The aim of a stratigraphic analysis is the reconstruction of the relative chronological sequence of an archaeological site irrespective of the space dimension. Most archaeologists will agree that of all approaches to reconstruct the chronological sequence, the Harris matrix diagram has become not only the most widely used but also the most effective. So why is there a need for discussion? In an e-mail discussion David Bibby and I identified two controversial issues: (i) The Harris diagram represents only one aspect – chronology – of the site, whereas the spatial component, e.g. the fact that some contexts are far bigger than other contexts does not play a role at all. We all know that in a large Harris diagram the probability of introducing an error is large. 3D recording or a 2D GIS of the site may help to identify errors and only when this spatial information is included the complete picture of a site can be presented. But how can we combine the spatial and the chronological information efficiently? Is voxel GIS a solution? How can we record interfaces in a voxel GIS? (ii) The Harris diagram is an interpretation of a site. If a dig was directed by someone other than the director who actually did the dig, the results might be a different one. How can we present a more complete and objective picture of a site? Consider the following excavation technique: Remove extremely thin artificial strata, take a photo suitable for rectification, remove the next slice and so on, so that you get slice by slice a 3D model of the site similar to a computer tomography., Can one, on the basis of such a model, identify boundaries of contexts and the interfaces more objectively? If you are interested in answers to these questions you are welcome to take part in the discussion.