John Cowan <[log in to unmask]> writes:
> Finnish is interesting here:  8 and 9 are "two less than *deksan*"
> and "one less than *deksan*", but "deksan" (which is obviously IE)
> is not the Finnish for 10.

Hmm?  I thought one of the lemmas of 'yksi' is 'yhd-' and one of
'kaksi' is 'kahd-'.  The numbers in certain forms show great
similarity to the '-deksan' numbers.

   'Kahden hengen'           = 'of two beds'
                               (phrase in genitive case)

   'Kahdeksi yöksi/päiväksi' = 'for two nights/days'
                               (phrase in translative case)

   'Kahdeksan markaa'        = 'eight Marks'
                               (phrase in nominitive case,
                               'markka' in partitive case)

Analogously: 'yksi'/'yhded'/'yhdeksi'/'yhdeksän'.

My analysis up to now was that 'kahdeksan' was something like 'two
off' / 'two away', without mentioning 'ten' in any way (no second stem
for 'ten' in that word), just some ending: i.e., the that
'-eksan'/'-eksän' was a certain ending on the lemmas 'yhd-'/'kahd-'.

Your analysis would surprise me, because there are no lemmas
'yh-'/'kah-', but only 'yhd-'/'kahd-' of these numbers IIRC.
There are 'yksi-', 'yhd-' and 'yht-'.

Maybe some finno-ugrist could clarify.  I'm confused now.