Why elite runners land in front of their COG

One of the strong beliefs within most of the schools of efficient running that developed subsequent to the ground breaking attempts of Gordon Pirie to understand how to run efficiently, is the belief that one should aim to land with the foot travelling backwards relative to the centre of gravity of the body (COG) so that the foot is travelling at zero velocity relative to the ground at foot-strike. This would be expected to avoid a braking action that is both wasteful of energy and also raises the risk of injury. Pirie is emphatic: ‘Over-striding is one of the most common technical afflictions of runners and is one of the most dangerous’ (Pirie, Running Fast and Injury Free, p18). However, observation of elite athletes reveals that at footfall the point of contact is in fact usually in front of the COG.

Some coaches have tended to relax the demand to land under the COG and have suggested that it might be best if the foot lands a little in front of the COG. For example, in a comment on my blog ‘Where should the foot land’ posted on January 2nd, Pose runner Bill McGuire, said:

‘Jack Becker, who is the main guiding voice on the Pose forum, has suggested a few times that landing a little in front of the COG is acceptable. I can’t say for sure, but I suspect his reasons are based on his formidable intuitive feel for Pose more than on mathematics.’

I too have a great respect for Jack Becker, and I believe his intuition is correct. My reasons are based on the principles of rotational motion. The issue can be discussed without mathematical formula, but it does require at least an intuitive understanding of Newtonian mechanics. Fortunately, at least with the hind-sight available to us in the modern world, most of Newtonian mechanics appears intuitively correct. Certainly Newtonian mechanics appears more in tune with common-sense that Einstein’s relativistic mechanics which superseded it. The reality is that for human-sized objects moving at running speed on the surface of the earth, Newtonian mechanics provides an excellent description of reality. So, let us re-examine the issue that I addressed in my blog on January 2nd in a little more detail

In the language of Newtonian mechanics, the arrest of the foot on the ground at foot-strike interacts with the impetus of the forward momentum of the torso to create a torque (a ‘twisting effect’) that results in rotational motion. That is, when the torque is applied, the angular momentum of the body increases. It is crucial to appreciate that a torque can only be applied by contact with an external object such as the ground. Re-arrangements of the relative placement of limbs without leverage on the ground cannot change the angular momentum. When the moving body contacts the ground, angular momentum can be imparted to the body. The angular momentum of the body will then remain constant unless another external torque acts on it. This is the law of conservation of angular momentum (see http://en.wikipedia.org/wiki/Angular_momentum).

If the initial torque is uncorrected, this rotational motion will result in a face-down crash within a few strides. In conclusion, during the period of stance, there must be an initial period in which the foot is arrested and a ‘head forwards and down’ rotation is generated, and this must be followed by a subsequent time interval in which an opposite torque is applied.

The angular momentum imparted will only be appreciable if the torque is applied for an appreciable length of time. In theory we might avoid the problem of rotation if we could support ourselves using only an instantaneous touch down. But an instantaneous period on stance would result in infinite vertical ground reaction force to ensure that the weight of the body can be supported. Therefore, we must remain on stance for an appreciable time. If vertical GRF is to remain less than three times body weight, we must be on the ground for at least one third of the gait cycle.

Where must the foot fall? The first thing that must happen during stance is the foot must be arrested. To achieve this, the ground reaction force must be directed backwards. This can be achieved by landing slightly in front of the COG, so the downwards pressure of the impact is directed downwards and slightly forwards. This induces a backwards directed ground reaction force (GRF) that arrests the foot and the lower leg. Once the COG has passed over the point of support, the downwards force exerted by the body on the ground is directed obliquely backwards, generating a forwards directed component of GRF. As discussed in my post earlier today, the impulse imparted to the foot by this forward directed ground reaction force is almost adequate to accelerate the foot and lower leg to a forward velocity matching that of the torso. Angular momentum is conserved. Rotation is corrected and very little energy is lost. Nonetheless, it is crucial to appreciate that unless we land in front of the COG, a face down crash will eventually occur. These considerations suggest that attempting to land immediately beneath or behind the COG is not only pointless, it is potentially injurious.

What observational evidence supports this conclusion? As mentioned previously, elite athletes such as Haile Gebreselassie tend to land with their foot in front of the COG. What happens in elite athletes who employ the Pose style? Recently Jeremy Huffman, elite sub-4 minute miler who has subsequently adopted Pose (and generously offers expert advice on the PoseTech forum and elsewhere) posted some excellent pictures of himself taken during early and mid-stance, on the Fetcheveryone website (http://www.fetcheveryone.com). In the picture taken very slightly after mid-stance (when the leg is bearing the maximum load) the point of support (the ball of the foot) is slightly behind the COG. However, in the earlier pictures, the point of contact of the foot with the ground is clearly in front of the COG. We cannot easily estimate how much ground reaction force is being exerted in the period before the COG passes over the point of support from these pictures. However, force plate data acquired by Cavanagh and LaFortune (Journal of Biomechanics, 1980) provide a strong clue. In the runners who landed on mid-foot there was a horizontal backward directed ground reaction force acting during the period from first contact with the ground until the point of maximum load bearing. After that time, the horizontal ground reaction force is reversed. A similar phenomenon was observed in heel strikers, though the shape of the GRF curve was somewhat different. As predicted in light of the law of conservation of angular momentum, in both mid-foot strikers and heel strikers, the average value of the backward directed GRF in the first part of stance was approximately equal to the average value of the forward directed horizontal GRF acting in the final part of the stance phase.

