Sabre-Toothed Cat: Weak Bite, Sabres at Risk of Breaking

Biology lens: built to hold

Forelimbs like a wrestler. Julie Meachen-Samuels and Blaire Van Valkenburgh measured the bones of Smilodon, 28 living cat species and the even larger American lion Panthera atrox on radiographs. The upper arm bones of Smilodon were reinforced by thicker bone walls than in any living cat or in the American lion; the thigh bone was thickened too, but within the range of other cats. The authors interpret this as a greater ability to subdue prey with the forelimbs – to keep the canines from breaking and to position the bite for a quick kill 1.

What injuries reveal. Caitlin Brown and colleagues studied over 1,900 diseased or injured bones from Rancho La Brea. The frequency of traumatic injuries was higher in Smilodon (4.3 %) than in the dire wolf (2.8 %), which implies that its way of killing carried a greater risk of injury. Smilodon, which the authors classify as an ambush predator, was injured more often than expected in the lumbar vertebrae and the shoulders; the dire wolf, a pursuit predator, more often in the limbs and the neck vertebrae 8.

Alone or in groups? Chris Carbone and colleagues compared the fossil counts at La Brea with predators that come to recordings of distressed prey in Africa. In both, two species dominate – the social lions and spotted hyenas in Africa, the presumably social dire wolf and Smilodon at La Brea – and the authors conclude that Smilodon was social 5. Christian Kiffner questioned the comparison: in the asphalt seeps, heavier animals were more likely to get stuck, the lure lasted longer than a playback, and the Ice Age predators of North America need not have behaved like those of an African savannah 9.

A family in Ecuador. Coastal deposits in Ecuador preserved at least two young Smilodon fatalis and one adult that probably died together. Both young have a rare extra tooth, a third premolar in the lower jaw, which, as Reynolds and colleagues report from earlier studies, occurs in only 2.4–6 % of Smilodon at Rancho La Brea. This suggests inheritance – a rare sign of relatedness in the fossil record. The young were at least two years old and were found with an adult that was probably their mother, which indicates prolonged parental care. According to Ashley Reynolds and colleagues, Smilodon may have grown about as fast as a tiger, but for as long as a lion 3.

Physics lens: a weak bite, a strong neck

Bite force. Colin McHenry and colleagues built one of the most detailed computer models of a vertebrate skull and simulated how Smilodon bit. The bite force produced by its jaw muscles was relatively weak – about one third of that of a lion of comparable size – and the skull withstood forces from struggling prey poorly. It was better suited to bites on restrained prey, in which the neck muscles add to the force of the bite. The authors conclude that prey was brought to the ground and held before the killing bite, which was most likely directed at the neck 7. Comparing many living and fossil predators, Stephen Wroe and colleagues likewise estimated a relatively low bite force for its body mass 10.

A stiff skull for a stab. Borja Figueirido and colleagues compared the amount of compact and spongy bone in the skulls of Smilodon and the scimitar-toothed cat Homotherium. Smilodon had an extremely thick skull with relatively little spongy bone, consistent with a large investment in cranial strength for a stabbing “canine-shear” bite. Homotherium had more spongy bone and probably killed more like a lion, with a clamp-and-hold bite 11.

How wide, how hard? From skull measurements that stand in for the size and leverage of the jaw muscles, Ashley Deutsch and colleagues estimate a bite force of about 1,280 N at the canine and about 4,670 N at the carnassial, the shearing cheek tooth. That is similar in magnitude to estimates for the much smaller jaguar – slightly lower at the canine, higher at the carnassial – not to those for the largest living cats. The angle to which Smilodon could open its jaw was significantly larger than in living cats, but the opening measured at the canine tips (linear gape) was not. The authors note that their estimates fit the view that muscles of the neck and body helped to power a canine-shear bite 12.

An alternative model. Not everyone accepts the neck-powered bite. In a study with a digitised replica of the skull and neck, Jeffrey Brown found that rotating the skull with the neck muscles does not close the jaws. He proposes instead that the lower jaw was braced against the neck of the prey while the forelimbs rotated the skull forward, like a lever 13.

