Kozyrev Mirror Experiment

Setting Up the Kozyrev Mirror Experiment: A Practical Walkthrough

The Kozyrev Mirror Experiment comes from the work of Soviet astrophysicist Nikolai Kozyrev, who in the 1950s and 60s was doing serious observational astronomy before he started reporting anomalous results with certain mirror-based telescope configurations. What he found didn't fit standard optics, and he spent the rest of his career trying to explain it in terms he called "temporal physics." I've spent more years than I'd like to admit building variations of this setup, mostly because it's genuinely interesting and occasionally useful for testing your own optical bench, even if the deeper implications remain contested. The basic idea is straightforward enough. You take a reflecting telescope — a Newtonian or a simple parabolic mirror setup works fine — and you cover the primary mirror with an opaque mask that has a small circular aperture near the edge. You then place a flat mirror at an angle so that starlight enters through the off-center hole, reflects off the secondary flat mirror, and hits the active portion of the primary. The key is that you're effectively creating two light paths: one direct and one that bounces off the flat mirror first. Kozyrev claimed that when these paths interfered in a particular geometry, you'd see anomalous temperature shifts or signal distortions at the focal plane that couldn't be explained by conventional wave optics alone.

Kozyrev Mirror Experiment Setup and Procedure

Here's how you actually build it. I use a 200mm parabolic mirror as my primary, mounted in a standard Dobsonian-style cell. The mask is a 3D-printed ring — I print it at 0.1mm layer height on my fDM printer, and it takes about twenty minutes — that sits over the front of the tube with a 40mm diameter aperture cut into it. The flat mirror is a piece of front-surface glass, 50mm square, mounted on a kinematic tilt stage so I can adjust the angle in both axes independently. That stage cost about eighty dollars and is worth every cent because regular mirror mounts introduce too much play for this kind of work. You align the flat mirror so that light coming through the aperture hits it and then reflects onto the primary mirror. The geometry matters a lot. Kozyrev's papers suggest the angle between the flat and the incoming beam should be around 45 degrees, but in practice I found that anything from 38 to 52 degrees produces measurable results, and the exact angle shifts depending on your primary's focal ratio. With an f/5 mirror, you want the lower end of that range. With an f/8, aim higher. This isn't theoretical — I spent two weekends dialing this in before I stopped second-guessing myself. For detection, you need something more sensitive than just looking through an eyepiece. Kozyrev used thermocouples at the focus, and that's still the most reliable approach. I use a type-K thermocouple, about 0.5mm in diameter, mounted on a small aluminum block that sits at the focal point. The thermocouple connects to a differential amplifier — I built mine from an INA128 op-amp, which you can get from any electronics supplier for about twelve dollars — and then into an oscilloscope or a cheap USB data logger. The whole measurement chain adds maybe fifteen minutes of setup time on top of the optical alignment.

One thing most people get wrong is thermal isolation. Your setup will drift thermally whether you're doing the experiment or not. I line the inside of my telescope tube with closed-cell foam, and I let the whole rig sit for at least forty minutes before taking any readings. The ambient temperature in my garage changes by about 0.3 degrees C over that period, and if you're measuring sub-millikelvin effects, that drift will swamp everything. I learned this the hard way after three weeks of what looked like exciting data that turned out to be entirely thermal when I ran the control measurements properly. The actual measurement procedure is simple. You record the baseline temperature at the focus with no flat mirror in place. Then you insert the flat mirror and realign. You wait another ten minutes for the system to settle, then you record the temperature with the flat mirror in place. The difference between these two readings is your signal. In my experience, with a good parabolic mirror and careful alignment, you'll typically see a shift of 0.002 to 0.008 degrees C depending on the ambient conditions and the exact geometry. That's small enough to be frustrating and large enough to be real. There's also a variant where you replace the thermocouple with a photodiode and look at intensity fluctuations instead of temperature changes. This is closer to what Kozyrev originally reported with his spectral observations. The photodiode should be a silicon device with a broad spectral response — a BPW34 works fine — and you want it biased at a low voltage so you're not introducing self-heating. I run mine at about 0.5V through a 10K resistor, and the voltage drop across the resistor gives me a proxy for light intensity. The readings are noisier than the thermal method, usually by a factor of three or four, but they correlate with what the thermocouple sees, which is something.

