That doesn't sound quite right. The quoted maximum SP separation is to ensure that scale expansion and contraction does not exceed 1% (from memory) and the implication must be that, if a larger SP spacing is chosen scale expansion and contraction exceeds 1%. Its nothing to do with 'accuracy of detail'. The scale is not 'absolutely correct', it is approximate, unless the large projection used to generate E(LO)2 is compensated for by having different, more accurate, scales quoted when sections of the projection are taken, more accurate for their latitude with respect to the SP. Still they will not be 'absolutely correct', only approximations.