Studia Islandica - 01.06.1962, Síða 184

182
and Magnúss saga berfœtts (4700 words), but not by Ólafs saga helga
(91 300 words), which is the very core of Heimskringla and, among
the Kings’ sagas, is the one which for many reasons is most comparable
to the great family sagas. Thus the need still exists for a large-scale
linguistic investigation. The present study is a new attack on the pro-
blem Snorri-Egils saga, differing in scope and method from the ear-
lier ones.
2. The scope and inethod of the investigation. (Pp. 14—23).
First a few words on the texts examined. Snorri is represented by the
whole of Heimskringla. But the voluminous work has been divided
into two parts of approximately the same size. The first one, containing
Ólafs saga Tryggvasonar and Ólafs saga helga, is referred to as Snorri
A; the second one, containing the short prologue and all the rest of
Heimskringla, is referred to as Snorri B. The main purpose of this
division is to provide a desirable, and indeed necessary, possibility of
checking and testing the result.
Besides Egils saga (Egla) the study includes the family sagas Lax-
dœla saga, Eyrbyggja saga, Njáls saga (Njála) and Grettis saga
( Grettla).
To use statistics one must know the exact size of the texts employed.
The present investigation is concerned with the prose only. Thus the
Skaldic poetry, both in Heimskringla and the family sagas, is not taken
into account. The number of words for the various texts, excluding the
poetry, is presented in the list on pp. 14—-15. In the column to the
right the sums have been rounded off to the nearest thousand; the
latter figure should be exact enough for the present statistical purpose.
Unlike Wieselgren’s and van den Toorn’s studies the present one is
based exclusively on the vocabulary, and does not deal with syntactic
traits. Previous observations on points of contact between vocabulary
and phraseology in different sagas have usually been of a rather spor-
adic and unsystematic character. The lack of a sufficiently broad basis
and of a statistical examination of the material allowed of no safe
conclusions.
The method introduced here aims at determining how many words
in the whole bulk of the texts occur in Snorri, on the one hand, and
in one only of the five family sagas consulted, on the other. The num-
ber of such words —- named pair words — is interpreted as a
measure of the affinity between Snorri and the saga in question. Let
us consider a fictitious case, and suppose that when Snorri A and Egla
are compared 100 such words can be found in them, whereas when
Snorri A and Eyrbyggja are compared 50 can be found, and when
Snorri A and Njála are compared also 50. However, these figures would