Most fluid machines can be considered as periodic noise sources when operated under constant conditions, which allows for a frequency domain representation of the source and the associated acoustic field In the duct. In such a representation, the source is characterized by frequency-dependent values of both strength and impedance. Although knowledge of these values can be gained by either experimentation or by modeling, one-port acoustic characteristics of an in-duct source with high flow velocity, high temperature, and high sound level can be measured only by the multiload method using an overdetermined set of open pipes with different lengths as applied loads. However, the problem is that negative source resistances have been often measured. This paper reviews the possible causes of the problem, with reference to experimental and theoretical results, in an attempt to clarify the issue. A new interpretation is given for the violation of basic assumptions and the defect in the algorithm of multiload method. The major cause and mechanism of the problem is due to the violation of time invariance assumption of the source and the load impedance can seriously affect the final measured result of source impedance.