Damage to the greater trochanter in all three specimens includes loss of most or all of the femoral tubercle, but in BAR 1002’00 its base is present, but nothing can be said about its shape or extent. It blends into the greater trochanter proximo-medially, in this respect being similar to other hominid femora, and different from Miocene hominoids such as Ugandapithecus and extant African apes, in which the tubercle is either absent or is not well developed 〚22〛 and 〚24〛. This feature could be a hominid apomorphy, and is likely related to locomotion.

The intertrochanteric line is present but has a low relief in Orrorin. It is palpable and clearly visible in slanting light. In this, it differs from extant and Miocene African apes, in which it is absent. The anterior part of the intertrochanteric line, which is the zone of insertion of the iliofemoral ligament, the function of which is to maintain the upright stance, is strongly expressed in humans 〚1〛. In BAR 1002’00, the intertrochanteric line penetrates further onto the neck than it does in humans or australopithecines (AL 288–1), a morphology that also occurs in SK 82 and SK 97 from Swartkrans, South Africa, and in AL 333–95 〚1〛 and 〚41〛.

In BAR 1002’00 and 1003’00, there is a swelling quite low on the anterior surface, which might be a continuation of the intertrochanteric line. In AL 288–1 it also extends quite low. The basal part of the intertrochanteric line is the zone of insertion of the pubofemoral ligament and part of the iliofemoral ligament. The pubofemoral ligament inserts onto the horizontal ramus of the pubis. According to Aiello and Dean 〚1〛 the superior pubic ramus in these early hominids is relatively longer than in modern humans and lies in a different angular relationship to both the ilium and the ischium 〚5〛. This morphology suggests that Orrorin might have possessed a pubis in which the horizontal ramus was morphologically close to that of australopithecines. In SK 97 and SK 82, the intertrochanteric line continues as small distal crests. Orrorin does not possess such crests but has a low swelling which fades out distally.

In Orrorin, the lesser trochanter is medially salient in all three individuals (Fig. 1A). In australopithecines, the lesser trochanter is usually posteriorly projecting, and is not visible in anterior view (Fig. 1B). It is well developed in all the specimens from Lukeino, its extremity is smooth and the cortex appears to be dense. It is most strongly developed in BAR 1003’00, and its anterior edge forms a lip that overhangs its base. A similar lip is present in BAR 1002’00. The medial extremity of the lesser trochanter is damaged in the Aragai specimen (BAR 1215’00), but, in its preserved part, it conforms with the two specimens from Kapsomin. It differs from the Maka femur and that of AL 288–1 in not having a groove at the level of fusion between the lower part of the trochanteric apophysis and the diaphysis 〚23〛 and 〚47〛.

BAR 1002’00 is the only specimen with the neck entirely preserved. About half of it is preserved in BAR 1003’00 and only its base in the Aragai specimen. The neck is elongated when compared to those of recent African apes and is closer in this respect to australopithecines and humans. It is longer than in any of the Miocene hominoids: Ugandapithecus 〚36〛, Proconsul 〚3〛 and 〚44〛, Afropithecus 〚15〛, Nacholapithecus 〚26〛, Kenyapithecus africanus 〚21〛, Dryopithecus 〚25〛 and Oreopithecus 〚19〛. We conclude that a greatly elongated femoral neck such as occurs in Orrorin represents an apomorphic condition among hominoids, and that functionally it is related to bipedal locomotion and upright stance with the femur being habitually hyper-extended. It is thus a defining character of Hominidae in the strict sense (i.e. excluding African and Asian apes).

The femoral head in Orrorin is larger relative to shaft diameter than it is in AL 288–1 and other australopithecines and great apes. In this feature, Orrorin is closer to the human condition than it is to the ape or australopithecine one. However, because there is some debate about the value of this character for determining relationships or locomotor repertoire in hominids 〚30〛 and 〚43〛, we do not give it any weight in the conclusions of this paper.

