RECENT galaxy redshift surveys1-3 offer evidence of a quasiperiodic network of 'sheets', with typical dimensions of ~100-1Mpe (h is the Hubble constant in units of 100 km s-1 Mpc-1). Galaxy formation models in which large-scale structure forms purely by gravitational instability from small initial irregularities seem to have problems producing structures on such large scales without violating constraints on the anisotropy of the microwave background, unless they start from apparently contrived initial-fluctuation spectra. Models in which galaxies and clusters have a non-gravitational (possibly explosive) origin seem to be better candidates for the progenitors of large sheets and bubbles, but one would still not expect a perfectly regular lattice. So how surprising are the observations, if they are interpreted in the framework of such models? Here I show that a random cellular model, which mimics the galaxy distribution formed as matter is swept up into shells around expanding void regions or by cosmic blast waves, reproduces the observed Great Walls, voids and periodic structures without undue difficulty.