> John Conway's helpful answer runs slightly astray at one point:
>
> "although the Greek prefixes tend to alter as they form compounds
> (so for instance "penta" becomes "pente" in "penteconta" = 50),
> this is less natural in English."
>
> This is the wrong way round: five is pente, fifty is penteconta (but in
> fact the 'e' in 'penteconta' is eta, while that in 'pente' is epsilon);
> in compounds of the type we were interested in (pentagon = with five
> corners) the form was always 'penta-', as in 'pentateuchos' = 'consisting
> of 5 books'.
It is because there is no distinction between epsilon and eta in English.
An example with no confusion (eta/epsilon) is the 7, 70:
Greek: 7: e(pta/ : hepta
70: e(bdomh/konta heptaconta-{-gon, -hedron}
>
> In just the same way, 'eight' = 'octo' while in compounds of this type we
> find 'octa-', e.g. 'oktaknemos' = 'having 8 spokes' So it's not right to
> say '"octagon" is the established English spelling for what would be
> "octogon" if taken directly from the Greek.'
No. It is right to say octagon. First, because octa- is a Greek prefix as
well, and secondly, for system's uniforming reasons:
tetra-
^
penta-
^
hexa-
^
hepta-
^
octa-
^
ennea-
^
all the endings are a (alpha)
>
> The Greeks will have had 'oktagon', so he who coined 'octagon' for English
> knew what he was doing. What needs explaining is why we have 'octopus', not
> 'octapus' . (The answer is that both 'oktapous' and 'oktopous' are found
> in reputable Greek writers, so even a pedant will probably admit that one
> is entitled to draw on either form.)
But the second o in the Greek octo is omega: o)ktw/
In Greek: o)kta/pus, and o)ktw/pus
I think that the octo- English prefix is not the Greek o)ktw/ but the Latin
octo- : the English octopous is from the Latin octopous < Greek o)ktw/pus
cf. English: encomium < Latin encomium < Greek e)gkw/mion
Conway's system is good, IMHO, and easily memorizable - where I disagree
is the case of 100: he prefers the type hecto (or something like that).
I think that it would be: hecatonta-; again for uniforming reasons:
triaconta-
tetraconta-
....
enneaconta-
hecatonta- (or at least: hecaton-)
Antreas