substrates are on the order of 10 i10 and 10 i11 m 2 , respectively. 2 From the experimental data, it is clear that silicone resin impregnation of FRCI-12 restricts internal gas e ow much more than phenolic resin impregnation of FiberForm. There are several possible mechanisms to explain this. First, pressure-driven internal mass e ow in a porousmaterialscalestothethirdpower (atleast)withthecharacteristic channel dimension. Thus, comparable resin loadings into substrates with comparable densities are expected to cause greater e ow restrictions in the material with the e ner microstructure. Secondly, whileresin impregnationpredominantlycoatsthe e bers,some webbing between e bers is possible and has been observed in scanning electron microscope images. Such webbing would be more likely in the e ner microstructure substrate (FRCI-12) and may contribute to the very low continuum permeability of virgin SIRCA. For PICA specimens, the continuum permeability is larger and the slip parameter is smaller (on average) along in-plane directions than along transverse directions. This result is consistent with the anisotropicmicrostructureoftheFiberFormsubstrate,whichcauses the e ow path to be less tortuous and the mean distance between gas-surface collisions to be longer along in-plane directions than transversedirections. 2 ThoughFRCI-12substrateshavemicrostructural anisotropies similar to FiberForm, the transverse permeabilities of the virgin SIRCA specimens are found to be comparable with or greater than the in-plane permeabilities. This e nding gives further evidence that the internal e ow properties of virgin SIRCA are dominated by the microstructural changes produced by resin impregnation. Asexpected fromthe mass loss accompanying pyrolysis, charred specimensofbothPICAandSIRCAofferlessobstructiontothee ow than virgin specimens. For PICA specimens, the relative changes in K0 and b due to the pyrolysis process are larger for the transversely oriented specimens than the in-plane oriented specimens. This result is consonant with the view that resin is likely to agglomerate or web at e ber intersections during impregnation. Because e bers are preferentially oriented normal to the transverse direction, agglomerations at e ber intersections would offer more obstruction to transverse e ow than in-plane e ow, and charring would thus have a greater effect on transverse than in-plane permeability. The continuum permeability changes dramatically for SIRCA specimens upon charring, increasing by three orders of magnitude. Moreover, the underlying anisotropy of the FRCI-12 substrate is recovered; in-plane oriented specimens of charred SIRCA are about twice as permeable as transversely oriented specimens.