12/18/2000 Fixed an inconsistency in the interpretation of relative contributions of different subpopulations of particles in c(s) versus c(M) distributions. (This is relevant only for the c(M) distribution when applied to broad heterogenous mixtures; but both the c(s) and ls-g*(s) functions are unchanged). The problem in the previous versions was that the calculation of relative populations in the c(M) method [after export to a spreadsheet program such as Origin] had to be performed by summation over the discrete data points in the relevant range of M-values. Sedfit now rescales the c(M) distribution properly to allow for integration instead of summation. This can make a difference, as the discrete M-values are not linearly spaced, but spaced on a 3/2-power scale. If not taken into account, previously this would have led to a slight overestimation of the larger population. In contrast, the c(s) and ls-g*(s) distributions are calculated on a linearly spaced grid of s-values, such that the relative populations calculated by summation or integration are identical. [Another, still existing problem with the c(M) distribution as opposed to the c(s) distributions is that it requires good estimates of the frictional ratios, which appears even more difficult for broad heterogeneous mixtures. Because the c(s) distribution is significantly more stable in this respect, it is in most cases probably much better.] A related question occurs when calculating diffusion coefficient distributions and Stokes-radius distributions from dynamic light scattering data. Both the I(rh) and I(D) distributions are calculated and displayed on a logarithmic scale, with a the differential distribution functions refering to the log(D) and log(rh) data, respectively. This allows an intuitive estimation of relative populations from the area under the curve in the log units. However, if the data are replotted in linear units (e.g. rh in nm) then the distribution also needs to be logarithmically rescaled to ensure that the integral over subpopulations still represents the correct relative contributions to the total intensity of the scattered light. There is now checkbox in the parameter input box that allows to determine if the distribution is scaled properly for linear integration (default) or for the logarithmic integration.