With the horizontal force crowd chanting Matt Brughelli as the leader of the speed revolution I had to respond to the data on the charts. Why are the strides 3.5 meters? Is Dwight Howard secretly training for the 2012 games in London? Why are strides contact lengths 200 ms long? Does the treadmill have magnets because the study was on the moon? You can get data on force plates if you do enough sam
The Brughelli Study: Super Strides?
-
-
Participant -
Participant
Carl, to be frank this study (https://www.fittech.com.au/downloads/treadmilldocs/Cronin_et_Effects_of_Running_Velocity_on_Running_Kinetics_8.pdf ) did not impress me. I am not greatly enthused by “horizontal vs vertical” at the best of times and the methodology and data presentation of this study was very poor in my eyes.
I will say I quite like the methodology of comparing of different % max speed for an individual. Needless to say, Brughelli et al decided not to include “Running Velocity” values in Table 1 titled “Running velocity, running kinetic, and kinematics.*”, and do not mention the actual running velocities of their subjects in the paper at all. Doing so would “potentially” have indicated the “limitations” of the study, particularly given the ad nauseam references to investigations on running velocities greater than 6 or 7ms-1 in the discussion section. The nature of their subjects’ terribly slow velocity is down to the torque treadmill (less true for curved toque treadmills like the Woodway 3.0 curve) where velocities are lower than overground.
We should note “Stride Length” refers to two steps, and length = distance the belt moved.
Stride Frequency at 100% is 1.67Ss-1 –> Step Frequency = 3.34StepSec-1
Stride Length at 100% is 3.27ms-1 –> Step Length = 1.635ms-1
This means speed at 100% velocity is 3.27*1.67 = 5.46ms-1
Additionally, Contact Time = 209.67ms-1
3.34 StepSec-1*209.67 = 700.2978ms total contact time
Leaving (1000-700.2978)/3.34 = 89.73ms average flight time
So Contact time/Flight time = 209.67/89.73 = 2.34?!
Flight times were not included in “Running velocity, running kinetic, and kinematics.*”
Brughelli has said https://bretcontreras.com/2010/08/sprint-research-biomechanics-and-practical-implications-an-interview-with-matt-brughelli/
“I don’t think friction is a problem with the Woodway treadmill. It’s possible.”
This is very possible to me.
“Sixteen semiprofessional Australian Rules football players participated in this research (age 23.3 6 2.1 years; height 184.8 6 12.4 cm; and weight 84.1 67.4 kg). The players were recruited from the West Australia Football League. All subjects provided written, informed consent within the guidelines of Edith Cowan University. The subjects had at least 2 years experience with resistance training, endurance training, and performed maximum effort sprints on a regular basis.
This is the first study, to our knowledge, that has investigated the effects of running velocity (up to maximum) on Fv and Fh in well-trained athletes who routinely perform a mixture of training methods, including maximum effort sprints, power training, strength training, and endurance training.”
5.46ms-1 Max V i.e. reflective of 19-20sec+ 100m performance or not far off velocity for a 5-6min run in this population? If it wasn’t friction at play was gravity 20ms-2 in the lab?
The big issue is the use of a torque treadmill. The study was not on the moon but nor were the researchers observing gait on a normal surface as the Woodway 3.0 is a non-motorized torque treadmill, which is very significant as “running speed” means treadmill belt speed not COM horizontal velocity, and velocity of the belt will require continued force application. Contact lengths (distance and time) may be longer and obviously translation and oscillation of the centre of mass may be somewhat different.
In particular oscillation of the COM in the vertical direction will be altered, potentially de-emphasising the role of vertical force and oscillation of the COM (the COM is not projected ala overground sprinting in a biomechanical sense). Given the treadmill is non-motorized, athletes will obviously have to apply force right through the ROM at toe off to keep the belt moving, and again this is key because the performance measure is belt speed not COM horizontal velocity.
As such we get an outcome for stride length with longer time to apply force likely yielding greater stability of the COM, and treadmill design necessitating a longer force application and exaggerated toe off, potentially creating an artificially longer contact distance and/or force in late stance.
From a biomechanical perspective, the essential components of running (in particular hip vs knee vs ankle vs MTP extension velocity ala Ito et al and others) are likely altered by the use of Woodway 3.0 non-motorized treadmill, and as such the results should not necessarily be extrapolated to sprinting on level ground as in track and field and team sports.
Brughelli has stated this:
I am in complete agreement with Karl Zelik (Buckley et al. 2010) that correlations and linear regressions should not be used as a surrogate for fundamental mechanical understanding of speed limitations. Instead of making such strong conclusions, I think Weyand and colleagues should have embraced the shortcomings and limitations of their study in order to motivate further research.
I too am in complete agreement with Mr Zelik 😉
Brughelli et al also state in their paper:
However, it has been demonstrated repeatedly by Cavagna and colleagues (3-7) that an increase in RFv leads to an asymmetrical rebound (i.e., the amount of time force exceeds body weight < amount of time force is less than body weight) during high-velocity running. As running speed increases, the asymmetry becomes greater, thus leading to a decrease in push-average power (i.e., the work done divided by the duration of positive work production), and a decrease in maximum running velocity.
This misses the mark as an "asymmetrical rebound" relies on the simple spring mass-model which as I have discussed previously is not applicable to sprinting. What will actually happen in overground sprinting as per Ito et al, is that the knee extension velocity will be lower compared to hip extension (eventually becoming negative) during the ground phase and the "asymmetrical rebound" is a fantasy, as kinematics influence kinetics and "amount of time force exceeds body weight < amount of time force is less than body weight" needs to be considered in terms of relative velocity of hip vs knee vs ankle vs mtp etc.
I'm not a great fan of the horizontal vs vertical debate. At the end of the day I would be mightily surprised if the basis of maximum velocity sprinting (and fatigue/speed endurance) is not borne out to be very much a combination of hip extension velocity and vertical stiffness during the ground phase (largely influenced by knee stiffness). There will be both power output components and postural/skill components related to achieving that model, and other factors may have small additional effects, but I do think a model based on a combination of hip extension velocity and vertical stiffness is likely the basis of maximum velocity sprinting. I look forward to robust, well-designed research which stands on the shoulders of real sprint research on the track.
-
Participant
.
