Why won't my perpetual motion
machine work?

by Donald E. Simanek.

The machine in this illustration is from an idea
of 18th century Scottish astronomer James Ferguson,
intended to show why perpetual motion wheels can't work.

I don't maintain a web message board. But many people contact me by email. Here's a compilation of questions people have raised over the last 15 years or more, and some reasonably short responses.

Is an "over-unity" machine the same as a perpetual motion machine?

These terms are used various ways and sometimes carelessly. Literally "perpetual motion" means "moving forever". One trouble with a perpetual motion machine is that it would take forever to test it. Seriously, if a device continues its motion for a very long time without any measurable decrease in its motion, we would have to say that it is indistinguishable from a perpetual motion device. The other trouble with such a hypothetical device is that it only has as much energy as you give it initially, so as soon as you extract any energy or work from one, its motion would decrease and soon stop. So it wouldn't be useful as anything but a fascinating curiosity. If an unpowered spaceship were moving in space, with nothing else that could affect its motion, presumably it could move forever in a straight line without slowing down. And if there were nothing else in the universe to affect it, there would be no one to observe it and no reference to measure its motion against. But the universe is filled with stuff, and everything continually interacts with other things. Even the planets moving about the sun gradually lose mechanical energy because of dissipative tidal forces. Atoms seem to persist indefinitely if not disturbed, but we can't really say (nor measure) whether anything within an atom is moving.

The fundamental laws of physics do not prohibit perpetual motion. But properties of physical objects conspire to prevent it. Friction and other dissipative processes convert orderly motion to disorderly motion, thwarting our efforts to achieve perfect (100%) energy efficiency in machines.

Inventors have little interest in merely producing perpetual motion. They want over-unity performance—unlimited energy output for free.

"Over-unity" means an energy efficiency greater than one. Suppose you could make a device with efficiency of 200%. It puts out twice as much energy as it takes in. Then you could take half the output energy and use it as input energy for the device, and keep it running forever. This would be a perpetual motion machine and would put out useful energy as well.

Energy flow chart of a hypothetical over-unity machine. The secret over-unity device (OU) increases the energy by some unknown physical principle. Some of the energy is fed back to the input, some wasted as thermal energy, and some output as useful work.

This is why "over-unity" and "perpetual motion" are often used as if they were synonyms. If you can achieve over-unity, you have also achieved perpetual motion, but not vice versa.

Wouldn't this be dangerous?


Suppose you have built such a 200% efficiency machine. Feed back half its output to its input and it then needs no other input power source. Take its remaining output and feed it into another such device, and then you get double the initial output. Cascade a bunch of these in series and you get unlimited output power for free, except for the cost of making the machines. For a modest investment in machinery you could power the earth. Or destroy it.

But, even simpler, just feed back all of that output to the input. The machine is now out of control, a case of positive feedback leading to instability. The power output could only be kept from going to infinite values by the machine's internal power-handling limitations, or by power-limiting safety systems that we build into it. Even if the over-unity is modest, say an efficiency of 110%, you could still get unlimited power output, it would just take a bit longer to build up. Don't try this at home!

A perpetual motion machine would be safer, for it doesn't increase energy.

Pierre Richard (engineer, Paris), 1858,
British patent No. 1870.
Notice the friction brake on the left side.
Dircks (1861), p. 482.
You may have noticed that no one has yet achieved this, or has even come close. I'm not worried about this doomsday scenario happening. I have noticed that several times in recent years some people have claimed they have achieved something like 135% efficiency in their machines. I strongly suspect that they don't know how to measure efficiency properly.

In the older literature one sometimes sees perpetual motion machine designs in which the inventor has included a friction brake, or a speed governor, presumably to guard against such catastrophes. Or maybe they were hoping that the load at the output would keep the machine from destroying itself. But in recent claims of presumably more sophisticated devices, the inventors seem to have completely ignored the possibility of instability due to over-unity performance. And yet we haven't heard of any of their devices suffering meltdown. Curious, isn't it? But then, we haven't heard of any confirmed reports of their wonderful devices producing useful work continuously either. Hmm...

The reader may recognize the similarity of this to the ancient mathematical fable of the Wheat and chessboard. One grain of wheat is placed on the first square of the board, 2 on the second, 4 on the third; each time doubling the previous amount. When all 64 squares are filled with wheat, there are 18,446,744,073,709,551,615 grains on the board. After 64 passes through our 200% efficiency over-unity device with energy feedback, the energy has been multiplied by that much.

If this could be controlled, wouldn't it be completely non-polluting?

Not necessarily. All machines waste energy through various dissipative forces. Friction and viscosity are examples. These processes convert mechanical energy to thermal energy, and this "heats up" the machine and its surroundings. This is a separate problem, and no one knows how to eliminate dissipative processes completely. So these wonderful over-unity machines proliferating around the world would still contribute to global warming.

This exposes a common misunderstanding. In the world of real machinery poor efficiency is thought to be only due to friction. Just reduce the friction to zero, and all we'd have is a perpetual motion machine with efficiency equal to one. But reducing friction and other dissipative processes to zero won't get you to efficiencies greater than one, that is, over-unity performance. To do that you'd have to find some way to "multiply" energy, or create energy. If that were possible some of the newly created energy could be useful output, but some would still be converted to thermal energy due to friction.

What does "closing the loop" mean?

