The Web, with its explosive growth, is becoming an efficient resource for up-to-date information for the scientific researcher. Informal online archives are repositories for technical reports. Proceedings are more and more commonly published on the Web. The collection of online journals is growing. Indeed, a good number of online journals are "born digital". Many researchers simply put their papers up on their own web site. The large volume of online material makes it quite desirable to be able to access cited documents immediately from the citing paper. Implementing this direct access is called "reference linking". Some reference linking services exist today. A number of commercial publishers, recognizing the significant value-added nature of reference linking, have banded together to form the CrossRef organization. The CrossRef publishers share their metadata, which enables them to interlink their journals. This metadata is not, however, available without a fee to organizations or individuals outside of CrossRef. The vast majority of online scholarly literature is accompanied by little or no metadata. Since it is desirable to link up this literature as well, the problem of automatically reference linking online scholarly literature in the absence of metadata and author intervention is a problem very much worth considering. This paper explores this problem in detail, and presents some algorithms for extracting metadata from online texts and linking full-text documents together. The extent to which reference linking of the online literature can be done automatically is therefore the main topic of this paper.