models, the size and dimension of the specimen rack varies according to the lamp size and the wattage at which it is operated to provide an irradiance at the face of the specimen of 1.10士0.03 W/(m=nm) measured at 420 nm or equivalent when exposed in standard holders. A3.1 Aging of the xenon burners or filters can result in changes in lamp spectrum. Changes in lamp spectrum may also be caused by accumulation of dirt or other residue in or on the burner envelope. A3.2 Filter-In order for xenon-arcs to simulate terrestrial daylight, filters must be used to remove short wavelength UV radiation. In addition, filters to remove infrared radiation may be used to preventunrealistic heating of test specimens that can cause thermal degradation not experienced during outdoor exposures. Filters toreduce irradiance at wavelengths shorter than 310 nm must be used to simulate daylight filtered through window glass. The instrument manufacturers' recommendations should be used to provide the appropriate spectrum (see A3.4 belo司. Replace filters when chipped, cracked, or when discoloration or milkiness devel-ops. Discard xenon lamp tubes and filters at the manufacturer's recommended time intervals or sooner, or when 20 AATCCFading Units can no longer be attained in 20士2 continuous light-on operating (clock) hours. A3.3 Spectral Irradiance of Filtered Xenon-Arc-Fig. A1 shows the desired relative spectral power distribution for filtered xenon-arcs comply with these limits. The acceptable limits for variationof the relative spectral power distribution shown in Fig. A1 are on file at the AATCC Technical Center. A3.4 Follow the device manufacturer's instructions for recommended maintenance