It means what the person using it wants it to mean. Seriously, in over-unity circles it usually means to feed a machine's output energy back to its input, in order to sustain the machine's motion without any other energy source. Suppose an inventor has a machine, with input and output, but he claims the output is greater than the input, say a 150% energy efficiency. Skeptics suppose that he's measuring or calculating the energies incorrectly, but the measurements are complex and subtle so some simpler test is wanted. Why not simply feed the output energy, or some of it, back to the input to replace whatever was supplying input energy before? Then if the device still runs, without extgernal energy source and without decreasing its energy output, that would be strong evidence that it was actually producing, or creating energy by itself. Of course no one has ever accomplished this, and inventors often refuse to even try. They offer excuses such as "the output energy is in a form incompatible with the input". That's lame, for we know many ways to convert the forms of energy, potential, kinetic, electrical, magnetic, nuclear, thermal, etc. etc. from one to the other and back again. In such cases we suspect the reluctant inventor is hiding something, perhaps a hidden energy source that is part of a scam.

Example: A motor is used to drive a generator, and the electrical output of the generator is fed back (closing the loop) to power the motor. Sounds absurd, doesn't it? It is. Even if the motor and generater were both 100% efficient, it would not work.

How can I calculate or measure the efficiency of my device?

First, be clear what efficiency you are talking about. Mechanical efficiency is the ratio (useful work out)/(energy in), the two being measured simultaneously over the same time interval. In some cases, where the output is steady, it is easier to measure (useful power out)/(power in) simultaneously. Efficiency is not the ratio of forces out and in. "Useful" work is that which can move objects by applying force to them. This said, the best way is to measure the output and input energies. The output energy can be measured several ways: (1) use the output power to lift a weight a measured distance, (2) use the output to heat a resistor/thermocouple arrangement and measuring its temperature change, (3)heat a glowing incandescent light bulb with the device's output power until its brightness matches that of an identical bulb powered by DC current, measuring the current and potential at that bulb and using P = IV. Similar methods may be used to measure the input energy at the same time. These methods help avoid complications due to non-sinusoidal waveforms, pulses, etc. But there are still pitfalls for the unwary, some of which I deal with in my document Testing perpetual motion machines.

My device has an energy efficiency of 95% when it has input power driving it, but without energy input it comes to a stop. If I increase the efficiency just a bit more, say just 10% more, won't it have efficiency of 105%, that is, over-unity performance? It could then run itself without input power by closing the loop. That shouldn't be difficult.

You need to learn more about how to use percents properly, and also how mathematics can be misapplied to the real world. If the input power is zero, then 105% of zero is still a power output of zero.

Light objects rise in liquids, the buoyant force overcoming gravity. Can we tap that buoyant force continually to do useful work?

I see many proposals with lightweight balls rising in a liquid column. Objects lighter (less dense) than a liquid will, if placed at the bottom of the liquid container, rise to the top. They do this against the pull of gravity. To the naive observer it seems that they are tapping energy from the liquid or from gravity. They aren't doing either one of these things. The energy they gain as they rise to the top comes from the work done when they were initially pushed to the bottom, working against fluid pressure.

Some inventors try to tap energy of the balls by letting them fall from the top of the liquid tank down some sort of ramp to the bottom, extracting energy from them as they fall down. Then the balls are inserted back into the water through some clever mechanical valve at the bottom of the tank. Unfortunately, the work required to push a ball through that valve, working against the pressure difference, is just equal to the energy it would gain as it rises to the top of the liquid. There's no energy gain in the process, only energy lost to viscous drag.

Simple mechanisms can multiply force. They can also multiply the distance through which a force acts. Can't we simply combine or reconfigure these to multiply both force and distance simultaneously, and therefore multiply work?

The simple machines known to the ancients. Typical presentation from a modern elementary physics textbook. IMA is the "ideal mechanical advantage", i.e., the mechanical advantage (ratio of output to input force) in the perfectly frictionless case.

The ancients did more than discover these machines. They also analyzed them to find out how they worked. In the process they learned that the product of force and distance could never be increased in any of these mechanisms or any combination of them. Work is the product of force and distance. Work out = Work in - Energy losses due to dissipative processes. Some perpetual motion machine inventors haven't yet caught up on this fact of nature.

My device will require magnets and magnetic shields. Where can I buy suitable magnetic shields?

Advertisement for a
magnetic corset.

Magnetic shields are useful for keeping the magnetic component of electromagnetic radiation away from sensitive circuits. They are not as effective for steady or slowly varying magnetic fields. Magnetic shields work by redirecting magnetic field lines so that they are mostly kept away from regions where we don't want them. They do this by re-routing magnetic field lines through the shielding material instead of somewhere else. Therefore the magnetic material still experiences forces from the magnets, and Newton's third law still applies. A magnetic shield and a permanent magnet are strongly attracted to each other. So in analyzing a device with magnetic shields you must include all forces acting on the shields and the forces shields exert on other parts of the device. Inventors usually totally neglect even considering the forces on the shields and the work done on and by these forces. For rapidly varying AC fields, the average force exerted on the shield can be nearly zero, but considerable mechanical energy is still lost by heating the shield. While the electric field component of AC radiation can be almost entirely shielded from a finite volume by a full metal enclosure around that volume (acting as a Faraday cage), the magnetic component can never be completely shielded.

