Gripla - 01.01.1975, Qupperneq 71

MANIFESTATIONS OF RAGNARS SAGÁ LOÐBRÓKÁR 67
the two variants. As a result of this procedure, Mageröy finds that the
number of words in common seldom exceeds 33V6% of the total
number of words in either of the two variants, and that when it does
exceed this figure, the circumstances are exceptional—either the tale
is of the chain-tale type, in which the element of regular and rhyth-
mical repetition is likely to give rise to a greater similarity of wording
between variants than would be found between variants of other types
°f tale,7 or—as happens in one case—one of the two variants, the one
containing a percentage of words in common higher than 33Vs%, is
exceptionally short (comprising only 66 words) in comparison with
the other (comprising 224 words).8 * A brief comparison of certain sec-
tions of variants of the Norwegian popular legend about Knut Skrad-
dar, which Mageröy also carries out, shows that the number of words
in common between these sections of variants does not exceed
33V6% either.0
Mageröy then points out that, in view of the freedom which writers
of medieval texts often felt in relation to their exemplars, surviving
texts of scribally interrelated written versions of a saga may some-
times show, in parts, few, if any similarities. It is always possible in
theory, therefore, that extensive differences between surviving texts
reflect written rather than oral variation. The pairs of folktale variants
examined by Mageröy nevertheless suggest, in his view, that in the
case of short narratives at least, oral variation is a possible alternative
to written variation where the number of words shared in common by
the surviving saga-texts is limited to roughly one-third or less. If the
surviving texts or parts of texts have more than roughly one-third of
the words in common, on the other hand, it is likely that they reflect
scribally interrelated written versions of a saga or story, and thus
provide examples of litterœr skyldskap, or rittengsl, particularly if the
texts in question can be shown to have in common many series of
more than six words in sequence, and several whole sentences.10
Mageröy then goes on to calculate the percentages of words in com-
7 See Mageröy, 237, 240.
8 See Mageröy, 244.
0 This is based on the variants of the legend printed by Liest0l in his Norske
œttesogor (1922), 169-82.
10 See Mageröy, 247-48.