which is important in view of the considerable expense which would be 

 involved in tests made on full-size pieces. 



The graph in Fig. 1 5 gives a general idea of our tests. From left to 



right there is recorded the relation 7 of different models. The bigger it 



is, the greater the stress it will experience for a given pressure. From 

 bottom to top are recorded the pressures applied during each of the 

 tests. Each of these tests is shown by a cross. Underneath the lower 

 curve is to be found the zone within which all the models resisted the 

 pressures applied without sustaining permanent deformation. Cross 

 No. 2 represents a test lasting eight hours, cross No. 3 one lasting 



ISOO-t 



•«--d--** 



Fig. 15. Diagram of strength tests made on model portholes 



eighteen hours. The other tests made were of shorter duration. 

 Between the two curves is to be found the zone where the plexiglas 

 sustained permanent deformations more or less considerable, but with- 

 out ever having been broken, thus without ever having let a drop of 

 liquid pass through. Experiment No. 4 alone, recorded above the 

 second curve, resulted in the destruction of the model. There is no 

 question in this case of the plexiglas having been defective. It was the 

 conical hole in the steel plate which splayed on the side where there was 

 low pressure and in these conditions it is clear that the plexiglas could 

 no longer resist. The conical socket had been machined in a plate of 

 mild steel much softer than the special steel of our cabins. As this 

 breakage took place under a load of 1270 atmospheres (corresponding 

 to a depth in the sea of more than 7-4 miles), this test was very reassur- 

 ing. The interior diameter of this model was twice as great as its thick- 



[ 150] 



