in more detail, which of these features (as well as others not mentioned) is 

 actually present in particular hydrodynamic experiments requires some means 

 of boundary-layer flow visualization. Surface films of various types have 

 traditionally been used in aerodynamic applications for this purpose and 

 in hydrodynamic work oil films have been found useful for detection of 

 laminar separation and transition (see, e.g., the review Acosta and Parkin 

 1975) . But a much more sensitive method for water is schlieren photogra- 

 phy. It is, of course, necessary in this method to provide a density 

 contrast; this can be achieved by injection of another liquid or, more 

 simply, by heating or cooling the surface layers. One sees then, in the 

 usual schlieren setup the thermal boundary layer which is for water at 

 room temperature about one-half the mechanical boundary layer' thickness. 



We were concerned at first about the effects of this heating or cool- 

 ing on the boundary layers themselves. Fortunately, this problem had 

 already been exhaustively treated by Wazzan et al. (1968), (see also Wazzan 

 and Gazley 1977) to show that under the expected laboratory conditions the 

 effect was negligible near a laminar separation but that some delay in a 

 turbulent transition might be expected. (We shall return to this point 

 later.) A schematic arrangement is shown in Figure 9; this is basically 

 the same as that of Arakeri et al. (1973) but with an improved spark 

 source. A photograph of the experimental arrangement in the Caltech Low 

 Turbulence Water Tunnel (LTWT) may be seen in Figure 10. 



Laminar Separation 



The schlieren technique was surprisingly easy to apply. For example, 

 Figure 11 shows a photograph of the flow past a 2-in. diameter hemisphere 

 body at a flow speed of about 22 ft/sec, well within the range of experi- 

 mental Reynolds numbers of the data of Figure 3. The clear laminar 

 separation on this body is easily seen, together with the turbulent re- 

 attachment of the free shear layer further downstream. The results really 

 exceeded our expectations for it could be quickly confirmed that a laminar 

 separation was present on the hemispherical body up to the top speed of 

 about 60 ft/ sec in the HSWT at Caltech (a body Reynolds number of about 



18 



