For what it's worth, we in the Old Norse-speaking world (or
at least that part of it in which I am active) treat this
situation (very common in our MSS) by using <supplied
reason="omitted">, the reasons being those mentioned in the
discussion, viz. that <gap> is intended for material which
is (or was) present in the MS but is not, for one reason or
another, included in the transcription, while <corr>/<sic>
are for the indication and/or emendation of errors in the
original. Neither of these is appropriate in the case of
spaces left for initials which have not been filled in,
whereas <supplied>, intended for text supplied by the
transcriber, is. We use the reason attribute, with the
values "omitted" and "illegible", to distinguish between
letters or words assumed to have been (inadvertently)
omitted by the scribe (which in our tradition would
customarily be placed in angle brackets in a printed
edition) and letters or words now unreadable but assumed
originally to have been in the manuscript (which in a
printed edition would be placed in square brackets). (We
use <supplied> only when the missing text can be
reconstructed with a reasonable degree of certainty; when
such is not the case we use <gap>.)

Makes sense to me.

Matthew

M. J. Driscoll
Arnamagnaean Institute
University of Copenhagen