Gazelles v gliders: Mirinda Carfrae v Chrissie Wellington

As described in a recent post, my attempt to recover some of the speed of my youth by engaging in a lifting program to re-build my leg strength was only partially successful.  I exceeded my expectations regarding gains in strength, but so far this has not been translated into increased speed.  Unfortunately a recurrence of arthritis has confounded my immediate hopes, and at the moment I am more concerned about re-building my aerobic base.  However a recent discussion of the merits of gazelles v gliders on the Fetch efficient running thread has prompted me to re-examine the issue of my loss of speed.

The most striking thing about the change in my running style as I have grown older is the fact that my stride length has shortened.  Now, whenever I try to increase pace, my cadence automatically increases, often going well above 200 steps per minute even at a modest pace. While many recreational runners might benefit from an increase in cadence, at least up to 180-190 steps/min, I am fairly sure that in my case, the increase in cadence reflects reduced ability to get airborne, leading to a stunted stride.  I had attributed this to lack of strength but maybe strength wasn’t the main problem.

The gazelles v striders comparison provides food for thought.  Here is a good illustration (though I do not agree with all of the comments by the commentator).   The crucial difference between gazelles and gliders is that gliders do not produce as much elevation of the body on each step as gazelles. Because they produce less elevation than gazelles, their stride is shorter and they employ a higher cadence.  While I am not sure that even my mother would have ever described me as a gazelle, there is little doubt that I have become a glider as I have aged.   This is what I have been trying to correct.  However, this video clip provides at least some grounds for questioning the need to overcome my tendency to be a glider.   As the video illustrates, Chrissie Wellington, without doubt the greatest female triathlete ever, is a glider.    In the video, Chrissie’s gliding is compared with the style of one of the most elegant triathlete gazelles, Mirinda Carfrae.

The costs of gliding

Could it be that gliding is efficient?  It is tempting to think that reducing elevation costs must be more efficient, but this would be far too simplistic.  When considering efficiency, we need to consider the three major energy costs of running:

1)            Overcoming braking.  Provided a glider increases cadence to ensure that time on the ground does not increase, the braking cost per step will be the same for both but the cost per  mile will be  greater for the glider because there are more steps per mile.

2)            Limb repositioning costs: these increase with cadence and will be higher for the glider

3)            Elevation costs:  Although the video commentary  incorrectly states there are no elevation costs, in fact elevation of the centre of mass occurs before lift- off as a result of extension of hip, knee and ankle during the late stage of stance.   Furthermore due to the higher cadence, the saving in elevation cost per step is partially offset by the greater number of steps per mile.  However, the elevation cost per mile will nonetheless be somewhat lower for a glider because elevation cost increases as the square of airborne time, so the saving in elevation cost per step is relatively greater than the extra cost due to more steps per mile.

In estimating the total cost, we need to balance the three variables: braking and limb repositioning costs are greater for the glider, but elevation costs are less.  At higher speeds, repositioning costs become the dominant cost and there is no doubt that at high speed (e.g. faster than 7 min/mile) gazelles are more efficient.  At intermediate speeds (7min/mile-10 min/mile) braking costs and elevation costs are both quite appreciable and at such paces too, gazelles are almost certainly more efficient.  At very low speeds (slower than 10 min/mile) braking cost become relatively small because the leg is never far from vertical and therefore the horizontal ground reaction force that produces braking is always low.  At such paces the major goal should be minimising elevation costs.  So on balance, I think it is only at very slow paces that gliders might be more efficient than gazelles.

So why is Chrissie Wellington a glider?   I do not know, but wonder whether it might be an unconscious attempt to decrease the risk of injury when tired in the late stage of an ironman.  With regard to injury, the issue is the relative risk of a larger number of smaller impacts for the glider compared with fewer larger impacts for the gazelle.  While the phenomenon of repetitive strain injury demonstrates that repeated small impacts can be damaging, I suspect that size of the impacts plays an even bigger role in damage.  Therefore, I am inclined to think that for a tired runner, (either in the late stages of an ultra or an ironman) the risks will be lower for the glider.

Overall, there is little doubt that the gazelle style is better for medium paced and faster running, but there is reason to debate whether or not the glider style might be beneficial for tired, slow runners. I am still eager to become as much like a  gazelle as my aging limbs will allow.  While I can no longer blame lack of strength, I wonder if maybe lesser ability to capture elastic energy at foot fall is the cause of my gliding.   In a future post I will describe my plans for the attempt to recover elasticity.   But in the short term, my focus is on re-building my aerobic base, and that is what my next post will address.

Finally, it is noteworthy that on the two occasions when Chrissie Wellington and Mirinda Carfrae went head-to-head in the world ironman championship (at Kona in 2009 and 2011), on each occasion Wellington won the overall event but on both occasions Carfrae ran the faster marathon. In my opinion Wellington is the greatest female triathlete ever but Carfrae is the more efficient, and faster, marathon runner.

