It is of fundamental importance for ice core interpretations to understand how near surface properties of firn microstructure are linked to the process of air entrapment at the firn-ice transition in 60 to 100 m depth. There is still an open debate with the consensus that temperature and accumulation rate are not sufficient to parameterize the pore close-off density/porosity especially its change over time at one specific site. There are suggestions that some anisotropic features (extended connections via vertical pores) the layering (increase of mean air content due to the sealing effect of dense layers) or differences in grain size (sintering of larger grains increases enclosed air volume) might affect the close-off density respective air content however without much experimental evidence. In this talk we will give experimental arguments for the layering hypothesis. The measurements on microstructure at various polar sites were done with the means of three different methods including 3D-X-ray micro-computer tomography (µCT), 2D- microstructure mapping (µSM) and 1D-absorption of gamma radiation (AGM). The MiCT measurements provide the first data set of three dimensional parameters on dry polar firn including the coordination number, the structural anisotropy, neck radii, grain clustering and open/closed porosity determinations. The investigations at the firn-ice transition reveal surprisingly large porosity variations on the cm-scale (large layering). However the estimates of open/closed porosity suggest constant critical porosities for firn microstructure (independent of grain size). Even no evidence for significant anisotropies at the firn-ice transition has been detected. The microstructure observations at the firn-ice transition are compared with results from a percolation model based on a regular bcc lattice which has taken the layered structure into account. The agreement between model results and observations supports the layering hypothesis and show that the mean critical density/porosity depends not only on the amplitude of porosity, respectively density variations with depth but also on the frequency of such fluctuations. We conclude that the mean porosity at the firn-ice transition might depend primarily on the magnitude of layering at the specific site.