The British Chess Problem Society
Caissa's Playthings — by T. R. Dawson
Back to — BCPS homepage.
This series of articles on “Caissa's Playthings” was the earliest partially systematic account of fairy chess published by T. R. Dawson, and it still provides an excellent introduction to the subject for beginners. It appeared in the extensive chess column edited by W. S. Branch in the Cheltenham Examiner newspaper. Dawson later expanded on this account in other publications, notably Caissa's Wild Roses (1935). The text is presented here in a single column justified format so as to give an impression of a newspaper column (though the original is much narrower and in smaller type). This also provides room to insert a few notes alongside.
Quick links to sections on this page:O Introduction
I Fers {1,1}
II Alfil {2,2}
III Wazir {0,1}
IV Dabbaba {0,2}
V Camel {1,3}
VI Giraffe {1,4}
VII Terror (Q + K)
VIII Chancellor (R + Kt)
IX Centaur (B + Kt)
X Dragon (P + Kt)
XI Grasshopper
XII Neutral
The Weekly Articles Notes


  The intention of this essay is to survey in as complete a manner as possible the chief “new” chess pieces that have been added from time to time to the orthodox seven.[2] The treatment is historical, dealing with early changes; and technical, pointing out the most prominent characteristics of each and exemplifying these in problems. The ground covered is restricted to Indo-European chess.
  At the present moment we have reached a settled period in which little change in chess principles is likely to occur. The settling of the dummy pawn controversy saw the end of important changes.[3] In the Morning Post in 1912 I noticed a letter advocating the reintroduction of the dummy pawn to promote originality.
  This leads us to glance at the justification of this or any other new power to a piece. The most obvious reason is the one just quoted, the promotion of originality. One need only read one of Mr A. C. White's essays of the classification of problems to find that all modern composers may hope for is technical novelty — the using of two bishops where a bishop and queen were required before, the doubling of what hitherto was only a single theme, and so on.[4] The infusion of fresh interest by a study of these “new” pieces should therefore be valuable. There is a deeper, more logical reason than this, however. The widest and most general definition of chess pieces as forces and of chess play as changes in equilibrium in these forces is an old and well-established one. Quite recently the analogy was carried so far that chess problems were termed equations.[5] This conception of chess is invaluable from the point of view of these new forces. To introduce such and their new conditions of equilibria into our present system is clearly quite compatible with the definition and no real objection can be raised to these fresh factors. In fact, to make a complete study of chess it is essential that we bring such factors into our reckoning. To omit them would be equivalent to a study of mathematics which stopped at algebra. There is a third reason most potent of all, the human reason. All of us like to dabble in these queer, out-of-the-way, concoctions. They fascinate us, even if we do not treat ... [6].

[1]Cheltenham Examiner 10 April 1913.

[2] In counting seven orthodox pieces TRD is presumably counting the Bishops on black and white squares as different. In CWR (1935) he reduced his count to six.

[3] Presumably Dawson is referring to more recent happenings than Allgaier's (1795) handbook or Staunton's laws for the 1851 London tournament.

[4] The supposed end of orthodox chess problems is a perennial prophecy, but new compositions are still published in great numbers.

[5] Can anyone identify the work described here?

[6] There are a few lines missing from the end of my photocopy of this section.

I — The Fers or Mediaeval Queen. [1]

  Strictly speaking, of course, our own present-day queen and bishop are most unorthodox pieces. The original chess had no such powers. Since, however, I write as a modern, it will be only fair to treat the early queen and bishop as forces new to our twentieth century chess.
  The Persian name for their fourth best piece was fers, or fars, signifying a grand vizier or minister, and derived from the Persian farzin — “distinguished”. The term was borrowed by the Arabic scribes, and though untranslatable in Arabic, the term “fers” travelled with the piece. [In a similar way we get “rook” from the Persian rukh — a war chariot or car.]
  A great deal has been written on this and other mediaeval pieces,[2] so, briefly, I may say it moved one square diagonally in any direction. At a later period a queen made by promotion could for its first move leap to the next square but one to its standing square in rank, file or diagonal, providing such square were unoccupied. These features are all illustrated in the following problem.


White mates in two; the Q at g8 is a new one.

  The solution commences 1 QB3. White now threatens QKt4 mate. If Black defends by QKt4, the new queen at Kt8 leaps to Kt6, giving mate. If QK4, new queen leaps to K6, giving mate. If RxQ or RKt6 or R5, then KtR4 mate, and if anything else, the threat.

