Endgame Study - J. Jespersen, 1907.

White to play and draw.

https://www.chess.com/blog/lodrac91/j-jespersen-1907

There is the obvious dual 1. Bd6 (.Bg3 works as well) which plays the same solution 2 moves faster. Hard to believe that was overlooked by the author. Could not find a simple repair though.

The initial position may be wrong? I will try to check it in HHdbVI.

What about this correction if really needed?

B on b8 is better I think.

Or maybe just a W pawn is missing on e4.

The version in #3 works but #4 is dualed. Good idea to check HH's database. I have no license for it so I can't get there. Let's see what it says!

HH's database gives the cook but no correction, I have updated the solution with the e5 pawn correction.

https://www.chess.com/blog/lodrac91/j-jespersen-1907

It's worth adding the variant 1 .. Kd5 2. e6! a2 3. Be4+! to the fix which has individual significance.