Numbers: Some Options

I see five options for our numeric system, each with advantages and disadvantages:

1) A separate basic root for each number at each decimal place, i.e. toru = "four," pima = "forty," etc.; pima toru = "forty four."

Pros: It's very clear, no possibility of misunderstanding.
Cons: It uses up far, far too many roots and requires an unreasonable feat of memorization on the part of the learner. Unacceptably ridiculous, in short.

2) Each numeral has a monosyllabic root, and these are simply stacked to build complex numbers. So if sa = "1," he = "2," pu = "5," then hesapu = "215."

Pros: Numbers are very short and simple.
Cons: This works unlike any other derivational process in Koa; it's kind of weird how "100" and "10" aren't overtly mentioned. Despite the fact that this was my initial idea back in 2001 or whenever it was, I don't think it feels right at this point.

3) Each numeral and decimal place has its own bisyllabic root, e.g. sulu = "four," kumi = "ten," súlukumi sulu = "forty-four."

Pros: Very clear; both numeral and decimal place are phonetically salient.
Cons: Resultant forms are the longest of any of our options, especially as we keep going -- súlupima súlukumi sulu = "444," etc. Perhaps more seriously, all numbers above ten have four syllables, the only four-syllable words (I believe) in Koa.

4) Each number is composed of two CV roots, one for number and the other for decimal place; these are arranged in sequence for building. Thus if su = "four," to = "x10^0" lu = "x10^1," then sulu suto or maybe sulu su = "forty four."

Pros: Number roots are nice and short, easy to remember and work with.
Cons: We have to choose between eliminating the vast number of resultant bisyllabic forms from our root array on the one hand (unacceptable), or having a large number of roots that mean one thing in numerical context and something else elsewhere. Sulu, for instance, could mean "forty" when followed by the counting word pi as in sulu pi cumo "forty squashes," but "helmsman" elsewhere: Ka sulu i si suo sulu pi cumo = "The helmsman ate forty squashes." Languages typically do just fine even with huge numbers of homophones (cf. Mandarin), so this might be okay.

5) Each base number has a bisyllabic root, which is then modified by a monosyllabic suffix to indicate decimal place. Thus, if sulu = "4," -ma = "x10^0" and -ho = "x10^1," then súluho súluma or súluho sulu = "44."

Pros: We're using a minimum number of roots, retain monosignificance, and have highly distinct, easily understandable and memorable forms. We retain the basic Koa derivational system, although the suffixes are being used in a different way than they would with nominal roots.
Cons: Numbers are kind of long -- súlune súluho sulu = "444" as opposed to e.g. sune sulu su -- though not ridiculous. Are they distinct enough from each other?

I suppose length isn't something to which we should attach prime importance given, for example, seitsemänsataa kahdeksankymmentäyhdeksän which seems to work just fine as "789" in Finnish.

I think we can throw options 1 and 2 away outright, which leaves us 3, 4, and 5 to choose between -- we'll think it over. In the mean time, note that either way words of quantity need to be separated from the noun by pi, with a slew of unanswered questions, chief among them being whether pi counts as an article. Is "four squashes"...

sulu pi cumo
sulu pi a cumo
a sulu pi cumo
a sulu pi a cumo

In parting, I'm pleased to see that, no matter how we end up doing this, we should have no trouble constructing phrases along the lines of E-o du instruistoj da studentoj, e.g. sulu mehe pi sahi = "four men's worth of beer."