zontally across Government, in special science programs, is being 

 studied for other essential areas. We have gone further in ocea- 

 nography than in some other areas of science. 



Essentially, the budgets will be provided through the Depart- 

 ment, but they are coordinated on a Government-wide basis in 

 oceanography. 



MR. PAUL D. FRELICH (General Instrument Corporation): What 

 is reliability? What do you mean when we are talking about relia- 

 bility? Can you discuss this in a general way and a quantitative 

 way? 



MR. G. JAFFE (HO): "Reliability, " for those of us who have 

 struggled with instruments for a number of years aboard ships and 

 on test programs and in laboratories, is simply the ability of the 

 instrument to perform over a reasonable period of time. The 

 durability and reliability problem in the oceanographic field is 

 slightly different than in the missile field in that the time scale is 

 significantly different. We would like our instruments to operate 

 over long periods of tinne, at sea, under relatively adverse 

 conditions, and this is contrary, of course, to the very short time 

 that nnost nnissiles are in flight. 



Generally speaking, I would define reliability in this case as 

 the ability of the instrument to operate under its environment for 

 a reasonable period of time, and with a certain amount of accuracy. 



THE CHAIRMAN: Adding to that just a little bit, Mr. Jaffe, would 

 you not agree, in a sense your reliability means reproducibility 

 within some kind of measurable linnits? These linnits might be 

 fairly broad, providing they can be measured; that is, a certain 

 accuracy with a standing measurement of error, which we can 

 compute. 



MR. G. JAFFE (HO): True. We may all be guilty of not seeing the 

 forest for the trees when we talk about the terms, reliability, 

 durability, and reproducibility. These are all part of the same 

 problem, the ability to have a device which will perform as the 

 specifications say they shall perform. This is really a package 

 problem. I do not think you can speak about reliability without 

 thinking of accuracy and without thinking of reproducibility and 

 long-term stability. 



113 