Nickel-Neodymium magnet on top of a
sheet of mu-metal magnetic
shielding easily supports
unmagnetized ferrous objects
on the bottom side.

Commercial shielding materials are ferromagnetic alloys. They cannot create or eliminate magnetic fields, only redirect them. You can't "block" the attraction of two magnets by placing such a shield between them. The magnets will then be attracted to the shield, experiencing forces in nearly the same direction as before.

If you place a strong magnet near one side of a magnetic shield (such as mu-metal), the region on the other side of the shield will not be field-free. In fact, a ferrous metal object on the "shielded" side will still be attracted toward the shield. Experimenter's note: Manufacturers of shield materials advise against placing strong magnets in contact with a magnetic shield, for this can induce a residual permanent magnetism in the shield. I place a 2 mm thick nylon washer between the magnet and shield.

When an inventor uses magnetic shields in an over-unity device proposal, he's imagining a kind of shield that does not exist. He assumes a "magic shield" that violates fundamental laws of physics. But most proposals I've seen still wouldn't work even if the magic shields did perform as the inventor hoped.

At least one company that sells magnetic shields warns buyers that these are primarily useful for radio frequency shielding. The company must have had quite a few inquiries from perpetual motion machine inventors. I can only imagine letters asking "Which of your many magnetic shielding materials is best for use in a perpetual motion machine?"

Antigravity shield
woven into a carpet.

The inexperienced inventor imagines that magnetic shields act on static magnetic field lines in the same way that an opaque object "blocks" (absorbs) light. They don't. Only for high frequency electromagnetic radiation does anything of that sort happen.

Before anyone asks, there's no such thing as a gravity shield. And magnetic adhesive patches do not relieve the pains of rheumatism, can not increase your gasoline mileage if you put them on the fuel line, nor ionize or detoxify water if placed on water pipes. Nor are all of those refrigerator magnets making the food inside the refrigerator any healthier to eat.

My device uses an innovative method to maintain continual overbalance of mass, force, and torque. But it still stubbornly sits there, unmoving, taunting me.

Much ingenuity has been wasted trying to design continually overbalanced devices—clever mechanisms that shift masses from one side of an axle to the other as the wheel turns. The idea is to continually keep more mass on one side of the axle. This can be done, and if you turn such a wheel by hand there's always more mass on one side. But the wheel never turns continually on its own. Why? The work required to shift masses from one side to the other is always at least as great as the work those masses will provide due to the overbalance. Such devices may turn just part of a revolution then settle down to an equilibrium position and stubbornly sit there at rest and in perfect equilibrium, even though they are in an apparently unbalanced condition. If examined carefully it is seen that the forces and torques within the structure of the device are, thanks to Newton's third law, perfectly in balance in force and torque equilibrium.

In any wheel-type device, each mass must complete a closed path. The work gained over part of the path as a mass falls is equal to the work required to raise it up again. To try to get around this fact of nature is as futile as finding a round trip walking path that is downhill all the way, in either direction.

The closest you can come to perpetual motion is a simple flywheel with frictionless bearings. It would turn forever, but produce no output of useful work. Any "improvements" you make on this, no matter how ingenious, using gears, shifting weights, magnets, fluids, quantum mechanics, etc., only reduces its performance.

Can I improve the performance of my over-unity wheel or belt device by making it larger?

Making the chain longer
and adding more weights won't help.

Larger is not always better. Making such a mechanism larger may seem a good way to increase the "overbalance" that you suppose will make it work, but increasing the size also increases the net load that needs to be moved and/or the distance it must be moved, and these factors are in direct proportion. So save money and make a small model. It will fail for much lower cost than the larger one. Some inventors have even built wheels as large as a carnival Ferris wheel. They only turn when the wind is blowing.

Some have modified an overbalanced wheel device into a belt device, then supposed that the overbalance could be made greater by making the belt longer. It can, but that also adds more mass that must lifted a greater distance up the other side of the belt. Nature has gotcha again.

My wheel with moving masses doesn't work. Would it work if I add more moving masses or make them heavier?

No. In fact, you can test your idea with fewer masses for less cost. See previous answer.

My wheel has many identical moving parts to achieve continual overbalance. I can't afford the expense of building a model.

First, even if you do achieve continual overbalance in all positions of the wheel, that won't initiate or sustain motion. Look at the center of mass of the sum of all those parts. If the center of mass never rises above the wheel's axle, it's a non-starter.

Second, consider testing just one of those moving parts in a simpler design, perhaps with a pendulum arrangement. It still won't achieve perpetual motion, but you might learn some physics hands-on.

I did a computer simulation of my idea and it works beautifully. But when I build it, the darn thing just stubbornly sits there at rest.

I've heard this story many times. Computer simulations are only as good as the information fed into them. GIGO (garbage in, garbage out.) So if the simulation shows a working machine, you know you've given the program incomplete or incorrect information. But even the best such simulation program, with perfect data input, uses known, reliable and well-tested physics laws, so it couldn't produce results that violate those laws, could it? All perpetual motion and over-unity devices must violate physics laws. So why do inventors even bother with such computer simulations?

Similar caveats apply to folks who say "my idea works fine on paper." Yes, and visual illusions of impossible objects also look fine on paper.

Too complicated to understand.

My design has gone through many changes and improvements. It's now so complicated that I don't understand how it works. But I'm certain that it will work. Can you help me?

