Daniel VanArsdale, 3/2004.
After Holzhausen & Sohege, 1899.
White wins. Vao c2. (4+5)

1. Vc2-g6+(a)  f x g6
2. d6          Sc1-d3 (b)
3. d7+         Ke8 x d7
4. Kf6-f7      Sd3-e5+
5. Kf7-f8 wins

(b)
2.  ...         a3
3.  d7+         Ke8 x d7
4.  Kf6-f7      a2
5.  e8(Q)+      Kd7-d6
6.  Qe8-e7+     Kd6-d5
7.  Qe7-a3      Kd5-c4
8.  Kf7-e6      h5
9.  Qh3 x c1+   Kc4-b3
10. Ke6-d5      h4
11. Kd5-d4      h3
12. Qc1-c3+ and Kd4-c4 wins

(a)
1. d6?         Sc1-d3
2. d7+ (c)     Ke8 x d7
3. Kf6-f7      Sd3-e5+
4. Kf7-f8      Se5-g6+

(c)
2. Vc2 x h7    Sd3-c5 (not a3 3. Vh7-g6+)
3. Vh7-b1      Sc5-e4+
4. Kf6-e5      Se4 x d6
5. Ke5 x d6    a3

(a)
1. Vc2-d1?     h5
2. d6          Sc1-d3

(a)
1. Vc2-e4?     Sc1-d3 (or a3)
2. d6          Sd3-c5
3. Ve4-c6      Sc5-e4+ and S x d6

[1. V-N6]

Below is the original study with all orthodox pieces. The key is of course g6 [P-N6]. We feel the study has been improved with the use of the Vao since: (1) the key in the revision is more paradoxical since one step has been added to the path of the Black Knight in its attempt to reach the key defensive square g6, (2) White's only piece is sacrificed, despite the potent threat of V-h5+ and (3) the Black Pawn on h7 is essential to prevent 1. Vc2-d1 from cooking the problem (if 1. ... Sc1-e2  2. d6 and d7+ wins). In the 1899 version the Pawn on h7 is not essential for any line of play, being mere window dressing to make the key appear even more paradoxical.

It may be objected that the Vao is incidental to the revised problem, being immediately sacrificed to block g6. But no other piece, such as a Bishop, would suffice, for it could simply wait for Black to play Se5-g6 and then capture the Knight. The Vao works in preventing such a refutation for the same reason that the Pawn did in the original study: both pieces can move to a square that they can not capture on. Probably many new remarkable endgame themes can be discovered by adding one or more Chinese line pieces to the board. And as in the above example, quite a few classical themes may find entertaining new manifestations.

The original Holzhausen & Sohege study has been reprinted as number 102 in "1234 Modern End-Game Studies" by M.A. Sutherland and H.M. Lommer.