[1]Cheltenham Examiner 10 April 1913.

[2] W. S. Branch follows this article with extensive historical notes about the fers, including similar pieces in Chinese, Burmese, Siamese and Japanese chess. I believe he had done extensive research on the history of chess himself, and was not relying on H. J. R. Murray's A History of Chess which was first published that year.

II — The Alfil or Mediaeval Bishop. [1]

  The alfil has possessed quite a number of names, these being chiefly variations in spelling in the old MSS. Such were alfin, alfyn, alphyn, and aufyn. Alfil is the Arabic from al fil — the elephant.
  The move is diagonal, being a leap to the next square but one to that on which it stands, whether the intermediate square is occupied or not. The illustrative problem is a hybrid — ancient and modern blend. I composed it at a time when I thought the Mediaeval pawn could promote to any mediaeval piece. (As a matter of fact, only a queen may be made.)


  The solution goes 1 PB8 (Mediaeval B), KQ2; 2 RB7ch, KK3; 3 PKt8 (Mediaeval Bch), mate! If 2—Kb3, 3 PR8 (Mediaeval b), mate! If 2—K else, 3 PR8(R) mate. If 1—else, 2 RB7 as above.

  I have treated these first two men very briefly, since so much has been written elsewhere of them and there are a large number of problems extant showing their peculiarities.

[1]Cheltenham Examiner 17 April 1913.

III — The Wazir. [1]

  Timur or Timour was a great tartar soldier who conquered Northern India and there discovered a game[2] which interested him so much that it is now known as Timur's Great Chess. It was played on a board 11 by 10 squares with two extra ones of special properties. There were a number of new pieces added to the normal chess array to fill in the vacancies.
  The Wazir, or Commander in Chief, was one of these. It could move like our rook, but only one square at a time. Shortly after Timur's period the name was changed in Central Asia, Northern India and Persia to Fish or Crocodile. A “Great Chess” is still played in India by native princes and higher dignitaries, so presumably this piece is still in use. Professor Brownson made some problems with this piece, under the name of Mero. They appeared in the seventies in his Dubuque Chess Journal and in his collection of problems dated 1876.
  Its chief characteristic is its slow going! Fancy having to zig zag fourteen moves to traverse a long diagonal. The slowest modern piece, the King, only takes half that number of moves. This superiority of the King is here illustrated:—


  The inverted rooks at a8 and h1 are wazirs. White mates in 10. The theme is simply “grab” — a duel between the white king and the W at h1. Play 1 Kb1, W.R7; 2 KKt1, W.R6; 3 KKt2, W.R5; ...; 8 KKt7, W checks; 9 KxW, and now either W.Kt1 or KR3 is forced when 10 RR5 mate.

  The only other point on which we need be curious about the wazir is his holding power. It only takes 5 queens or 8 rooks to guard the whole of a chess board, but no less than 20 wazirs are required. There are a great variety of placings; here is a symmetrical one. Put wazirs on: a2, a5, b2, b5, b7, b8, c5, d1, d2, d3, e6, e7, e8, f4, g1, g2, g4, g7, h4, h7.

[1] Cheltenham Examiner 24 April 1913.

[2] In a following note, Branch queries H. J. R. Murray as to whether Timur was acquainted with his game before he saw India. Murray opines: “I think there can be little doubt that the game originated in Persia (possibly in Turkestan), not long before Tamerlane's day. No work before Al-Amuli's encyclopedia knows it, or of any game with citadels. Al-Amuli died in 1352. He gives it among his five varieties of chess.”

IV — The Dabbaba. [1]

  This piece is also from Timur's Great Chess. The name means a movable shed or tower on wheels. This was pushed up (before the days of artillery) against a town or castle wall in a siege, either to act as a scaling ladder or as a cover to mining. In European warfare it was named a “sow”.
  The piece moves rookwise, but only leaps to the next square but one to that on which it stands, just like the mediaeval bishop in fact, but rectangularly instead of diagonally. Its most striking characteristic, of course, is its leap, illustrated in the subjoined position.


  The inverted rooks, two black, one white, are all dabbaba. White mates in two by 1 KKt3! — B any; 2 DB2 mate — RR5; 2 DxR mate — QKt any 2 KtB3 mate — KKt any; 2 KtKt4 mate.
  Here are no less than four smothered mates. A weakness of the dabbaba is also seen in this problem. He is easily blocked in by his own men, as witness both black Ds in this diagram.

