Ozone measurements made with different kinds of instruments in clean air during the past few years indicate that total ozone amounts measured with Dobson spectrophotometers on AD wavelengths may be too high by about 5%. An examination of the consistency of ozone data determined from A, B, C, and D wave-length ozone absorption coefficients adopted for use at the beginning of International Geophysical Year (1957--1959) suggests that the problem stems from either a significant error in the determination of the effective ozone absorption coefficient at the short (305.5 nm) wavelength of the A wavelength pair, the value of the effective wavelength itself, or from the presence in the atmosphere of an anomalous absorber other than ozone. Since Dobson spectrophotometers serve as reference standard instruments for measurements of total ozone, it is imperative that the source of the systematic error be identified, and its magnitude assessed.