3 On castling and attacking the castled position in general
The origin of castling
Castling, the most distinctive move in the whole of chess, affects the nature of mating attacks in various ways and because of its importance in this connection, the general significance of the move is worth examining.
Castling is not of any great antiquity. It is also obvious that such a complex and distinctive operation as the modern castling move was not born in a day, but rather developed gradually from century to century with the evolution of the rules of the game themselves.
In the Indian game of chaturanga (pre-chess) there is no trace of the move, nor can it be found in the Arab shatranj. The first traces of any unusual or exceptional movement being made by the king, out of which the castling move evolved, are to be found in mediaeval European chess, which in its first stages can be described as the first and not very significant revision of shatranj towards providing the pieces with greater mobility. A historical source attesting to this first revision is a work by the Lombard monk Jacobus de Cessolis, in which are recorded the rules governing the movements of the individual chess pieces at the time. There it is stated that king, queen, and pawns have the right to make an initial move of two squares as well as the normal one. From this early reform the double opening step taken by the pawn survives to this day, while in the case of the queen it became obsolete when the piece was given its present powers of movement during the great reform of the game at the end of the fifteenth century. In the case of the king it was further developed and altered until it eventually took its present-day shape as castling.
There is extensive confirmation of the introduction of the king’s move from other sources after Cessolis, though the particular conditions governing the move vary. The form most frequently recorded is that where the king moves like a knight (‘Freudensprung’), with the limitation that it is not allowed to go beyond the second rank. The next form, which made its appearance in Italy in the sixteenth century, already included a movement by the rook in the same move – a step nearer to the castling of today. The king’s use of the knight’s jump disappears at this point and the rules state that the king is free to move from e1 to g1 or h1 and the rook from h1 to f1 or e1. This type of ‘free castling’ or ‘Italian castling’ persisted in Italy right up to the nineteenth century, when it was superseded by the present rules, which were introduced in France during the eighteenth century.
This short survey of the historical evolution of castling is a good illustration of the gradual development of the rules of chess and it makes it particularly clear that these rules have been changed and perfected with the desire of enriching the game; the general tendency towards giving the pieces more dynamic movement reflects the increased animation of social movements in Europe from the Middle Ages onwards.
In addition to this general line of development in the laws of chess, we also owe the introduction and eventual adoption of the castling move to the indirect influence of the great reform of the fifteenth century. This reform created a powerful queen out of the fers (which moved one square) and the modern bishop out of the alfil (which moved exactly two squares). In this new type of chess, with three long-range pieces, the fighting ability of the king declined; once in the days of pre-chess it had been the strongest piece of all. The king now became a hindrance with its slow movement, and its position in the centre of the back rank stood out immediately as a fault in the game’s structure. On the one hand, it had to be safeguarded against the formidable new forces, and on the other it had to be removed from the centre so as not to hamper them. For this reason it was necessary for the king to get away from the centre as quickly as possible, and it was to this end that the king’s double move was built on and transformed into the full castling move. Therefore, not only the development of this most distinctive move can be traced through the history of chess but also its logic: it exists as the necessary complement to the reform of the other pieces’ movements. We start a game of chess today with the pieces placed in the ancient order which derives from chaturanga, and then we switch over by castling to a new position, which is better suited to the alterations made in the game’s rules.
The dynamic and positional significance of castling
Chess moves in general can be looked at from the dynamic or the positional point of view, depending on whether we are considering them as part of an active combination or as static features.
The castling move without doubt powerfully increases the dynamic potential of a position. A player who has castled at once obtains a whole range of possibilities which he did not have before. He has increased the influence of the rook, cleared the square e1 for a possible attack along the e-file, and on top of this he has prepared the ground for the quick establishment of ‘communication between the rooks’, i.e. the entire clearing of the back rank, as a result of which the rooks cover each other or can combine together in attacks.
