Marc Millis, former head of NASA’s Breakthrough Propulsion Physics project, recently returned from another trip to Germany, where he worked with Martin Tajmar’s SpaceDrive project at Germany’s Technische Universität Dresden. Recent coverage of the ongoing experimental work into spacedrives in both the popular and scientific press has raised public interest, leading Millis to explain in today’s essay why and how the techniques for studying these matters are improving, and how far we have to go before we have something definitive. Millis is in the midst of developing an interstellar propulsion study from a NASA grant even as he continues to examine advanced propulsion concepts and the methodologies with which to approach them.

by Marc Millis

Two recent articles, one in Scientific American [1] and the other in Acta Astronautica [2], prompted this update about the experimental tests of possible spacedrives. In short, the experimental methods are improving, but definitive results are not yet in hand. While this update is mostly on the “Mach Effect Thruster,” it also touches on the infamous “EmDrive,” as well as a refresher on the general quest for spacedrive physics.

First, what is a spacedrive? Presently, a spacedrive is still a goal rather than a proven device. The ambition is to find a fundamentally different way to propel spacecraft rather than rockets or sails. Rockets are limited by having to carry their entire journey’s reaction mass with them (propellant). Sails are limited by one-directional photons (or particles) from an external source. Imagine, instead, if there was some way for a spacecraft to interact with its surrounding spacetime to move in any direction and be limited only by the amount of available energy. That ambition is the essence of a spacedrive.

That detail – of interacting with spacetime to induce motion – is a matter of undiscovered physics. That makes it harder to grasp, harder to explain, and harder to solve. It’s easier to grasp engineering challenges that are based on known physics, since there are already operating principles to cite. With spacedrives, the operating principles are works-in-progress – more akin to lines of inquiry than having complete packages ready for scrutiny. Though theories for faster-than-light warp drives do exist (one type of spacedrive), the physics of the required negative energy is still debated – which itself is a prerequisite to devising how to engineer a warp drive. In addition, though there are experimental replications of thrusts from possible spacedrives, separating experimental artifacts from actual thrusts is also, still, a work in progress – and the main point of this update.

Before getting to the latest experiments, here is a bit more background behind the challenges of a spacedrive. At first blush, such wishful thinking might seem to violate conservation of momentum – a crucial detail. Conservation of momentum is easy to grasp for a rocket; the rearward-blasted propellant matches the forward momentum of the spacecraft. The situation is less obvious with spacedrives. There are a least 3 approaches to address conservation of momentum: 1) using a reaction mass indigenous to space or spacetime, 2) negative inertia, or 3) exploring the physics about inertial reference frames – the backdrop upon which the conservation laws are defined.

The majority of this update is related to the 3rd option – inertial frames. For new readers, a more complete introduction to various approaches and issues of both spacedrives and faster-than-light flight are spelled out in the book Frontiers of Propulsion Science [3]. If you’re curious about that broader coverage, that book and subsequent papers are one starting point.

Back to inertial frames and conservation laws: An inertial frame is such a ubiquitous property of spacetime that it is often taken for granted. It is what allows accelerated motion to be felt – the reference frame for Newton’s F=ma and the subsequent conservation laws. If you’ve never thought about it before, this can be hard to grasp because it’s so foundational. One useful book is Mach’s Principle: From Newton’s Bucket to Quantum Gravity [4], which articulates several different attempts to represent how inertial frames exist. What makes this book particularly useful is that it compiled workshop discussions about the differing approaches. Those discussions are illuminating.

One of those attempts is called “Mach’s Principle,” which asserts that the surrounding matter of the universe gives rise to the inertial frame properties of space. Or stated differently, “inertial here, because of matter, out there.” A similar perspective is something called “inertial induction.” The implication of these is that inertia is more than just a property of mass. Inertia is an interaction between mass and spacetime – and perhaps with undiscovered nuances.

Perhaps an analogy might help. When you plot trajectories on graph paper, you usually don’t give much thought to the paper. The paper is just some fixed, reliable background upon which the more interesting details are plotted. But what if the paper was not uniform nor constant over time? What if the trajectories might vary because of the properties of the paper itself? In this case, the rules for plotting on graph paper would have to be updated to account for the rules about the paper itself. Here, the graph paper is analogous to an inertial frame and “plotting trajectories” is analogous to Newton’s F=ma and subsequent conservation laws. If there are deeper details about inertial frames and their effect on inertia, then Newton’s F=ma and the conservation laws would have to be refined to incorporate those finer details.

