Although computation time was greater than it would have been on the 

 Univac or other large machine, the problem as done on the Logistics Computer 

 was conceptually simpler, because of the ability to store ultimately all data 

 needed at any computation stage, and more economical (even if the problem had 

 been chargedfbr) due to the smaller cost per hour of this machine. 



In conclusion, ihe authors would like to thank Louis Grey, Anatole Holt 

 and William Turanski for the help and advice they have given in the Univac pro- 

 gramming, Gordon J. Morgan of the Logistics Research Project, William W. 

 Ellis and Bernard Chasin who helped with the IBM card operations needed to 

 provide the tables in this report. 



The original spot height data furnished by the U. S. Navy Hydrographic 

 Office, the leveled data, the values of Q(p, q), L*(r,s), and U(r, s) for the ori- 

 ginal computations, and tie values of Q(p, q), L(r, s), and U(r, s) for the re- 

 duced data, are given in the following tables. (Note that in order to compute 

 the U(r, s) values, the L(r, s) values must be used, and not the L*(r, s) values.) 

 The tabulated values of L*(r, s) should be doubled on all borders except at the 

 corners where they should be quadrupled to obtain the L(r, s) values. These 

 values are also available in a deck of IBM punched cards at the Research Divi- 

 sion of the College of Engineering, New York University. The raw data and 

 the leveled data are given in 540 cards per run, for Data Sets 2 and 3 and 

 350 and 3&0 cards for Data Sets 2A and 3C, respectively; and the covariance 

 surface, the spectrum, and the smoothed spectrum are given on 841 cards 



per run. Thus 11,882 cards are available in all. All Logistics Computer prov 



grams used are in the possession of George Stephenson. 



94 