If you don't understand it, how can you be so certain that it will work? No, I can't help you. A wise colleague used to say, "The perpetual motion machine inventor concocts a device so complicated that he can't see any reason why it wouldn't work. So therefore he assumes it must work."

My magnet wheel won't turn. Should I buy stronger magnets?

Strong magnets can be obtained at small cost, so buy some to experiment with and learn how magnetism works. Be careful, though, for some are strong enough to injure you if your finger gets pinched between them. And keep them out of the hands of small children, who might swallow them. You will soon learn that even with larger magnets your magnet wheel won't turn around even once by itself.

My wheel-type device doesn't turn by itself even once around. Will reducing friction help?

Reducing friction won't fix an unworkable device.
© Raymond James.

Friction is never the sole reason for the failure of a supposed over-unity device. Even if you could remove friction entirely it wouldn't work. Look for the real reason for its failure. You can be sure the reason isn't friction.

Likewise, viscosity is never the sole reason for failure of perpetual motion and over-unity devices using liquids. Assume a machine that is free of friction, viscosity, and all other dissipative processes. Analysis will always show that it still can't work even if completely idealized.

Most machines depend on friction to work. Imagine a world without friction of any kind. You couldn't walk, vehicles could not move, belts would slide over pulleys, knots would come undone, structures would collapse.

Removing all friction may not even be a good idea.
Cartoon © 1987 by John Holden.

Looking at books and websites, I conclude that all the simple perpetual motion ideas have been tried, and all have failed. Can some of these ideas be modified, improved or combined to be successful?

Designer hubcaps don't improve the performance of square wheels.

That has been tried, too. Any clever improvement or ingenious mechanical gimmick increases a device's mechanical complexity and degrades performance. The closest you'll ever come to a perpetually turning wheel is a simple flywheel with frictionless bearings. Any "improvements" you add will bring it to a stop sooner.

I want to tap energy from natural sources. Which would be the best source, gravity or magnetism?

Neither. These are not sources of energy. They are natural forces important to the operation of many machines, but no working cyclic machine has ever extracted any energy from gravity or from magnetism. All the machinery of mankind has not diminished the strength of the earth's gravitational field by even a smidgen. If you want sources of energy from nature, try something that moves, like wind, tides, or falling water. Or something that can be burned, like coal or oil. Or something that varies in temperature naturally. Or something that actually emits energetic particles, like the sun, or radioactive ore.

But I hear someone object. "When I ski down a snowy mountain slope I gain kinetic energy. Doesn't that come from gravity?

When we are talking about perpetual motion we are dealing with cyclic machines, devices that complete a closed cycle of operation, indefinitely. When you go to the ski resort the potential energy you have at the top of the ski run came from walking up the mountain, driving up in an automobile, or taking a ski lift. That potential energy is what gives you kinetic energy as you ski down the slope. Gravity was not the source of that energy; it was an intermediary agent.

But couldn't gravitational fields provide an unlimited source of energy? All our machinery operates in a gravity field, and some depend on it, yet the gravity is not diminished at all by this.

Force fields are a mathematical way of describing what will happen when things are placed in the field and move in that field. They are not sources of continual energy. No one has ever extracted energy from a gravitational field. The gravitational field of the earth acts downward toward the earth's center. Always downward. You never will see a stone rise up from rest by itself. Any cyclic motion of a body in this field will show that the body gains kinetic energy when falling, and loses the same amount of energy when rising the same distance. The net gain or loss of its energy is zero during each cycle. Even an earth satellite orbiting the earth without power in an eccentric orbit shows the same thing, gaining speed as it approaches earth and losing speed as it moves back away.

Someone may bring up waterwheels. Isn't that a cyclic motion dependent on gravity? It is indeed, but it isn't a cyclic process and it doesn't extract energy from gravity. The energy comes from water flowing from higher to lower elevation. The water then flows down streams to lakes or oceans, where radiant energy from the sun evaporates some of it and atmospheric circulation (also sun-driven) moves it elsewhere and dumps it as rain. Some of that rain falls at higher land elevations, forming streams which power waterwheels, and so on. It is a cyclic process, but not a closed one. It requires energy input from the sun. And gravity, though necessary to the process, is not a source of energy. The energy came from the sun.

The fact that gravity is not diminished by all of our machinery, space satellites, etc. should tell you that all of these processes aren't stealing any energy from gravity. Now some things may steal a bit of energy from the rotating earth (they'd have to be pretty massive events), slowing it slightly. But that doesn't come from the earth's gravity and it doesn't diminish the earth's gravitational strength.

But don't magnets have unlimited stored energy? A refrigerator magnet will support itself on the wall of the refrigerator forever, continually exerting force against gravity to keep itself from falling. So isn't it capable of unlimited work?

So I suppose the nail driven into the wall is also doing unlimited work supporting the picture frame hanging from it? I have heard the "refrigerator magnet" example from many people over the years, and find it incredible that they can so confidently make this absurd claim without even thinking of obvious counter-examples.

Force and work are different things. Work requires motion. A force that produces no motion does no work, and consumes no energy.

Some magnet motor and magnet engine proposals have continually moving magnets. Can't these extract energy stored in the magnets?