Because the greater trochanter is missing in all three Lukeino hominid femora, we cannot provide accurate estimates of the depth or length of the superior notch (above the femoral neck and between the femoral head and the greater trochanter). However, we can obtain minimum measurements (length 20+ mm, depth ca 12.5 mm), and thus an idea of its shape. The superior notch of the femur in Orrorin is of the hominid type, and diverges greatly from the African ape and Miocene hominoid model. This is not surprising, because the shape of the superior notch is related to the length of the femoral neck, the shallow elongated notch occurring in species in which the femoral neck is long, and a short, deep notch in species with shorter femoral necks.

The gluteal tuberosity is sometimes referred to as the third trochanter 〚24〛, but we question its homology to this structure in other primates. The gluteal tuberosity is best preserved in BAR 1002’00 (Fig. 1I), but it is also present but damaged in BAR 1003’00. In the Aragai specimen, it is not clearly expressed due to breakage, but part of it is nevertheless visible as a swelling below and posterior to the base of the greater trochanter. In the Kapsomin specimens, it is proximo-distally elongated and raised into a ridge with a rugose surface. Distally it coalesces with a long crest that extends to the distal extremity of the preserved part of the shaft. The proximal part of the tuberosity is swollen more than is usual in humans. In South African Paranthropus, this part is damaged, but appears to form a more or less vertical crest, and is thus quite similar to Orrorin. The distal part forms an elongated swelling.

In the morphology of the gluteal tuberosity and its coalescence with the distally leading crest, Orrorin is unlike any of the extant African apes and Miocene hominoids (save in some aspects Ugandapithecus 〚17〛), and is closer to humans and australopithecines. In chimpanzees, for example, there is a lateral spiral pilaster between the grooves for insertion of the vastus lateralis and gluteus maximus and there is no gluteal tuberosity. As pointed out by Lovejoy et al., 〚24〛 and Stern and Larson 〚39〛, the variation in muscle attachments in this area of the femur reveals a fundamental difference between hip and thigh musculature in apes and humans, which almost certainly relate to the major postural and locomotor repertoire of the hominoids involved – quadrupedalism and semi-orthograde posture in African apes, and bipedalism and fully orthograde posture in humans and australopithecines. Orrorin is clearly more human-like and australopithecine-like in the morphology of the gluteal tuberosity and the regions surrounding it, from which we deduce that it was probably bipedal and orthograde, although in ways that were somewhat different from australopithecines and modern humans.

In posterior view, the lesser trochanter is medially projecting (Fig. 1C), more than is usual in australopithecines (Fig. 1D) and in those extant humans that possess marked femoral shaft anteversion (mainly from the New World) in which the lesser trochanter projects posteriorly. In posterior view, the base of the lesser trochanter of Orrorin is seen to have several elongated but low and slightly sinuous ridges leading towards the base of the greater trochanter (i.e. supero-laterally). The extremity of the lesser trochanter is well separated from the neck, as in humans, gorillas and Pongo, and the insertion of the ilio-psoas muscle does not prolong onto the inferior surface of the neck as it does in chimpanzees and gibbons 〚3〛.

In the Orrorin femora, the pectineal line is clearly marked. The spiral line is the insertion of the vastus medialis muscle. The spiral line that departs latero-distally from the area below and medial to the lesser trochanter to join the linea aspera in humans is present in Orrorin, but it is located some distance below the base of the trochanter, as is usually the case in humans 〚14〛. In AL 288–1 (Fig. 1D) and the Maka femur 〚24〛 and 〚47〛, this line cuts into the base of the lesser trochanter, forming a prominent groove in its inferior surface and this is a significant difference from Orrorin. A similar curved line occurs in chimpanzees, but there is no associated groove at the base of the trochanteric apophysis. The main difference between Orrorin and humans is that, in its distal portion, the spiral line does not closely approach the crest that descends from the third trochanter, thereby not forming a high relief linea aspera, as in humans. As it is, the closest approach is 6 mm in BAR 1002’00 some 100 mm below the third trochanter.

