An exact axon order test showed no clear gain beyond count controls in one newborn mouse source table

We tested a structural prediction of ordered motor axon activity. We used one released newborn mouse graph with 20 axons, 229 target rows, and 1,848 links. The export differs from the article totals, so the scope is this source table. An exact subset method fitted a line order on 114 targets. The score counts missing axons between the first and last input to a target. We averaged the score across all tied training optima. The mean score on 115 separate test targets was 5.943. Two control chains kept every target count and every axon count separately in training and test data. Each of 998 controls repeated the exact fit. Their mean score was 6.134. The observed decrease was 3.1%. Of the controls, 134 scored at least as well. The lower-tail Monte Carlo fraction was 0.135. Thus, this test did not establish added line order beyond the specified counts. The result does not measure recruitment or exclude weak structure. We provide a small exact tool for up to 20 axons, a method proof, figures, control scores, source provenance, and repeat instructions. Direct small-case checks passed. A replay from a fresh source download reproduced all numerical result files.

connectome · neuromuscular junction · synapse removal · motor development · axon order · fixed counts · exact optimization · reproducible analysis