Application of the PC matrix yields a coordinate system in the direction of the physical axes but in pixel units. Conversion to physical units is performed by applying the scaling specfied by the CDELTn.
In some past applications of this process, steps (2) and (3) have been combined into a single step, using a matrix that is the product of the CDELTn, which may be thought of as a diagonal matrix, and the PC matrix. This matrix, called the coordinate description (CD) matrix, described the tranformation from the array coordinates in pixel units to the physical coordinate in physical units. It was represented using keywords of the form CDi_j, by the Space Telescope Science Data Analysis System/Image Reduction and Analysis Facility (STSDAS/IRAF), or CDiiijjj, proposed by HW. This method was described in the previous version (3.1) of this Guide. FITS readers should be prepared to encounter it in some existing data sets.