Upper body aids propulsion on a Cruzbike bicycle

At last. Empirical, reproducible evidence to demonstrate what experienced Cruzbike riders have known all along.

The essential tool for this proof has been riding quietly with me for thousands of miles, but I did not recognize its capabilities until recently, after stumbling upon an article about a special feature of the Stages crank-based power meter called “high-speed data streaming”. Stages released an app that runs on my smart phone that receives and collects torque data from the left crank 64 times per second. That’s a crank torque measurement every 0.0156 seconds. The data not only flashes across the screen, but is saved to a CSV file for detailed analysis and graphing. This is a game-changer. Previously, I could analyze power data in 1-second increments, but that’s far too long for analyzing what happens during a small portion of a pedal stroke.

First, I sat on my V20 with the left crank in (or near-to) the 12:00 (straight-up) position, I applied a constant force with my left foot to the pedal and watched the torque numbers stream by on my device. The brake was engaged so the bike wouldn’t roll away and the crank would stay in a static position. The streaming torque numbers stayed relatively stable with minor variations as long as I held my foot pressure constant. Then I pulled gently on the left handlebar, I’d call it a “tug” or an impulse of a fraction of a second. The torque numbers shot up and then settled back down. Every pull got the same result… an increase in torque at the crank. Torque at the crank is propulsive force, so Q.E.D. (proof completed, or in Latin “quod erat demonstrandum”). But was I done? Had I proven it beyond a reasonable doubt, meeting the standards that Cruzbike’s toughest critics will demand?  Of course not.

The obvious flaw is that the increased torque could very well have been due to me momentarily applying more force with my left leg (involuntarily or reflexively) every time that I pulled on the handlebar. To eliminate that possibility, I needed to replace my leg with a constant force of the same vector, but not subject to voluntary or involuntary changes. I attached the crank/pedal to a cable and pulley, and attached a 25 lb dumbbell to the other end of the cable.

Stages Power-meter crank cabled to a 25 pound dumbbell


Now there was a constant force on the crank, but voluntary (and reflexive/involuntary) leg forces had been removed from the system. The result with the dumbbell produced the same result, which was about a 30% spike in torque at the crank with each tug on the handlebar. Here is a graph of the actual data, with torque in Nm (Newton meter) on the Y-axis and time on the X-axis (because there are 64 data points/second total time is about 13-14 seconds):

graph of torque vs. time
Tugging gently on the handlebar produces a significant increase in torque at the crank.

The dumbbell created about 13 Nm of torque on the crank. You can see a slight oscillation due to the dumbbell gently swinging. Then you can see the four times that I pulled on the handlebar. Each pull significantly increased torque up to a total of 16 to 18 Nm. That’s about 30% more torque from just a little tug on the handlebar.

This video will show you the basic set up of the testing. My son, Will, helped out with the demo while I recorded video.

The next phase of the testing I replaced the dumbbell with three long bungee cords stretched out to a length of about 25 feet. I wanted to see if the torque increase would work the same with a mass-less force just the same as it did with the 25 lb dumbbell.  Here is the graph of the high-speed data from the bungee testing:

Torque vs. time
Torque at the crank with bungee cords, and handlebar tugging – 26 second data-stream.

I wanted a long bungee cord length because a long length is going to give a nearly constant force when pulled a very short distance. A short bungee will yield increasingly more force as it is stretched a short distance.
As the graph above clearly shows, the torque went up when I tugged on the handlebar. There is a downward baseline drift of the torque, which I ascribe to using old bungees that don’t have truly elastic properties. Or else the crank arm moved from being 12:00 to something less, thus shortening the lever-arm length, and thus lowering the torque measurement. I did not quantify the force of my “pull”, but this could be done with a fish-scale… hook one end to the end of the handlebar, and pull on the other end. I tried to use an amount of force easily produced while riding.


The upper body can do real mechanical work, specifically functional, propulsive work, by pulling or otherwise applying torque to the handlebar. This has now been shown by measuring an increase in torque at the crank, and illustrated by the elevation of a dumbbell, with handlebar tugging/turning. This is a major milestone in understanding the biomechanics of the Cruzbike. Whether or not Team Cruzbike sets a record in the 4-person RAAM (Race Across America) in June, Jason Perez beats everyone (again) in California’s toughest mountain double-centuries, or Cruzbike racers sweep the men’s recumbent categories at the 6-12-24 Hour World Championship (again)… this might be the most important Cruzbike milestone of 2017.


These results are easy enough for anyone with a Cruzbike and a Stages crank to verify. If you attempt to verify, please use eye protection in case a cable or bungee cord snaps or flies off.

While this testing only analyzed the increase in torque to the left crank with left handlebar tugging during the pedal downstroke, it would be expected to increase the torque in the left crank during the upstroke with right handlebar tugging (assuming the rider is clipped-in).