Many sabres, many solutions. Sabre-like canines evolved several times in different groups of vertebrates over the last 265 million years. Simulations of gape angle, bending strength and bite force by Stephan Lautenschlager and colleagues show a remarkably high functional diversity: the lineages differed in function, and different ways of killing evolved even within the same lineage 14.

Mathematics lens: the sabre as a beam

A tooth is a cantilever. A canine is fixed in the jaw at one end and free at the other – a cantilever beam. Blaire Van Valkenburgh and Christopher Ruff modelled canines in this way, with an elliptical cross-section, and found that the canines of sabre-toothed cats resemble those of the living dog family (canids) more than those of living cats in shape and strength 15. Tseng used beam theory as well, together with finite element simulations, to follow the sabre of Smilodon through its eruption 6.

A force F at the tip of a tooth of height h bends it with the moment M = F · h. In this simple model, the largest stress occurs at the base, on the surface farthest from the bending axis:

Here W is the section modulus, which describes how well the cross-section resists bending. For an ellipse with semi-axis a front to back and b side to side, standard beam theory gives (a calculation for this article):

Tseng writes the same quantities with the full length x and width y of the cross-section (I = π · x · y³ / 4); the constant differs, but the ratio of the two stresses does not 6.

The weak direction. The ratio of the two stresses for the same force is simply a/b: a tooth twice as long as it is wide is stressed twice as much from the side as from the front. In Tseng’s digital eruption model of specimen AMNH FM55576, the cross-section of the fully erupted canine at the gum line is 32.17 mm by 18.00 mm (Table S1) 6 – so a sideways load stresses it about 1.8 times as much as a load from the front. This ratio alone does not show that the sabre was fragile: every flattened tooth is weaker from the side. The comparison with living cats comes from Van Valkenburgh and Ruff: cats have relatively stronger canines than dogs, particularly against bending about the front-to-back axis (in our reading of the abstract: by a load from the side), while the canines of sabre-toothed cats resemble those of dogs 15. Meachen-Samuels and Van Valkenburgh likewise describe the long sabres as more vulnerable to fracture than the canines of living cats 1.

Growing into strength. Tseng found that the cross-section of the sabre grows with its height in a way that increases its bending strength, although its sideways deflection under a load increases as it grows. According to Wysocki and colleagues, the milk canine and the erupting sabre were present together for about 11 of the roughly 22 months the sabre took to erupt – about half of the eruption period 16, which Tseng describes as more than half. Standing side by side, the two canines effectively widened the tooth, which brought Tseng’s estimates of strength and stiffness closer to theoretical optima 6.

Chemistry lens: what teeth remember

A clock in the enamel. Tooth enamel is formed layer by layer, and it locks in the ratio of the oxygen isotopes ¹⁸O and ¹⁶O, written δ¹⁸O, in the body when it forms. Measured along a growing canine, the changes in this ratio can be used as a clock for how fast the enamel grew. For the sabre-toothed cat Smilodon gracilis, Robert Feranec found a growth rate of about 5 mm per month over roughly 16 months 17. For Smilodon fatalis, Aleksander Wysocki and colleagues combined a previously published oxygen-isotope growth rate with CT scans of developing teeth: the upper canine erupted at about 6.0 mm per month. The rest of the permanent teeth were in place by 14–22 months of age, but the sabres finished erupting only at about 34–41 months – a delayed dental development compared with living cats of similar size 16.

A menu in the enamel. Carbon isotopes (δ¹³C) record the plants at the base of the food chain. In one canine of Smilodon gracilis from Florida, the carbon isotope values point to prey that fed only on so-called C3 plants; this probably rules out some prey animals such as horses and mammoths, and the values may indicate prey caught in a closed habitat – or prey species not yet analysed, according to Feranec 17.

