Fossil evidence for the origin of Homo sapiens


  • Jeffrey H. Schwartz,

    Corresponding author
    1. Departments of Anthropology and History and Philosophy of Science, University of Pittsburgh, Pittsburgh, PA 15260
    • Departments of Anthropology and History and Philosophy of Science, 3302 WWPH, University of Pittsburgh, Pittsburgh, PA 15260
    Search for more papers by this author
  • Ian Tattersall

    1. Division of Anthropology, American Museum of Natural History, New York City, NY 10024
    Search for more papers by this author


Our species Homo sapiens has never received a satisfactory morphological definition. Deriving partly from Linnaeus's exhortation simply to “know thyself,” and partly from the insistence by advocates of the Evolutionary Synthesis in the mid-20th Century that species are constantly transforming ephemera that by definition cannot be pinned down by morphology, this unfortunate situation has led to huge uncertainty over which hominid fossils ought to be included in H. sapiens, and even over which of them should be qualified as “archaic” or as “anatomically modern,” a debate that is an oddity in the broader context of paleontology. Here, we propose a suite of features that seems to characterize all H. sapiens alive today, and we review the fossil evidence in light of those features, paying particular attention to the bipartite brow and the “chin” as examples of how, given the continuum from developmentally regulated genes to adult morphology, we might consider features preserved in fossil specimens in a comparative analysis that includes extant taxa. We also suggest that this perspective on the origination of novelty, which has gained a substantial foothold in the general field of evolutionary developmental biology, has an intellectual place in paleoanthropology and hominid systematics, including in defining our species, H. sapiens. Beginning solely with the distinctive living species reveals a startling variety in morphologies among late middle and late Pleistocene hominids, none of which can be plausibly attributed to H. sapiens/H. neanderthalensis admixture. Allowing for a slightly greater envelope of variation than exists today, basic “modern” morphology seems to have appeared significantly earlier in time than the first stirrings of the modern symbolic cognitive system. Yrbk Phys Anthropol 53:94–121, 2010. © 2010 Wiley-Liss, Inc.

Our species Homo sapiens has never been subject to a formal morphological definition, of the sort that would help us in any practical way to recognize our conspecifics in the fossil record. To understand why, a bit of history is helpful. The earliest surviving comparisons between humans and animals using both differences and similarities are those of the Greek polymath Aristotle (see review in Schwartz, 1999). On the subject of human distinctiveness, Aristotle wrote:

“Now, man, instead of forelegs and forefeet, has, as we call them, arms and hands. For he alone of the animals stands upright, on account of his nature and ousia [= “substantial being” or “defining character”] being divine, and the function of that which is most divine is to think and reason; and this would not be easy if there were a great deal of the body at the top weighing it down, for weight hampers the motion of the intellect and the common sense” (Aristotle, 1945, xxx IV. 12 693a25–31).

Although Aristotle's comparisons were limited to humans and other living animals, he nonetheless articulated three major features—bipedalism directly, and the freeing of the hands in locomotory behavior, and the reasoning power of the brain by implication—that would long stand as defining characteristics of our species H. sapiens, and would provide as well the morphological cornerstones of the eventual discipline of paleoanthropology. In Aristotle's view, a Prime Mover pushed the psyche of each organism, on its rung of the Ladder of Life (Scala Naturae), to follow its destiny and to strive to achieve impossible perfection. Although perched on the uppermost rung of this ladder, humans, no less than any other organism, failed to achieve a perfect state.

During the Dark Ages that replaced the Greco-Roman tradition of individual thinking and exploration with spiritual inquiry and divine revelation, the Scala Naturae was more or less directly transformed into the Great Chain of Being, in which an ascendant ordering from the inorganic through the organic world reflected the creation story of Genesis (Lovejoy, 1942). The early systematists who labored to elucidate this chain achieved their goals through equally idiosyncratic classifications. One way to recognize the nearly divine status of humans was to exclude them altogether from the classification. This route was chosen in the 16th Century by Konrad Gesner (the inventor of the genus rank), and also in the 17th Century by Francis Willughby (see Schwartz, 1999), who nevertheless had clearly considered human characteristics in his comparisons, describing as “man-like” a number of features he thought aligned the broad categories of “baboon” and “monkey.” In 1632, Ioannes Jonstonus (Jonstonus, 1632) became one of the first taxonomists to discuss humans directly in comparison with other animals, but only more than a century later were humans classified not in their own higher category, but in the same group as other man-like mammals.

In 1735, Carolus Linnaeus struck what to a religiously minded “scientific” world was a deep blow to the sacrosanct, by placing the species to which he belonged within a taxonomic group that John Ray had actually named for other animals: the order Anthropomorpha (Linnaeus, 1735). Only later (Linnaeus, 1758) did Linnaeus change the ordinal name to Primates, meaning “chiefs of creation.” Although raising the ire of other taxonomists, Linnaeus was really just taking a logical step. But as outraged as his fellow taxonomists were, Linnaeus had rejected neither special creation, nor the belief that his own species, which he dubbed H. sapiens, had been created last among Primates and in the image of its God. Still, it was not just in grouping humans in the same taxon as other mammals that Linnaeus broke with broad tradition. More specifically, it was in his presentation of the genus and species H. sapiens that Linnaeus abandoned his usual practice of providing a diagnosis for each taxon. For, his only comment about his own species was: nosce te ipsum (know thyself).


Nicholas Steno had demonstrated as early as 1669 the structural similarities between fossil bones, teeth, and shells, and their counterparts in living vertebrates and invertebrates. But, well into the 19th Century, humans were denied any antiquity beyond the historical present as recounted in a literal reading of the Book of Genesis (see review in Schwartz, 1999). Thus, when in the late 18th Century, Johann Friedrich Blumenbach (Blumenbach, 1969) published his treatise on the morphological features that united all “races” of H. sapiens, his comparisons were solely among extant taxa. Although Blumenbach generally praised Linnaeus's groundbreaking taxonomic work, he felt not only that Linnaeus had been too focused on features of the mammal dentition, but also that he had left an unfortunate gap by not providing even a single feature on which to base either the genus Homo or the species H. sapiens. It was this lacuna that Blumenbach set out to fill.

Blumenbach discussed, among other things, the “external conformation of the human body,” its “internal conformation,” and “those points, in which man is commonly, but wrongly, thought to differ from the brutes.” On the first topic, Blumenbach discussed a number of osteological features distinguishing humans from other primates and that were indeed fundamental to a diagnosis of H. sapiens: 1) Erect posture that develops naturally and spontaneously, which is associated with an anteroposteriorly shallow but laterally broad thoracic cavity, widely separated shoulder joints, short sternum, and scapulae that lie posteriorly on a rib cage that does not fully encase the viscera; 2) A broad and flat pelvis with broad and expanded ilia in which the ossa coxae (Blumenbach's ossa innominata), together with the sacrum and its coccygeal bones, form a basin that cups the viscera (according to Blumenbach, a “true” pelvis); 3) Two hands, each perfect (harking back to Aristotle) in possessing a long thumb (the basis for Blumenbach's order Bimana, in contrast to Quadrumana, which subsumed the “four-handed” nonhuman primates; 4) Two feet with large and nonopposable first toes; 5) Vertically implanted lower incisors accompanied by serially aligned, close-set, and short canines; 6) Molars with rounded rather than pointy cusps; 7) A short mandible with a prominent chin; and 8) A single (not twinned) opening in the palate that is situated just posterior to the upper incisors. Among soft tissue features, Blumenbach suggested humans were unique in possessing swollen lips and earlobes [though Schultz (Schultz, 1968) would later point out that chimpanzees occasionally develop earlobes]. Although Blumenbach disagreed with his colleague Johann Wolfgang von Goethe over the significance of the absence of a premaxillary bone in humans, the two were intellectually united in the belief that the most important attribute separating “man” from the “brutes” is reason—the same quality to which Edward Tyson (1699) had resorted on discovering the anatomical similarities between a juvenile chimpanzee and humans, and to which others would repeatedly turn in their attempts to capture the distinctiveness of H. sapiens.


The constraining influence Genesis had on considerations of human antiquity was reflected in the silent rejection by the scholarly community of Charles-Philippe Schmerling's (1833) astonishingly insightful interpretation of human-like bones from the Belgian cave sites of Engis and Engihoul as both fossilized, and contemporaneous with the remains of extinct mammals. Even when Charles Lyell (1863) later studied these caves and came to agree with Schmerling, the case for human antiquity fell largely on deaf ears. Indeed, the discovery in 1857 of the Feldhofer Grotto Neanderthal remains still failed to fully resolve the issue of human antiquity.

The saga of this discovery is too well known to need repeating in detail. But, it is nonetheless important to emphasize that Carl Fuhlrott, into whose possession the Neanderthal bones first came, and Hermann Schaaffhausen, their scientific describer, were adamantly at odds regarding their antiquity. Fuhlrott believed that their state of mineralization and their apparent co-occurrence with fossils of extinct mammals demonstrated their ancientness. But despite describing them as different from present-day humans in substantial aspects of their preserved morphology, Schaaffhausen advanced a series of arguments that culminated in a resounding rejection of this individual's antiquity. For Schaaffhausen, the Feldhofer Grotto remains could be accommodated easily by stories of barbaric and savage H. sapiens who once inhabited Western Europe (Schaaffhausen, 1861).

The first analysis of the Feldhofer Grotto specimen by an acknowledged evolutionist came in 1863, in Chapter 3 of Thomas Henry Huxley's monograph Man's Place in Nature, “On some fossil remains of man.” Before turning to the Feldhofer remains, Huxley introduced the specimens from Engis and Engihoul. Although glossing over both the Engis child's partial skull and the material from Engihoul, Huxley did discuss in some detail the adult and partially complete Engis crania and accompanied these passages with a lithograph that illustrates clearly the morphological details of a bipartite brow (Schwartz, 2006), emphasizing that the configuration of the supraorbital region, in conjunction with general contours of the weakly muscle-scarred braincase, indicated that this partial cranium had “belonged to a man of a low degree of civilization: a deduction which is borne out by contrasting the capacity of the frontal with that of the occipital region.”

Turning to the Feldhofer Grotto skeletal remains, Huxley listed a large number of differences distinguishing the fossil skullcap from modern humans. Given his theoretical predispositions toward the significance of morphologically discrete features and evolutionary saltation, as well as his rejection of Darwinian gradualism (Schwartz 2005, 2006), one might well have expected Huxley to conclude that here was an extinct relative of modern humans. Instead, via some remarkable special pleading, he claimed that it was possible to assemble a sequence of human skulls, from around the world, which represented a progression from the most primitive to the most advanced. Having, thus, produced a graded series based on perceived differences and similarities in cranial shape, he could then declare that “A small additional amount of flattening and lengthening, with a corresponding increase of the supraciliary ridges, would convert the Australian brain case [at the bottom of the series] into a form identical with that of the aberrant fossil” (Huxley 1863, p 179–180). As a result, “the fossil remains of Man hitherto discovered do not seem to me to take us appreciably nearer to that lower pithecoid form, by the modification of which he has, probably, become what he is” (Huxley, 1863, p 183). In this way, the most prominent comparative anatomist of his time simultaneously denied the distinctiveness of the Neanderthaler and introduced the notion that the morphology of H. sapiens encompassed an almost unimaginably broad range.

A year later, William King (1864) took Huxley to task. In his view, the distinct morphologies of the fossil specimen were without counterparts in any living human. But even as discoveries mounted that showed the Feldhofer specimen was no isolated occurrence, the influence of Huxley's conclusions quietly grew. Indeed, the mindset he fostered is still alive and well today. Still, what we find perhaps most remarkable about the early discourses on the Feldhofer Grotto remains, postcrania included, is that all almost entirely neglected Blumenbach's distinguishing features of H. sapiens. Only in one of Huxley's (1863) three chapters in Man's Place in Nature, “Man's relation to the lower animals,” was Blumenbach even mentioned, and then merely to claim an allegedly undue emphasis on external differences between the quadrumanous primates and humans in their (hind) feet. Despite those differences, Huxley argued, the skeletal details of the human foot narrowed the “gap” between Blumenbach's Bimana and Quadrumana.


The intellectual environment of the late 19th and early 20th Centuries also sanctioned the identification of subsequently discovered human fossils as “racial” antecedents of presumed modern races, from sites such as Grimaldi Cave (northeastern Italy), Boskop (Southwest Transvaal, South Africa), and Wadjak (East Java, Indonesia), as early representatives of specific racial groups of H. sapiens. In turn, this provided free license to luminaries such as Sir Arthur Keith (e.g., Keith, 1931) to publish evolutionary trees that not only schematically depicted scenarios of racial differentiation, but also positioned individual fossils in distinct evolutionary lines leading to modern racial groups.