Permanent magnets are used in motors and generators worldwide, and none of these machines ever extracts any energy from their magnets. The magnets merely facilitate the conversion of mechanical to electrical energy or vice versa. After many years of operation, the permanent magnets in these devices still retain their original magnetic properties.

The stored energy in a magnet is only that due to the magnet's manufacturing process. It is a small amount. In normal use, the internal stored energy of a magnet is not used or diminished at all. Heating or hammering the magnet can, however, destroy its internal domain alignments, and therefore, its magnetic effect.

Besides, if the magnet did "contain" such a tremendous amount of energy, it must have required at least that much energy to manufacture it, and magnets would be far more expensive.

It's irrelevant, but interesting, to consider just how much energy is stored in a small experimenter's magnet. That information isn't easy to find on the web. I was astounded at how small it is, and asked Rick Hoadley to do an independent calculation, which agreed with mine.

Rick considered an Alnico-5 magnet, in the form of a bar magnet is 6” x 1” x ¼” in size, a volume of about 2.5 × 10-5 cubic meters. The energy stored in an Alnico-5 magnet bar of that size is 1.2 Joules. [If you used a stronger magnet, say an NdFeB magnet bar, it would store about 14.7 Joules.]

Quoting Rick:

1.2 Watts for 1 second is = 1.2 Joules = 1200 Watts for 1 ms. A typical hair dryer uses 1200 Watts while it is running. If there is an easy way to take the energy out of a permanent magnet with 100% efficiency, it could run the hair dryer for 1 ms. That’s all the energy in the magnet, 1.2 Joules.

If, however, you had a similarly sized NdFeB magnet, it could run the same hair dryer for almost 13 ms! Wow, one hair might get dry!

Thanks to Rick Hoadley, The Magnet Man, rhoadley@execpc.com for the calculations.

So anyone supposing they could "extract" considerable energy from magnets to solve the energy crisis had better rethink the matter.

Is centrifugal force a good energy source?

Centrifugal force is a widely misunderstood concept, often badly presented in physics courses. "Centrifugal" means "fleeing outward". It is not some exotic kind of force found in nature. It is nothing more than a convenient mathematical concept used when physicists and engineers do analysis of rotating systems using non-inertial rotating coordinate systems. Forces are never sources of energy. Forces occur when bodies interact, and, if motion of either body occurs, that interaction may result in one body losing energy and the other gaining an equal amount of energy. No energy is ever created from a force.

How about converting momentum to energy?

Momentum and energy are two different concepts, and are not convertible one to the other. They have different physical dimensions and units. Mathematically, momentum is a vector and energy is a scalar. Energy is conserved in every closed system we have ever studied, and energy is neither created nor destroyed. Momentum is also conserved in such systems, and the two conservation laws represent independent facts about nature. In the early history of physics when these were not yet understood, there was much debate over which was the "better" or "proper" way to describe motion. This debate was settled in the 17th century, when we realized that both concepts are necessary to fully describe how mechanical things work and how bodies interact. Many physical problems simply cannot be solved using only one, but not the other, of these concepts. Both concepts must be used simultaneously.

Could we convert angular momentum to linear momentum, or vice versa?

Some have tried to convert rotational momentum to linear momentum. The Dean Drive was one such example. Norman Dean was taken in by a stick-slip friction phenomena that he didn't understand. His device, if it actually worked on the principle he claimed, would violate not only energy conservation but momentum conservation as well. Indeed, others still hold out hope of making such a third-law-violation device (sometimes called a "reactionless thruster"). But most inventors totally ignore momentum of all kinds because they simply don't know anything about it. They may not even realize that the conservation of momentum law is just as solidly established in physics as the conservation of energy law that they generally despise.

Rotational kinetic energy is just ordinary kinetic energy, since kinetic energy is a scalar and does not depend upon the direction of a body's motion or whether the path of a moving body is straight or curved. So there's nothing more to say about that.

Energy, angular momentum and linear momentum are all different beasts. They have separate conservation laws, different dimensions and units, and aren't convertible one to the other.

In your analysis of perpetual motion proposals you never include centripetal and centrifugal forces in the math. Isn't it possible that if you did include them, you could show that the idea could really work after all?

I have never seen a perpetual motion machine proposal where it was necessary to deal with centripetal or centrifugal forces in the analysis to conclusively show why the device wouldn't work. Nor have I ever seen a proposal where the inventor claimed his idea depended on them. But, rest assured, that if you did a full free-body force and torque analysis of the device, the outcome would be the same: the device won't work. To do that much analysis would be "using a sledgehammer to crack a walnut."

I'd like to build a prototype, but I don't have much money and don't have a machine shop.

Nearly all the devices people describe to me can be built from readily available materials with simple tools. Identify the feature of your device that is the reason you think it will work. Isolate that and build a prototype to test it. Suppose your device is a wheel. Most such perpetual wheel devices can be tested in the modified form of a pendulum, easily built with Erector or Meccano parts. Curiously, very few perpetual motion machine proposals are in the form of pendulums. See Building perpetual motion machines for examples.

If you are clever enough to invent such an original device, you should be clever enough to build an inexpensive prototype that would conclusively show whether it works as you expect.

Be aware that some people become so obsessed with an idea that they are blinded to all else. They spend money and time on a quest that leads nowhere but to failure. This is especially true if they choose to work in isolation, and never listen to reasonable and informed criticism of their ideas.

