I guess that you want to show proportionality (=linear regression through origin) and not just a linear relationship. An alternative way to look for proportionality is to do linear regression on log-transformed data:
LogY=a*LogX+B
The criteria for proportionality is then that "a" is 1, or at least not statistically different from 1. It is difficult to tell whether this approach will help you. It is often a great approach when looking at data sets that span more than 3 orders of magnitude, and where the observations in the high range are dominating the figure and the regression.
First of all you should plot a graph. Visual inspection is almost the best way to see if the linear approach is the most suitable. If there are some ambiguities you should plot standardized residuals that can be helpful in revealing possible outliers as well as to check for the presence of any trends like curve line or straight lines decrease or increase, or weighted increase in residual error. High values of R2 are meaningless if inspection of residuals suggest using curve line calibration. Besides R2 values, you should give values of standard deviation, number of calibration points, as well as F i.e. corresponding probability P value that states that R is statistically significant.
You can use to study the coefficient of linear correlation Excel or a software like SPSS or StatGraphics, all what you have to do is calculate R square according with the regression analysis between Absorption and concentration, using different concentration of an standard solution similar to the sample you are going to study, if the R square is near 99% there s a good correlation between your standards and the absorption at a wave length using the same cuvette.