Added note (4th May 2013)

In his comments below, Robert had pointed out that I have not provided adequate justification for my claims about the energy costs.   My claims are based largely on calculations based on applying Newton’s equations of motion to the simplified model of running described in my posts in January and February 2012.    At a pace of 4 m/sec, (which is close to that of Carfrae and Wellington in the world ironman championship in 2011) the calculations demonstrate that the combined cost of elevation and braking is 6% greater for a Glider than for a Gazelle, assuming that that the Glider has an increased cadence sufficient to produce a similar time on stance.  Observations suggest that Gliders do increase cadence to maintain a similar time on stance, and furthermore, if a Glider did not have increased cadence they would be even less efficient because the longer time on stance would produce an even greater increase in braking costs (as indicated by my post of February 2012).

The crucial questions is whether or not the simplifications I used in performing these calculations lead to serious errors. There are two respects in which the simplifications might lead to error.  First, I assumed a sinusoidal shape for the variation of ground reaction force with time.   Variations in the shape of this curve introduce small changes to the results of the calculations, but in the absence of force plate data for each runner, I cannot do a more precise calculation.  However, the error due to this simplification is likely to be small.

The other potential source of error is that my calculations are for the total energy expended on elevation and overcoming braking. I have not subtracted the energy saving expected via elastic recoil.  If a runner maintains greater tension in the leg muscles, as recommended in the BK running style, it is possible to recover a greater proportion of the required energy via elastic recoil.  I cannot exclude the possibility that a Glider might maintain greater tension in the leg muscles, but think that in general this is unlikely as increased tension in the leg muscles results in a greater rate of rise in ground reaction forces and potentially increases the risk of injury.  The advocates of the BK style recommend thorough preparation using plyometrics before attempting this.  I think that increasing leg muscle tension would not be a good strategy for a tired ultra runner.  Thus I doubt that Gliders make greater savings via elastic recoil than Gazelles. I suspect that the opposite might be the case, though this is speculation.  However even if equal efficiency of recovery of energy via elastic recoil is assumed, the Glider incurs a 6% greater cost for elevation and braking than the Gazelle.

Also, it should be noted that my calculations do not include limb repositioning costs. These costs are almost certainly higher for the Glider because repositioning cost increases with cadence (as discussed in my post in April 2012) and furthermore, the trajectory of the foot of a Gazelle results in a shorter lever arm of the swinging leg, further increasing efficiency.

On balance, I think it is likely that my calculations provide a fairly realistic estimate of the relative costs of elevation and braking for the two styles.

28 Responses to “Gazelles v gliders: Mirinda Carfrae v Chrissie Wellington”

  1. Robert Osfield Says:

    I don’t find the classification “Gazelles v Gliders” useful. The classification is really down giving a silly name to observed differences in time in the air and what effects this has on other parts of the gait to accommodate this. Silly names just confuse things and would much rather people just properly explain the differences without layering on extra marking style crap, to me it suggests they don’t really understand running mechanics correctly so have to invent stuff.

    I also am not convinced a runner with short time in the air vs long time in the air will be any less efficient at the paces that these two talented ladies are running at. Running with short time in the air also typically have shorter time on stance than runners with longer time in the air which will counter out any increased limb repositioning cost associated with a higher cadence. The only way to know where the balance lies for a particular athlete would be to measure the running economy.

    Without the measurements you’d have to run a theoretical model that is validated against real data. Using an uncalibrated theoretical model to predict tradeoffs might provide indications of trends but won’t be accurate enough to know when change overs in maximum efficiency would occur i.e. you won’t be able to accurately predict what cadence and time in air/time on stance loading is most efficient at different speeds.

    From what we do know about experimental data is that higher cadence and shorter time on stance are the two key changes as people run faster and the maximum speeds are dictated largely by how short a time on stance a runner can generate. Also shorter time on stance will also be expected to increase running economy.

    As you point out the tradeoff with long time in air/short time on stance is increased maximum loads and loading rates. However, if time in air is reduced inline with reductions in time on stance the loads will not change so a short time on stance doesn’t not directly mean higher loads.

    On the efficiency balance front I would also expect lower loads associated with short time in the air would also put lower demands of the muscle fibres which in turn would mean the balance of utilization of slow and fast twitch fibres would change, enabling more slow twitch fibres to be engaged, or simply less fibres of any sort. Given this I would expect a short time in the air to be more aerobic than long time in the air and as I result I would expect better endurance performance.

    Muscle damage is not a linear progression, a 10% increase in loads will increase muscle damage faster than 10%, so again this will tilt things in favour of the short time in air for long endurance races.

    The aerobic and muscle damage sparing aspects of low time in the air relative to time on stance is something you notice in ultra runners “ultra shuffle” gait. An extreme shuffling gait is likely to be mechanically less efficient but after many hours of running being able to run at all is a challenge. There is a paper on a the mechanical efficiency of ultra runner before and after a massive multi-day journey and found him, at the end, to be less efficient favouring a gait that required lower loads.

    • canute1 Says:


      Thanks for your comment.

