38 THE STUDY OF SPEECH CURVES. 



In the work of tracing, it is necessary to balance several factors. 

 The enlargement ehosen for the vowels will not give the curves of the con- 

 sonants; that necessary for the consonants makes the vowel waves so high 

 and steep that they lose all detail. In selecting the longitudinal magni- 

 fication (time equation) such a balance must be kept with the vertical 

 magnification that the waves come out properly for the purpose in hand. 

 For studying curves with the eye it is desirable to condense them hori- 

 zontally (as in the Cock Robin plate). For accurate measuring and 

 analyzing, they must be more extended. Such an extension takes up too 

 much room in printing, a single vowel being often a meter or two long. A 

 convenient mean is that adopted for the curves shown in plate vii. The 

 problem of cost in printing puts limits to the magnification, as will be 

 readily understood from the following facts. The Mitchell vowel record, 

 traced with the time equation 1mm. = 0.0002s. and the magnification 258, 

 gives a strip of tracing 717 meters long (nearly half a mile), which makes 

 104 plates of the size of plate vii, containing 1,938 lines, each 37cm. long. 