  The holding power is very easily studied. Notice a D on c4 can only command a4, c2, c6, and e4, and can only reach in all a system of sixteen squares, all white. Similarly the other sixteen white squares form a system on which a D might travel, and the black squares make two more like networks. Hence we only need to study one system and multiply by four to learn the constants of the whole board. If four Ds be placed on a6, c8, g4, c2, the whole of one system is guarded or occupied. If six Ds be placed on b4, b6, f2, 4, 6, 8, the whole of another system is guarded. Thus we require sixteen Ds to guard and occupy or 24 Ds to entirely guard the board. This compares unfavourably with the Wazir, which, as we saw, took 16 and 20 respectively.

[1] Cheltenham Examiner 8 May 1913.

V — The Camel. [1]

  This is still another of the Timur pieces. It is very similar to the knight in move, but reaches slightly further, traversing the diagonal of a rectangle 2 by 4 squares.[2] Thus a camel on e4 commands d1, f1, h3, h5, f7, d7, b5, b3. It has full power to leap, of course, and so is yet another piece capable of giving smothered mate. Here is an example of this.


  The inverted Kts are camels. White mates in 2 by 1 Q-Kt5! threatening 2 QxKtP mate. If 1 PxQ 2 CxP mate. If 1 KtxR 2 CxKt mate. If 1 P-Kt3 2 QxBP mate. If 1 else; 2 threat.
  This problem also illustrates the unhandiness of the camel in defence. It reaches too far out and both the White Queen and Rook easily slip inside its guard with impunity and unhampered freedom.

  Two Camels and a King can not force mate against a lone opposing King as can be proved thus. Put Black K on b8, White K on b6, and white Camels on f7 and d4.[3] Note the Cf7 is necessary to keep Black K from c8. Now if Black move to a8 — suppose Cf7 could check at b5. K would be driven back to b8 and C to a5 would be mate. But observe in checking at b5, the C guards c8 just as it did on f7. Now a Camel, just like a Kt, if it guards a square, must take two moves to gain another square which guards the first, so that from any square guarding c8, the Camel must take at least two moves to reach b5. hence it follows the C can never guard c8 one move and check on b5 next, and so two Cs cannot force mate.

  Here, however, is an amusing exercise in which one C and one Kt between them force mate, this being possible because the C is far enough away to let WK get near on one hand, and because the Kt is handier in changing colour on the other!
  White:— K on c5, C on e6, Kt on c6, Black:— K on a8.[4] White mates in 6 at most by 1 KtK7, KR2; 2 KKt5, KR1; 3 KR6, KKt1; 4 KKt6, KR1; 5 C-Kt5 ch, KKt1; 6 KtB6 mate.

  A final point worth mentioning of the Camel is that he can make a tour of his 32 square domain similar to a Kt's tour. Here is a specimen which is re-entrant, meaning the 32nd square is just a Camel's move from the 1st.[5] Play from a1 to b4, a7, d8, g7, h4, g1, f4, e5, f6, c7, f8, g5, h8, e7, b8, a5, b1, e1, h2, e3, d6, c3, f2, c1, d4, g3, h6, e5, b6, a3, d2, and so back to a1.

[1] Cheltenham Examiner 15 May 1913.

[2] Strictly speaking the line of the camel move is not the diagonal but from centre to centre of the corner squares of the 2 by 4 rectangle. It is now more usual to describe the camel move as {1,3}, indicating a move equivalent to 1 rook step followed by 3 rook steps in the perpendicular direction.

[3] The position described is:

[4] The problem position is:

[5] Here is a diagram of the tour.

VI — The Giraffe. [1]

  This is the fourth piece from Timur's chess. Its move is one square rookwise beyond the camel's move. This means that the giraffe is another leaping piece, which traverses the diagonal of a rectangle 2 by 5 squares.[2] It will be seen it changes the colour of its square each move, like a knight. Owing to its exceptionally long legs — to say nothing of its neck — the giraffe can never guard more than four squares on the board at once. Thus placed centrally on e4 it only reaches f8, d8, a5 or a3. Moved partly to the edge on f2 it has g6, e6, b3, b1, or right at the edge on d1 it has c5, e5, h2 — never more than four, and in a corner only two, like the Kt and Camel. The following two-er illustrates these long sweeping moves and guards:—
  White: K f2, G f6, P e3; Black: K a1, R b1, G's a3, c1, P's a2, b2, e4; White mates in two.
  Notice if G c1 any; 2 G-Kt5 mate, and if G a3 any; 2 G-K2 mate. We require a waiting move with King for key. The G's guard g2, e2. Therefore 1 K-Kt3.
  The advantage of being so remote from the point of attack is well illustrated in the following little three-er, where the White queen is able to get between the King and the G's to administer mate, even though one of the G'c commands two squares in the king's field on the side further from the White pieces.