In Orrorin, the intertrochanteric crest is clearly visible in BAR 1002’00, but between the two trochanters it has lower relief than is usually the case in humans, there being a saddle shaped depression between the bases of the greater and lesser trochanters with virtually no overhang of the intertrochanteric crest superiorly, although there are some pits in the surface of the bone immediately superior to the margin of the crest (Fig. 1C2). As a result the sub-trochanteric fossa of Orrorin is weak, as in South African Paranthropus. In humans, the crest between the two trochanters is continuous and superiorly it overhangs the sub-trochanteric fossa. This saddle shaped depression in the intertrochanteric crest is variably developed in humans and chimpanzees, but it is seldom as low as it is in Orrorin. Despite damage to BAR 1003’00 and BAR 1215’00, it is possible to see that the saddle shaped depression is present in all three femora from Lukeino, and it is presumably a constant feature of the species. The intertrochanteric crest of Orrorin is almost at a right angle relative to the long axis of the femoral neck (Fig. 1C2). This morphology recalls the pattern seen in australopithecines.

An obturator externus groove is clearly visible in posterior views of BAR 1002’00 (Fig. 1C2) and BAR 1003’00. The surface of the bone in the Aragai specimen is not well enough preserved to demonstrate the presence of the groove.

In BAR 1002’00, from the fossa where the obturator externus attaches to the femur, the groove leads medially across the lower part of the femoral neck to terminate at its inferior margin. It is about 5 mm wide (proximo-distally) and some 25 mm long (latero-medially). There is no swelling or tubercle on the neck superior to the groove on its medial end. The groove is highly polished where it emerges from the fossa and it diminishes slightly in depth and polish towards its medial end.

The obturator externus is an adductor, flexor and external rotator of the thigh. One of its functions is to shorten the distance between the pelvis and the femur thereby steadying the hip joint. Day 〚10〛 and Robinson 〚30〛 suggest that the groove is indicative of full extension of the femur. The groove is caused by pressure of the obturator externus tendon onto the back of the femoral neck, and has been taken to be a sign of habitual extension of the hip joint either during posture or in locomotion 〚10〛 and 〚22〛. Similar grooves exist in other Plio-Pleistocene hominids such as OH 20 from Tanzania 〚10〛, SK 82 and SK 97 from South Africa 〚30〛 and other East African specimens 〚11〛, 〚12〛, 〚13〛 and 〚43〛, as well as AL 333–95 and AL 288–1 from Ethiopia 〚23〛, but are usually absent in apes. A similar groove occurs in a few other primates such as Trachypithecus, Papio, and Ateles 〚3〛. Stern and Larson 〚39〛 and 〚40〛 and Stern and Susman 〚42〛 suggest that the groove can be indicative only of frequent bipedalism, even of a monkey-like or ape-like nature.

The presence of this groove in Orrorin reveals that the femur was habitually in a hyper-extended position. In humans, the obturator externus runs from the posterior side of the femur to insert onto the anterior part of the pelvis 〚45〛. When the femur is hyper-extended (e.g., in a standing posture or when walking and running) the tendon presses onto the back of the femoral neck. The groove is not due to abrasion, but is caused by haversian remodelling of the bone. In humans hyper-extension of the femur occurs even in infants that are not yet walking. When the femur is flexed, as in humans sitting or in bending-over postures, the tendons do not press as much on the femoral neck. In habitually quadrupedal apes, such as chimpanzees and gorillas, there is little or no pressure from the obturator externus on the femoral neck, and thus no groove is produced. But even in bipedally walking chimpanzees, the femora are not as hyper-extended as they are in humans, and any pressure produced on the femoral neck by the obturator externus is weak compared to that in humans. The limited bipedal activity undertaken by chimpanzees is too transient to lead to remodelling of the outer contour of the bone, so no obturator externus groove occurs in this species.

