When the abscissa errors are smaller than a single sampling interval, subtraction becomes more difficult to use. One can circumvent this problem by decreasing the sampling interval. However, this solution may be unsatisfactory, because the considerable efficiency associated with collecting, transferring, and storing small data sets must be sacrificed. For the authors' purposes a technique was desired that could be applied to any set that meets the Nyquist criterion. In addition, it must be capable of being applied to the fairly noisy data that are obtained in Raman measurements. Experimental evidence indicates that the least-squares fit substantially reduces the effect of noise. It should be understood that although the least-squares-fit results bear a resemblance to the subtraction results, they are suitable for qualitative use only. The actual values measured are affected by the form of the probe function.