      While it is true that the most reliable answer regarding efficiency can be obtained by direct measurement, I think it is possible to make meaningful estimates of the likely costs of the two styles. I think this worthwhile because many ultra runners raise the question of whether a style with reduced elevation is more efficient. In my experience, when discussing this, they do not take account of braking costs. The main point of my post was to draw attention to the braking costs, and furthermore to emphasize the implications of increasing cadence.

      The relationship between time in the air and time on the ground depends on how the runner adjusts cadence. At constant cadence, short time in the air is associated with longer time on the ground, at a given pace. However, as illustrated in the link I provided, Gliders tend to increase cadence relative to Gazelles in a way that results in similar time on the ground for both styles of running.

      Based on the simple model presented in my posts of Jan and Feb 2012, for a pace of 4m/sec, the sum of braking + elevation costs for a gazelle with a cadence of 180 steps/min and a vertical ground reaction force of 3 times body weight is 1203 Nm /Km. A glider with a cadence of 200 steps /min needs to achieve a vGRF of 2.77 times body weight to achieve the same ground contact time. At 200 steps/min and vGRF = 2.77 x body weight, the cost of braking + elevation is 1278 Nm/Kg. This calculation ignores energy saving due to elastic recoil, but the energy saving via elastic recoil is likely to be greater for the gazelle because elevation cost is a large proportion of the total for the Gazelle, so it is likely that after allowing for elastic recoil, the saving in total costs for the Gazelle will be even greater than 75 Nm/Kg. Furthermore, limb repositioning costs will be less for the Gazelle (as discussed in my post of 27th Feb 2012.)

      Thus, as far as can reasonably be estimated the energy costs at 4 m/sec (typical of Carfrae and Wellington in the marathon run during an ironman) the costs for a Gazelle are likely to be 6-8% lower than for a Glider who adjusts cadence such that both runners achieve similar ground contact times.

      Insofar as a particular runner might have better ability to recover elastic energy or a more efficient trajectory of the arms and legs, it might be that in an individual case, a Glider might improve their efficiency, but I believe that it is worthwhile demonstrating that the argument that gliding might be more efficient due to decreased elevation costs is not true.

      With regard to risk of injury, we are in agreement. I also agree that tired ultra runners do tend to adopt a style with lesser elevation, though unfortunately, they often also decrease their cadence, resulting in even greater inefficiency. Unless the runner is very fit, a major priority in the late stages of an ultra is often surviving without injury. However, if one has the level of fitness typical of elite ironman triathletes, I believe that the Gazelle style is likely to result in a faster marathon.

      It is also worth noting that for slower runners, the loss of efficiency of the Gliding style is less serious.

      • Robert Osfield Says:

        I am not convinced that your analysis is sound enough to support the conclusions you are making. A few odd things jump out:

        The losses to breaking forces being greater for the runner with lower time in the air but the same time on the ground is something that you assert but don’t substantiate. Just because a runner is taking more steps doesn’t mean that losses to breaking are higher, a runner with the same ground contact time but lower time in the air will have lower breaking forces so the losses per stride are lower, balancing out the increases cost per unit distance.

        You also suggest that a runner with high time in the air for a given time on stance will save more energy due to elastic recoil because the proportion of energy required for elevation is higher. This doesn’t make sense as recoil returns as percentage will be similar for both cases, and in both cases you aren’t getting energy for free, it’s simply the recoil reduces the actual costs – half of 10 units is always going to be more than half of 8 units, even if you say that you are saving 5 instead of 4 units.

        The reduction in energy costs due to recoil also effect elevation and breaking costs – both costs are born by the same mechanical system and the energy return works via that exact same system. You only mention that recoil in the context of elevation costs so it’s not clear whether in your analysis you’ve applied it to breaking costs.

        Energy cost of higher cadence is likely to be something that is tuneable, a runner that is runs habitually at a lower cadence will have tuned their muscle activation and stiffness for this cadence and changing cadence up will make then relatively less efficient. A runner with habitually higher cadence will likely to be tuned to have a stiffer torso as the range of motions required are lower and will have a lower positioning costs for high cadence than the habitual low cadence runner. To do accurate experimental measurement of limb positioning costs one would really need long period of time for the runner to retune their gait to the new cadence.

        Finally there is the aerodynamic costs that will be similar in both cases and add to total costs. While these are additive to the cost of breaking, elevation and limb positioning, at 4m/s these costs cease to negligible and will substantially change any % differences in overall costs.

        Putting all this together and I’d suggest that one has to be very careful about making predictions based on s implied theoretical model. Making general comments about one gait over another being a more or less efficient at a specific pace isn’t something I’d support.

        I have to say the youtube commentary about the Glider vs Gazelles makes a number of statements that just don’t make sense, such as elite marathoners having to use a longer time in air to provide the stride length to reach high speeds. In reality elite’s typically have a higher cadence than typical runners, and with higher cadence the time in air is proportionally lower. Statements that are exactly the opposite of what is really happening rather undermines the whole commentary.

    • canute1 Says:

      Thanks for your additional comments.

