Tests of the standard NCER multiplex system recorded and played back on both the Bell and Howell 3700B (about 0.1% tape speed variation) and on the Sony TC-126 cassette recorder (about 1% tape speed variation) showed that subtractive compensation employing a reference frequency multiplexed on the data track was remarkably effective in reducing tape-speed-variation-induced noise 1 and, hence, in increasing the dynamic range of the record/playback system. Further tests suggested that the 0 to 30 Hz bandwidth of the standard system (8 data channels) might be increased substantially, at the 'price' of reducing the number of data channels to 3 or 4, without serious loss of dynamic range if subtractive compensation could be implemented effectively with the broader-band system.