The collapse strengths of Cylinders 1, 3, \, and 5 were affected by 

 the closure plates. Their average collapse strength was about 15 percent 

 higher than would be expected for a semi-infinite cylinder. This is in 

 good agreement with the theory of Sturm for simply supported cylinders of 

 critical length under external hydrostatic pressure. 



There is no concrete explanation for the cracks which were observed 

 in Cylinders 1 and 6 although several possibilities exist. The most 

 probable explanation is that the cracks were caused by high local stresses 

 occuring at the ends of the cylinders as they punched through the lead 

 and came in direct contact with the steel end plate. This assumption is 

 supported by the fact that each time a cracking noise occurred^ it was 

 heard while pressure was being applied to the cylinders. It is also 

 possible that the glass cracked while unloading although no noise from 

 the pressure chamber was heard while pressure was being released. 



If the cracks were caused by high local bearing stresses, it appears 

 possible that this drawback may be overcome. By proper design of the end 

 closures, the bearing stresses can be minimized and the cracks may possibly 

 be eliminated. Although the cracks had no apparent influence on collapse 

 strength, they must be eliminated before glass can be seriously con- 

 sidered as a hull material for deep-depth vehicles. This is true since 

 water would very likely flow through the cracks when on or near the 

 surface and would prohibit recovery of the vehicle. 



No definite conclusions can be made concerning any systematic effect 

 of initial imperfections on collapse strength. Obviously, more elaborate 

 measurements than those taken during these rather exploratory tests are 

 required before any definite conclusions can be drawn. However, several 

 general observations can be made concerning the effect of initial im- 

 perfections from the tests described herein. As shown in Figure la, the 

 observed collapse strength agrees very well with classical theory when 

 based on minimum measured thickness. This is not surprising since the 

 eccentricity of the bore relative to the exterior surface caused the minimum 

 thickness to occur at a particular angular orientation along the entire 

 length of the cylinder. Although the limited measurements indicate that 

 varying degrees of out-of-roundness were present in the cylinder, there 



