This variation carries information about the composition of a scattering medium such as sea water. Finding links between the slopes of the NU7441 spectra and the type of scattering particles would require a number of additional studies to be carried out. However, these graphs (Figure 1) provide insight into the diversity of these spectra, showing that the spectral effects of light scattering in such quantities

as the scattering coefficient and backscattering coefficient derived from scattering at different angles should not be neglected. This was the motivation for the considerations presented below. Knowing the measured particle VSFs and their integrals (bbp and bp), one can then find a spectral relation between them. In Figure 3 SB203580 measured values of bbp were plotted against the particle VSF for 117° ( Figure 3a) and against the particle VSF for 140° ( Figure 3b). One can see that all the points in Figure 3a can be fitted with one linear equation (the best linear fit does not depend on wavelength λ) with a good correlation coefficient R2, whereas in Figure 3b each wavelength requires a different linear fit (the ratio of bbp to βp(140°) varies with wavelength). The linear regression lines as well as the correlation coefficients R2 were put in the figure for each wavelength. On the basis of all available measurements

made in southern Baltic waters it was found that for a scattering angle θ = 117°, function χp can be approximated by a single value of 1.07 for all the wavelengths examined. Boss & Pegau (2001)

proposed a value of χ(117°) = 1.1; these authors claim that this value is the same regardless of whether we consider the particle only (water removal approach) or both the particle and the water (total approach). According to the uncertainty of measured VSFs, which is about 5%, these values are in good agreement. For a scattering angle Fenbendazole θp = 140° the value of χp(θ) changes from 1.06 to 1.19 in the range of wavelengths examined; χp(θ) increases almost linearly with increasing λ. This relation can be described by a simple equation (with a high correlation coefficient R2 > 0.99): equation(5) χpθ=140∘=0.3λ443+0.76. The spectral variabilities of χp(θ = 140°) and χp(θ = 117°) are shown in Figure 4. The standard deviations shown in Figure 2b indicate that for longer wavelengths the value of βp(117°) is better for obtaining the backscattering coefficient than βp(140°) because of its greater accuracy. This is consistent with the results of Sullivan & Twardowski (2009), who examined millions of VSFs obtained from MASCOT. Their measurements were carried out with a low angle resolution and for one wavelength only. My results ( Figure 2b) show that for θ = 117° standard deviations of χp are 0.05 for all the wavelengths, while for θ = 140° the standard deviations are higher (for the longest wavelength of 620 nm the standard deviation is also the highest).