Studia Islandica - 01.06.1962, Side 187

185
quarter of the frequency for Egla. Perhaps the relationship stands out
most clearly in percentage figures. With a fairly equal distribution of
pair words in the five family sagas, each of them ought to get 20%
of the total number. But of the 489 pair words (according to the ad-
justed column to the right) Egla lays hold of 39.5%, Laxdœla 18.0%,
Eyrbyggja 16.5%, Njála 9.5%, and Grettla 16.5%. In other words,
Egla takes twice and Njála half as big a share as they should have
done by an equal distribution of the pair words. The remaining three
sagas are a few per cents below 20. — Still more significant, perhaps,
than the total numbers for all categories is Eglds outstanding pre-
dominance within each category separately. If the shares are expressed
in percentage figures, Egla gets of the total sums (adjusted figures) of:
adjectives 33.5%; nouns for persons 41.5%; verhs 39.0%; ahstract nouns
43.0%. (Pp.26—27).
Tahle II presents the number of pair words between Snorri B on
the one hand, and the same five family sagas as in Table I on the
other. If this checking comparison with another voluminous Snorri-text
gives the same clear outcome as the confrontation with Snorri A, it
must be regarded as a very strong confirmation of the validity of the
method used and of the reliability of the main result. Though Snorri B
from some points of view is less suited than Snorri A for a direct com-
parison with the family sagas, the new test reveals much the same pic-
ture, and thus corroborates convincingly the unique affinity between
Egla and Snorri. (Pp. 27—28).
Table III, finally, is in some respects the most important and con-
clusive one. If the method used is to be regarded as decisive, Snorri A
and Snorri B ought of course to exhibit an especially high frequency
of pair words. Thus one has to arrange a confrontation between the
two Snorri-texts. This is done in Table III and commented upon on
pp. 28—30. One notices that Egla has disappeared from this table. The
reason is that as a possible Snorri-text Egla might complicate the di-
rect comparison between Snorri A and Snorri B. As the statistics have
been arranged, Egla and Snorri B in respectively Tables I and III are
confronted on identical terms with Snorri A on the one hand and the
remaining four family sagas on the other.
A glance at Table III should be enough to melt even the most stub-
born scepticism. The profile of Snorri B in Table III could hardly
look more like that of Egla in Table I. Both of them present more
than twice as high a total frequency of pair words with Snorri A than
any other saga can show; in both cases Laxdœla comes next. Egla as
well as Snorri B demonstrate a marked preponderance within all cate-
gories, though not quite as pronounced in the groups adjectives and
verbs.