AXAF telescope witness mirror calibrations have been carried out on ALS Beamline 6.3.2 in July 1997 and Jan-Feb 1998 to ascertain whether sufficient beam purity and signal to noise were available over the energy range 50-1000 eV. An overall accuracy of 1 percent is necessary on reflectivities to meet AXAF program requirements. Some beamline variations were incorporated - based on two previous test runs. Various filter/order-sorter combinations were used to control beam purity. A test involving strength of the N-edge dips in reflectivity versus sample angle to the beam showed the NVII, NVI, NIII, NII, and NI edges in roughly expected proportions. This verified adequate purity in all filter/order-sorter channels except the 260-454 eV channel that includes the NV and NIV edges just above the carbon K- edge. We discuss a future solution of this puzzle by other test for evaluation of the carbon contamination layer. Oxygen contamination is also ubiquitous on Ir mirrors and foils. These problems can also be handled through optimization of the data acquisition and use of control samples. In reflection Ir NIII appears out of the noise as two doublet-like bumps. Ir NII has so far been observed near the noise level < 2 percent. Averaged over many scans it too appears to be doublet-like. Scans through the NI edge near the noise level are compatible with a doublet-like or broad feature from 685-707 eV. transmission data give the NI and NIII edges better. Ir NIV and NV edges are not yet measurable for systematic reasons. NVI and NVII appear to be normal edges. Angle scans to derive the Ir and Cr layer thicknesses are feasible at approximately 900-920 eV. Reflectivities can be reliably measured at small angles to 1300 eV. This will be useful for overlap with measurements at NSLS. Optical constants derived from reflectivities in most segments of this energy range will meet requirements for AXAF. Transmission measurements are being pursued in addition, for confirmation and to augment some segments.