You may be wondering if I think tugging on the handlebar increases power (or torque) by 30% overall. The answer is I think it’s less than that. The 30% torque increase is during a snapshot in time. If that tug-impulse lasts about 20% of the downstroke, that would be approximately a 6% increase in overall torque. Therefore I’m going to estimate approximately a 6% increase in power is a practical and achievable estimate from the upper body technique. This may seem a small number, but it can make a huge difference in a race; especially a long race. More testing is definitely needed.

Optimal timing and intensity of the handlebar tugging are acquired skills best developed through deliberate practice during riding on the road, but can be practiced on a trainer, even though the motion of the front-end is highly constrained on a trainer. The aim is to turn the handlebar slightly so that the left pedal/BB moves slightly toward the left hip simultaneously as the left foot is making the powerful downstroke. A split second later the technique is used on the right, during the right foot’s downstroke.

Use of the upper body to add propulsive force is optional. Extra power is never “free”. However, there are great advantages to being able to distribute workload across numerous muscle groups. Muscles become depleted of metabolites. By-products like lactic acid build up within exhausted muscles. Maximum power can be increased and duration of exertion can be increased when more muscle groups are involved in the activity.

Comparison to other bicycle formats:

Rear-wheel-drive (RWD) recumbents cannot gain any power through this technique because tugging on a handlebar only turns the front wheel and imparts no motion (or torque) to the cranks. Most people think the much longer drive-chain and the extra idlers/chain-guides are why RWD recumbents may be slower at climbing. I believe the lack of upper body input is a bigger disadvantage.

On the other hand, standard bikes have this same capacity as Cruzbikes for torque on the handlebar causing increased propulsive torque at the crank. Actually, they have more of it because the standard bike frame may be leaned farther than a Cruzbike boom should be swiveled. However, they have a much higher energy cost to pay when they use this technique: 1) they MUST lift their body off of the saddle to allow the bike to rapidly lean side-to-side; and 2) they become much less aerodynamic in this position. Cruzbike riders can essentially use the technique continuously without either of these penalties. This advantage, in addition to the well known aerodynamic and ergonomic advantages, is why the Cruzbike is becoming the choice of performance-oriented recumbent racers and is the fastest climbing recumbent on the market.





  1. From the last paragraph: ” … the standard bike frame may be leaned farther than a Cruzbike boom should be swiveled.” -Any plans to overcome this disadvantage? Will we see a Cruzbike that “is the fastest climbing [bike] on the market.”

    1. As I understand the differences in the way a standard bike and the Cruzbike use upper body input for propulsive power production, I wouldn’t say one method is overall disadvantaged compared to the other, they each have their own constraints. Can the Cruzbike design be improved to climb faster? Probably so, but I think our riders have more to gain from practicing on the current designs, than, say, waiting for a slightly lighter front-end.

  2. I think every force applied will always generate a reaction in the opposite direction. This experiment just confirms what physics says.

    Since for a moving bottom bracket (MBB) bicycle design like the cruzbike, the bottom bracket is pivoted on the steering head tube, every sideways force applied on the pedal has to be countered on the opposite direction if the wheel has to remain straight relative to the main frame.

    The handlebar has therefore to be stabilised by the arms when the legs apply the pedal force. Hence the upper body input. Secondly, when the pedal force is not big it can also be stabilised by the opposite leg and torso when riding hands free.

    This is, to some degree, similar to riding a DF off the saddle.

  3. However you don’t need the handlebars to steer a Cruzbike. Forget about the pulling and tugging. Install an additional hand crank and spin arms and legs simultaneously. Full body drive.

    1. Even though I can ride my Cruzbikes “no-hands” I still need my handlebars for steering most of the time. Adding a second set of cranks, chain, etc., for the hands to spin would add a lot of weight and complexity. I like how simple the Cruzbike design is, and yet how functional it is.

  4. There is also an other thing I would like to point out.
    When an ordinary person on an upright bike is pedalling , one rest the non power given legs on the pedal which make the power given legs to overcome that with its ovn weight.
    Proff. bikers have therefor there feet closed to the pedals to give power all around pedalling. This power is given from the stomach muscles. On a recumbent bike (and specially a cruzbike) this power is given for free, based on the fact that sitting or laying position gives a less power jobs for the not powered legs resulting in more power from the power given legs and less work from the stomach muscles. Try to think about that next time you bike.
    This from a new T50 biker from bikers country Denmark and an old long chain recumbent biker.

    1. Hi Joern,
      That’s an interesting theory and one I have NEVER heard before. Another way to say it is that our pedaling motion requires less work to overcome gravity because much of the thigh/leg mass is moving horizontally rather than vertically. I like it! Thanks for pointing that out.


    2. I don’t think this makes a difference. On a upright bike you have the gravity force of the other leg compensating the gravity force if the leg that needs to be lifted. The two forces cancel.
      However, as you say the professionals do, you should also buy click pedals and then push and pull with each leg. It really makes the bike go faster.