I've made a wheel with carefully positioned magnets, and it turns continually when I hold another magnet near it in just the right position. But when I clamp that same magnet in that same position, so I don't have to hold it steady, it doesn't work. Why?

Because when you hold the magnet in position, you are supplying the energy by doing physical work on the magnet you are holding. You really can't hold it steady, but are continually making small motions to try to keep it steady. That's what keeps the wheel turning. No, it's not psychic energy, or any of that sort of moonshine. It is a process of brain/muscle delayed feedback, sometimes called the ideomotor effect. When you bring the magnet near the wheel, it begins to turn, and this changes the position of the wheel's magnets and the force they exert on the magnet you are holding. You sense the motion this force causes and you try to compensate for it in order to keep your magnet in the same position. But there's a slight delay in your muscular response. The magnets' strengths, their spacing around the wheel, the wheel speed, the mass and strength of the magnet you are holding, as well as the delay time of your nervous system and muscular response—all of these determine the period of the small oscillation you impart to the magnet, and if all of these are just right, you can maintain rotation of the wheel. We frequently see such demonstrations on YouTube, and some people really think they are on the verge of creating a perpetual motion wheel. With just a little more refinement... See: Howard Johnson magnet motor.

This is often compared to the table turning or table tipping phenomena reported during séances in the heyday of Spiritualism. Gullible people sat around a table in a darkened room with their fingers pressing on a small table. They were instructed to try to prevent the table from moving. Sometimes the table moved, often vigorously. (Often with a little help from the spiritualist medium who also sat at that table.) Sensing slight motion, the sitters would try to prevent the motion, but because of the delay in their responses, they just caused a rocking periodic motion of the table.

Pendulum divination. The string is held by the fingers with the hand relaxed.One (of many) charts for use with pendulum divination.

This has been compared to the ancient "pendulum divination" game of holding a finger ring (or mystical-looking pendant) on a string suspended from your finger. It supposedly answers questions by its mode of swing. But there's a difference. In the magnet motors and the table turning the nervous and muscular system response time plays a crucial role. In the pendulum divination game the person holding the pendulum can subconsciously (or consciously) control the nature of the motion produced with very slight finger motion. The pendulum has several modes of motion, all with very nearly the same natural frequency. If its support isn't rigid it can switch slowly from one mode to another. Also, seeing a small deviation toward a change of mode, the person holding the string can subtly encourage or discourage that to make the ring "answer" whichever way is desired. See also Ouija board.

Model of the Hamel spinner.

A neat version of this perpetual motion deception uses a large steel ball bearing with a ring magnet placed on it, the whole thing resting on a very smooth table. Another magnet is held above, causing the ball bearing to move so that the ring magnet is near the top. The ball bearing may start to rotate slowly, then speed up, as you try to hold the magnet above it in the optimum position. To make this work the magnets' strength, ball bearing weight and ring magnet strength and weight must be balanced. When turning, the ball bearing/magnet assembly is in precarious equilibrium, and just a bit of tilt will change its contact point with the table. So the device is a delicate magnet-gyroscope. Perpetual motion machine scam artists have used this in public demonstrations of the "principle" of their motors. All of these work best if the natural rotation period of the physical system matches the natural period of the hand holding the magnet. This is sometimes called "parametric excitation by hand".

I've seen this called the Hamel Spinner. For a picture of this toy, see David Hamel spinning device. Don't fuss too much about the dimensions of the parts, so long as they are in proportion to the diagram. When I first built one I used a relatively weak 1.25" ceramic ring magnet on a 2" steel ball, and a very strong magnet above, which must be strong enough to lift and keep the ball-magnet assembly upright, but not so strong that it lifts it off the table. It worked well. But I was once careless and the ball was yanked up to the magnet, breaking the ring magnet. The upper magnet need not be a ring magnet.

Rodney Brian has done some experimentation on this device. See his video and review. He shows quite persuasively that (a) the device is not over-unity, (b) energy from the hands drives it, (c) the steel ball can be replaced by a glass or plastic ball, and finally (d) magnets aren't necessary. The key to the toy's behavior is (1) A round ball, weighted above, rotating about a slightly tilted rotation axis, and (2) slightly out of phase motion of the hand, supplying energy to sustain the toy's motion.

Click here to see this effect at work in the Minato motor. Watch Minato's hands "working". There's a simple test to see what is happening. Instead of holding the magnet in the hand, clamp it to a solid support. Then the wheel, once started, will spin for a long while (like a flywheel) but eventually slow to a stop. There are many videos of such devices on the web.

Might dark energy and dark matter provide unlimited sources of energy?

Dark energy and dark matter are speculative entities for which there is as yet no direct experimental evidence. They are hypotheses that seem to account for certain observations about the expansion rate of the universe. There are also competing hypotheses that do not invoke dark matter and dark energy, so the verdict on their "reality" is not in yet. The popular media love to ballyhoo such exotic ideas from speculative theoretical physics. Even textbooks pander to student interest in science fiction by including such ideas alongside established and tested physics, without clearly distinguishing speculation from established science. Even if such hypothetical concepts turn out to have reality equal to that of ordinary matter and energy, no one has the slightest idea whether they can be in any way "tapped" or "converted" to forms of energy that are capable of doing useful work, like running machinery, generating electricity, etc. And if they can, there's no clue how we'd go about doing it. So if any free-energy or over-unity-device huckster claims his device is really running on dark matter or dark energy, or zero point energy, or "etheric energy" you can be sure he's talking moonshine and mumbo-jumbo—and hang onto your wallet.

