Independent release gave a PD2–PV5 order bound in ten fly records with equal quanta, no noise and site CV≤3. Three separate omissions removed it.

We tested how anatomical counts affect an initial release estimate in ten fly paired recordings. We used 93 first responses and two supplied count panels. We fixed the observed response moments. With equal site probabilities, PD2 mean release probability exceeded PV5 for all 20 count pairs. Ratios ranged from 3.403 to 18.180. We then allowed probabilities to differ between sites. Release was independent, each connection had one fixed quantum size, and no noise was added. At a site probability CV cap of 3, the lower PD2 mean exceeded the upper PV5 mean by a ratio of at least 1.104. The bound was exact for the tested moments. However, three single-record omissions removed it. Trial resampling also showed unstable bounds. A signed window mean gave a stronger bound and passed all record omissions. Trial resampling still gave unstable results. Matched background voltage variance exceeded response variance in two weak-response records. Explicit models at a CV cap of 4 matched both peak moments in all records and reversed the order. These models did not fit the full amplitude distributions. The work quantifies a conditional count robustness result and its sample limits. It does not establish a biological release order beyond the source model. The package preserves compact voltage data, both count panels, exact bounds, model witnesses, code, figures, source hashes and repeat instructions.

connectome · structure and function · quantal analysis · release probability · synapse count · partial identification · Drosophila · sensitivity analysis