News flash: Dmanisi hominids were not short

11 minute read

By now, the news of the Dmanisi hominids' small size has been out for years. There was a National Geographic feature on the story more than four years ago -- before my twins were born. If you think about early Homo, you've been incorporating the small body sizes represented by the Dmanisi postcrania into your thinking for some time now. The resulting conclusion has been repeated in lots of stories: "Early humans didn't need long legs to leave Africa."

So it came as no surprise when this week's report by Lordkipanidze and colleagues confirmed the short stature of the Dmanisi hominids:

Stature and body mass of the Dmanisi individuals calculated from various independent long bone measurements yield estimates between 145-166 cm and 40-50 kg, respectively (Table 1 and Supplementary Information 8). Their small stature might be interpreted in two different, but non-exclusive, ways. On the one hand, it might represent a plesiomorphic character shared with earliest Homo (cf. H. habilis) (125-157 cm and 32-52 kg), whereas the KNM-WT 15000 specimen appears to be derived in this respect (150.5-169.1 cm and 45.5-70.6 kg). On the other hand, differences in stature between the Dmanisi and KNM-WT 15000 hominins might reflect adaptation to different palaeoecological contexts (Lordkipanidze et al. 2007:308).

Except for one thing: They're not short.

Like too many papers these days, the details are hidden away in the supplements. Nobody's ever very interested in them, I guess. The supplements to this paper give most of the details about how the authors estimated mass and stature for the three individuals: the subadult represented by the D2680 humerus and D3160 femoral shaft fragment, the "large adult" reresented by the D4507 humerus, D4167 femur, and D3901 tibia, and the "small adult" represented by the D3442 first metatarsal.

Body mass estimates were calculated using the equations for femur, humerus, tibia, and metatarsal I [ref. 72, this is McHenry and Berger 1998]. The inferred body mass of the large adult individual is between 47.6 kg and 50.0 kg. The body mass of the small adult individual, calculated from the first metatarsal (D2671) is 40.2 kg. Based on humeral and femoral dimensions, the body mass of the subadult is between 40.0 kg and 42.5 kg.
Stature estimates for the subadult Dmanisi individual were obtained with prediction equations for juvenile samples; estimates based on humeral length (D2680) yield a value between 144.9 cm and 161.4 cm. Stature estimates for the large adult individual were obtained from humeral, femoral, and tibial dimensions, yielding a range of 146.6 cm - 166.2 cm. Stature estimates based on the length of the first metatarsal (D3442) yield a value of 143.0 cm (Lordkipanidze et al. 2007:S14).

Americans are handicapped to various extents because they lack an intuitive grasp of how long a meter is. The stature range for the subadult individual, 145 to 161 cm, is equivalent to a range from 4'9" to 5'3". For the "small adult", the single stature estimate of 143 cm is equivalent to 4'8" -- remembering that this is for a single foot bone. The "large adult" range of 147 to 166 cm is equivalent to a range from 4'10" to 5'5".

We can take a number of perspectives on these stature estimates. The Dmanisi adults were a bit shorter than the average American. According to the CDC, the average stature of American men aged 20 years is 176 cm (5'9"), with only 10 percent of men shorter than 167 cm at this age. Women aged 20 years have an average stature of 163 cm (5'4"), with 10 percent of women shorter than 155 cm at that age.

The Dmanisi subadult is a different story. American girls aged 12 years have an average stature of 151 cm (4'11"), and 95% of girls are taller than 139 cm. There's nothing very unusual about a 12-year-old who is 4'9" tall (145 cm), and the upper 95 percent confidence limit of 5'3" (161 cm) would have made this 12-year-old several inches taller than my wife Gretchen at that age. Twelve-year-old boys are not taller than girls -- they average around an inch shorter. The Dmanisi subadult skeleton is not short for a living human -- in fact, if the individual was a boy, he may have been a bit tall.

But living Americans are hardly the right comparative sample. Estimates of body size in early Homo have been framed around the question of whether the hunter-gatherer adaptation requires large bodies. For this question, we shouldn't compare the Dmanisi body sizes to fat Americans with their Flintstones childrens' vitamins, but instead to prehistoric hunter-gatherers.

