Codd himself found that in some situations it was in fact partial or missing information that required representation, and worked around the issues that arose when he wanted to extend these capabilities to empty sets or missing values in a series of papers (Codd and Date, 1993; Codd, 1986, 1987) working within the bounds of the traditional forms (e.g. portraying NULLs as operations on values of unknown statuses). However, these factors are crucial to defining and representing knowledge.