How does changing the datum affect accuracy?
Therein lies a (pretty obvious, if you think about it) secret of good trimsheet (in fact, any graphical system) design.
For a longhand (or PC) calculation, for all reasonable purposes, datum position doesn't matter an iota. However, that is not the case when it comes to drawing pictures ...
Consider the datum position -
(a) if a long way forward of the envelope, the envelope graph will be long and thin and stretch from lower left to upper right - typical GAMA POH style which, in turn, is based on an ICAO sampler
(b) if somewhere inside (or near) the envelope, the envelope will be squarish and boxy in shape
(c) if a long way aft of the envelope, it will be like (a) but sloping from lower right to upper left - can't imagine that anyone would bother doing this
Now ..
(d) if you are not convinced that this is the case, take your favourite aircraft's AFM and rework the CG envelope (weight by moment) for such datum positions ... then plot the co-ordinates.
(e) without running any sort of error analysis, consider the way one can scale the envelope variations ... and the resulting physical chart dimensions.
(f) in case (a) one might have, say, 10 inches of CG travel (forward to aft limit ... represented as moment) drawn with a dimension of, say, half inch, between forward and aft limits ... simply because you have to fit in the sloping characteristic on the page
(g) in case (b) the slopey bit goes out of consideration and you can stretch the envelope laterally so that, say, the 10 inches might be scaled to fit on, say, 5 inches of paper
(h) when completing a trimsheet, which of (a) and (b) would you reckon might give you the more accurate answer ? (where "more accurate" = "lesser error")
(i) re (b), the designer ends up having to be careful that he/she doesn't go overboard and make the scale too generous.
As this thread develops we may end up doing a critique on whatever sheet the originator has in mind .. and all will become patently clear.