The currently accepted theory of spectroscopic noise causes most spectroscopists to accept the statement that when the limiting noise source is constant detector noise (e.g., as in mid- and near-IR spectroscopy) the noise of a transmittance spectrum is also constant, and independent of the sample transmittance. A careful examination of the effect of noise on spectra, however, reveals that most of the previous derivations have neglected the effect of noise in the reference reading on the noise content of the spectra. The consequences range from the fact that even with constant detector noise, the error of a transmittance spectrum is NOT constant, to the fact that the optimum value of sample transmittance, to obtain best relative precision, is not 36.8%, as usually stated, but 33%. We also present the results of experiments performed to distinguish between the results predicted by the two theories and to verify that taking the reference noise into account does, in fact, make a difference. “The trouble with the world is not that people know too little, it's that they know so many things that ain't so.” (Mark Twain)
On-line analysis of chemical composition using an FT-Raman spectrometer in the near-IR