ABSTRACT
It is hardly necessary to point out the size of the archaeological problem in London, effectively a single site, with 677 acres inside the City boundary, of which 330 are inside the Roman wall. The real problem, however, is one of policy what one is studying the pottery for and the way that it interacts with the quantity. The sole use of description, even so-called 'objective' description, as a basis for the allocation of sherds to fabrics is therefore suspect, if one cannot lay out all the pottery at once. In mathematical terms, if one think of individual fabrics as points in a multidimensional 'fabric space', then Common Names are represented by clusters of points in that space. The pottery code has already been mentioned in connection with the Fabric Type Series and the Fabric Index. The code forms the logical structure of the Fabric Type Series. The pottery is stored in context groups. Proxy labels represent missing items.