-
Participant
TSCM,
I still don’t know what the top stride length distance is after the calculations.
CV
-
Participant
That interview is just incredible, and I barely stopped laughing the whole way through the first time I saw it. To be fair, I did think the second sentence of this quote showed sound judgement:
“I have a few ideas for more research on the limitations on maximum running velocity. However, after that I will probably never visit the topic again.”
I’ve personally come to the conclusion that horizontal vector training may actually be a reference to the improvement curve over time, which will be a flat horizontal line. Specific Adaptation to Imposed Demands 😉 ?
Perhaps Matt and Bret can have athletes hip thrust their way to glory at the Woodway MaxV sprint world championship?
I do think that traditional strength and conditioning (S&C) can improve speed and acceleration. However, I do NOT think the improvements are due to greater vertical GRF’s during running. I think Weyand et al. 2010 has made a very strong case for this point. In their new study they have reported that runners apply sub-maximal forces (i.e. vertical GRF’s and extensor muscle forces) during maximum velocity running. So I doubt if you improve an athletes squat, he/she will produce greater vertical or extensor muscle forces during running. I would like to see someone do a correlation between squat strength and vertical GRF’s during maximum velocity running. I doubt there would be any relationship.
Who needs more vertical forces and stiffness during acceleration? Just emphasise that “horizontal vector” and hip thrust your way straight into the ground!
Who would have thought that Weyand et al might find force production in 80-100ms of contact time during sprinting is less than in in a different motor pattern like hopping with longer contacts?
I hear some crazy cat once suggest a concept called power which equals force/time, can anyone confirm this?
I hear some other crazy cat suggested that forces do not magically appear but are in fact produced by muscles, and the way muscles act can change if the movement is inherently different 😮
It is of course well known that horizontal forces in sprinting are at such incredibly high levels as Matt demonstrated with a mean of 360.9N at 100% Speed in his own study!
Does Matt mean to suggest that there are possible confounders with correlation and transferability of maximal squat strength to maximal sprinting speed? Outrageous 😮
How could others be so short sighted with their vertical suggestions? Clearly the solution is to be face down on the track from lack of vertical forces so short-sightedness is no longer a problem!
There also seems to be a lot of talk about hip hyperextension, which is so great because there is so much evidence that better sprinters show greater hip extension amplitude!
As flight time and belt speed are not worth presenting in the paper itself, I’m happy to attempt to deduce (the flight time and contact time data really do not make sense for upright running and I would love to see the posture that was used on video for each speed 40-100%)
40%
Contact time = 301.78. 150.89*1.60 Step Rate = 482.85ms contact/sec running
Flight Time = (1000-482.848)/1.60 = 323.22ms/step (??)
CT/FT = 301.78/323.22 = 0.94
SF * SL = 0.80*1.70 = 1.36ms-1 belt velocity
60%
CT = 280.45. 280.45*2.30Step Rate = 645.04ms contact/sec running
FT = (1000-645.035)/2.30 = 154.33ms/step ??
CT/FT = 4.18
SF * SL = 1.15*2.12 = 2.44ms-1 belt velocity
80%
CT = 248.29. 248.29*2.90Step Rate = 720.04ms contact/sec running
FT = (1000-720.04)/2.90 = 96.54ms/step??
CT/FT = 7.46
SF*SL = 1.45*2.57 = 3.73ms-1 belt velocity
100%
CT = 209.67. 209.67*3.34step rate = 700.298ms contact/sec running
FT = (1000-700.298)/3.34 = 89.73ms/step??
CT/FT = 7.80
SF*SL = 3.27*1.67 = 5.46ms-1 belt velocity
Let’s note at the outset that it appears that on belt velocity
“40%” is in fact 1.36/5.46 = 25.4%
“60%” is in fact 2.44/5.46 = 45.6%
“80%” is in fact 3.73/5.46 = 69.7%
The CT and FT are completely off the mark as far I can see for data on upright running and the values really seem very strange.
The mean flight time decreases markedly from 154.33ms at 45.6% belt speed, to 96.54ms at 69.7% belt speed to 89.73ms at 100%. CT/FT goes from 4.18 to 7.46 to 7.80.
Concurrently, mean vertical force stabilises with values of 1922.7, 1942.3, 1983.7N. This is because we are basically describing a sled push more so than running.
Flight time decreases greatly while vertical force is barely increased at much higher belt speeds. Therefore, as per the laws of physics, it is apparent the take off angle and position of the COM must be lower at “80%” and 100% than at “60%” in Brughelli’s study.
Obviously this is not what we observe in real sprinting. Horizontal force will of course be higher when we are pushing out the back with low postures on a Woodway torque treadmill at the grand rate of 5.46ms-1 treadmill belt speed. However this likely has very little relevance for improving horizontal velocity of the COM in overground sprinting.
-
Participant
Carl, the data presentation is certainly not user friendly, and this is a bit of trend with some researchers.
“Stride Length” =3.27m but this refers to two steps as a full stride cycle = 2 steps. Therefore the step length is 3.27/2 = 1.635 or 1.64m accounting for significant figures.
“Stride Frequency” = 1.67 but this again refers to two steps. Therefore the step frequency is 1.67*2 = 3.34 steps/second.
On of the many problems with the data presentation is the contact time values seem to refer to a single contact (I can’t really see it referring to two contacts which would be very unconventional, though the paper does show some odd methodology in general) while the stride parameters of the table concern a full stride cycle.
To be perfectly honest, I think if they’d presented data in a more conventional way ala 1.64m Step length, 3.34 step freq, belt velocity 5.46ms-1 at 100% Speed, the limitations of the study would have been even more blatant. That readers appear to need to go and deduce belt velocity, flight time and so on for themselves is not great. That there is no mention of the belt velocity values anywhere in the paper but constant reference to belt velocity and how well trained the athletes are is ridiculous.
Calculating from their data, for one step at 100% the mean was 1.64m belt displacement from touchdown on one leg to touchdown on the other leg, and seemingly ground time 209.67 ms, flight time 89.73ms (the flight time and contact time data really do not make sense and I would love to see the posture that was used on video for each speed 40-100%).
-
Participant
tscm,
“There also seems to be a lot of talk about hip hyperextension, which is so great because there is so much evidence that better sprinters show greater hip extension amplitude!”
