What are the 1.shortest and 2.longest accurate rounds A)possible and B)actually played?

"accurate" as in blunderless, and preferably no objectively 'weak' moves.

And played to natural conclusion, i.e. Draw per 3fold, 50move, et c. but not by 'alternate agreement's premature ly

•  a far as longest possible, that would be predicated on 50-move rule: *16 pieces + *(maximum pawn moves?) + * 16( pawns/pieces promoted from pawns)... obviously reduced immensely with restriction of _accurate_ play (which could go otherwise accurately much longer without 50move rule" as suggested by drawn endgame tablebases surpassing 100 moves from larger tablebases and suggested by Checkmate-in-200+ positions).

One possible contender for the shortest non-imperfect round actually played that I discovered (with thank to Yigor, https://www.chess.com/forum/view/chess-openings/acuteness-of-openings) is the Immortal Draw game (http://www.chessgames.com/perl/chessgame?gid=1250160), which at a mere 18 moves played by either side was acutely forced into a draw within 16 turns commencing with aggressive 3..Bxpf2+.  I call this round 'non-imperfect' instead of the stronger term 'flawless' or further yet 'perfect' because Black's initial sacrifice on turn3 is probably weak, indicated by stronger candidate moves in that position. However, all preceding half-moves played in that round, as well as all to follow( starting with 4.KxB and concluding with 18..Bb2+ two moves short of three-fold repetition / perpetual check), in accord with the 'acuteness' of them, were strong and in fact the only 'strong' moves available (even though other ones may have still resulted in draw, they were substantially weaker as measured by engine, and indeed much deviation would blunder into loss for misstepped side).

While shorter forcible non-imperfect rounds are perhaps possible, I doubt much shorter than that considering necessitation for many. accurate/necessary exchanges or sacrificings ---notwithstanding the unlikely possibility for an acute 3fold or perp.check to be possible to arrive in a blunderless game earlier yet (and at that, even if performing a repetitive move concluding draw is quantifiable 'acute', if there is another possible move in the non-obtuse position that would not reduce the analyzed evaluation to lower than -0.9 or so i.e. still with odds for a draw and potentially even victory, then would that not be 'stronger' since obligating opponent to play on potentially to make errors? otherwise why not dejure default-draw right away).