Although the Piltdown forgery for decades complicated interpretation of the Neanderthals, and especially of Eugene Dubois' Pithecanthropus erectus, the search for human-like fossil remains proceeded apace. By the 1930s, not only had a plethora of Neanderthal specimens been discovered at sites in western and eastern Europe but diverse fossil specimens of other human relatives had been unearthed at widespread locales: the Mauer mandible in Germany (holotype of H. heidelbergensis); a fairly complete skull and various other skeletal elements, from what is now Zambia, on which the species name H. rhodesiensis was bestowed; an isolated molar from a site near Beijing (then Choukoutien, now Zhoukoudian) that served as the namesake of the genus and species Sinanthropus pekinensis, into which a number of partial crania, fragmentary jaws, and isolated teeth were subsequently folded; and, from various sites in South Africa (first, Taung, followed by Kromdraai, Sterkfontein, and Swartkrans), cranial and mandibular specimens that became the holotypes of various species distributed among three different genera (Australopithecus, Plesianthropus, and Paranthropus).

During the 1940s, the recovery of such diverse specimens fueled the practice of bestowing new species and even new genus names on each new fossil. Thus when, in the 1950s, the systematist Ernst Mayr turned his sights onto the still-nascent field of paleoanthropology, he found himself dumbfounded and befuddled by a “bewildering diversity of names” in its literature. Following in the footsteps of Theodosius Dobzhansky (Dobzhansky, 1944), the geneticist and fellow architect of what became known as the Evolutionary Synthesis—who advanced the notion that the capacity for culture removed all hominids from evolutionary processes that would otherwise lead to divergent speciation—Mayr waded vigorously into the arena of human evolution.

Mayr's reasons for suggesting that human evolution was a single, nondiversifying continuum of change were these. An educated systematist would recognize that, regardless of apparent morphological differences, all hominids possessed the same adaptation: bipedal locomotion. Because this systematist would also know that a genus is defined by the ecological specializations of its constituent species, all hominids should be subsumed in the genus Homo because they all uniquely share the same locomotor form. Further, present-day H. sapiens is an incredibly varied and geographically widespread species that has successfully occupied all available econiches. By extension, hominids of the past must have been as morphologically variable as living humans, if not more so. Because, as he (Mayr, 1942) had previously argued, diversifying speciation (as opposed to linear transformation) requires that subspecies (defined as incipient species) had to invade vacant econiches for new selection pressures to orchestrate their gradual acquisition of new adaptations, hominids were not, and would never become, taxically diverse.

As a result, Mayr concluded that, because only one species of hominid would have existed at any point in time, the entire course of human evolution could be characterized as a highly variable, polymorphic continuum of transformation comprised of three time-successive species (Mayr, 1950). These were H. transvaalensis (for the earliest hominids, which were then only known from South African sites), H. erectus (which subsumed Sinanthropus, Pithecanthropus, and the Mauer jaw), and H. sapiens (everything younger than H. erectus, or for whatever reason not considered to be part of it; these included the Neanderthals and the Ngandong specimens from Java). In striking contrast to Blumenbach's focus on features that might distinguish the species sapiens from other mammals, Mayr's argument is interesting in that it presumes a transformation series of species; species neither morphologically defined nor diagnosed. Instead, Mayr redirected the focus away from the species, which until then had been the center of taxonomic debate (e.g., see Huxley, 1940, 1942; Dobzhansky, 1941; Mayr, 1942), and up to the genus, which he discussed only in the broadest of terms, with regard to locomotor behavior.

Mayr's reason for disregarding species in the hominid case may have had something to do with the repudiation of what was then very recent history, the ugly face of “race” and “racism” (Schwartz, 2006). Thus, Mayr claimed, echoing Darwin (1871), that even though we all know that “Congo pygmies” and Watusi are members of the same species, H. sapiens, without this prior knowledge, even a competent morphologist confronted with the skeletal remains of these “clearly different” humans might easily, yet mistakenly, conclude that each human group represented a distinctly different species. The implication of Mayr's folding a cornucopia of synchronic but morphologically dissimilar specimens into time-successive species was this: If groups of apparently disparate morphology are more or less universally agreed on to be members of the same species, it is scientifically ludicrous (and racist) to attach biological, systematic, and thus evolutionary meaning to the differences between them.

Notwithstanding Mayr's good intentions in reacting to the horrors of ethnic cleansing that were part of World War II, what is more relevant for the question “What constitutes H. sapiens?” is Mayr's promotion of a version of Linnaeus's “undiagnostic diagnosis” of our species: nosce te ipsum. In other words, because we “know” that short “Congo pygmies” and tall Watusi are members of H. sapiens, there is no need to offer a morphological diagnosis of our species because, well, we just know who we are. The same can be said of Huxley's dismissal of the Feldhofer Grotto individual. Huxley “knew” that the historically recent specimens in his study were representatives of H. sapiens. Consequently, although he faithfully illustrated features that we might now regard as restricted to H. sapiens, there was no compelling reason for him to discuss or describe them in any detail. Rather, he could concentrate on the thought experiment of how a Feldhofer-shaped calvaria might be transformed into that of what he considered the most archaic of human races, the Australian Aborigine.

The major point here is that in the cases both of Huxley and of Mayr, but especially of the latter, reifying a purely intuited species actually leads one back to the next lowest taxonomic rank: namely, the genus. For Huxley, Homo was the only genus available; although for Mayr, having compressed all previously proposed hominid genera into one, Homo, this was his focus by default. And the species, or the lineage of ever-changing chronospecies, had become entirely secondary. Indeed, even when Robinson's (Robinson, 1953a) wide-ranging defense of “the multitude of genera of Australopithecines proposed by Broom” (Mayr, 1953) obliged Mayr to recognize that “forms with Australopithecine characters existed not only in South Africa, but also in East Africa and Java” (p 281), and thus to envisage the possible coexistence of, and competition between, the two hominid genera Australopithecus and Homo), his focus remained on the ecological adaptations of each genus (Mayr, 1963b). Mayr's vision of the genus Homo never wavered: an entity within which each subsumed species became smoothly and continuously transformed into another, culminating in extant H. sapiens.

Despite Mayr's partial recantation, the influence of his original 1950 article is clearly evident in Leakey et al.'s (1964) revised diagnosis of the genus Homo, which from Keith onwards had been based largely on the assumption that a “cerebral Rubicon” of brain size divided humans from apes: This latter notion, in turn, clearly reflected the emphasis of earlier (nonevolutionist) scholars on “reasoning,” and thus on the brain as the ultimate barrier between humans and the “brutes.” Although Leakey et al.'s revised diagnosis was unfortunately filled with descriptors such as “variable,” “usually,” and “overlaps,” and phrases such as “very strongly marked to virtually imperceptible,” it is significant that they also emphasized many of the features that Blumenbach offered to distinguish the species H. sapiens from other species of mammals—and which Mayr, via his adaptation-centered model, later promoted as the systematic core of the genus Homo. According to Leakey and colleagues, in Homo:

“… the pollex is well developed and fully opposable and the hand is capable not only of a power grip but of, at the least, a simple and usually well developed precision grip…the anterior symphyseal contour varies from a marked retreat to a forward slope, in which the bony chin may be entirely lacking, or may vary from a slight to a very strongly developed mental trigone: the dental arcade is evenly rounded with no diastema in more primitive members of the genus … the canines are small, with little or no overlapping after the initial stages of wear” (Leakey et al., 1964, p 8).

Such declarations as these emphasize that, although Blumenbach's efforts were confined to species diagnoses that emerged from comparing living organisms, once human fossils were brought into the systematic equation features that had been presented as species-specific became increasingly descriptive of a chronologically and geographically diverse assemblage of specimens, none of which necessarily represented H. sapiens. This opened the way to the diagnosis of other species of the genus Homo without reference to H. sapiens, which remained as it always had been: simply, our species.

With that species firmly established as the pinnacle of primate and, more narrowly, of hominid evolution, there seemed, perhaps, even less need to define it. Because, implicitly, all features of living H. sapiens must be derived, or “advanced,” relative to any now-extinct hominid relative. Because the accepted scenario featured the theme of transition from H. erectus into the earliest H. sapiens, this assumption anticipated a linear pattern of acquisition of increasingly derived features in our lineage. In turn, this expectation provided a springboard for an array of publications that sought to trace a transformation or trend toward becoming totally “anatomically modern” in, for example, metrical attributes of the dentition (Wolpoff, 1971), the facial skeleton (Smith, 1984), the postcranial skeleton and inferred bipedality (Robinson, 1972), and brain size (Pilbeam, 1972). The title of Pilbeam's (1972) once-widely used textbook on human evolution, The Ascent of Man, clearly implies a morphological transformation that is remorselessly advancing toward the modern human condition.

What is most odd about this history is that anyone actually familiar with even a small portion of the human fossil record would ever even consider embracing Mayr's bizarrely influential assertions about human evolution. For, the signal of that record, even as it existed in the 1950s and 1960s, did not support Mayr's view at all. Nevertheless, most paleoanthropologists not only succumbed to Mayr's dictates but became intellectually constrained by them, apparently for the most part at least as a result of the weight of authority Mayr had gained, along with Dobzhansky and the paleontologist George Simpson, with the triumph of the “hardened” version of the Evolutionary Synthesis. This intellectual victory resulted in the almost complete suppression of competing evolutionary ideas, emanating primarily from Germany and the United Kingdom, that were in many ways much more “synthetic” (Schwartz, 2009a, b; Schwartz, in press) than the Synthesis itself. In the United States, especially, the prominent physical anthropologist S. L. Washburn (Washburn, 1951) was highly influential in publicizing the virtues of replacing old-fashioned “typology” with “population thinking” (e.g., Simpson, 1949). As a result, paleoanthropologists seem not to have noticed that the routine recognition and delineation of three different chronospecies of Homo was becoming ever more artificial and arbitrary as the hominid fossil record expanded.


A result of all this was that a vaguely defined but intuitively attractive practice came into vogue, of conceiving evolutionary change in terms purely of populations (Mayr, 1942, 1963a). This was neatly encapsulated by Simpson in his comment that evolutionary change was the “selection-influenced accretion of genetic changes in populations” (Simpson, 1949, p 389). Indeed, Simpson went so far as to dismiss non-Synthesis evolutionary ideas as “typological systematics,” which he equated with an equally unscientific pre-evolutionary essentialist mindset. Although partly well intentioned, in questioning the proliferation of named species that differed from one another often only in morphological minutiae, the focus on population thinking, whether in neo- or paleozoology, raised a major operational problem. Namely, how does a systematist work with the slippery concept of the species as an entity that is always slightly but continually changing, and whose boundaries at any point in time are determined by a biological definition imposed by entirely external constraints? For Simpson (e.g., Simpson 1944, 1961) and Mayr (e.g., Mayr, 1969, among others, the “lineage” thus became somewhat interchangeable with the “species.” There is no need to rehash here the debates over species definition that ensued from the late 1970s through the 1980s (see Tattersall, 2009). But it is worth repeating how influential the hardened Synthesis was on paleoanthropology, and thus on the determination of which fossil specimens should be considered H. sapiens (Tattersall, 1986).

An early consequence of population thinking in paleoanthropology and modern human origins was seen in later adaptations of Franz Weidenreich's notion of human evolution (Weidenreich, 1946, 1947) that portrayed morphologically disparate and geographically far-flung fossils as the ancestors of living races of H. sapiens currently occupying the same regions of the Old World. What held these oddly assorted specimens together as a single hominid species, despite their striking morphological differences, was the Mayrian notion that some amount of gene flow had maintained the biological link between these lineages. The evolutionary unity of diverse agglomerations of fossils as members of one and the same species was, thus, derived from an unknown degree of presumed genetic continuity, whereas the morphological differences among them were seen as reflecting each group's partial geographic isolation and adaptation to different environmental circumstances.

An extreme interpretation of the regional racial continuity model pervaded the work of the much-reviled Carlton Coon (e.g., Coon, 1966), but more recently Milford Wolpoff and colleagues have maintained a recognizably Weidenreichian multiregional model, with the origins of modern human population diversity rooted in H. erectus (Thorne and Wolpoff, 1982; Wolpoff, 1989b, 1992, 1996; Wolpoff et al., 1994b, 2006; Frayer et al., 2006). The primary assumption on which Wolpoff and his colleagues rely is the existence of continually changing and occasionally interbreeding and gene-exchanging lineages within a species exhibiting morphological variations that are due to differing ecogeographical circumstances (e.g., there is a Neanderthal phase in Europe and the Near East, but not in Asia). From this perspective, these authors ultimately argued that if H. sapiens is today the result of nearly 2 million years of post-H. habilis lineage transformation, it is nonsensical to recognize H. erectus as a distinct taxon. Rather, if human evolution after H.habilis and into the present was indeed continuous and genetically interwoven, one should refer all non-H. habilis specimens to the species to which their living descendants belong: namely, H. sapiens.

Erik Trinkaus' (2006) scenario for H. sapiens origin is, in essence, a version of the multiregional model. However, rather than referring specifically to fossils that other paleoanthropologists might consider representative of distinct taxa, Trinkaus keeps his language vague. Consequently, his (p 598) “general model of Pleistocene genus Homo phylogeny” begins with the emergence of “early Homo” in Africa during the late Pliocene, followed in the early Pleistocene by the dispersal of populations of early Homo throughout Africa and into Southern Eurasia, extending from the Atlantic to the Pacific by the end of the early Pleistocene. “Archaic Homo” of the Middle Pleistocene expanded this range geographically and acquired regional variations in craniofacial morphology and body proportions in the process. Trinkaus explains this supposed transformation from the perspective of a populational emphasis on “intraspecific differentiation through isolation-by-distance” (p 598).

