586 
Proceedings of the Royal Society 
ch, and gave a single reversible sound. “ Ho,” when turned back, 
sounded distinctly “ och.” 
Since both vowels and consonants sound alike backwards and 
forwards, it appears that articulate speech may be divided into a 
series of successive parts, each having a character of its own, and 
each reversible so far as sensation is concerned. 
The separate reversible sensations correspond in the main to the 
letters of our alphabet, but our letter i represents at least two rever- 
sible elements, and ch , ?^y, or th , on the contrary, are single sounds. 
Clear, clean sounds, as spoken by educated people in all countries, 
seem remarkable for the small number of the elements that enter 
into each word. An uneducated man, speaking a dialect or patois , 
will make each nominal vowel consist of a large group of simple 
reversible sounds running into one another. 
6. The wave form for the letter r has been recorded. It agrees 
with that obtained by Donders on the phonautograph, and consists 
of a series of simple curves, similar to harmonic curves, which gradu- 
ally increase to a maximum, and then gradually decrease to a 
minimum. The number of waves separating the maximum from the 
minimum intensity of the sound was in the examples observed 7 or 
8. The length of the whole period was therefore 14 or 16. 
7. The wave forms obtained from the letter o in figs. 4 and 5 
illustrate the difficulty which the ear, even of trained musicians, 
feels in determining the octave in which a given note is sung. It 
will be seen that each period consists of two very nearly equal parts, 
and that the amplitude of the second partial has obviously been 
large as compared with that of the prime tone. To one ear it might 
seem natural to call the pitch that of the second partial, since this 
was the loudest tone ; to another the pitch would be that of the 
prime, because this tone was lowest in pitch. Naming a given note 
by the lowest of its component tones, irrespective of the relative 
intensity of the partials, seems to be a convenient but conventional 
rule. 
8. The instrument as at present made does not record all sounds. 
This is shown by the fact that its reproduction of the French u is 
very imperfect, and ee is only sometimes well heard from it. 
The curves laid before the Society were obtained by multiplying 
the traces on the tinfoil some 300 times by means of a system of 
