There's something I remember being attributed to Dr Johnson, but I can't confirm that it is. Treat it as a joke or an urban legend until someone can find it in Boswell or find the correct source. Anyway, someone -- I always imagine Edna Everage glasses and a strong Texan accent (sorry) -- says, "Oh, Dr Johnson, tell me, is the plural of octopus, octopi or octopodes?"
To which he replies, "Madam, that shows an ignorance of three languages."
First point. Latin does not form the plural of -us by changing it to -i.
This is true only for second-declension nouns, which are the majority of those ending in -us, but there are plenty in the third declension (corpus ~ corpora, Venus ~ Venera, opus ~ opera), and plenty in the fourth declension (status ~ statûs, hiatus ~ hiatûs). In Latin, the second declension -us ~ -i was never generalized into the other declensions. It was and remains wrong to form *corpi or *stati, Classical Latin, Mediaeval Latin, or New Latin.
The Greek word pous 'foot' had plural podes. The Greek oktopous was taken into Latin as octopûs (not octopus) and would have been a third-declension noun with plural octopodês.
Now in post-Classical Latin, the distinction between long and short vowels was lost. Octopûs would have changed to octopus. At this point it would have looked like a second-declension noun -- in the nominative singular anyway. They might have lost the declension octopodem, octopodis etc. and created a new one octopum, octopi etc. This actually happened with the related word polypûs, which has given modern 'polyp', a sure sign that it had strayed into the second declension at some point.
To sum up: 'octopodes' is Greek, 'octopi' isn't Latin (or is bad post-Classical Latin, and scientific New Latin always strives to use good Latin)... and 'octopuses' is English. It's an ordinary English word, not technical, so there's no reason not to use the English plural.
As for a collective of octopuses: an army? a handful? a clutch?
