Fibrous networks comprise a range of materials, from F-actin and tubulin, in eukaryotic cells, [1][2][3] to cellulose, in paper. [4] In the last decade, carbon nanotubes (CNTs) have emerged as a technologically important class of fibers, with nanoscale dimensions of width but micro-to macroscopic dimensions of length. CNT networks enhance the strength, [5,6] conductivity, [7][8][9] and flame resistance [10] of polymer composites, and have a spectacular effect on the rheology of CNT suspensions. [11][12][13][14][15][16][17] With recent progress in the ability to control CNT length, aspect ratio, defined as the ratio L/d, where L is nanotube length and d is diameter, has emerged as a key parameter in determining CNT percolation, [6,7] optical response, [18] toxicity, [19,20] and flow alignment. [21,22] Although in most commercial applications the network forming attributes of CNTs are critical, the influence of aspect ratio on the mechanics of such networks has received only limited attention. Here, we measure the length-dependent mechanics of CNT networks, and demonstrate a cross-over from strain-induced deformation governed by aggregates of short rigid nanorods to that governed by individual long semi-flexible nanotubes. Our results are in agreement with simple scaling arguments, and suggest that short CNTs can provide the same network forming attributes as long CNTs at slightly higher concentrations, but with significantly improved flow processability due to a substantially lower network yield stress. Our results have important implications for the efficient flow processing of CNT fluids and composites. We use two distinct batches of multiwalled CNTs (MWNTs) grown under the same conditions in the same reactor, [23] with uniform diameter distribution, around 50 nm, and ''short'' (L ¼ 4 mm) and ''long'' (L ¼ 60 mm) lengths, where the former are obtained by sonicating the initially long MWNTs in solution for an extended period of time. The insets to Figure 1 show scanning electron microscopy (SEM) images of the as-grown MWNTs from two different perspectives prior to dispersion in the Newtonian epoxy solvent. No surfactant was used, as this reduces the electrical conductivity of the network by several orders of magnitude.[24] The suspensions prepared in this way have conductivities of 0.1 (V m) À1 at 0.1% CNT. Suspensions were prepared with 0.015% < f < 2% and 0.015% < f < 7% for the long and short MWNTs, respectively, where f is the nanotube mass fraction. These concentrations correspond to a wide range of semidilute to concentrated regimes, [25] implying significant overlap and mechanical entanglement. To measure the network morphology, we used small-angle neutron scattering (SANS) and ultrasmall-angle neutron scattering (USANS), as shown in Figure 1. Scaling the scattering intensity by f collapses each data set onto a single curve, suggesting that the nature of the nanotube structure is relatively independent of concentration. Both the short and long nanotubes show Q À4 behavior, at large scattered wave-vector (Q...