For Trinkaus, the process of regional differentiation continued into the late middle and early late Pleistocene. He finds evidence of this in the appearance of “late archaic humans (Neanderthals)” in Western Eurasia, of “less-well-documented late archaic humans” in Central, Southern, and Eastern Asia as well as in northwestern Africa, and of “early modern humans” primarily in Eastern Africa. Rejecting cladistic theory and methodology with the assertion that this approach to phylogenetic reconstruction is tautological, Trinkaus (Trinkaus, 2006) turns to the traditional stratophenetic approach (Gingerich, 1976) to determining character polarity: i.e., an accepted temporal sequence of fossils is the true and unerring arbiter of primitiveness versus derivedness. From this transformationist perspective, Trinkaus concludes that paleoanthropologists who embraced a cladistic orientation misinterpreted Neanderthal features as being derived and those of modern humans as being primitive, when things were clearly the other way round.

The counterpoint to a multiregional model (the single-origin, out-of-Africa notion) has been championed by C. B. Stringer (Stringer et al., 1984; Stringer and Andrews, 1988; Stringer and McKie, 1996), in part on the basis of late Pleistocene hominid fossils, but also because mitochondrial DNA sequence data were interpreted to indicate a single African origin for all modern human populations (Vigilant et al., 1991; Hedges et al., 1992; Stoneking, 1993). [Because the assumptions underlying the use of mitochondrial DNA in phylogenetic reconstruction are still debated (e.g., Awadella et al., 1999; Hagelberg, 2003; Schmitz et al., 2005), we shall deal here only with the relevant fossils and their morphology.]

Stringer and colleagues began by embracing specimens primarily from the late Pleistocene Levantine sites Jebel Qafzeh and Skhūl and the penecontemporaneous African sites Omo-Kibish (Ethiopia) and Border Cave (South Africa) as representatives of early H. sapiens. They then took these specimens as evidence of an African origin of modern humans to the exclusion of Neanderthals, which they accepted as an independent entity marked by numerous apomorphies. Although Stringer and colleagues did not provide any unifying features unique to the Levantine and African specimens, they allocated them to the species H. sapiens because all possess highly vaulted neurocrania, relatively small faces, and thin cranial bone. They additionally suggested that a feature unique to extant H. sapiens is that each superciliary arch is comprised of two distinct moieties (i.e., is bipartite).

This configuration contrasts, for example, with the Neanderthal brow, which is characterized as relatively uniformly tall and smoothly continuous from side to side. Recently, we have provided an alternative description of this region in extant H. sapiens (e.g., Schwartz and Tattersall, 1996b, 1999a, 2000b), as consisting of a swollen, anteriorly facing “butterfly-shaped” mounded region, of which the “body” coincides with the glabellar region and whose “wings” are delineated or undercut obliquely and superolaterally by a more planar lateral portion whose inferomedial extremity may coincide with the supraorbital notch/foramen.

Schwalbe (1901), who had earlier considered a bipartite brow a feature of H. sapiens, also mentioned that some specimens in his study did not express much if any morphological detail in the supraciliary/supraorbital region. We are fully aware of the range of expressed supraorbital detail in H. sapiens from faint to marked, but we also appreciate that “faintly developed” often characterizes females of various extant populations (see Schwartz, 2007b); even so, some trace of a glabellar butterfly is normally palpable. Consequently, because degree of expression of the “brow” and various other sexually dimorphic features of H. sapiens represent states of developmental continuum that spans between the hypo- and hyperostotic, we suggest that it is more biologically relevant to focus on the “glabellar butterfly” rather than the supraorbital notch as reflecting the presence of a bipartite brow. Otherwise, because the nonclosure of this notch to form a foramen is (primitively) widespread among anthropoid primates (Schwartz, 2007b), an especially pronounced notch, as in, e.g., KNM-ER 1813, may be erroneously taken as evidence of the bipartite configuration. Interestingly, although Stringer and colleagues are often perceived as radical opposites to the multiregional school of thought, they do converge in a mutual acceptance of a number of fossil specimens attributed to “early” (as opposed to “archaic”) H. sapiens. We will return to this in a moment.


If H. erectus indeed gradually transformed into H. sapiens, the question arises of when and where this occurred. Historically, this has been particularly perplexing because, when all Asian specimens of generally late to middle Pleistocene age were collapsed into the single taxon H. erectus, together with the apparent contemporary represented in Europe by the Mauer jaw, it appeared that the geographical range of this extinct hominid had once extended across most of the vast Eurasian continent. Indeed, this prediction seemed to have been validated with the 1960s discovery at Olduvai Gorge in Tanzania of a ca. 800 ka calvaria (OH 9), which many paleoanthropologists thought was remarkably robust, but yet with H. erectus-like features. Among those, most characteristics most frequently remarked in this context were the long and lower cranial vault, the massive and ledge-like brows, a somewhat distended occipital, and the “puffed out” cranial sides with margins delineated by rugose temporal muscle markings (e.g., see reviews in Santa Luca, 1980; Schwartz and Tattersall, 1999b, 2000c, 2003).

From Swartkrans, South Africa, a small (compared with Paranthropus) mandible that was originally attributed to “Telanthropus capensis” was allocated to H. erectus largely for the reason that something this small had to be Homo—and the only species of Homo available at the time was H. erectus (Robinson, 1953b). When, approximately two decades later, much older specimens were discovered in the area of Koobi Fora (especially the crania KNM-ER 3733 and ER 3883) on the east shore of Lake Turkana, Kenya, and then at Nariokotome on the west shore (the skull, mandible, and unusually complete postcranium of KNM-WT 15000), the range of this species thus appeared firmly to include the African continent and to span virtually the entire Pleistocene. Fossils subsequently recovered from the 1.8 Ma site of Dmanisi, Republic of Georgia, have been suggested as filling in the geographical “gap” in the record of H. erectus (Rightmire et al., 2006).

But although one might argue for the existence of a geographically and even temporally wide-ranging species H. erectus, this does not demonstrate that H. erectus actually “evolved” into H. sapiens. Indeed if, as by systematic necessity we must, we turn to the type specimen of H. erectus (the Trinil 1 skullcap) for the defining characters of this species, we are obliged at the very least to entertain the possibility that its array of apparent apomorphies [such as the smooth transition from superoinferiorly thin and nonprotruding but laterally continuous brows into the low and long frontal behind, the depressions on either side of bregma that give the false impression of a definitive elevation, the distinctly “V”-shaped occipital protrusion, the “lamination” of temporal bulges on the sides of the cranial vault, and the development of a neurocranium that when viewed from behind is wider than tall (Schwartz and Tattersall 2000c, 2003)] preclude it from being ancestral to any other known species of Homo, H. sapiens included (e.g., see Santa Luca, 1980; Schwartz and Tattersall, 1999b, 2000c, 2003).

No less important for this discussion is the possibility of greater taxic diversity represented among specimens attributed to “H. erectus.” For instance, although the allocation of East African specimens to the species H. ergaster (Groves and Mazák, 1975) might have seemed radical to some paleoanthropologists, the obvious morphological differences among the three specimens usually presented as representing this taxon, KNM-ER 3733 and ER 3883 and KNM-WT 15000, may additionally be systematically relevant (see Schwartz and Tattersall, 1999b, 2000c, 2003). Not to mention, of course, the fact that the type specimen of H. ergaster is a mandible (KNM-ER 992) that differs in dental morphology from WT 15000 (Schwartz and Tattersall, 2000c, 2003). In general, though, the tendency in paleoanthropology has been to preserve H. erectus as a geographically widespread and morphologically very variable species that precedes the emergence of H. sapiens (e.g., Lieberman et al., 2002).

Clearly, it will take time to climb out from under the shadow of Mayr's idea of the genus as a rank reflecting broad ecological adaptation, in this case represented by “modern” body proportions and bipedalism. Witness, for instance, Wood and Collard's (1999) thoughtful attempt to define the genus Homo on the basis of “striding bipedalism.” Although laudably trying to define, or at least to delineate, the parameters of this locomotor construct in the broad context of overall body proportions (the skeletal “forest”), this upward taxonomic focus neglected also to take into consideration specific differences of skeletal, especially pelvic and femoral, morphology (the skeletal “trees”). Nevertheless, inspection of pelvic and, especially, femoral morphological detail reveals that some of the very features that have for decades now been noted as specific only to australopiths (e.g., posterior orientation of ilia, long femoral neck, posteriorly directed lesser trochanter, and severe “carrying angle”; see review in Schwartz, 2007a) are present in at least two specimens attributed to Homo: the subadult KNM-ER 15000 from Nariokotome, Kenya (cf. Walker and Leakey, 1993; Schwartz, 2007a;) and a subadult partial skeleton from Dmanisi associated with skull D2700/D2735 (Lordkipanidze et al., 2007).

The point here is that any endeavor to trace the origin of H. sapiens directly to H. erectus depends on which specimens one includes in the latter species, which then directly affects scenarios of when, where, and, depending on the degree of speculation, even how H. erectus might have given rise to H. sapiens. But as already mentioned, if one diagnoses the species H. erectus on the basis of derived features preserved in the Trinil specimens [not only the calvaria (see above) but also the variably complete femora, which are laterally compressed throughout much of the shaft to a degree typical of tibiae (JHS, personal observation)], then the hypodigm of the species is reduced considerably. Indeed, it comprises primarily the specimens from Sangiran, which, when the petrosal region is preserved, present the clearly derived configuration of grooves for an arborizing rather than large and single sigmoid sinus (Schwartz and Tattersall, 2000c, 2003). What makes this latter hypothesis interesting is that some of the derived features of calvarial morphology (e.g., the posterior profile of the neurocranium, which is much wider than high) are also exhibited in Dmanisi skull D2282, whereas the derived condition of grooves for an arborizing sigmoidal sinus is present in D2280 (Schwartz and Tattersall, 1999b, 2000c, 2003). How other specimens that have been allocated to H. erectus since 1950 may cluster as morphs, and how these morphs may be related to one another (if indeed they all are), remains unclear to us, but the hint of a possible clade of which Trinil and Sangiran H. erectus plus specimens from Dmanisi are a part is neither biologically nor geographically implausible.

An alternative to the “H. erectus as ancestor of H. sapiens” notion has been proposed by Bermúdez de Castro et al. (1997), who have argued that the ancestry of H. sapiens lies in the species they named H. antecessor from specimens at the ca. 780 ka levels of the Gran Dolina at Atapuerca, northern Spain. Indeed, one specimen, a partially reconstructed subadult skull, and a zygomatic bone in particular, indicates to them that H. antecessor gave rise both to H. neanderthalensis and to H. sapiens. Their argument is that, as in juvenile and adult H. sapiens, the external infraorbital surface of the zygoma of their subadult H. antecessor is “indented” or depressed. However, as in adult H. neanderthalensis, an adult zygoma (Bermúdez de Castro and Arsuaga, 1999) from Gran Dolina is not thusly depressed. Bermúdez de Castro et al. interpret this array of subadult and adult morphological conditions of the zygoma as indicating that the indented infraorbital region in adult H. sapiens represents a neotenic retention during descent from H. antecessor, whereas the change from subadult to a different adult zygomatic configuration reflects ancestry and descent between H. antecessor and H. neanderthalensis.

We can certainly agree with Bermúdez de Castro et al. that H. sapiens and H. neanderthalensis represent different species. For if one interprets the phylogenetic relevance of Neanderthal morphology in a broader context, and without first imposing on it a scenario of ancestry and descent, specimens we would call Neanderthal emerge as unique and distinctive in cranial as well as postcranial morphology (see reviews in Tattersall and Schwartz, 1998, 2009; Schwartz and Tattersall, 1999a, 2003, 2006; Schwartz et al., 1999). For example, juvenile and adult Neanderthals are distinctive in developing a protruding, wedge-shape “snout” that is puffed out bilaterally on its sides because of maxillary sinus expansion (so much so that the medial orbital wall is typically involved); a vertically oriented growth of bone (“medial projection”) from the lateral wall of the nasal cavity that projects medially into the nasal cavity; and a well-defined and pitted suprainiac depression (Fig. 1). In addition, in frontal view, the adult Neanderthal lower face tapers medially from the zygomatic arches toward the alveolar margin, and the occipital bears a partially delineated nuchal crest, the superior margin of which is marked only by the superior border of the suprainiac depression, whereas its inferior “margin” exists because the nuchal plane undercuts the occipital plane (Fig. 1). Dentally, the major cusps on all permanent molars and the deciduous first molars are peripherally placed, thus opening up basins (trigon and talonid), whereas, in their lower counterparts, distinct mesial basins are bounded by thick paracristids and protocristids (Fig. 1). Postcranially, Neanderthal clavicles and pubic rami are relatively the longest among primates, the termini of the distal row of manual phalanges are broadly rounded (not tapered) and unusually dorsoventrally compressed, the groove for the teres minor muscle typically lies dorsally on the infrascapular border, the pubic symphyseal region is superoinferiorly tall and thin anteroposteriorly, and the greater sciatic notch is essentially uniformly “U”-shaped.