In terms of Einstein’s general relativity – an established refinement of Newton’s laws – inertial frames and momentum conservation are treated only locally. I’m not sure quite how to put this in words, so I’ll defer to examples. With the warp drive, Einstein’s equations describe the local effects on spacetime from the warp drive itself, but cannot describe how (or if) momentum is conserved across a whole journey, encompassing the departure and arrival points as a total picture. Similarly, the momentum conservation of traveling through a wormhole cannot be described. While the local effects at each throat can be described, the bigger picture encompassing both the entry and exit throats and the mass that went through, cannot. There is room for more advances in physics.

Mach’s Principle and Inertial Induction are still open investigations in general physics, though not a dominant theme. Their relevance to spacedrives is because Mach’s Principle was a starting point for what is now called the “Mach Effect Thruster.” It began around 1990, when a reexamination of Mach’s Principle led to new hypotheses about fluctuating inertia, which then led to a 1994 patent for a propulsion concept [5]. Experiments followed. By 2016, three other labs were observing similar thrusts, which led NASA to award a 2017 NIAC grant for further investigations.

The original theory, from James Woodward of the University of California at Fullerton, showed that the inertia of a mass would fluctuate with a change of power of that mass. At first, varying the power of the mass took the form of charging and discharging a capacitor – where the capacitor was that mass. By doing this with two capacitors, while also changing the distance between them (via a piezoelectric actuator), a propulsive force was claimed to be generated (see figure and caption).

Figure 1. Transient inertia applied for propulsion: While the rear capacitor’s inertia is higher and the forward capacitor lower, the piezoelectric separator is extended. The front capacitor moves forward more than the rear one moves rearward. Then, while the rear capacitor’s inertia is lower and the forward capacitor higher, the piezoelectric separator is contracted. The front capacitor moves backward less than the rear one moves forward. Repeating this cycle shifts the center of mass of the system forward.

Since the center of mass of such a system moves without the opposite motion of a reaction mass, it appears to violate conservation of momentum, but does it? Since inertia is no longer constant, the usual equations do not fit without some reconsideration. This is a debated issue – debated in a constructive way. One version asserts how momentum conservation is indeed satisfied [6]. Others would prefer that the original fluctuating inertia equation be further advanced to explicitly address the conservation laws. Another desired refinement is to have the original equations explicitly connected to the experimental hardware – to show what dimensions of that hardware are the most critical.

Armed with an apparently working device, Woodward and his team concentrated on improving the experiments rather than that additional theoretical work. Over the years of making modifications to the device to amplify the effect, the ‘fluctuating inertia’ capacitors and the piezoelectric actuator were merged. Now a stack of piezoelectric disks serves both the functions of the inertial fluctuations and the oscillating motion. The power that affects the inertia now includes the mechanical motion too.

This is where the Scientific American article is worth mentioning. That article gives a decent review of the history and status of the Mach Effect Thruster (which also goes by the name “Mach Effect Gravity Assist (MEGA) Device”) as conducted by Jim Woodward and Heidi Fearn. It includes some perspectives that are useful to read separately, instead of needing to repeat those here. It addresses other aspects of the bigger picture of pursuing these kinds of research inquires.

The other article that prompted this update is in the journal Acta Astronautica. In addition to Woodward’s team, a group at the Technical University of Dresden, Germany, led by Martin Tajmar, secured funding for a broader project to research spacedrives in 2017. That group is one of the 3 labs that replicated the Woodward results in 2016. The recent Acta Astronautica article is an update on their experimental hardware and procedures, in preparation for careful testing of the Mach Effect Thruster, the EmDrive, and other possible spacedrive effects.

A preceding work by Tajmar that fed into this latest update was an attempt to advance Woodward’s original fluctuating inertia equations into a form that mapped to the experimental hardware [7]. With such equations a new thruster could be designed to maximize the thrust and experimental predictions could be made for the existing hardware. To span the possibility of debated assumptions (such as what kind of power affects the inertia; mechanical, electrical, other?), more than one version of such equations was derived for future tests.

Though this paper is more about the testing methods, in the course of that preparatory work, it became evident that none of the analytical models match the data. The models predicted correlations between the thrust and operating frequencies that was not observed. If the Mach Effect Thruster is indeed working, it is not producing thrust per these models derived from the original theory. Hence, that thruster is now considered a “black box” – a term used to denote a device whose operating principles are unknown, and where the test program concentrates on seeing if, and under what circumstances, it functions.

