Previous page: Two Authors
The above scenarios involved only two authors, and also assumed that text A was not itself copied or edited from any other texts. What happens when there is a third text, created by a different author, aC, and A, B, and C each contain several passages (e.g., descriptions, events, and stories)? Each text includes passages not included in the others, and also common passages. By extension from the scenario involving two authors, if C is created after both A and B, it could contain text copied or edited from either or both A or B:
C can contain text copied word for word from either A or B. Barring mistakes, the copied text in C will be identical to that originally in the text from either A or B. However, because B’s text was originally partly copied or edited from A, and partly written by aB, the text in C that was copied will vary in style depending on what part of B’s text was copied;
C can also contain edited portions of texts A and B. Because some of B’s text includes some of A’s text, the profile of the edited portions in C will be a combination of the profiles of A, B, and C.
C can also contain text written independently on exactly the same subjects as were previously written by aA and aB. The independent text in A, B, and C will have the profiles of the three authors, with no influence from each other.
We can therefore see that when three authors (aA, aB, and aC) produce text on the same subject, the profiles of these three texts will vary even more than when there are only two authors, again depending on how much one author copies from or edits the text of another. With three texts each category can be represented by a three-digit number, with the digits representing text in A, B, and C respectively.
As with two texts, a ‘2’ represents words appearing in a text, a ‘1’ represents a place in a passage in one text which does not include a word used in the parallel passage in another text, and a ‘0’ represents the absence of a parallel piece of text, as shown in the Venn Diagram above. With three authors each text is split into 9 categories (instead of only 3 when there are 2 authors), for a total of nineteen categories, depending on whether each of the authors either write their own text, or copy or edit text from one or both of the others. Taking text A as an example, with two authors A could be split into c20, c21, and c22, which together was termed c2X, and with three authors these three categories are each split into three depending on the actions of aC, so that:
A = c2XX (or c20X + c21X + c22X); and
c20X = c200 + c201 + c202
c21X = c210 + c211 + c212
c22X = c220 + c221 + c222
One of the main findings when just considering two authors was the ‘ownership’ of the shared category, c22. The same still applies with an additional author. Whatever copying or editing takes place between A and C, or B and C, we can still apply the same tests to determine the ‘ownership’ of c22, except that with three authors c22 becomes c22X (where the 'X' means any of the three possible values, i.e. c220, c221, and c222) because we are not concerned with the words in C, and a similar change applies to the other categories.
The results of these tests are not affected by the actions of aC. However, as described above c22X consists of c220, c221, and c222, and by combining these in other ways we also have c22A (= c220 + c221), and c22N (= c221 + c222), all of which do depend on the actions of aC. Therefore, there are additional tests that can be performed that do take account of aC’s actions. For example:
These tests can be used to determine direction of transmission between A and B in just those parts of A and B that do not have parallels in C. In the same way as with only two authors the ‘X’s can be replaced by 1’s, 2’s, A’s, or N’s respectively to determine direction of transmission when aC has copied or edited from A or B:
1’s: Words in passages in A or B that have parallels in C, but where C has different words;
2’s: Words in passages in A or B that have parallels in C, where C has the same words;
A’s: Words in passages in A or B that have no parallels in C, or where C has parallels containing different words;
N’s: Passages in A or B with parallels in C, where C has either the same or different words.
This gives a total of 48 tests when comparing A and B, which may provide additional information. For example, we may be able to determine that aB edited A in passages for which there are no parallels in C, but that aA edited B in passages for which there are parallels in C.
We can employ the same tests as with A and B, except that, for example, c22 becomes c2X2.
Then, as before, the X’s can be replaced by 0’s, 1’s, 2’s, A’s, or N’s to determine the effect of the actions of aB.
In this case c22 becomes cX22, (etc.), and again the X’s can be replaced, this time corresponding to the actions of aA.
As with correlations involving just two authors, it is possible that the styles of the various categories may not correlate with each other, or perhaps that the correlations are mixed and provide inconclusive results, and (again) we may not be able to tell the chronological order of A, B, or C without further analysis.