Figure 1.

La Ferrassie 1 cranium (left and middle columns), illustrating cranial features of Neanderthals: e.g., inferomedially tapering lower face in frontal view; anteriorly protruding, wedge-shape snout; medial projection from lateral wall of nasal cavity; well-defined and pitted suprainiac depression that is defined below by a partial torus; and undercutting of the occipital by the nuchal plane. Le Moustier adolescent maxilla (top right) and Krapina 58 mandible (lower right), illustrating typical Neanderthal dental features: e.g., in the upper molars a large internally situated protocone that truncates the trigon basin and that also expands the tooth distolingually, peripherally placed protocone and metacone; and in the lower molars a truncated trigonid with distinct paracristid that delineates distal to it a mesiodistally thin but buccolingually wide trigonid “basin” (crease, really), peripherally situated talonid cusps that subtend mesiodistally long and buccolingually broad basins, and round distal ends. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Yet although it is important to recognize that the numerous autapomorphies of H. neanderthalensis not only preclude it from the ancestry of any other known hominid species, but also presumably from any successful and biologically significant hybridization with them, it is also necessary to remember that H. sapiens and H. neanderthalensis were not necessarily the most closely related of known hominid sister species. Indeed, if one considers other cranial features that have at one time or another been put forth as being potentially distinctive of H. neanderthalensis, such as relatively superoinferiorly thick, double-arched brows that are continuous across glabella (Stringer et al., 1984; Stringer and Andrews, 1988), or a long, horizontally oriented parietomastoid suture (Schwartz and Tattersall, 1996a, b, 1999a), we are obliged to turn rapidly to other non-H. sapiens specimens in any attempt to discover our closest relative (Schwartz and Tattersall, 2003).

It may be reasonably argued that Neanderthals are members of a larger clade (Schwartz and Tattersall, 1996b, 1999a, 2002b, 2003, 2006; Tattersall and Schwartz, 1998) that also includes the Steinheim skull and the Sima de los Huesos hominids. Both have some apomorphies of H. neanderthalensis, but not all. Embracing this larger clade has nontrivial implications for the Gran Dolina specimens. For it is at best difficult to delineate specific features that would unite these Spanish fossils with the Neanderthal clade, especially given the marked differences in dental morphology (cf. Bermúdez de Castro et al., 1997; Bermúdez de Castro and Arsuaga, 1999; Falguères et al., 1999). Another alternative, no less intriguing, is provided by the detailed similarities between the lower dentitions of the relevant Gran Dolina specimens and the teeth that are preserved in the three mandibles from the penecontemporaneous Algerian site of Tighenif (Ternifine), originally designated as H. mauritanicus (cf. Arambourg, 1955; Hublin, 2001; Schwartz and Tattersall, 2003). Although Bermúdez de Castro et al. (2007) have pointed to a few differences between the Gran Dolina and Tighenif specimens to support retention of the species H. antecessor, this does not then mean that the original hypothesis “H. antecessor is ancestral to both H. neanderthalensis and H. sapiens” is thereby also validated.

If H. antecessor is not a junior synonym of mauritanicus, the next most likely hypothesis is that the two taxa represent closely related species, distinct from those otherwise known in Western Europe. If this hypothesis is viable, so too is the possibility that the affinities of the Gran Dolina and Tighenif hominid/s lie closer to H. sapiens than to other hominid species. For, as seen most clearly in Tighenif 2, the North African specimens bear a vertical keel along the mandibular symphysis that, in association with attendant morphologies, is characteristic of H. sapiens (see discussion below and Schwartz and Tattersall, 2000a, 2003). Because most of the Iberian Peninsula is climatically and faunally North African rather than European, it is not surprising that these seemingly far-flung specimens might represent the same hominid, or at least a closely related hominid pair. Questions that remain to be answered, however, include that of whether either fossil sample represents a population directly ancestral to H. sapiens, and, thus, which region represents our geographical site of origin.

With the matter of H. sapiens origins still in limbo, and if we can exclude as ancestor both H. erectus (however constituted) and H. neanderthalensis or a member of its larger clade, and possibly also the Gran Dolina and Tighenif specimens, a last resort might be the apparently cosmopolitan species H. heidelbergensis. But, as with H. erectus, this question boils down to how narrowly or broadly one casts the taxonomic net within the time range of ca. 600–300 ka. For, even though the crania from Kabwe, Petralona, Bodo, and Arago are frequently presented together or in some combination as representing H. heidelbergensis (Fig. 2), often forgotten in this exercise is that the type specimen of this hominid species is the Mauer mandible (Fig. 3). Consequently if, and only if, one can demonstrate a connection between the type specimen and of these or other skulls offered as heidelbergensis, can we make any case for this taxonomic allocation.

Figure 2.

Crania from Arago (top left), Bodo (top right), Petralona (bottom left), and Kabwe (bottom right). Note in all superiorly delineated margins of tall but not necessary anteriorly protrusive supraorbital margins that are essentially flat on their anterior surfaces, tallest circum-midorbit, and which “twist” superolaterally. Note also, e.g., differences in lower facial expansion or “puffiness” and length of nasal bones (and thus expanse of nasal aperture). Not to scale. am. [Color figure can be viewed in the online issue, which is available at]

Figure 3.

Comparison of Mauer (left column) and Arago 13 (right column) mandibles. Although the former is larger and more robust, both are uniquely similar in possessing: low and vaguely defined articular condyles; sigmoid notch crests that are deepest below the condyles; broadly arcuate but also posteriorly truncated gonial regions; posteriorly situated large mental foramina that lie at the termini of laterally thickened corporal tori; lateral corpora tori that delineate below them an anteroposteriorly long sulcus that is defined below by a thickly everted inferior corporal margin; and an inferior corporal margin that terminates in an inferior marginal thickening that serves as the “tethering point” of a upwardly arcing inferior margin that “lifts” the lower margin of the anterior surfaces upward. Note also similar disparity in morphology between P1 and P2 and rounded mesial and especially distal ends of elongate molars that bear buccally situated hypoconulids and filled-in, crease-like talonid basins. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Fortunately, the gracile Arago 13 mandible is sufficiently well preserved to show that, in details of both teeth and jaw, it shares unique morphologies with the more massive Mauer mandible (Schwartz and Tattersall, 2002a). Among these are the broad symphyseal region that arcs superiorly between two well-defined inferior tubercles, posterior to which the inferior margin of the corpus is thickened outwardly, creating a distinct somewhat horizontal sulcus above; the huge, low-lying and posteriorly situated mental foramen; the mandibular head lying below the level of the tip of the coronoid process (which was artificially shorted by animal gnawing); the very broadly rounded gonial region; the long and ovoid lower molars, of which M2 is the largest, with long but buccolingually quite truncated talonid basins; the hypoconulid that lies in all molars just buccal to the midline of the crown; the trigonid basin large only in M2 and M3; and the elongate first premolars that taper mesiodistally whereas the short and rather ovoid second premolars are wide buccolingually (Schwartz and Tattersall, 2000a, 2002a, 2003) (Fig. 3).

Because it seems clear that the Arago sample represents one single hominid (Schwartz and Tattersall, 2003), it is reasonable to extend the name heidelbergensis to all these specimens, which include the partial cranium Arago 21. This makes feasible comparison with specimens known only from crania (Schwartz and Tattersall, 2002a, 2003) (Fig. 2). Particularly compelling in Arago 21 is the configuration of the brow, which exhibits an unusual anteroposterior twist of its undifferentiated anterior surface, and a continuous superior margin that is defined by a distinct edge or corner (in contrast, for instance, to the smoothly “rolled” brow of Neanderthals and Steinheim); because the brow is not markedly protrusive anteriorly and especially superiorly, the post-toral sulcus behind is rather shallow. Favorable comparisons can be made between Arago 21 and cranial specimens from Petralona (Greece), Kabwe (Zambia), Bodo (Ethiopia), Dali, and Jinniushan (both China). With the exception of the Jinniushan specimen, the brows of all these specimens, including Arago 21, are very tall superoinferiorly, reaching their maximum height near mid-orbit.

Still, there are also differences among specimens within this assemblage (Fig. 2). The nasal bones are shorter and less protrusive, and the aperture is narrower and situated higher on the face in Arago, Petralona, Kabwe, Dali, and Jinnuishan, than in Bodo. Although possessing shorter nasal bones than seen in Bodo, Petralona and, especially, Dali display broad nasal apertures. In Petralona and Bodo, the lower face is swollen infraorbitally and toward the nasal region, whereas Petralona exhibits much more expansive sinus inflation than Kabwe does, not only in the face but also superiorly into the frontal bone. Further, internally, both Arago and Bodo present a H. sapiens-like configuration involving a well-excavated hypophyseal fossa that distinctly separates a horizontal and long anterior cranial fossa from the middle cranial fossa. In contrast, Kabwe and Petralona are similar to various australopiths as well as to apes and other anthropoids in that the anterior cranial fossa is virtually in the same plane as the obliquely oriented clivus, with a shallow and ill-defined hypophyseal fossa that barely distinguishes the two intracranial regions (Seidler et al., 1997; Schwartz and Tattersall, 2002a). Whether this similarity reflects primitive retention or secondary derivedness awaits further study.

Clearly, however, even though specimens referred to H. heidelbergensis would seem to be strongly united on the basis of their derived supraorbital morphology, it is also evident (especially on the basis of intracranial morphology) that this species subsumes a substantial morphological variety, possibly even suggesting the presence of more than one hominid morph. On the reasonable assumption that all specimens in this group are at least closely related, one might suggest a relationship of this potential H. heidelbergensis clade with the Neanderthal clade, but aside from presenting a brow that is thickened superoinferiorly to some extent (a descriptor that can be applied to a number of other hominids), we are hard put to delineate specific morphologies that support this hypothesis. More importantly, the same suite of apomorphies that distinguishes the “heidelbergensis” group would certainly also preclude any of its members from being ancestral to H. sapiens and would likely exclude all from a clade that included the latter.


The notion that an archaic phase of H. sapiens emerged from H. erectus, subsequently transitioning into “early anatomically modern” H. sapiens and then into “recent anatomically modern” H. sapiens, has occupied reams of text over decades. Frequently, still included in the archaic phase of this continually changing lineage-species H. sapiens are the specimens just discussed from Arago, Petralona, Kabwe, Bodo, Saldhana, Singa, Jinniushan, and Dali. Because the conception of archaic and “modern” phases of a single species is unusual in paleontology, even when “species” is taken as equivalent to “lineage” (should we actually be entertaining notions of archaic versus anatomically modern Tyrannosaurus rex, Plesiadapis tricuspidens, Omomys carteri, or Proconsul africanus?), its anomalous persistence in paleoanthropology most likely reflects the unusual history of the latter, dominated from the beginning by transformationist notions (e.g., Huxley, 1863; Mayr, 1950; Dobzhansky, 1955, 1962). Yet, clearly the persistence of this tradition does not do justice to what has in recent years become a vastly better-known human fossil record, the analysis and systematic interpretation of which should be based on the available morphological evidence, not a priori on debatable phylogenetic scenarios.

The first proposed evidence of a morphological transition from archaic to anatomically modern H. sapiens came from specimens recovered from the Levantine sites of Tabūn and Skhūl (McCown and Keith, 1939). Although McCown and Keith originally grouped the Tabūn and Skhūl hominids together as members of a highly variable but distinct species, Palaeoanthropus palestinensis, the placing of these specimens in a preconceived morphological continuum from “more archaic” to “more fully modern” H. sapiens has persisted in the literature, with the Tabūn material (possibly 122 ± 16 ka) often being regarded as more morphologically Neanderthal and the Skhūl specimens (stage 5, ca. 119 ka) as overall more anatomically modern (e.g., Howell, 1958; Stringer, 1978).

Two partial crania from the Omo Kibish Formation, Ethiopia [Omo I and Omo II, both now believed to be ca. 195 ka (McDougall et al., 2005)], were originally described by Day (1969), who regarded them both as variants of H. sapiens while admitting that Omo I was more modern in cranial shape and Omo II more archaic. On reconstructing the Omo I specimen, Day and Stringer (1982) aligned Omo I with H. sapiens and Omo II with H. erectus, but they later emphasized the dissimilarities between the former and other specimens allocated to H. sapiens (Day and Stringer, 1991).

Particularly following their description by Vandermeersch (1981), specimens excavated from another Levantine site, Jebel Qafzeh (early stage 5, probably between 100 and 90 ka), have typically been viewed as representing a single population that was almost, if not fully, anatomically modern (Howells, 1974; Stringer, 1974; Vandermeersch, 1981; Trinkaus, 1984). Also from the Levant, the Zuttiyeh (Galilee) frontal (possibly 200+ ka), which Keith (1927) suggested was morphologically Neanderthal-like, has subsequently been promoted as “archaic H. sapiens” (Vandermeersch, 1982) and as ancestral to a Qafzeh/Skhūl group (Wolpoff, 1989a) or even, by association with Skhūl V, as modern human (Zeitoun, 2001).

