How thickness is measured. A sample thick enough to separate its first internal reflection from the main pulse gives two independent timings — the transmitted delay (n−1)d/c and the echo delay 2nd/c. Two equations, two unknowns: thickness and refractive index come out together, with no assumed index and no calibration against a reference sample. Below that, the echo hides inside the main pulse and the two stay correlated; the tool says so and falls back to the full Fabry–Pérot fit, where (n−1)·d is still exact to a few parts per thousand and a measured thickness pins down the rest. Absorption depends on amplitude rather than timing, so beam coupling, alignment and a stale reference all degrade it — where a figure is marked upper bound, it is not a measurement. Atmospheric water-vapour lines are flagged and excluded from every fit. Validate against a known standard before quoting absolute numbers.