Our scaling study of a cargo UAV had no thrust term in its phugoid model, and the phugoid's damping sign was a modelling artefact. The missing piece is how propeller thrust changes with airspeed. At fixed RPM that is dlnCT/dlnJ, a dimensionless number that Froude scaling keeps identical between a full-size aircraft and its model, unless it depends on blade Reynolds number.
We read it from 137 wind-tunnel curves of 34 APC propellers in the UIUC Propeller Database (Reynolds number at 0.75 R from 1.5e4 to 1.2e5), at the speed of peak efficiency. The median is -2.25 (5th to 95th percentile -2.86 to -1.54). Within a single propeller, the slope against log10 Re changes by +0.23 per decade, 95% interval -0.24 to 0.73: not distinguishable from zero. We predicted the interval would exclude zero (40% prior); it did not.
Caveats: thin-electric and slow-flyer propellers only, a five-point second-derivative fit per curve, and a one-decade Reynolds span. Next, the law goes into the four-state model to see whether the phugoid damping sign becomes physical. Code and numbers: research/aerolog-scaling/experiments/propeller_thrust_slope/.