Studia Islandica - 01.06.1962, Qupperneq 188

186
Even supposing that Snorri B and Egla have the same author, it is
a priori likely that Snorri B should reveal a somewhat higher fre-
quency of pair words with Snorri A. Snorri A and Snorri B as Kings’
sagas and parts of a coherent literary work have certain traits in com-
mon, which distinguish them from the family sagas and may influence
the phraseology. In fact, it is rather surprising that the difference be-
tween Egla and Snorri B on the point in question should prove so
small: Egla according to Table I lays hold of 39.5% of the pair words;
the corresponding rate for Snorri B according to Table III is 43.0%.
A certain increase one had to take into account in advance, and one
should have imagined it rather greater than 3.5%.
The statistics over the pair words have demonstrated that Egla when
put on trial against extensive comparative material reveals almost the
same traits as an authentic Snorri-text — in very sharp contrast to
four of the most well-known family sagas. If such an outcome cannot
be accepted as proving beyond doubt that Snorri Sturluson dictated
Egla, there seems to he little hope left that a linguistic study will ever
lead to conclusive results in this or similar questions of authorship.
4. The result eheeked wilh tlie aid of Hallfreðar saga, Heiðar-
víga saga and FóslbræSra saga. (Pp. 30—36). A possible objection
to the conclusion of the pair word statistics could be, that the demon-
strated linguistic affinity between Egla and the Snorri-texts might de-
pend not on their being written by the same author, but on chronolo-
gical vicinity. According to the prevailing opinion among scholars they
both belong to the oldest stratum of the saga texts in question — from
the beginning of the 13th century.
However, the strength of such an objection can be tested. It may be
done by using further saga texts of admittedly early origin for com-
plementary tests. HeiSarvíga saga, HallfreSar saga and FóstbrœSra saga
have here been chosen for that purpose. As to the dating of these sagas
and the editions used see p. 31.
The method employed in these tests is as follows. The three new
sagas regarded as a whole (48 000 words) are excerpted. The number
of pair words found there of those which have already been listed in
the Appendices I—XII and counted in the Tables I—III, is established.
The original pair word lists are correspondingly reduced. If this con-
frontation should result in a considerable elimination of pair words
between Egla and Snorri A (or Snorri B) or between Snorri A and
Snorri B, the conclusions above might have to be revised. On the
other hand, if Egla and Snorri B do not lose proportionally more pair
words than the other sagas, i.e. Laxdæla, Eyrbyggja, Njála and Grettla,
this must be interpreted as an important confirmation of the original