Singer and Wymer (Singer, 1953) described all of the material from Klasies River Mouth, South Africa (ranging from perhaps as old as 120 ka to ca. 60 ka, with the majority dated to at least 80 ka), as completely anatomically modern H. sapiens. Many paleoanthropologists continue to reiterate this conclusion, although others have challenged it (e.g., Wolpoff et al., 1994a). From our study of the Klasies River Mouth material, we concluded that two morphs were actually represented in the sample (Schwartz and Tattersall, 2000a).

The most recent discoveries attributed to early anatomically modern H. sapiens were found at Herto, Middle Awash, Ethiopia, and date to 154–160 ka (White et al., 2003). Of the three adult crania, only one (BOU-VP-16/1) was sufficiently complete to allow White et al. to present systematically useful morphology. Although they described the supraorbital region of this specimen as bipartite, they based their suggestion that the Herto adult is more modern than archaic on metrical comparisons with African fossils they accepted as representing both H. erectus and archaic H. sapiens, plus non-African modern humans and various Neanderthal specimens. The reconstructed partial skull of a child (BOU-VP-16/5, estimated to have been 6–7 years at death), found in fragments on the surface, was interpreted via craniometric analysis as similar to the adults from Herto, basically in not presenting an (unspecified) Neanderthal complex of features. Also central to the describers' conclusion was Herto's chronological intermediacy between older, archaic H. sapiens African sites and specimens from younger Late Pleistocene sites.

The absence of morphological detail in this brief overview of specimens attributed to H. sapiens once again reflects the peculiar history of paleoanthropology, in which chronology, not morphology, is typically regarded as the crucial determinant. In terms of morphology, what we glean is a general understanding that, at least craniodentally, any specimen classified as anatomically modern should be relatively thin-boned and small-toothed. It should have a neurocranium that is relatively vaulted (in frontal and lateral view), tall and parallel-sided (in rear view), and not too elongate (in profile) as well as a relatively small face (and a “reduced” brow), a maxillary incisura (canine fossa), a non-torus bearing occipital, and poorly developed muscle scars (particularly from the temporal muscles, whose lines should lie low on the parietals). In addition, the mandible should be small, gracile, and weakly muscle-scarred and should bear a swelling in the symphyseal region.

But although this description might seem on the face of it sufficient, it really does not approach the level of detail routinely demanded in systematic studies of other organisms. This is hugely to the detriment of our knowledge of the origin of H. sapiens, and it is not a situation that will be rectified easily or quickly. Nonetheless, we can offer a few suggestions.


Although we are still far from understanding the details of developmentally regulated genes and signal transduction pathways, or even the effects of hyper- versus hypoexpression of transcription factors on the development of morphology (which is all we have for fossils), the existence of a continuum from the molecular to the morphological is now well documented (e.g., Gerhart and Kirschner, 1997; Ronshaugen et al., 2002; Davidson and Erwin, 2006; Newman, 2006; Stern et al., 2006). It thus follows that any ontogenetic information should be explored because it reflects some aspect of this continuum in the emergence of final form. Here, we review the evidence relating to the development of the supraorbital and symphyseal regions as examples of this perspective and its relevance to understanding the unique morphology of H. sapiens.

The supraorbital region

A broad survey of anthropoid primates reveals that, regardless of the specific morphology of the adult supraorbital region, in very young individuals (i.e., even as late as M1 eruption) the generally smooth supraorbital region invariably gives little or no hint of the morphology that will eventually characterize the adults (dimorphic or otherwise) of that species (Schwartz, 1997 and unpublished data). Indeed, it is virtually impossible to predict from the neonate what its adult supraorbital morphology will be.

Adult conformations of this region vary widely. In papionins (see Schwartz, 1997), the adult supraorbital region presents a distinctly defined, superoinferiorly thin, and anteriorly projecting bar-like torus that runs essentially straight across from side to side. In African apes (Pan troglodytes, P. paniscus, and Gorilla spp.), the supraorbital region grows into an anteroposteriorly thick torus, with a markedly vertical component that produces a post-toral sulcus behind and which is indented over glabella to varying degrees (Fig. 4). The list of examples is endless, but from the undistinguished supraorbital region of the juvenile, myriad distinctive adult configurations emerge developmentally, ranging from the projecting, “goggle-like” circumorbital region of gibbons and siamangs, through the superomedially- and laterally raised partial circumorbital rims of Pongo, to the more fully but differently rimmed orbits of various New World monkeys, such as Cebus and Alouatta (Schwartz, 1997).

Figure 4.

Emergence of supraorbital morphology in Gorilla, from the nondescript to the specific (counterclockwise, from top left to top right), and in H. sapiens (right column), from 7 month fetus to adult (in the adult, arrows point to the medial “glabellar butterfly,” which, in this specimen, extends into the field of the laterally flatter plane). Not to scale. [Color figure can be viewed in the online issue, which is available at]

Even though one might expect living H. sapiens, the poster-child of paedomorphosis, to be the most neotenic of anthropoids in supraorbital morphology, it is not. Rather, one also finds little or no supraorbital embellishment not only in the small marmosets and tamarins, but also in various colobine monkeys (thereby ruling out size as a factor in lack of supraorbital morphology). Still, in H. sapiens, albeit closer to the onset of adulthood than in other anthropoids, the previously featureless region of glabella (Fig. 4; also Fig. 12) swells anteriorly (even if only slightly), and from each side of this mounded midline a wing-like swelling may also emerge, its inferolateral extremity terminating generally at the supraorbital foramen/notch (i.e., near the midpoint of the superciliary arch) and the superolateral extremity extending somewhat beyond this point laterally (Fig. 4). Altogether, this mounded protrusion forms a “butterfly”-like shape, which elsewhere we have described as a “glabellar butterfly” (e.g., Schwartz and Tattersall, 1996b, 1999a, 2002b; Tattersall and Schwartz, 1998; Antunes et al., 2000). On each side, the superciliary region lateral to the “butterfly wing” is flatter and more plate-like, with perhaps also a slight posterior declination to its surface (Fig. 4). This then constitutes the “bipartite” brow, the development of which is unique to H. sapiens compared with all living and almost all fossil primates.

Among fossil hominids, the available sample is adequate to allow us to track the emergence of the thick, double-arched, and laterally continuous brow seen in H. neanderthalensis (Schwartz and Tattersall, 1996b, 2002b, 2003). As in extant anthropoids, the supraorbital region of 3–4-year-old specimens (Engis, Pech de l'Azé, Roc de Marsal, and Subalyuk) is featureless in this species (Fig. 5). Only in slightly older individuals (La Quina child and Teshik Tash) can one discern with any confidence the beginning of supraorbital swelling, which developmentally expands bilaterally from the region of glabella (Fig. 5). In the Le Moustier adolescent, the laterally continuous brow typical of Neanderthal adults can already be detected (Fig. 5). What cannot be known, though, is whether the brow of this individual would have become much more anteriorly distended as it matured, as in La Ferrassie I and especially Guattari, or if it would have remained relatively low as in Gibraltar I or Krapina C (Skull 3) (Fig. 5).

Figure 5.

Growth sequence demonstrating emergence of the “double arched” and smoothly rolled supraorbital region typical of adult Neanderthals (La Quina child, top left; Teshik Tash juvenile, bottom left; Le Moustier adolescent, top right; and Krapina C (skull 3), bottom right). Not to scale. [Color figure can be viewed in the online issue, which is available at]

Nevertheless, what is important about these specimens is that they are consistent with a picture of taxon-specific, postnatally achieved, supraorbital morphology. Consequently, although we cannot reliably infer adult supraorbital form from the study of juvenile hominids, such as those from Herto, Modjokerto, Taung, Skhūl I, and Dikika, we can state with some confidence that the specific supraorbital morphology of any adult was acquired during growth from a previously featureless frontal bone. With this in mind, we turn to fossil specimens that have been considered anatomically modern H. sapiens, in search of those that present a bipartite brow.

Among the fossils that most of us were taught were uncontestable early representatives of our species are specimens from Qafzeh and Skhūl. From Qafzeh, the specimen most frequently cited and illustrated is the fairly complete skull Qafzeh 6. Yet this specimen lacks a bipartite brow, possessing instead a superoinferiorly somewhat tall brow that is anteriorly low and mounded, and continuous across an equally tall glabellar, region. Thus, although the neurocranium of Qazeh 6 is rather globular, and relative to it the face is not massive, this specimen conspicuously lacks the one particular apomorphy that would cement its allocation to H. sapiens (Schwartz and Tattersall, 1996b, 2000b) (Fig. 6).

Figure 6.

Comparison of Qafzeh 9 (left column), Qafzeh 11 (middle column), and Qafzeh 6 (right column). The former two specimens appear to have had a bipartite brow, which is clearly lacking in Qafzeh 6. See text for detail. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Less frequently discussed and illustrated are the broken, but reasonably reconstructed, adult Qafzeh 9 and the less complete and subadult Qafzeh 11 crania (Fig. 6). Yet for our discussion here, these specimens are interesting and frustrating in equal measure. The adult Qafzeh 9 specimen is so damaged that we can only surmise that the fragment that is apparently correctly placed medially in the superciliary arch on the left side is mounded or somewhat swollen, suggesting the “butterfly” configuration. Bone laterally in both superciliary regions is more clearly flat and plate-like, also suggesting that Qafzeh 9 possessed a bipartite brow. Although the subadult Qafzeh 11 is less damaged, and thus presents a more pristine supraorbital region than Qafzeh 9, crucial morphological detail in this region was not yet fully developed. Nevertheless, we feel confident in identifying the glabellar “butterfly-shape” swelling characteristic of a bipartite brow.

Along with Qafzeh 6, Skhūl V has often been presented as a representative of “early anatomically modern” H. sapiens (Fig. 7). Like Qafzeh 6, the skull is generally rounded and vaulted in profile, and the largely reconstructed face does not present itself as excessively massive relative to the neurocranium. Also as in Qafzeh 6, what is preserved of the supraorbital region of Skhūl V does not present a bipartite configuration. Interestingly, the brow of Skhūl V is less tall superoinferiorly, and much more anteriorly protrusive in the form of a torus, than its counterpart in Qafzeh 6 (Fig. 7). The Skhūl II frontal fragment (which retains the glabellar region together with some of the left supraorbital region and most of the right) and the small portion of the left supraorbital region of Skhūl IV are strikingly similar to Skhūl V (Fig. 7). However, in Skhūl IX, the preserved right supraorbital region, with most of glabella, is superoinferiorly thin but arced rather than relatively straight, and barely protrudies anteriorly (Fig. 7). Skhūl VII retains the lateral portion of the right orbital region, but the anterior surface of the superciliary arch is missing; the brow appears to have been arcuate and may have been taller than in the other Skhūl specimens. Predictably, the supraorbital region of the juvenile cranium Skhūl I is featureless. Thus, whatever the exact morphology of each of the Skhūl specimens, there is no trace of a glabellar “butterfly” in any of them.

Figure 7.

Comparison of Skhūl V, Skhūl IX, and Qafzeh 6 (top, left to right), and LH18 (Ngaloba), Skhūl II (lower middle, anterior and oblique views), and Liujiang (bottom, left to right). Although the superciliary region in Skhūl V is the most protrusive anteriorly (followed by Skhūl II) and is the least tall superoinferiorly, neither it nor the other specimens present a bipartite brow. The “crease” in the right supraorbital region of Skhūl V is due to damage. Note also the distinctly teardrop-shaped bulge, rather than inverted T-shape of the H. sapiens “chin.” See text for detail. Not to scale. [Color figure can be viewed in the online issue, which is available at]

With regard to other specimens that have been identified as “early anatomically modern” H. sapiens, we could confidently detect a glabellar butterfly only in the Liujiang cranium (>67 ka, possibly 101–227 ka) (Fig. 7). In the otherwise distinctive LH 18 (Ngaloba) calotte (108–129 ka) (Fig. 7), there appears to be something resembling this structure, the more robust and superoinferiorly thicker lateral portion forming an antero-obliquely facing plane (Schwartz and Tattersall, 2003). The variably complete crania of Omo Kibish I and II, Singa, and Jebel Irhoud I, and the Klasies River Mouth frontal fragment (Figs. 8 and 9), are broadly contemporaneous with, or older than, the Liujiang and LH 18 specimens and have been suggested as at least representing a precursor to anatomically modern H. sapiens. None of these specimens, however, displays a supraorbital configuration that could be described as bipartite, or as possessing a butterfly-shaped glabellar region.

Figure 8.

Comparison of Omo Kibish I [left and middle columns; frontal, mandible (anterior, right profile, inferior views), and lateral view of occipitoparietal region] and Omo Kibish II (anterior and right lateral cranial views, right column). The two may represent different morphs, but they are similar in lacking a bipartite brow. The symphyseal region of Omo Kibish 1 clearly lacks H. sapiens's features. See text for detail. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Figure 9.