  Thus 1 Q-R2, P-R4; 2 K-R4, K-Q5; 3 Q-Q6 mate! If 1— K-Q5; 2 Q-R5, K-Q6; 3 Q-Q1 mate.

  Finally, I see no reason why a giraffe tour of the board cannot be made analogous to the knight's, but I have not succeeded in working one out myself at present.[3]

  Before leaving this portion of my subject — the ancient “new” pieces, I may mention several interesting facts in connection with Timur's Great Chess.
  The king's pawn in this game had the privilege of promoting to a Prince, a piece having all the king's power but subject to capture in the ordinary way.
  There was also a special Pawn of Pawns which was held in hand and, as one move, could be placed on any desired square. At its first promotion this pawn became a second edition of the king's pawn just mentioned.
  There also existed a piece, the scout or talia — meaning a spy — which moved just like our bishop.[4] This was 200 years before our modern bishop was established. Also about the same date that the Wazir was re-named the Fish, were introduced Lions, Bulls, and a Sentinel. The moves of these last men are not clearly known, as the MS mentioning them gives but scanty information. It is surmised, however, that the Sentinel was similar to the modern queen. the nomenclature of fish, lion, bull, was probably derived from the signs of the zodiac.

  There is a great deal more that could be said about the development of Indian chess in China, Japan, and the Far East generally, but this would take us too far from our subject and must be left to a future occasion. The remaining “new” pieces are all of comparatively modern suggestion, and one at least appears to be really new.

[1] Cheltenham Examiner 22 May 1913.

[2] Similar comments apply as for the Camel move. Also the ‘Zurafa’ in Timur's game (according to Murray's History) was somewhat different from the modern Giraffe: after the leap it could continue forward like a rook.

[3] Subsequently, in l'Echiquier 1930's Dawson achieved a 63-cell giraffe tour. In fact a complete giraffe tour is impossible. G. P. Jelliss gave a proof in Chessics #2 1976. Dawson does not seem to have been aware of closed tours of a 10×10 board by {1,4}- and {2,3}-movers (i.e. Giraffe and Zebra as we now call them) given by A. H. Frost in 1886.

[4] Other sources (e.g. Murray) say that the Talia could leap over a piece on a diagonally adjacent square. There was also a bishop-moving piece, the Courier, in the earlier German game of ‘Courier Chess’ played on a widened board with 12 files.

VII — The Terror. [1]

  In 1840 Herr Tressau of Leipsig played on a board 11 by 11 squares with three new pieces. One of these was a General, a combination of Q and Kt. This is a piece one reads of in text books as being capable of mating against a lone King, but so far as I am aware no example of its power was published until, in 1912, I wrote a semi-humorous paper naming the combination a Terror[2] — a particularly appropriate name, as any one who tries its properties will find. Mr G. C. Wainwright has since published a problem embracing this piece.
  The only prominent feature of the Terror is its enormous power. The following problem shows this and illustrates the lone mate.[3] White mates in two.


  1 K-Q4, Black any move; 2 Terror to QB3 mate. It will be seen that Black King has for his move a choice of seven squares.
  In Wainwright's problem by using two Terrors the Black King is given eight flight squares without a checking key.[4]

[1] Cheltenham Examiner 29 May 1913.

[2] The article referred to is probably one of Dawson's series that appeared in Reading Observer that year. The name ‘Terror’ is still occasionally used, but ‘Amazon’ is now more popular.

[3] A note in the next issue reads:— In No VII, there should be a White Pawn on QKt4 on the diagram illustrating “the Terror”. Without the Pawn Black can prevent mate in two by moving P-Kt5. — We show the corrected diagram.

[4] Does anyone have details of this problem?

