SATELLITE-TRACKING PROGRAM — ^HAYES 335 



The differences between the theodolite readings for the satellite and 

 any one of the reference stars give a value for the satellite position. 

 The mean of the values obtained for all the reference stars is accepted 

 for the final position of the satellite. The smallest readable unit on 

 the Van Biesbroeck goniometer is 1 second of arc. 



With the Mann machine, the film is placed on the comparator near a 

 zero-degree orientation; i.e., with the oscilloscope edge toward the 

 measurer. The satellite image is brought to a point near the center 

 of the target screen. The stage of the Mann machine is rotated mitil 

 the trail of the satellite is as nearly parallel with the horizontal cross- 

 hair as is possible, and the stage is locked. The satellite image (or 

 central break) is brought to the cross-hair intersection. The two 

 plane coordinates, x and y, of the reference stars and the satellite are 

 then measured. The stage of the Mann machine is then rotated 180° 

 and the measurements are repeated. This is done to eliminate the 

 magnitude error — a systematic but not a constant error of the observer. 



For the computation of the 6 plate constants, the measurer used 6 

 st ars, employing the least-squares method to compute the 6 constants 

 from 12 equations. When there were large residuals, one or two refer- 

 ence stars were sometimes omitted. If large residuals still remained, 

 he repeated the measurements, never using fewer than four comparison 

 stars. 



A measuring accuracy of 1 micron (which corresponds to 0.4 second 

 of arc on the Baker-Nunn films) or better can be achieved with the 

 Mann comparator. 



Before the introduction of the completely automatic equipment the 

 X and y coordinates were read by eye and written down by hand. 

 These data as well as the catalog data on the reference stars were 

 punched on tape by a Flexowriter and the position of the satellite was 

 computed by a Burroughs E-101 computer using the Flexowritei- tape 

 as input. The computation with this machine took about 30 to 40 

 minutes. 



As a preliminary step, the two machines were used to locate "un- 

 known" stars from the Yale catalog by measuring their positions 

 relative to nearby reference stars also selected from the Yale catalog. 

 By this means, the nature and extent of several sources of error could be 

 determined. First, of course, there were the errors inherent in the 

 machines themselves. For example, the Mann engine was operated by 

 means of a periodic screw and a secular screw ; each of these mecha- 

 nisms had to be evaluated. 



Second, there was the human element. Each person using the 

 machine would do so in his own particular way ; he would handle the 

 machine in an individual fashion and would be more or less accurate 

 compared to other measurers. The personal error could in general 

 be eliminated, however, by making direct and reverse measurements of 



