To study the overall structure of the peptidoglycan fabric of the sacculi of gram-negative and gram-positive walls, actively growing cultures of Escherichia coli and Bacillus subtilis were treated with boiling sodium dodecyl sulfate solutions. The sacculi were then treated with enzymes to eliminate proteins and nucleic acids. These intact saccoli were probed with fluorescein-labeled dextrans with a range of known molecular weights. The penetration of the probes could be monitored by the negative-staining appearance in the fluorescence microscope. At several chosen times, the molecular weight fraction that allowed barely observable entry of the fluorescein-labeled probe and the molecular weight fraction that penetrated to achieve almost, but not quite, the concentration of probe in the solution external to the sacculi were determined. From three pairs of times and molecular weights that met one or the other of these two criteria, the effective pore size could be calculated. The minimum size of protein molecule that could diffuse through the pores was also calculated. Two mathematical models, which gave essentially the same results, were used to interpret the experimental data: one for the permeation of random coils through a surface containing holes and the other for rigid spheres diffusing through water-filled cylindrical pores. The mean estimate of the effective hole radius in walls from E. coli is 2.06 nm, and that of the effective hole size in walls from B. subtilis is 2.12 nm. These results are supported by experiments in which the loss of preloaded cells was monitored. Various fluorescein-labeled dextran samples were mixed with samples of intact cell walls, held for a long time, and then diluted. The efflux of the dextrans was monitored. Neither large nor small dextrans stained under these conditions. Only with dextran samples of a sufficiently small size were the sacculi filled during the preincubation period, and only with the largest of these could the probe not escape quickly. From the pore (or mesh) size, it can be concluded that the wall fabric of both organisms has few imperfections and that the major passageway is through the smallest possible pore, or "tessera," formed by the maximal cross-linking of the peptides from glycan chain to glycan chain compatible with the degree of rotational flexibility of the chains of repeating disaccharides of N-acetyl muramic acid and N-acetyl glucosamine. A tessera is composed of two chains of eight saccharides cross-linked by two octapeptides. The size of a globular hydrophilic molecule, if it did not bind to wall components, that could pass freely through the meshwork of an unstretched sacculus of either organism is roughly 25 kDa. We stress that this is only a rough estimate, and it may be possible for proteins of less than 50 kDa to pass through the native wall during normal growth conditions.