Homonymous entries of the database get a number specified in the field ‘homonym number’. Although homonymous entries are necessarily distinct by other fields, too, this number is needed for two purposes:

• Reference to lexical entries: for instance, from the entry look₁, there is a reference to peep₃.
• Sorting of the entry list: even though there are other relevant ordering criteria (see below), these do not suffice, and the number is the last resort.

Words that share their meaning and differ only in their word class, like look (v/n), are often treated as homonyms. Consequently, word class is one of the ordering criteria. However, many homonyms share their word class, so that this criterion is not sufficient. Moreover, it is not the primary criterion, either.

While the database is compiled, there are no criteria for adequate numbering of homonyms. However, it has to be specified if cross-referencing among entries works textually (rather than by record IDs). Therefore, during the compilation, homonym numbering is provisional.

In a printed dictionary, homonyms are numbered, too. Therefore, the homonym number will be exported when the database is converted into a dictionary. At that point, the numbering is definitive and should therefore obey a principle. This probably necessitates renumbering homonyms by hand before the export.

A common order principle for homonyms is as follows:

1. meaning,
2. word class.

E.g.:

• peep₁ (v.) ‘make a short, shrill sound’,
• peep₂ (n.) ‘a short, shrill sound’,
• peep₃ (v.) ‘look furtively’,
• peep₄ (n.) ‘furtive look’.

Since the primary criterion is not formalizable, homonym numbering must be done by hand.