Two adult worm maps retained wider model growth intervals than sampled controls with fixed reciprocal degrees

We tested whether reciprocal connections account for a reported excess of stable transient growth in two adult C. elegans chemical wiring models. Each binary map had 180 common neurons. We generated 199 independent swap chains for each map, control type and chain length. Controls kept each cell input and output degree. A stricter control also kept each cell reciprocal degree. At 100 attempts per edge, the map growth interval widths were 0.2418 and 0.2189. The reciprocal-degree control means were 0.1842 and 0.1721. No sampled control reached the map width. Degree-only control means were 0.2045 and 0.1865. Results at 20 attempts per edge agreed. Thus fixed reciprocal degrees did not remove the excess in these finite ensembles. All 1592 graphs passed constraint checks. Four independent random-stream replays matched. This is a positive linear model with equal binary weights and equal leak. Two partial maps, no matched activity, and no proof of uniform graph sampling limit the result. Original code, scalar results, a figure and pinned source hashes support reproduction; source and derived graph records stay outside the publication.

C. elegans · connectome · reciprocal degree · transient growth · null models · positive linear system