Note on the Critical Bandwidth

Exceptionally small values of the critical bandwidth are found from masking thresholds of a tone in narrow bands of noise, when the test tone's frequency remains at the center of the band of noise components. Numerous other experiments—by no means limited to masking—have led to the conception of a markedly wider critical band. In this communication the evidence on this ambivalence. especially as regards masking at the central frequency of a noise band, is critically examined. It appears that the critical band cannot be expected to be accurately determined from central-masking experiments. The most important reason is that the effective power of masking noise is not the sole determinant of masking. The inherent fluctuations of random noise, becoming less rapid and more noticeable at small bandwidths, exert a considerable influence on the detection of tones. The extent of this effect can be evaluated from the intensity difference limen of narrow-band noise. It is to be concluded from the discussion that the larger values of the critical bandwidth are to be preferred, and that the smaller values should be rejected.