Tests of synapse distances rejected 7.7 to 9.2 percent of random label controls at a 5 percent cutoff in two C. elegans maps

We tested whether a rank test of synapse distances keeps its stated error rate when whole sites receive random type labels. We used 29 neuron pairs from the JSH L4 and N2U adult C. elegans nerve-ring maps. Each pair had at least three chemical sites and three electrical sites. We fixed all positions and type counts and enumerated 145,542 site label assignments. At a 5 percent cutoff, the distance test rejected 7.71 percent of assignments in JSH and 9.17 percent in N2U, with equal pair weights within each map. The largest pair rate was 22.28 percent. An exact whole-site test of the same rank statistic gave mean rates of 4.17 and 3.81 percent. Seven observed distance tests passed Holm correction within their maps. Only two exact site tests passed that correction; none passed correction across all 29 pairs. A six-site proof gave an exact distance probability of 1/5,005 but a whole-site probability of 0.1. Thus, that mismatch did not depend on a normal approximation. Axial spacing, 3D midpoint, count, and annotation checks show where results persist and where small panels limit them. This is a conditional calibration study. It does not reproduce earlier cable-distance figures, establish an age effect, or prove a lack of biological separation. The package preserves site tables, exact distributions, code, source hashes, figures, independent checks, and repeat instructions.

connectome · chemical synapse · gap junction · synapse position · permutation test · dependent distances · exact enumeration · Caenorhabditis elegans