136 THE STUDY OF SPEECH CURVES. 



but also because it constantly furnishes new suggestions for the perplex- 

 ing phenomena found in the actual speech curves. 



It is interesting to consider what such synthetic work may ultimately 

 lead to. A successful S3'nthetic apparatus or system of computing would 

 be able to counterfeit the curve of any vowel; inscribed on a gramophone 

 disc such a curve would produce a vowel sound that had never been pro- 

 duced by the vocal organs. New vowels could be produced that have 

 not been — and perhaps never will be — present in any language. 



The development of the method of synthe- 

 sis will lead ultimately to systematized collec- 

 tions of curves which will include all the usual 

 forms found in speech curves. When this has 

 been accomplished, the collection can be used 

 as an index for analyses. When a curve is 

 found to coincide with a synthetic curve, the 

 formula of synthesis can at once be used as 

 the result without performing the actual anal- 

 ysis. Although such an index would be of 

 almost incredible utility — furnishing an anal- 

 ysis after a few minutes' search instead of the 

 10 or 12 hours' minimum of labor regularly 

 '''■ '"L^sZ^'"" "' required for each wave-it is unfortunately 



not to be hoped for in the near future. Ihe 

 curve of a frictional si u?oid (regardless of phase) contains three arguments 

 to be varied: period, amphtude, and factor of friction. A moderately 

 thorough treatment would require perhaps 100 variations of each argu- 

 ment. This would require tables for 300 fundamental curves, and addi- 

 tions for 50,000 curves. Such an undertaking is at present out of reach. 



