The neuter is one of the three gender
s in Indo-European
languages. In the ancestral Proto-Indo-European
, all nouns belonged to one of three classes. This system was preserved in many of their descendants, such as Latin
, and German
Two of the classes are called masculine and feminine because all male and female creatures belong to them, and the third is called ne-uter "not either". But the distribution of inanimate nouns across the three genders is baffling. There are never any simple rules. For example, none of the Indo-European languages has the obvious rule saying that things that are neither male nor female are neuter. (English has the pronoun "it" for this purpose, but isn't divided into noun classes.)
Many other languages have class systems like this: Bantu languages (such as Swahili) have more than ten, North-East Caucasian languages (such as Chechen) have more than three, and Afro-Asiatic languages (such as Arabic) have two. The division into masculine/feminine/neuter is, as far as I'm aware, restricted to the Indo-European family.
There are three peculiarities of the neuter. One is in all languages that have it: that the nominative and accusative are identical. That is, the noun has the same case ending whether subject or object of a sentence.
Also, the neuter is nearly identical to the masculine: it generally has the same declension apart from the nominative/accusative.
Another peculiarity is that in Greek (and if I recall rightly in Vedic Sanskrit) a plural neuter subject is followed by a singular verb.
In Latin the commonest neuter ending is -um, plural -a. This is the so-called second declension. It is identical to the masculine -us apart from the nominative/accusative. An example is verbum 'word', plural verba. In Greek the corresponding ending is -on, plural -a, as in phenomenon and criterion. The Greek and Latin both came from an earlier Indo-European ending -om.
In the third declension one common neuter ending is -us. This is a big source of errors about Latin plurals, because these are not the more familiar second declension masculine -us whose plural is -i. Examples of this are corpus 'body', plural corpora, and opus 'work', plural opera. But there are numerous other shapes in this subclass: mare 'sea', plural maria (as on the Moon); animal, plural animalia; caput 'head', plural capita; etc. etc.
The corresponding Greek subclass is similarly misleading. The neuter ending -os has plural -ê: as in telos 'end, purpose', plural telê; anthos 'flower', plural anthê (this means those pretty coloured things in next door's garden are polyanthe, not polyanthi); epos 'epic verse', plural epê; etc. Also in Greek you get other subclasses with similar declension, including -ma plural -mata: so words like stigma and lemma belong here.
As there is no common theme running through the neuter, the origin of the gender system has always been rather mysterious. But in recent decades, studies of the deep history of Indo-European have suggested an answer. This is by no means certain, but it looks okay.
The idea is that very early Proto-Indo-European was an active language (see typology for a bit more detail). It marked nouns not for whether they were subject or object of a sentence but for whether they actively initiated the action or not.
The ending -s marked active. The ending -m marked inactive (patientive -- or you could say "passive" but that's usually associated with verbs). So some nouns, like "master" (Latin nominative dominus, accusative dominum) could be either: they could do things or have things done to them.
But others were always inactive. Neuter nouns have a slight (and only slight, but it may be archaeological evidence of an older stage) tendency to be inert: look at Latin mel 'honey', marmor 'marble', rus 'country(side)', cadaver: these don't actively do things, even if you make them the subject of a sentence.
In the plural, the neuter -a is the same as the feminine singular -a, and this is now believed to be no coincidence. It comes from an earlier stage ending in a laryngeal, -aH, and the -H marked a collective. If the neuter plural was originally collective, this explains why some languages used a singular verb after it.
It must be stressed that this analysis does not apply to familiar languages like Latin, Greek, and German. This hypothetical reason for the split into genders is deep within an early stage of the ancestral Proto-Indo-European, and once it had become established, it changed, and words switched genders for phonetic reasons and for reasons of analogy. Anyone who knows a bit about both Latin and Greek will appreciate that related words can be of different genders: knowing the Latin often misleads you if you try to guess at the Greek. In particular, the supposed "collective" meaning of the -a ending was hijacked by some female word(s) and came to be reinterpreted as a feminine ending.
The earliest known branch of the Indo-European family is the Anatolian, which includes the Hittite language. This branched off from the common stock much earlier than the rest: the group is sometimes called Indo-Hittite to emphasize this. Hittite does not have a masculine/feminine/neuter system. It has a two-class system, animate and common. It is believed that this might reflect the conditions in the common ancestor that gave rise to the three-gender system by the process explained above.