20 



of exposure of the liquid to pressures below vapor pressure. However, an 

 examination made by the writer and Mr. Crump of the original data of Reference 

 37 reveals that the liquid was exposed to tensions for times that increased 

 with increasing absolute velocity. Furthermore, the times of exposure were of 

 the order of TO -3 seconds — apparently sufficient, in any case, for nuclei to 

 reach critical size before being carried into the high-pressure regions. Thus, 

 the apparently anomalous result was obtained that the higher the tension the 

 longer the water could withstand the tension before rupture. It would appear 

 that the water was well de-nucleated even with relatively high air content 

 although no special methods were used in treatment — the only treatment being 

 settling of the water for several hours before each experiment. However, 

 these trends can often be obtained in this nozzle even after the water has 

 been disturbed although under these conditions they can be maintained for only 

 short periods. It would be of interest to conduct similar experiments in 

 which the time of exposure is held constant but the velocity and, thus, the 

 pressure reduction varied. 



In the case of the experiments of Numachi and Kurokawa, a nozzle 

 with an abrupt expansion was used so that a vortex was induced in the separa- 

 tion zone just downstream of the nozzle. Although the ambient pressure rises 

 at this location, the pressures at the core of the vortex might fall well 

 below the pressure in the minimum section of the nozzle. Thus, with increas- 

 ing stream velocity, the strength of the vortex may be expected to increase 

 with resultant decreasing core pressure. 8 ' 40 Since the fluid in the separa- 

 tion zone is not swept away as rapidly as that in the main stream, the results 

 obtained by Numachi are quite tenable as to trend despite other objections as 

 to the size of nozzle used and to his method of analysis. If the actual pres- 

 sure at the vortex core could have been measured and used as the inception 

 pressure rather than the pressure in the constriction, the apparent discrep- 

 ancy might have been resolved. Mr. Crump has agreed with this explanation 

 (see Reference 37) > and, at the writer's suggestion, has conducted experiments 

 with a model of the Numachi nozzle (to be reported) to test this hypothesis. 

 He has found that, although the numerical results differ somewhat, the trend 

 is as predicted above. The possibility of cavitation occurrence in separation 

 zones at pressures above those predicted by analysis of the boundary is dis- 

 cussed more fully on page kf in connection with designing for cavitation pre- 

 vention. 