On the other hand, castling is also a positional move, and one which provides a particular element of permanency to the position. For it is unique and irrevocable; it can only be carried out once in a game, which means that the king will as a rule stay on the side on which it has castled. In special cases it can, it is true, travel with a greater or lesser degree of difficulty from its castled position to various parts of the board. Thus castling is only relatively irrevocable (unlike a pawn’s move, which is absolutely so) but for practical purposes the king’s position after castling is, at least until the arrival of the endgame, fairly definitely a permanent positional feature. Before castling, the king has three possible courses of action: to remain in the middle or to castle either to right or left. Once it has castled it no longer has any choice; it then has a definite shelter, a ‘permanent address’, a fact which the opponent naturally takes into account. It is certainly safer against attack, but it has at the same time committed itself positionally to one particular area of the board; thus the undoubted advantages of castling are offset by certain disadvantages.
Artificial castling
Chess also contains the term ‘artificial castling’, which involves a player creating a position the same as or similar to that reached after genuine castling. It is attained not by a single castling move but a series of moves by the king and the rook. In the continuation from the following diagram we shall encounter the simplest example of ‘artificial castling.’
Black, to move, played:
1 ... Nxe4
White’s best would now be 2 Nxe4 d5 3 Bd3 dxe4 4 Bxe4, in which case the prospects would remain even. However, before taking the knight on c4 White decides to use his bishop to prevent his opponent from castling – an idea which, in this position, is faulty.
2 Bxf7+ Kxf7 3 Nxe4
At first sight White’s plan appears to have succeeded, since, as well as maintaining the material balance, he has prevented Black from castling. However, the further play shows that Black can carry out ‘artificial castling’ without difficulty and has made a clear gain in that he is left with strong pawns in the centre.
3 ... d5 4 Ng3 Rf8 5 d3 Kg8
Now the superiority of Black’s position is quite obvious, and one can easily appreciate the part played by ‘artificial castling’; Black has admittedly expended three moves on it, but White has derived no advantage from this fact, since he has made three moves of even less value, i.e. Bxf7+, Nxe4, and Ng3. In a sense, Black was castling while White was transferring his knight from c3 to g3.
The correct moment for castling
Beginners are often advised to castle as quickly as possible. This is a useful and sensible piece of advice in the majority of instances. Certainly, less experienced players offend against this general precept extremely often, postponing castling unnecessarily and as a result suddenly finding themselves in an awkward situation, by which time it is far too late to castle. The present writer once made a statistical survey of late castling using a number of simultaneous games and came to the conclusion that a common characteri
stic of so-called ‘simultaneous massacres’ (when the master wins all, or nearly all, of the games) was precisely the lateness of castling. Around the twelfth move of one such simultaneous display the situation was as follows: the master had castled on eighteen of the boards and not yet on two, while his opponents had castled on three boards but not on the rest. The result: 20-0 to the master – a real massacre! This is a typical example which indicates the disadvantages resulting from a delay in castling. In other simultaneous displays, where masters had a harder task and poorer results, the statistics on castling, taken around the twelfth to fifteenth moves, on the whole showed smaller discrepancies.
The general rule that one should castle as soon as possible is therefore quite in order, though it must at once be emphasized that there are also many exceptions to it. Every chess player of greater experience is well acquainted with those particular situations when it is correct to postpone castling or when it is altogether unnecessary. To make the matter clearer, I shall set out some situations of this kind and support them with examples:
1) Castling is postponed or not carried out at all, because some other action is more useful. This might be quite simply the capture of an opponent’s piece, the spoiling of his position, or, indeed, an attack. If such an attack is sufficiently strong and profitable, the attacker often never castles, because he is victorious first. This kind of situation needs no particular examples.
2) Castling is postponed, because for the time being it is still dangerous. It is better to prepare it by removing the danger first, e.g. by exchanging the opponent’s threatening pieces or by some other manoeuvre.
As an example of the dangers of castling too early, let us take the following variation of the Giuoco Piano.
1 e4 e5 2 Nf3 Nc6 3 Bc4 Bc5 4 d3 Nf6 5 Bg5 h6 6 Bh4 d6 7 0-0?
At this point castling is premature and incorrect. There is a danger of Black’s replying ... g5 and following up with a pawn attack against White’s castled position; this danger ought to be averted first. 7 Nc3 is a step in this direction; if Black continues with 7 ... 0-0? 8 Nd5 g5, then White can reply 9 Nxg5! Nxd5 10 Qh5 hxg5 11 Bxg5 with an overwhelming attack. Moreover, if Black answers 7 Nc3 with 7 ... g5 8 Bg3 and then ... Bg4 or ... h5!?, the situation is less dangerous because White has not yet castled and can therefore prepare to castle on the queenside, which is not threatened.
7 ... g5 8 Bg3 h5! 9 Nxg5 h4!
Black does not worry about the consequences of Nxf7 but continues his attack consistently. In fact, he gives up his queen for the bishop on g3; this would not be a good exchange if White had not castled, but against the castled position the attack is decisive.
10 Nxf7 hxg3 11 Nxd8
If 11 hxg3, then 11 ... Qe7 12 Nxh8 Qh7 followed by ... Ng4, with a decisive attack.
11 ... Bg4 12 Nf7 Rxh2! 13 Qd2 Nd4 14 Nc3 Nf3+ 15 gxf3 Bxf3 16 Qh6 Rg2+ 17 Kh1 Rxf2+ 18 Kg1 Rg2+ 19 Kh1 Rg1#
3) The player does not castle at once because, although he is able to castle on one side, he actually wishes to castle on the other side and needs to make further preparations. Alternatively, he delays until his opponent castles, and only then decides which side to castle. If he is aiming for an attack, he may decide to castle on the opposite side to his opponent; if he is aiming to forestall an attack, then he will more likely castle on the same side as his opponent.
There are many examples in the games of the masters of this kind of pause before castling. It is assumed, of course, that this pause does not entail any dangers. In open positions such dangers are more frequent than in closed ones.
The following situation will serve as an example of this pause before castling.
It is White’s move and he may, if he wishes, castle short immediately. But he calculates as follows: after 1 0-0 Black will play ... 0-0, and then how can I attack him? The pieces by themselves can achieve nothing in this blocked position; h3 and g4 still need to be played and must be prepared by Qd2. But in the meantime Black will consolidate his position on the kingside (e.g. by means of ... g6 and ... Rf8-f7-g7) after which the attack will have lost its power. There is no doubt that for an attack based on g4, White’s king is not well placed at g1, and it is better for that purpose to castle on the queenside and then deploy the rooks on the g- and h-files. However, if White chooses 1 Qd2 and then 2 0-0-0, Black can play 2 ... 0-0-0, whereupon the attack with g4 no longer applies; White will have to play a hasty b4, in which case it will be quickly seen that his king is badly placed on c1. Consequently, he must put off castling and wait. White therefore plays 1 Qe2, and if 1 ... 0-0, then 2 0-0-0, while if 1 ... 0-0-0, then 2 0-0. He castles on the opposite side from Black in order to get the maximum freedom for carrying out his attack.
4) Castling may be postponed, and perhaps forgone forever, in certain positions where the centre is permanently blocked.
The reason for being able to postpone castling in such positions is, of course, that the pawns form an obstruction in the centre and so make it impossible to open the central files and diagonals without painful sacrifices. With the centre blocked, the kings are also safer in the middle, at least for some time. The logical course of attack in such cases tends to be on the unblocked flanks, and there are times when the king is more exposed by being in a castled position on the wing than in the middle, where it is protected by the blocked pawns.
