Today's progress, on the LDev003 and LDev004 files:

1. Re-generated them from the Collate files with my application, which fixed the problems caused by commas vs semicolons etc.
2. Categorized all the readings using my app -- that works fine. I've followed the categorizations Cara added last time around.
3. Worked through Cara's comments in the original files, and fixed remaining problems where I could.
4. Added comments for all items I'm not sure about, or where there needs to be further input from Cara and/or Ray.

Sent the results to Cara, with explanation, and the following notes:

One fairly common problem relates to additions in the witnesses. The word "added" is often, but what that means is not usually clear. For example:

Perhaps the "addition" is actually an addition in the witness (which should be tagged with an <add> tag), or simply a word which exists in the witness but not in the base text (in which case it should just be in the reading). For instance, if the base text has this:

    one three

and the witness has this:

    one two three

then the app tag should look like this:

    <app>
        <lem>one three</lem>
        <rdg>one two three</rdg>
    </app>

although it could also look like this:

    <app>
        <lem>one</lem>
        <rdg>one two</rdg>
    </app>

or this:

    <app>
        <lem>three</lem>
        <rdg>two three</rdg>
    </app>

The main point is that there must be enough in the <lem> tag to show where the extra bit in the witness actually shows up.

However, if the witness itself has "two" in the form of a scribal addition, then the app tag should look like this:

    <app>
        <lem>one three</lem>
        <rdg>one <add>two</add> three</rdg>
    </app>

Does that make sense? I've tended to use the <add> tag, especially where you mentioned using it, but I suspect that usually we're not dealing with additions in the witnesses, we're dealing with bits of text that show up in the witnesses but not in the base text -- not the same thing at all.

I think these files are close enough for me to use as the basis for writing XSLT/output code for the Web view of the MS, though, so I'll start on that next week.