If the spec stays as is, I think the guidelines and this note in the 
spec for <teiCorpus> should be changed:

    Must contain one TEI header for the corpus, and a series of TEI
    elements, one for each text.

The note says <teiCorpus> /must /contain such and such and the data 
model says it /need not /contain that.

I’ve opened a new issue on GitHub titled “teiCorpus: make guidelines and 
data model consistent.”


On 10/1/2018 10:58 AM, Syd Bauman wrote:
>> It does seem to me that a <text> in <teiCorpus> without a
>> <teiHeader> would be an ill-formed encoding.
> Absolutely, you should not have a <text> in <teiCorpus> without a
> <teiHeader>, and indeed the content model does not allow that
> combination. It does allow just
>    <teiCorpus>
>      <teiHeader> <!-- ... --> </teiHeader>
>      <text> <!-- ... --> </text>
>    </teiCorpus>
> which is a bit silly, because it could have just been <TEI>, not
> <teiCorpus>. But also nothing particularly wrong with it if the
> content of <text> is, in fact, a corpus. :-)