Variably complete crania of Jebel Irhoud 1 (top left), Dar es Soltane II (bottom left), Border Cave 1 (top right), Singa (bottom right), and Klasies River Mouth frontal 6103 (middle). None displays evidence of a bipartite brow. Not to scale. [Color figure can be viewed in the online issue, which is available at]

We have not been able to examine the adult cranial specimen (BOU-VP-16/1) from the somewhat older site of Herto, Middle Awash. However, we need to mention it because it has been allocated not only to H. sapiens, but also to a new subspecies, H. sapiens idaltu (White et al., 2003). As is clear from the excellent published images, BOU-VP-16/1 has a more robust and superoinferiorly taller brow (including the lateralmost extremity) than any fossil in which we can confidently describe a bipartite brow replete with glabellar “butterfly.” As seen in the published photographs, the more completely preserved right superciliary arch of BOU-VP-16/1 presents a slightly postero-obliquely oriented “crease” that delineates medial and posterior moieties, with the medial portion more anteriorly facing and the posterior portion inclining posteriorly. Atypical of any bipartite brow, however, is that the anterior surface of the medial supraorbital moiety of BOU-VP-16/1 is vertically flat from top to bottom and appears to “twist” toward its medial extremity so that it ultimately faces rather laterally. The superior margin of this anterior moiety also bears a distinct margin that continues onto the glabellar region, thus partitioning the two supraorbital sections as separate entities.

Among chronologically younger specimens that have been considered definitively anatomically modern H. sapiens are the incomplete crania Border Cave 1 and Dar es Soltane II (Fig. 9). Although we have in the past agreed with this interpretation (Schwartz and Tattersall, 2003), our reassessment of these specimens has made us much more tentative now in both cases. Among the variably complete Pleistocene crania that we also viewed as morphological H. sapiens in our 2003 study, we still confidently include in our species the relatively recent specimens from Abri Pataud, Brno, Chancelade, Combe Capelle, Cro-Magnon, Dolni Věstonice, Engis (the adult), Grimaldi, Isturitz, Mladeč, Pavlov, Predmostí, Svitavka, Tuinplaas, Velika Pécina, Vogelherd, Wajak, Zhoukoudian Upper Cave, and Zláty Kůn (Fig. 10). Unaligned with typical H. sapiens on supraorbital conformation are the very recent specimens from Fish Hoek and Boskop (Schwartz and Tattersall, 2003) (Fig. 11). The latest estimate of 6891 ± 37 BP for Fish Hoek (Stynder et al., 2009) makes this atypicality all the more intriguing.

Figure 10.

Crania and associated mandibles of various fossil H. sapiens: Abri Pataud, Mladeč 1, and Dolni Věstonice XV (top, left to right); Grimaldi 5, Cro-Magnon 2, and Wadjak 4 frontal and Wadjak 23 mandible (bottom, left to right). Note variable expression both of a bipartite supraorbital configuration and of an inverted T-shaped chin. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Figure 11.

Boskop (left column, including left partial mandible in symphyseal, left lateral, and inferior views, and Fish Hoek (middle and right columns). Note nonbipartite supraorbital configuration. Note in anterior and lateral views of the mandible the smoothness of the symphyseal region and, in inferior view, the relatively uniform anteroposterior thickness of the bone from the symphyseal region onto the corpus. Not to scale. [Color figure can be viewed in the online issue, which is available at]

The “chin”

From at least the time of Blumenbach's (1969) treatise on features that distinguish H. sapiens from other living animals, the human “chin” has received particular attention from comparative biologists (Schwartz and Tattersall, 2000a). Unfortunately, the focus has typically been on the presence of some (any) anterior protrusion in the region of the mandibular symphysis. This has led to such unhelpful comments as that the only living mammals that develop a chin are humans and elephants (Enlow, 1982). In the search for evidence of the emergence of anatomically modern from more archaic H. sapiens, any three-dimensional perturbation of a symphyseal surface that in other mammals is typically smooth or flat (whether vertical or posteroinferiorly slanted) tends to be taken as evidence of an incipient chin. In light of this emphasis on simple anterior protrusion, rather than on morphological detail, even modern-day humans who fail to achieve the requisite anterior growth of the mandible have been seen as anatomical curiosities (Enlow, 1982). Nevertheless, although the form and development of the feature that is truly unique to H. sapiens, the chin, has been illustrated and described in textbooks for centuries, its systematic and phylogenetic significance has been obscured by the endeavor to create a sequence of morphological transformations from extinct to extant humans.

As reviewed elsewhere (Schwartz and Tattersall, 2000a; Schwartz, 2007b), the major features of the human chin are visible prior to the fifth fetal month, well before the right and left sides of the mandible fuse along the symphysis (Fig. 12). Specifically, early on, there is an anteriorly raised inferior symphyseal margin that continues laterally for some distance, creating the depression that in adults is identified as a mental fossa. Before birth, and continuing afterward until coalescence is complete, the right and left raised symphyseal margins fuse from top to bottom (Fig. 12). The result is an inverted “T” configuration in which the stem of the “T” is represented by the raised but now joined right and left symphyseal margins, and the arms on either side of the fused symphysis are the thickened inferior margins. The mental fossae lie on either side of the stem of the inverted “T” and above the thickened inferior margin.

Figure 12.

Growth series illustrating the configuration of the symphyseal region in H. sapiens. Five-month fetus (upper left): note everted symphyseal and inferior margins and large and deep bilateral mental foramina; also, the symphyseal sides have only begun to fuse superiorly, a process that will continue inferiorly. Two- to 3-year old (bottom left): anteriorly, note everted inferior margins with mental fossae above and in the midline a modest triangular swelling of bone; inferiorly, note that the symphyseal region is thicker anteroposteriorly than the corpora immediately lateral to it. In the 5-year-old (middle) and adult (Abri Pataud) (right), note that that the thickness is maintained inferiorly, whereas, in the symphyseal regions, the inverted T is variably expressed. Pre-adult specimens are in uncatalogued teaching collections, American Museum of Natural History. Not to scale. [Color figure can be viewed in the online issue, which is available at]

With continued growth and bone remodeling, and the emergence of the first set of anterior teeth, the mandibular alveolar margin grows superiorly away from the tip of the stem of the inverted “T.” The crisp and thin rami of the inverted “T” also thicken, especially along the inferior margin and at the juncture of the stem and the arms, forming what in the adult is commonly identified as a “mental trigon” (Fig. 12). Sometimes the lateral extremities of the arms also thicken and are then referred to as mental tubercles. Although the superior limit of the stem often becomes less distinct, it never extends to the alveolar rim because its development is from the basilar, Meckel's cartilage-derived bone of the mandible, whereas alveolar bone derives from the neural crest-derived mesenchymal cells that give rise to the teeth and their attendant soft and hard tissue structures (Ten Cate and Mills, 1972). Of further note in H. sapiens is that, from the juvenile through the adult, the symphyseal region is noticeably thicker anteroposteriorly than the bone of the corpora on either side when viewed from below (Fig. 12). Thus, in H. sapiens, the anterior region of the mandible differentiates early on in fetal development into the basic configuration that will be retained to varying degrees of crispness in the adult.

In anthropoid primates, the anterior region of the neonate mandible is essentially as featureless as it will remain in the adult (Schwartz, 1997, 2007b; Schwartz and Tattersall, 2000a). Indeed, as seen for instance in the Taung child and Swartkrans SK 3978, the featureless symphyseal region is consistent with the equally featureless symphyseal regions of australopith adults, regardless of the morph they represent (Fig. 13). Consequently, it is reasonable to conclude that if the adults of a species present a morphologically blank symphyseal region, the juveniles did too.

Figure 13.

Anterior views of mandibles of young and adult australopiths illustrating their characteristically featureless symphyseal regions. Taung child, Swartkrans SK3978 child, and Makapansgat MLD 2 subadult (top, left to right); Swartkrans SKW 5, Peninj, and SK 12 (bottom, left to right). Note also, as in juvenile anthropoids generally, the supraorbital region of Taung is essentially featureless. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Although some adult Neanderthal specimens may exhibit some anterior symphyseal topography, it is significantly absent in both the Le Moustier adolescent and known juveniles, Gibraltar 2 (Devil's Tower), Pech de l'Azé, Roc de Marsal, Amud, and Teshik Tash (Schwartz and Tattersall, 2002b, 2003) (Fig. 14). Indeed, the vertically oriented symphyseal regions of these specimens are similarly broad and shallowly curved from side to side, with a smooth profile across the midline. This configuration is retained in a number of adults, notably La Ferrassie I, La Chapelle-aux-Saints, and various mandibles from Krapina. But in other adult specimens, the midline in lateral profile may slope down and back (e.g., Tabūn C1); the anterior teeth may protrude anteriorly farther than the bone below (e.g., some Krapina specimens); or a subincisal fossa may produce apparently protruding anterior teeth and, immediately below (but well above the inferior margin), a gentle bulge or swelling (e.g., Spy 1, Shanidar 1) (Fig. 14) (also illustrations in Schwartz and Tattersall, 2002b). Also of note, from juvenile into adult, is that when viewed from below the Neanderthal symphyseal region not only is broad and variably straight across or shallowly arced from side to side, but is also typically (though not invariably, in Regourdou and Kebara, for example, bone thickness is consistent) thinner anteroposteriorly than the bone of the corpora on either side (Fig. 14). In any event, from a developmental perspective, the variability in details of adult Neanderthal symphyseal configuration clearly emerged with growth from a morphologically undistinguished symphyseal surface. Thus, no Neanderthal adult specimen with a bulge (invariably well above the inferior margin) provides any insight into the “evolution” of the human chin.

Figure 14.

Growth series from Neanderthal children to adult in anterior and inferior views to illustrate characteristic symphyseal configuration. Amud infant, Pech de l'Azé child, and Le Moustier adolescent (top, left to right); Krapina 58 and La Ferrassie 1 adults (bottom, left and right). Note variability in anterior “overhang” of anterior teeth versus smooth surface in adults; note inferiorly, especially in pre-adults, that the symphyseal region may be thinner anteroposteriorly than the bone of the corpora to either side. See text for detail. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Given the obvious differences between H. sapiens and Neanderthals, the mandible of the Skhūl I child is of particular interest (Figs. 14 and 15). For, although partially reconstructed in the symphyseal region, the preserved bone on the inner surface demonstrates that it was very broad and gently arced from side to side; it was also thin anteroposteriorly and thicker farther along the corpora. Although not extending across the midline, the bone preserved externally on the right side is smooth and shows no sign either of a mental fossa or of a rise toward the symphysis. In addition to this atypical (for H. sapiens) symphyseal morphology, the exposed right M1 presents not only the peripherally placed cusps and broad and long talonid basin characteristic of Neanderthals, but also a well-developed centroconid, as in the Tabūn C1 M1 (Schwartz and Tattersall, 2003) (Fig. 14). These comparisons are particularly interesting in light of the fact that, even though slightly distorted, the outline of the cranial vault viewed from behind and the morphological details of the occipital region are not typical of Neanderthals, as is also the case with the adult specimen Skhūl V.

Figure 15.

Partial skull and mandible of Skhūl I child compared with inferior view of mandible of Tabūn C1 (lowermost right) and close-up of Tabūn C1's left M1 (insert with mandible of Skhūl I). The numerous views of the Skhūl I mandible demonstrate that reconstructed Neanderthal-like shape of the symphyseal region and its thinness anteroposteriorly relative to the thicker corpora is accurate. Also note evidence of a centroconid in the Tabūn C1 molar, which is pronounced in the Skhūl I child. Note also in Skhūl I, as in all juvenile anthropoids, the featureless supraorbital region. Not to scale. [Color figure can be viewed in the online issue, which is available at]

With regard to the Skhūl adults, the Skhūl V mandible (Fig. 16; also Fig. 7) is damaged along the incisor roots, but the reconstruction of these teeth as slightly anteriorly inclined, and of the subincisal fossa immediately below, seems to be accurate. In profile, an anterior bulge emerges below the subincisal fossa, reaching its most anterior extent around the inferior margin. In front view, this bulge is teardrop-shaped, and it transitions smoothly into the surrounding bone all around it (Schwartz and Tattersall, 2000a, 2003). In inferior view, the Skhūl V mandible is clearly uniformly thick anteroposteriorly throughout the broad symphyseal region, becoming somewhat thicker more laterally along the corpora.

Figure 16.

Skhūl IV, II, and V (top, left to right) and Border Caves 2 and 5 (bottom, left and right). Note variably developed teardrop-shaped bulge in symphyseal region of Skhūl specimens and anteroposteriorly uniformly thick bone from symphyseal region onto corpus in all. The symphyseal region of Border Cave 5 is essentially featureless, whereas that of BC 2 is enigmatic in presenting a bulge versus a developmental derivative of an inverted T. Not to scale. [Color figure can be viewed in the online issue, which is available at]

The less well-preserved Skhūl IV mandible is broken between the right C1 and P1, but the intact symphyseal region shows the same curvature and anteroposteriorly uniform thickness as Skhūl V (Fig. 16). In Skhūl IV, the anterior teeth are not truly forwardly inclined, but are instead undercut by an extremely shallow subincisal fossa below which, in left profile, a bulge not unlike that in Skhūl V emerges. In anterior view, some damage notwithstanding, the bulge is broader than in Skhūl V and less well defined, but it too merges smoothly with the surrounding bone. Only the anterior portion of the mandible of Skhūl II (Fig. 16) is known (better along the right corpus than the left), but the relative uniformity of anteroposterior thickness throughout this region is preserved. Unfortunately, plaster reconstruction occupies much of the upper portion of the anterior surface of the symphyseal region, but it is clear, as seen particularly on the right side, that the surface between the mental foramen and the front of the jaw is not hollowed out or in any other way fossa-like. Further, rather than being thickened with some anterior distension, the inferior margin of the symphyseal region curves smoothly down and then posteriorly.

