Studia Islandica - 01.07.1963, Síða 104
102
(67). None oí the other sagas has suffered a reduction in any way
comparable with this.
As to Laxdæla, the latter table exposes its affinity with Knýtlinga
in a most striking manner. ff one assumes a quite equal distribution,
each of the five family sagas ought to account for 20% of the total
sum of pair words. In the former series Laxdœla reached 29.5% (72
out of 246), in the latter one its share has increased to 39.0% (35 out
of 90). (For comparison it may be noted that Egla in the pair word
series with Heimskringla — divided into two parts, Snorri A and
Snorri B, treated separately — showed corresponding percentage fig-
ures of 39.5 and 37.0 respectively.) Moreover, as a glance at the
table may reveal, Laxdœla has a clear preponderance over the other
four sagas in each separate word category. (The absolute figures
should not mislead us here. Thus, Njála with its 8 adjectives and 7
verbs may seem to come close behind Laxdœla with 10 and 8 respec-
tively. But adjusted according to the rule given on p. 101 above, the
figures for Njála would be 5 and 4 respectively, i.e. exactly half of
those for Laxdœla.)
As indicated above, it seemed advisable in the previous paper on
Snorri and Egla to divide Heimskringla into two parts, approximately
equal in size, thus acquiring a valuable possibility of checking the
result. In the beginning it was not intended to use such an arrange-
ment for Knýtlinga, since it seemed rather small for that purpose —
48085 words. But there is of course nothing to prevent our making the
division afterwards. Thus, pair word tables are now given for both
the first and second halves of Knýtlinga, here named Knýtlinga A
and Knýtlinga B. As for the saga as a whole, two lists are given for A
(p. 57) and B (pp. 57—58) respectively. It is particularly noticeable
that the figures are much lower in the B-series, more or less for all
the five sagas — a fact which seems hard to explain at first sight,
but which need not trouble us here. As to the crucial point, the posi-
tion of Laxdœla, the correlation between A and B is very satisfactory.
In the reduced /1-series Laxdœla accounts for 40.5% (24 out of 59)
of the pair words; in the reduced B-series its share is 36.0% (13 out
of 36). (These percentage figures may once more be compared with
those of Egla in its pair word series with Snorri A and B: 39.5 and
37.0 respectively.)
The pair word statistics here presented confirm the result of the
previous statistics for those 19 adjectives (adverbs) and 16 nouns,
characteristic of Laxdæla. Two methods — quite independent of each
other, as far as can be seen — lead unequivocally to the same con-
clusion. Henceforth quite new arguments, and very strong ones, must