Energy-dispersive XRF analysis provides a fast, non-destructive way to determine element distributions of samples. Once a calibration is done, the method provides an almost effortless way to achieve quantitative information. However, these calibrations are most often not linear and do not pass the origin thus they are only applicable in a very narrow range around the used standards. With the 2008 presented log-ratio calibration equation (LRCE) a promising method that aimed to get rid of this non-linearity has been introduced. The current work investigates how well the method achieves this goal by measuring a set of seven standards with a micro-focused XRF setup and a commercially available hand-held XRF device from Bruker. Parameters goodness-of-fit R2, slope alpha and y-intercept beta have been calculated and compared with both XRF devices as well as the Compton peak correction and the raw data set. Prior to the micro XRF measurements the setup had to be rebuild due to a relocation, thus requiring a check of the system performance. This was done by a long-term drift measurement, ensuring that the same results are achieved in every measurement. Apart from these calibrations, measurements with the hand-held device have been carried out in the historic japanese tower in Brussels. The drift measurements pointed out a slight drift for high count rate elements but the error could be rendered insigni cant when correction methods are applied. The calculations of the micro XRF setup reveals that the LRCE correction is favorable to the Compton peak correction. The performance is especially good for poorly correlated elements. For the hand-held device only elements with low signal and strong air absorption perform better with the LRCE. The measurements of the historic site in Brussels could identify some of the applied materials an techniques on paper and wood basis.