Most of the heat was a camera setting
A 2014 report on Rossi’s E-Cat reactor presents 32 days of output 3.13 to 3.75 times the electrical input, read from a thermal camera. The camera was told an emissivity alumina has across all wavelengths at that heat, about 0.40; across most of the camera’s own band, alumina is nearly black. Re-read with the in-band value, the report’s own temperatures give about 1.2 times the input. The friendliest reading we could build gives 1.7.
The verdict
Nothing there
Re-read with the emissivity alumina has in the camera's own band, the report's temperatures give about 1.2 times the electrical input, not 3.1 to 3.7, short of the 1.5 the claim needed to survive. The friendliest reading, calibrating the camera on the report's own dummy run, gives 1.7: still half the claim, and on an emissivity the one measured spectrum contradicts.
How the verdict was set. The run’s RESULTS proposed nothing there, provisionally, and on 2026-10-12 Ben held the verdict open until alumina’s in-band emissivity at 1300 °C is pinned down. On 2026-10-18, at his instruction, the lab’s verdict engine replaced the hold: the excess and the measured-spectrum readings fail the 1.5 line, and the dummy-calibrated reading, 1.74, clears it as a robustness check. The in-band emissivity at 1300 °C is still unmeasured.
Reimplemented, the report’s own procedure reproduces its Table 7 on all 16 files: total output within 0.38 %, the ratio within 0.35 %.
With alumina’s in-band emissivity, the ratio falls to 1.19–1.25 at 0.95, 1.25–1.31 at 0.90 and 1.16–1.22 at 0.98. The body’s temperature drops from a reported 1260 and 1399 °C to 682 and 738 °C.
A simulated tube making no extra heat, read through the report’s procedure, prints a ratio of 2.00 and 2.14 for the two runs modelled. The report printed 3.13 and 3.59.
Give the simulated tube 1000 W of real extra heat and the corrected route recovers 2.07 against a true 2.29. The report’s method on the same tube prints 12.1.
The claim, as presented. The report presents 32 days of excess heat from Rossi's E-Cat reactor, with output about 3.2 to 3.6 times the electrical input, measured by thermal camera with a surface emissivity near 0.7 taken from the literature.
Levi, Foschi, Höistad, Pettersson, Tegnér and Essén (the Lugano report), “Observation of abundant heat production from a reactor device and of isotopic changes in the fuel”, October 2014 · the episode
The registry’s note on sources. Read from the source named here by a research subagent (2026-10-10); paraphrase, not quotation. No video was watched.
Re-read before the run, 2026-10-12. The report gives output 3.13 to 3.75 times the input over 32 days (its Table 7). The output was computed from surface temperatures read by an Optris PI 160 camera, which sees 7.5 to 13 µm, set to alumina’s total emissivity from a literature curve: about 0.40 at 1260 to 1400 °C. A dummy run without fuel, at about 450 °C, balanced in and out. The report’s isotope claims are outside this episode.
Preregistered in PREREG.md (commit b991d98) before any code; Addendum 1 (569a8cc) before any code; Addenda 2 and 3 fix two crashed runs that produced no numbers; Addenda 4 to 6 after the run, declared as such, each before the diagnostic it governs. Self-graded, with no outside review. Lab files are listed at the foot of the page.
Band cameras and emissivity
A thermal camera measures the radiance in its own band and turns it into a temperature using the emissivity it is told. Alumina is nearly black between 7.5 and 10.5 µm and somewhat less so at 11 to 13 µm. Its total emissivity at 1300 °C is low because most of the power at that heat is at short wavelengths, where alumina is partly transparent.
Tell a long-wave camera 0.40 when the surface is about 0.9 in its band, and it reads hundreds of degrees too hot. The report then multiplies that temperature to the fourth power by 0.40. At 450 °C the two emissivities are closer, so the dummy run is less affected.
