Conclusions about DIV types:
(a) That they are primarily useful for navigation in a book and understanding of it, and only secondarily useful as a means of searching or of limiting searches. The main reason that they are not more useful for retrieval is the lack of control on the vocabulary used for TYPEs. If we could establish greater control, either ourselves or by use of an external controlled vocabulary, we could improve the reliability of TYPEs for retrieval. We've made some efforts on this, at least with regard to frequently recurrent features (imprimatur, letter, poem, tract, play, sermon, license, publisher's catalogue, errata, title page, half title, appendix, preface, to the reader, encomium, etc.)
(b) That the usefulness for navigation is to some extent potential rather than real because much of the information is suppressed in the current interface. But imagine an alternative interface. Imagine one, for example, in which HEADs are suppressed, but TYPEs and Ns are always displayed. We would like the book to be navigable and intelligible in that form of display as well.
So ignore the presence of headings, and the quirks of the current display, to a large extent in deciding how much detail to give in the TYPE and N.
(c) That the hierarchical ordering of the DIVs allows information
about the parent to inform our understanding of the siblings. This means
that if it is obvious from
the parent (and/or the N values) what the parts are, and there
is no convenient name for the subordinate parts, one may take the subordinate
div as relative to the
superordinate DIV.
E.g. <DIV1 TYPE="index">
<DIV2 TYPE="part"
N="A">
<DIV2 TYPE="part"
N="B">
</DIV1>
makes perfect sense. There is no need to invent, say <DIV2 TYPE="lettered part of index">. The obverse is also true, however: if the superordinate DIV does not supply much information, then a greater burden is placed on the subordinates to do so.
E.g. <DIV1 TYPE="miscellany">
<DIV2 TYPE="part"
<DIV2 TYPE="part"
is less useful than
<DIV1 TYPE="miscellany">
<DIV2 TYPE="song">
<DIV2 TYPE="proverb">
<DIV2 TYPE="letter">
So think of the intelligibility of the outline as a whole when trying to decide how much detail to give at any particular DIV level.
(d) That a DIV should be typed more distinctively insofar as it is useful (in making the structure of the book intelligible) to distinguish DIVs from one another in the same book. The vague bit here is 'useful': it is probably more useful in *retrieval* to use more generic terms (e.g. poem and hymn, rather that epic, epode, ode), but probably more useful in *analysis* to use the more specific terms. And sometimes it seems more useful in navigation to slide over slight differences in the interests of grouping like things with like. Do we want to say:
<DIV1 TYPE="poems">
<DIV2 TYPE="part">
<DIV2 TYPE="part">
<DIV2 TYPE="part">
<DIV3 TYPE="section"
<DIV3 TYPE="section"
<DIV1 TYPE="poems">
<DIV2 TYPE="poem">
<DIV2 TYPE="poem">
<DIV2 TYPE="poem">
or
<DIV1 TYPE="poems">
<DIV2 TYPE="ode">
<DIV2 TYPE="pastoral">
<DIV2 TYPE="epigram">
<DIV2 TYPE="epithalamion">
<DIV2 TYPE="songs">
<DIV3 TYPE="sacred song">
<DIV3 TYPE="secular song">
All represent the structure intelligibly, but the last clearly gives the best 'flavor' of the structure and allows much more intelligent navigation.
Maybe we might want to think about including multiple keywords.
We already do that in some of our phrasal type: TYPE="list" vs.
TYPE="list of ships"; "list" is included in both. But we could
expand this, to say:
<DIV1 TYPE="poems">
<DIV2 TYPE="poem (ode)">
<DIV2 TYPE="poem (pastoral)">
<DIV2 TYPE="poem (epigram)">
<DIV2 TYPE="poem (epithalamion)">
<DIV2 TYPE="poems (songs)">
<DIV3 TYPE="sacred song">
<DIV3 TYPE="secular song">
(e) Finally, the question of all-in-one DIV1s. Actually, this is perhaps really two questions: what to do when the body is essentially undivided, and the DIV1 is only a formality, i.e. <BODY><DIV1>..[no divs]..</DIV1></BODY>; and what to do when there is a DIV1 coterminous with <BODY>, but it is itself further subdivided. In the latter case, the subordinate DIVs potentially pass some information back up to the parent.
TEXT
SERMON
SERMON
SERMON
MASQUE
JEST
POLEMICAL TRACT
/TEXT
obviously says more than
TEXT / TEXT
But in either case, the TYPE of the toplevel DIV1 is something a little different from the other DIVs, since it has no sibling and bears in only a limited vertical way on the structure of the whole. Its structural role is essentially always 'the whole thing'. For that reason, and also the practical one that is even more difficult to genre-type whole books than it is to structure-type parts of books, and that the genre-typing of whole books is more the responsibility of cataloguers creating catalogue records than it is ours, I have been reluctant to include much specific TYPing of the all-inclusive DIV1, and recommended that in most cases one simply uses TYPE="text".
Nevertheless, this is something that we can perhaps
really do something about if we wanted to. We would need a list of genre terms
that are simple, few,
distinctive, and inclusive enough that their application should be
straightforward in most cases. And they should represent so far as
possible the actual classification that contemporaries would have
used. E.g.:
POEM
PLAY
LETTER
BALLAD
ROYAL PROCLAMATION,
EPISCOPAL PROCLAMATION,
PARLIAMENTARY PROCLAMATION,
WRIT
CHRONICLE (different from history?)