Competition – a debate. At Rancho La Brea, Larisa DeSantis and colleagues combined several lines of evidence (a multi-proxy approach) from hundreds of carnivore and herbivore fossils. They conclude that dire wolves ate prey from more open habitats than the cats, and that sabre-toothed cats and dire wolves competed little for prey – overturning an earlier view 4. That would place the cats’ hunting grounds in more wooded, less open country – our inference from the prey, not a direct finding. Blaire Van Valkenburgh and colleagues challenged this conclusion, pointing to problems with inferring a lack of competition from the dental enamel data 18; DeSantis and colleagues replied to these concerns 19. The question remains open.

Radiocarbon and the end. Radiocarbon dating uses the decay of the radioactive isotope carbon-14 in once-living material. Robin O’Keefe and colleagues obtained 172 new radiocarbon dates from megafauna at Rancho La Brea. Seven extinct species of megafauna disappeared there by 12,900 years ago. The authors link this to a shift of the ecosystem after aridification and vegetation change, time-series models point to large fires as the primary cause, and growing human impact may have been the trigger – a hypothesis, not a settled result 20.

Computer science lens: stress in a growing sabre

How much does the direction of a load matter, and does a longer sabre become weaker? The program below takes the cross-sections of a fully grown Smilodon fatalis canine that Tseng took from digital 3D models at successive heights of a simulated eruption 6, and calculates, with the beam formulas from the mathematics lens, the bending stress at the base for a load from the side and from the front. Assumptions: the cross-section is an ellipse; the load of 100 N at the tip is an assumed round value for comparison, not a measured force; only the ratio of the stresses and their trend are meaningful, not the absolute values.

import numpy as np
import matplotlib.pyplot as plt

# How strong is a sabre? The upper canine of Smilodon fatalis as a cantilever beam with an
# elliptical cross-section, measured at the gum line of each stage of a simulated eruption (the adult tooth shifted step by step into the jaw).
# Data: Tseng (2025), Table S1, "morpho" models (article licensed CC BY-NC-ND 4.0; measurement values quoted with citation): (erupted height, front-to-back
# length x, side-to-side width y) in mm.
CANINES = {
    "AMNH FM55576": [(40.62, 21.58, 10.28), (50.23, 24.64, 11.18), (58.62, 26.75, 11.63),
                     (68.15, 28.78, 12.81), (78.31, 30.34, 13.93), (89.28, 31.08, 15.07),
                     (97.68, 31.58, 15.66), (106.83, 32.16, 16.44), (118.43, 32.55, 17.48),
                     (127.78, 31.99, 17.70), (137.76, 32.17, 18.00)],
    "KU S2255": [(20.20, 14.89, 7.73), (31.31, 19.81, 9.67), (42.91, 22.99, 10.58),
                 (54.85, 26.80, 11.96), (65.21, 29.56, 13.25), (75.16, 31.41, 13.94),
                 (85.12, 33.21, 14.94), (95.03, 34.63, 15.65)],
}
SPECIMEN_INDEX = 0          # 0: AMNH FM55576, the longest “morpho” series in Table S1 (11 stages, 41–138 mm); 1: KU S2255
SPECIMEN = list(CANINES)[SPECIMEN_INDEX]
LOAD_N = 100.0              # force at the tip (N) – an assumed round value for comparison
ROUND = False               # True: same cross-sectional area, but a round tooth


def bending_stress(force, height, x, y):
    """Peak bending stress (MPa) at the base of an elliptical cantilever.
    Semi-axes a = x/2 (front-to-back) and b = y/2 (side-to-side), all in mm.
    Side load bends about the long axis: section modulus π·a·b²/4.
    Front load bends about the short axis: section modulus π·b·a²/4."""
    a, b = x / 2, y / 2
    moment = force * height                      # N·mm
    side = moment / (np.pi * a * b**2 / 4)       # N/mm² = MPa
    front = moment / (np.pi * b * a**2 / 4)
    return side, front


data = np.array(CANINES[SPECIMEN])
height, x, y = data.T
if ROUND:                                        # same area, circular: x = y = √(x·y)
    x = y = np.sqrt(x * y)
side, front = bending_stress(LOAD_N, height, x, y)