The checks
Every check below was written down before the code. Each planted fault had to turn its check red once.
| Check | Result | Numbers |
|---|---|---|
| The report’s procedure on its dummy run (Table 3) | pass | 314.48 W vs 316.50; with rods 444.34 vs 446.36 |
| … on all 16 active files (Tables 6 and 7) | pass | total within 0.38 %, ratio within 0.35 %, every file |
| … with the emissivity curve shifted by +0.10 | red, as it should be | every file misses by 19 % or more |
| Camera model, told the true emissivity | pass | 5×10−13 K |
| … told 0.40 on a 0.95 surface at 900 K | red, as it should be | reads 1424 K, 524 K too hot |
| The claim: ratio with in-band emissivity above 1.5 | fails the claim | 1.19–1.25 at 0.95; 1.25–1.31 at 0.90; 1.16–1.22 at 0.98 |
| Energy audit of the simulated tube | pass | residual 1×10−16; final imbalance under 10−4 of input |
| … leaving out the end caps’ radiation | red, as it should be | 8.5×10−2 |
| Null: simulated tube read with the right emissivity | pass | 0.964 and 0.961, within 1 ± 0.15 |
| A plain tube through the report’s procedure | missed | 2.00 and 2.14 against the report’s 3.13 and 3.59; outside ±30 % and ±150 K |
| … at the dummy run’s input | missed | 1.72 (gate 0.85–1.15); camera 750 °C where the report read 445 |
| Positive control: 1000 W of real extra heat | pass | true 2.29, recovered 2.07 (−9 %); the plain tube recovers 0.97 |
| Finer grid and time step | pass | temperature 5×10−5, ratio 1×10−4 |
| Time to settle after a +103 W step | reported | 1450 s; the report’s Plot 5 suggests about 400 s |
The misses. The simulated plain tube runs hotter than the real reactor did: at the dummy’s input it reaches 492 °C, where the report’s camera saw 445 °C. So either the model loses heat less readily than the reactor, or the in-band emissivity is lower than 0.95. The null check passing rules out a plain mismatch between the model’s formulas and the report’s.
The open question: the dummy run
Diagnostics added after the run, declared as such, with no gate. The ratio for two active runs and for the dummy, by the emissivity the camera is assumed to see.
| Camera-band emissivity | File 1 | File 6 | Dummy |
|---|---|---|---|
| 0.40, the report’s own setting | 3.36 | 3.69 | 1.55 |
| 0.60 | 1.86 | 1.97 | 1.07 |
| 0.66, where the dummy balances | ≈ 1.7 | ≈ 1.75 | 1.00 |
| 0.70 | 1.57 | 1.61 | 0.95 |
| 0.74–0.75, where the active runs cross 1.5 | 1.50 | 1.50 | ≈ 0.88 |
| 0.80 | 1.39 | 1.41 | 0.86 |
| NASA TN D-3115 spectrum (band average 0.915) | 1.23 | 1.25 | 0.78 |
| 0.95 | 1.19 | 1.21 | 0.76 |
The dummy run does not balance under any measured in-band emissivity: the correction reads it at 0.76 to 0.79 of its input, where it should read 1.0. Either alumina’s in-band emissivity at 450 °C really is about 0.66, which the report’s own Figure 7 hints at, or the report’s convection and rod accounting reads 20 to 25 % low at that heat. This work cannot tell which.
If the emissivity stays near 0.66 up to 1300 °C, the excess is about 1.7 times: a large effect if true, and still half the claim. The one measured in-band spectrum the lab found, for flame-sprayed alumina at 880 K, says 0.92, a different sample at a cooler temperature. Settling it needs the in-band emissivity of the reactor’s own alumina at 1300 °C, or a calorimetric run.
- The report’s isotope analysis.
- The resistor coils, and the Inconel cables inside the rods.
- Whether the report’s total-emissivity curve is right. It is granted throughout.
Read: the Lugano report (pages 1–28, 2026-10-12) · Liebert, NASA TN D-3115 (1965), Fig. 2. Prior art: Valat, Goldwater, Greenyer, Higgins and Hunt, J. Condensed Matter Nucl. Sci. 21 (2016) 81–96, replicators sympathetic to the field, found 1.1–1.3 with an emissivity near 0.95; this re-analysis reproduces their range independently. No novelty claimed.