Stern and Susman 〚41〛 considered that the grooves on AL 288–1 and AL 333–95 (Australopithecus antiquus) 〚23〛 were not true obturator externus grooves. They suggested that the presence of a small tubercle at the base of the femoral head gives the illusion of a short obturator groove. There is no evidence of such a tubercle or bump in Orrorin, whereas in Paranthropus from South Africa there is a swelling but no tubercle between the groove and the femoral head. The occurrence of a groove in a few catarrhine and platyrrhine primates is due to convergence and not to phyletic relationship, and thus it does not weaken the value of this character for determining functional anatomy within the Hominoidea. We conclude that within the latter superfamily, the presence of a clear obturator externus groove is a derived character related to orthograde posture and bipedal locomotion. For this reason, the presence of an obturator externus groove in Orrorin is highly suggestive of habitually hyper-extended femora, and thus of orthograde posture and bipedal locomotion.

Only the inferior part of the base of the quadrate tubercle is preserved in Orrorin, but the tubercle itself has been damaged by carnivore chewing. We can say nothing about its morphology.

In lateral and medial views, it is evident that the femoral shaft of Orrorin is anteriorly convex from proximal to distal (Fig. 1E and I). Breakage accompanied by slight opening of the breaks on the anterior aspect of BAR 1002’00 exaggerates the impression of outbowing of the femoral shaft, but each undamaged segment shows distinct curvature, as does the shaft of the undamaged specimen BAR 1003’00. In Gorilla and Pan, the femoral shafts are more rectilinear in lateral view, whereas in extant and fossil Homo the shaft is anteriorly convex 〚12〛. Thus, in this feature, Orrorin is closer morphologically to humans than to African apes.

In humans, the pilaster forms the anterolateral margin of the insertion for the gluteus maximus.

In lateral view of the Orrorin femora, the region below the greater trochanter shows no sign of a spiral pilaster and its associated obliquely spiral grooves for the vastus lateralis and gluteus maximus insertions that are typical of Pan 〚24〛. Instead, there is a vertically oriented gluteal tuberosity, which coalesces into a low crest that descends the length of the femoral shaft. This crest starts descending the femoral shaft from the posterior part of the lateral surface just below the gluteal tuberosity and it gently curves onto the posterior surface as it descends. It is still well developed 179 mm below the proximal end of the gluteal tuberosity. The same morphology occurs in BAR 1003’00, but the crest is more robust and has a rugose surface. In Orrorin, therefore, the form and orientation of the pilaster is human-like and not at all ape-like.

None of the Orrorin femora has a hypotrochanteric fossa as voluminous as those frequently developed in humans. It is more like an enlarged but low relief surface bordered by the crest that descends the shaft from the gluteal tuberosity. It is slightly excavated laterally and posteriorly, so that its border stands up as a low crest with concave sides, especially the posterior one. This morphology could well represent a precursor of the hypotrochanteric fossa that typifies human femora. In Orrorin, this crest is better developed than in Pan, and is closer to the human morphology.

The position and orientation of the lesser trochanter is variable in humans, with some New World populations having it posteriorly oriented and Old World populations more medially located (Fig. 1E and F). Lovejoy et al. 〚24〛 suggest that the position of the lesser trochanter in humans is related to the degree of anteversion of the femoral shaft, which, according to them, relocates the lesser trochanter medialward. However, this explanation may not apply to australopithecines, and if not, then the posterior projection in australopithecines and some humans is not homologous. In the Miocene hominoids Ugandapithecus and Afropithecus, the lesser trochanter projects medially, but unlike Ugandapithecus, in Orrorin the medial border which supports the lesser trochanter does not form a right angle with the anterior surface of the femur.

The fovea capitis in BAR 1002’00 is located at the most medial part of the femoral head just below the midline of the femoral neck in anterior view and in line with the superior edge of the femoral neck in superior view (Fig. 1E and G).

