[6] There is no trace of it at all in Gothic, where the distinction between the short high and mid vowels had become allophonic (Proto-Germanic /e/ and /i/ merged).
[8] Variation is found within dialects too with doublets such as Old English spora : spura "spur", spornan : spurnan "to spurn", cnocian : cnucian "to knock"; Old Icelandic fogl : fugl "bird", goð : guð "god", goll : gull "gold."
[11] The mutation is rare in Old Norse, e.g. verr "man", heðan "hence", neðan "from below" in contrast to niðr "down(wards)"[1] and perhaps jafn "even."
Instances where a-mutation has failed to occur in Old Norse can mostly be explained as analogical forms,[12] although a palatal stop /ɡ/ or /k/ immediately preceding the /i/ in a short-root syllable has a tendency to block or reverse the process.
In the West Germanic variety that gave rise to Old English, a-mutation did not affect the second element of the diphthong */eu/ (for which the earliest Old English texts have eu): treulesnis "faithlessness", steup- "step-" (Epinal Glossary 726, 1070); but in other branches of West Germanic */eu/ eventually became */eo/ unless followed by */w/, e.g. Old Saxon breost "breast" vs. treuwa "fidelity.
Old English derives from a type of Germanic in which single */m/ had the same effect on preceding */u/ and */e/ as a nasal stop followed by another consonant.
"[19] Lloyd, for example, proposed an alternative explanation for all apparent instances of a-mutation of */i/; he suggested that "the partial overlapping in Germanic of the two phonemes /i/ (represented in all environments by [i]) and /e/ (with the allophones [e] and [i]) led to the occasional development of an e-allophone of i by systemic analogy".
[2] Cercignani, on the other hand, argued that "no 'umlaut' phenomena can be assumed for Proto-Germanic", preferring to ascribe these changes to "the prehistory of the individual languages.