print(f"{SPECIMEN}, load {LOAD_N:.0f} N at the tip")
for h, s, f in ((height[0], side[0], front[0]), (height[-1], side[-1], front[-1])):
    print(f"  height {h:.0f} mm: {s:.1f} MPa from the side, {f:.1f} MPa from the front")
print(f"at the last stage, a side load stresses the tooth {side[-1] / front[-1]:.1f} times more than a front load")
print(f"from first to last stage, the side stress changes by a factor of {side[-1] / side[0]:.2f}, "
      f"while the height grows {height[-1] / height[0]:.1f} times")

plt.rcParams["font.size"] = 13      # large type: the chart stays readable on a phone
fig, ax = plt.subplots(figsize=(6.4, 5.6), layout="constrained")
ax.plot(height, side, "o-", lw=2.5, color="#c0392b", label="load from the side")
ax.plot(height, front, "s--", lw=2.5, color="#1f6f8b", label="load from the front")
ax.set_ylim(0, 1.15 * side.max())
ax.axvspan(20, 40, color="grey", alpha=0.15)
ax.text(30, 1.12 * side.max(), "typical\ncat canine", ha="center", va="top", fontsize=11)
ax.set_xlabel("erupted height of the canine (mm)")
ax.set_ylabel(f"bending stress at the base (MPa)\nfor {LOAD_N:.0f} N at the tip")
ax.set_xlim(15, 145)
ax.set_title("Sideways, the sabre is the weak spot")
ax.legend(loc="lower right", fontsize=11)
plt.show()

• Open in Colab
• Download .py
• How to run this code

Output · what the code prints, re-run at every build

AMNH FM55576, load 100 N at the tip
  height 41 mm: 18.1 MPa from the side, 8.6 MPa from the front
  height 138 mm: 13.5 MPa from the side, 7.5 MPa from the front
at the last stage, a side load stresses the tooth 1.8 times more than a front load
from first to last stage, the side stress changes by a factor of 0.74, while the height grows 3.4 times

What the result teaches:

• Sideways is the weak direction. For the fully erupted canine, a load from the side causes 13.5 MPa, a load from the front 7.5 MPa – the side load causes about 1.8 times as much stress. The side is the weak direction at every stage, because the ratio of the two stresses is just the tooth’s length-to-width ratio – between 1.8 and 2.3 in this specimen.
• A longer sabre is not a weaker sabre. In Tseng’s simulated series the erupted part becomes 3.4 times taller, yet the side stress falls to 0.74 times its starting value: the section at the gum line widens fast enough to more than compensate for the longer lever. This matches Tseng’s finding that the cross-section is optimised for bending strength 6 – although, by his results, it also bends more easily sideways.
• Robust and limited. The direction effect follows from the measured shape alone. The absolute numbers depend on the assumed load and ignore the jaw bone and the milk canine that supported the growing tooth. And the series is one adult tooth shifted step by step against a hypothetical gum line, with the same load at every stage – not a record of a growing tooth.

Try it yourself:

• Another specimen: set SPECIMEN_INDEX = 1 (KU S2255). Its tooth is even flatter; at its last stage (95 mm), the side load causes 2.2 times the stress of the front load, and from the first to the last stage the side stress falls to 0.49 times its starting value.
• A round sabre: set ROUND = True – the same cross-sectional area, but circular. The direction no longer matters (ratio 1.0), and the side stress of the fully erupted tooth drops to 10.1 MPa.

KU S2255, load 100 N at the tip
  height 20 mm: 23.1 MPa from the side, 12.0 MPa from the front
  height 95 mm: 11.4 MPa from the side, 5.2 MPa from the front
at the last stage, a side load stresses the tooth 2.2 times more than a front load
from first to last stage, the side stress changes by a factor of 0.49, while the height grows 4.7 times

AMNH FM55576, load 100 N at the tip
  height 41 mm: 12.5 MPa from the side, 12.5 MPa from the front
  height 138 mm: 10.1 MPa from the side, 10.1 MPa from the front
at the last stage, a side load stresses the tooth 1.0 times more than a front load
from first to last stage, the side stress changes by a factor of 0.80, while the height grows 3.4 times