Le registre fossile du vivant dévoile les anatomies, les adaptations et les écologies passées. Une des contributions les plus précieuses de la paléontologie est l'utilisation de ce registre pour explorer les innovations générées par l'évolution dans des mondes aujourd'hui disparus. Toutefois, un mode alternatif d'interprétation de ce registre tyrannise la paléoanthropologie tout comme d'autres sciences historiques. Ce mode se base sur la comparaison avec les données actuelles. En effet, si le présent éclaire le passé d'une multitude de façons, les riches données du monde actuel peuvent aussi compliquer la reconnaissance et l'analyse des innovations de l'évolution survenues dans le monde préhistorique. Un paléobiologiste imprudent peut facilement interpréter faussement les organismes anciens en utilisant des concepts inappropriés basés uniquement sur les formes et fonctions actuelles.

Par exemple, la dichotomie entre des locomotions « bipèdes » opposées aux locomotions « quadrupèdes » s'applique de façon satisfaisante aux hommes modernes et à leurs proches parents africains, les grands singes. Mais qu'en est-il des organismes éteints qui, situés à la base du clade hominidé, vécurent au cours du Miocène récent ? Cette dichotomie néontologique ne contribuerait-elle pas plutôt à masquer des innovations évolutives, documentées uniquement dans le registre fossile ? Il s'agit là d'un problème fondamental pour l'étude de l'origine et des premières phases de l'évolution des hominidés.

La mise au jour d'extrémités proximales de fémurs dans la formation de Lukeino (Tugen Hills, Kenya) en 2000 [17] était un événement crucial en paléoanthropologie. Enfin, un élément clé de l'anatomie locomotrice d'un hominoïde du Miocène récent venait d'être découvert. En particulier, le fossile BAR 1002′00 était porteur de grands espoirs, étant donné l'importance du fémur proximal pour la locomotion, étant donné aussi la longue histoire des découvertes et interprétations de cet élément du squelette dans l'étude de l'évolution humaine.

The fossil record of life on earth reveals ancient anatomies, adaptations and ecologies. One of paleontology's most valuable contributions is the use of the fossil record to explore evolutionary novelty in all-but-vanished worlds.

Archaeologists who study prehistoric cultures know the danger of viewing the past through the prism of modern ethnohistorical observations. Archeologists call this tendency of the modern world to distort their interpretations of the past a “tyranny of the ethnographic present” [29].

A kind of parallel interpretive tyranny exists in paleoanthropology, as in other areas of historical science. Here, the rich detail of the modern world compared to the paucity of the prehistoric world can serve to obscure the recognition and analysis of evolutionary novelty. The present illuminates the past in myriad ways. However, the unwary paleobiologist can easily misinterpret past organisms by using inappropriate interpretive constructs based solely on modern form and function.

For example, the dichotomization of locomotor modes as ‘bipedality’ versus ‘quadrupedality’ works well to describe modern humans and their close African ape relatives. But what about now-extinct organisms who lived during the Upper Miocene, near the base of the hominid clade? Does this neontological dichotomization actually serve to obscure evolutionary novelty that is accessible only through the paleontological record? This is a fundamental issue for the study of hominid origins and early evolution.

The recovery of proximal femora from the Lukeino Formation in the Tugen Hills of Kenya during 2000 [17] was a milestone in paleoanthropology. At last, a key part of the locomotor anatomy from an Upper Miocene hominoid had been found. In particular, the BAR 1002′00 fossil held great promise due to the key role that the proximal femur plays in locomotion and the long history of discovery and interpretation of this skeletal element in human evolutionary studies.

