The interpretation of experiments for very large Na clusters (sizes of a few thousand atoms) suggest that these have icosahedral structures, and that the transition to the bcc structure of the bulk has not yet taken place. We argue here that the reason for this is the fact that the screening cloud around a Na+ion in a finite cluster depends on the cluster size, and even for clusters with about ten thousand atoms it has not yet converged to its bulk form. The same can be said for the effective interionic potential and cluster structure. PACS: 36.40+d 61.50.Lt 61.50.Cj.