A convenient example to illustrate this can be obtained by re-examining the above diagram, which we ought to do anyway because we left White’s move 1 Qe2 unanswered. Black’s best reply also consists of a postponement of castling, e.g. 1 ... a6 or 1 ... Nd8. The postponement of castling is feasible for both sides in this position, and for a very obvious reason: the pronounced blockade in the centre. Attacks can only take place down the flanks and as a result the kings are at the moment relatively safe in the middle. Play from this position might develop as follows:
1 Qe2 Nd8 2 h3 Nf7 3 Qd2 0-0-0
It is less dangerous for Black to castle on the queenside now; if White plays 4 0-0, Black replies 4 ... Rdg8 threatening ... g5, while after 4 0-0-0 the danger of an attack on the flank disappears and Black can relax.
In this case the preliminary skirmish has in fact ended with castling, but there are blocked positions where castling does not occur at all and the kings stay permanently in the middle. Even in the above position White has the alternative of not castling and instead co-ordinating his rooks by Kf2; later on he can even take his king up to e3.
5) Castling is unnecessary, because the endgame is already not far off, in which case the king is well placed in the middle of the board.
These cases too are not infrequent in master chess; we shall select two examples from opening theory, where the white king forfeits the right to castle as a result of the exchange of queens.
1 d4 d5 2 c4 e6 3 Nc3 Nf6 4 Bg5 Nbd7 5 e3 Be7 6 Nf3 0-0 7 Rc1 c6 8 Bd3 dxc4 9 Bxc4 Nd5
Capablanca’s freeing manoeuvre in the Orthodox Defence to the Queen’s Gambit.
10 Bxe7 Qxe7 11 Ne4 Qb4+ 12 Qd2 Qxd2+ 13 Kxd2
This variation was played in a number of games from the 1927 World Championship Match between Capablanca and Alekhine. White does not worry about castling, as the position is approaching an endgame and the king is fairly safe in the middle of the board.
A second example is supplied by my analysis of a variation of the Slav Defence.
1 d4 d5 2 c4 c6 3 Nc3 dxc4 4 e4 e5 5 dxe5!
In my Modem Opening Theory I suggested that this was better than 5 Bxc4.
5 ... Qxd1+ 6 Kxd1!
White could have retained his right to castle by playing 6 Nxd1; however, it is better to use the king since castling is no longer necessary in this position, and it is more important that the knight should remain on c3. For example, after 6 ... b5 7 a4 Black is not obliged to continue with 7 ... b4, whereas after 6 Kxd1 b5 7 a4 he has nothing better than 7 ... b4, whereupon White obtains the advantage by 8 Nb1 Ba6 9 Nd2.
In this instance White, in playing Kxd1, does not gain the ‘centralizing tempo’ which he did on the thi
rteenth move in the previous example, but he prefers to lose his chance to castle just to keep his knight on c3.
Many occasions for renouncing castling in this way arise in practice, but one should avoid taking the king too far afield. This comment applies to so-called ‘queenless middlegames’ and not to true endgames, in which the king is nearly always better placed in the centre. One should keep one’s eyes open and weigh up the opponent’s opportunities for an attack on one’s king, for they can easily develop in ‘queenless middlegames’. Safety from attack and a genuine prospect of making further piece exchanges – these are the two important criteria affecting a decision on the forfeiture of the right to castle.
The cases I have mentioned in which castling can be justifiably postponed or dispensed with are meant to induce the reader to think about the problem more deeply, for every rule in chess should be applied flexibly. To make a correct decision on exactly when to castle one needs considerable knowledge and experience.
Finally, one more hint: even in cases where there is justification for a delay in castling, it is still a good thing to have the opportunity to castle quickly on one side at least. When the pieces have been moved out on one side of the back rank, we have an opportunity to castle ‘at hand’, and this castling will cost just one tempo. However, if there is one piece which has still not left the back rank then obviously castling will cost us two tempi. Now there is a big difference between one tempo and two, if one’s opponent can discover some concealed means for a sudden attack on the king before it has reached its castled residence! For this reason it is good policy to get ready to castle as soon as possible, and then consider whether it can still be put off for a while.
Art of Attack in Chess Page 5