To test the thrusters, they are placed on the end of a torsion beam that can twist horizontally (vertical axis). The term “torsion” means that the beam is sprung, its rotation is limited and proportional to how much thrust occurs at the tip. This is the same concept as the Cavendish balance that measured Newton’s gravitational constant. When the thruster is pointed one way, the beam deflects one direction. When pointed in the other direction, the beam deflects in the other direction. And the third important orientation is when the thruster is pointed in a direction where it should not deflect the beam. By comparing the actual deflections in each direction (and under different operating conditions), the performance of the thruster can be assessed.

Deciphering actual thrust from all the other things that can look like thrust is difficult. A major clue for a false positive is if the beam is deflected when the thruster is not pointed in a thrusting direction. Another major clue is revealed when the power is delivered to a dummy device instead of to the thruster – to see if simply delivering the power through the apparatus affects the apparatus. Another possible effect is from the peculiarities of the balance beam itself while powered up (e.g. thermal drift of the electronics). When testing the thruster in a thrusting direction, there might be slight shifts in the center of mass as the thruster warms up – where that thermal effect might look like thrust. And then there is the challenge of how vibration might shift the position of the balance beam. There are more possible side-effects than these, but these are the major ones.

Another false positive that merits separate mention is confirmation bias. Confirmation bias is not an instrumentation phenomenon, but a psychological phenomenon. After people reach a conclusion, they tend to filter evidence to fit their preconceived notion, rather than letting the data speak for itself. It happens way more often than it should. It is so insidious that we seldom know when we are guilty of it ourselves. Our bias skews, well, our bias. The important lesson here, for you the audience, is how to spot those biases when you come across new articles. If an article sounds like they it’s trying to prove or disprove, rather than decipher and conclude, then its findings are likely skewed.

The Acta Astronautica article comes across like an investigation in progress, rather than a conclusion in search of evidence (or advocacy). The article outlines the performance limits of their hardware and the procedures used to distinguish the aforementioned side-effects from potential genuine thrust. To measure a claimed thrust of 2 µN, the thrust stand has demonstrated a sensitivity of 0.1 µN, as well as plots of the background noise showing less than ± 0.02 µN. The procedures include calibration with known forces before and after each run, measuring the thermal drift of the electronics, and automated operation that repeats a set of runs 140 times to get ample data to average. The tests are conducted in vacuum and the thrusting directions can be changed during a test sequence remotely without having to break vacuum or risk affecting other configuration settings.

Other than the aforementioned conclusion that the Mach Effect Thruster is not following analytical models, there are no other conclusions to report. Sample data is shown for the Mach Effect Thruster (more than one version) and the EmDrive, but only to illustrate the measurements that can be made, rather than any attempt to report on the viability of either of those thrusters.

In closing

Conferences are coming up where more progress will be reported. Consider this article a preparation for interpreting these next series of papers. Carl Sagan’s adage, “Extraordinary claims require extraordinary evidence,” is exactly the tactic here. The results only have as much substance as the fidelity of the tests. This most recent progress bodes well for that fidelity. The prior tactic of “quick and cheap” experiments to test other claimed devices turned out to be neither quick nor cheap. Promotional material and sensationalistic articles are easy to create. Reliable findings are harder, less glamorous, and take longer.

The implications of a genuine new propulsion method, plus the independent replications, are driving the perseverance to wade through these complications. If it turns out that a new propulsion method is discovered, then not only will we have a more effective way to propel spacecraft, but also a new window into the lingering mysteries of physics. The less obvious value is if it turns out to be a false positive. In that case the years-long ambiguity will be resolved, and the lessons learned will make it easier to assess future claims of new thrusters.


1. Scoles, S. (2019). The Good Kind of Crazy. Scientific American, 321, 59-65.

2. Kößling, M., Monette, M., Weikert, M., & Tajmar, M. (2019). The SpaceDrive project-Thrust balance development and new measurements of the Mach-Effect and EMDrive Thrusters. Acta Astronautica, 161, 139-152.

3. Millis, M. G., & Davis, E. W. (Eds.). (2009). Frontiers of propulsion science. American Institute of Aeronautics and Astronautics.

4. Barbour, J. B., & Pfister, H. (Eds.). (1995). Mach’s principle: from Newton’s bucket to quantum gravity (Vol. 6). Springer Science & Business Media.

5. Woodward, J. F. (1994). U.S. Patent No. 5,280,864. Washington, DC: U.S. Patent and Trademark Office.

6. Wanser, K. H. (2013). Center of mass acceleration of an isolated system of two particles with time variable masses interacting with each other via Newton’s third law internal forces: Mach effect thrust. J. of Space Exploration, 2(2).

7. Tajmar, M. (2017). Mach-Effect thruster model. Acta Astronautica, 141, 8-16.