Why do you think this expression is increasing in popularity?
-
Participant
Rich,
It’s symptomatic of the internet age where consuming intuitively appealing but ultimately baseless rubbish, like the hip hyperextension for horizontal velocity stuff, is extremely easy for those who find it all too demanding to attain a good general understanding of the sport sciences along with a specific understanding of their sport/s, as well as think for themselves. For those that do make the effort to read research, it can then be difficult to critically analyse studies like the above and others I’ve mentioned which show confused or flawed methodology. This struggle in interpretation can apply to anyone from the teenage novice without any interest in the sport sciences to highly experienced researchers who consistently show heavy cognitive bias and/or put out poor quality work.
There is a big focus in some sprint research on simple outcomes like force output and contact times/distances and so on, but at the end of the day biomechanics is the mechanics of the body and productive biomechanics research looks at the big picture of how outcomes are achieved, attempts to identifying essential components of performance, and delves much deeper than peak force values as the human body is not a simple system.
In strength and conditioning particularly, there’s a very strong culture of the “real world”, “hardcore” or “meat head” types who are quite incapable and/or resistant when it comes to learning and critical thinking, and this is only growing with sites like T-Nation, Elitefts and related blogs. Realistically, the strength and conditioning field is to a large extent the masculine equivalent of being a “personal trainer to the stars”, and the simplicity and tone of the articles on such websites make them appealing to novice coaches and athletes not only on an intellectual level but also very much on an emotional level. The same could be said for many conferences purportedly about education, which in many cases really involve feeling social validation with like-minded people as much as becoming a better coach. This is quite a positive and healthy thing, but the amount of learning (as opposed to a feeling of consumption or fulfillment) that actually occurs through so called “continuing education” with certain conferences and DVDs/ebooks etc from various coaches is pretty questionable and it is no substitute for real learning.
Reading research critically or learning biomechanics/biochemistry/biostatistics etc is very important but very dry and mentally demanding in comparison to being wowed at a Perform Better conference or by the latest T-Nation revelation or the “death of the squat” or “horizontal vector training”. Often many coaches reading research papers do not delineate properly between opinion based comment in the “introduction” and “discussion” sections (which could perhaps be more accurately titled “agenda” and “speculation” in many cases) and the meat of the study in the methods/results. It can be difficult to effectively analyse the quality and meaning of the methods/results without adequate background knowledge of general sport sciences and the specific topic (including related papers), and many do not even think to question and simply take the “expert opinion” of the authors in the intro/discussion/conclusion as the objective reality.
-
Participant
Simply awesome post tscm.
Rich,
It’s symptomatic of the internet age where consuming intuitively appealing but ultimately baseless rubbish, like the hip hyperextension for horizontal velocity stuff, is extremely easy for those who find it all too demanding to attain a good general understanding of the sport sciences along with a specific understanding of their sport/s, as well as think for themselves. For those that do make the effort to read research, it can then be difficult to critically analyse studies like the above and others I’ve mentioned which show confused or flawed methodology. This struggle in interpretation can apply to anyone from the teenage novice without any interest in the sport sciences to highly experienced researchers who consistently show heavy cognitive bias and/or put out poor quality work.
There is a big focus in some sprint research on simple outcomes like force output and contact times/distances and so on, but at the end of the day biomechanics is the mechanics of the body and productive biomechanics research looks at the big picture of how outcomes are achieved, attempts to identifying essential components of performance, and delves much deeper than peak force values as the human body is not a simple system.
In strength and conditioning particularly, there’s a very strong culture of the “real world”, “hardcore” or “meat head” types who are quite incapable and/or resistant when it comes to learning and critical thinking, and this is only growing with sites like T-Nation, Elitefts and related blogs. Realistically, the strength and conditioning field is to a large extent the masculine equivalent of being a “personal trainer to the stars”, and the simplicity and tone of the articles on such websites make them appealing to novice coaches and athletes not only on an intellectual level but also very much on an emotional level. The same could be said for many conferences purportedly about education, which in many cases really involve feeling social validation with like-minded people as much as becoming a better coach. This is quite a positive and healthy thing, but the amount of learning (as opposed to a feeling of consumption or fulfillment) that actually occurs through so called “continuing education” with certain conferences and DVDs/ebooks etc from various coaches is pretty questionable and it is no substitute for real learning.
Reading research critically or learning biomechanics/biochemistry/biostatistics etc is very important but very dry and mentally demanding in comparison to being wowed at a Perform Better conference or by the latest T-Nation revelation or the “death of the squat” or “horizontal vector training”. Often many coaches reading research papers do not delineate properly between opinion based comment in the “introduction” and “discussion” sections (which could perhaps be more accurately titled “agenda” and “speculation” in many cases) and the meat of the study in the methods/results. It can be difficult to effectively analyse the quality and meaning of the methods/results without adequate background knowledge of general sport sciences and the specific topic (including related papers), and many do not even think to question and simply take the “expert opinion” of the authors in the intro/discussion/conclusion as the objective reality.
-
Participant
Carl, the data presentation is certainly not user friendly, and this is a bit of trend with some researchers.
“Stride Length” =3.27m but this refers to two steps as a full stride cycle = 2 steps. Therefore the step length is 3.27/2 = 1.635 or 1.64m accounting for significant figures.
“Stride Frequency” = 1.67 but this again refers to two steps. Therefore the step frequency is 1.67*2 = 3.34 steps/second.
On of the many problems with the data presentation is the contact time values seem to refer to a single contact (I can’t really see it referring to two contacts which would be very unconventional, though the paper does show some odd methodology in general) while the stride parameters of the table concern a full stride cycle.
To be perfectly honest, I think if they’d presented data in a more conventional way ala 1.64m Step length, 3.34 step freq, belt velocity 5.46ms-1 at 100% Speed, the limitations of the study would have been even more blatant. That readers appear to need to go and deduce belt velocity, flight time and so on for themselves is not great. That there is no mention of the belt velocity values anywhere in the paper but constant reference to belt velocity and how well trained the athletes are is ridiculous.
Calculating from their data, for one step at 100% the mean was 1.64m belt displacement from touchdown on one leg to touchdown on the other leg, and seemingly ground time 209.67 ms, flight time 89.73ms (the flight time and contact time data really do not make sense and I would love to see the posture that was used on video for each speed 40-100%).