      At a given pace, the braking cost per step is estimated from the integrated value of horizontal ground reaction force (hGRF), which in turn is largely determined by the obliquity of the leg and the horizontal momentum of the body. For a given pace, the obliquity of the leg is determined by the time on the ground, which is virtually identical for both styles. The computation using the simple model confirms that braking cost per step is similar for both styles, but the number of steps per mile is greater for the glider, giving a greater braking cost per mile.

      The issue of any difference in elastic recovery is not a essential part of my argument. For the purpose of the calculated energy costs, I assumed that elastic recovery was the same for both styles. I suspect that the Gazelle will gain more from elastic recoil based on my estimate of the degree of eccentric contraction in occurring in the relevant muscles that contribute most to horizontal v. vertical push, but this is very speculative. Hence I did not make elastic costs an essential part of my argument.

      I acknowledge that the model on which I based the calculations is a simplified model, and therefore the calculations are not precise. Nonetheless, the elevation costs determined largely by the height that the centre of mass rises, while braking cost is determined largely by horizontal momentum of the body and obliquity of the leg integrated over the ground contact time from footfall to mid-stance. I think that the simple model takes these into account well and therefore, the simple model provides a fairly good estate of the relative contribution of elevation and braking. However as described above, elastic recovery is less easily quantified. It might be that runner who is prepared to sustain a very high initial vGRF by landing with a stiff leg, as is recommended on BK style, might manage to increase efficiency of elastic recovery. I suspect that few ultra runners do this, and therefore I consider that the simple model captures the essential features that matter for ultra runners.

      With regard to the views of the commentator on the video clip, we are in complete agreement.

      • Robert Osfield Says:

        Canute, I’m afraid you must have an error in your maths, as the following statement suggests that there is an problem:

        “The computation using the simple model confirms that braking cost per step is similar for both styles”

        I know this is wrong because while the angles and distances in stance are the same regardless of time in air when the stance time is the time, the actual forces applied both are higher in the case where the time in the air is higher – which means the vertical and horizontal forces are also higher. Please sit down and do an analytical model of the forces.

        When you do it you’ll find the cost of breaking per step is less for the low time in air relative the high time in air, and it’s in proportion to the difference in cadence. The two cancel each other out and you end up with the same total cost for a given period of time distance.

        Please don’t, yes but my computer model tells me yet that it’s the same. Clearly the maths is wrong somewhere, you need to go back to and sort the problems out. With the problems sorted out only they can you start to make more representative conclusions. However, I’d still caution on making any specific conclusions though as theoretical models are just that.

    • canute1 Says:

      Thanks for your further thoughts.

      I do not agree with your logic. The angles are not the same. If the height of the centre of mass (COM) at mid-stance is the same for both running styles, the COM is higher at foot-fall and at lift-off for the Gazelle, and the ratio hGRF to vGRF will be lower.

      There is no reason to suspect an error in my maths. However perhaps the assumption that height of the COM at mid-stance is the same in both styles should be examined empirically. It gets back to the question of tension in the legs. But if Gliders do maintain lesser tension in their legs I would suspect that they will be less efficient at recovering elastic energy.

      • Robert Osfield Says:

        Hi Canute,

        You haven’t done the re-appraisal of the maths yet have you… trust me, there is something wrong. Once you look back at it and fix the maths you’ll have a eureka moment.

        The angles are very similar for a short time on stance, the ratio of hGRF and vGRF will be very similar. The biggest differences in the magnitude of the forces, which is proportional to (1+ta/ts) there ta is time in air, and ts is time on stance. The cadence is equal to 1/(ta+ts), and if you multiply these terms together you get ((ta+ts)/ts)/(ta+ts) which is 1/ts, the ta term councils out completely. Once you add in the other terms you’ll get breaking force proportional to ts * v^2, and time in the air does not figure.

        What you you claiming is that somehow magically the very slight differences in angles exactly equals the increase in steps due to cadence i.e. 1/(ta+ts) which is nonsense, the trig maths are completely different and will be of much smaller magnitude.

        As for runners with a lower time in air having lower muscle forces/tension this will be true, but this doesn’t necessarily mean they will have lower muscle stiffness, and it’s the stiffness and range of motion that will determine the efficiency of elastic recoil. A runner with higher time in air and high forces will have a greater range of motion of their joints and hence muscles – the collapse more on stance, only once you take this into account will it be possible to determine which gait has the great muscle stiffness.

        I get the sense that you are stopping short of the necessary level of analysis that is required to make useful predictions and hence conclusions. You are making some mathematical mistakes and they stopping short when you get the answer you want, rather than doing the proper analysis and double checking your results and then making conclusion from what the results tell you.

    • canute1 Says:


      vGRF is proportional to (1+ta/ts). That is not true for hGRF, which is the force that determines braking costs

      Contrary to your concern about possible bias, it would of course suit me personally if Gliding was more efficient, but so far the evidence from observation of comtemporary elite distance runners and from carefully considered computation suggest that the Gazelle style is more efficient. Note that even if Salazar and Clayton were Gliders, their times have been eclipsed by the present generation of marathon runners.

      I accept that the implications of muscle stiffness are tricky to estimate. That is why in my main argument I have assumed equal efficiency of savings from elastic recoil. However I do accept that it might be that some runners achieve a greater efficiency of recoil. I consider that is unlikely in the case of ultra Gliders.

