The measurement of fan sound power levels has been done in accordance with many national and international standards, like BS 848-2:1985, DIN 45635-38 and -9, AMCA 300:90 and ISO 5136. All these standards define the acoustical properties of the measurement environment, to identify the physical relationship between the sound pressure level, which can be measured, and the sound power level to be determined. Unlike other noise sources, anyway, the inlet and outlet side of fans are acoustically connected, and the acoustical impedance of the environment on the non-measurement side influences the sound power level which can be observed on the measurement side. As a result, the repeatability of the sound power measurements on fans can be compromised using simple, acoustically untreated test ducts, on the non-measurement side. A series of experimental measurements were carried out in a reverberating room, in accordance with the AMCA 300:90 standard, on the suction side of a centrifugal fan with scroll, connected to an airflow-measurement chamber with a simple "short outlet duct'' on the discharge side of the fan. Third-octave band sound power spectra where measured, at the same flow coefficients but at different fan speed values, with the test setup otherwise unmodified. The measured sound power levels were then reduced to "non dimensional sound power spectra'', as functions of "non dimensional frequencies'', according to the ISO 13348 computation method. Satisfactory results could be achieved across most of the sound spectrum, but not around the blade-passing frequency and its second-order harmonic, i.e. where the fan is generating its major pure-tone components. The blade-passing frequency amplitude is rising and dropping with the change of the fan speed and fundamental rotation frequency, and the variation of the amplitude can be related with the ratio between the BPF wavelength and the physical length of the simple, straight outlet duct, demonstrating to be a simple case of axial resonance along the outlet duct. These test results confirm the importance of the new ISO 13347 family of standards, introducing the simplified anechoic termination of non-measurement ducts, to standardize the acoustic properties of the test setup, also on the non-measurement side. Notwithstanding its merits, this method is not widely adopted by the industry, which is still struggling to find a practical solution to make measurements in all the many different duct sizes, required to cover an extensive fan range, with a still-acceptable investment and a manageable stockpile of different acoustically-treated test-ducts.