In Orrorin, the trochanteric fossa is preserved in all three femora. It narrows sharply towards the attachment of the obturator externus muscle and there is no sign of its deep penetration into the shaft as occurs in chimpanzees 〚24〛. At most there is a shallow depression inferiorly (Fig. 1C2 and G). In this, Orrorin is close to humans and australopithecines (the fossa is as marked as it is in South African Paranthropus) and quite different from Pan. Deep penetration of the trochanteric cavity into the femoral shaft does not occur in Gorilla nor is it present in Miocene hominoids, and it is probable that its presence in chimpanzees represents an apomorphy of the genus. Being plesiomorphic among hominoids the morphology of the trochanteric cavity observed in Orrorin does not imply any special relationship between this genus and humans, but it does represent a difference from chimpanzees.

In the digital fossa, there is a slight depression anterior to the insertion of the obturator externus. Bacon 〚3〛 states that in humans and Paranthropus, the common tendon of the obturator internus and gemelli is located antero-superiorly to the obturator externus fossa, whereas in Australopithecus it is not. Even though the morphology is difficult to observe in Orrorin, it appears that in this feature this genus is more human-like than australopithecine-like.

The femoral head is almost spherical, its greatest diameter being 32.1 mm and its least diameter 31.7 mm. The greatest diameter is proximo-distal (i.e. parallel to the shaft).

In superior view, the femoral head of Orrorin is posed on the femoral neck in such a way that more of the head is anterior to the neck than is posterior to it (Fig. 1G). This gives the impression that the head is deflected slightly anteriorly. For this reason, the groove round the base of the head is appreciably shallower on the posterior side than it is anteriorly. In African apes and Miocene hominoids, the head is usually posed almost symmetrically on the neck, or at least more symmetrically than in Orrorin. Even if the orientation of the head varies in humans and chimpanzees 〚2〛, the anterior twist in Orrorin is clearly outside the range of variation that occurs in the latter species. This feature may well be functionally related to bipedalism for the same reasons that the femoral neck is anteroposteriorly compressed and possesses an obturator externus groove, the three features comprising different parts of the same overall morphological package.

According to Jenkins 〚20〛 and Stern and Susman 〚41〛, in modern humans the articular margin of the head typically passes from antero-lateral to postero-medial, but in some specimens the margin is more or less parallel to the long axis of the neck. In chimpanzees, the reverse is the case. In AL 288–1 it is less like modern humans, the articular margin of the femoral head passing from antero-medial to postero-lateral as in non-human hominoids 〚41〛 (Fig. 1H).

The femoral neck in Orrorin is antero-posteriorly compressed (Fig. 2), even more so than in AL 288–1 (Australopithecus antiquus). In BAR 1002’00, the supero-inferior diameter is 22.5 mm compared to the antero-posterior diameter, which is 15.6 mm, whereas in BAR 1003’00 the homologous measures are 26.4+ mm by 16 mm. In BAR 1215’00 the measures are 24.6 mm by 15.2 mm. Australopithecines and humans have antero-posteriorly compressed femoral necks, but chimpanzees, gorillas, and orang-utans have almost circular or only slightly compressed femoral necks. Proconsul nyanzae has a slightly compressed femoral neck (anteroposterior diameter 15.2 mm, supero-inferior diameter 20.1 mm 〚44〛) but not as greatly compressed as is usual in hominids. The functional meaning of this restructuring of the neck in hominids is probably related to the fact that in bipeds the bulk of the body weight is transferred to the ground only through the hip joint, whereas in quadrupeds the transfer of body weight is shared between anterior and posterior limbs.

For two reasons, it is not coincidental that in hominids the presence of a flattened femoral neck correlates with the presence of a groove for the obturator externus. Firstly, the flattening of the neck effectively shortens the path that the obturator externus takes on its way from its insertion on the femur to its termination on the anterior part of the pelvis, and secondly, it reduces the effect of a convex obstacle from its path (the almost circular femoral neck that occurs in apes) that would not only impart a bend into the muscular system, but would lead to the creation of increased stresses in the obturator externus muscle where it passes over the femoral neck.

