by Klebanoff^ for the velocity fluctuations in the boundary layer. This type of behavior 

 diminished with increasing frequency and was not noticeable for values of f 5*/Uq exceeding 

 4 X 10~'^. Further support for the possibility that intermittency plays a role at the lower fre- 

 quencies can be found in the fact that at these frequencies, the measurements of the convec- 

 tion velocity yielded a value of about 0.8 U^. That part of the boundary layer that moves with 

 a mean velocity of 0.8 U^ is strongly intermittent. 



The validity of the above conjectures could be established by repeating this study 

 using fully developed pipe flow. 



In view of the preceding, it would have been desirable to have procured data at lower 

 frequencies. Unfortunately, this was not possible due to the tunnel noise which limited the 

 lowest usable frequencies to between 100 and 200 cps. The signal-to-noise ratio above this 

 range was always in excess of 20 db. This figure was obtained by placing the pressure pick- 

 up in the middle of the working section of the tunnel where the flow was essentially laminar. 

 Actually, some noise was undoubtedly induced by the flow around the microphone, so that 

 a better signal-to-noise ratio might actually have existed at the wall. 



Vibration pickup by the transducer was not a problem. By simply plugging up the hole 

 in the transducer, the vibration excitation could be measured, which in this case was too low 

 to even consider. 



LONGITUDINAL CROSS SPECTRAL DENSITY 



In this work it was decided to study the longitudinal cross spectral density rather than 

 the longitudinal correlation. Since this is a novel approach, a few words of explanation are 

 needed in order to explain the measurement procedure. 



The cross spectral density of the pressure fluctuations is defined as 



where the asterisk denotes the complex conjugate, and where 



T 



-T 



T 

 -T 

 with the subscripts 1 and 2 denoting the two observation points. 



