Developing historical dictionaries for the era of the semantic web presents a number of problems, in particular, many dictionaries use distinct methodologies and have their own data models as a result. There are existing standards for digital dictionary entries, in particular TEI and Ontolex. These have limitations for historical lexicography. The key problem is that historical dictionaries include citations that reference authoritative texts (editions and/or manuscripts). The existing standards do not handle or standardise these references in a way that allows the kind of semantic, two-way referencing that builds the Semantic Web. The Dictionary of Old Norse Prose (onp.ku.dk) was designed before the emergence of digital lexicographic standards but is built on long-standing historical lexicographic practice. The data model it uses includes two-way linking between citations, articles and sources, not only the Old Norse sources but also in many cases foreign parallels for translated works. The rich relationships between words and texts can therefore be made available in new ways. For example, without the addition of any further data, ONP can produced parallel glosses of all its texts by exploiting this two-way linking. It also uses linked data techniques to dynamically connect its own data with external projects, linking from external source digital texts to produce concordances and linking other dictionaries to supplement it where it is currently incomplete. This paper will demonstrate some of these capabilities as well as discuss the potential of extending the existing standards to encompass these possibilities.