In historical linguistics, the law is a sound law in Proto-Indo-European where word-medial consonant clusters containing underlying tend to lose one of the nasal consonants or . This sound change was first documented by Johannes Schmidt in 1895 and is named for the Avestan reflex . The law, along with Stang's law, is one of Proto-Indo-European's manifestations of a phonotactic restriction against multiple adjacent sonorants in the coda of a syllable.
Affected affixes.
The law is usually invoked to explain the disappearance of in the oblique case forms or thematic derivatives of animate-gender and nouns, which have cases where a sequence should appear. Thematic derivatives of neuter nouns were also affected.
Phonological conditioning.
The conditions under which the law is triggered remain controversial and unsettled. It is usually supposed to be triggered when the sequence follows a consonant; if the is preceded by a short vowel, the rule does not apply.
Byrd has losing the if the preceding syllable was accented and the if the following syllable has the accent. This account is rejected by Tijmen Pronk. Alexander Nikolaev also rejects the post-tonic -loss part of this account but instead posits that rule applied to delete either one of or (more often ) when the next syllable is accented. The law would also be generalized to thematized derivatives of any nasal-stem noun, including -heteroclitic nouns, where the base noun would lose the stem-final nasal in composition.
Dissimilatory loss of in when the root contained a labial consonant has been used to explain many cases of the rule. Pronk posits this dissimilation as the sole mechanism of the rule.
Adiego alternatively proposes that the rule was conditioned not by the accent, as Byrd believes, but instead by the ablaut grade of the root; the zero grade would cause the deletion of but with other grades, the would be lost instead.