¹ Andrade makes no attempt to deny this fact.

² As in some of the cases mentioned by Andrade.

³ L'analisi elettroacustica del linguaggio, Milan 1934.

⁴ Concerning the usefulness of the work for the determination of phonemes, consider the following extracts from a review by Treviño, American Speech 10.220 f.: ‘Vowels have the greatest number of distinguishers, voiced consonants have fewer, while some voiceless consonants have none. . . . [a] is the most complex, while [i] and [u] are less so. . . . The vowels [e] and [o] are considered combinations of the three fundamental vowels because of the variability observed in them. . . . The technique employed apparently does not differentiate between open and close vowels. This does not substantiate the fine discrimination claimed for it. . . . The linguistic material consisted of isolated vowels, twenty-five words, and ten phrases. . . . In general, neither the linguistic material nor the type of subjects used seems to furnish a broad enough basis for the conclusions reached on the structure of speech.‘