. . . . "Each archive's SQLite database was extracted. For every frame with tracked cells (about 17 h, 200-245 frames per wing; the frames table lists further time points without cell data), each cell's neighbours were listed from the directed bonds and their conjugates; cells sharing a bond with the margin cell were dropped, leaving 950-1,400 interior cells per frame (767k cell-frames). Per frame: fraction of six-sided cells f6, mean and SD of neighbour number, CV of apical area, and the number of daughter cells appearing by division in that frame, expressed per hour per 1,000 interior cells and smoothed over 12 frames for the correlation. Per wing: f6 averaged over the first and the last 2 h and its change; Spearman rho of f6 against time, against the smoothed division rate and against area CV across frames; the fraction of interior vertices with three bonds; and at every 20th frame the chord angles at interior three-way junctions (SD and the fraction of junctions with all three angles within 15 degrees of 120), compared between the first and the last 2 h." . . "Measured: 0.7636 median over wings of the fraction of six-sided interior cells in the last 2 h (n = 3 wings, 766,923 interior cell-frames, range 0.742-0.773 across wings; first 2 h 0.435-0.479). Interpretation: The first run of this claim (top-level result.json) refuted it as a sufficient rule in a proliferating epithelium where the size condition was only marginal. This run asks when, if ever, the rule turns on: the pupal wing is the classic case (Classen et al. 2005, Dev Cell 9:805) in which an irregular epithelium repacks into mostly hexagons between about 15 and 32 h APF as divisions cease, a repacking that the authors attribute to planar cell polarity-dependent junction remodelling rather than to surface tension alone. The test treats Thompson's conditions as the state of the tissue at the end of the window (no divisions, confluent, under tension) and asks whether the hexagon pattern follows; the secondary correlations ask whether it tracks the removal of division and size variation. Whole-wing interior cells are used; veins and inter-veins are not separated. The database format is identical to the ch07-09 first run, so the same code path is used. Data are not yet downloaded." . "Claim test of thompson1917-ch07-09 (pupal-wing-timelapse): supports, confidence 0.8" . . . "Public, expert-corrected segmentations with cell tracking and lineage of a confluent epithelium followed through the end of its proliferative phase, not a tissue the claim's author used as an example, not produced by a segmenter shown in this repo to round junctions." . "claim-test/0.1.0" . "biological" . "The six-sided fraction stays below 0.50, or falls, after division ceases, so the end of proliferation and continued cortical tension do not produce a pattern of hexagons." . "Data licence for the archives is not stated at the download location; they are distributed publicly as TissueMiner example data and are used here for analysis only. Data were downloaded after the pre-registration commit (1805066a) and are not committed (2.2 GB); SHA-256 of each archive is recorded." . "Mechanism is not tested. Classen et al. 2005 showed the same repacking requires planar cell polarity genes and junction remodelling (T1 transitions), and Etournay et al. 2015 showed the wing blade is under anisotropic tension from hinge contraction during this window; Thompson's mechanism is isotropic surface tension. The data show that the hexagon pattern appears when his conditions hold, not that it appears for his reason. The division-rate correlation is also consistent with the division-topology account of the low hexagon fraction in proliferating tissue (Gibson et al. 2006)." . "Pre-registered support criterion met in 3 of 3 wings: f6 rises from 0.435-0.479 in the first 2 h to 0.742-0.773 in the last 2 h (change +0.26 to +0.34; required >= 0.60 and >= +0.10), with 99.7% three-way vertices. All three secondary expectations also hold in every wing: rho(f6, division rate) -0.65 to -0.91, rho(f6, area CV) -0.57 to -0.79, and junction regularity rises (all three angles within 15 degrees of 120: 24-27% early, 42-49% late; angle SD 21 degrees early, 15 degrees late). The polygon distribution in the last 2 h is 12% pentagons, 76% hexagons, 11% heptagons; SD of neighbour number falls from 0.88 to 0.50." . "Taken with the first run, the two runs bracket the claim: the same cell type is 38-40% hexagons while dividing (area CV about 0.5) and 74-77% once division stops and area CV falls to 0.34, with the hexagon fraction tracking both. Thompson's conditions (similar size, complete contact, uniform force) are met at the end of the pupal window and the hexagon pattern follows. Hexagons remain below the compound eye (0.99 in claim-sources/cell-geometry) and junction angles (SD 15 degrees) remain less regular than corneal endothelium (10 degrees), so 'a pattern of hexagons' here is a 3:1 majority, not a lattice." . "The division rate is counted as daughter cells appearing by division (two per division) per hour per 