TRACT
TREATISE (how does distinguish a treatise from a
tract?)
SCIENTIFIC TREATISE
ALCHEMICAL TREATISE\
OPTICAL TREATISE
THEOLOGICAL TREATISE
MISCELLANY (but what about collection?)
ALAMANAC (but what about emphemeris?)
POLEMIC
We could do this, but someone would have to find and decide on the scheme. And if we wanted to have multiple keywords, we might want to use a different tagging scheme altogether, and add tags to the TEXT tag rather than TYPEs to the DIV1.
E.g. <TEXT>
<KW>DRAMA</KW><KW>TRAGEDY</KW><KW>MASQUE</KW>
<BODY>
rather than
<TEXT><BODY><DIV1 TYPE="drama tragedy masque">
or <TEXT>
<KW>POLEMIC</KW><KW>LETTER</KW>
rather than
<TEXT><BODY><DIV1
TYPE="polemical open letter">
The latter would confuse the navigation function. Perhaps
this kind of generic or subject indexing is really a different kind of thing
from the navigational TYPing
and in fact would interfere with the latter in some cases.
So the suggestion is: use a very limited list of controlled genre terms for the all-in-one DIV1; use only one (except perhaps with parenthetical qualification), and otherwise use TYPE="TEXT".
*TYPE="argument" -> summary
*TYPE="bible verse examination" -> type (in the theological sense,
in a book discussing the "types of Christ")
*TYPE="biogrpahy" -> biography
*TYPE="catalogue" -> list of ... (wardens, substances, materia medica)
|
publisher's advertisement
*TYPE="contents" -> table of contents | summary of contents
*TYPE="ditty" -> song
*TYPE="editions of books cited" -> list of editions cited
*TYPE="imprint" -> colophon
*TYPE="index of chapters" -> table of contents
*TYPE="life" -> biography
*TYPE="list of contents" -> table of contents
*TYPE="listing of conspiratours" -> list of conspirators
*TYPE="meane" -> mean
*TYPE="measures" -> list of weights and measures
*TYPE="names of authors" -> list of authors
*TYPE="poem for the author" -> encomium
*TYPE="poem for the composer" -> encomium
*TYPE="poem in figure" -> concrete poem
*TYPE="poem to the author" -> encomium
*TYPE="synopsis" -> summary
*TYPE="table of Ireland" -> index of Ireland
*TYPE="table of Scottish people and places" -> index of Scottish persons
and places
***
In Hooke's Micrographia (account of investigations with
a microscope), changed
<DIV2 TYPE="section"><HEAD>Observ. VII. ...</HEAD>
<DIV2 TYPE="section"><HEAD>Observ. VIII. ...</HEAD>
etc.
to
<DIV2 TYPE="observation" N="7"><HEAD>Observ. VII. ...</HEAD>
<DIV2 TYPE="observation" N="8"><HEAD>Observ. VIII. ...</HEAD>
etc.
And in long summary at end (in <BACK>), changed
<DIV1 TYPE="summary">
<DIV2 TYPE="item">
<DIV2 TYPE="item"> etc.
to
<DIV1 TYPE="summary of observations">
<DIV2 TYPE="observation" N="1">
<DIV2 TYPE="observation" N="2">
etc.
In Psalm commentary, changed
<DIV1 TYPE="biblical commentary">
<DIV1 TYPE="biblical commentary">
<DIV1 TYPE="biblical commentary">
<DIV1 TYPE="biblical commentary">
etc.
(which was no help to navigation)
to
<DIV1 TYPE="commentary" N="Psalm 34">
<DIV1 TYPE="commentary" N="Psalm 35">
<DIV1 TYPE="commentary" N="Psalm 36">
<DIV1 TYPE="commentary" N="Psalm 37"> etc.
Also, where two pages were missing, including the beginning of the notes
on Psalm 103, closed a div and opened a new one, to make it clear
that the bit after the missing pages belonged to the commentary on Psalm
103, not 102. Also provided a heading for this section by
stealing a running header. (Don't be afraid to use running title as a clue
to what kind of TYPE to assign to a division.) Like this:
<GAP DESC="missing" EXTENT="2 pages">
</DIV1>
<DIV1 TYPE="commentary" N="Psalm 103">
<PB ...>
<HEAD>The ciii. PSALM.</HEAD>
<P>...
***
Question: I've used "mittimus" as a DIV Type. Should I choose something simpler, like document?
On the one hand, 'mittimus' is what it is called in the book, and is an entirely legitimate term; on the other, it might be thought obscure enough to cause confusion as to what is being inserted. Since a 'mittimus' is a kind of writ, expand the TYPE and say TYPE="writ of mittimus" or TYPE="mittimus writ."
This is in general. Within this book, there is probably little need to
be very specific, either with this document or with the other insertions,
since they mostly have very specific heads that make it very clear what they
are: <HEAD>Mittimus</HEAD> <HEAD>Examination and Confession</HEAD>
etc. Since specificity would duplicate the info in the HEADs, you could get
away with "document.".
***
Question: Should we distinguish publishers' and printers' lists of books with other things labeled 'Advertisement'?
Yes. Publishers' and printers' catalogues should be given TYPE="publisher's
advertisement"; other kinds of thing headed "advertisement" should normally
be given
some other TYPE, e.g. TYPE="notice" or TYPE="publisher to the reader" (etc.).