Suppose my machine actually produces more energy than it takes in. Might it be tapping some previously unknown energy source that's invisible all around us, one we hadn't previously detected?

If so, then, your machine would have detected that energy source. Let me know when you achieve that. Perhaps I'll hear about it when you are awarded the Nobel Prize. But first, have your measurements and calculations independently checked. You just might have made a blunder.

I think I see why gravity wheels don't work. And magnetic wheels, too. But what if I combine gravity and magnetism?

In this game, any clever combination of any number of unworkable systems is also guaranteed to be unworkable.

I understand that an over-unity machine would violate conservation of energy. But would it also violate conservation of momentum?

This question is largely overlooked by perpetual motion machine inventors. Yes, an over-unity machine would necessarily also violate conservation of momentum. And if it is in the form of a rotating wheel, it also violates conservation of angular momentum. Of course it would also violate all of Newton's three laws. A clean sweep of the foundations of classical physics! All of the laws that have so successfully formed the basis of our industrial society would be invalidated. Indeed, one should wonder what laws might remain that would even apply to the perpetual motion device.

When you turn on an over-unity wheel it increases its speed from zero. As it gains speed it gains rotational energy, and angular momentum, all without any energy or momentum input except the slight force of flipping the "on" switch. If it drives a vehicle, it increases linear momentum as well.

An interesting subset of unworkable devices includes those that supposedly work by creating momentum. They are called "reactionless drives", "reactionless thrusters" "fuel-less thrusters" or "internal propulsion engines". Harry Bull (1932) gained media attention with a clever demonstration with a pendulum, two masses and a spring. He claimed it moved the center of mass without any external force acting on the system. Norman Dean (1961) got a lot of press in Popular Mechanics magazine with his "Dean Drive". He was misled by the "stiction" (stick and slip) property of friction. Other such devices include Robert L. Cook's inertial propulsion system US Patent 4,238,968, and James Woodward's theoretical proposal of a reactionless propulsion system, US patent 5,289,864 and US patent 6,347,766. See Dean Drive for descriptions of the first two of these.

It is no surprise that all of these turned out to be misinterpretations of the experiments and perversions of physics to justify those misinterpretations. In all cases the devices obey classical physics, and nothing unusual is happening.

Doesn't the patent office refuse to patent perpetual motion devices?

I keep hearing this alleged "fact" repeatedly asserted by folks who pretend to be knowledgeable without bothering to check their facts. It is simply not true. Every year patent offices around the world issue patents for unworkable and useless devices. True, most inventors avoid using the words "perpetual motion" or "over unity" in their patents, but frequently they do say things like "energy efficiency of 125%", or, my favorite, "highly efficient unlimited source of energy", which amounts to the same thing. Patent offices say that a patentable device must be "new or original" (not previously patented), and also "useful", but judging by the patents they actually grant, they don't follow that policy scrupulously. I have even seen patents from the European Patent Office that give lists of patents for "similar devices", and sometimes even references to unworkable devices described in the book "Perpetual Motion, History of an Obsession" by Ord-Hume. The patent examiners are fully aware of what's going on, and simply don't care.

For examples, see this sampling of patents for unworkable devices. Certain devices, like the gear and lever, and the mechanisms of Archimedes (see above) aren't patentable because they are so old and have been in common use so long.

Why not apply simple logic. A perpetual motion machine would run forever. An over unity machine would do better than that, putting out excess energy forever. That would be an infinite amount of energy over its lifetime. And there's nothing infinite in the universe, so this is an impossible thing.

I stand in awe of such logic.

What are the most important physics principles that a perpetual motion or over-unity inventor should know, but often doesn't?


  1. A force does no work on a body unless it moves that body in the direction of the force. The work done on a body is W = F•d, the scalar product of the force and displacement vectors. It is of size Fd cosθ where θ is the angle between the force and displacement. So a force acting perpendicular to the direction of a body's motion does no work on that body.
  2. Newton's laws and how to use them in real-world situations. Especially one needs to be able to do vector algebra properly in order to do free-body force analysis of systems. A good course in elementary physics covers that material, but many students don't ever really understand it well enough to use it properly.
  3. If you think some force or torque drives your system, look carefully for reaction forces and counter-torques. You might easily overlook them in your enthusiasm.
  4. Nature abhors over-unity devices. All the laws of nature conspire to make such devices impossible.
  5. "Overbalanced" wheels can easily be designed that have continual imbalance of weight, force, or torque. But these don't initiate motion.
  6. If a mechanical system moves through a closed cycle, and the final and initial states of the system and all its components are indistinguishable, then the wheel won't initiate motion. (Simon Stevin's Principle.)
  7. If the center of mass of a wheel is always below its rotation axis in any possible position of the wheel during a cycle, then the wheel won't initiate or sustain cyclic motion.
  8. If a wheel works equally well when pushed in either direction, it won't initiate continual motion by itself.

If perpetual motion and over-unity machines aren't possible, why are so many people on web forums talking about ways to do it?

Because they haven't a clue how to accomplish it. If they had any useful ideas, the machines would have been already built and independently tested, and there would be no need for more idle speculation and empty talk. They should cut out the talk and start tinkering with their ideas hands-on. Then they might learn something about how nature works and how it doesn't.

