Circe: A fairy chess condition

The realm of fairy chess encompasses countless types of unorthodox rules and pieces, and in previous columns some accessible examples have been introduced, namely Kamikaze Chess and the Grasshopper and the Nightrider. This time we examine one of the most popular and fruitful of all fairy forms: Circe. The Circe condition has spawned many sub-variants but all are based on the idea of rebirth after a capture. Under the rules of standard Circe, a piece that gets captured is immediately reborn on its home square. Thus, a captured white queen instantly reappears on d1 and a black one on d8. A rook, bishop, or knight that’s taken returns to its home square that’s of the same colour as the capture square. A captured pawn is reborn on its original second rank, in the file where the capture occurred. For instance, in the first diagram, if the white queen captures the bishop, the latter (on a white square) reappears immediately on c8, and the move is written as Qxb1(c8). The black pawn on e2 returns to e7 when captured: Qxe2(e7). If the rebirth square of a captured piece is occupied, the unit disappears as in normal chess, so here Qxa3 is orthodox because h8 is blocked; a piece that’s taken on its own home square would also be removed.

Circe rules enable interesting effects that are not possible in orthodox compositions. An oft-seen motif is that of immunity from capture for a unit whose rebirth would place the capturer's king in check. Suppose the d1-square is vacant here, then 1…Bxd3(d1)?? is illegal as the reborn queen would be checking the black king on g1. This problem actually presents another special idea in Circe, one relating to underpromotion. In orthodox two-movers, black promotions to rook and bishop are generally ignored as defences since they can’t be preferable to a queen promotion (there’s no time for Black to concoct stalemate). But in a Circe two-mover, such black underpromotions could produce distinct play because the new piece and a queen will have different rebirth squares.

Before analysing the main promotion variations, let’s consider a few non-thematic tries. Strong captures on e2 involving multiple threats are dealt with by plain black checks: 1.Qxe2(e7)? Bxf5(f2)+! and 1.Rxe2(e7)? Qxd4(c1)+! (all rebirths are incidental here). The e3-knight controls flights on f1 and g2, and if White opens the B + S battery prematurely, we find 1.Se~+? Kf1 2.Qxe2(e7), but 1…Kg2! refutes due to 2.Qxe2(e7)+ Bxf5(f2)+! when the queen check is intercepted by the reborn pawn. Using an opponent’s unit to block a check against oneself is another Circe-specific tactic, one that will be seen repeatedly in this and the next problem. The correction try 1.Sc2+? cuts off the black bishop for 1…Kg2 2.Qxe2(e7), but now 1…Kf1! defeats since the knight has also interfered with the b2-rook.

The thematic defences are prominent checks by the e2-pawn when it captures the knight and promotes to a queen or a bishop. Either move gives White access to d1 and opens white lines of guard to f1 and g2 (freeing up the e3-knight), and this results in two potential recapture mates by the queen and the knight. In the set play, 1…exd1=Q+ allows 2.Qxd1(d8) but not 2.Sxd1(d8)+? because of 2…Qxd4(c1)+!, while 1…exd1=B+ gives 2.Sxd1(c8) but not 2.Qxd1(c8)+? in view of 2…Bxe6(f1)! In both cases the wrong recapture enables the reborn black piece to counter the white check. The actual play begins with 1.Bd5!, which threatens 2.Rh1. Black’s only defences are the two promotion checks, which lead to remarkable changed mates brought about by the placement of the key-bishop. Now 1…exd1=Q+ permits 2.Sxd1(d8) and not 2.Qxd1(d8)+? Qxd5(f1)!, while 1…exd1=B+ admits 2.Qxd1(c8) and not 2.Sxd1(c8)+? Bxf5(f2)+! Again White has to recapture carefully to avoid spoiling defences by the reborn black pieces. Since White has reversed the replies to the two defences compared with the set play, this problem is an elaborate demonstration of the reciprocal change theme.

In this helpmate position, White wants to deliver a rook mate that will cover the a1-flight, but with no direct way of moving either rook to the first rank, we expect Black to assist by capturing one on a white square to transfer it to h1. To avert the self-check such a black capture entails, White can first close the rank with the d2-bishop, which then forms a battery with the reborn rook. So let’s try 1.Qd6 Bc1 2.Qxa6(h1), but now it turns out that the bishop cannot execute a battery mate because wherever it goes, Black will be forced to capture the piece and bring it back to c1, e.g. 2…Bh6+ 3.Sxh6(c1). (There’s no difference had the d2-bishop moved to e1; and White could even try with the other bishop: 1.Qd6 Bd1 2.Qxa6(h1) Bb3+ compels 3.cxd2(c1).)

The solution involves a similar plan but begins with an immediate capture of the white bishop: 1.cxd2(c1), the advantage being that d2 is left unguarded by Black and hence it becomes a safe square for the battery-firing bishop. Now with this bishop already on c1, White has a spare move with which to shift either rook to a white square where it can be captured by the queen for rebirth on h1. However, the options for where to sacrifice the rook all seem to result in a disruptive black queen check, e.g. 1…Ra8 2.Qxa8(h1)+, or 1…Rxg8 2.Qxg8(h1)+. Only 1…Rc6! works because the check 2.Qxc6(h1)+ is answerable by 2…Bxd2(d7) – a Circe-specific kind of cross-check. For part (b), in which the white bishop is placed on e3, Black again starts by capturing the bishop, 1.dxe3(c1). Looking ahead at the mating capture on e3, we realise that the reborn pawn on e7 can be similarly used to neutralise a queen check; therefore 1…Re6! 2.Qxe6(h1)+ Bxe3(e7). The remaining twins show perfectly analogous play: (c) 1.exf4(c1) Rh5 2.Qxh5(h1)+ Bxf4(f7), (d) 1.Bxg5(c1) Rxg8 2.Qxg8(h1)+ Bxg5(f8), and (e) 1.Sxh6(c1) Ra8 2.Qxa8(h1)+ Bxh6(b8). In this top-notch battery creation problem, the black queen checks the king from five different directions, and every part shows a pair of reciprocal captures between the white bishop and a black unit.