      • Robert Osfield Says:

        Hi Canute,

        > vGRF is proportional to (1+ta/ts). That is not true for hGRF,
        > which is the force that determines braking costs

        vGRF and hGRF and simply the horizontal components of the GRF that is generate by the leg, the GRF increases till it peaks near mid-stance at which point hGRF is close to zero and thus vGRF approximately equals vGRF.

        The GRF increases from zero to the max vGRF and is proportional to the max vGRF, and since hGRF is simply the horizontal component of GRF the hGRF is directly proportional to max vGRF, but of course varies due to shape of the GRF and angles involved. The shape of the GRF and angles involved will be similar for low and high airborne times with the same time on stance, but the max GRF will be big variable that is different.

        Again, I ask you to go back to look at the maths. Once you get a clearer understanding of the maths more appropriate conclusions can be drawn. I wouldn’t be pressing this matter if I didn’t believe they you’d have something important to gain from this additional effort.

      • canute1 Says:

        The component depends on the angle and as previously pointed out, the angle of the line joining COM to point of support at footfall and at lift-off is not the same for the two running styles. I believe that this fact is the source of our disagreement. Nonetheless, I have re-examined the maths. I am confident that the mathematical formulae are correct. I cannot rule out the possibility that the computer program that performs the numerical integration of the equations of motion might be faulty, but I did check it extensively in January and February of 2012, and see no reason to distrust it now.

      • Robert Osfield Says:

        While the angles are not exactly the same they are similar, the difference is cannot magically account for the horizontal breaking forces being the same despite the large disparity in vertical forces. You keep making the same mistake, you need to take a step back and then take a better look at model. Your understanding of running mechanics will benefit from this inquiry.

        I will get round to putting together a computer model and open source so it can be used and tested by others. However, I won’t be able to get to it right away as I have a number of other topics to complete first on the blogging front let alone family and work responsibilities. This is shame as it’s not really fair pointing out problems without providing a solution.

      • canute1 Says:

        Robert, I look forward to the results of your computations. Thank you for your challenges as these do prompt me to reconsider the situation carefully, though in this situation, so far I see no reason to change my conclusions.

  2. Ewen Says:

    Canute, this is a great subject. I’m in the same boat as you in that I’d like to increase the length of my stride (which I agree has shortened over the years). I’m sure my cadence is pretty much the same as it was in 1992 (when I was running my fastest), yet my stride is much shorter.

    I think you’re onto something with the thought of somehow improving elastic recoil either through training or a slight technique change. The brain probably plays a big part in changing our technique over time more towards that of a ‘glider’ — gliding feels like it produces less impact trauma (in spite of a faster cadence at the same pace) so we probably drift towards running that way as a means of self preservation.

    It’s notable from the video that many of the runners, both gazelles and gliders, are running with a heel-first ground contact and most appear to be wearing shoes with a fairly substantial mid-sole. For the gazelles, this would give the feeling of not impacting the ground so hard (the brain would have the limbs/tendons absorb more of the impact if the runner were wearing less substantial footware). So I think for we older runners, the answer *could* be to run with a heel touch and wear cushioned shoes (or ‘over-cushioned’ shoes such as the Hokas or Adidas Boost). In that way we could run with greater abandon (as we did with the fresh legs of youth), impact the ground harder and generage more elastic recoil.

    I think it’s worth noting that there have been successful fast gliders – Alberto Salazar and Derek Clayton are two that come to mind – both capable of 5 minute miling over the marathon distance.

    • Robert Osfield Says:

      Hi Ewen,

      I believe Canute’s maths is wrong somewhere and hence the conclusion about longer stride length/longer time in the air being more efficient is not sound. Your examples of runners that used higher cadence and short time in the air suggest that it’s perfectly possible to be efficient without needing to bound along.

      W.r.t footware the Hokas are so soft they are likely to increase ground contact time and make you less efficient. The Boost by contrast looks like to have an efficient enough midsole to provide a good enough energy return to not negatively impact ground contact time. The key to efficient is ground contact time so it’s the parameter you want to minimize and seek running gaits and footware that minimize it.

      The gait changes that can help reduce ground contact time will be to land on the mid-foot or forefoot, as from a theoretical standpoint elastic recoil is enhanced and human studies have shown mid-foot & forefoot runners on average have short time on stance.

      Lightweight footware helps significantly with efficiency as it helps reduce the cost of repositioning the foot on each step, with lower costs of repositioning you use less energy and also can use a higher cadence which in turn results in lower time in the air and potentially lower time on stance. Lower time in the air is good for reducing loads, lower time on stance is good for efficiency.

      Unfortunately the Adidas Boost while having an efficient mid-sole is heavy and looses all of it’s potential efficiency gains by being several onces heavier than a normal racing flat. The high heel will also make an efficient mid-foot/forefoot gait less likely.

      So… an ideal shoe would be a very light, zero drop shoe made with the Boost mid-sole. Ideal gait will by mid-foot, with high cadence and don’t worry about being a “glider”, just work on short time on stance.

