94 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1957 



were chosen for maximum life, compatibility with each other, and for 

 best adaptability to the design intent. Specifications particularly adapted 

 to this use were set up for all of the some 50 different metals and non- 

 metals employed in the enclosure design. In general, the methods es- 

 tablished for proving the integrity of the materials are more elaborate 

 than usual commercial practice. In most instances, such as that of cop- 

 per container tubes, the extraordinary inspection for defects and weak- 

 nesses with its resulting rejection rate, resulted in high cost for the usable 

 material. 



TESTING 



A substantial part of the development work on the repeater enclosure 

 was concerned with devising tests that give real assurance of soundness 

 and stability. It is beyond the scope of this paper to discuss how each 

 part is tested before and after it is assembled but certain outstanding 

 tests deserve mention. 



Steel Ring Tests 



Each of the inner steel rings, before installation, is required to pass a 

 magnetic particle test to find evidence of hidden metallurgical faults. 

 Each ring is later a participant in a group test under hydraulic pressure 

 simulating the crushing effect of ocean bottom service but exceeding 

 the working pressures. The magnetic particle test is repeated. 



Helium Leak Tests 



Both glass and rubber seal assemblies, before being installed in re- 

 peaters, are required to undergo individual tests under high-pressure 

 helium gas. Helium is used not only because its small molecules can pass 

 through smaller leaks than can water molecules but because of the ex- 

 cellent mass spectrometer type of leak detectors commercially available 

 for this technique. While helium is applied at high pressure to the 

 outer wall of the seal, the inner wall is maintained under vacuum in a 

 chamber joined with the leak detector. The passage of helium through a 

 faulty seal at the rate of 10~^ milliliters per second can be detected. 

 Stated differently, this is 1 milliliter of helimn in 30 years. The relation 

 of water-leak rate to helium-leak rate is dependent on the physical na- 

 ture of the leak, but if they were assumed to be equal rates, the amount 

 of water which might enter a tested repeater in 20 years would be 0.66 

 grams. A desiccant within the repeater cavity is designed to keep the 



