The photolysis of chlorine peroxide (ClOOCl) is understood to be a key step in the destruction of polar stratospheric ozone. This study generated and purified ClOOCl in a novel fashion, which resulted in spectra with low impurity levels and high peak absorbances. The ClOOCl was generated by laser photolysis of Cl-2 in the presence of ozone, or by photolysis of ozone in the presence of CF2Cl2. The product ClOOCl was collected, along with small amounts of impurities, in a trap at about -125 degrees C. Gas-phase ultraviolet spectra were recorded using a long path cell and spectrograph/diode array detector as the trap was slowly warmed. The spectrum of ClOOCl could be fit with two Gaussian-like expressions, corresponding to two different electronic transitions, having similar energies but different widths. The energies and band strengths of these two transitions compare favorably with previous ab initio calculations. The cross sections of ClOOCl at wavelengths longer than 300 nm are significantly lower than all previous measurements or estimates. These low cross sections in the photolytically active region of the solar spectrum result in a rate of photolysis of ClOOCl in the stratosphere that is much lower than currently recommended. For conditions representative of the polar vortex (solar zenith angle of 86(o), 20 km altitude, and O-3 and temperature profiles measured in March 2000) calculated photolysis rates are a factor of 6 lower than the current JPL/NASA recommendation. This large discrepancy calls into question the completeness of present atmospheric models of polar ozone depletion.