1,000 interior cells; the absolute value is affected by cells entering the tracked region (appears_by MovedIntoMask) and by the smoothing window, the correlation with f6 is not. The first and last 2 h are compared within each movie; absolute developmental time is taken from the paper (imaging began at about 16 h APF), not from the databases." . "Whole-wing interior cells were used; vein and inter-vein cells were not separated although they differ in shape (Etournay et al. 2016). The three wings come from one lab and one genotype; the Dye et al. discs of the first run come from the same lab, so the two runs share segmentation conventions but not tissue stage." . "When cell division stops in a confluent epithelium under tension, the six-sided fraction rises to a clear majority (>= 0.60) and junctions approach 120 degrees, in step with falling division rate and falling cell-size variation." . "Pre-registered. Refuted if in at least two of three wings f6 over the last 2 h is < 0.50, i.e. hexagons do not become a majority even after proliferation ends; or if f6 falls by >= 0.05 from first to last 2 h. In between (f6 in [0.50, 0.60), or wings split) is 'mixed'. If the databases cannot be obtained or do not cover the window up to >= 28 h APF the verdict is inconclusive with reason data-unavailable or underpowered." . "0.8"^^ . . "supports" . "If similar-sized cells in complete contact under a uniform force tend to form a pattern of hexagons, then in the Drosophila pupal wing blade, where cell division ceases during the imaged window (16-32 h APF) and the epithelium remains confluent under cortical tension, the fraction of six-sided cells must rise from the proliferating-epithelium value (about 0.4, measured in the first run of this claim) to a clear hexagonal majority (>= 0.60) by the end of the window, and the rise must track the fall in division rate and in cell-size variation." . "Per wing: f6 in the first and last 2 h and its change, three-way vertex fraction, Spearman rho of f6 against division rate and area CV across frames, chord-angle regularity early versus late; verdict by majority of wings against fixed thresholds." . "Pre-registered. For each of the three wild-type pupal wing movies WT_1, WT_2, WT_3 (TissueMiner databases distributed with Etournay et al. 2016; the wings of Etournay et al. 2015, 16-32 h APF, 5 min frames), compute per frame for interior cells (no bond with the margin cell 10000) the fraction of six-sided cells f6, the mean and SD of neighbour number, the CV of cell area, the division rate (cells appearing by division per hour per 1,000 cells) and, every 20th frame, the fraction of interior three-way junctions with all three chord angles within 15 degrees of 120. Supported if in at least two of three wings f6 averaged over the last 2 h is >= 0.60 and exceeds f6 over the first 2 h by >= 0.10, with the three-way vertex fraction >= 0.95; secondary pre-registered expectations, reported but not required: Spearman rho(f6, division rate) across frames <= -0.3, rho(f6, CV_A) <= -0.3, and the fraction of near-120-degree junctions higher in the last 2 h than in the first 2 h." . "false"^^ . . "three wild-type pupal wings imaged every 5 min from about 16 h APF for about 17 h, the window in which the last division wave ends and the wing blade is remodelled under tension" . "not stated for the data files; the TissueMiner repository is BSD-licensed (Scientific Computing Facility, MPI-CBG) and the archives are offered for public download from http://bds.mpi-cbg.de/tissue_miner/" . . "Dataset: TissueMiner example datasets WT_1, WT_2, WT_3: the three wild-type E-cadherin-GFP Drosophila pupal wing time-lapses of Etournay et al. 2015 (eLife 4:e07090), segmented with TissueAnalyzer and tracked, as TissueMiner SQLite databases (Etournay et al. 2016, eLife 5:e14334)" . . . . . "sha256:119d0facc43b480a7673ef4b8e0405368cccca28fc9456ed0a8767d77c4792dd" . "sha256:994779314e39a906b9e171c7f2da0b12cacd94708bb09d6a70236ca62cec53ae" . "sha256:db5de2a3b6a1547eac5a19227c90cd7408de4b54db6a8992847963996faa8df8" . . "delphisci claim tester (claude-fable-5.1; cursor local session, hexagon follow-up)" . . . "claude-fable-5.1" . . . . . . . . . "2026-11-08T16:24:31.902Z"^^ . . . . . "Claim test of thompson1917-ch07-09 (pupal-wing-timelapse): supports, confidence 0.8" . . "RSA" . "MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAyFHtPB6p0mCjDozxlOY9CmO3nBm7Te3jFbXyLmM+G1tGe47Da0jYIbhkl9Oo1ZjxmCa3LCwlEqK8gj6hEIcL4ahW30JXh9HycfKVk1bugkzMEV15tJdTzAU3sGMFnbNsn43xY2LFLy45tApkRm5byEiM7oRRrooIdwoQKnP4Wzq/bfAoJ9qvW+MGkelyOTd1jbSt+KfmT7xKmDfWzAHddUlWwMX2QJOLxDaffOteTop6ZRgPiNmPJKA/S62Gac9XtSYT3oDkZ6/nr5cxxNLpOyxTf5FPISiCe11xYO/unYjY+mwnOfA8yhJIyx1l0O9I0vpOpn+E5TvyHJV+oSTrvwIDAQAB" . "x8o9OjOfXybQe6+XbCz4MpzZYpde7l8N0yYmG1I7GmjhgHGGRindVEmf66GsRjcMvbmKlYiXuO+fPYNIxApRx5fR7UNQrSS5JpWC5HYhMVVCFON0gXgbTSFckchb9fH3ow6Prm1jQ1F1sN0mB4CyXZtw+BzhgXNpPQh3rkPxqwilWLo5efwQDO+/2wdrHMxexPQXk3Q+ITOFz3L//yv9taB+iGmYJQv9mXWepXdBqpgMCkMqI3nRBW9iTwWgvaLn6n+Bwzym5L/DbPDR5xVMnmrUTHR7AEUKt+6Ggb8ZjGcJ8f/bdLcJbzbHh4kIi1Ej1YLz7MHGioRSHdCh2H4iAQ==" . . .