What it means phylogenetically, is that Orrorin shares with australopithecines and humans a restructured hip joint, of which antero-posterior compression of the femoral neck is just one aspect.

The transverse section immediately below the lesser trochanter is medio-laterally expanded and there is a slight widening of the diaphysis linked to the flattening and to the strength of the gluteal tuberosity. In Orrorin, the section is rectangular with rounded corners, which makes the diaphysis platymeric (mediolateral diameter measured 3 cm below the base of the lesser trochanter is 25.5 mm, anteroposterior diameter is 21 mm) Paranthropus from South Africa is platymeric as are humans. Some Miocene hominoids such as Ugandapithecus 〚17〛 have flattened sections below the lesser trochanter. Platymeria of the femur may thus be a primitive feature within hominoids and not a derived condition of Hominidae as sometimes thought. The absence or weakness of platymeria in extant African apes may represent the derived condition among hominoids.

CT scans of the Orrorin femora carried out at the Clinique Pasteur, Toulouse, reveal that the distribution of cortex in the femoral neck is asymmetric (Fig. 2). In BAR 1002’00, in which the neck is complete, the cortex is thickest inferiorly and thinnest superiorly. However, the cortex superiorly, anteriorly and posteriorly is relatively thicker than it is in humans, but appreciably thinner than it is in African apes. As shown by Ohman et al. 〚27〛 in their tomodensitometrical study of 35 hominoid femora: “Throughout the femoral neck H. sapiens displays thin superior cortical bone and inferior cortical bone that thickens distally. In marked contrast, cortical bone in the femoral neck of African apes is more uniformly thick in all directions, with even greater thickening of the superior cortical bone distally. Because the femoral neck acts as a cantilevered beam, its anchorage at the neck-shaft junction is subjected to the highest bending stresses and is the most biomechanically relevant region to inspect for response to stress. Cortical distribution in the African ape indicates much greater variation in loading conditions consistent with their more varied locomotor repertoire. Cortical distribution in hominids is a response to the more stereotypic loading pattern imposed by habitual bipedality.” For this feature, Orrorin is closest to humans and australopithecines, and quite different from African apes.

In apes, the femoral neck is almost horizontal if the femur is oriented in its normal position during locomotion (i.e. with the distal femoral condyles held at the same horizontal level and the shaft almost vertical). In this configuration, body weight impinging vertically on the femoral head is transferred laterally along the neck before it passes down the almost vertical femoral shaft towards the knee. There are thus two marked angles in the bony link that transfers body weight from the pelvis to the knee, one between the femoral head and the neck, the second between the femoral neck and the shaft, the neck comprising the ‘cantilever beam’ of Lovejoy 〚22〛. In this configuration, the cortex of the femoral neck has to be thick all round as tensional stresses are marked, especially along the superior surface of the neck.

In humans, in contrast, when the distal femoral condyles are oriented horizontal with respect to each other as in normal fully orthograde posture and bipedal locomotion, the femoral shaft lies at a marked angle from the vertical and the inferior part of the femoral neck is oriented almost vertically. Thus, in humans body weight is transferred from the pelvis to the femoral shaft by way of the femoral head through the neck without the two right angle bends that occurs in apes, but by a more direct path, admittedly slightly sinuous 〚7〛 and 〚27〛. The biomechanical result of this is that femoral neck cortex is thickened inferiorly and reduced superiorly 〚22〛.

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. It is not identical to that of humans and australopithecines, but is clearly much closer to these hominids than it is to that of apes. Because of the biomechanical implications, we deduce that the asymmetric distribution of femoral neck cortex in Orrorin provides good evidence that the femoral shaft was oriented at a marked angle in normal orthograde posture and locomotion, and thus that Orrorin was orthograde and that its main locomotor repertoire while on the ground consisted of bipedalism.