A different group of six young Royal Navy enlisted men listened two periods a day for a week to each of three tests, made up of nine different complex steady state sounds. On one test all nine sounds had fundamental frequencies of 55 Hz; on the second, 110 Hz; and on the third, 220 Hz. Three of the nine sounds were completely meaningless and remained so on all three tests. Three sounds were segments of the intoned vowels, a, i, and u when the fundamental frequency was 110 Hz. When shifted in frequency one octave the resultant sounds were perhaps vowel-like but were not a, i, and u. Three sounds were the tones of musical instruments when the fundamental frequency was 220 Hz. Subjects identified each sound by an arbitrarily assigned number.

The group that heard the “real” vowels always made fewer errors in assigning numbers to them. When all sounds were shifted upward or downward one octave, errors among the three classes of sounds were equally divided for the first test session. At the end of one week there were fewer errors on the vowel-derived sounds sounding down one octave and on the meaningless sounds sounding up one octave at 220 Hz.