A somewhat better indication of the accuracy of TRANSIT 
tracking is obtained by a sort of closed-loop conSistency calcula- 
tion. To accomplish this consistency check, the data for one 
particular pass at one station are held in reserve and the orbit 
is determined by the remaining passes during a particular time 
period. Then, using this orbit and the reserved Single-pass data 
the location of the ground station is computed - just as would be 
done in a typical navigation fix. The result of this position 
determination is compared with the known ground station location 
and the assumption is made that the error in satellite position is 
no greater than this resulting error in station location (actually 
it should be appreciably less, on the average, since there would be 
some error in station location even with a perfectly known orbit). 
based on data from TRANSIT 2-A, and the resulting sigma is well under 
one-quarter mile. 
It might be thought that the calculations described in 
the preceding paragraph are a direct test of the ability to perform 
navigation fixes by TRANSIT. Unfortunately this is not quite the 
case. For, in order to use TRANSIT for navigation in the practical 
operating system, it is necessary to base the calculation on an 
orbit determined previous to the time of the navigation fix (based 
on prior data). In other words, the practical use of TRANSIT for 
navigation always involves orbit extrapolation. This extrapolation 
process introduces a further error which obviously depends on the 
state of knowledge of the forces acting on the satellite and in 
particular on the knowledge of the gravitational field. Calcula- 
tions similar to those of the previous paragraph but where the data 
used for the position determination were obtained a day later than 
those used for the orbit determination (thus requiring a full day 
orbit extrapolation) indicate that the resulting error in position 
determination is now about one-half mile. Extending the orbit 
4/2 .B.Kershner, "The TRANSIT System" Proceedings of the Institute 
of Radio Engineers, Sept. 1960 
112 
aye 
Large numbers of these ''closed-loop" calculations have been performed— 
