Evolution of throwing

The capacity for high-speed, accurate overhand throwing is one of the most distinctive and consequential motor abilities in the human lineage. No other animal can match the combination of velocity and precision that a trained human thrower achieves: professional baseball pitchers routinely release balls at over 40 metres per second (approximately 145 km/h), while even untrained adults far exceed the projectile speeds attainable by any other primate.¹ This ability is not a simple by-product of general upper-body strength or intelligence. It depends on a coordinated suite of derived anatomical features—in the shoulder, torso, waist, arm, and wrist—that evolved over the course of at least two million years. Understanding the evolution of throwing illuminates not only the biomechanics of the human body but also the origins of hunting, the development of projectile technology, and possibly the evolution of the brain itself.

The uniqueness of human throwing

Throwing is not, in the strictest sense, unique to humans. Chimpanzees throw objects—sometimes overhand—in agonistic displays, during intergroup encounters, and occasionally in aimed contexts such as nut-cracking. Wild chimpanzees at Fongoli, Sénégal, have been observed fashioning and thrusting sharpened sticks into tree hollows to probe for bushbabies, and captive chimpanzees frequently throw objects at zoo visitors.⁶ Yet the kinematic data reveal a gulf between chimpanzee and human throwing. Chimpanzee throws are slow and inaccurate, typically generating ball velocities under 20 metres per second and showing poor consistency in aim. Their throws rely almost entirely on the arm, with minimal contribution from the torso, hips, or legs.⁴ The difference is not merely one of practice or motivation. It is structural: chimpanzees lack the anatomical features that allow humans to store and rapidly release elastic energy during the throwing motion.

The key distinction lies in the kinetic chain. In a skilled human throw, energy is generated sequentially from the legs through the hips, trunk, shoulder, elbow, and wrist, with each proximal segment decelerating as it transfers momentum to the next distal segment. This "whip-like" cascade of energy transfer allows the hand to reach velocities far in excess of what any single muscle group could produce in isolation.^{1, 11} The entire sequence unfolds in less than a second, and the release window during which the ball must leave the hand to travel accurately to a target may be as brief as a few milliseconds. This level of timing precision is unmatched by any other voluntary human movement.

Biomechanics and elastic energy storage

The biomechanical foundation of high-speed throwing was elucidated in a landmark 2013 study by Neil Roach and colleagues, published in *Nature*. Using three-dimensional motion capture to analyse the throwing kinematics of collegiate baseball players, the team demonstrated that the human shoulder functions as an elastic energy storage mechanism during the cocking phase of the throw. As the arm is drawn back into extreme external rotation, the tendons, ligaments, and elastic elements crossing the shoulder joint are stretched like a slingshot. When the arm transitions from cocking to acceleration, this stored elastic energy is released, producing angular velocities at the shoulder that can exceed 9,000 degrees per second—the fastest motion produced by any joint in the human body.¹

Several specific anatomical features make this elastic energy storage possible. The human shoulder joint is oriented more laterally than in apes, positioning the glenohumeral joint in a way that permits extreme external rotation during the cocking phase. The relatively low, broad waist of humans—in contrast to the funnel-shaped ribcage of apes—allows independent rotation of the hips and trunk, enabling the torso to "wind up" relative to the pelvis and store rotational elastic energy in the oblique muscles and spinal connective tissues. The increased torsion (twist) of the humeral shaft positions the elbow and hand optimally for the throwing plane when the shoulder is abducted. And the mobile, extensible wrist permits a final "snap" that adds velocity at the moment of release.^{1, 10, 11}

Roach and colleagues experimentally confirmed the importance of these features by fitting subjects with braces that restricted shoulder external rotation, waist rotation, or both. When either degree of freedom was limited, throwing speed dropped significantly. When both were restricted simultaneously, subjects' throwing performance converged with the low velocities observed in chimpanzees, demonstrating that it is precisely these derived features—absent in the great apes—that account for the human throwing advantage.¹

Fossil evidence and the timing of throwing evolution

If the capacity for high-speed throwing depends on specific skeletal features—lateral shoulder orientation, a tall mobile waist with decoupled hip-trunk rotation, and increased humeral torsion—then the fossil record should, in principle, reveal when these features first appeared. Roach and colleagues argued that the full suite of throwing-relevant morphology is first present in *Homo erectus*, which appears in the East African fossil record by approximately 1.9 million years ago. The Nariokotome boy (KNM-WT 15000), a remarkably complete *H. erectus* skeleton dated to approximately 1.5 million years ago, exhibits a tall waist, a barrel-shaped (rather than funnel-shaped) thorax, a laterally oriented glenoid fossa, and humeral torsion values within the modern human range.^{1, 10}

