Radius sampling distance changes inferred branch matching in selected Allen mouse and human dendrites

We tested how radius sampling distance affects a passive branch-matching score in Allen Cell Types reconstructions. We used the same basal branch points at distances from 1 to 20 micrometres. In 32 selected mouse spiny cells from 32 donors, the donor-weighted geometric mean daughter-to-parent ratio fell from 1.680 to 1.519. In 46 human spiny cells from 18 donors, it fell from 1.939 to 1.495. Ideal matching is 1 under uniform passive cable assumptions. The paired change in mean log ratio was -0.101 in mouse, with a 97.5% donor bootstrap interval of -0.184 to -0.019, and -0.260 in human, with interval -0.324 to -0.201. These intervals adjust for two primary groups. A geometric proof shows that continuous arms at a shared positive junction radius force this ratio toward 2 at zero sampling distance. Descriptive fitted exponents also changed with distance and reached the grid limit near the junction. A larger 5-micrometre support sample retained distance sensitivity. This is a measurement-method result, not evidence that real dendrites violate a physical matching law. Radius interpolation, selected long arms, tissue cutting, and absent electrical measurements limit interpretation. Code, derived radius samples, sample audits, input hashes, checks, and retrieval instructions are included.

Allen Institute · dendritic branching · Rall rule · radius interpolation · passive cable · measurement sensitivity