    • canute1 Says:

      First, with regard to Robert’s claim that my maths are in error, I do not think there is any need for concern. As pointed out in my response to Robert, there is an error in the logic that led to his assertion.

      I do accept that if runners adopt differing degrees of tension in their legs it might be possible to make greater savings via storage of elastic energy. I do not think that ultra or ironman Gliders usually maintain a high level of tension, but this remains a matter for speculation. I remain confident that the Gazelle style is more efficient at moderate and fast paces, except perhaps if a Glider maintained very high tension in the leg muscles.

      I was not aware Salazar was a Glider. His two current most famous protégés, Farah and Rupp are Gazelles. Note that at sub 28 min 10,000 pace, even Gazelles often have a cadence approaching 200, but they nonetheless also have a long stride and the elevation typical of a Gazelle. However I agree that Clayton was famous for his ‘shuffling’ stride. Although Clayton was a contemporary of mine, the nearest I got to see him was when I competed in the Australian National Marathon championship in Melbourne in 1970. But to my disappointment, Clayton did not run. He won that event in both 1969 and 1971. I guess that even if he had run in 1970, I would have only had a back view, but it would have nonetheless been interesting to see.

      While I agree that our brains probably automatically encourage us to minimise impact forces, I think that there is quite strong evidence that the optimum thickness of EVA padding is only about 5-6mm. My own experience from my younger days is that the thin-soled Onitsuka Tigers were noticeably better than any other road running shoe I ever wore. I think that now I am older, the preferred approach is to recover the ability to withstand the eccentric forces that I used to be able to withstand in those days. If that proves impossible, I would rather explore other ways of adjusting muscle tension before adding extra weight in my shoes.

      • Robert Osfield Says:

        Hi Canute,

        One observation that I saw online somewhere picked out the some of these distance elites look they they are spending lots of time off the ground because they feet are well off the ground whilst they are the air, but because of their high cadence they actually don’t spend long in the air.

        What is happening is their knees are highly flexed on pull through so relative to their hips and the ground their feet are relatively high off the ground. However, with the high cadence their hips aren’t vertically oscillating a great deal – they aren’t bounding along like a gazelle.

        The faster one runs the quicker the pull through has to be and the higher the feet will be, so the quicker you go the more this effect is observable. However the quicker you go the higher your cadence so time in the air actually goes down. Cadence doesn’t increase as rapidly as speed so distance traveled per stride goes up but it’s not down to them bounding along.

        Given this observation I don’t believe the “gazelle” and “glider” terminology is useful. A gazelle has a large vertical oscillation and glider has lower vertical oscillation, but the key controller of vertical oscillation is cadence. However, at high speed even a glider will outwardly look like gazelle because the knee will fold up on pull through and make it look like they have a high vertical oscillation when they don’t.

        What I would say is that someone will look more like a “gazelle” if their cadence is low relative to their speed, and look more like a “glider” when their cadence is high relative to their speed. What they look like really isn’t what makes someone efficient or not though, the crucial factor here is time on stance.

      • canute1 Says:

        You have created two parallel discussions of your opinions on this issue. My primary responses to your comments are given above. However, your statement in this paragraph that the key controller of vertical oscillation is cadence is misleading, and your statement that as you go quicker that air time actually goes down is contrary to the evidence. Fig 2 of Weyand ( J Appl Physiol 108:950-961, 2010) shows that airborne time increase with pace up to 6m/s and then flattens. The primary determinant of time in the air is mean vGRF and duration on stance, since the product of these is the vertical impulse. Speed increases as vGRF increases, and, up to around 6 m/sec, airborne time also increases. At speeds faster than marathon pace, the combination of increasing vGRF and decreasing time on stance results in an almost constant airborne time.

        I strongly agree with the view that many recreational runners will benefit by increasing cadence. However, the point of my post was to point out that one must consider all three of the energy costs, and simplistic rules about optimising one variable are inadequate. In particular, limb repositioning costs become very high when cadence and pace are high. Whether not one finds the terms gazelle and glider helpful is perhaps personal taste. Recognising the need to optimise vGRF, time on stance and cadence is what matters, and is determined largely by Newtonian mechanics, though the viscoelastic properties of tendons and muscle do add some intriguing complexity to the situation.

    • canute1 Says:

      I just had a look at your training diary and notice that you have had a distressing time with your arm. I hope it settles quickly and that the cause turns out to have been some unaccustomed trauma – were you doing pull-ups or other upper body exercises with the Speedy-Geese on Monday?
      Best wishes, C.

  3. Ewen Says:

    Hi Robert,
    It’s possible to be efficient without bounding (e.g. Clayton/Salazar), but as Canute postulates, there’s a speed threshold where a glider is less efficient than a gazelle. The majority of fast elite track runners up to 10,000m are gazelles (both males and females). I don’t think the slight increase in ground contact time would make a difference to we weekend warrior runners if we wear cushioned shoes.

