I was interested to read Bryan Cather's thread about these the other
day as I too had been recently mulling over the same points he raised
mainly in response to Robert Perry's having recently made available
a MIDI scan of the original.  It's always nice to revisit the Joplin
material since I believe so many of us came to love mechanical music
through one ragtime revival or another -- I certainly did!

Someone long-ago took two unrelated facts (Joplin was syphilitic
and Joplin's rolls sound uneven) and made a flawed false connection
(Joplin's rolls are uneven because he had syphilis).  Joplin's
declining health is oft-quoted as a reason for the uneven nature of
these recordings.  I say that's just folk repeating rubbish writing
from years ago -- it simply doesn't withstand scrutiny.

Pianists that loose the ability to play like that lion at the keyboard
they once were simply do not deteriorate to play in the manner heard
on these rolls.  Keyboard note accuracy technique goes, the ability to
play complex runs and riffs dwindles, the wide scope of playing narrows
but the sense of time typically stays.  If Joplin was so ill his timing
was that sloppy believe me there'd be a lot more wrong with these
performances than just sloppy timing.  Listen to the audio recordings
Eubie Blake made in the 1970s and you'll hear what I mean about
declining technique.

Whilst I can't support the "Joplin was so ill his rolls stink" theory,
I can support the "bad recording" theory.  I don't dispute any of
Robert Perry's perforation technical statistics as they are facts.
What I can say for certain is that by 1916 and despite Aeolian have
been selling hand-played rolls for 4 years it's another fact they most
definitely had not mastered the art of it.

Whilst Robert has in mind "hundreds" of hand-played recordings of the
period that are "noticeably smoother" there are also large quantities
that either are equal or frankly worse.  They can't all have had
syphilis surely?  Or did they?  "Parisienne" played by Tierney is one
such example if you need one to go hunt for.

I am making full allowance for the fact that the pianist might also
have been lousy to start with however it is a fact that there are
numerous rolls of lively 2/4 and 4/4 music that display the same odd
timing faults that the Joplin rolls show.  The sloppy timing is an
artificial mechanical recording artifact: a timing fault which living
pianists of any ability (pr lack thereof) simply don't do.  These
subtle wow-and-flutter ripples even extend through the first several
years of Duo-Art roll recordings, all the (related) Uni-Record and
Metro-Art series, Connorized and even the QRS 100000 series.

Whilst some are better than others there are plenty examples available
to listen to as midis off the various roll scan web sites so go download
them, compare and listen for yourself for the proof.  You will soon
find even Felix Arndt playing with exactly the same loping tempo as
Joplin.

Basically, as J L Cook is rightfully quoted, it took an inordinate
amount of effort to straighten these out.  So much effort, in fact, that
roll companies abandoned all hope of using this method reliably for
dance music which leads you on to all the hand-played strictly
metronomic dance rolls of the 20s and 30s.  They're delightfully
perfect and we all have come to love them but no live pianist ever
plays like that however good.

The fault is with lining up the rows per inch of one recording against
another and paying off the error margin here and there.  It's like
overlaying two big graph paper grids one with squares every 4 inches
and the other every 5.  Every so often their lines coincide and the
rest of the time they don't but are closer or further apart.  Where the
lines coincide the editor gets it spot on.  The rest of the time
placing the note is made to the nearest "best fit" position.  The
compromise is a little out and so you hear the slight tempo
wow-and-flutter even where the original recording might have been
perfect.

This is all that there is wrong with the Joplin rolls.  If you count
1-2-3-4 in strict time over the irregularities of the roll you will see
that you arrive that the final bar of each section where you expect to
without extra half bars of hesitation or anything else drastic.  If you
use MIDI editing software you will know that this is "quantizing" in
modern terminology and you can actually replicate this loping tempo
sound by badly quantizing any live-recorded MIDI quite easily.