I thought the data was a misprinted as 1.6m strides is Dwarf Howard not Dwight! 3.3 hertz is hardly world class. Thanks for your monster posts TSCM.
-
Participant
Carl, it’s certainly lower speed “running” on the torque treadmill, and a maximum 5.46ms-1 means the subjects are not using normal mechanics period.
As this horizontal force promotion from Brughelli and Contreras is such a prime example of the issues with biomechanical research directions and critical thinking ability, and is taken as valid by many, I think it’s worth the effort to really break it down it terms of the evidence cited. Both Brughelli and Contreras’ intentions may be in the right place, but the information being put out is not good at all in my opinion.
The basic keys are:
1. A big picture perspective of biomechanical analysis of a movement (many sources, motor control context etc).
2. Critical analysis of research.With the addition of my recent study, there are now three studies that have directly investigated the effects of running velocity on vertical GRF’s over a range of velocities up to maximum ([b]Brughelli et al. 2010; Kuitunen et al. 2002; Nummela et al. 2007)[/b]. Each study used an athletic population and reported that vertical GRF’s (i.e. peak and average GRF’s) remained constant after reaching ~70% maximum velocity. [b]This is direct evidence against the claim that maximum running velocity is limited by vertical GRF’s.[/b]
I’d also like to point out that non-motorized treadmills have been shown to be valid in comparison with overground running for maximum running velocity. [b]They have also been shown to have similar running kinetics or kinematics to overground running, and have excellent reliability.[/b] In a side note, Weyand et al. 2000 used a “motorized” force treadmill. Motorized force treadmills have been shown to alter running kinematics compared with overground running (McKenna et al. 2007)
One more study I wanted to mention. Peter Weyand has published a new study in the Journal of Applied Physiology (Weyand et al. 2010) on the same topic. [b]In this study, forward hopping and backward running were compared with maximum running velocity.[/b]
McKenna et al (2007) “A comparison of sprinting kinematics on two types of treadmill and over-ground” Scand J Med Sci Sports 17: 649-655
McKenna et al compared torque treadmills (a different model of toque treadmill to that used by Brughelli so Brughelli’s method is not validated, and the speeds reported by Brughelli are a great deal slower than McKenna) to motorized treadmills and overground running at a given speed. Notably Brughelli is unable to support the statement “I’d also like to point out that non-motorized treadmills have been shown to be valid in comparison with overground running for maximum running velocity.” with a reference.
In comparing kinematics at a given speed across conditions (which is an invalid method given max speeds differ between conditions and therefore a given speed means variable intensity as %max):
McKenna et al found statistically significant reductions in hip extension at toe off in torque treadmill running compared to overground sprinting.
McKenna et al also found statistically significant reductions in ankle extension at toe off in torque treadmill running compared to overground sprinting.
McKenna et al did not find statistically significant differences for knee flexion velocity but the figures were disparate at 6.88 mean (5.33-7.80 CI) for torque treadmill versus 9.08 mean (6.60-11.60 CI) for overground sprinting.
Therefore Non-Motorized force treadmills (and again McKenna and Brughelli used different equipment so Brughelli’s method is not validated) have also been shown to alter running kinematics compared with overground running.
(βflx = knee flexion velocity, βext = knee extension velocity) :
Subject variation was a main effect for βflx (p =0.022, ES = 0.89, Power = 0.90) and βext (p=0.01 , ES = 0.95, Power = 0.94); however, instead of appearing as a main effect, the inherent inter-subject variance possibly located itself within the interaction between subject and condition, which exhibited significant results for all kinematic variables with the exception of Tp and βext. This implies that, while there are clear main effects of condition and velocity, subjects reacted idiosyncratically to the various running surfaces.
Given the “subjects reacted idiosyncratically to the various running surfaces” (and effects are very robust) it is clear that to state “non-motorized treadmills have been shown to be valid in comparison with overground running for maximum running velocity” is misguided.
The authors continue (TGc = ground contact time, Tb = braking time, βo = knee angle at touchdown, α1 = hip angle at toe off, βflx = knee flexion velocity):
Finally, the relationship between velocity and some kinematic variables varied with condition for TGc (p = 0.002, ES = 0.63, power = 0.91), Tb (p<0.001, ES = 1.02, power = 1), βo (p = 0.001, ES = 1.02, power = 1), α1 (p = 0.028, ES = 0.47, power = 0.67) and βflx (p=0.001, ES = 0.72, power = 0.97). For example, the knee angle at footstrike becomes increasing flexed with increasing velocity for the conventional and overground running conditions, but on the torque treadmill, the knee becomes more extended at foot strike as the velocity increases (Fig. 4). The combined effect of these variations is a null result for the main effect of βo with velocity.
McKenna et al seem to share Brughelli's approach of not actually presenting any mean or individual values regarding maximal running speed on the torque treadmill, which is obviously a strange approach if the purpose of the paper is investigating the limiting factors for maximal sprinting performance. It is however clear from the Figure 4 that the maximal velocity in the torque treadmill condition was less than in the overground or conventional treadmill conditions.
So in terms of using the torque treadmill as a biomechanical model investigating limiting factors in maximum sprinting:
1. The maximal speed is decreased, and therefore intensity is increased at a given speed.
2. Essential Components of Performance are altered as relationships between kinematic variables have changed.
3. The surface is potentially invalid and unreliable for extrapolation to overground sprinting as "subjects reacted idiosyncratically to the various running surfaces".[/b]
McKenna et al in reality have found that sprinting on a torque treadmill represents an inherently different motor task to overground sprinting.
Brughelli conducted his research at Edith Cowan University, where Rob Newton (who does some very interesting work) is a professor. The previous PDF link of his study is from a commercial site for a fitness technology company, which sells the Woodway 3.0 treadmill used. Newtown gives a lengthy video discussion of the treadmill on the website https://www.fittech.com.au/videos/woodway_rob1.wmv
In the end everyone is free to draw their own conclusion on the objective validity of the Woodway 3.0 for use in the Brughelli study. It would appear findings on a torque treadmill are not equivalent to overground sprinting in many ways, contrary to the conclusion stated by McKenna et al.