    Canute, yes I recall watching the ’84 Olympics and noting Salazar’s shuffling gait (even recognisable from the air). Clayton was a glider (a shame you didn’t get to race him!), but a 2:08 marathoner. I think there could be 2:05 marathoners with a glider/gazelle cross type of gait, but most would be gazelles.

    Regarding more cushioned (thicker midsoles), I think they could be of great value to older runners if they run in them correctly (more gazelle-like) and not just allow the cushioning to absorb energy. The Hokas, while looking heavy, are surprisingly light – this one claims a weight of 295 grams. My Free 5.0s (a light shoe for me!) are 298 grams in size 13. My lightest shoe is the Vitarra at 230 grams and my old Asics 1120s are 390 grams. A friend’s husband used the Hokas in a recent half marathon and ran a lifetime PB of 1:38 (in the 60-64 age-group).

    Not sure what the cause of my arm swelling was. They’re doing further blood tests and a catscan in a few weeks. DVT in the arms is rare, but it could be a one-off thing for me. I can’t run for 2 weeks and will be on the drugs for 4 months.

  4. canute1 Says:

    Thanks. It is interesting to hear about your friend’s husband’s HM PB in Hoka one ones. I had not appreciated that they were so light, nor that the heel to toe drop was only 4mm, though even this much drop is greater than I would prefer. It is also puzzling that the text describes a 4mm drop, while the summary information in the side bar states 6mm. It appears there are two different inserts.
    I was impressed by the apparent impartiality of this review:
    His conclusion that they are great for downhill running make sense, but so too does his concern that they are a bit sluggish on the road. This review implies that they might be best for off-road ultra’s. Price is clearly an issue and I would also be concerned about durability, despite the lugs. I like the potentially wide toe-box, though there seem to be differing opinions about this. Is this due to the differences between tarmac and trail models or inserts? Your point about the aging athlete is a good one. Overall, following your prompting to look into the details, I am intrigued but not yet convinced.

    There is a little (not yet conclusive) evidence that marathon runners who fly long distances to races have increased risk of DVT due to hypercoagulability. The few cases I know of were leg DVT, possibly because local trauma also plays a role, but also simply because leg DVT is much more common. Dehydration and perhaps low resting heart rate might also contribute. I hope your problem is a one-off event, but if I were you I would ensure that I remained well hydrated and moderately mobile – in both upper and lower limbs – if you fly to GC for the 10K or HM this winter.

  5. canute1 Says:


    Here is an interesting statement by Alberto Salazar when interviewed by Amby Burfoot in 2010.

    “There has to be one best way of running. It’s got to be like a law of physics. And if you deviate too much from that–the way I did in my career–it can be a big handicap. …. You can be efficient for a while with bad form–maybe with a low shuffle stride – but eventually that’s not good for your body. It’s going to produce tightness and muscular imbalances and structural problems. Then you get injuries, and if you’re not careful – if you don’t take care of the muscular and structural issues – the injuries can put you into a downward spiral.

    This quote was repeated by Runblogger at:

    • Ewen Says:

      Thanks Canute. I have seen that Salazar quote — I think his own awkward form (and eventual injuries) is why he’s so concerned about tuning the runners he coaches forms.

      There’s another review on the Hoka ‘Bondi’ shoe which mentions good and bad points. Apparently the various models are quite variable in feel (and weight). The ‘Bondi’ was the one worn for that HM PB. The sizing is also a bit different to normal shoes so I wouldn’t buy a pair without trying them on. I’m interested that this type of shoe might allow the older wearer to run with more abandon (not worrying about impact, or perhaps running with the most impact-causing style – that of the gazelle), just as the ultra trail runners report running recklessy downhills.

      Yes, I was wondering if low resting HR (esp when sleeping) and the fact that I sleep on my right side may have caused it, plus maybe some dehydration from our unseasonably warm weather. A friend just noted that a friend of his got DVT when flying back from the Paris Marathon. I haven’t entered GC yet, so will wait until the last minute.

      • Robert Osfield Says:

        Hi Ewen,

        There have been studies that have found that increased cushioning leads to increased force generation, however, the findings vary from individual to individual, so while the average might be for increased force generation you might not respond in this way.

        Another recent study showed that a small amount (10mm) of cushioning produces better running economy than no cushioning and less than more cushioning, so as Canute points out there will be an optimum level. I expect this again will be very individualistic, so how much is optimum will again vary from person to person, and likely also the pace and gait they use.

        Studies have shown also linked weight of running shoe to running economy, a general rule of 100g equally 1% change in running economy is a reasonable guide to use.

        Both the weight and excessive amount of cushioning provided by the Hoka’s will impact on running economy. I also believe it’s likely that it’ll they have so much cushioning that they’ll increase ground contact time.

        If you want to improve running economy you are best finding lightweight shoe (less than 200g) with a level of cushioning that feels responsive. You are also best to work on your gait, potentially changing footwear to something more minimal to help with gait re-training. Pete Larson of Runblogger wrote a good article reviewing a study about the benefit of this:

        For you own personal experiment of 1, I’d suggest reflecting on what might be limiting your performance the most – and focus on these areas. Canute’s example of establishing strength as lacking and then focusing of strength training is a good example of this. Good luck.

