The Third Grammatical Treatise (TGT) was composed around the middle of the thirteenth century by Óláfr Þórðarson, a nephew of Snorri Sturluson. It is divided into two parts: the first is a paraphrase of standard Latin grammatical treatises (specifically Priscian’s Institutiones Grammaticae IH), supplemented with information about the tunic alphabet. The second part is a much closer adaptation of the Ars Major of Donatus on the faults of speech (ID, 1-6), illustrated with examples of Norse-Icelandic skaldic poetry. Of the poetic examples, a large amount of material is not found elsewhere and a large proportion of that is anonymous. The anonymous material consists of mostly very short fragments (usually two lines) on a variety of subjects. Despite the existence of six editions of the treatise and a number of scholarly commentaries on aspects of it spanning a century and a half, there has been no explicit attempt to identify, classify or contextualise the anonymous verses in TGT apart from one brief and one indirect attempt by Finnur Jónsson (1920-4 and Skj). This paper will begin to undertake this project. I will identify some sources, propose some reclassifications and assess Finnur’s attempts at grouping and determining the provenance of the anonymous material. A number of these verses are interesting in themselves, but the general significance of the anonymous material in 7GT lies in the information it can provide about its status and transmission. This issue was first explored by Gisli Sigurdsson at the Akureyri saga conference. A revised version of the article was published in 2000; it contains a chronology of the life of Óláfr and a detailed account of the poetry of known provenance. I therefore refer the reader there for some of the details not discussed in the present paper. Gísli concentrated on the verse of known provenance, concluding that Óláfr's oral sources—and by extension literary knowledge in general—derived largely from either the royal courts of Scandinavia or from the author’s own neighbourhood in Iceland (2000, 112-13). The present paper will show that this conclusion can be extended to the anonymous material, but requires some refinement with respect to that material.

The significance for Olafr of Norse poetry in general, and by extension that of the treatise, is made quite explicit in the opening chapter of the second section (I have used Finnur Jónsson's edition of 1927 for the sake of providing a normalised text for these pre-prints; translations are my own unless otherwise noted): Í þessi bók [Donatus, Ars Major I] In this book it may be clearly má gerla skilja, at oli er ein listin understood that everything is the skáldskapr sá, er rómverskir one art: the poetry which Roman spekingar námu í Athenisborg á orators learnt in Athens in Greece