600 



BELL SYSTEM TECHNICAL JOURXAL 



The Fourier sitIos representing the 84-cycle %va\e contained 30 terms, 

 the component frecjuencies being odd multiples of 84 up to a limit of 

 4,956 cycles; for the series representing the 153-cycle wave 17 terms 

 were used covering the range from 153 to 5,049 cycles. The agreement 

 between calculated and observed output waves would have been more 

 exact, particularh' at the corners of the wave shapes, if calibrations 



/www^y\ 



Fig. y — Oscillogram of s:|iiare-to|)|)C(l acoiislic waves as rcconkil 1)\ tin- a|)[iaratiiB 



and calculations had lieeii carrii'd to frequencies considnalilN ahow 

 5,()(K). As it was, the jjerformance was considered good; it indicated 

 that the uncorrected records of speech waves as taken were suiHiciently 

 accurate for most purposes, while if harmonic analysis of the records 

 was planned accurate results couitl be obtained over the range from 

 80 to 5,000 cycles, if the correction factors delerniiiied by ihe calibra- 

 tion were api)lii'd. 



In this description of (he reci)rding ai)paratus the emphasis has been 

 placed on the dynamical characteristics of the apparatus and its 

 calibration, but some of its other working features may briefly be 

 mentioned. 'Ilu' apparatus was sufficiently powerful to record sounds 



