Rafał L. Górski wrote

[...]

> 1 .There are superfluous passages, which are deleted *by the editor*, for
> purely philological reasons, that is they do not make any sense in this
> particular context.

[...]

I think you will need get a clearer statement from the editor what is going
on and why before you can make any definite encoding decisions. "Deleted by
the editor" here presumably means that the editor does not want the passages
to appear as part of the body text visible to a reader in the default view
or presentation. That is a somewhat special usage of "delete" though, as is
emphasised when you go on to say that the editor apparently requires the
"deleted" text to be marked up so as to include corrections to it. In other
words, it is not meant to be deleted at all, but merely excluded from one
view of the text, but retained, along with some level of editorial
annotation/intervention, in some other view of the text which the markup is
required to encompass and retain.

If that is the case, exactly how this might best be encoded is rather
dependent on what is meant by saying "they do not make any sense in this
particular context". Presumably, since these passages are there in the
source, they must have made some sort of sense to someone, unless they are
the outcome of gross mechanical errors of a kind normally associated with
more modern print technologies than you are dealing with. So, again making
the best guess on the information so far provided, it sounds as though the
claim that they do not make sense is an editorial judgement rather than
something that could be understood as an uncontentious recording of an
indisputable state of affairs. If that is so, they their content may belong
in critical apparatus of one kind or another. Without a clearer idea of
what is intended, it is hard to say more.

If it does emerge that something related to P4's sic/corr approach is
appropriate, this would seem to be a very plain case for considering the P5
alternatives. As you say, in the sic/corr way of doing things, one view is
encoded as the value of an attribute. It is hard to see this ever being
appropriate where the material concerned can be described as a "passage"
(which, physical extent apart, implies it is highly likely that it will need
to contain markup and so not be feasible as an attribute value). It is
pretty plain that the devisers of these tags were thinking of fairly short
strings, maybe even less than a lexical word in many instances, when they
proposed this mechanism. P4 does have suggestions for how to deal with
alternative views involving " large or heterogenous spans of text", viz the
use of the various segmentation and alignment methods, but as far as I
can see P4 has little to offer for the sort of intermediate situation you
are describing. Hence the attraction of P5's "choice" element, for which
see the Alpha P5 Guidelines Chapter 6. Because this uses element text
content for both views, it is capable of dealing with text containing markup
and of any desired length (as well as making it straightforward to render
alternative representations under user control if the material is being
delivered interactively). If this approach seems to meet your needs, but
you are not willing or able to consider moving to P5, there is no reason in
principle why the <choice> element should not be introduced to P4 by way of
customisation. (Although the idea of incorporating a "choice" mechanism into
P5 may well have been inspired by the way "choice" is used in some flavours
of schema, it is in no sense dependent on validation via schema as distinct
from DTD, so there is no obstacle to inserting it into a P4 DTD on that
score).


> I encode a text which is preserved in one manuscript, however
> some pages fall out of it and were found separately. In the book
> edition fragments contained by these separate pages are marked
> eg. _here starts the page found by Jan Kowalski_. Maybe I wasn't
> reading the P4 carefully enough, but I could not find any advice for
> such a case.

Again, more clarity about what the case actually is would make it easier to
offer advice. What, precisely, is being encoded? If the answer is "the print
edition" then the "marking" you refer to is just one sort of editorial
intervention, and how best to encode it depends on how you choose to encode
such things in general. It might be a simple as a <note>. If you need to
mark the boundaries of the restored folia without disrupting markup that
flows across them, you might want to make use of linked <milestone>s and/or
<anchor>s to demarcate the restored material. But basically, the task is to
translate into encoding the way the print editor chose to insert an annotate
the restored folia. This is a different question from the more general one
of how best to markup restored folia when preparing an original digital
edition from MSS sources. However, the phrase "I encode a text which is
preserved in one MS..." seems to suggest that you see yourself as encoding
the actual MS using the print edition as a species of witness to the text
rather than your canonical source. While doubtless feasible, that route has
many pitfalls, and I'm not sure the Guidelines could offer much detailed
advice on following it.

Michael Beddow