[Bicycle Bob]

Thanks for the source postings. The bible I've been going by is "Theory of Wing Sections" by Abbott and Von Doenhoff, which takes up the history from just after that first chart. Being concerned with wings, it makes sense for them to not go beyond 21% thick, as they want to leave something in hand for some angle of attack. That had persuaded me to go to 25% on a strut myself.

The NACA tests progressed to "laminar flow shapes" of which the 67-series is the most extreme. These are almost lab curiosities, first achieved in production on the P-51, using bondo over the flush rivets. No such shape can tolerate a blemish, let alone a wheel opening, without tripping the boundary layer into turbulence in a wake spreading 15 deg to each side. The "standard roughness" in these tests, which double the drag, consists of .011" grit, covering 5-10% of the area, just within 8% of the length from the leading edge. (like small bug strikes)

Here are a few selected data points from my book:
The lowest cd shown was for a shape that didn't make a stable wing, so there may be much room for exploration. That wing is the 0010-35, in the old system, 10% thick. At Re from 3 to 9 million, the cd is .003 near-polished, and .009 "rough."

The 2424 shape is similar to the 2412 It gets a cd of .008 smooth and .013 rough, at 24% thickness, while the 12% thick shape gets .0065 smooth and .010 rough.

Moving on to the later laminar shapes, which rely on perfect conditions to maintain a laminar boundary layer back as far as the second term in the identifier, usually the widest part of the shape. A 63-006 shape is a 6-series, (laminar) with 30% laminar flow, no camber for lift, and 6% thick. - ie: 100" long, 6" thick. The 67-025 is 70% laminar, no camber, 25% thick. The 66 and 67s are slipprier than the 63 and 64s, but who can wait for 70% of the way back for the first seam, wheel opening, or other feature?

Shape: cd around 6 mil Re. cd with standard roughness
63-006 .0043 .0087
63-021 .0055 .011
64-009 .0043 .0087
64-015 .0047 .0097
64-021 .0054 .018
66-006 .0032 .0085
66-021 .0041 .013

These all show a clear preference for the slender shape, despite the extra surface drag. The text helps illustrate the trends with increasing speed, and gives examples that improve, such as the 65-418, with less separation as you go faster. This is not uncommon, but the devil is in the details, and these shapes are only a rough indication about designing practical cars.

Re: underflow - it is good if done right, but blocking it off is a cheap fix that gives downforce, which is good for road racing, not cruising. Underflow is not really hard to visualize; it is like a duct, but with the boundary layer on the flat side moving and helping stabilize things. After F1 banned the vacuum cleaner cars, they did quite well with underbody venturi systems.