Moving black's king to h8 loses immediately to Bxg7. Capturing the knight loses two moves later after Qxh6 (not counting the spite check with black's rook).

<32. Nh6+ gxh6>

(32...Kh8 33. Bxg7#)

<33. Qxh6> 1-0 as Black has no satisfactory way of defending himself from mate:

A) 33...fxe6/...f6 34. Qg7#

B) 33...Bf8 34. Qg5+ and 35. Qxg7#

C) 33...Bf6 34. Bxf6 and mate is unavoidable.

D) 33...Rc2+ 34. Kh3 Rxh2+ (34...Rxb2 35. Qg7#) 35. Kxh2 and mate cannot be stopped.

<i)b. 34...Rxb2 or fxe6 35. Qg7 or even Qg5#>

Only 35. Qg7# should be played! Not 35. Qg5+?? Bxg5 0-1

<ii) 32. Kh8 33. Nxf7#>

33. Nxf7+ is not mate and is inferior to 33. Bxg7#

I apologize if this note seems insensitive, but you need to take care when commiting the ideas that have have welled forth from you into the written format.

You are quite right <35. Qg5> is erroneous in this context and it did not belong here. It should have been 35. Qe7# alone.

Now that we are at it I will show you how it crept in.

I was thinking of the following line; (I didn't even write it as it looked outright extravagant and inane)

32. Nh6+ gxh6 33. Qxh6 Bf6 34. Bxf6 Rc2+ 35. Kh3 Rxh2+ 36. Kxh2.

now when I was at this node of thought I said to myself 'both 36...Nxf6 and 36...fxe6 are useless as the White Q mates from g7 with the help of either minor that survives. I suppose Black may just as well make a silly looking move like 37.. a6, Qxb3, Nc6 or Nc5 and White would mate from 37. Qg7# or even Qg5#'.

And that last move 37. Qg5# was a bit of residual memory that I inserted in the wrong place. (Here even 37. Qg5# works).

The solution is refuted by 1...Qxf8 to which we could find no answer(for a 4-mate).>

Yes, you're (both) right - there's no mate in 4 after 1. Rf8+ Qxf8 (!). A fairly typical piece of carelessness on Reinfeld's part.

Your suggested move is playable and keeps an advantage (and a likely win - only after a fairly difficult endgame though). The game continuation is better - see analysis....