The first clue indicates that the noise is low in frequency. The second and third clues both indicate that the noise sources are multiplicative; that is, the noise scales linearly with the signal. The SNR of spectra measured using a chopper (AM) is much worse than the SNR of spectra measured by phase modulation (PM). PM is a derivative technique which removes from the detected signal a large DC component of radiation not modulated by the interferometer. This DC component can carry multiplicative noise, so it is desirable to remove it. As noted in Slide 4, the noise level in dynamic difference spectra is 1 to 2 orders of magnitude lower than in the corresponding single-beam spectra. This also points to a multiplicative noise source which is discriminated by the second demodulation. The final clue indicates that a significant part of the noise is below 1 Hz. In certain experiments the averaging time, t, at each step was increased from 0.5 to 8 seconds, and resulted in an increase in SNR significantly greater than the expected Öt.