Guidelines, p. 388 - "<wit>[unattested]</wit>"

I agree that this example seems wrong, not only because of the unlicensed use of special notation but more seriously because the existence of the hypothetical reading is not relevant to the point which this set of examples is discussing, namely the use of the reading group element. I suspect that expert advice may be needed on how to improve it however (if we can get Peter Robinson's attention again...)

Why wouldn't one want to use <witness> for conjectural witnesses? i.e. MsA was copied as MsB and MsC was copied from that, but MsB isn't extant. We assume the existence of MsB by the textual variations that are evident in the extant witnesses, but what would be wrong with having a <witness> that isn't extant?

-james

Agreed to revise and expand example: firstly, use a straightforward witness; secondly add an example for the "unattested" case.

James: "what would be wrong with having a <witness> that isn't extant?"

Perhaps it is just my positivist inclinations that lead me to believe that witnesses attest to something "real" (which can be transcribed and interpreted in different ways, of course), whereas a conjecture only attests to the ingenuity of the editor, who proposes a theory about a reading, a reading which may never have been actualised in any document.

Anyway, I did not raise any point about <witness> (which describes witnesses), but about <wit> (which refers to witnesses). According to the Guidelines, "<wit> contains a list of one or more sigla of witnesses attesting a given reading, in a textual variation." and this does not fit the notation "[unattested]".

More generally, I would like to see clearer advice in the Guidelines about how to encode conjectures within <app>. This both applies to any conjectures one might make oneself and conjectures made by others that one wishes to transcribe.

Jens makes a good point: the guidelines need to be clearer about this (cf. my last comment on ticket http://purl.org/tei/bug/3496958\).

In EpiDoc (which almost never concerns itself with MS variants, but often attributes conjectures and records deprecated readings/conjectures), rdg is used with @resp to record a reading, interpretation or conjecture by a specific editor. I think we need to *EITHER*

(a) give examples of and discuss this usage in the TEI guidelines, *OR*

(b) consider inventing a new element <cnj> or similar to record conjectures in <app>. But is this now overlapping with the designs of the appcrit WG?

Concretely, for this ticket, I propose to: (1) replace the example Jens objects to with a case involving a real witness, and (2) add a second example of an unattested witness (if someone can remind me how we agreed this should be done).

Council agress that Gaby's proposal a) below is what makes most sense for now.

-James

This ticket has been assigned to me for some time, but I confess I'm not sure exactly how to fix it. Should the <rdg> in question just have an @resp instead of @wit? That seems the most straightforward solution.

(Note that this example occurs three times in the prose as well as once in the elementspec. Should we change all 4 at the same time?)

After offline discussion with James, and looking back at my suggestion a bit further above, it looks like this is indeed the best solution. I have therefore changed the line:
<lem>Eriment<wit>[unattested]</wit></lem>
to
<lem resp="#ed2013">Eriment</lem>
in all four instances (at [r12274]).