SATELLITE-TRACKING PROGRAM — HAYES 337 



Meanwhile, the staff was working on the problem of measuring films 

 with the Mann machine. The positions of nine images were reduced 

 with a probable error in right ascension of 1.05 seconds of arc, and 

 in declination of 0.54 second of arc. At the same time, a program 

 was written for the reduction of measurements made by this machine 

 so that computations that required 1 or 2 days by hand could now be 

 performed on a Burroughs E-101 electronic computer in some 15 

 minutes. This was the first step toward automating as much of the 

 procedure as possible. 



As a further step to facilitate the work, a special project was under- 

 taken to assign Yale catalog niunbers to the BD and CD star charts. 



Precision reduction of the films continued, so that in the first quarter 

 of 1959, a total of 155 satellite images were measured and in the 

 second quarter 109. Meanwhile, however, the Baker-Nunn stations 

 were taking films at a considerably faster rate; during the same 

 6-month period, more than 4,000 films were received in Cambridge. 

 Clearly, more rapid and efficient means of measuring the films remamed 

 to be found and put into practice. 



COMPUTATIONS 



Before Explorer I was launched early in 1958, the Observatory had 

 developed two computer programs that were to be the basis for the 

 determination of orbits and the preparation of predictions for the next 

 year and a half (see Part 2 of this history ^). 



From a set of observations of a satellite the Herrick-Briggs-Slowey 

 initial orbit determination program was used to derive the orbit with- 

 out any previous knowledge of it. With a program of this type, the 

 accuracy camiot be high since usually only three observations are 

 used for the calculation of an orbit. However, an initial concept of 

 the elements of the orbit can be obtained. 



Two major improvements were soon made in the program. First, 

 an empirical correction for air drag used an expression for the nodal 

 period as input and computed the corrections to the observations 

 necessary to give the osculating orbit at the time of the first observa- 

 tion. The second provided an alternate method of interpolation when 

 the usual method failed. In this mode of operation, the program must 

 find any and all elliptical solutions in a given range that fitted the 

 observations. 



By mid-1958 the program was fully debugged, tested, and com- 

 pletely operational in all of its essential parts. Proof of the usefulness 

 and accuracy of the program was demonstrated by its application to 

 the tracking of 1958 Delta. The program was used not only to obtain 

 an initial orbit but also to follow the changes in the orbital elements. 



2 See footnote 1 on page 315. 



