This is an earlier experiment. Read the amendments below before treating its findings as current.

The question

A layer at 20 m made the fleet lose 3–9 times more craft. Where and why?

What was tested

Collision records, hard height limits, a lifted stack, and the low layer alone or under one other layer at 40 to 120 m; also the instant-response craft.

What was found then

Each layer's height band was soft: craft are pulled back into it, not stopped. Hard limits made losses 17–45 times worse. Most extra deaths were low-layer craft at ground level. Lifting the stack cut losses on both layouts (in 5 of 6 versions on the second). Even lifted, the low layer sank about 26 m. Alone it held height; under another layer it sank, 16 m with a 40 m gap and 4.5 m with a 120 m gap. A 'safe spacing' claim was drafted and withdrawn (one spacing rule tested). The instant-response craft held at every gap.

What we learned later

LATER / CYCLE 235

Giving the wing the instant-response craft's smaller spacing rule removed about two thirds of the sinking, not all.

research/design/prio_sag_pin2_record.md

LATER / CYCLE 243

This line of work on the sinking low layer closed with the cause still open.

research/design/prio_sag_line_close.md

What changed

nothing

Numbers and source references
12
hard-wall runsresearch/design/prio_floor_band_record.md
17–45×
extra losses in the two-layer sky with hard wallsresearch/design/prio_floor_band_record.md
1.5–7×
loss reduction from lifting the stack 25 m, first seedresearch/design/prio_floor_shift_record.md
5 of 6
versions with fewer losses on the second seedresearch/design/prio_floor_shift2_record.md
26 m / 27–31 m
lifted low layer's drop in 10 s / later flying height (box 40–50 m)research/design/prio_floor_shift2_record.md
16 / 8.5 / 4.5 m
sag under one other layer at 40 / 80 / 120 m gapsresearch/design/prio_band_sag2_record.md
−2.3 / −0.7 / 0.0 m
point craft's offset at the same gapsresearch/design/prio_sag_standard_record.md
5 / 0 / 8 vs 46 / 18 / 32
craft lost, point vs wing, at 40 / 80 / 120 mresearch/design/prio_sag_standard_record.md
Original structured record ↗