FIGURE 6 — FIELD SELECTION AND CHRONOLOGICAL REFERENCES

This module visualizes existing results without changing the numerical models.
Panel a reads every row of the archived 19,939-field fixed-scale census, plots
the empirical cumulative distribution of the best RMS, and enlarges the eight
lowest scores. The separate free-proportion rank is computed from ratio_E;
that metric is not overlaid on the degree-valued RMS axis.

Panel b evaluates the unchanged S11 model between 3100 and 2450 BCE in half-year
steps for (28,72,16) and (28,72,20). It reads the two full-window minima from the
archived denominator audit and rechecks their scores through sky_model.word.
The shaded intervals are 3000–2920 cal BCE (main reference 2) and 2635–2575 cal
BCE (main reference 33). Woodhenge selection was informed by its known date;
main Methods 2.5 and S11 describe that retrospective information order.

After extracting Supplementary Data 1, run:
  python -m pip install -r Figure6_Sky_Comparison/requirements.txt
  python Figure6_Sky_Comparison/render_figure6.py

The default data root is the module's parent. It expects Research_Record_S1_S10.tar.xz
and S11_Woodhenge there. It reads one CSV directly from the tar.xz; no full archive
extraction or external download is needed. Optional arguments are --data-root
PATH and --output-dir PATH. Outputs contain the PNG, vector PDF, sampled curves,
eight displayed candidate rows and a numerical/source-hash record.

No census is rerun and no additional archaeological observation is introduced.
The distribution is a description of the defined candidate family, not a null
distribution or probability of prehistoric intention.