Additionally, regarding extrapolating the Weyand findings, hopping and backward running are not the same as maximum velocity sprinting, and a big picture perspective not just contact times is needed.
-
Participant
Next we have Nummela et al 2007 ( https://www.kihu.fi/tuotostiedostot/julkinen/2007_num_factors_re_10000.pdf )
Protocol for max speed trials:
“[b]In the first test [b](8 × 30 m)[/b], the athletes were able to accelerate 50 m to ensure a normal and steady running gait throughout the 30-m measurement section. The speed of the first run was 5.0 m• s-1 and, thereafter, the speed was increased by 0.40 m• s-1 after each run until the sixth run (7.0 m• s-1).[/b] The running speed was regulated by small lights placed on the next lane at intervals of 4.0 m (Naakka Ltd., Lappeenranta, Finland). The athletes were instructed to adjust their speed to coincide with the lights, which were switched on and off along the track. [b]The athletes were asked to run the last two runs at maximal speed. Each running bout was separated by a two-minute recovery period during which the runners returned to the start of the sprint course. [/b]In order to measure running speed, stride rate and stride length, a photocell contact mat [22] and two photocell gates were placed on the sprint course (Ivar Ltd., Tallin, Estonia). A special 9-m long force platform system was placed in the middle of the 30-m measurement section.”
The “maximal speed” trial involves a 50m variable acceleration into a 30m fly, where data is recorded in the “middle 9m”. This trial is the 7th or 8th in a series of 8, which are separated by a whole 2 minutes rest in which athletes must return to the start line.
Additionally the population is described as follows:
“Twenty-five young male endurance athletes (ten distance runners, eight orienteers and seven triathletes) volunteered as subjects for the present study. All the athletes belonged to the national junior team and their mean ± SD age, height and body mass were: 19.8 ± 1.1 years, 1.82 ± 0.07 m and 69.4 ± 7.5 kg, respectively.”
The limitations are clear:
There are possible kinematic changes with fatigue in the protocol, and inherent kinematic changes due to training background.
The intensity of runs 1-6 as per “The speed of the first run was 5.0 m• s-1 and, thereafter, the speed was increased by 0.40 m• s-1 after each run until the sixth run (7.0 m• s-1)” will vary depending on if the athlete’s max speed is 7.70ms-1 or 9.40ms-1.
There maybe fatigue during the 30m fly particularly in later trials, and the middle 9m of 30m may not represent max speed.
There will also be variation in acceleration pattern of different subjects with metabolic consequences.
Nummela’s plot (Fig 1 C and D) shows vertical effective force rises to 7.0ms-1 (i.e. through trial 6 of 8*30), but then levels off at “max speed”, while vertical impulse declines sharply, which is hardly astounding given what we know about fatigue patterns in repeated sprint tasks and the 400 metres.
Apparently a comparison of 9m of sprinting in a 30m “max speed” section, using endurance athletes (not even simply distance runners) ranging 1.82 ± 0.07 m, 69.4 ± 7.5 kg and maximal 30m speed from 7.70 to 9.40ms-1, and using a protocol where fatigue is very likely, is an example of robust evidence. Apparently also there was no need to compare individual progression but a scatter plot of all trials was sound, despite the fact that both inherent characteristics and the intensity of the protocol will vary greatly within the sample.
This study does not measure “Running mechanics as a function of speed” in a way that is relevant to sprint performance, but is confounded by the extent (variable between athletes) of fatigue ala the Hanon papers from the Bleed Runs thread. The use of pooled data to make inferences on maximal velocity mechanics is utterly invalid in the context of the methodology.
-
Participant
Kuitunen et al (2002) “Knee and angle joint stiffness in sprint runners” MSSE 34; 166-173.
The irony of this citation is interesting as there are some very clear lessons for Brughelli and Contreras in this investigation.
Some were no doubt delighted to read the following:
Vertical (Fz) and horizontal (Fy) GRFs were measured…The peak Fz was constant but the peak Fy increased both in braking (p<0.001) and in propulsion phase (p<0.01) with increasing speed (Fig 2).
Seasoned observers will note that "Fy" is an odd index for horizontal force 😉
For such an unusual finding (note this study was published in 2002, 5 years prior to any of the others mentioned) it is strange that an experienced group of researchers make no mention whatsoever of this apparent revelatory effect of horizontal force in their discussion, conclusions or abstract. The reason for this is that a typographical error likely occurred and the labels are back to front.
What does rate a mention in the discussion is:
In the present study, the vertical stiffness increased linearly with the running speed. It is consistent with the previous results in slow running (2, 12). Increase in the vertical stiffness was accompanied by decreased vertical displacement of the COM during the eccentric phase (-37.9%, p<0.01).
It would be quite curious for this to be observed with vertical force remaining constant 😉
The researchers also included EMG plots for various muscle activation patterns which demonstrated changes with increasing speed. The only muscle showing significant increases for 100% vs 90% speed throughout the ground phase was biceps , while others showed changes through portions of ground phase, and others only during swing phase.
Soleus showed increases:
Average EMG of SOL muscle increased both in preactivation (p<0.01)) and during the contact phase (p<0.05) with increasing speed
As did the vastus medialis and rectus femoris (note implications for preactivation allowing knee stiffness during braking)
EMG activity of the VM and RF muscles increased with increasing speed (p<0.01) in preactivation and the peak activity also occurred earlier at high speeds in VM
And the biceps femoris
The activation of BF increased also with running speed, but the increase was statistically significant only in the contact phase (p<0.05)
However the gluteus maximus didn't seem to contribute to outrageous hip hyperextension at toe off for increased speed, falling to virtually no activation after the first part of ground phase (very similar was observed by Tidow et al https://www.nacactfca.org/sprinting.pdf , see "rear support" on page 12 figure). Instead activation during swing is elevated at higher speeds.
The preactivation of GM muscle increased with running speed (p<0.05) and then decreased during the first half of ground contact
So it seems some will overlook typographical errors and jump to conclusions but ignore very strong evidence relating to their favouritr claims for which they possess none.
Not only is force very low at terminal hip extension, but it has been known for quite some time that gluteus maximus activity is somewhat high during swing phase (lumbopelvic control for optimal ground phase mechanics in my opinion as per other threads) but falls to almost zero by about 5ms of ground phase in sprinters. Why Brughelli or Contreras might think using the glutes for hip hyperextension will increase maximal sprint velocity is beyond me as the evidence is utterly to the contrary.