  6. Ewen Says:

    Robert, thanks. I think the thing that’s most limiting my own performance at the moment is aerobic base, however I’m also concerned about my decreasing stride-length. However, I know I can’t have a longer stride without being better aerobically. We would have all witnessed the runner in a race with a beautiful long stride, however they’re ‘miles’ behind an efficient glider. They have a long stride, but lower cadence to bring them to the limit of their aerobic ability.

    Agree re the need to tune cushioning to the individual. As I said earlier, the Hoka Bondi is a fairly light shoe (would have been categorised as a light-weight trainer) in the old days. It’s around 300 grams, so only 40 grams heavier than my Kinvaras which I use for racing. It only looks heavy! My Asics 1120s are 390 grams.

    • canute1 Says:

      Aerobic base is a major factor in performance at distances from 3K to marathon and beyond, and it is worth maximising it. However, sprint speed is also a strong predictor of performance from 3K to marathon, but is only minimally related to aerobic base.

      Last year, after a period of several months base-building at low intensity and moderate volume (around 50Km per week) I reached a point where I required around 630 heart beats/Km during aerobic running, which for me indicates good aerobic capacity, yet I could not run 4K in less than 20 minutes. After a brief period of strength conditioning , Magill drills and stride-outs (together with a good taper) I ran a HM at a substantially faster pace than I could maintain for 4Km only a matter of weeks before. I also ran a 5K at a much faster pace a few weeks later.

      For the recreational distance runner there is no doubt that the first target is developing aerobic capacity but my own experience demonstrates that for me, aerobic capacity should not be the only target.

      Although I remain a little sceptical about Hokas for road racing, I can see potential advantages for high volume training on hard surfaces. However if you do get some Hokas I would recommend that you also do fairly regular barefoot sessions (maybe on that lovely grass at Stromlo) to ensure that your foot muscles remain well conditioned

      • Ewen Says:

        I recall that sequence of events leading up to your good half last year. I agree on the need to keep developing speed ability on a year-round basis (especially for older athletes). I particularly like the idea of integrating sessions of Magill drills and short sprints whilst doing aerobic training – that’s what I intend to do once I’m over this health issue.

        Interesting about your 630 h/beats per k producing 5 minute ks, then later well under 5 minute ks for a half. Grellan noted a recent pacing marathon where he ran 5 minute ks at 620 h/beats and 125 ave HR. In the Melbourne Half last year when I was just under 5 minute ks my h/beats per km was 708. I think I have room for improvement there.

        Unfortunately the Hokas aren’t sold in Canberra and I wouldn’t buy a pair without trying them on first. I’ll have to visit Wagga or Mittagong. I’d continue to rotate shoes, using the Free 3.0s at Stromlo, also the Kinvaras and Virratas. I think the Hokas would be especially good for long runs going by the improved recovery people have noted.

  7. EternalFury Says:

    It seems this discussion happens around 2 postulates:
    1) Physics dictates there is an optimal running form.
    2) Maximal running economy will be achieved by complying with innate neurological patterns.

    While 1) may be demonstrably true, the real question is: Is the plasticity of the motor cortex of an adult human being sufficient to successfully adopt that running form?

    If it is not, then trying to comply with that running form will be an exercise in futility, which will lead to reduced running economy.

    Now, it may be that champions have 1) and 2) line up innately.
    It may further be true that the best among them also have a very high VO2max and a highly trainable LT.
    These last statements ring true with the way natural evolution usually happens. The fittest is not fittest due to superior will ; the fittest is fittest due to their innate ability to fit. (a.k.a., “if you want to be a champion, pick your parents carefully”)

    The question remains: How does one figure out how to make the most of one’s ability?

    Short of measuring running economy in a controlled environment (a lab) while using various running forms, I don’t see how the question could be answered practically.
    If the plasticity of the motor cortex of an adult human being does permit adaptation, how long would that period of adaptation take? (lab assessments would need to be taken at regular intervals during the period of training, which could take months or even years)

    It could be that running like a gazelle is measured to be inefficient at first, but more efficient once adaptation has fully happened.
    Or, it could be that the plasticity of the motor cortex is extremely limited at birth or after a few years and no amount of training will ever allow the subject to run as a gazelle more efficiently than they do some other way.

    Sorry to raise more questions than answers, but I think we don’t need to understand how the magic happens, we just need to measure and compare its output.

    • canute1 Says:

      Thanks EF. Interesting thoughts.
      I agree that the fittest are fittest because of their innate ability to fit. I think Darwin was mainly referring to ability to produce surviving offspring, and maybe at some point in human history, running was important for the survival of offspring. But I think that humans developed an even more useful feature than running ability: the drive to get better at things by understanding how they work.

      The human brain (like the rest of the body) is fairly plastic, even perhaps in old age. But there are several different ways that brain can change. In infants the process involves conscious focus on the goal while leaving non-conscious processes to work out the mechanism. As we get old, simply wanting to achieve goals does not work so well, so maybe it is time to focus more on understanding the mechanism.

Leave a Reply

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

You are commenting using your 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: