Confidence interval about regression parameters

Neither. These tell you about the linear relation between y and x, true, but in analytical chemistry you are rarely testing the linear model. The standard error of the regression (Sy/X) is a useful number to quote, or calculate 95% confidence intervals on parameters and estimated concentrations of test solutions. Plot residuals against concentration if you are concerned about curvature or heteroscedacity. (Sections 5.3.2, 5.5)... 

It is often important not only to calculate the values of unknown constants but also to statistically evaluate the obtained results. Such estimation tells a lot about the quality of the solution. For instance, the calculated kinetic parameters are sometimes reported with the corresponding confidence intervals. A few formulas from regression analysis are enough to find these intervals. For example, one can calculate a sum of square errors SSE of the experimental data from calculated regression line (the so called least squares method) ... 

---