Although they do not present a configuration of the symphyseal region comparable with the inverted “T” of H. sapiens, the Skhūl adults also do not exhibit the typical anteriorly thin, and broad and gently arced, symphyseal region of Neanderthals (cf. Figs. 14 and 16). Thus, in the symphyseal region, as in the configuration of the brow, Skhūl adults are characteristic neither of H. sapiens nor of Neanderthals. What is more, although the Skhūl I mandible might on its own be interpreted as a juvenile representative of the population represented in the adult form by Tabūn C1, features of the vault make this unlikely. What can be said with some confidence, though, is that the Skhūl I mandible is not very plausibly the basis from which adult Skhūl mandibular morphology emerged. All in all, this assemblage gives us a very good example of the difficulties of sorting out the systematics of a morphologically diverse but evolutionarily close-knit group.

To add yet another wrinkle to this puzzle, the presence of a human-like chin in the Tabūn II mandible, which is larger and taller in both corpora and rami than the Tabūn C1 mandible, has been actively debated (see review in Schwartz and Tattersall, 2000a). As also seen in profile in various Neanderthal specimens (see above), the anterior teeth of Tabūn II are set well forward of the symphyseal region below. This anterior tooth “displacement” is emphasized further by a superoinferiorly tall subincisal fossa, from the lower part of which a modest bulge emerges anteriorly. Unfortunately, the inferior margin across the entire symphyseal region is reconstructed in plaster. However, on each side of the reconstructed portion, there is a distinct tubercle on the inferior margin. When viewed from below the mandible appears to be uniformly very thick anteroposteriorly throughout the somewhat tightly curved symphyseal region, and it is perhaps slightly thinner along the corpora. An additional piece of the mystery is seen in the full exposure of the third molars in front of the anterior margins of the rami (although on both sides this margin is indented posteriorly by a marked preangular sulcus that accentuates the exposure of the M3's).

Just as the cranial specimens from Qafzeh probably represent more than one morph, the mandibles from this site suggest at least two different morphs. The mandible of the juvenile Qafzeh 4 is unfortunately somewhat crushed and slightly incomplete in the symphyseal region, though its anterior portion clearly lacks the inverted “T” configuration and attendant mental fossa characteristic of H. sapiens (see images in Schwartz and Tattersall, 2000a, 2003), something that would be expected to have begun forming in a H. sapiens of this individual's age (approximately 5–6 years, given the eruption of the first molar). Instead, the anterior surface of this mandible is featureless. Yet the slight medial crushing of the partial left corpus cannot obscure the fact that, as in H. sapiens, when viewed from below the mandible would have been thicker anteroposteriorly in the symphyseal region than lateral to it.

The partial adult mandible Qafzeh 7 (Fig. 17) presents a short and shallow subincisal fossa, below which a fairly large but low teardrop-shape bulge flows smoothly and gently into the surrounding bone; the inferior margin of the symphyseal region is missing, but in lateral view it is clear that this bulge recurves posteriorly. The mandible is sufficiently intact in the symphyseal region to show in inferior view that the bone is uniformly thick anteroposteriorly in the midline, and may also have been so farther laterally. The symphyseal morphology of the Qafzeh 8 fragment is undecipherable.

Figure 17.

Qafzeh 7, 9, and 11 (top, left to right). Note tear-drop shaped bulge on symphyseal region of Qafzeh 7 and apparent uniform thickness of bone anteroposteriorly from symphyseal region to corpora; apparent outward deflection of bone of symphyseal region of Qafzeh 9; and thicker symphyseal region on Qafzeh 11. Also note dental similarities between Qafzeh 9 and 11, but not between them and Qafzeh 7. See text for detail. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Although cracked and reconstructed, the adult Qafzeh 9 (Fig. 17) mandible is clearly different from Qafzeh 7. The anterior teeth of Qafzeh 9 are somewhat anteriorly displaced, creating a short subincisal fossa below their roots, but there is no sign of a teardrop-shaped swelling. Rather, pieces of undeformed bone on either side of the midline clearly curve anteromedially, as if rising toward a midline keel (Schwartz and Tattersall 2000a, 2003). What is preserved of the inferior margin suggests that the symphyseal region would have been thickest inferiorly. Thus, although it is impossible to describe the configuration in inferior view with confidence, we believe the most straightforward interpretation of symphyseal morphology is that it bore the inverted “T” and the associated mental fossae characteristic of H. sapiens.

The partially reconstructed mandible of the subadult, Qafzeh 11, presents a thin midline keel whose uppermost extremity lies well below the alveolar crest of the central incisors (Fig. 17). This keel is confluent with a symphyseal swelling that fans out as it approaches the inferior margin (rather than curving medially to form a teardrop shape); on each side is a shallow depression. In lateral view, this triangular swelling projects anteriorly above the inferior margin, although, primarily because the inferior margin is elevated above, a small but distinct tubercle is seen on the left side. In inferior view, the mandible is thickest anteroposteriorly in the symphyseal region, with thinner bone laterally along the corpora. We believe that Qafzeh 11 exhibits a subtly configured inverted “T,” with attendant mental fossae.

In contrast to the Skhūl mandibular specimens, among which it appears most likely that Skhūl 1 does not represent a juvenile form of the adults from that site, it is possible to hazard that Qafzeh 4 is the juvenile of whatever adult morph Qafzeh 7 represents. Certainly, it is neither a juvenile version of Qafzeh 9 and 11, nor of a Neanderthal: the symphyseal region of Qafzeh 4 lacks any of the morphological adornment characteristic of H. sapiens; yet, when viewed from below, it is not thinner anteroposteriorly than the bone lateral to it along the corpora as in Neanderthals.

The most compelling reasons for associating Qafzeh 4 and 7 come from M1 morphology (Schwartz and Tattersall, 2003). In both specimens, the protoconid sits more mesially than the metaconid; the mesiodistally long entoconid courses from just mesial to the midline of the distal margin so that it lies across from the apex of the metaconid; and the relatively small but bulbous and slightly buccally placed hypoconulid juts out from the perimeter of crown. The M1's of Qafzeh 9 and 11 are mesiodistally longer on the buccal than on the lingual side, the entoconid is smaller than and lies opposite the hypoconid, and the wedge-shape hypoconulid is positioned distally in the midline of the crown (Fig. 17). This in itself is interesting, but beyond providing insight into the hominid assemblage represented at Qafzeh, the association of Qafzeh 4 with Qafzeh 7 permits us to suggest that the adult individuals from Skhūl may also be aligned with this potential hominid morph.

Among specimens penecontemporaneous with Skhūl and Qafzeh, only one, a mandible fragment from Klasies River Mouth (AP 6222), exhibits an anteroposteriorly thickened symphyseal region bearing the inverted “T” configuration (Fig. 18). Other Klasies River Mouth mandibular specimens that preserve some, if not the full extent of the symphyseal region [KRM1/AP 6100, AP 6101 (21776), AP 6102, and AP 6223], present neither the H. sapiens structure nor the teardrop-shape configuration seen in the Levantine non-H. sapiens morphs (Fig. 18). Rather, in these Klasies River Mouth specimens, the symphyseal region tends to be rather straight and vertical in lateral profile. Interestingly, although this region [best preserved in KRM1/AP 6100, AP 6101 (21776), and AP 6223] is strongly curved laterally (and is not as wide from side to side or as straight across as in the probably chronologically older Jebel Irhoud 3 juvenile mandible or the Neanderthals), when viewed from below (Fig. 18) it is thinner anteroposteriorly than the corpora immediately to its sides, as in Neanderthals. Although weathered and slightly damaged, the yet older Omo 1 mandible is strikingly reminiscent of these latter specimens from Klasies River Mouth, i.e., it lacks morphological detail, and the strongly curved and uniformly thick symphysis is narrower than the corpora to its sides.

Figure 18.

Anterior and inferior views of mandibles from Klasies River Mouth. KRM 6222, 6100, and 6101 (top and middle rows, left to right); KRM 6102 (anterior view only); and KRM 6223 (bottom row, left to right). Note in KRM 6223 a distinct triangular development in the symphyseal region and, in inferior view, that the bone of the symphyseal region is somewhat thicker anteroposteriorly than the corporal bone immediately lateral to it. In the other mandibles, although the symphyseal region in some presents various three-dimensional configurations, none approaches the inverted T-shape and, inferiorly, one sees that the bone of the symphyseal region is either thinner anteroposteriorly or of equal thickness compared with the thickness of the bone to its side. Not to scale. [Color figure can be viewed in the online issue, which is available at]

Of the two Border Cave mandibles, only one [BC 2 (A1102), not BC 5] presents a symphyseal keel that swells laterally and anteriorly toward the inferior margin. However, as in BC 2, the BC 5 mandible is quite strongly curved in inferior view, and the corpora are as thick anteroposteriorly as the uniformly thick symphyseal region (Fig. 16). Thus, among the African specimens just discussed, we feel confident in identifying the morphology typical of living H. sapiens only in Klasies River Mouth AP 6222.

Among hominids of more recent age, all diagnostic features of H. sapiens mandibular symphyseal morphology are unambiguously present in specimens from Abri Pataud, Brno, Chancelade, Combe Capelle, Cro-Magnon, Dolni Věstonice, Grimaldi, Haua Fteah, Isturitz, Predmostí, Svitavka, Vogelherd, Wajak, Zhoukoudian Upper Cave, and Zlaty Kun (Fig. 10). Apparently excluded from this group are the very recent specimens from Fish Hoek and Boskop, neither of which exhibits any detailed symphyseal morphology (Fig. 11). Again, the redating of Fish Hoek to less than 7 kya (Stynder et al., 2009) adds another element to the puzzle of human diversity in the Holocene.

In this section, we have focused on two of the more striking features of cranial osteology that seem to characterize the living species H. sapiens. Still, we are fairly sure that, on close inspection, we might well discover similar patterns of distribution and variation in those other osteological features that have at one time or another been claimed as unique to our species. Clearly, whatever the structure under consideration, one might reasonably expect developmental outliers to exist within any species, and, as a result, we are unable to say with any confidence whether, for example, the lack of the modern human chin and glabellar butterfly in the Fish Hoek specimen indicates its exclusion from H. sapiens in a broader biological sense. If we were to insist on tossing out all characters that are not very tightly correlated indeed with other features used in the analysis, we might find ourselves totally bereft of any characters to work with; still, on strictly morphological considerations (which is all we have that is directly systematically relevant), we can do no more than say securely that the Fish Hoek specimen and others like it (including the Omo 1, Herto, Border Cave 5, Boskop, and Klasies specimens, except for AP 6222) belong to the same close-knit clade as modern H. sapiens.

Because at least the vast bulk of living humans are plainly derived in these features, whereas the fossils are not, on the basis of morphology, we are able neither definitively to include such specimens in H. sapiens, nor assuredly to exclude them from the living species. Our inclination is to err on the side of inclusion, which, in operational terms, should not greatly distort the overall evolutionary pattern we perceive among our closest relatives (it has the effect of simplification, leading to a more robust result than potentially needless complication). We think it is important to remember that there must be a systematic signal buried somewhere in the mass of morphologies with which our immediate clade presents us. But for the time being, it seems, we are limited to a fairly gross level of resolution in clarifying that signal.


We have focused in the preceding section on configurations of the supraorbital and mandibular symphyseal regions because, even though the number of subadult fossils is small, in these features, available specimens provide considerable insight into the emergence of, as well as the differences among, the more frequently preserved adult morphologies. We look forward to the day when a burgeoning understanding of developmental continua from the molecular level through the emergence of structure (see review in Schwartz, 2009a) will provide a better foundation for the comparative study of a wider array of adult morphologies. Nonetheless, it is already very clear that living H. sapiens is an anatomically highly distinctive entity in many osteo-dental respects. It is with this distinctive entity that we must necessarily compare any putative fossil members of our species. Partly because of the peculiar history of interpretation of the hominid fossil record already adumbrated and partly because of operational difficulties stemming from the fact that in the case of H. sapiens and its putative close fossil relatives we are dealing with a phylogenetically very close-knit group, there is currently no satisfactory or even agreed-on osteological diagnosis of our species. Nevertheless, we and others have previously pointed to a list of features that are potentially significant as autapomorphies of H. sapiens (Day and Stringer, 1982; Stringer et al., 1984; Schwartz and Tattersall, 1999a, 2000b, 2003; Lieberman et al., 2002).