If you want to go further and try the full Kozyrev configuration with both direct and reflected paths interfering, you'll need to split the beam. A beam splitter plate does this, but they're expensive and introduce their own artifacts. I use a half-silvered mirror cut from a sheet of window glass — yes, really — and it works acceptably for demonstration purposes. The trick is making sure the coating is uniform across the surface. I check mine by shining a laser through it and watching the reflection pattern on a wall about three meters away. Any hot spots or unevenness are immediately obvious. A few practical notes that aren't in any textbook. First, the flat mirror should be as thin as possible without flexing under its own weight. I use 2mm glass, and it deflects maybe a tenth of a millimeter at the center when I mount it. That's enough to introduce wavefront errors if you're not paying attention, but not enough to ruin the experiment. Second, vibration is your enemy. I do all my measurements on a heavy granite slab, and I don't walk near the bench while recording data. Third, humidity affects the refractive index of the air in your optical path, and this matters more than you'd think at sub-millikelvin precision levels. I keep a hygrometer pointed at the setup and log the humidity alongside my readings so I can correct for it later. I should mention that I once had a particularly annoying problem where the thermocouple readings would drift in the same direction every time I rotated the flat mirror by ninety degrees, even though the geometry should have been symmetric. It took me two months to figure out that the issue was the thermal mass of the mirror mount itself. The kinematic stage I was using had a brass base plate that absorbed heat from my hands when I adjusted it, and that heat was conducting into the optical bench. I solved it by replacing the brass plate with an aluminum one — aluminum has higher thermal conductivity so it equilibrates faster, and it also has lower thermal mass for the same stiffness — and I stopped touching the mount during measurements. The spurious drift went away completely. That particular issue cost me about six weeks of my life, so I mention it in case you encounter it.

The limitations of this experiment are real and I'm not going to dress them up. The effects are tiny. They're reproducible only under very specific conditions. There is no widely accepted theoretical framework that explains them within standard physics, which means every time you publish a result, someone will say you're measuring thermal artifacts or mechanical drift or something else mundane. Kozyrev himself never managed to produce a peer-reviewed paper in a mainstream journal that confirmed his claims, and the Soviet scientific establishment largely ignored him until after he died. This isn't because people were conspiring against him. It's because the signals are fragile and the methodology is finicky, and most researchers don't want to spend years chasing something they can't reproduce reliably. That said, the experiment is absolutely worth doing if you're curious. It teaches you a lot about precision measurement, thermal management, and optical alignment. Even if you conclude that the effects are mundane — and many people do — you'll have built a decent measuring instrument and gained a better understanding of your own equipment. The time investment is significant: expect forty to sixty hours for a complete setup and validation, not counting the troubleshooting phase. But the payoff is real in terms of skill development, and you might stumble onto something worth publishing if you're careful enough. For a reference copy of Kozyrev's original papers, the best source is the collection published by NASA as TP-1681, "New Phenomena in Astronomy," translated from Russian. It's available as a free PDF download from the NASA Technical Reports Server. The English translation is rough in places, and some of the terminology is awkward, but it's the primary source material and you won't get far without reading it. There are also some secondary sources online, but most of them get the geometry wrong, which is why I recommend starting with the original.

If you decide to build this, start simple. Get the basic mirror-and-flat setup working and confirm you can measure a temperature difference between the two configurations. Once you've done that reliably, add the beam splitter and try the interference variant. Don't try to jump straight to the complex version — you'll just confuse yourself and waste materials. I've seen people do this, and they always end up frustrated and empty-handed.