[aerohead]

I went back through Hucho.
What I was remembering, turned out to be about the 1951 VW microbus, compared to the 1969 Vanagon.
Pawlowski published on minimum leading edge radii requirements in 1930.
Lay's research in 1933 confirmed the efficacy of Pawlowski's work.
M'o'ller, in 1951, did not use Pawlowski's geometry for the VW bus.
It turned out that, the 1951 design would have had no greater drag, if the front radii had been just a fraction of what was used.
--------------------------------------------------------------------------------------
With respect to scale models:
Hucho suggests that anything below 1/4-scale would be good only for 'qualitative' investigation. You would not be able to do 'quantitative' work below that 'size'.
In the United States, the typical minimum model scale is 3/8ths, as of 1986.
Hucho said that 1/4-scale results are identical to 1:1-scale.
In 1991, Texas Tech University was using 1/24-scale in their small water tunnel, for food coloring-injected flow imaging. Only.
For underwater tow-testing and actual drag measurements from an overhead gantry bridge- supported sting, they used a 3/8-scale model, and SCUBA divers for close observation and image capture.
Ford paid $ 68,000 for a highly accurate 3/8-scale Ford Taurus model, with clear plastic hood ( bonnet ), complete with all engine bay structures, to develop CFD code. Model velocity was 2-mph.