According to this sound law, an original *s changed to *š (a sound similar to English ⟨sh⟩) after the consonants *r, *k, *g, *gʰ and the semi-vowels *w (*u̯) and *y (*i̯), as well as the syllabic allophones *r̥, *i, and *u: Specifically, the initial stage involves the retraction of the coronal sibilant *s after semi-vowels, *r, or a velar consonant *k, *g or *gʰ.
In Indo-Iranian *r and *l merged, and the change worked even after the new sound; e.g. Avestan karš-, Sanskrit कर्षति kárṣati 'to plough' < PIE *kʷels-.
[4][3] A later extension of RUKI was particular to the Iranian languages: *s, *z shift to *š, *ž also after the labial stops *p, *b, including even secondary *s from Proto-Indo-Iranian *ć < PIE *ḱ.
[2][5] The Ruki rule also displays a rather different behavior in Nuristani, conditioned by the following factors:[6][7] According to Orel (2000: 62),[8] Albanian shows a limited RUKI-like development, where *s becomes sh only after PIE *y, *w (including their vocalic counterparts *i, *u).
Dash has a disputed etymology, with rival versions attributing the word not to Proto-Indo-European *dʰews-om but instead *dʰeh₁-l-,[9] or *demh₂ from *dmh₂ "to tame".
Michiel de Vaan (2015) instead has a Proto-Albanian *ʃ emerging from different means, which barely resemble a RUKI law: Indo-European *ks shares the fate of simple *s in becoming *ʃ before *t (as occurred for jashtë "outside" and gjashtë "six", but not other cases with *ks where *t did not follow), with *t as the conditioning factor, rather than the prior *k.[13] Meanwhile, the development of *s itself is highly disputed,[14] but in contrast to Orel's view that it was conditioned on a RUKI-like phenomenon, De Vaan[13] prefers Kortlandt's[15] view that *s became *ʃ when either followed by an unstressed vowel or intervocalically, regardless of the quality of nearby vowels.