stiffeners the cracking would not occur: but the thermal stresses 

 would not be done away with. 



After having studied a whole series of constructions we adopted the 

 following solution, suggested by Mr. Loser : it satisfies all the require- 

 ments set out here. 



The partitions are made of metal having semicircular corrugations 

 with a radius of curvature of 3-84 in. and with their axes vertical. 

 Thus they provide sufficient resistance to variations in pressure (of 

 which we have spoken). Calculations and tests with models com- 

 pletely reassured us on this point. If we consider a hull with a given 

 circumference but which without too great an effort can become 

 slightly oval, and if we allow the partitions to be rigid in the vertical 

 direction and extensible or contractile in the horizontal direction, it 

 can be seen that we satisfy all requirements. If the temperature of the 

 partitions rises, the vertical diameter of the float will increase, while the 

 horizontal diameter will decrease slightly in such a way as to allow the 

 hull freely to assume a form which suits its circumference and which is 

 determined by its own temperature. 



The engineer will understand that our construction is in its main 

 lines isostatic, while with flat partitions it would be very heterostatic. 



At first sight it might be feared that our float might not be able to 

 resist bending moments which would act on the horizontal plane. That 

 is not the case, however. The circumference of the hull being given 

 and its vertical diameter being fixed by the corrugated partitions, the 

 horizontal diameter is determined by these two dimensions. Our 

 partitions, although corrugated, prevent all deformation of the hull, 

 as much in the lateral direction as in the vertical direction. 



It can thus be seen that our partitions in corrugated sheeting do not 

 offer any disadvantage, and perfectly answer all that we demand. 



Appendix 5 



Thickness of Metal Sheets of the Float 



I fixed the thickness of the upper plates of the hull at j in. The 

 thicknesses suggested by the majority of specialists varied from 2~5" in. 

 to ^5- in., but some go as far as f in. I think that 2-5" in. would have been 

 enough : but it must not be forgotten that if the tensile strength of two 

 metal plates of j in. and of ^5- in. are in the relation of 5 to 4, or i -25, 



[ 163] 



