In Notker’s writings, <p, t, k> are found in word-initial position if the word is preceded by an obstruent. The graphs <b, d, g> representing Old High German /b̥, d̥, g̊/ and <t> representing Old High German /t/ are found word-initially following vowels and sonorant consonants. In a radically new interpretation of Notker’s Anlautgesetz, Lahiri/Kraehenmann (2004) argue that Notker used the graphs <b d g> word-initally for derived geminate allophones after sonorants and <p, t, k> elsewhere for underlying geminates. In contrast to Lahiri/Kraehenmann (2004), the present article argues for the traditional analysis that Notker’s rendering of initial OHG /b, d, g/ as <p, t, k> reflects a neutralization of the contrast between OHG lenis /d̥/ and fortis /t/ when preceded by an obstruent, the same pattern found in modern Swiss German (Moulton 1986). It will be argued on the basis of graphemic and historical evidence that Notker maintained a three-way contrast between /d̥, t, tt/. Therefore, phonological length alone is not sufficient to describe the contrasts between stops in Notker’s dialect of Old High German.