with Westinghouse Electric Corporation, a deep- towed speed log which 

 should yield a better estimate of ground speed than speed logs operating 

 in the near-surface waters. 



CONCLUSIONS 



The need for a shipboard computer with the capacity of a 70 90- type 

 machine is not far off. It is presently possible to define the basic require- 

 ments of future shipboard computer systems. It is not possible, however, 

 to anticipate all the things one would like to do a year hence, thus versa- 

 tility and flexibility of design are extremely important if the ability to 

 follow up on experience is desired. Multiprogramming capability is 

 essential, including the ability to time- share real-time and off-line pro- 

 grams. The program should be written with a modular construction and 

 preferably in a Fortran- like language with sampling and control instructions. 

 This would allow the user greater ability to modify and expand the program 

 as desired. Core memory should be made as large as practical; the 

 tendency is for best estimates to be too small. Equipment and programs 

 should exist that will prevent failures in peripheral units or external input 

 units from halting or destroying the operation of the other activities of the 

 computer system. It also appears that the use of several similar output 

 devices for the visual recording of the various types of scientific data 

 processed is preferable to one large unit recording everything in a 

 necessarily less flexible format. And last, but not least, reliability 

 should be maximized to the extent that urgency demands, and design and 

 money permit. 



REFERENCES 



Bernstein, R. and C. O. Bowin, 1963, Real-Time Digital Computer 

 Acquisition and Computation of Gravity Data at Sea: IEEE 

 Transactions on Geoscience Electronics, Vol. GE-1, No. 1, 1-10. 



Bowin, Carlo. , 1963, WHOI Reference No. 63-40, Supplementary 



Report on the Automatic Gravity System Used Aboard R/V CHAIN 

 July 1962 - September 1963. 88 pp. 



374 