VIII — The Chancellor. [1]

  The history of this, the most popular new piece ever used in modern days, I quote piecemeal from Mr B. R. Foster's “Chancellor Chess”, a little book published in St Louis, U.S.A., in 1889:—

  “There are four instances where the Chancellor under different names was used on different boards.
  Carrera in 1617 inserted two new pieces, a Campione, having the moves of R and Kt, and a centaur, combining the moves of B and Kt, on a board 10 by 8 squares.
  The Duke of Rutland in 1717 used a board 10 by 14 squares, and introduced two new pieces, a Concubine, possessing the power of R and Kt, and a crowned Rook, with the moves of K and R.
  L. Tressau of Leipsig, in 1840 played on a board 11 by 11 with three additional pieces, an Adjutant moving as B and Kt, a general with the move of Q and Kt, and a Marshall, having the moves of R and Kt.
  And several years ago Mr H. E. Bird, the veteran chess master, suggested a board 10 x 8 and two new pieces, a Guard (R and Kt) and Equerry (B and Kt).
  The Campione, the Concubine, the Marshall and the Guard were old names for the Chancellor”

  This piece is easily the most widely used of all “new” pieces. Many games have been printed showing its powers, and in 1887, in the St Louis Globe Democrat, twenty-seven problems were entered in a Chancellor tourney. There are more than forty problems in “Chancellor Chess”, and in my eccentricity collection are a considerable number more.
  As Illustrating its powers in a very complete manner I use the following task problem, in which the Chancellor is seen giving its maximum of twelve different mates. The Queen can do no more than this.


  The inverted Rook is the Chancellor.[2] White mates in two. The key is 1 C-K4 and according to the defences the C mates on b4, c5, d6, e7, f6, g5, f4, e3, d2, c3 and d4, e5.

[1] Cheltenham Examiner 5 June 1913.

[2] The name ‘Chancellor’ is still popular with variant chess players, but problemists tend to use the name ‘Empress’ which was used by Dawson in most of his later work.

IX — The Centaur. [1]

  As we saw in the historical notes to the Chancellor, the B-Kt combination was suggested at similar times as R-Kt union. Still more recently Mr W. P. Turnbull published a puzzle in Chess Amateur, using two griffins (B and Kt), and in the curiosity column of the Strand Magazine appeared a tour of B and Kt, making a magic square, by Mr E. Wallis of London.
  The piece is a comparatively powerful one, and in suitable positions relative to the black king may give eight different mates. The next problem shows this. The inverted bishop is a Centaur. White mates in two. The key is 1 P-B4, leading to Centaur mates on b2, b3, c2, c3, b5, b6, c5, c6, according to Black's play.


  If the C can get the opposing king into a corner he may mate him there alone, and in any case can keep him in perpetual check. Thus in the following ending,[3] White forces a draw from an otherwise helpless position:—
  White; K on a4, C on f5.
  Black: K on h8, Q on a6, B on b7, Kt on a8, P's on a5, b6, c7. White draws by 1 C-Q4 ch, K-Kt1; 2 C-K6 ch, K-R1; 3 C-Q4ch, K-R2; 4 C-B5ch, etc., forcing perpetual check.

[1] Cheltenham Examiner 12 June 1913.

[2] The name ‘Centaur’ is now seldom used for this piece (it is preferred for the K + Kt), instead Dawson's later term ‘Princess’ is preferred.

[3] This is the ending position:

X — The Dragon. [1]

  The combination of Knight and Pawn leads to some amusing points of strategy. It was first used so far as I am aware in a short paper by myself in the Reading Observer last year.
  At the best it can only give four direct mates, but these are possible from three positions relative to the black king. The most interesting is that in which the D is above the king, as shown in the next problem.


  The inverted Kt is the dragon, which has all the power of Kt and P except promotion. White mates in two. The key is 1 R-R2. If 1 PB4 or P-K4, then 2 DxP, e.p. mate! If P-B6, 2 D-K3 mate; if P=Kt6 2 D-B3 mate, etc. Here in two variations the D moves as a pawn and mates as a Kt, whilst in the other two it reverses these functions.
  The other two positions from which it may give four mates are with black king at e5, say, and D on d2, mating on c4, d3, d4, f3, or with D on e3, mating at c4, g4, d4 or f4, after suitable black moves.

  The king and one D alone can not force mate, but two D's by themselves are able to do so, as in the little twin problem:—[2]
  White: D's on b5, d7, P on g7. black: K on b7. (i) White mates in 2; (ii) Put WP on g3 and mate in 4. In (i) simply play 1 P=B, etc. In (ii), 1 P-Kt4, K-R1; 2 D-B5, K-Kt1; 3 D-B6ch, K-B1 or R1; 4 D-Kt6 mate. This is a very clear illustration of the holding and mating power.