The first recovered proximal femoral fragment of an early African hominid was part of the Sts 14 individual. It was described in Nature by Broom and Robinson in 1947 [1 (p. 431)]: “The femur is much crushed, and has not yet been removed from the matrix. It is rather slender, and has a few characters not quite human.” Broom and Robinson never elaborated on the nature of the novelty they initially perceived, because, as they stated in 1950: “It seems unwise to describe this badly crushed femur in detail as the California University Expedition has found the beautifully preserved upper third of a femur apparently of an Australopithecid from a deposit at Sterkfontein, about a couple of miles from the caves where we are working. We have seen this specimen before it was cleared of matrix; but, as Prof. Camp or one of his assistants will doubtless be describing it in detail, we do not feel at liberty to say anything more about it.” [2 (p. 62)].

In the next paragraph, however, Broom and Robinson asserted differences between the proximal femur of Australopithecus and other hominids, noting differences in the top of the greater trochanter (which they admitted was missing from the only known specimen, Sts 14), the shallowness of the trochanteric fossa, the lack of a well developed trochanteric crest, and the placement of the lesser trochanter. The femur recovered by Camp was never described, and its whereabouts remain unknown today.

The next early hominid proximal femora were found in a nearby but geologically younger breccia at Swartkrans. The deposit yielded an abundance of Australopithecus robustus craniodental remains, so the two proximal femora SK 82 and 96 are usually (uncritically) attributed to that species. Because of their relatively early chronological recovery and good preservation, these specimens became a comparative foundation for later studies of the proximal femur in human paleontology, and they continue to be studied today [20].

John Napier was the first to describe the Swartkrans proximal femora in 1964 [13]. This was a fact that Napier called “somewhat surprising in view of the light they throw on the walking ability of australopithecines and, hence, on certain aspects of their phylogeny” (p. 691). Napier described the novel external architectural features of the Swartkrans femora, calling particular attention to the small size of the head relative to the length of the neck. In 1972, 35 years after Sts 14 had been recovered, Robinson published definitive descriptions of the three South African proximal femora. He thereby integrated this skeletal element into his functional and phylogenetic synthesis for A. africanus and A. robustus [19].

By the time Robinson's treatise on the South African postcrania was published, the pace of fossil discovery in eastern Africa had rapidly accelerated. The first proximal femur to be added to these early hominid collections was Olduvai Hominid 20, a proximal femur without a head from Olduvai Gorge. The specimen was similar in size and morphology to preserved parts of the two Swartkrans femora. O.H. 20 was described by Day in 1969. He noted the functional significance of the groove for the obturator externus muscle [4].

Additional discoveries in the Lake Rudolf (now Turkana) basin in the early 1970s added many new proximal femora to the early hominid fossil inventory. In 1973 Walker presented his reconstruction of an Australopithecus femur based on the then-available parts [25]. The same year, at the 9th International Congress of Anthropological and Ethnological Sciences in Chicago, Lovejoy presented a paper (that would be published in 1975) entitled Biomechanical Perspectives on the Lower Limb of Early Hominids [10]. The proximal femur played a central role in his synthesis. Lovejoy summarized research on the early hominid proximal femur to that point, itemized those anatomical features in which Australopithecus differed from later Homo, and concluded: “the lower limb skeleton of Australopithecus points to a long history of bipedalism among hominids” (p. 324). The next three decades witnessed the extension of that history deep into the Pliocene.

Beginning with the recovery of the A.L. 129 hominid femur from Hadar in 1973, the recovery of the partial A.L. 288 (‘Lucy’) skeleton in 1974, and the discovery of A.L. 333 the next year, a wealth of even earlier hominids became available. The publication of these materials only five years after they were found [9] generated a large and still-ongoing debate about the locomotor capabilities of early hominids. The proximal femur figured prominently in the debate.

One of the first to draw functional and phylogenetic interpretations from the Hadar postcranial remains was Brigitte Senut. Her early access to these unpublished fossils allowed her to present an analysis of the early hominid humerus and its articulations as a ‘thèse de 3^e cycle’ examined in Paris in 1978 [21]. In that and all her many subsequent works, Senut has contended that Homo is a phylogenetically ancient line. She concludes that Homo existed contemporaneously with Australopithecus, and that for ‘Lucy’, “the chimp-like elbow-joint morphology (probably convergently derived with Hylobates, Pan, and Gorilla, and not primitive) does not permit us to consider this hominid as a direct human ancestor” [23 (p. 193)].