As gluteus maximus activation falls rapidly with continued hip extension in the ground phase of sprinting, I think it's fair to say "findings" of supramaximal glute activation at the top of a hip thrust are truly irrelevant to sprint training. Alternatively, you could suggest use the hip thrust exercise for sprint performance will promote very detrimental intramuscular and intermuscular co-ordination.
I think coaches should be excited by the recent findings about vertical forces, and be open to implementing additional training methods for improving speed, acceleration and overall athletic performance. I think coaches should start implementing more horizontal strength and power exercises, hip extension/hyper-extension exercises, proper eccentric exercises, and continue to implement single-leg exercises.
Bret Contreras says:
October 2, 2010 at 12:38 am
Stuart, most coaches and CSCS's never read any journals. In fact, during the course of a year an overwhelming majority probably read zero full papers. Most rely more on conferences, online articles, and gossip for the latest methodology. They basically wait for stuff to get popular then jump on the bandwagon. There are some who read the journals and these guys are often trailblazers. There are some that publish in the journals too, and these guys are very valuable. The science/math is very difficult for most to understand (at least for studies related to Biomechanics). It always takes SOOOOOOO long for trends and "the truth" to spread. Crazy.Mark Young says:
August 25, 2010 at 12:19 am
Seriously, who gives references in an interview??? That is friggin' awesome! This guy doesn't mess around. Excellent information!Bret Contreras says:
August 25, 2010 at 7:37 am
Yeah, these guys are in another league Mark! Thanks!Needless to say I think these statements are a bit questionable.
Again I'd come back to keys in applying science of:
1. A big picture perspective of biomechanical analysis.
2. An ability to critically analyze research.If anyone reads a paper with the expectation of relying on the author's themselves in learning the background and analyzing critically, the consequences can be severe, whether it's a novice or a professional academic doing the reading. Relying on expert opinion and peer-review is not always reliable as there are people involved and people make mistakes, since science is simply a human construct and we might as well be in a church listening to a figurehead preaching if we simply accept opinion as fact and don't think for ourselves (no offence intended to religious members).
-
Participant
This study just popped up on Charles’ Poliquin’s and Bret Contreras’ websites:
https://www.ncbi.nlm.nih.gov/pubmed/22422028
Any thoughts on this one? Perhaps this could be a separate thread?
-
Participant
This study just popped up on a couple sites:
https://www.ncbi.nlm.nih.gov/pubmed/22422028
Any thoughts on this one? Perhaps this could be a separate thread?
Bret’s done a blog post on this one too: https://bretcontreras.com/2012/03/why-is-christophe-lemaitre-so-damn-fast/
-
Participant
Edited to be a bit less harsh as that probably doesn’t help communication of ideas.
I disagree fundamentally with the sentiments of
https://bretcontreras.com/2012/03/why-is-christophe-lemaitre-so-damn-fast/ and
Eur J Appl Physiol. 2012 Mar 16. [Epub ahead of print]”Mechanical determinants of 100-m sprint running performance.” Morin JB, Bourdin M, Edouard P, Peyrot N, Samozino P, Lacour JR.1. Are we actually studying limitations to sprinting performance in Lemaitre vs National class vs non-sprinters?
The aim of this study was to investigate the detailed mechanical variables associated with field 100-m performance and discuss recent hypotheses about the mechanical determinants of sprint performance
Morin has chosen to present group data for the 3 national level sprinters, specifically mean and SD for a group of three national-level sprinters performing a single trial in each condition (overground and torque treadmill)on two occasions, rather than individual data. Use of standard deviation within a sample size of 3 is obviously ridiculous. This “may” have been done because in fact there is no significant maximal velocity performance difference between Lemaitre and the national level sprinters apparent in his measures (especially compared to what might be expected), note (2SD is a normal level of “statistical significance” see https://en.wikipedia.org/wiki/Standard_deviation):
Christophe Lemaitre’s maximum velocity (ms-1) during the 100m field test was within “1SD” of the 3 national level sprinters, 11.21 vs 10.78(0.37).
It is only on the torque treadmill, that his maximum speed was greater than “3SD” superior to the national level sprinters, 8.67 vs 8.13 (0.18).
Additionally, Morin has grouped the sprinters based on all time best, but with Lemaitre doing 10.35 for 100 in the study itself and the others 10.92 with “SD” 0.20, Lemaitre is not nearly the outlier in this study that might be assumed.
McKenna et al: This implies that, while there are clear main effects of condition and velocity, subjects reacted idiosyncratically to the various running surfaces.
Morin has (unwittingly?) illustrated this:
Christophe Lemaitre’s average velocity (ms-1) for the field test was “>2SD” different (SD with 3 subjects? WTF?) of the national level sprinters, 9.66 vs 9.16 (0.17). However on the torque treadmill, his average speed was within “1SD” of the national level sprinters, 7.08 vs 6.77 (0.21).
Christophe Lemaitre’s maximum velocity (ms-1) during the 100m field test was within “1SD” (again SD with 3 subjects? WTF?) of the national level sprinters, 11.21 vs 10.78(0.37). However on the torque treadmill, his maximum speed was greater than “3SD” than the national level sprinters, 8.67 vs 8.13 (0.18).
Needless to say, Lemaitre’s maximal velocity is not 8.67ms-1 and step length is not 1.53m in overground sprinting. Again if Morin had presented individual data, variations in individual speed relative to others with the change in running surface would likely have been clear and so we are left with the hilarious presentation of a group mean and SD within the group of 3 national level sprinters.
Bret concludes:
What makes Christophe Lemaitre so damn fast? … his shows that throughout the race and especially at high speeds, he continues to produce high levels of net horizontal force.
The only problem with that is they don’t have any valid measures relating to a race.
Morin attempts to calculate a strange index of “theoretical max horizontal velocity” and “theoretical max horizontal force” for which individual values are not presented in the paper, and compared to the values actually measured in the field test we see for “horizontal velocity”:
Actual overground 7.80-11.2, “Thereotical” from torque treadmill 7.71-14.0. You don’t need to have a statistics background to get the impression this represents some random error, given the minimum drops while the maximum gains considerably.
