The first point is, in my opinion, having clear the purpose of what one is doing. Does one want to reconstruct the author's text? The one commented upon by its main commentator? The one copied by a well-known intellectual?… The question amounts to a basic dichomoty:
- 1. Reconstruction of the Ur-text (with Urtext understood lato sensu, as every possible stage of the text, provided that it is linked to a precise person, time and place).
- 2. Reconstruction of the Tradition.
In the second case, it will make sense to list as many variants as possible, virtually also all variants. They will help one to throw light on, e.g., the fortune of the text, the kind of copysts who copied it, their geographical and social origin, etc.
In the first case, by contrast, I would suggest not to list variants which do not make sense and/or which are due to obvious misreadings.
For instance, in the first case I would not list variants such as sakti instead of śakti, kamma instead of karma, or even na instead of sa if the context is clear and they are very similar in the manuscript. In the second case, it would be interesting to note that the text had been copied in a region where palatal ś were pronounced as s, by copysts who were not extremely accurate in reproducing consonant clusters and possibly in a telegu (or similar) script.
If, as often the case, one wants to do both, one runs the risk of producing a critical apparatus so full of variants, that it will be neglected altogether by readers, etc. Hence, I would add a couple of additional suggestions:
—If the text has never been published before, I would avoid producing an unreadable edition, i.e., one in which there is only one (or 2, 3, 4) line(s) per page and the rest of the page is occupied by variants.
—Many, possibly most, insignificant variants could be included in the manuscript's description. One could explain there that the copyst of X tends to duplicate t-s or tends to confuse voiced and unvoiced stops, etc.
—If the number of insignificant variants is enormous, but the manuscript(s) one is working on is very important for the reconstruction of, e.g., the cultural history of a certain area, one could consider producing a separate diplomatic edition.
—I would always produce a translation of the text I have edited, so that readers can better see the rationale of my editiorial choices.
As for conjectural emendations, as already explained, I tend to think that the received text is "innocent until proven guilty". I would not emend it if it makes sense as it is and would anyway emend it as little as possible. Obviously enough, the degree of emendations strictly depends on the kind of manuscripts one is working with. But also on the purpose one has (virtually no conjectures, if one wants to reconstruct the tradition over the text).
If you want to read the post which lead to this one, with the interesting comments of an anonymouys reader, click here.