The classic "ascent of man" image has our lineage progressing upwards. First, we used to knuckle-walk like modern chimps. Over time we gradually became more and upright, straightening our backs and striding forwards as modern man. It's a noble image of our origins and almost ubiquitous with the field of human evolution.
Except this famous picture is probably wrong. And not just because of its European male focus. There is growing debate over whether our ancestors ever did knuckle-walk like a chimp.
Modern primates
Modern primates move about in a range of different ways. From swinging through the trees to walking along branches to that classic knuckle-walk and much more. Since Darwin wrote about evolution researchers have been arguing which one of them was the precursor of bipedalism.
Indeed a lot of early attempts to find the fossils of our ancestors was based in Asia because scientists noted that human bipedalism is very similar to orang-utan bipedalism. Based on this many thought we evolved from them in Asia, where orang-utans lived. That's why a lot of the first hominin fossils from the late 19th/early 20th centuries (such as " Java man ") are from that part of the world.
With the discovery of Australopithecus in Africa, the focus shifted to that continent instead. There, people began to notice chimps and soon worked out that they were actually our closest living relatives. Because of this, they became the de facto model for early hominin behaviour, including the source of bipedalism. Chimps do sometimes walk upright and when they do so they walk with a "bent-hip, bent-knee" posture. From this semi-squat bipedalism evolved?
Walking in this manner is woefully inefficient. Keeping the legs and hips bent requires that the muscles there be working continuously which is both costly and painful. It isn't all bad, being a lot more stable than human bipedalism. However, you can clearly see that there would be some pretty strong selection pressures to make it more efficient.
Challenging chimps
However, there are some strange aspects to chimp bipedalism which some researchers argue is incompatible with this model. In particular, there's the fact that chimpanzees walk the way they do not because of any anatomical failing in their legs or hips but because their spines aren't flexible enough. They can't bend their back to alter their centre of gravity so instead, have to bend their knees.
This lack of flexibility stems from the fact chimpanzees have lost a vertebra, shortening their lower spine. Also, that spine is trapped between their tall iliac blades. This is to make their back more rigid, allowing them to knuckle walk with significantly less effort. Bonobos, an offshoot of chimps, have also shortened their spine but have done so in a different way. Chimps have fused two of their vertebrae together whilst bonobos have moved the spine upwards, shortening the lower part of it.
Given that out of humans, chimps and bonobos 2/3 of them have long spines it seems reasonable to conclude that their common ancestor had a long spine and the chimpanzee shortening is relatively recent. As such that last common ancestor would have been better at walking bipedally than modern chimps and have less "evolving" to do in order to become more efficient. There would also have been only a small amount of evolution necessary to become better at knuckle-walking.
Did fossils knuckle-walk?
This obviously raises the question of how did this adaptable common ancestor move? Did it move in a different way which split into knuckle-walking and bipedalism? Alternatively was it a knuckle-walker that became better at knuckle-walking and also split into bipedalism, or a bipedal animal that became better at bipedalism and split into knuckle walking?
To figure out which is the right answer requires looking at the hominin lineage since our ancestors are the closest (time-wise) to the last common ancestor. The chimpanzee lineage is woefully incomplete forcing us to rely on our own to delve back to the splitting point.
The closest specimen we have is Sahelanthropus tchadensis which is so near to the split some even suggest it may be the common ancestor. However, we only know it from a partial skull, so sadly it isn't much help for studying locomotion. The next up is Orrorin tugenesis which is a bit more useful. Their femur shares enough similarities with our own to indicate it was bipedal.
Ardi to the rescue
Ardipithecus ramidus provides the best information on this topic with over 45% of the skeleton being recovered. Dating to ~4.4 million years ago it falls almost halfway between modern humans and our common ancestor with chimps. Thus one might expect it to be morphologically halfway between humans and chimps. About halfway upright on "the ascent of man", if you will.
Yet it isn't. Ardi, as one find is affectionately known, was able to walk upright pretty well. Whilst "she" wasn't as good at it as later hominins her bipedal capabilities were still better than chimps. There's no evidence that it engaged in bent-hip, bent-knee locomotion.
Indeed, there's no evidence she would ever knuckle-walk. However, there does seem to be some evidence that it spent a fair bit of time in trees and whilst it did so it walked on all fours across branches. Not knuckle walked, but with the foot flat. It also appears to have done quite a lot of swinging whilst up there based on its arm adaptations.
Conclusion
Given all of this, it seems we can "redraw" the ascent of man (although new finds are still changing things so this is far from the final redraft). Since it turns out chimps evolved to knuckle-walk later, our common ancestor instead moved more like Ardi did in the trees. Swinging from branch to branch and walking flat footed on all fours across the branches she arrived at.
As the "locking" of the spine (which prevents bipedalism) also appears to be a feature unique to chimps the common ancestor would also have been fairly decent at bipedalism had it ever gone to the ground. Whilst it would've been no Homo erectus it would not have had to rely on the "squat" walk of chimps.
Two lineages began to emerge in this common ancestor. One appears to have focused on the quadrupedalism used in the trees, starting to use it on the ground. To make this method more efficient this branch eventually turned the flat-footed quadrupedalism into the knuckle-walking quadrupedalism present in chimps today.
The other focused on their bipedal ability, gradually making that more and more efficient until they could stride out of Africa and spread across the world. Their brains expanded and they began to wonder where they came from, eventually looking back to the common ancestor who had started it all by dropping out the trees.
References
Lovejoy CO, & McCollum MA (2010). Spinopelvic pathways to bipedality: why no hominids ever relied on a bent-hip-bent-knee gait. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, 365 (1556), 3289-99 PMID: 20855303
