In the new version of our data, we extend the kinds of parallels from the old “full” and 'partial" to include:
- Full (a text or section of text parallels another text or section of text)
- Resembling (one text “resembles” another)
- Retelling (a story or event is “retold” somewhere else)
- Mention (a text or passage is “mentioned” somewhere else)
The new system makes it fairly easy to extend these types of relation.
Our recent discussion of T 121 raises the possibility of another kind of relation:
- Fragmentary (only fragments of a text exist)
In the case of SN 16.3 = T 121 etc., one of the parallels is SF 74. This is a Sanskrit fragment that includes only a small portion of the text. Even the part that is available is about 30% reconstruction. So the parts that actually parallel the sutra text are in reality just a few lines.
This is currently listed in parallels.json with ~sf74, indicating a resembling parallel. However, it seems, in fact, to be pretty much the same as the Pali text, or it would be if it existed in full. To identify it by section seems overly complex and also not entirely accurate, as a sectional parallel is for sections of text transcluded in others, not for fragmentary survivals.
Would we be justified in creating a new class of relation for such cases? By specifying it as a “fragmentary parallel” we are clearly indicating the kind of relation.
The problem is, of course, that such fragments exist on a spectrum, from a scrap of just a few characters all the way to mostly complete texts. So the texts would have to be defined in a somewhat arbitrary way; something like:
- A fragmentary parallel is used for cases where the text only exists in fragments that are significantly less than the full text.
For example, SF 74 would count as a fragment, whereas SF 73, which contains a full text and just a few reconstructed phrases, would not be a fragment.