Senut has consistently relegated Australopithecus (including A.L. 288-1 ‘Lucy’) to a phylogenetically peripheral side branch in human evolution. This is a stance championed by the late Louis Leakey and promoted by his son Richard (as documented by Pickford [16]). Senut held firmly to this position through the 1980s and 1990s. She has not wavered, and in the current millennium, posits a ‘Praeanthropus–Homo lineage’ closer to Homo sapiens, contemporary with the dead-end lineages of Australopithecus [22].

Debates about early hominid systematics and locomotion during the 1980s and 1990s centered on the Hadar postcranial sample. An important proximal femoral specimen was added to A. afarensis during the 1981 Middle Awash field season. Dating to approximately 3.4 million years in age, the MAK-VP-1/1 specimen from Maka, Ethiopia represented, until recently, one of the oldest proximal hominid femora known. In the spring of 1982, this specimen was subjected to CT scanning. The results were dramatic. First reported in 1984 [3] and [26], these radiographic results provided strong internal support to interpretations formerly based solely on the external morphology. Virtually all contemporary workers now interpret the proximal femur of Australopithecus as indicative of some form of bipedal locomotion.

The 1990s witnessed an intensified interest in the internal structure of the early hominid proximal femur, accompanied by astonishing advances in evolutionary developmental biology, and the beginnings of an integration of this new information into studies of early hominid fossils [11]. These developments set the stage for recovery and analysis of proximal femoral remains belonging to still earlier hominids.

During the 1992 and 1993 field seasons, a substantial collection of craniodental and postcranial remains was made at the Aramis site, Ethiopia by our Middle Awash research team. These remains date to approximately 4.4 Ma and represent a new genus and species of hominid (Ardipithecus ramidus). They are interpreted to represent cladistic hominids. They were announced as such in Nature in 1994 [27] (the genus name, initially Australopithecus, was changed to Ardipithecus the following year) [28].

Three years later, in 1997, even earlier hominids were discovered nearby. Haile-Selassie and colleagues began to recover largely craniodental remains from Upper Miocene deposits along the Western Margin of the Middle Awash study area. These were published in the genus Ardipithecus in 2001 [7] and elevated to species status in 2004 [8].

Unfortunately, as of Spring 2005, the Middle Awash has not yet yielded any head or neck of a proximal femur belonging to Ardipithecus ramidus or its apparent ancestral chronospecies, Ardipithecus kadabba. Therefore, all knowledge of the anatomy of the early hominid femur head and neck derives from Kenyan specimens found in 2000.

I was in Ethiopia when I received the first media reports of a December 2000 press conference in which these approximately 6-Ma discoveries by Brigitte Senut and Martin Pickford were announced to the world. The BAR 1002′00 specimen represented the first well-preserved proximal femur of an Upper Miocene fossil hominoid. For this reason, it was of enormous significance for understanding hominid origins and early evolution. I immediately congratulated the authors via an e-mail to Martin Pickford on January 20, 2001.

He replied the next day, noting that the ‘Millennium Ancestor’ femora were more human-like than Lucy's. The fossils were initially published shortly thereafter by Pickford and Senut in the South African Journal of Science [17]. They claimed: “the proximal femoral morphology of the Lukeino hominid is closer to that of humans than it is to that of Lucy.” This conclusion was accompanied by the statement: “6 million years ago in Kenya, there was a bipedal hominid which was closer in size and morphology to extant humans than the much younger australopithecines and Ardipithecus ramidus” [17 (p. 22)].

When the first scientific publication of the new Kenyan fossils in Comptes rendus de l'Académie des sciences, Paris [24] appeared later that year, Senut and colleagues assigned the proximal femora to a new genus and species, Orrorin tugenensis. Indeed, this new genus was differentially diagnosed, in part, with the contention that the proximal femur was “more human-like than those of australopithecines or African apes” [24 (p. 139)].

