(Model 6). It is possible that a premature failure occurred as a result 
of the machining error in the third frame of the model. The irregularity 
may have created a weakness in the hemisphere due to an increase in stress 
set up by a deviation in the desired boundary condition. The strain- 
sensitivity coefficients shown in Table 4 at the 2-deg longitudinal locations 
for Model 6 indicate greater membrane deflection than at any other location 
on the hemisphere. It is possible that analogous action resulted from 
the pressure loading of Model 4A due to the overcutting of the inside of 
the cylinder wall; a discrepancy in the desired boundary condition in this 
model may have effected a Similar decrease in its collapse pressure. 
The elastically designed Model 7A developed a collapse pressure of 
290 psi. This is approximately 74 percent of the classical theoretical 
value of Zoelly and Timoshenko and 107 percent of the empirical elastic 
value of Krenzke. 
The theoretical values of pressures presented in this report were 
calculated using the nominal values of the radius and the thinnest measured 
values of the spherical thickness taken for each model. AS indicated by 
the ratios of experimental collapse pressure to the TMB empirical buckling 
collapse pressure P,, in Table 5, results for all the models showed reasonable 
agreement over a ae range of ratios of hemisphere thickness to radius. 
This consistency may be attributed to the high standard of precision main- 
tained in the machining process. Since the models were not stress relieved, 
residual machining stresses were present, but they are considered negligible 
in view of the very small sphericity departures indicated in Table 2. The 
good correlation of experimental to theoretical collapse value obtained for 
Model 7A (to which the most rigorous conditions were adhered in the 
machining process) tends to support the suggestion of Reference 4 that if 
machining standards could be maintained sufficiently high to satisfy the 
rigid assumptions of classical buckling theory, then the classical formula 
could be used for thin spherical shell design. However, even if such 
high quality control could be obtained, cost and time requirements would 
undoubtedly be prohibitive for prototype construction. 
The experimental results for the models pertaining to this report 
are plotted nondimensionally in Figure 6. The ordinate is the ratio of 
experimental collapse pressure to the equilibrium yield pressure Ey? as 
