6 PREFACE. 



but the matter must be left to physicists, as it leads rather aside from the 

 more strictly phonetic problems. 



The second chapter gives an account of my method of obtaining speech 

 curves. In addition to its accuracy, a special feature of the method is that 

 the process is automatic and can go on day and night continuously. This 

 is important for the more specially phonetic and psychological problems, 

 for which whole speeches and conversations must be studied. When we 

 consider that a moderate enlargement of the speech curve for a four- 

 minute conversation requires a tracing a quarter of a mile long, and that 

 to obtain each individual wave accurately the apparatus must run slowly, 

 it is easy to understand why the apparatus must be automatic. The 

 account of the apparatus is intended to be detailed enough to ena])le a 

 skillful mechanic to construct duplicates; the methods of testing its 

 accuracy are described. 



The chapter on qualitative analysis indicates how phonetic facts 

 can be read directly from the speech curves without measurement. This 

 will probably he a new thought to philologists. Anyone with a knowledge 

 of the principles of phonetics can in a short time acquire the art of reading 

 curves. What he will find in them will surprise him. The speech curves 

 present to the e,ye of the phonetician the flow of language as it is actually 

 spoken. It is evident that the investigator can proceed on a much surer 

 foundation than at present, where he uses only two methods; for modern 

 languages hie can do nothing but gather vague impressions b,y the ear, while 

 for the ancient languages he must rely on the typographical representations, 

 which give no details and which may be both erroneous in their origin 

 and misunderstood by himself. The great need in phonetics at the present 

 day is a supply of accurate speech curves of various languages published 

 in the form of atlases for purposes of study by specialists. 



In Chapter IV, I have explained how such fundamental factors of 

 speech as melody, duration, and amplitude can be obtained from the curves 

 by simple methods of measurement. The problems are of interest to the 

 psychologist as well as the phonetician. 



In Chapters V, VI, and VII some difficult but necessary problems are 

 discussed. In the first place, the method of harmonic analysis is care- 

 fully considered, as it is the basis of all work on wave analysis. I have 

 had the advantage of advice from Professor Weber, of the Swiss Poly- 

 technicum (who was the first person, with Schneebeli, to apply the Fourier 

 analysis to a vowel curve), and of Professor Wolfer, of the Swiss National 

 Observatory at Zurich. Although Helmholtz had distinctly stated that 

 the application of the harmonic anal3'sis to a vowel wave is a purely 