In that publication [24], these new Kenyan Upper Miocene fossils were interpreted as exclusively ancestral to the Homo lineage. Senut and Pickford interpreted both Australopithecus and Ardipithecus as side branches in the human family tree (they depicted Ardipithecus as an ancestral chimpanzee, and ‘Australopithecines’ as an extinct clade [24 (Fig. 2, p. 142)]. Given the historical context, it was as if their new phylogeny had been drawn by Louis Leakey himself.

Senut, of course, was being consistent with a phylogenetic interpretation about Australopithecus that she had held for over twenty years. At last, she was describing Upper Miocene fossils that she and the Leakeys had long predicted would represent the early human lineage. These fossils, particularly the Orrorin femora, were being used to support a Leakey family contention that Australopithecus played no role in the evolution of Homo. But did the new fossils actually confirm Leakey's and Senut's predictions that hominids more derived than Lucy would be found in the Upper Miocene?

Unfortunately, the initial 2001 presentation and analysis of Orrorin [24] were flawed by a variety of confusing and inaccurate statements. Among these, enamel thickness was reported as 3.1 mm “on the apex of the paraconid” [24 (p. 140)]. This was important because the enamel thickness was one of only two characters purported to differentially diagnose Ardipithecus from Orrorin. However, since the paraconid has long been absent from the molars of all higher primates, it was unclear where the published measurement had been taken, and what it represented.

Furthermore, for some reason, the original description of the Orrorin proximal femur appeared to characterize the functionally critically important obturator externus groove (so important for interpreting locomotion; see above) as an “intertrochanteric groove” (p. 141), (which it is not, and whose citation, oddly, was not made to the paper by Day [4]; see earlier, but to others who emphatically disputed its significance as a marker of bipedal locomotion). Given these irregularities, it is therefore not surprising that many readers were unconvinced of the accuracy of the first functional and phylogenetic claims made by Senut and Pickford about Orrorin.

After puzzling over the nomenclature and metrics presented by Senut and colleagues in their initial Comptes rendus de l'Académie des sciences descriptions, I e-mailed Martin Pickford on March 13, 2001, to again congratulate his team, and to ask for clarification. I also asked Pickford whether their described “intertrochanteric groove” was actually the obturator externus groove, and requested clarification of the published 3.1-mm enamel thickness value. I asked about the distribution of cortex around the femur neck as follows: “Did you photograph the femur before it was glued together, or was it found in one piece? If it wasn't broken all the way thru, do you have X-rays or CT to show what the distribution of cortex is around the neck? We have really great views of this in the Maka femur, and broken A. afarensis ones, and would like to compare.”

Martin Pickford and colleagues replied by e-mail on April 3rd, 2001. They stated that the femur had, indeed, been found in four pieces, and that photographs, sketches and measurements of cortical thickness had been made before the specimens were glued together. They have never been made available. They went on to say that the fossils could be unglued if necessary, and that they had a complete series of CT scans at 0.5-mm slices through the femoral neck. Their response made it clear that the obturator groove was well expressed, that the 3.1-mm enamel thickness was inaccurate, and that further studies were underway. In a follow-up e-mail on April 12^th, Senut and Pickford asked us not to publish any of the observations that they had shared with us, pending a more detailed study they were preparing.

I wrote to Pickford again on August 5th, 2001, urging him to focus on the cortical distribution in the femoral neck: “Here, because of the Maka femur, we have been working intensively, particularly Owen Lovejoy. One thing we've learned is that broken surfaces are ALWAYS superior to CT scans in evaluating relative thicknesses. We think that great care is needed to present the relevant data both numerically and photographically, not only where the scans are, but also photographs of the unglued/broken originals with the cortex showing clearly.”

