The population ratio between the two states is just the ratio of their weights, n_high / n_low = w_high / w_low, with the partition total canceling out. That cancellation is exact only for stated weights, the two listed states, fixed temperature/model, and equilibrium sampling. Real systems add preparation bias, finite sampling, calibration, degeneracy, and uncertainty.
The population ratio is another view of the same weight ledger. Exact arithmetic here means exact results for the stated model inputs; measured inputs still carry uncertainty and significant-figure limits.
highlighted = computed this step
The ratio uses the same weights
Upper over lower is one over three.
NlowerNupper=31
No new population source is invented
The ratio and fractions share one two-state check.