If we accept that it is inevitable that we must experience a braking force early in the stance phase, how can this be done safely? As recently suggested in the Fetcheveryone discussion of efficient running by coach emjaybee, the really important goal is having the knee flexed so that the point of foot contact is behind the knee joint. This will facilitate absorption of much of the impact energy in the quadriceps muscles, allowing for subsequent recovery of the energy at lift-off, and minimizing injurious jarring of the knee and other joints. (http://www.fetcheveryone.com).


5 Responses to “Why elite runners land in front of their COG”

  1. Simon Says:

    Hi Canute, I largely agree and I have a few points that may be worth considering;

    1. I believe Jack may have said that the foot fall can be forward of the GCM so long as it is not bearing weight. To me, that means the foot can make contact but then muscular activity must minimise loading until it is under the COG. The kind of Pose forum posts I have seen this arise in are when a runner is being discussed and someone says the runner has landed ahead of their GCM. A closer look at the shape of the foot and leg reveal that it is not bearing noticeable weight (foot pronation and more knee compression indicating an increased load).

    2. In theory, would applying conservation of angular momentum to the asymmetric pattern of running (much more rearward extension than forward) mean that the forward contact that stops the angular momentum would have to have a much larger impulse and would show up in the force plate data?

    3. In practice, I can’t observe angular movement continuing in the airborne phase when I watch videos of runners and I wonder if it is either tiny or somehow countered during the final instances of the support phase? Would a horizontal rearward force at the foot as it passes from mid to terminal stance tend to counter the affect of generating angular movement?

  2. canute1 Says:

    Simon, Thanks for your comments.
    I think the reason you do not see rotation during the airborne phase is because there is an equal and opposite torque applied during the stance phase. If we ignore possible changes in the distance from point of support to the COG, then equal and opposite torque in early and late stance implies equal and opposite impulse from horizontal GRF during early and late stance. (By impulse, I mean the integral of the GRF over time, or in simpler terms, the average force multiplied by the time for which it acts) The available force plate data from Cavanagh and LaFortune suggest the early backward directed impulse almost exactly balances the later forward directed impulse, so we would not expect to see rotation when airborne.

    Your comment raises the question of whether the equal and opposite impulse might be produced by some mechanism other than landing in front of the COG. In principle this is possible. If foot-fall immediately under the COG was followed by a strong quadriceps contraction accompanied by simultaneous hip extensor action sufficient to stabilise the thigh, the foot would tend to be pulled forwards generating a backward GRF. If the hip flexor action then increased further, the subsequent effect would now be a force pulling the foot backwards, generating a forwards GRF. Thus a well timed action of quadriceps plus hamstrings followed by increased hamstring action could probably generate the desired balancing of backwards and forwards impulse in early and late stance. However, this sequence of muscle contractions would probably be more consumptive of energy that simply landing a little in front of the COG, storing the energy in the quads and then releasing it, accompanied by some hamstring contraction in late stance. Thus, on balance I think it would more efficient to land a little in front of the COG.

    The question of whether or not the leg of a Pose runner bears a substantial load between foot-strike and the point where COG passes over the point of support could best be answered by examining force plate data collected from a Pose runner.

  3. Simon Says:


    Yes, I see what you are saying; because there is a rearward rotation introduced by early stance, the forward rotation in late stance only cancels it. It also fits nicely with observations of runners that stay on support longer – they nearly always extend further to the front too and I believe this is a prime candidate to explain why. Thanks for that – another piece of the puzzle is clearer.

    This leads on to your second point about landing under the COG. The ideal of landing exactly under the COG looks like it would require a very early pull to avoid generation of forward rotation i.e. the pull would have to happen instantaneously (leading to infinite GRF among other impossibilities). So, if my understanding is correct, landing very close to the COG will require very little time on support to minimise forward rotation which in turn will require a very high cadence to minimise the ratio of airborne time to support time. Does that sound right?

  4. canute1 Says:

    Simon, Yes, I agree.
    As you suggest, we can minimize time on stance by increasing cadence, though issues such as the rate at which energy can be transferred to muscles and ligaments and then recovered by elastic recoil will probably set a limit on how short the time on stance can be. My rough estimate is that once the time on stance is less than 90ms we would not be able to recover energy by elastic recoil effectively. It would be interesting to examine videos from some sprinters to see what time they spend on stance.
    Also we would require a very powerful action to lift-off at very high cadence. Although recoil might provide a substantial proportion of the energy required to accelerate the lower leg and foot to match the speed of the torso, that estimate did not the energy required to increase the gravitational potential energy of the leg, nor the energy required to move the thigh etc, so in practice the maximum cadence will be limited

  5. Rebbeca Ammer Says:

    I really happy to read this post,I was just imagine about it and you provided me the correct information .

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: