You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
What changes if we do it: about 1,500 film frames on nine 1985 rolls stop being drawn at nominal scale if, as looks likely, their scale is the wrong field — the same scale_wrong class #72 tabled. What happens if we never do: they stay on the nominal fallback at roughly 1/r of their true width (r ≈ 1.6–1.8), which is the documented behaviour for an unmeasured roll-height, and the note's bound keeps saying so.
Problem
#72 settled the r ≈ 2 mass per roll-height, but only over the sweep's near_upper stratum (2 < r above sea level ≤ 3). Just under its floor sits a family it never sampled:
1,534 frames on nine 1985 rolls (bc85063, bc85078, bc85080–bc85083, bc85089–bc85091) at 6,096 m (20,000 ft), 305 mm, 1:10000. Ratio above sea level 6096 / 3050 = 1.999, against the sample's lowest 2.025.
Two of them surfaced among the random draws beyond the band: bc85083 r 1.73, bc85090 r 1.63.
Sample MRDEM under the family's frames (they are not in inst/extdata/flying_height_sweep.csv) — or add a stratum to data-raw/height_calibrate-flying_height_slip.R for 1.6 < r_asl ≤ 2.
Fetch and transcribe the remaining rolls' logbook pages, blind. bc85080 and bc85081 are already in data-raw/flying_height_logbooks.csv, including their 20,000 ft strips.
Run them through settle() under tail = "near_upper" with the same rule: logbook factor 1, no legible lens conflict, and spacing that fits the reported height and rejects nominal.
Before starting, check whether the stratum holds other families. The "1.6 < r_asl ≤ 2" band was never sampled at all, so this family is the largest known group, not necessarily the largest.
See inst/notes/terrain-correction.md, "The r ≈ 2 mass is two defects", Bound.
What changes if we do it: about 1,500 film frames on nine 1985 rolls stop being drawn at nominal scale if, as looks likely, their
scaleis the wrong field — the samescale_wrongclass #72 tabled. What happens if we never do: they stay on the nominal fallback at roughly 1/r of their true width (r ≈ 1.6–1.8), which is the documented behaviour for an unmeasured roll-height, and the note's bound keeps saying so.Problem
#72 settled the r ≈ 2 mass per roll-height, but only over the sweep's
near_upperstratum (2 < r above sea level ≤ 3). Just under its floor sits a family it never sampled:bc85063,bc85078,bc85080–bc85083,bc85089–bc85091) at 6,096 m (20,000 ft), 305 mm, 1:10000. Ratio above sea level 6096 / 3050 = 1.999, against the sample's lowest 2.025.bc85083r 1.73,bc85090r 1.63.bc85079–bc85081roll-heights are the same camera, lens and scale at 21,000 ft, and the pages read in Half the r≈2 mass is a wrong scale, not a mislabelled lens, and the fallback draws it at half width #72 log 20,000 ft forbc8508085-145 andbc8508181-97 — the height these frames carry.So they are likely
scale_wrong(true scale nearer 1:17000), and unmeasured.What would settle it
The #72 route, extended to this stratum:
inst/extdata/flying_height_sweep.csv) — or add a stratum todata-raw/height_calibrate-flying_height_slip.Rfor 1.6 < r_asl ≤ 2.bc85080andbc85081are already indata-raw/flying_height_logbooks.csv, including their 20,000 ft strips.settle()undertail = "near_upper"with the same rule: logbook factor 1, no legible lens conflict, and spacing that fits the reported height and rejects nominal.Before starting, check whether the stratum holds other families. The "1.6 < r_asl ≤ 2" band was never sampled at all, so this family is the largest known group, not necessarily the largest.
See
inst/notes/terrain-correction.md, "The r ≈ 2 mass is two defects", Bound.