Fortunately, there have been many analyses of stature in recent and prehistoric hunter-gatherer populations. Some of the comparisons in the current paper fit this criterion -- the North African Epipaleolithic sites of Afalou and Taforalt are in their comparative samples, which also include the bones of some early agriculturalists from Turkey. So to get an indication of the way the Dmanisi statures compared with these populations, we can look directly at Figure 3 of the paper. Here's the first panel, Figure 3a, which shows the Dmanisi tibia as a six-pointed star, and human tibiae as the letter "Z":

There, you can see the D3901 tibia is considerably shorter than the entire human sample. Except, oops! The figure is wrong. Table 1 reports a range of human tibia lengths from 290 mm to 374 mm; this figure shows a range from around 320 to over 440.

The correct range of tibia lengths is shown in Figure 3c, plotted as the y axis with femur length as the x axis:

There you can see the star for the D2901/D4167 individual, right in the middle of the recent human comparative sample. It's not short at all -- it's in the middle of the distribution.

The same thing goes for the D4507 humerus, illustrated along with the D4167 femur in Figure 3b:

A few comparisons with other hunter-gatherer samples confirm that the Dmanisi statures are typical of recent populations. Pretty and colleagues (1998) studied an archaeological sample of Aboriginal Australians from the Murray River region. Using stature estimation methods for the tibia, femur and humerus, they found that males in their sample (n=55) had an average stature of 166 cm and females (n=40) an average of around 153 cm. Wells (1952) reported a mean for !Khu (Northern Bushmen) males of 158 cm and females of 148 cm, both with standard deviations around 5 cm. Ruff (2000) puts the average stature of males at Pecos Pueblo at 161.2 cm with a range from 155 to 168 cm. In the KNM-WT 15000 monograph, Ruff and Walker (1993) report the average stature of African population samples, excluding Pygmies, as 162.3 cm. And although it is common knowledge that the Early Upper Paleolithic people of Europe were tall, the average male stature in the Late Upper Paleolithic was around 166 cm, and the average female stature around 153 cm (Formicola and Giannecchini 1999) -- virtually the same as Australians.

At their expected values, the statures of the Dmanisi adults were approximately the same as !Khu and Pecos Pueblo, and around four inches shorter than the averages (but taller than more than 10 percent) of these other groups. Compared to living people, they just weren't short.

That is all assuming that the "large adult" specimen is actually a male. Lordkipanidze et al. (2007) support this assignment based on the proximity of the remains to the D2600 mandible, which is clearly a large male. I don't have any reason to doubt the assignment, although the stratigraphic details in the paper don't clearly show the association -- the "large male" remains including D2600 appear clustered, but the specimens aren't labeled and don't all seem to be represented. If the skeleton turned out to be female, it would be an inch or two taller than average for the larger groups above.

I have focused on stature rather than mass, mainly because it is more reliably estimated from bone lengths than mass is from articular breadths, but also because it is more heritable. Still, the same basic observations apply: hunter-gatherer populations are not heavy people, and a mass estimate of 50 kg would not be exceptional for a male.

So why is everybody saying that these individuals are small? The real contrast is not between Dmanisi and living people, but between Dmanisi and the large East African "H. erectus" specimens, like KNM-WT 15000, KNM-ER 1808, KNM-ER 736, KNM-ER 739, and OH 28. And yet, these large specimens are hardly typical in East Africa: they are the upper end of a range of variation in postcrania extending down to Lucy's size, barely more than a meter tall. We have often assumed that these larger specimens belong to H. erectus, and I have argued for such an assignment in print (Hawks et al. 2000). But I think that the lower end of this range of variation is completely up for grabs -- especially considering the small size of the KNM-ER 42700 cranium.

