CALC-086

Two scale-free ratios on SRC-085's Figure 18

Supports CLM-0867 and CLM-0868 (low confidence). A calculation is required: both claims state values produced by arithmetic performed here. From an independent adversarial pass that did not read the model code and worked from the primary PDF, with a sabotage check run on every step.

Both quantities are ratios, and that is the point of them. A ratio of two radii on the same curve, or of a peak to a mean over the same disc, cancels any uniform error factor and any scale error. So neither result can be argued away by an assumption about losses or about the figure's x scale - only the x origin matters, and every result below is stated across all candidate origins.

1. Peak to mean inside 5 km, from the source's own three numbers

peak / mean(inside 5 km)  =  S0 / [ f * P / (pi * 5000^2) ]

SRC-085's own 230 W/m2, 88 pct, 6.72 GW       3.0547
SRC-085's own 230 W/m2, 89 pct, 6.72 GW       3.0204   (89 pct is what two of its own chains give)
model, SRC-085's specified 10-step taper      2.8663
model, continuous 10 dB Gaussian              2.8719

Discrepancy 5 to 7 per cent. A uniform error factor cancels from this ratio, so no error budget reconciles it: the source's stated peak, stated collection efficiency and stated radiated power over-determine the aperture. The calculation does not say which of the three is wrong.

2. Main-lobe skirt ratio, r(peak/230) / r(peak/23)

Matching the model would require the x origin at 195.8 PDF units, which the same figure's sidelobe level and its own annotated radii both exclude. The disagreement is therefore as origin-robust as the sidelobe-position agreement recorded in the same pass. The error budget pushes the wrong way: filling the null raises the skirt and would increase the ratio.

3. The control that agrees

Verification note, VER-1093: the control agrees AT THE EXTREME OF THE BAND, not flatly. Model 1.5258 matches the TOP of the measured band, 1.526, to 0.013 per cent, and misses the BOTTOM, 1.490, by 2.40 per cent. It is still a real control - unlike the skirt ratio it is not a disagreement at any origin - but "AGREES" flat overstates it.

A label trap in SRC-085, found while checking this and recorded so nobody walks into it. The source calls its illumination taper 10 dB throughout. Its own Figure 16 step table runs 22.14 down to 2.46 kW/m2, a ratio of exactly 9.0000, which is 9.5424 dB. The adversarial pass transcribed the table rather than the label, so its model is built on the real values - but anyone later modelling "the 10 dB taper" from the source's own words will build a different aperture from the one the source designed.

The most origin-insensitive ratio available, r(sidelobe 2) / r(sidelobe 1), agrees: model 1.5258 against 1.490 to 1.526 measured across the same origins. A pass that found only disagreements would be suspect; this one did not.

4. Two figure annotations not previously recorded

Figure 18 marks 5.7 km at the 0.1 mW/cm2 (1 W/m2) contour, which is the rectenna site exclusion boundary value CLM-0020 records, and 13.6 km where the pattern falls through 0.01 mW/cm2, the USSR guideline drawn on the same figure. The model puts the 1 W/m2 contour at 6.02 km beam-normal.

5. The unresolved residual, stated rather than tidied

Figure 18's x origin is not resolved better than about 4 per cent. Six routes span 178 to 192 PDF units: the figure's own sidelobe level and annotated radii point to 179 to 180, the drawn axis and the 0 label to 183 to 186. Nothing in the figure supports the 194 that the prose's "approximately 9 km" first sidelobe would need. No conclusion above depends on resolving it, and nothing short of a cleaner scan would settle it. None was sought.

Cite this