Richard Stibbons mentioned that the rolls he obtained with his Hupfeld
Animatic-S piano contained a combination of round and square punches.
These sounded familiar, so Richard posted me one of his rolls for
inspection.

 [ See Richard's article and the roll image, and hear the MIDI files
 [ at http://mmd.foxtail.com/Sounds  -- Robbie

These rolls were probably produced in the UK by the "Up-to-date Music
Roll Company", which I believe originated as an offshoot of the UK
branch of Imhof & Mukle.  They match other examples I have seen or have
in my oddments box.  The company seems to have specialised in copying
or creating rolls for coin-operated pianos, as well as the more usual
domestic product.  Keith Prowse, whose name is on the piano, probably
had rolls made for the UK market.

"Up-to-date" rolls are seen on various labels, dating (so far as it is
possible to guess) from the 1930s through into the 1950s.  The earlier
ones usually are identified as products of Up-to-date ('Muvis' and
'UTD' labels, for instance), but the later ones are more enigmatic.
Typical later products are piano rolls on 'Parex' or 'Pianostyle'
labels -- the latter uses a copy of the original American Pianostyle
label, but carries an address of 'Chadwell Heath, Essex', a suburb to
the east of London!  The musical contents and numbering of these two
labels is identical.  Information on them is non-existent.

The use of a mixture of ordinary round punches and oblong punches is
quite distinctive.  The oblongs are about 2 punch rows long, and are
used only in instances where a note of precisely that length is
required.  They are not used in chains or extended notes.

It's hard to know precisely why the rolls were made in this way.
It would seem that the complexity of using two separate punches would
outweigh the saving from reducing the number of punch rows.  I wonder
if anyone has any idea of what sort of machine they used to make these
rolls?

Anyway, I would be reasonably confident that this odd style of
perforation is a whim of the roll maker, not an intrinsic part of
the expression system!

Julian Dyer