Semi-

Numeral or number prefixes are prefixes derived from numerals or occasionally other numbers. In English and other European languages, they are used to coin numerous series of words, such as unicycle – bicycle – tricycle, dyad – triad – decade, biped – quadruped, September – October – November – December, decimal – hexadecimal, sexagenarian – octogenarian, centipede – millipede, etc. There are two principal systems, taken from Latin and Greek, each with several subsystems; in addition, Sanskrit occupies a marginal position. There is also an international set of metric prefixes, which are used in the metric system, and which for the most part are either distorted from the forms below or not based on actual number words.

In the following prefixes, a final vowel is normally dropped before a root that begins with a vowel, with the exceptions of bi-, which is bis- before a vowel, and of the other monosyllables, du-, di-, dvi-, tri-, which are invariable.

The cardinal series are derived from cardinal numbers, such as the English one, two, three. The multiple series are based on adverbial numbers like the English once, twice, thrice. The distributives originally meant one each, two each or one by one, two by two, etc., though that meaning is now frequently lost. The ordinal series is based on ordinal numbers such as the English first, second, third. For numbers higher than 2, the ordinal forms are also used for fractions; only the fraction ½ has special forms.

For the hundreds, there are competing forms: those in -gent-, from the original Latin, and those in -cent-, derived from centi- etc. plus the prefixes for 1–9.

Sesqui- is used in Latin combinations for 1½ (e.g. sesquicentennial) and quasqui- for 1¼; multi- and poly- are used in Latin and Greek combinations, respectively, for 'many' (multilateral, polygon).

*For Latinate 21, 22, etc., the pattern for the teens is followed: unvigint-, duovigint-, etc. For higher numbers, the reverse order may be found: 36 trigintisex-. For Greek, the word kai ('and') is used: icosikaihena-, icosikaidi-, pentacontakaipenta-, etc. In these and in the tens, the kai is frequently omitted, though not in triskaidekaphobia. (The inconsistency of triskaidekaphobia with the table above is explained by the fact that the Greek letter kappa can be transliterated either "c" or "k".)

...
Wikipedia