of the total corrected sum of squares for BIOMASS. When pH is used in a model with the other four variables, however, its partial sum of squares, 1,917,306, is only 10% of the total sum of squares and less than 15% of the regression sum of squares for all five variables. On the other hand, the partial sum of squares for SALINITY is larger than the simple correlation between BIOMASS and SALINITY would suggest