Mrs. Elvis, the Homo erectus pelvis

Scott Simpson and colleagues describe their find of a 1.5-million-year old, relatively complete pelvis of early Homo from Gona Ethiopia. The paper is in Science this week.

[UPDATE (2008-11-15): I’ve added Figure S4 from the data supplement, which is a nice comparison of the new pelvis reconstruction, BSN49/P27, on the top row, with the reconstructed pelvis of AL 288-1, “Lucy”.]

BSN49/P27 pelvis, top, compared to AL 288-1, bottom

The first thing I want to say about this paper is the complete stupidity of the journal in placing almost every graph, measurement, and piece of analysis in the online supplement. There is a decently detailed paper here, with some good illustrations, but it’s broken up into fragments by the publication style.

In fact, when I read through the paper the first time, my thought was, “Gee, that’s a pretty superficial report – there must be two or three more papers to write here somewhere.” In the online supplement, there are extensive comparisons, but none of them appear in the printed article.

That’s no discredit to the authors, since after all it’s good to have your paper printed in Science. But wow, if this is the way that the science has to go, it’s ridiculous. The lack of comparisons in the printed article has to be glaring, even to readers from outside the field.

The anatomy

The pelvis is pretty much within the size range known for other early Homo specimens. Its bi-iliac breadth (roughly, the width of the body across the hips) is just under a foot, at 288 mm. That’s not nearly the largest known for fossil hominids – the Sima de los Huesos male pelvis (Pelvis 1, Arsuaga et al. 1999) has a bi-iliac breadth of 340 mm; the Jinniushan female os coxa may correspond to a pelvis of nearly that size or even a bit larger. The BSN49/P27 pelvis is only 3 cm broader than Lucy’s, but that is enough to make it larger than average for recent human females. Fossil Homo had broad pelves with widely flaring ilia, a consistent observation across all Pleistocene specimens.

The most interesting thing is that the pelvis has itty-bitty acetabula. The acetabulum is the socket for the head of the femur; it’s part of the hip joint. So naturally, the size of the acetabulum reflects the size of the femur head, and both of these reflect (imperfectly) the magnitude of forces passing through the joint. The most stable components of these forces come from the weight of the body and the muscles that stabilize the hip, and the largest forces come from dynamic loading as the person runs or jumps. For this reason, a small acetabulum probably means small body size.

Simpson and colleagues estimate that their acetabula, with a supero-inferior diameter of 41 mm, would correspond to a femur head diameter of around 35 mm (33.4 to 36.8, estimated by regression). In the context of the fossil record, that is an exceptionally small femur head diameter.

  1. A number of fossil femora attributed to early Homo are quite long – following the long-limbed KNM-WT 15000 model. These femora, including KNM-WT 15000 itself, as well as Likewise, other acetabula of early Homo have much larger diameters, including KNM-ER 3228 and OH 28. The acetabular diameter of BSN49/P27 is much smaller than these large specimens.

  2. There are other femora attributed to early Homo that are shorter than those of KNM-WT 15000, with smaller head diameters. For example, the Dmanisi D4167 femoral head has a diameter of 40.0 mm, KNM-ER 1472 is 40.0 mm, and KNM-ER 1481 is 43.4 mm. The estimated femur head diameter for BSN49/P27, at most 37 mm, is smaller than any of these.

  3. However, in the context of living small-bodied humans, the acetabular diameter of BSN49/P27 is not unusual. McHenry (1992) reports femur head diameters for a small number of recent Khoisan (36 mm) and Pygmy (33 mm) individuals, and Berger et al. (2008) report the mean femur head diameter of a sample of Andamanese as 37.3 mm. Each of these mean sizes for contemporary populations would be consistent with the acetabular diameter of BSN49/P27.

  4. On the other hand, it isn’t obvious that the bi-iliac breadth of BSN49/P27 would fit within these small-bodied populations. For example, Ruff (1994) reports bi-iliac breadths for a number of Pygmy individuals, all of which are at least 30 mm smaller than the 288 mm value estimated for BSN49/P27.

