Robbie's musings on the rollography makes me want to put down here
some of the notions that I have taken for granted on this subject,
and which probably stem from my experience editing Charles Davis
Smith's Welte catalog and from 30 years of book cataloging.

I'd call the work a catalog, if it's aim was limited to listing and
indexing in an organized fashion a particular person's or organization's
holdings.  The degree to which the catalog would attempt to be
authoritative, noting and correcting errors of spelling, wording,
and attribution, would be an ad-hoc decision; it would not be essential
for the utility of the catalog.

I think of a rollography as a systematic listing of all known rolls
intended to form an authoritative guide to the subject.  Which means
that source errors would be noted and corrected.

Some important data fields which ought to be included in a catalog
or rollography are:

* manufacturer/label/number series
  roll or item number
  title
  composer(s)
  performer(s)
  type of composition (waltz, overture, march, etc.),
  dating (of roll certainly, possibly also of underlying composition)
  recording location information
  location in collection (for inventory catalog)
  known owners (for a 'locating' or 'existence' list
  physical condition/characteristics (for catalog)

 [ * The rollography of Welte rolls contains duplicate numbers
 [ issued by different companies, thus "label" is needed.
 [
 [ "Recut" or "Original" roll should be in a catalog, since this
 [ indicates quality and replacement value.  -- Robbie


Anybody in his right mind who is undertaking a large catalog or rollo-
graphy had better give serious consideration to the method he chooses
to support his data base.

It would be a serious blunder to use an unsophisticated word-processor
type of system, with the result that one would create several concurrent
data files (one for roll number, one for title, one for composer, one
for performer) and enter each roll separately into each file.  This would
not only quadruple the work, but also make it impossible to carry out
corrections uniformly in each of the files, when it becomes necessary --
as it certainly will -- to correct a tile or a credit, not to mention the
fact that human error can creep in 4 times as often as it would otherwise.

The data base needs to be capable of taking one entry per roll and, by
means of tagging or fielding of the various elements in the entry, be able
to generate the needed listings or indices from that master data set, and
to handle the alphabetization automatically (including skipping in the
alphabetization the initial articles "the," "an," and "a.")  It is a BAD
compromise to have to omit the initial articles from the listings because
the chosen software isn't capable of sorting them properly.  And it is,
in my mind, unacceptable to list titles under the initial articles.  Of
course, if the list contains titles in foreign languages, the initial
article problem embraces more than the three English articles.

I leave out for now discussion of basic catalog structure questions as
provision of cross references to lead users from where something isn't,
but could be, to where it is.  Also standardization of spelling of Russian
proper names, like Chaikovsky or Tschaikowski or Tchaikowsky, [which vary
depending upon which Western European language was used when the Cyrillic
was transliterated into the Roman alphabet.]

The idea of reading into a recorder for transcription by a clerk is
certainly efficient, but there are dangers -- especially in handling
the kind of abstract and non-contextual data like you will be dealing
with. I can imagine how a clerk, hearing "Creole Belles, by J. Bodewalt
Lampe" through the headphones, will render these data on his keyboard.
Possibly "Creole Bells, by Jay Bodeywalt Lampey" or "Three Ol' Belles, by
Jabo DeWalt Lampie" or who knows what.  And in order to deal with the
specific spelling of Russian composer names, the name will have to be
spelled out by the speaker. This problem is more critical in rollography
production than in catalog production, I think.  But it exists to a degree
there too.

These are some ideas to consider.

Matthew Caulfield

 [ Reading the title is a redundant check, actually.  If the roll number
 [ is correct the full title will be taken from the rollography file.
 [ -- Robbie