[1] Cheltenham Examiner 19 June 1913. This column is preceded by an editor's note:—
  To understand the following position and solution one must realise that in acting as a Pawn as well as a Knight the “Dragon” can capture a pawn en passant. Having done so it checks as a Kt would.

[2] This is the problem position:

Note that in (ii): by “Put WP on g3” he means move the WPg7 to g3, not add another WP!

XI — The Grasshopper. [1]

  This is another piece I have worked on quite a considerable time, though for the first time publishing any material on it. It is somewhat similar to the Chinese “cannon” which only attacks an adverse man if some other man intervene. The grasshopper moves queenwise, but only to a square immediately beyond one man in the line.
  As a general introduction to its possibilities, let us examine the following position. Each of the inverted pieces is a grasshopper, moving as described. White mates in two.


  The key is 1 K-Kt7. If now G at h3 hops to f3 he gives check but he enables white G to reach e4, which stops the check ! and gives mate.
  Next the G h8 can come to a1 or go to d8, in either case letting the white G go to a1, mate. If G at e8 to e1, then Gd1, mate. If G at h6 to h4 or P-h4, then G to h4, mate. If P-d5, GxP(c6) mate; and if P-c5, G-a8 mate.
  It will be seen that the mates are all in the nature of smothered mates, and this follows from the nature of the insect.

  It may show quite well defined focal character. Thus in this problem:—[2] White: K at g5, R at d1, Kt at e3, P's at c2, f2, f6. Black: K at e4, G at a3, P's at b5, c5, d6, e5, f3. White mates in 2.
  The key is 1 Kt-B5, and now the Kt could mate on d6 or g3 were it not that the G guards each of these squares. The G is thus showing focal action, the two central squares d6, g3 being termed foci. If G-g3, he still retains guard on d6, thus forcing KtxG, mate. Similarly P-B5 takes away the G's power on d6, leaving the same mate.
  There are many other interesting strategic points about the grasshopper which lack of space forbids me to enter upon.[3]

[1] Cheltenham Examiner 3 July 1913.

[2] The problem position is:

[3] A note by W. S. Branch follows:
  We suppose the above are the first printed problems in Grasshopper Chess! Mr Dawson is like the sentry in Gilbert's opera, who thinks of things that would astonish you. His next paper, concerning the “Neutral”, will conclude the series.

XII — The Neutral. [1]

  This is the newest of all new pieces, being used by myself late last year for the first time. The definition is simple but the results are wonderfully complex. Thus:— Either player may treat any neutral as of either colour.
  I give a detailed solution of a simple two-mover, and must leave the numerous fine details of strategy for the time being. The inverted (B) is a neutral bishop, and inverted (P) a neutral pawn.


  White plays 1 Kt-Q7, threatening Kt-B7 mate with the other knight. If black play KtxKt, then white assumes the (P) is black and the (B) white and plays (B)x(P) which is mate for, suppose black say the (B) is black he can only move it on the long diagonal, and white on his next move promptly calls it white and plays (B)xK. If black 1, calls (P) black and plays King to Kt2, white at once calls it white and (P)xK. Also suppose white calls (P) and (B) both white and attempts to mate on the move by (P)-B7, black at once reverses colour and plays the (P) back to B3. This two-er is particularly interesting in that the neutral (B), as already described, gives a direct mate after KtxKt, for a direct mate is a difficult thing to give with a neutral. Ordinarily black would simply reverse the colour and move the neutral back where it came from. Here, of course, he can do that, but it does not save his king.

  That concludes what I have to say in detail on new pieces. I can only hint vaguely at such things as the Ship, the Ghost, the Phantom, etc., which form further arrows in my quiver, leaving them all to some future consideration. What I have written, I trust, will convince readers that Caissa is a very energetic young lady, capable of supplying from her toy shelves the requirements of many more generations of problemists. To Caissa I pay this small tribute.


[1] Cheltenham Examiner 10 July 1913.

[2] The two neutral pieces, although shown as white are not counted as White pieces. Some chess problem magazines now show them by special half white and half black symbols.

[3] The diagram on the left originally had the white king at f1. Henrik Juel points out that there is no solution due to 1...nBg2+ and that the problem is sound with wK on g8. A check in the TRD archive in the BCPS Library shows that Dawson originally had the wK at g8 and that an error must have crept in during the original preparation of these page. Apologies to all and thanks to Henrik Juel.