GRAPHIC RECORDING OF SPEECH VIBRATIONS 
89 
was deposited in soot on a revolving tambour. But the soot 
record is not wholly satisfactory because different portions of the 
flame deposit separate superposed records not wholly agreeing 
in form, as only the lower edge of the flame gives a true sinusoid. 
So far as published, Brown’s apparatus was applied only to a 
few simple vowel sounds and its power of recording finer and 
higher speech sounds remains to be tested. But the extreme 
sensitiveness of the manometric flame as demonstrated by 
Nichols and Merritt, makes it probable that his apparatus could 
be developed and used along this line. It would therefore be 
of the utmost importance if this method could be so developed 
as to eliminate the superposed soot figures and give a single 
clear sinusoidal line. Perhaps this could be effected by using a 
generally clear flame in which a single tiny soot-producing source 
is introduced, and from which a fine line of soot would continu- 
ously pass to the recording tambour. If the method can be thus 
perfected, it would give a cheap, efficient and manageable way 
of producing clear and legible records of connected speech. 
How important this would be will appear plainly if we reflect 
that even our best equipped and endowed laboratories find 
serious mechanical and financial difficulties in making extensive 
records of connected sentences and words. To be legible the 
sinusoids must not be foreshortened, and when this condition 
is fulfilled, the record of even a single syllable becomes several 
feet long, complete words and sentences correspondingly longer, 
while a discourse may be measured by the mile. The mechanical 
difficulty and expense of producing photographic records on this 
scale is prohibitive. Hence the importance of developing a 
cheap and efficient means of producing extensive permanent 
records of connected speech. It would seem to be worth while 
for different laboratories to cooperate in experiments directed 
to this end. 
II. APPLICATION TO SPECIFIC PROBLEMS 
As to the application of perfected types of apparatus to specific 
linguistic problems, the chief need is undoubtedly a more exten- 
sive and intensive study of the phenomena of colloquial speech 