How can we hear what Joplin really sounded like?  Unfortunately we
don't have the original that came right off the cutter as he played.
On QRS roll 9725 you can hear a slightly cleaned-up version of Joplin's
other performance of Maple Leaf.  People have slated this but I think
it deserves a second listening.  I assume it to be the work of J
Lawrence Cook.  Whilst we can't ask Cook (he died in 1974) close
inspection of 9725 does seem to show it as a little labor of love on
his part - he'd already done a purely Cook version (roll 7308) earlier
on anyway.

What Cook didn't know about jazz or ragtime piano
performance probably wasn't worth knowing.  Whereas Connorized and
Aeolian had at best 5 years experience with the machinery Cook probably
had over 30 by the time he did 9725.  The note lengths and playing
style are very close to the Aeolian version - crisp short and punchy.
This perfectly suits the piece which (being from 1899) is close to the
sound of a banjo rag being played which the earliest piano ragtime
originally sought to follow to a degree.  Everything fits perfectly.

What else is apparently odd with 9725 apart from the bluesy final chord
(which Cook added as his own stamp on it clearly)?  There is a very
very slight swing to the rhythm! Again, Cook's decades of experience
may have been able to decipher this from the Connorized and Aeolian
rolls.  He may have seen (from his own experience of working with the
same sort of machinery) that the results of the early editors were an
early inability to replicate this very slight swing into the final
master roll which translated into a jerky strict tempo instead.  There
are precedents for this sort of thing happening elsewhere with other
roll brands : compare Artempo's J P Johnson "Carolina Shout" recording
with the QRS version.

As to Joplin playing a piece by Handy I don't find this unusual.
Whilst Joplin is today such a cultural icon we would, if we could go
back in time, perhaps ask him "Why play that?" about the Handy piece.
The reality is that in 1916 Joplin was just a fading pianist / composer
and probably would have responded "Why not?"  People wonder the same
thing about Gershwin playing Kern on piano roll etc.  By 1916 Handy
must have been considered more than "up-and-coming" -- a few full years
since his Memphis Blues went on sale.

 [ Frank Himpsl and others have independently surmised that Scott
 [ Joplin composed the second strain of "Memphis Blues" for W.C. Handy,
 [ which might explain why he recorded it.  -- Robbie

With regard to the tempo of rags and the over-academic meal some folk
make of all this, I despair.  Ragtime music has two charms: harmonic
and rhythmic.  Most rags focus on one or the other (Joplin) but some
have a heavy load of both (James Scott).  Where the writing is harmonic
play the music slow enough to enjoy those qualities.  Where the music
is essentially rhythmic it should be played not so slow that the
rhythmic charm falls apart.

As a pianist known for playing ragtime for years that is the total sum
of my academic conclusion on the music -- about 999 pages less than
Perfessor something-or-other on the web and 85 bound volumes less than
various university music professors fabricating ragtime's academic
detail where there is none.  Ragtime is fun to be enjoyed.  I never
ever saw anyone who sat listening to it like it was a profound Mahler
symphony.

"Maple Leaf Rag" is essentially rhythmic and comes to life at a more
brisk tempo.  "Gladiolus Rag" (which in notes-on-the-page terms is
often compared to "Maple Leaf") has a higher load of harmonic beauty
so benefits from being played slower to reveal it's charms.

Many rags can be played acceptably at a range of speeds but not all.
From a pianistic point of view playing Joplin is easier than Scott.
Whilst you can afford not to rush the technically not inconsiderable
"Gladiolus", for much James Scott you need to play with the same
technical efforts but fast enough to enliven the rhythmic charms side
of the music.  This is probably why one hears Joplin more than Scott,
which is a shame in many ways.  As to roll tempos, they are usually
appropriate in line with the music considering all these factors.

This really applies to all music : if you sense the piece is essentially
rhythmic, play the tempo at the speed that best releases that quality.
If the piece is non-rhythmic and stacked full of melody and wonderful
harmony then play it slow enough to release all of that beauty.  If the
music has neither of these it's Stockhausen or Lloyd-Webber.

Regards,
Adam Ramet
http://www.themodist.com
http://pianola.forumer.com
http://www.playerpianogroup.org.uk