Such features include the following, in addition to the bipartite brow and chin: extension of a tall, sheet-like vaginal process to the lateral margin of the tubular ectotympanic, and its approximation to the mastoid process; very lateral placement of the styloid process with the stylomastoid foramen lying posteromedially at its base; a narrow and high occipital plane; retention into the adult of a mound-like arcuate eminence on the petrosal; and segmentation of cranial sutures, with deep interdigitation of various segments. To these features, we can add an incompletely closed off subarcuate fossa and an inferior orbital plane that tilts down and back from the inferior orbital margin (Arsuaga et al., 1997). Fossil hominids that fulfill all of the criteria just mentioned include all of those listed earlier as possessing bipartite brows and chins. As we have seen, our subjective preference would be to include in this group the Omo, Herto, Fish Hoek, and other specimens discussed at the end of the last section despite their lack of human chins and definitive glabellar butterflies. Sadly, though, taking this inclusional step hardly simplifies our understanding of the immediate phylogenetic context of H. sapiens. For remarkably, although we have good fossil evidence for close relatives of the Neanderthals (Sima de los Huesos, Steinheim), modern H. sapiens itself appears oddly isolated.

Although Lieberman et al. (2002) suggested, on the basis of morphometric analysis, that cranial globularity and facial retraction are autapomorphic for anatomically modern H. sapiens, their sample allegedly showing these characteristics included Qafzeh 6, Skhūl V, and Jebel Irhoud 1, none of which shows the morphology of the living species. Similarly, in a subsequent study of the middle cranial fossa and its external cranial wall, Bastir et al. (Bastir et al., 2008) used the Singa, Skhūl V, and Mladeč 1 crania to represent H. sapiens. But as we have already noted, among these cranial specimens, only Mladeč 1 of possesses the bipartite, glabellar butterfly supraorbital configuration, and, where preserved, none except Mladeč 1 presents the basicranial features of H. sapiens listed above. The problem of recognizing modern human morphology is further complicated by Bastir et al.'s (2008) inclusion of Bodo, Arago 21, Kabwe, and Petralona as H. heidelbergensis in their principal components analysis of middle cranial fossa configuration, but only of Kabwe in their transformation comparison with a modern human (specimen unidentified) and the Guattari Neanderthal. For, as we have already seen, Kabwe (and Petralona) differ significantly from Bodo and Arago 21, and thus also from extant humans, most especially in the detailed morphology of the anterior and middle cranial fossae (Schwartz and Tattersall, 2002a; Seidler et al., 1997).

What is more, both the Skhūl V mandible and Jebel Irhoud 3 fail to present the inverted “T” configuration of the symphyseal region, while displaying a marked retromolar space. It is worth pointing out that Lieberman and Bastir and their colleagues (Lieberman et al., 2002; Bastir et al., 2008) also based their conclusions about the developmental distinctiveness of human cranial morphology on the transformation of one adult skull shape into that of another, which does not reflect the actual ontogeny of structure. What is potentially most interesting about these studies, however, is the implication that various late Pleistocene hominids may have independently developed basic neurocranial and craniofacial shapes and relationships similar to those seen in H. sapiens.

Species are the basic units of phylogenetic analysis, and patterns of genetic and morphological diversification within them are fundamentally different from those expressed among them. It is thus critical to know whether or not the diversity one inevitably sees in the fossil record is significant at the species level. The late Pleistocene doubtless saw repeated contact between resident hominid populations in Eurasia and populations of anatomically modern H. sapiens newly emerging from Africa. There has, thus, naturally been considerable interest in whether there was any hybridization among the hominids involved, in which case, any morphological definition of H. sapiens based purely on its living representatives is doomed. The records of the latest Pleistocene in Europe and the Near East are the only ones well enough documented for this question to be asked, and it has led naturally enough to a search for fossils possessing both anatomically modern H. sapiens and Neanderthal features. Fossil individuals recently said to represent Neanderthal/modern hybrids (or some variation on the theme) include those from the ca. 24.5 ka site of Abrigo do Lagar Velho, Portugal (Duarte et al., 1999) and from the ca. 35 ka site of Peştera cu Oase, Romania (Trinkaus et al., 2003a, 2003b; Rougier et al., 2007).

In the case of Lagar Velho, the largely complete skeletal remains are those of a child who died at the age of ca. 4 years, some thousands of years after the last documented occurrence of Neanderthals in Iberia. Duarte et al. (1999) suggest that this individual was ultimately derived from a population showing some evidence of hybridization between Neanderthals and anatomically modern humans. Abundant modern features include a distinct mental trigon, associated with a thickened inferior margin and a pair of mental fossae. Allegedly intermediate between modern humans and Neanderthals is a juxtamastoid (occipitomastoid) crest that is subequal in projection to the very short mastoid process. On the Neanderthal side are some retreat of the symphyseal region and, most especially, robust postcranial bones.

Peştera cu Oase has yielded a complete mandible (Oase 1) with three right and the last two left molars (Trinkaus et al., 2003b). Oase 2 consists of a fairly complete facial skeleton with M1–2 in place and M3 erupting bilaterally, plus an associated partial frontal, partial but complementary right and left parietals, and virtually intact right and left parietals. Oase 3 is represented by a relatively complete left temporal bone (apparently missing the uppermost part of the squamosal suture) (Trinkaus et al., 2003a). Most of the analysis of these remains involved metric comparisons, interpreted as presenting a picture of mixed “late archaic” and anatomically modern human features (Trinkaus et al., 2003a).

The morphological case made for archaic/modern admixture in both cases is substantially less than compelling, and, for the moment, the simplest hypotheses seem to be the most robust. The clear signal emerging from the various contributions to the edited monograph on the Lagar Velho child (Zilháo and Trinkaus, 2002) is that, the editors' conclusions notwithstanding, these remains fall within the envelope of variation reasonably expected for “modern” children of that age, but with some interesting hints at difference. For example, although some aspects of tooth development of the Lagar Velho child approach those noted in Neanderthals, others approach or are even identical with the pattern of H. sapiens (Bayle et al., 2010); interpretation of the significance of these observations, however, requires a broader comparative context among hominids. And although the Peştera materials are not thoroughly typical modern H. sapiens, they most emphatically do not differ from us in the direction of Neanderthals. Of course, nobody knows just what a Neanderthal/modern hybrid should be expected to look like [studies of living primate species that have been observed to interbreed have unhelpfully focused on differences in pelage patterning, external elements of the gluteal and anogenital region, body size, and various facial dimensions (see review in Schillaci and Froehlich, 2001)]. Still, the expectation that seems to guide the interpretation of particular fossil specimens as presenting a mixture of “Neanderthal” and “H. sapiens” features may be reminiscent of Darwin's (1868) acceptance of the then-popular notion of parental character blending in offspring. And as Bateson (Bateson, 1909) demonstrated over a century ago, and as Darwin actually already well knew (Darwin, 1868) from extensive interaction with animal breeders, the external features of hybrids (even the offspring of very similar parents) tend to retain the identifying traits of those parents, which, if relevant, would make it hard to argue that any of the perplexing Levantine fossils were hybrids, either.

Green et al. (2010) have recently announced a draft of the entire Neanderthal genome. Their comparative analysis of a portion of this sequence, particularly in the region that codes for metabolically active proteins (which represents 2–3% of a metazoan's genome) (Eisen, 2000), suggests to them that there may have been a minimal amount of introgression (some 1–4%) of Neanderthal genes into the genome of the H. sapiens who invaded their European homeland some 40 ky ago (and, oddly, not in the other direction). Whether or not this conclusion will stand as more data come in, it is most likely on the basis of current knowledge both of the fossil record and of the effects of hybridization on morphology that H. neanderthalensis made no identifiable morphological contribution to any known fossil (or modern) population of H. sapiens. Yet, even so, there is nonetheless quite evidently a great deal more morphology out there than easily fits into our received (rather reductionist) schemes. And this surely adds a note of caution to any conclusion about the milieu out of which our species emerged. For, although our eyes do not deceive us, our traditions may.


A coda: Our overview here is limited to the fossil evidence for the emergence of a morphologically recognizable species H. sapiens. But, as Tyson, Blumenbach, and Goethe all knew long ago, the property that distinguishes us above all from everything else in nature, and which certainly gives us the feeling of being so different, is the way we process information in our heads. We possess reason. And if evidence in the archaeological record of symbolic behaviors (Tattersall, 2009) may be taken as a reliable proxy for the possession of “reason,” then this acquisition substantially followed the appearance of H. sapiens as a morphological entity as it is conceived here. The earliest H. sapiens, certainly those from Herto and Omo Kibish, were evidently doing business much as their predecessors had done (Klein, 2009). Only later were complex “modern” behavior patterns adopted. The simplest hypothesis seems to be that the biological underpinnings for the cognitive peculiarities of H. sapiens were attained exaptively, as part of the genetic reorganization that led to the emergence of the very distinctive bony anatomy of our species. But their radical potential was only “discovered” post hoc, much as birds discovered the potential of feathers for flight, and the tetrapods discovered the use of limbs for terrestrial walking, well after their initial acquisition. Thus, yet again, however remarkable, and apparently demanding of special explanation as we may think ourselves, we see in our own origin the operation of entirely routine evolutionary processes, emphasizing the importance of remaining vigilant against unconscious “human exceptionalism.”


For the variety of historical reasons explored above, our species has contrived to elude satisfactory morphological definition. Through a sort of self-reinforcing process, whereby each reasonably large-brained extinct form that was shoehorned into the taxon has appeared to enlarge its permissible morphological limits, a huge variety of morphologies has been admitted into H. sapiens, albeit sometimes into archaic varieties of the species. Acceptance of this muddled variety has been facilitated by a view of evolution that emphasizes gradual transformations in lineages, in which species are basically units of convenience rather than of biology, and that are expected in principle to be undefinable in morphological terms. To systematists studying other groups of mammals, this situation would be untenable; it would, indeed, effectively prevent them from plying their trade using currently fashionable approaches, but paleoanthropologists have remained fairly unperturbed because, after all, as human beings we “know who we are,” and do not really need to be told, which absolves us, of course, from having to find out.

Our view is that the practice of systematics in paleoanthropology urgently needs to be brought into line with systematics as practiced elsewhere in zoology. The burgeoning human fossil record increasingly demonstrates that hominid history was one of diversity, of vigorous evolutionary experimentation, rather than one of gradual transformation over the eons, and our approach to understanding human phylogeny needs to change commensurately. Species, including our own, need to be restored to their rightful place as actors in the evolutionary drama, and, if we are to do this, we have to be able to define them adequately. We fully recognize that any living species has a dynamic behavioral dimension, but, alas, morphology is all that any extinct form bequeaths us that is directly relevant to determining its identity—neither geological age nor inferred ecology has the direct relevance to species recognition inherent in morphology.

H. sapiens is a variable species in space, and it is reasonable to suppose that it was comparably variable in time. But, it is equally true that living H. sapiens is a highly distinctive morphological entity, even compared with some of its closest apparent relatives in the fossil record, for example H. neanderthalensis. We and other researchers have identified a suite of osteological characters that seems to be distinctive for living H. sapiens; when we apply these features to putative relatives in the fossil record, we find a strictly limited number of candidates for inclusion in H. sapiens. Indeed, they eliminate everything that has in the past been referred to as “archaic H. sapiens,” and certainly all fossils above about 200 ky old, as well as many younger ones. The situation is complicated by the fact that some fossil hominids within this time range (some of them remarkably recent) have the full suite of these features except for the most distinctive of them (the “bipartite” brow and the chin), but, even after relaxing the morphological definition of H. sapiens to include these specimens, we find our species remarkably isolated. Various putative relatives in the African record may have one or two features reminiscent of the living species, but they certainly do not provide a close systematic context for it in the way, for example, that the Sima de los Huesos (Atapuerca) specimens or the cranium from Steinheim, provide context for H. neanderthalensis.

Various fossil hominids in the 100 ky range and below have been suggested at one time or another to be the possible result of hybridization between H. sapiens and another hominid form, most commonly H. neanderthalensis. The currently available literature on documented interspecies hybrids is not very helpful in this regard, focusing largely on size and external and soft-tissue features (possibly because hard-tissue characters of the kind that would preserve in the fossil state are not very noticeable). However, what little is known suggests that hybrids tend to preserve parental features more or less intact. Our examination in this light of the best-known putative hybrids between the highly morphologically distinct H. sapiens and H. neanderthalensis suggests that most of the fossils concerned are indeed distinctive in some way, but that this distinctiveness is unlikely to be the result of hybridization. All in all, our analysis suggests that there is a great deal more morphology out there that is relevant to the origin and immediate evolutionary context of H. sapiens than paleoanthropologists have been prepared to come to grips with in systematic terms and that a change in priorities, or at least an addition to them, would be a salutary development. Certainly, a better African fossil record will be necessary for clarifying the background from which our species emerged; so also will closer scrutiny of modern human morphology, and the application to it of the methods that animate the systematics of other mammalian groups.


The authors thank Bob Sussman for inviting them to undertake this overview, and several anonymous referees whose suggestions improved it. Above all, they thank the many curators and managers who made their collections available to us for study.