Earlier hominins appear to lack one or more of these features. Australopithecines, including *Australopithecus afarensis*, retained a cranially oriented (upward-facing) shoulder joint more similar to that of extant apes, a funnel-shaped ribcage, and a relatively short waist. While australopithecines were bipedal and their freed upper limbs could certainly have been used for throwing objects, the biomechanical analysis suggests they would not have been capable of the high-speed, elastic energy-dependent throwing characteristic of later *Homo*.¹ The implication is that selection for throwing ability intensified during the transition from australopithecines to early *Homo*, coinciding with a broader shift toward increased carnivory and stone tool production in the archaeological record.

The pattern of humeral torsion provides an additional line of evidence. In modern humans, the degree of humeral torsion is partly developmental—influenced by throwing activity during growth—but the baseline range is substantially higher than in chimpanzees. Fossil humeri from *Homo erectus* show torsion values consistent with habitual throwing, whereas earlier hominin specimens and Neanderthal humeri show somewhat lower values. This has led some researchers to suggest that Neanderthals, despite their formidable physical strength, may have been less proficient at high-speed throwing than contemporary *Homo sapiens*, relying instead on thrusting spears used at close quarters.^{9, 10}

Archaeological evidence for projectile technology

The most dramatic archaeological evidence for the importance of throwing in hominin life comes from the site of Schöningen in northern Germany. In 1994–1998, Hartmut Thieme recovered a series of carefully crafted wooden spears from deposits dated to approximately 300,000 years ago, associated with the remains of butchered horses. The spears, which range from 1.8 to 2.5 metres in length, are tapered at both ends with the maximum diameter and centre of gravity located in the forward third of the shaft—a design that optimises aerodynamic stability and distance when thrown overhand, closely resembling the balance of modern javelins.^{3, 13} Additional wooden artefacts recovered from the same deposits include a double-pointed throwing stick, approximately 64 centimetres long, dated to roughly 300,000 years ago, which experimental analyses confirmed could be thrown with accuracy over distances of 30 metres.¹⁵

Whether the Schöningen spears were thrown or thrust has been debated. Steven Churchill argued that the force-delivery characteristics of the spears, combined with the close-range ambush hunting strategies inferred from the site's landscape context, were more consistent with thrusting than throwing.⁹ However, the javelin-like weight distribution, the experimental evidence that replicas fly accurately when thrown, and the presence of the dedicated throwing stick at the same site strongly support the interpretation that at least some of these weapons were designed as projectiles.^{3, 15}

Stone-tipped projectile points provide another important category of evidence.¹² At the site of Gademotta in the Ethiopian Rift Valley, Yonatan Sahle and colleagues reported stone points with diagnostic impact fractures—indicating use as projectile tips rather than handheld tools—from deposits older than 279,000 years.¹² By the Middle Stone Age in Africa, beginning roughly 300,000 years ago, composite hafted points suitable for use as throwing spear tips become increasingly common, suggesting that projectile technology was central to *Homo sapiens* subsistence strategies from an early date.^{8, 12} The subsequent invention of the atlatl (spearthrower) and the bow further extended the range and lethality of projectile weapons, but the fundamental biomechanical capacity that made all of these technologies effective—the human throwing arm—was already in place by the time of *Homo erectus*.

Selective advantages of throwing

What drove the evolution of high-speed throwing? The most widely accepted hypothesis is that throwing conferred critical advantages in both hunting and anti-predator defence.¹ On the African savanna, where *Homo erectus* lived alongside large-bodied carnivores, the ability to injure or kill prey from a distance would have dramatically reduced the physical risks associated with close-range confrontation. Even inaccurate projectiles, thrown in volleys by cooperating groups, could harass and drive off competing predators from carcasses, a behaviour observed in modern hunter-gatherer societies.⁵ Throwing thus extended the effective "strike zone" of hominin hunters beyond the reach of teeth and claws, permitting the exploitation of large game that would otherwise have been inaccessible.

