[The three main types of Indo-European roots follow one and the same pattern in their vowel alternation; the differences they present historically must be the result of phonetic changes. The monosyllabic long-vowel roots (heavy bases) are generally recognized as a special case of the short-vowel roots. In the ablaut system of dissyllabic roots, the second syllable seems to be ignored. It must have arisen either (1) from vocalization of an old laryngeal, or (2) from metathesis of the full vowel in a number of old monosyllabic roots containing a sonant.]

In discussing another problem, Monro 141–2 warns us against allowing ourselves to be led ‘into the fallacy of thinking that relations which are distinct to us, because expressed by different language, were distinctly conceived by those who expressed them all in the same way;—the fallacy, in short, of supposing the distinctions of thought to be prior to the language which embodies them.‘ The warning seems to me of special value in dealing with my present problem.

[Two allophones of Germanic k became separate phonemes after the i-umlaut had changed the distribution of palatal vowels in Proto-Old-English. Not the phonetic development of palatal [k'], but the distributional change involving velar [k] caused the completion of the phonemic split.]

[It is well established that some assibilation of Germanic k occurred in Northumbrian as in other dialects of English, but the extent to which non-initial k was affected has remained open to question. A study of the complete place-name material indicates that in the north of England k was retained as a stop in final position. In the county of Northumberland only, medial k was assibilated before i/j, and in the sequence -ik- + palatal vowel; elsewhere in the North medial k was in all situations retained as a stop.]

[The reconstruction of PMP Dewha explains the vowel of Jv. -ro ‘two’ and the first vowel of Tg. -lawà ‘two’, both heretofore considered to be irregular.]