The first step was a precise characterization of the error in the main mirror. Working backwards from images of point sources, astronomers determined that the conic constant of the mirror as built was −1.01390±0.0002, instead of the intended −1.00230.[60][61] The same number was also derived by analyzing the null corrector used by Perkin-Elmer to figure the mirror, as well as by analyzing interferograms obtained during ground testing of the mirror.[62]
Because of the way the HST's instruments were designed, two different sets of correctors were required. The design of the Wide Field and Planetary Camera 2, already planned to replace the existing WF/PC, included relay mirrors to direct light onto the eight separate CCD chips making up its two cameras. An inverse error built into their surfaces could completely cancel the aberration of the primary. However, the other instruments lacked any intermediate surfaces that could be figured in this way, and so required an external correction device.[63]..."
http://en.wikipedia.org/wiki/Hubble_Spac…