There is one good argument for associating East African "Homo erectus" exclusively with the large-bodied specimens: KNM-ER 1808 and OH 28 are both apparently female (based on their pelves), but both have tall statures, based on their femora. McHenry (1991) puts KNM-ER 1808 at 180 cm and OH 28 at 171 cm. It is the large size of these female specimens that argues for a reduction in sexual dimorphism and average large body size in Homo erectus. It is that association -- low sexual dimorphism and large body size -- that argued for a significant increase in home range size and dispersal potential in this species. I'll call it the "long-legged colonists" hypothesis: the idea that hunter-gatherer ecology, large body size, and low sexual dimorphism were linked to each other, all enabling long-distance dispersal and the initial colonization of Eurasia. The Dmanisi body sizes refute this hypothesis.

But looking back, the "long legged colonists" hypothesis was half incorrect chronology and half wishful thinking. Why would early humans have needed statures near the extreme of modern human populations, if recent hunter-gatherers have relatively small bodies? Recent hunter-gatherers have maintained large home ranges, sexual division of labor, and large mammal hunting with statures no larger -- and often smaller -- than the current global average. The largest stature estimates for early Homo fossils are well above the average statures for any but the very tallest human populations.

Even the tallest modern human populations average substantially shorter than the tall East African fossil stature estimates. Ruff and Walker (1993:259) report the mean for living Africans "of tall stature" as 166.6 cm. That's a midsex average of 5'6" for tall populations. The tallest population in the world now is the Dutch, where 21-year-old males average 184 cm. That's virtually the same height as estimated for KNM-WT 15000 as an adult, but remember that the Dutch stature is an average; as it stands, KNM-WT 15000 is an extreme. Early East African Homo was not as tall as late-twentieth century Dutch; they must have averaged substantially less.

And as for chronology: all of the tall-stature early Homo specimens are now substantially later in time than Dmanisi. Only KNM-ER 1808 might approach Dmanisi in age. The rest of these tall stature specimens are at least 200,000 years younger.

We are left with a remaining question about variability: Were these early humans (Homo erectus) unusually variable in size? I don't think so. If anything, they appear to have exhibited less variation in stature than human populations today. No ancient population was as tall as the Dutch. It is not even clear that early Pleistocene East Africans were as tall as recent East Africans, although they may have been so. No fossils yet assigned to Homo erectus were as short as Pygmies; although some Homo habilis-associated postcrania were even shorter. If the species boundaries are drawn right, there may be no problem of variability in the postcrania.

That may be a big "if". The limited degree of variation is fairly remarkable considering that the fossils in question span over a half-million years of time, in East Africa and Eurasia. Maybe there ought to be more variation than anyone is now assigning to H. erectus, and the species boundaries are wrong after all...


Formicola V, Giannecchini M. 1999. Evolutionary trends of stature in Upper Paleolithic and Mesolithic Europe. J Hum Evol 36:319-333.

Fredriks AM, Van Buuren S, Burgmeijer RJF, Meulmeester JF, Beuker RJ, Brugman E, Roede MJ, Verloove-Vanhorick SP, Wit, J-M. 2000. Continuing positive secular growth change in the Netherlands 1955-1997. Pediatric Res 47:317-323.

Lordkipanidze D and 17 others. 2007. Postcranial evidence from early Homo from Dmanisi, Georgia. Nature 449:305-310. doi:10.1038/nature06134

Lieberman DE. 2007. Homing in on early Homo. Nature 449:291-292. doi:10.1038/449291a

Pretty GL, Henneberg M, Lambert KM, Prokopec M. 1998. Trends in stature in the South Australian Aboriginal Murraylands. Am J Phys Anthropol 106:505-514. doi:10.1002/(SICI)1096-8644(199808)106:4<505::AID-AJPA5>3.0.CO;2-H

McHenry HM. 1991. Femoral lengths and stature in Plio-Pleistocene hominids Am J Phys Anthropol 85:149-158.

Ruff CB, Walker A. 1993. Body size and body shape. Pp. 234-265 in The Nariokotome Homo erectus skeleton, Walker A, Leakey R, eds. Harvard University Press, Cambridge MA.

Ruff CB. 2000. Body size, body shape and long bone strength in modern humans. J Hum Evol 38:269-290. doi:10.1006/jhev.1999.0322

Wells LH. 1952. Physical measurements of northern Bushmen. Man 52:53-56.