It comes as no surprise that no individual data is presented once again. Bret did however share:
Lemaitre possessed a theoretical max horizontal velocity and theoretical max horizontal force of 14.0 m/s …whereas the slowest sprinter possessed a theoretical max horizontal velocity…of 8.28 m/s.
I don’t think we need to look out for 14.0ms-1 in London, and it unlikely the 15.03 sec performer can “thereotically” manage 8.28ms-1. Needless to say comments made by Bret and Morin on “thereotical” indices and the DRF are without basis as they are from a torque treadmill and also ignore some fundamental physics (see next post).
Morin et al have used these theoretical values to compute their pet index of DRF on the torque treadmill. In other words, they calculate DRF on the assumption that Lemaitre can run 14.0ms-1 on the torque treadmill, despite his measured maximum velocity being only 8.67ms-1! What’s more the 14.0ms-1 is thereotically achieved with zero power output (details in next post), i.e. there are no GRF’s and we are observing flight without energy cost!
Additionally, the kinematic alterations noted on a torque treadmill make comments like this invalid:
First, though the National sprinters along with Lemaitre did possess better vertical stiffness than the non-sprinters, leg stiffness did not vary between Lemaitre and the National sprinters and surprisingly the non-sprinters as well.
To use force data from the torque treadmill to describe differences in overground sprinting performance is quite invalid.
All Morin and Contreras have described and presented in terms of kinetics and kinematics is data from the torque treadmill, and there is no valid data on overground sprinting kinetics and kinematics whatsoever!
All the additional concerns I have written about below are essentially moot points, because the validity of the research design and practical applications are non-existent at the outset!
-
Participant
2.The use of correlation co-efficient to illustrate differences
In a sample of 13 we must be very careful with the possibility of outliers skewing a correlation co-efficient. Needless to say, in a sample size of 3 national level sprinters, a correlation co-efficient is limited, but again the presentation of individual values for what ever reason, is not a better option from the authors’ perspective.
Needless to say, within such a heterogenous group of 4 sprinters vs 9 non-sprinters there is a large number of extreme outlying values including those Bret points out in the blog.
3.The Use of the “DRF” index of force application technique
Again, this DRF index is calculated from technique on the torque treadmill, which as per above is invalid for comparison to overground sprinting and shows a hugely idiosyncratic response from Lemaitre in particular. Again, Morin does not present individual values for this in the published paper that has been peer-reviewed, and all we have is what Contreras shares in a blog.
In the published paper Morin et al rely upon a correlation co-efficient (difficult to justify in a group of 3 as per above points), and rather than presenting a table (4 sprinters doesn’t make for a huge table) or even a scatter plot of all subjects or the 4 sprinters, which would readers a slight impression of the data Morin et al instead present:
Fig. 1 Typical linear force-velocity and 2nd degree polynomial power-velocity relationships obtained from instrumented treadmill sprint data for the fastest (100-m best time: 9.92 s, 100-m time of 10.35 s during the study: black and dark grey circles) and slowest (100-m time of 15.03 s during the study: white and light grey circles) subjects of this study.
That is to say the most feasible data presentation regarding DRF appears to be a comparison of force application technique in one of the most talented and highly trained sprinters on the planet to a 15.03 sec runner!
Fundamental Issues with the DRF Figure
As per the above in computing the DRF Morin et al have forced the line of regression through 14.0ms-1 maximal velocity to get a good correlation.
The data points start at around 2ms-1 and so Morin et al but despite this the line of regression has been forced through 0ms-1.
They have used Y-intercepts as follows (thia “Coyle fallacy” seems to be spreading 😉 ):
At 0ms-1, Horizontal force is >8N.kg-1 for Lemaitre and >6N.kg for 15.03 guy.
At 0ms-1, Power Output is 0W.kg-1, Force is 0W.
At thereotical maximum velocity, horizontal force, power output, and force/velocity are all ZERO!
The theoretical values presented in the DRF calculations completely ignore the basic laws of physics.
On this basis Lemaitre will be floating to gold in London!
These intercepts are not based on any data as the first data points are at around 2ms-1, and the data ends at 8.67ms-1 for Lemaitre.
Moreover, they are impossible in real running and contradictory!
1. Apparently the two subjects are producing large and different amounts of positive horizontal force without moving?
2. Apparently at 0ms-1 both sprinters have ZERO power output and ZERO force but are still producing horizontal force?
3. Apparently power ouput, force etc are ZERO at maximum velocity?
3. Apparently in a static standing position power output and force is ZERO (I guess we will soon see Newton disproven with a “death to gravity” campaign 😉 )?
Obviously the DRF index is problematic. Once the line of regression is adjusted to reflect reality, the correlations and differences between athletes are greatly affected. All references for validation of DRF are from Morin himself, and to be honest if this were replicated the issues mentioned would likely be very clear.
It would appear the torque treadmill is potentially not only unaffected by friction, but causes subjects to be unaffected by gravity as well!
Do faster sprinters have higher body masses, thereby allowing them to produce greater force?
This is a very strange question to be asking. Sprinting is a mass specific task, muscle produces force (body composition is not homogenous in groups as varied as that of Morin et al), and specific muscles are more responsible for locomotion than others. To be honest I think Contreras is overly taken with specific research and needs to understand the basics first:
French researcher JB Morin is leading the way with some amazing research over the past several years…This study is one of the coolest studies I’ve ever read and I commend the researchers for putting together such a comprehensive study in order to advance our understanding of sprint biomechanics.
“Amazing” and “coolest” are difficult descriptors to apply objectively, but needless to say I think more focus needs to be paid to factors like “validity” and “relevance”.
Back to the findings
The ability to apply the resultant force backward during acceleration was positively correlated to 100-m performance (rs [0.683; P0.018), but the magnitude of resultant force was not (P = 0.16).
Having dealt with the first sentence, let’s look at why resultant force might not correlate at very high intensities prefectly (Will Hopkins is one individual who Bret might seek advice from on issues like whether to reject p=0.16 out of hand, especially given the sample size and characteristics)
Obviously time will play a role as Power = Force/Time
On the torque treadmill (and again at the outset this is not relevant to overgground sprinting) Lemaitre’s contact time average is 0.121.
The national level sprinters’ mean and “SD” (SD is hardly valid in a sample of 3 sprinters) are 0.129 (0.003).