The energetic implications are significant. Meat and marrow from large animals are calorie-dense resources, and the shift toward increased carnivory in early *Homo* is associated with the evolution of larger bodies, larger brains, and reduced gut size—the suite of changes described by the expensive tissue hypothesis.⁵ If throwing enabled more reliable access to these high-quality resources, it would have been under strong positive selection. Roach and colleagues noted that the evolution of throwing anatomy in *Homo erectus* coincides temporally with both the appearance of Acheulean stone tool technology and evidence for increased meat consumption in the hominin diet, suggesting that these developments were functionally linked.¹

Calvin's throwing hypothesis and brain lateralization

In 1982, the neurophysiologist William Calvin proposed a provocative hypothesis linking the evolution of throwing to the evolution of the human brain. Calvin argued that the millisecond-level timing precision required for accurate throwing—particularly the narrow release window within which the projectile must leave the hand to reach its target—would have placed intense selective pressure on the neural circuits responsible for sequential motor planning. Because accurate throwing requires the coordination of dozens of muscles firing in a precisely timed cascade, Calvin suggested that the brain regions controlling throwing would have expanded under selection, and that these same neural circuits were later co-opted for other sequentially structured behaviours, including tool manufacture and language.^{2, 7}

Calvin's hypothesis specifically predicted that throwing ability and handedness should be linked, because the lateralised motor control required for throwing would favour the development of a dominant hand. The observation that approximately 90% of humans are right-handed—a degree of population-level handedness not seen in any other primate—is consistent with this prediction, and some researchers have explored the connection between throwing, handedness, and brain lateralization empirically. William Hopkins and colleagues demonstrated that in captive chimpanzees, individuals that threw more frequently and more accurately showed stronger hand preferences and greater asymmetry in the motor cortex of the brain, suggesting that even in non-human primates, throwing behaviour and hemispheric specialisation are correlated.¹⁴

Calvin's hypothesis remains speculative in several respects. The neural substrates of throwing overlap with but are not identical to those of language production, and the leap from motor timing precision to syntactic recursion is large. Nevertheless, the core insight—that the extraordinary demands of throwing on sequential motor planning may have been a driver of brain evolution—continues to be discussed in the literature and has influenced subsequent thinking about the co-evolution of motor skill, tool use, and cognition.^{2, 7}

Neanderthals and the throwing debate

The question of whether Neanderthals were effective throwers has significant implications for understanding their hunting strategies and, potentially, their competitive relationship with *Homo sapiens*. Neanderthal postcranial anatomy differs from that of modern humans in several respects relevant to throwing biomechanics. Their humeri show lower torsion values on average, their shoulders were broader and more anteriorly oriented, and their body proportions were adapted for cold climates and close-range power rather than long-range ballistic performance.^{9, 10}

The archaeological record associated with Neanderthals is dominated by Mousterian stone points and Levallois flakes that show patterns of damage consistent with use as thrusting spear tips rather than projectile points. The high frequency of upper-body trauma observed in Neanderthal skeletal remains—comparable to that seen in modern rodeo riders—has been interpreted as evidence for habitual close-range encounters with large prey, a pattern more consistent with thrusting spears than with throwing.⁹ However, the Schöningen spears, which predate the classic Neanderthal period, demonstrate that their probable ancestors or relatives possessed throwing technology, and some researchers have argued that the dichotomy between Neanderthal thrusting and *sapiens* throwing has been overstated.^{3, 13}

What is clear is that *Homo sapiens* populations in Africa developed increasingly sophisticated projectile technologies—hafted stone-tipped spears, throwing sticks, and eventually atlatls and bows—over the course of the Middle and Later Stone Age.^{8, 12} Whether this technological advantage in ranged weapons contributed to the competitive displacement of Neanderthals after 50,000 years ago remains debated, but it is one of several hypotheses for Neanderthal extinction.

Significance and synthesis

The evolution of throwing is not a marginal curiosity in human evolutionary history. It is deeply intertwined with some of the most consequential transitions in the hominin lineage: the shift to carnivory, the development of cooperative hunting, the elaboration of projectile technology, and possibly the reorganisation of the brain for sequential motor planning and language. The anatomical evidence indicates that the capacity for high-speed throwing evolved by approximately two million years ago in *Homo erectus*, making it one of the earliest distinctively human adaptations. The archaeological record documents the progressive elaboration of projectile weapons from the simple wooden spears of Schöningen to the complex composite projectile systems of the Upper Palaeolithic and beyond.^{1, 3, 8, 15} In a lineage defined by its tools, throwing was the original weapon system—the adaptation that first allowed a relatively small, slow, and poorly armed primate to compete with the great predators of the African savanna.