but the "eyeballed" estimates can only be off at the very very most 5 hp for any given rpm value (the middle of the dashes on the gridlines increase by 10 hp). the car being dynoed might vary this much just between different runs. after a point getting a better estimate of the regression, and the integral is pointless, unless it is to be averaged with many more integrals calculated from many more dynos under the exaxt same conditions. just my opinion.