As Bret and Morin have been pointed out this is greater than a”2SD” difference.
When we see that kind of difference, at contact times that low, accounting for it in analysis is wise, instead of simply suggesting in an abstract “magnitude of resultant force” was not correlated to performance as this can be somewhat misleading.
When coaches want to improve athletes rather than study differences between athletes, the methods will end up getting them producing more force in less time at a variety of contraction velocities and adressing factors in isolation (especially when they are irrelevant) is unrealistic.
Summary
Torque treadmill research is not useful for studying/improving overground sprinting performance.
Science is about answering a question; in this case which factors limit 100m sprint performance, not dancing around a question and going off on tangents.
Certain “leaders” need make an understanding of basic concepts taught in physics and applied biomechanics the top priority.
-
Participant
Thanks for the thorough posts as it takes some work to produce them.
https://www.youtube.com/watch?v=aTG84QUAJaw&feature=relmfu
When Charles and Bret posted this study I just don’t know how we can look at our french sprinter as an example of strength and conditioning being the reason for his sub-10 performance. (see last 10 seconds). I did notice that a crouch start and 6 seconds (45-55 meters) of data was collected. With acceleration being highly biased towards horizontal forces this paper seems to be a poor example of posterior strength or strength in general for maximal speed development.
-
Participant
No worries Carl, hopefully examining research and ideas can allow methodology can improve, and result in more practical and valid information for many passionate and hard working athletes and coaches.
I would agree the effect of training for Lemaitre does not seem overwhelming as per year-to-year progressions below. Clearly Lemaitre is an outlier from the rest of the subjects studied by Morin et al and from other elite sprinters in many ways. Some researchers seem to ignore that with a sample size of 1 (as with Lemaitre’s comparison here or others involving contact length and Usain Bolt) individual characteristics are extremely influential. Morin admittedly tried to control for anthropometry but I’d suggest it was again misguided.
There are many fundamental problems with the methodology which promote a bias to horizontal force, and 6 sec on the torque treadmill with VMax 8.67ms-1 using a crouch start is unlikely to be even 40m of belt distance, and who knows what is happening metabolically with postures staying low and getting lower at higher speeds (clear implications apparent confusion on vertical stiffness from Morin et al). While Contreras or Brughelli may persist (though some already appear to have decided presenting running speed will be not strictly necessary..), there is simply no validity to the data from torque treadmills with reference to overground sprinting from any kinetic or kinematic perspective as the essential components of postural evolution, force application and metabolic outcomes are lost.
As far as I’m concerned, these guys have completely lost any direction concerning the biomechanics of sprinting and are engaging in truly irrelevant research, as evidenced by Morin et al’s use of a theoretical model (DRF has appeared in more than one paper) which is completely without merit and Brughelli’s very clear misunderstanding of the literature. For Morin et al and all the others involved in the peer review process to not realize that the DRF model relies upon an assumption of zero power output and force at both 0ms-1 and maximum velocity is perhaps the most pertinent example anyone will ever see of missing the forest for the trees .
To facilitate comparison between subjects, the leg length to standing height ratio (L/H) and body mass index (BMI, kg m-2) were used.
The sample studied included a strong majority of 9 non-sprinters out of 13 subjects and we are not likely to see meaningful statistical correlation on leg length/height as various other physical/physiological variables have not been optimised and this will confound the relationship. These are the basic characteristics reported on subjects:
Physical education students [age (mean ± SD) 26.5 ± 1.8years; body mass 72.6 ± 8.4 kg; height 1.75 ± 0.08 m]
French national-level sprinters [age (mean ± SD) 26.3 ± 2.1 years; body mass 77.5 ± 4.5 kg; height 1.83 ± 0.05 m].
A world-class sprinter (age 21 years; body mass 81.0 kg; height 1.91 m)
If looked at this somewhat out of context of the sport and research, which seems to be Morin et al’s approach, we might say Lemaitre is significantly (>2SD) younger than both groups so maybe younger age is decisive in sprinting. Or equally that being taller and heavier is very influential and linebackers need to get on the track.
Obviously no-one would conclude on any of this because there are so many confounding variables at play and some are so obviously irrelevant but that’s the limitation of small, heterogeneous samples, let alone findings hinging on a single athlete.
As for Lemaitre’s improvement or future improvement, I think to a large extent puberty has done work so far on acceleration and max velocity and there’s been little change since his teenage years:
DOB 11/06/1990
100m:
2009 10.04 wind 0.2
2010 9.97 wind 0.9 at Rieti aka PR central
2011 9.92 wind 2.060m:
2010 6.55
2011 6.57
2012 6.57 -
Participant
Been a while since I have been on here, but several things I think people misunderstand about max V in human populations are as follows starting from zero.
1. Horizontal impulse diminishes each step in acceleration.
2. Vertical impulse increase each step in acceleration.
3. The end of acceleration is max velocity.
4. Vertical impulse diminishes as we decelerate
5. Horizontal impulse diminishes as we decelerate after reaching max velocity.These horizontal and vertical impulses are important in relation to step length and rate which give us the common equation for speed of step length / step rate.
Furthermore, we have observational evidence of runners being faster by completing their races in less steps, based on individual parameters and not on emprical parameters, ie.. one could theoretically train to do 100m in 39 or 40 steps like Bolt, but never come close to reaching their own personal record. This is the important distinction that should be made. The net gain from horizontal impulse during acceleration is what gives us our speed, the increase in vertical impulse on each step allows us to keep producing net gains in horizontal impulse as they diminish on each step. There is an optimal balance to be made, because increasing vertical impulses make ground contact times shorter which decreases the amount of time to achieve A net gain in horizontal impulse, because of a net loss due to friction at initial ground contact which must be overcome on each step. Sprinting, max velocity and acceleration are both skills that must be learned and is a motor learning and motor control issue with regards to optimization. If a sprinter doesn’t produce gains in net horizontal impulse during acceleration he will not achieve a higher max velocity, but if he doesn’t increase his ability create greater vertical forces he will not be able to alter his ability to accelerate to a higher max velocity and be stuck with the problem of accelerating to his max v faster and then holding that max v longer to produce faster times.
-
Viewing 19 reply threads
- You must be logged in to reply to this topic.