(a–c) Measurement of the length over which axons in LM fluorescence data are faithfully reconstructable (see Fig. 1d). (a) Experimental timeline: Injection of a fluorescent-protein (FP) expressing adeno-associated virus at postnatal day 28 (AAV.eGFP into M1 cortex, AAV.tdTomato into S2 cortex of a wild-type C57BL/6j mouse); transcardial perfusion, fixation, sample extraction from cortical L1 at postnatal day 46 followed by confocal volume imaging. (b) Rendering of a subvolume of the three-channel confocal image stack showing axonal long-range projections from M1 cortex (green), S2 cortex (red) as well as stained blood vessels (blue). (c) Histogram of the reconstructable pathlength () for the fluorescently stained long-range axons (red, green) in the example LM dataset (b) based on two independent expert annotations (light and dark colors; median of 28 µm and 33 µm (green channel, n = 115 and 88) and 88 µm and 107 µm (red channel, n = 56 and 47) for the two annotators, respectively). Dashed lines: cumulative axon fraction per annotator and channel. Note that annotators were explicitly asked to stop reconstructions whenever any ambiguity or uncertainty occurred, thus biasing these reconstructions to shorter length. (d–k), EM-based measurements of the length over which the trajectory of axons becomes unique in several cortical layers. (d) Renderings of three 3D EM datasets in layers 1, 2/3 and 4 of mouse cortex. (e) Skeleton reconstructions in the L1 3D EM dataset of all 220 axons (black) that traversed a center bounding box (blue) of edge length = 5 µm. (f) Determination of the uniqueness length for a given axon (black) by counting the number of other axons (gray) that were within the same seeding volume (blue box) and remain within a distance of no more than from the reference axon for increasing distances d from the center box. was defined as the Euclidean distance from the center box at which no other axon persistently was at less than distance from the reference axon (red: at least one neighboring axon, blue: no more neighboring axon). (g) Axonal uniqueness length in cortical L1 for all n = 220 axons from the center bounding box (see e). Number of neighboring axons for each axon (black) and the combined fraction of unique axons (blue) over Euclidean distance from the bounding box center. Note that at = 35 µm the trajectory of 90% of all axons has become unique (dashed blue line). For three of 220 axons, one or two other axons remained in proximity within the dataset (red triangles). (h,i) Axonal uniqueness length in two additional 3D EM datasets from L2/3 (h, n = 207 axons) and L4 (i, n = 128 axons), labels as as in g. (j) Box plot of uniqueness length per axon for cortical layers 1-4 (from g-i). Boxes indicate (10th, 90th)-percentiles, whiskers indicate (5th, 95th)-percentiles, red crosses indicate outliers. (k) Effect of the coarse alignment precision on axonal uniqueness length (resampled for smaller = 1, 2, 3, 4 µm, see Materials and methods).