Studia Islandica - 01.06.1962, Side 185

183
not give a correct idea of the relative frequency, because of the vary-
ing length of the texts, Egla having 62000 words, Eyrbyggja 38 000,
and Njála 97 000. Using the ordinary rule of three formula (which ad-
mittedly is an oversimplified model in the present case), and employ-
ing Egla as the standard of comparison, we should get the following
“corrected” number of pair words for Eyrbyggja:
50 x
--------- = ---------; x = 82.
38 000 62000
For Njála the “correction” would lead to a reduction, thus:
50 x
--------- = ---------; x = 32.
97000 62000
So in this case the numbers 100:82:32 would present a more correct
picture of the affinity between Snorri A and the three sagas in question,
than would the absolute numbers 100:50:50.
Certain implications of the method employed and some other possi-
bilities of classifying the material for statistical purposes are discussed
on pp. 17—18. The choice of “pair words” is motivated on pp. 18—20.
Firstly, these words form a fairly well defined category. They are
clear-cut statistical entities in the texts; in principle there can be no
doubt when we have to do with a pair word or not. Secondly, and this
is the important point, the pair words have been chosen with a parti-
cular view to the specific character of the actual texts and their manu-
script tradition. Pair words, of course, are relatively rare words —
though they need not be striking in any way. It seems probable that
such uncommon words should have been more exposed than others to
manipulations hy copyists — owing to negligence, thoughtlessness or
conscious choice of a more current word or phrase. There is little
reason to suppose that ordinary copyists would have changed from the
usual to the unusual; the contrary is more likely. This, however, im-
plies that such comparatively uncommon words as the pair words are
more likely than the average vocubulary to represent the original read-
ing. (Cf. the argumentation on p. 181 in this Summary.)
The bulk of nouns for concrete things has been omitted here. These
words are often closely connected with special occupations and sur-
roundings. They depend mainly on the subject matter of the text in
question, but are less significant of an individual phraseology. The rest
of the vocabulary has been divided into four different categories: 1. Ad-
jectives, both concrete and abstract. 2. Nouns denoting persons or groups
of persons. 3. Verbs or verb constructions. 4. Abstract nouns. The ma-