Don't be so negative. Isn't anything possible if you are clever enough?

No. If you need an example, try to draw a circle whose circumference is only 3 times its diameter. The geometry of our universe makes many things impossible, and physics laws are all constrained by that geometry. You can't make a plane triangle with equal sides but unequal angles. You can't design a walking path in a closed loop that is downhill all the way, in either direction. Nature has many impossibilities. The laws of physics that we have discovered tell us what nature can, and can't do..

How can you be so certain that conservation laws and Newton's laws are inviolate?

A good question, one that may get at the heart of the perpetual motion inventor's philosophy and motivation. First, let's be clear that any laws we write about nature should not be understood to be some absolute truth chiseled in stone. All physics laws arise from observations of what nature does, and how it behaves when observed with our best measuring tools. From this mass of data we look for regular and reliable patterns that we call basic principles and laws. We especially treasure those principles we discover that apply to a wide variety of phenomena, and are logically linked to other such principles. Some of these are so reliable that we have never seen exceptions to them, no matter how cleverly we devise experiments to test them. Some are found to be the same here on earth as well as elsewhere in the universe, based on the evidence reaching us from as far as our telescopes can see. We often call these laws "universal", meaning we have no evidence of any exceptions to these laws anywhere in the observed universe.

Now might there be unobserved places where such laws do not apply? We can't deny the possibility. But science doesn't deal in notions that are unobservable, and especially does not deal in imagined notions that have no known connection to what we can observe. This doesn't mean we reject such possibilities, but that we have no way to observe them or their imagined effects, so scientific confirmation is (at present) futile, and speculation about them is idle daydreaming.

For this reason, if someone tells us that there might be somewhere in the universe where gravity is proportional to 1/r3 we don't drop everything we are doing to investigate that possibility. We ask some skeptical questions first:

  1. What experimental evidence suggests this idea?
  2. What other reasoning supports it?
  3. What experiment could be devised to measure it or its consequences?
  4. How does it mathematically relate to other established and tested laws?
  5. If this idea were true, how would established laws have to be modified to maintain the mathematical consistency of physics? How could we test that?
Established physics is "accepted" because it works, and no exception has been found. This is not a "belief", but a "provisional acceptance" based on the overwhelming evidence of certain underlying regularities of nature's behavior. It would be perverse to deny that fact. Should we be looking for evidence to deny it? Well, we should certainly not blind ourselves to such evidence if it were to appear, and certainly not try to deny it or wish it away. But there are simply more productive things to do in science than to devote a lifetime to searching for something for which there's not the slightest shred of evidence, and no guidance where and how to look. There's an old joke about the philosopher who goes down into dark coal-bin without a light on a moonless night to search for a black cat that isn't there.

While there's a core of physics that is rock solid, critics often fail to distinguish that from the less certain areas of physics understanding. Certainly our ability to predict next week's weather is poor. This is because weather systems are complex, data is spotty and often poor, and there are just too many interacting variables to deal with. But as we improve our understanding of these processes, will any of this constitute "new physics"? No. The underlying physical laws will remain unchanged. Newton's laws will be untouched by any advances in weather forecasting.

It will take an infinite number of revolutions to achieve perpetual motion of a wheel.

Advances, even revolutions, in physics have been made, such as atomic theory, relativity, and quantum mechanics. And still, the laws of conservation of momentum, angular momentum and conservation of energy still apply as firmly as ever.

The very notion of someone undertaking research to disprove Newton's laws seems absurd.

This example may seem trivial and obvious. Other examples could be cited: stellar evolution, elementary particle theory, quantum mechanics, medicine, nanotechnology. Whatever advances are made in these areas, the classical mechanics that we see operating in everyday life will remain untouched. Whenever a perpetual motion inventor presents us with a mechanical device, we know it will not violate classical physics laws, and if he claims it has 200% energy efficiency we will know he made a blunder in measurement or calculation. But if this machine has a "black box" in it that supposedly operates on "quantum principles" then the sensible thing is to independently test it. Testing methods are simple. See Testing Perpetual Motion Machines. If the test shows its energy efficiency is less than one, we conclude that those claimed quantum principles, or other magical processes, operating in that black box aren't doing anything remarkable or useful.

If subatomic particles and processes do strange things (quantum weirdness) at small scale and small time intervals couldn't this be used to build devices that do similar strange things at the macroscopic level, even to violate macroscopic laws?

This is a tempting thought. Even those who understand this stuff haven't a clue how to bring that about. In fact, the more we learn about it, the more it seems that quantum weirdness behavior stays at the quantum level. Now the very fact that we can do experiments on it is evidence that quantum mechanics does communicate information to the macroscopic world (else we wouldn't know about it). But that doesn't translate to machines such as perpetual motion machines, over-unity machines, gravity shields, reactionless thrusters, instantaneous matter transporters, time machines, or double-slit diffraction of baseballs. Electrons and photons are not the same as baseballs and do not behave in the same way, even though we too often carelessly think of them as if their behaviors were analogous.

Input and suggestions are welcome at the email address shown here. When responding to anything on these web pages please indicate the specific document by subject, name or filename.

    © Donald E. Simanek, Feb, 2010, revised, April 2016.

Return to the Museum of Unworkable Devices.
Return to the main menu.