts and some other ungulates carried very large skulls relative to their torsos, whereas long-legged pursuit predators followed different proportions. Without a humerus or femur there is no secure regression to select. The title “largest terrestrial carnivorous mammal” is therefore not demonstrated.

It is safe to say that the adult animal was large and had an exceptionally large head. Shoulder height, limb length and body mass can vary by several times across otherwise plausible models. A numerical range based on contradictory relatives would hide rather than solve that uncertainty.

What Andrewsarchus ate

Canines and shearing regions confirm an ability to consume flesh. Crushing surfaces allow carrion with bone and potentially more varied foods. Exclusive hypercarnivory, routine pursuit of large herbivores and specialised bone cracking have no direct support from stomach contents, prey traces or the missing lower jaw.

The animal may have taken carcasses, caught accessible prey and exploited eggs or other resources. These are mechanically possible scenarios, not a list recovered with the fossil. No stable-isotope dietary series or repeated bite record is tied to the holotype.

If its relationship to entelodonts is correct, omnivory becomes more plausible, but ancestry alone does not dictate a menu. A massive head might help process a carcass without telling us running speed, hunting frequency or whether the animal displaced other scavengers.

What the body may have looked like

Only the head and upper dentition are direct. Strong neck muscles are likely from the attachment areas at the back of the skull. Hair is expected for a terrestrial placental mammal, but length, density, colour and markings remain unknown.

Long or short legs, number of weight-bearing toes, tail form and the line of the back all depend on which relative is chosen. The most honest artwork exposes this uncertainty instead of copying a wolf, bear or entelodont in full. The surrounding Cenozoic landscapes can be reconstructed from sediment and associated fossils, but they do not provide the absent skeleton.

The distinction is especially important because the skull encourages visual exaggeration. A plausible illustration can show one tested hypothesis while its caption makes clear which parts belong to the fossil and which are comparative.

What remains unknown

AMNH 20135 cannot reveal a growth series, sex, pathology range or individual variation. It gives no direct evidence for pace, posture, social behaviour or reproduction. Even its exact geographic distribution is open because there are no second and third confirmed specimens.

The animal is not scientifically empty: the skull securely records size, tooth pattern and several relationships. Its value lies precisely in showing how much can and cannot be concluded from a spectacular but incomplete fossil. This evidence discipline is also central to the guide on how fossils form and why missing parts are not neutral gaps.

Evidence, inference and reconstruction