The numerals and derived numbers of the Proto-Indo-European language (PIE) have been reconstructed by modern linguists based on similarities found across all Indo-European languages. The following article lists and discusses their hypothesized forms.
The cardinal numbers are reconstructed as follows:
Other reconstructions typically differ only slightly from Beekes and Sihler. A nineteenth-century reconstruction (by Brugmann) for thousand is . See also Fortson 2004.
The elements (in the numerals "twenty" to "ninety") and (in "hundred") are reconstructed on the assumption that these numerals are derivatives of *deḱmÃÂ¥(t) "ten".
Lehmann believes that the numbers greater than ten were constructed separately in the dialect groups and that originally meant "a large number" rather than specifically "one hundred."
The numbers three and four had feminine forms with the suffix , reconstructed as and , respectively.
Special forms of the numerals were used as prefixes, usually to form bahuvrihis (like five-fingered in English):
The ordinal numbers are difficult to reconstruct due to their significant variation in the daughter languages. The following reconstructions are tentative:
The cardinals ending in a syllabic nasal (seven, nine, ten) inserted a second nasal before the thematic vowel, resulting in the suffixes and . These and the suffix spread to neighbouring ordinals, seen for example in Vedic "eighth" and Lithuanian "ninth".
Reflexes, or descendants of the PIE reconstructed forms in its daughter languages, include the following.
In the following languages, reflexes separated by slashes mean: