The CSV has one spectrum per power. I fitted the two peaks in each spectrum with Lorentzians and a local sloping baseline; the fitted heights and positions for every power are in [lorentzian_peak_fits.csv](<BENCH>/raw/commercial/luna-high/bench/t-pl/run-03/lorentzian_peak_fits.csv).

The plots show the allometric fits for both peaks: [intensity vs power](<BENCH>/raw/commercial/luna-high/bench/t-pl/run-03/intensity_vs_power.png) and [peak position vs power](<BENCH>/raw/commercial/luna-high/bench/t-pl/run-03/peak_position_vs_power.png). For the main peak, intensity scales as \(P^{0.925}\); the second peak’s intensity exponent is \(0.853 \pm 0.079\). Uncertainties are 95% confidence interval half-widths from the allometric regression.

RESULT: b = 0.925 ± 0.027