Fabula 



The time for raggedness of Polyox solutions to disappear was not noticeably 

 affected by the range of solution preparation technique. The time depended, of 

 course, upon the value of x/M of interest. Roughly speaking, the times for rag- 

 gedness to disappear at x/M = 10.7 were about the same as the times for a 50% 

 loss of specific viscosity, according to Fig. 11, independent of the solution 

 preparation technique. 



With JlOO solutions prepared in the same way as the P301B solutions, no 

 significant decrease in -q^ was found over the longest period tested, viz., one 

 week, and as has been mentioned, there was no decrease of signal raggedness 

 for the same period. 



An effort was made to eliminate the raggedness by purifying the solvent. 

 "Softened" and demineralized tap water were tried with P301B, with no detect- 

 able effects. 



No dependence of raggedness upon sensor overheat or angle of attack could 

 be detected. Raggedness was only found when the moving sensor reached liquid 

 disturbed by the grid or by other obstacles moving at similar speeds, such as a 

 circular cylinder or the sensor strut itself. 



Spectral Effects of Signal Raggedness and Solution Noise 



The spectral effects of raggedness and solution noise are illustrated in 

 Figs. 12 to 16. Relative spectral levels are given for x/M = 20.2 in water and in 

 Solution P301-3 at various solution ages in Solution P301-3 (WSR-301; direct 

 dispersal; 18 ppm). The conversion from relative levels versus frequency to 

 absolute spectral levels versus wavenumber is not needed. There is a common 

 reference for the relative levels in all five figures, and the change in sensor 

 sensitivity corresponds to only a 0.7-db shift. This is less than the standard 

 error of the single-run spectral measurements of about ±1 db. In each figure 

 the data of three runs are shown: (a) a grid-turbulence test with recording of 

 the unfiltered signal (e.g.. Run C6 in Fig, 12), (b) a grid-turbulence test with 

 recording of the filtered signal in order to determine the high-frequency end of 

 the spectrum (e.g.. Run C9), and (c) a noise test with recording for the same 

 filter setting as in (b) (e.g.. Run ClON). Noise tests without filter are not shown 

 because the resultant noise corrections are negligible. The plotted spectral- 

 level data points are always the values without noise correction, and the associ- 

 ated fairing curves (the solid lines) are with noise corrections. Thus the noise 

 correction produces the departure of the fairing curves from the data points. In 

 later figures only the points for the signal-plus-noise spectral levels and the 

 noise-corrected fairings will be shown. The latter are always ended when the 

 noise correction becomes about 4 db, which occurs when the noise level is 2 db 

 below the signal-plus-noise level. 



In Figs. 13 to 16 the water noise level from Fig. 12 is shown by the curve 

 labeled W.N.L. The increase in noise in the solution is about 5 to 10 db at the 

 frequencies shown and is fairly independent of solution age. Figure 12 shows 

 that in water the noise correction would have reached 4 db at about 250 cps. In 

 the 4-hour-old solution, the noise correction is negligible up to above 250 cps 



56 



