affecting the collapse pressure and consequently nondimensional curve plots, 
to some degree. Also, due to the impracticability of constructing separate 
cylinders for each of the models of Reference 7, a Single cylinder is used 
for all. AS a result, none of the models is deSigned with the ideal bound- 
ary conditions of membrane deflection and no rotation, thereby further 
altering the stress pattern in the fabricated spherical shells. 
It should be emphasized that the plotted portion in Figure 6 should 
not be used for design purposes at this time because data are insufficient 
to verify its accuracy. Under no circumstances should the curve be used 
for design purposes unless the proposed spherical shell is to be designed 
and fabricated with the same accuracy as those models presented in this 
report; this precludes all but near-perfect machined models with ideal 
boundary conditions (where applicable). The curve may be used to design a 
spherical shell of any material provided it has the same shape stress-strain 
curve as those for inelastic failure design illustrated in Figure 1. 
CONCLUSIONS 
1. The experimental collapse pressures agree within + 7 percent of 
the theoretical buckling pressures obtained from the inelastic empirical 
buckling formula for spherical shells developed at the Model Basin. 
2. The experimental collapse pressures of Models 6 and 4A may be 
misleadingly low because of slight deviations from design due to errors in 
fabrication which occurred during the process of machining. 
3. The design of ideal boundary (membrane deflection and no 
rotation) at the juncture of the hemisphere with the cylinder appears 
effective in preventing the cylinder from altering the stress pattern (and 
consequently the collapse pressure) in the hemisphere from the pattern that 
would be created in a sphere of the same material, radius, and shell 
thickness. 
4, The empirical portion of the nondimensional plot of results for 
machined high-strength steel hemispherical shells (Figure 6) should not be 
used for design purposes pending verification of its accuracy. 
5. Upon verification of the accuracy of the empirical portion of 
the curve of Figure 6, it may be used for deSign purposes only when the 
proposed spherical shell is fabricated with the same accuracy as the 
