This article deals with the problem of finding names or phrases to describe a concept. This is a difficult problem, approached in different ways by different individuals. Fundamentally, it is the cause of many indexing difficulties, where an indexer and a searcher may not think of the same terms, and leads to decreases in recall.
The article presents empirical data from several domains regarding naming. Discussion then goes on to possible solutions. The main one advanced is essentially that of thesaurus construction - finding a number of words and phrases that can be considered as alternatives when describing a given concept. The article suggests soliciting data from a number of individuals as a reliable way to get coverage, which can be automated and possily improved by a ``failure analysis'' approach such as when ``unlimited aliasing'' is built into the system.