That e-mail referred to the Middle Awash team's work on the Maka femur that we had found in 1981, scanned and X-rayed in 1982, published in 1984, and been working on ever since (see above). Pickford replied via e-mail the next day (6th August, 2001). He said that since the femur of Orrorin had been broken across the neck, it could be unglued and studied. Because we were about to submit the final detailed description of the Maka fossil to the American Journal of Physical Anthropology, we made the results of that analysis available to the Orrorin team by sending them an electronic pre-submission copy of the manuscript the same day.

Martin Pickford e-mailed again on September 6th, 2001 to confirm that he would unglue the femoral neck in October in order to make a direct visual assessment. Later, on November 30th (after the Orrorin group's field season), I e-mailed to ask Pickford whether he had unglued the femur neck. He replied, on December 1st, that the conditions he was working under did not allow him to unglue the neck because he did not want to cause any damage. He stated that the CT scans in the team's possession were sufficiently clear to reveal the cortical thickness.

Martin Pickford and colleagues re-described the proximal femur of Orrorin in C. R. Palevol the next year [18]. This time the anatomical descriptions were more standard and detailed. Information on the internal anatomy of the femur was presented for the first time via the CT scans they had written us about. The descriptions and illustrations of the proximal femoral anatomy in this second publication made it clear that Orrorin shared several derived characters with later hominids, and thus was likely to be a cladistic hominid. Furthermore, external morphology (particularly the obturator externus groove) was in keeping with some sort of bipedal posture or locomotion.

What was missing, however, was the photographic or radiographic evidence that had, by 2002, become critical in the evaluation of locomotor mode among early hominids. The first CT scans presented by the Orrorin team in C. R. Palevol [18 (Fig. 2)] were of poor quality and nonstandard orientation. Nevertheless, despite an apparently thick inferior cortex through the neck, the poor scans appeared to show a substantial cortical thickness through the superiormost femoral neck. This is a cortex distribution not seen in Australopithecus or Homo, but more reminiscent of great apes.

Nevertheless, the text accompanying this figure claimed: “The pattern of distribution of femoral neck cortex in Orrorin is close to that of humans and australopithecines and radically different from that seen in chimpanzees and gorillas.” The authors went on to echo their earlier phylogenetic claims as follows: “If we are correct, then Australopithecus may represent a side branch in hominid evolution that became extinct without giving rise to Homo, a hypothesis that has already been suggested by Coppens 1981 and Senut 1980 and [18 (p. 202)]⋯”

On August 28th, 2002 I wrote Pickford to congratulate him on the publication of the Palevol article with the CT scans of the femur of Orrorin. I asked: “On the relative cortical thicknesses, I thought that the reviewer held you up for conventional X-rays or xeradiographs. Do you have these, and could we get a copy to compare to ours of Maka? And do you have the digital data for your CT scans? We were wondering what would happen if the actual cross-section were taken closer to the orientation that we advocate in the Maka paper. The cortical thickness through the superior neck looks very high to us, and we were wondering why this was.” Pickford replied the next day that they lacked conventional X-rays, but that they were hoping to have a detailed paper on the scans out soon.

I wrote again to Martin Pickford on October 15th of 2002, after another of his fieldwork trips to Kenya, to again inquire about whether he had been able to unglue the femur neck or get a conventional radiograph of the femur neck when he was in Nairobi. He replied the next day that he had not yet done this. My repeated attempts to ascertain the distribution of cortex in the neck of the Orrorin femur had proven frustrating and puzzling, but I assumed that the discovery team's third description of the specimen, when published, would reveal it.

The detailed 2002 description and interpretation of the Maka proximal femur of Australopithecus afarensis [12] represents the most exhaustive analysis of the early hominid proximal femur to date. Available to the discoverers of Orrorin before its submission, that work makes it abundantly clear that the distribution of cortical bone in the neck of the proximal femur is a crucial feature in the interpretation of locomotor modes among early hominids [14]. So, when Galik and colleagues (including Senut and Pickford) published External and internal morphology of the BAR 1002′00 Orrorin tugenensis femur in Science during September of 2004 [6], it was extremely disappointing to realize that no new CT data had been taken from this crucial fossil, and no photographs, sketches or conventional radiographs were being presented.

