With normalized count weights and assumed 25% threshold gaps, a twofold source rate range allowed reversed order in 185/206 pairs in one fly leg map

We tested how source rates limit recruitment order in one fly leg connectome. We used all 69 motor targets and 292 identified local sources with the transmitter label Ach in a pinned FANC v840 table. We normalized each target count vector to sum to one. Within each motor module, we gave the target with more selected input a threshold 25% higher than the other target. These thresholds were assumptions. An exact bound found that source rates between one and two permitted reversed order in 185 of 206 pairs. Each pair used a separate source pattern. Giving sources with the same released source label one rate reduced the count to 93. Shared rates within each hemilineage reduced it to 78. The median smallest source cohort for the 185 separate-rate cases had six sources. In the SETi-FETi example, 14 sources sufficed at the twofold range. Uniform random source rates reversed only 0.526% of pooled pair-by-draw events. Thus the possible reversal count does not estimate a natural reversal frequency. Threshold, source-subset, link-weight and source-version checks retained the difference between separate and shared rates. A separate linear program checked 618 cases, and a fresh download repeated all 14 numerical files exactly. The result gives a conditional limit on recruitment claims from wiring. It does not show reversed recruitment in a live fly. The package preserves code, exact integer certificates, activation cohorts, figures, source hashes, proofs and repeat instructions. Original source frames remain external under their source terms.

connectome · motor neuron · recruitment order · premotor input · Drosophila · source activity · linear model · exact bound · FANC