Studia Islandica - 01.06.1962, Qupperneq 190

188
As in the preceding chapter the original pair word series are con-
fronted with the new text and reduced in a similar way. If the affinity
between Snorri and Egla should depend mainly on the fact that Egla
has a more “historical” character than the other family sagas, then
the confrontation with Oddr’s work ought to result in a comparatively
greater loss of pair words between Snorri and Egla than between
Snorri and the four other sagas. Whereas, if no such effect can be
demonstrated, that would be strong support for the opinion that Snorri
and Egla have important linguistic traits in common, quite irrespec-
tive of possible points of contact in their topics.
The pair words eliminated when the Snorri A-series (App. I—IV)
is confronted with the new text are enumerated on p. 38. The losses
will reduce the figures of Table I to those which are given in Table
VII. Of 193 pair words Egla has lost 28 or 14.5%. The remaining four
sagas have lost in all 48 or 17.0% of 285 pair words. Proportionally
this means a little “gain” for Egla. In fact, Egla’s share in the total
sum of pair words has increased from 39.5% in Table I to exactly
40.0% in Table VII. These numbers reveal an astonishing stability in
the relations which appeared in the original pair word series.
The corresponding confrontation with the Snorri B-series and the
latter Snorri /1-series is demonstrated on pp. 39—40 and 40—41, and
the resulting new pair word numbers are listed in the Tables VIII
and IX respectively. They only stress once more the extraordinary
constancy of the profile which stood out originally.
To sum up, the new test demonstrates that a supposed greater like-
ness in matter between Snorri and Egla has not played any conceiv-
able róle in the unique affinity which the pair word series display.
6. Further checking of the method. Are tlie pair word series Iielp-
ful in tracing out mutual relations between the family sagas? (Pp.
41—51). In the two preceding chapters an attempt has been made to
fix some imaginable sources of error, which could possibly affect the
main result. But neither the proximity in time between Heimskringla
and Egla, nor Egla’s contact with the topics of the Kings’ sagas can
be shown to have had any statistically relevant effect. The testimony
of the pair words remains just as convincing.
However, it would be of great interest if one could also prove the
usefulness of the method in questions other than that of authorship.
Is it possible that the statistics of the pair words might reveal impor-
tant facts about the mutual relations between various sagas, and per-
haps contribute to the solution of chronological problems in that field?
The argument here is founded on the fact that there exist certain
generally accepted connections between the sagas. Thus leading scho-