Furthermore, despite years of our appeals to the discovery team for release of this crucial evidence, the Science paper presented only adjustments to the original scans, correcting for improper specimen orientation in the 2002 C. R. Palevol publication. As we noted in our Letter to the Editor of Science, published on February 11th, 2005, “the adjusted results are still inadequate to determine whether the femoral neck's cortical thickness conforms to a human, intermediate, or chimpanzee pattern.” We challenged the Orrorin team to provide: “(i) photographs, measurements, and drawings of its broken neck; (ii) conventional anteroposterior X-rays; and (iii) higher-resolution CT scans obtained with proper femoral orientation” [15 (p. 845)].

It is unclear why the Orrorin discovery team and its associates will not publish the comparatively very simple conventional radiography and conventional photography of the unglued BAR 1002′00 femoral neck that we have urged on numerous occasions (see above) since 2001. Martin Pickford and Brigitte Senut mysteriously did not join the list of authors who responded to our last, published request for these data in our February 2005 letter to Science. Their American colleagues responded: “it is our understanding that the initial studies were carried out under serious constraints of time and other resources […] and we have made it clear that we plan to rescan and study the existing fossils if funds are made available” [5 (p. 845)]. We were again disappointed because we had asked for the publication of new data, not the promotion of a funding request for documentation long overdue.

The distribution of bony cortex in the femoral neck is among the most important available evidence for ascertaining locomotor mode. Was the bipedality that the early hominid Orrorin practiced like that employed by modern humans and Australopithecus, or was it a novel and previously unknown form of locomotion? The evidence so far made available suggests the latter. This is because of the distinctive cortex visible on the top of the BAR 1002′00 femoral neck in the published CT scans. But the ‘inside story’ from that key anatomical region of the early hominid hip cannot be accurately or reliably told without better data. Why have those data not been made available, particularly when they require only an ungluing of the femoral neck or a simple anteroposterior conventional X-ray?

The Galik et al. Science paper was published on the 3rd of September, 2004 [6]. Fortuitously, the Inter-Academy Conference, Climats, cultures et sociétés aux temps préhistoriques, de l'apparition des Hominidés jusqu'au Néolithique, was held at the French Academy of Sciences a mere ten days later. Professor Henry de Lumley had graciously invited me to present a paper in the second session entitled The First Hominids. Because Dr. Brigitte Senut of the Orrorin team was a participant in the session, it represented an ideal forum for a productive exchange of data and ideas. Unfortunately, Dr. Senut once again presented the same ambiguous CT datasets. She thereby left all those in attendance wondering what really was the inside story on the early hominid proximal femur?

The last decade has witnessed an impressive advance in our knowledge about the earliest hominids due to discoveries in Chad, Kenya, and Ethiopia. It is clear that all three sets of fossils represent cladistic hominids (as opposed to cladistic chimpanzees or gorillas). This is because fossils in all three sets exhibit derived characters exclusively shared with later hominids such as Australopithecus. Whether the fossils from these three countries actually represent three separate genera (Ardipithecus, Orrorin, and Sahelanthropus) will necessarily require the discovery of additional fossils [8].

From associational and anatomical considerations, it is evident that these early hominids preferred more closed habitats than their descendants. It is also clear that they are different from any modern apes or previously known hominids--these earliest hominids were neither humans nor chimpanzees. And thus, forcing them into preconceived locomotor, species, or dietary categories based on the species-poor world of modern humans and relict apes almost certainly obfuscates the novelty that these Upper Miocene fossil primate remains may reveal.

The external anatomy of the Tugen Hills femora and the Middle Awash foot phalanx both suggest that some form of bipedality was being practiced by these earliest hominids. Did they share the highly evolved bipedality of Australopithecus? This is not clear. By forcing these organisms into narrowly preconceived categories based on organisms half as old (or younger), and by accommodating the new evidence in decades-old phylogenetic constructs, we sacrifice the insight that only paleontology can contribute to the study of our origins and evolution.