The charts and graphs found in the AFM originate from numerous data plots from flight testing. As these are not always nicely ordered, and a few 'wild cards' have to be eliminated, a process of 'curve fitting' either by hand, suitable algorithyms, or computer generation are used, with all due consideration of 'break-points' (which are numerous and highly significantt if interpolation is to be used).
This 'curve fitted' data is then broken down into millions of data points at small intervals, for use in Computerised applications.
Data points at much more 'pilot useable' intervals are taken from these. Where the performance engineer has to exercise extreme caution is to ensure that linear interpolation will yield at least, accurate data, and at worst, conservative data. In numerous cases, worst case assumptions must be made where graphic or algorythmic data is simply far too complex to put into a 'daily use' pilots manual.
For the reasons of the necessary conservatism mentioned, manually prepared data by the pilot will inevitably be conservative. I've done considerable work in applying 'pure' performance manual data to the process of RTOW creation following airport analysis, this applies all of the bends and twists in the real data with accuracies down to a fraction of a Kg. The general 'day to day' charts for the same aircraft I've created for pilot use, make many conservative and general assumptions. Be assured that if you find a better weight from the general data than the computer generated RTOWs provide, you've made a big mistake. At best, they should just be equal.