  5. Hi Joern and Jim,

    Wouldn’t that just be
    Diamond Frame “normal” bicycle:
    force of gravity pushing leg down + force leg muscle – force required to counteract gravity pushing opposite leg down = force of leg muscle. That is, force of gravity helps on one side while it is exactly the opposite on the other side all other things being equal?

    Which on a recumbent would be just plain force of leg muscle? (The same opposing forces due to gravity just they are more easily taken out of the equation because the applied force is roughly horizontal.) The differences seems like it would be more in what you can brace against to push that leg down whether it is pushing against gravity (a potential energy bank) and arm-stomach pull vs. pushing against seat and arm-stomach pull?

    I am probably missing something but it seems like there are slightly different applications of the forces but the general concepts and angles relative to the body or to gravity remain in play they just are rotated in the case of the body forces. These thoughts and discussion seems to point us again to the advantages of Cruzbike’s Moving Bottom Bracket design with it’s more diverse use of body forces (more arm and stomach in addition to leg) in comparison to other recumbents and possibly to it’s closer similarities to diamond frame bikes. Maybe further study or a closer look at the particular interactions of muscle groups and forces when comparing gravity vs. horizontal pushing and pulling of some of our elite riders would reveal things that could be of benefit.

    Joern, thank you so much for igniting my thoughts on this issue, I will be pondering this during my rides for quite some time.


  6. Hi Jim
    I have been thinking about your theory that arm pull helps for propulsion for some time now. In addition I was on a steep ride last week (I live in the Alps) with a T50. My conclusions:
    1) As long as you just apply a force (and no path) with your arms, there is no energy applied, only force. (As energy = force x path). This means: you use your muscels, therfore use energy, but don’t get any energy for propulsion. This is inefficient.
    2) If you start to pull some length with your arms, you can indeed generate additional propulsion. However, I think it is inefficient, as it is only possible to pull the handle a very short distance (maybe an inch). It really helped to do this on very steep slopes, but it was also excessively exhausting. I even had to walk after a short time doing this.
    3) The reason the force increases when you pull the handle bar in you experiment could be 1) prior to the pulling the handle bar, the wire does not pull in line of the pedal movement. The wire force causes a small angle between wire direction and bike axis. This means some of the wire force gets lost and is not measured. Once you pull the handle bar, the wire and bike axis are perfectly alinged an therefore 100% of the force is measured. 2) When pulling the handle bar, you lift the the weight which means you slightly accelerate it, which requires an additional force.
    Conclusion: you should think in terms of work (energy) rather than force.
    Here is the ultimate experiment: Build a new bike from the vandetta but change the following:
    – make a trike instead of a bike
    – this allows you to do back wheel steering instead of front wheel steering
    – now you can get rid of the front wheel steering, fix the handle bar an front wheel to the frame
    You now created a bike that still has the short chain but no longer requires pulling on the handle bar for compensating the off-center leg forces. I predict that even though the two back wheels are less aerodynamic, you will be faster because you no longer waste energy for compensating the leg force with your arms. This would be true for the T50 where the arms are much more bent than on a Vendetta, which leads to a much higher waste of energy. (Mucels require energy to hold a force, tendons don’t)

    1. Hi Mark,
      Pulling on the handlebar with the proper timing and coordination does, indeed, create force and the force moves (path, as you call it) the crank in a propulsive direction.
      If you are not experiencing this, if you conclusion is that you are wasting energy, then you need more practice. Beginners often feel like they are wasting energy, and they probably are because their efforts are not coordinated.

      You are correct that the motion accomplished by the arms is small (maybe an inch), which is very little compared to a standard bike. However, on a standard bike, leaning the frame can only be accomplished after lifting the body off of the saddle (at a high energy cost), and by putting the body in a very un-aerodynamic position. On a Cruzbike, the upper body input is less, but there are no penalties as on a standard bike, so the technique can be applied indefinitely.

      This past Saturday, I won a 100-mile Championship race by attacking on a climb. The RWD recumbents could not keep up. Give your new T50 about 500 miles and then write me again.


    2. There is also a twisting force within the body which is counteracted by engaging the upper body and core muscles. This force on the handle bars is more necessary on a Vendetta, I imagine, where the seat back provides less of an opposing force to the pedalling force. In your case Mark, the T50 has a more vertical seat which you can push against so there is less need to apply a force to the bars to increase peddling force except to counter the turning effect on the steering. In all cases we must push against something and the harder we want to pedal the more we must push against something. Even when climbing seated on my df bike I apply a lot of force to the bars to counteract my body’s twisting motion and the changing direction of the forces applied by both feet pulling and pushing.

      My suggestion is to try climbing those hills with no hands on the bars and try applying a strong force to the pedals and see how that goes. Good luck!

Leave a Reply

Your email address will not be published. Required fields are marked *