The third point is enough for me – what the specimen really says (along with many others) is that the variation in body size among Early Pleistocene Homo was extensive, like that of living people. Still, the fourth point does seem to indicate a difference in pelvic (and femoral) proportions compared to humans. Let’s assume for a moment that the specimen really represents an apparently small, broad female individual. What does that mean?

For one thing, it really does have to cast doubt on the “standing tall” theory for the evolution of early Homo. Many articles were written in the 1990’s and early 2000’s to explain why early Homo was tall and thin, and Australopithecus was short and thick. These papers followed the discovery of KNM-WT 15000, which really influenced people’s thinking about early Homo. The explanations included thermoregulation, water conservation, climbing effectiveness, home range size, gut/brain energetics, predator confrontation, infant body mass, and life history constraints.

Strangely (perhaps), nobody ever actually tried to test which of these differences were more important than others. They were often content to draw up predictions about the consequences of a KNM-WT 15000-like body shape, compared to Lucy’s (AL 288-1) body shape. In some ways, the situation was similar to explanations for the origin of bipedality – there are many possible explanations, but few attempts to test them in a quantitative way.

Could one pelvis really throw all these arguments into disarray? Well, honestly, they’re already in disarray – the Dmanisi hominids were enough to tip things over the edge. The fact is that early Homo erectus simply didn’t look uniformly like KNM-WT 15000. There are many body sizes represented in early Homo, even within Africa, considering the other new small-bodied African Homo erectus specimens, like KNM-ER 42700.

Plus, the arguments never grappled with another obvious blind spot: We have no reason to think that male australopithecines had size/shape proportions exactly like Lucy’s, or Sts 14’s, or even the small Homo habilis skeleton OH 62. I’m not talking about limb proportions, here, but stature/bi-iliac breadth and other gross proportions of the body shape. It is doubtful that large australopithecine individuals would have had the same proportions as the smallest specimens known, and yet that is the model many have used.

This provokes an obvious question: Is the new pelvis, BSN49/P27, an australopithecine? To be sure, it’s a lot bigger than the relatively complete female australopithecine pelves, like AL 288-1 (Lucy) and Sts 14. But its acetabular diameter does fit easily within the size range of australopithecines. Mayer and van Gerven (1978) provided an estimate for the vertical acetabular diameter of SK 50 (which was malformed by a probable dislocation) of 41 mm, the same as the new pelvis. SK 50 even has a large ilium, although probably not large enough to make a 288 mm bi-iliac breadth.

But no, it’s not an australopithecine. BSN49/P27 is compellingly female, based on its large and round pelvic inlet, large pelvic outlet with wide greater sciatic notches, and large subpubic angle. Plus, it has a more prominent, thicker iliac pillar than known australopithecine os coxae. This is not as robust as the early Homo specimens KNM-ER 3228 or OH 28, but together with the rounded pelvic inlet it suggests affinity with Homo rather than Australopithecus. So, small, broad-hipped human it seems to be – based on acetabular diameter, we would infer this to be a smaller individual than those represented at Dmanisi, or those represented by KNM-ER 1472 and KNM-ER 1481. Again, all those arguments about the “tall, thin” body shape of early Homo erectus are out the window.

Now, I’ve gone through this whole write-up without discussing the “infant head size” estimates featured in the paper. I don’t have anything to add to that issue, other than to point out that this pelvis could have birthed a good fraction of modern human infants – not all, certainly, but many. That’s one way to look at developmental variability: Ancient Homo would not have been uniform in brain growth rates; nor are living humans.


Berger LR, Churchill SE, De Klerk B, Quinn RL. 2008. Small-bodied humans from Palau, Micronesia. PLoS ONE 3:e1780. doi:10.1371/journal.pone.0001780

McHenry HM. 1992. Body size and proportions in early hominids. Am J Phys Anthropol 87:407-431.

Ruff CB. 1994. Morphological adaptation to climate in modern and fossil hominids. Yrbk Phys Anthropol 37:65-107.

Simpson SW, Quade J, Levin NE, Butler R, Dupont-Nivet G, Everett M, Semaw S. 2008. A female Homo erectus pelvis from Gona, Ethiopia. Science 322:1089-1092. doi:10.1126/science.1163592