You heard it here first :: hobbits are australopithecines!

This is Richard Roberts in an Australian radio interview (the interview is formatted in one-sentence paragraphs, this is a single contiguous excerpt):

Let's take a point of argument that this particular individual with a small brain is a microcephalic individual, is such an individual.
They have other features that indicate they're not suffering from microcephaly, they have unusual tooth structures - three roots to the teeth.
You find those in three-million-year-old people like Lucy in Africa, that only exist in very early Homo erectus.
You don't find those with modern humans.
We don't suddenly develop three roots to the teeth.
Nor do you suddenly develop long arms if you have microcephaly.
And that's what the hobbit has, they have slightly longer arming compared to ourselves.
The pelvis is wider than in modern humans.
They have very thick eyebrow ridges.
None of these are features of microcephaly.
When you look at a complete set of features of the skeleton, one or two of them might be credible as being microcephalic problems, but the rest of them can't be explained by microcephaly.
If you pick some of the ones like Professor Jacob has done I can understand how he reached that conclusion.
But not on the basis of all available features.

In my book, this fits the "quacks like a duck" definition of australopithecines. Nothing really new here (except a confirmation of the second jaw earlier in the interview). It's not clear to what extent these comments really reflect the biologists'; for example, LB1 has no preserved arms, and the radius from the site is not all that long.

Trawling through the original Nature article, we find these descriptions of australopithecine-like similarities.

There is a strong posterior angulation of the symphyseal axis, and the overall morphology of the symphysis is very similar to LH4 A. afarensis and unlike Zhoukoudian and Sangiran H. erectus. (Brown et al. 2004:1058)
The femur shaft does not have a pilaster, is circular in cross-section, and has cross-sectional areas of 370 mm2 at the midshaft and 359 mm2 at the midneck. It is therefore slightly more robust than the best-preserved small-bodied hominin femur of similar length (AL288-1). Distally there is a relatively high bicondylar angle of 148, which overlaps with that found in Australopithecus (1059)

. </p>

I would add to these the prominence of the anterior superior iliac spine, and I would quadruple the weight of the brain size in the decision--I just can't see a Homo-like brain size reducing to the smallest known for any hominid as a consequence of dwarfism.

Against these, we have a number of Homo-like features.

Although LB1 has the small endocranial volume and stature evident in early australopithecines, it does not have the great postcanine tooth size, deep and prognathic facial skeleton, and masticatory adaptations common to members of this genus. (1060)

But all of these specifically Homo-like similarities are problematic, because they all are correlated with a single feature: dental reduction. And several of these features are not shared with the members of early Homo to which the specimen is proposed to be related.

The tooth sizes of the LB 1 specimen are generally shared with humans. This means, significantly, that they are not synapomorphies of early Homo--it would take nearly as much change to generate this pattern from an early Homo dentition as from an australopithecine or habiline.

I haven't made up my mind entirely, and much of my consideration runs with the brain size. Again, I would say that pathology must be rejected for the brain size before turning to other hypotheses. But the recovery of additional similar specimens would make pathology seem infinitesimally unlikely as an explanation for all of them. We aren't there yet, but we must be on the cusp of it. A divergence date from mtDNA would help a lot.

More on the Flores hominids