QUOTE(BusyBee @ May 20 2007, 04:41 PM)

I thought it would be 8 semiquavers written with a 6 over the top ie 8 notes in the time of 6 but just looked this up in the AB Theory Book One by Eric Taylor.
It suggests a dotted crotchet can be divided into 4 semiquavers with a 4 over the top so perhaps 8 semiquavers can be two groups of this. 4 notes in the time of 3 twice. It's on p40.
I think this adds up.

The figure on top of the notes should be the number of notes actually notated (and not what the time value they should be played in). Thus, the figure on top should always be 8.
There is no rationale for writing two groups of notes of 4 as the beat is a dotted crotchet and not half a dotted crotchet.
QUOTE(sonataform @ May 20 2007, 04:43 PM)

QUOTE(Frederic Chopin @ May 20 2007, 03:42 PM)

That would be a publishing house's and performer's/sight reader's nightmare...!

Not sure. I'm sure I've encountered something like that in sightreading (but can't remember where) and not had a problem with it.
QUOTE(Frederic Chopin @ May 20 2007, 03:42 PM)

But that does not add up to a dotted crotchet!!!
Hrmph. No, it doesn't, does it? 8 dotted demisemis then.
Yes, that would be an alternative. Not convinced it would be popular with the publishing houses.