This article is from the sci.lang FAQ, by Michael Covington (mcovingt@ai.uga.edu) and Mark Rosenfelder (markrose@zompist.com) with numerous contributions by others.
-Miguel Carrasquer Vidal (adapted by markrose)]
Despite what one might think, these are independent developments.
The English s-plural comes from the PIE o-stem nominative plural ending *-o:s,
*-o:s, apparently extended in Germanic to *-o:s-es by addition of the PIE plural
suffix *-es (*-o:s itself comes from *-o-es). This *-o:ses became Proto-Germanic
*-o:ziz or *-o:siz, depending on the accent, which gave the attested forms--
Gothic -o:s, Old English -as, Old Saxon -os, and Old Norse -ar (with the change
to words that were not a-stems, a tendency which has since become nearly
universal.
The n-plural of German is generalized from the PIE n-stems (*-on-es --> -en).
It was still present in Old English n-stems, and survives today in a few words
like 'oxen'.
The Romance s-plurals (-as, -os, -es) are derived from the accusative (PIE
*-a:ns, *-ons, *-ens). Old French still had separate nominative and oblique
(accusative/ablative) forms, but in the end, grammatical cases were dropped
completely, and usually only the oblique forms were retained.
In Italian and Romanian, final -s was phonetically lost, and the plurals are
based on the nominative. The Latin nominative plural, at least in the o- and
a:-stems, was based on PIE *-i, of pronominal origin, not *-es as in most other
IE languages.
 
Continue to: