Check out Grant Acedrex, our featured variant for April, 2024.

This page is written by the game's inventor, David Jagger.


Chogo is my tentative reconstruction of the roman game of Latrunculi (a possible ancestor of chess). Chogo features an unusual 'co-enclosure capture'. Chogo-44, the version entered for this competition, is played on an 8x7 board with 3 squares missing from each corner.

Chogo-44 Rules in Brief:

Each side starts with 8 soldiers/pawns and a general/king. The aim is to capture the enemy king, or immobilise his whole side.

A side's pawns initially occupy the 8 home spaces (at each file's end). A side's first move consists of dropping its king onto an empty cell not beyond the halfway line or rank.

Thereafter, all men normally move orthogonally like chess rooks though kings are limited to a single orthogonal step.

Capture is by what I have called co-enclosure. Both sides can contribute to the enclosure of an individual who, however, can only be captured by an enemy move, when at least two enemy men must be involved in his complete enclosure.

A man may initially be trapped (i.e individually immobilised) in custodian fashion - that is by having two enemy men immediately adjacent to him on opposite flanks.

A straight line (or a solid square or rectangular mass) of men of one side may move orthogonally together in any direction with full rook power (even if a king is part of the moved contingent), even if some of them are trapped - but if they are all trapped the line cannot move at all.

Exceptionally, an untrapped man may be taken up and dropped elsewhere anywhere within otherwise inaccessible territory, but only when all enemy men exist within secure structures and are immune to conventional rook-wise attack in their current positions.

Chogo44: empty board

Chogo-44 Rules in Full:

(Orthodox chess rules apply where any uncertainty remains.)


The 44-square board is as illustrated, made up of a 8x7 rectangle with 3 squares missing from each corner.


There are nine men per side, comprising eight pawns and one king.

Objects and Outcomes:

The winner is the player who captures his opponent's king, or who deprives his opponent of any legal move by immobilising all his men. If no capture has been made after 50 moves by both sides, the game is a draw. Draws can occur by agreement. A player may resign.


Players take alternate turns, black moving first, and they may not pass on their go.


The start position is shown below.

Chogo 44 setup

First Move:

Black moves first. A side's first move must be to drop its king onto any empty cell not beyond the halfway line or rank.

Below is a sample position after both kings (black K, white K) have been dropped on move 1:

Rook Move:

All men normally move as individual units like orthodox chess rooks, any number of spaces orthogonally forwards, backwards, or sideways, without leaping, except that the king may move only a single orthogonal space when moving on his own.


When a player's man comes to be flanked on opposite sides by two enemy men (whether as a result of his own or his opponent's move) that man is trapped (i.e individually immobilised), and cannot be moved as an individual unit.

A man may freely pass between two enemy men. (However, if it stops between them it becomes trapped.)

A man is trapped (and so cannot move as an individual unit) until it is no longer flanked on opposite sides by enemy men.

It is possible for one move to trap more than a single man, and for one move to free more than a single man from entrapment.


Here, if it is white's move, the white pawn P at d2 could be moved to d5 to trap the three black pawns P at c5, e5 and d6.

Tandem Move:

An orthogonally-connected straight line (or solid square or rectangular mass) of men of one side may move orthogonally together with full rook power (even if a king is part of the moved contingent), even if some of them are trapped - but if they are all trapped they cannot move at all. The line as a whole can move either in column or side-on like a phalanx, and a solid mass similarly, in any wholly free orthogonal direction.
Here both sides have several opportunities for tandem moves -
e.g bc3-bc5, fg34-de34, d567-d345, fg5-gh5, g34-h34,de6-fg6 etc.

Here a black pawn P is trapped between a white king K and pawn P:-

Here is a completely immobilised black phalanx of two (consisting of a trapped king K and a trapped pawn P):- VVVV

This central black phalanx of three Ps can still move forwards or backwards with full rook power because not all its members are trapped:-

And the same is true for this black phalanx of a K and 2Ps:-

Co-enclosure Capture:

Only orthogonal contacts matter. So it takes 2 contacts to enclose a corner man, 3 an edge man, 4 a central man.

A man is captured (and must be immediately removed) if it becomes enclosed on all sides as a result of an enemy move, provided that at least 2 of the enclosing contacts are enemy men.

If only one of the contacts are enemies, the man is not captured or removed.

If a man becomes enclosed on all sides as a result of a friendly move, it is not captured or removed, however many enemy contacts it has.

Completing an enclosure of yourself in this way may buy you time for better defence or counter-attack. In this case a capture position would have to be remade by the opponent for him to secure the capture.

However, beware of the slide capture!

Here it is black's go. White is threatening to capture his P on f4 with c3-f3. If black plays g3-f3, (completing the surround of his own man on f4), white could then use the slide capture of e45-e34 to remove the man on f4 immediately anyway. White will have remade the full-enclosure position (around f4) with a single move.

black to move
(white threatens c3-f3xf4)

black g3-f3

white e45-e34xf4


A move of a single man can capture more than one individual.
Here, if the lowest white P is moved upwards to sit between the two black Ps, they would both be captured and removed.

Here, if the white phalanx on fgh5 moves to cde5, the three black pawns at c6, d6, e6 would all be captured and simultaneously removed.


Functional Man:

A functional man is one that is not completely immobilised - i.e one that can move freely or one that, though he cannot move freely as an individual unit (because he is trapped or completely hemmed in), can still move in conjunction with other men.

Secure Structures:

A secure structure exists where the (unchanging) disposition of a group of men of one side renders them immune to any conventional rook-wise enemy attack, however prolonged.

That is, if their side were not to move again, no combination of rook-wise attacks could ever capture any of the men forming that secure structure.

There are two recognised forms of secure structure - the city and the redoubt.


One side's secure structure which seals off an area of the board containing empty territory and in which there are no functional enemy is called a city.

Cities can exist within cities (of the same side). What was once a city is disqualified as a city if it becomes occupied by even a single functional enemy unit.

Examples of Cities:
Here white has three cities (city-spaces Us) and black one (city-spaces Us):

Here white has a single city and black none:

Here neither side has any cities:


Any other secure structure which is not a city is called a redoubt.

A straight line of comrades occupying a whole edge of the board is immune from enemy capture, and is an example of a redoubt.

Redoubts may be attached to edges or cities or may even be within cities.

A king as well as a pawn can contribute to both redoubts and cities.

Examples of Redoubts:-
(A pawn is shown here as an R/R if belonging to a redoubt.)

Here white has one city with one P and the K forming its barrier, and also 2 separate redoubts, one of one man, the other of 4 men. Black has a single redoubt consisting of 8 men, but no cities.

White here has 4 cities and 2 redoubts.
Black has a single city and 3 redoubts.

Mole Move:

Exceptionally, for a player's move, one of his untrapped men (whether pawn or king) may be taken up from the board and dropped elsewhere anywhere within one of his opponent's cities, but only when all enemy men exist within secure structures i.e when all are immune to conventional (rook-wise) attack.

(By definition, there must not already be any functional friendly men in any enemy city - i.e any friendly men that are already in an enemy city must already be completely immobilised.)

There is nothing to prevent a mole drop move being made into a position in which the dropped man is thereby trapped, or into a position in which he is fully enclosed.

Example 1:
Here white has 2 cities, 1 of them an interior city. Black has no cities. All white's men are immune to conventional attack. If it is black's move, any one of black's men could be dropped anywhere within either of white's cities. A drop into c5 would be into a trapped position. A drop into c7 would be into a fully enclosed position.

Example 2:
Here black has one city. White has one city, which contains a trapped black pawn at e6. If it is black's move, any one of black's untrapped men could be dropped anywhere within white's city. A drop into d5 would be into a trapped position. A drop into f7 would result in the capture of the white pawn at f6.

Example 3:
Here black has no cities. White has two cities, one of which is an interior city. White also has an insecure pawn at c3 which is not immune to conventional (rook-wise) attack. Therefore, as not all white's men are secure, black cannot use the drop move to enter either of white's cities.


'Polis' is greek for 'city-state' and is used in Chogo as a courtesy word. A player should say 'polis' before his move, if he has noticed that all his opponent's men are in secure structures and a mole drop move is permitted. Having said 'polis', however, he is not then obliged to use a mole move that turn. There is no penalty for miscalling 'polis', or for forgetting to say it.


A stronger player may agree to start with a pawn or two or three less than the full complement. It is up to the player which pawn or pawns to dispense with, though I would suggest wing pawns initially.

In a series of games it might be agreed that the player winning a game starts with a pawn less for the next game, or the player losing starts with one more (if room).

Here endeth the rules for Chogo-44.

Note on Endgames:

The movement rules for Chogo-44 were designed with endgames in mind. With the rules as stated, assuming reasonable play, we would usually expect the following results:-
KvK      impossible
KPvK     win
KPvKP 	 draw	
K2PvKP 	 draw	
K3PvKP 	 win	
K2PvK2P  draw
K3PvK2P  win		

K3PvKP can present a tricky endgame situation. The key to handling it would seem to be the attacking forces getting into this kind of position:-

The study of Chogo endgames on the generally larger, square or rectangular Latrunculi boards suggests that we would struggle to have K3P beat KP using the rules of movement as for Chogo-44 (which I have tried to keep as dynamic as possible).

A simple rule modification, however, solves the problem for Chogo on Latrunculi boards up to 8x8, namely that a part-trapped line of comrades can only move a single step together. This generates its own interesting endgame flourishes.

The equivalent key diagram for Chogo played on an 8x8 board would be something like:

For Chogo on Latrunculi-type boards greater than 8x8, the increased board size means that a forced win for K3PvKP would usually be unachievable.

We can however require that the first player who has been reduced to King and 2 Pawns resigns once his opponent's king has gained the safely of the central area and is unencumbered by any contacts from his own men - i.e once it is effectively invulnerable to capture.

In fact on these larger boards (having shed any notion of trying to capture every last man), Chogo might be better played allowing the king to move with full rook power, just as the pawns do. It may be that the12x8 Latrunculi found near Colchester was played to rules that were closer to this less constrained form of Chogo. See Sample Game 2 gif of a game which incorporates the game position of the Colcheser find.

Coming full circle, the smaller board Chogos and Chogo-44 itself could also be tried using kings with full rook power, and a single step untrapping tandem move, with automatic resignation when reduced to no-hoper king plus 2 pawns.

Design Background:

The deadline for the chessvariants 44-cell competition for 2004 gave me the necessary spur to get my ideas for a reconstruction of Latrunculi down in presentable form.

Several attempts at reconstructions already exist. Those I have seen (one even bought at Hadrian's Wall) differ markedly from each other and from mine. There is now, however, what seems to be a generally agreed starting arrangement (based on the discovery of a populated, probable Latrunculi board near Colchester) of the 'king' standing somewhere in front of a back line of 'pawns'.

The fact that the game may have been played on boards of various sizes (7x8, 8x8, 10x8, 10x12 etc.) led me to believe that I might be able to create a version small enough to fit 44 cells. I tried several other boards, i.e (6x7)+2, (7x7)-5 and (8x6)-4 (and almost gave up) before arriving at (8x7)-12 which plays surprisingly well.

I had not been fully convinced by any of the Latrunculi reconstructions I had seen. At least it seemed to me that there might still be room for other attempts. Then recently (December 2003) I hit on the idea that the usually accepted capture-type for Latrunculi might be wrong - and that the few fragmentary descriptions (all we have) which had been taken as suggestive of a custodian capture could be reinterpreted as signifying something quite different, perhaps involving custodian immobilisation linked with some kind of full-enclosure capture.

I then experimented with tying-in custodian immobilisation with a Go-like surround capture. This would allow men, if given full rook movement, to still be capturable - and what seemed to be a very playable game ensued. (At this initial stage I had granted all pieces, both pawns and kings, full rook movement.)

But with game testing, various problems emerged. Though singletons and groups of two could be surrounded fairly easily, larger groups needed too many surrounding men to capture them (for the number of starting men available). A player could just create a long line of men (king included) to kill the game. I seemed to be stymied, yet the underlying power of the game still intrigued me. Could I salvage anything from it?

Several classical references seemed to refer to enclosure, blockading, stalling up, penning in, so I still figured I couldn't be far from the truth with the Go-surround idea, and I persisted with similar approaches.

One such approach, just surrounding the ends of branches of connected groups to capture individuals from groups larger than a pair, seemed somewhat inelegant and over-complicated - yet contained the germ of what was to become co-enclosure capture.

An idea from John Conway's cellular automata "The Game of Life" flickered in my brain. In 'Life' too many neighbours can be a bad thing - they can contribute to a cell's downfall. Being penned-in o vermuch by one's own comrades could also prove fatal in battle perhaps.

With the advent of this idea I realized that I should abandon the idea of group capture as in 'Go' altogether, but concentrate purely on the enclosure of an individual, and that by allowing both players' men to contribute to the complete penning-in of an individual man, I didn't need so many captors to deal with enemy masses.

Also falling behind in material was not such a severe disadvantage as you would still have all your opponent's men too as possible collaborators in capture. The game became playable down to fewer men per side. This proved to be a major breakthrough.

Playtesting suggested that requiring at least 2 men of the capturing side to be involved in a co-enclosure worked well, and neatly applied to all types (corner/edge/central) of capture situations.

The playtesting of end games with small numbers per side convinced me that the king's power had to be reduced to that of a single-step rook. Luckily this did not seem to have a particularly adverse effect on the dynamics of the early parts of the game.

There was, though, another major design problem/opportunity to be solved/seized - how to prevent players from retiring behind their impregnable city walls (or at least penalise them for doing so). I considered several alternatives - namely (when all enemy men form secure structures):-

(a) Forbid a player from ever having all his men in secure structures.
(b) Require that a player cannot both start and end his move with all his men secure.
(c) Allow a man to make a knight's move or (d) bishop's move or (e) some other kind of jump move into
an enemy city.
(f) For his go, allow a player to remove or (g) reposition any enemy man.
(h) For his go, allow a player to drop any one of his enemy captives or (i) one previously lost man 
of his own or (j) any one of his men already on the board or (k) an entirely new man of his into an 
enemy city.
(l) Allow a player to replace any enemy man on the board with any one of his own from the board or
(m) with a entirely new man of his own.
(n) Allow a straight orthogonally connected line of men to act as a ram which pushes back a lesser
line of enemy men one space.
(o) Allow a player's king alone special jumping privileges.

Solutions (b), (c), (g) and (j) appealed most, and I finally settled on (j) - i.e to allow a player to drop any one of his (untrapped) men already on the board into an enemy city - the mole move.

One might expect that, with an evolving family of Chogo-like games (if any such family existed), several of these barrier-discouraging options (and probably others too) would have been tried. Some are suggestive of ways that the chess family (and indeed other game families) might have evolved.

It would also be quite possible to have variants which allowed an enemy city to be attacked as soon as it was formed, whether or not all other enemy men were secure. I experimented with some such variants, which seemed to offer a more dynamic though less structurally stable game, but I tended to prefer my original, stricter versions.

History, Legend, Speculation:

Latrunculi (soldiers) may have shared features with several other ancient board games, for instance, the greek Poleis (cities), known too as Petteia (pebbles), the egyptian Tau (robbers), the hebraic Iskundree (Iskander i.e Alexander), the early persian Nard (wooden block or board)/Nardshir (lion-board) /Takhteh Nard(battle-board), and the celtic Fidchell(wood/board craft), for all of which the rules remain speculative. There may also be an historical connection to the nordic Tafl family of games, tibetan Mi/Mig/Ming Mang (many eyes) and ultimately to the chinese Wei-qi(surrounding game), virtually identical with japanese Go.

It would not be surprising if there were different local variants of Latrunculi or indeed of others of these ancient games (as there are for draughts and chess of course). If I am right about the presence of the co-enclosure form of capture, it may have been common to several games and their variants in classical times.

Some people have argued that Latrunculi is identical with or closely related to the greek game of Poleis (cities). I tend to concur. In my view literary references to Poleis (or Petteia) are consistent with the presence of a royal piece (or captain).

Some consider that there may have been an earlier version of either or both games which did not include a royal piece. If this was the case, such a game (if with the same linear starting arrangement of pawns as Latrunculi) could not have permitted a sideways phalanx move, for a mass advance on the very first move to close down the opposition would have resulted in an instant win through mass immobilisation. Perhaps, though, a game without kings could have existed if all pieces were initially dropped successively anywhere on the board. There may, too, have been more pieces to start with. Once all were present, orthogonal movement could ensue. (Indeed a king could have been present and been dropped last of all.)

Petteia is said to have been invented by the sage Palamedes at the time of the Trojan war. Though he is sometimes considered to be a legendary figure, the idea is not beyond the bounds of possibility. Troy itself (apparently) was unearthed by Heinrich Schliemann (1822-1890) who believed that there was a factual basis to Homer's epics. A seige of 9 or 10 years of intermittent warfare interspersed with much boredom would have been a good time for the invention, transmission and later dissemination of a board game throughout the ancient world, especially one involving soldiers and cities. Soldiers along Hadrian's Wall are known to have played Latrunculi, which might be a derivative. Several boards have been found along the wall and in other roman outposts. One could even wildly speculate that the legend of the wooden horse of Troy, the ruse whereby the Greeks were able to finally breach the sacred walls of the city, might be a tale suggested by an exceptional horse-like leap move used to enter an otherwise impregnable city in the game of Poleis/Petteia.

One must, however, be on one's guard regarding the ascription of ancient games to legendary or divine figures. Thoth (egyptian god of science), Lugh (celtic god of light), The First Emperor of China, and Lao Tzu (chinese sage, 6th century BC) are among others credited with the invention of popular boardgames (Senet, Fidchell, Wei-qi and T'shu-p'u respectively). It is quite conceivable, though, that (along with Poleis and Latrunculi) several of the mysterious lost games which are associated with sizeable, gridded, often-rectangular boards, may share a common ancestry, perhaps a prehistoric one.

A quick look at the number of often independent design decisions affecting fundamental aspects of the game to be made when reconstructing Latrunculi, tells us that there are potentially thousands of different rule combinations that can be tried, and that there may be scores or even hundreds of reasonable solutions lurking in that game-space. Any game that proved to be so popular for centuries in classical times, however, has surely to satisfy certain stringent criteria which must be our guide in whittling down the number of realistic candidates. These would include: simplicity of rules, clarity, depth, variety and interest of game situations, continued opportunities for the player behind in material, convenient duration, and no doubt several other things too, besides, of course, (and very importantly) having to be compatible with the classical references.

Hopefully some of you might be tempted to test Chogo against these all too demanding criteria, and maybe some others of you, too, to design better Latrunculis of your own!

The Name:

The name Chogo is meant to be suggestive of both Chess and Go (with perhaps just a hint of Shogi). If there is any common heritage between them, it is just conceivable that a game something like Chogo, perhaps also involving co-enclosure capture, is the missing link, ancestral to both master games and maybe already in existence sometime around 1200BC, the period of the Trojan War.

The word itself, I find, is used in tibetan (balti) with the meaning big/high/great, as in such place names as Chogo Ri for Great Peak (actually the mountain K2, Karakoram2) and Chogo Lungma for Great Valley. Rather than commit to whether any such ancestral game might have emerged from the eastern or western sides of the great tibetan barrier let me diplomatically (with the use of the Chogo name) sit on that mighty fence!

And if it was actually tibetan, well that's just chogo.


Chogo can be played on many other boards just as Latrunculi perhaps was, a smaller board tending to a shorter game. The most popular boards for Latrunculi seem to have been 7x8, 8x8, 8x10, 8x12 and 10x12 and all these, and others, can be tried for Chogo. On rectangular boards I guess that the men would usually have been laid out along the longest edge for more concentrated action, and Chogo can follow suit.

For Chogo on fuller or bigger boards than for Chogo-44 the rule for tandem movement of a part-trapped contingent (as mentioned earlier) is different. A part-trapped column can move together one space orthogonally only.

On the larger boards, Chogo can (and perhaps should) be tried allowing the king to move with unrestricted rook power.

Other (untested) variants might allow the trapping of not just one but more than one man, by flanking either end of connected straight lines of enemy men.

Chogo could be tried too with fixed starting positions for the kings - e.g on the rank in front of their pawns facing each other on a central or near-central file, or offset from each other near-centrally.

Also, a different rule could be adopted when a side's men are all in secure structures, rather than rule (j) which gave us the Mole Move used within Chogo-44. Rule (b) - "a player cannot both start and end his move with all his men in secure structures" - and rule (c) - "allow a player to use a knight's move against an opponent with all his men in secure structures" - and rule (g) - "for his go allow a player to reposition any enemy man when all are within secure structures" - all seem good candidates for further investigation.

It is quite possible that forms of Chogo could be developed on hexagonal (and even other) grids, incorporating other forms of 'phasic' immobilisation/capture.

Finally you might like to experiment with beginning the game with an empty board, and having an initial phase where men are successively dropped onto the board - captures might or might not be allowed in this phase - with the kings being dropped last (or perhaps you could play without kings). Only once the drop phase has ended would normal rook-wise movements begin. You could vary the number of men per side, perhaps increasing the numbers until you find the most lively game. Such a game naturally begins to resemble Wei-qi/Go.

The 8x12 Colchester Latrunculi:

12 by 8 board Chogo

The game position shown on the board recovered from a tomb near Colchester in 1996 (see diagram below) is compatible with a Chogo variant where black plays first and where first moves are drops of each side's king anywhere in their own half. Alternatively the position could reflect kings having full rook power and each being situated in front of their pawns near-centrally, either on the same file or on neighbouring files (and not necessarily being moved first).

I have created a gif of a game of Chogo incorporating this position (Ed.: here as avi-file), using full rook-power kings, and allowing part-trapped columns only a single space move. An interesting (though no doubt amateurish) game developed. See Sample Game 2.

I've also included a 12x8 board for those who would like to try out the larger game.


(in no particular order)

Opening gambit in early version of draughts took 2,000 years to emerge
Roman board game found at burial site by Nigel Hawkes, Science Editor
Greek Board Games by Roland G. Austin (University of Liverpool, England)
Antiquity, 14, September, 1940, Pages 257-271
Game: Latrunculi Implemented by L. Lynn Smith, July 2001.
This version attempts to utilize the rules as described by Dr. Ulrich Schaedler.*/Latrunculi.html
Latrunculi Article by James Yates, M.A., F.R.S., 
on pp670-671 of William Smith, D.C.L., LL.D.: A Dictionary of Greek and Roman Antiquities, John Murray, London, 1875. 
Game: Military Latrunculi Reconstructed and implemented by Roger Cooper, September 2001.
Roman Board Games by Wally J. Kowalski
Some remarks on the Roman Board Games Internet Site.
Ulrich Schädler.
Ben Levick and Mark Beadle January 1992
A proposed Chess birth scheme
A proposed Chess birth scheme (part 2)
Hnefatafl - the Strategic Board Game of the Vikings
An overview of rules and variations of the game by Sten Helmfrid
Version 2. May 1, 2000.
by Eoin Mac White
John Conway's Game of Life


Board and Table Games from Many Civilizations - R.C.Bell - Dover Publications, Inc. 1979.
The Oxford History of Board Games - David Parlett - Oxford University Press 1999.
New Rules for Classic Games - R.Wayne Schmittberger - John Wiley and Sons, Inc. 1992.
A History of Board-Games other than Chess - H.J.R.Murray - Oxford University Press 1951.

Sample Gameplay:

A king's position is underlined.
*= mate threat.
> mole drop move.
# traps king.

Sample Game 1 Chogo-44:-
(see gif1)
	black		white
1 	b3, 		f5;
2	c1-c6,		b6-b4;
3	d1-d4,		e7-e3 *;
4	a3-a4,		e3-g3;
5	e1-e5,		a5-c5;
6	h3-h4,		g6-g4;
7	e5-e7,		f7-f6;
8	f1-f3,		g3-h3xh4;
9	d4-d6xd7,	b4-b6xc6;
10	b23-b45,	c5-c2;
11	d6-c6xb6,	c7-d7;
12	b45-d45,	f56-f67;
13	c6-e6,		d7-d6;
14	e67-e23,	c2-c5 *;
15	d4-e4,		h5-e5;
16	e23-b23,	f7-c7;
17	b23-b56,	c7-c6;
18	a4-c4xc5,	polis c6>f2; 
19	g2-g3,		f2-g2xg3;
20	e4-e2,		f6-e6;			
21	c4-e4,		de6-cd6;
22	e2-f2,		h3-g3;
23	f23-f56,	g23-b23;
24	b56-b45,	g4-f4;
25	f6-e6xe5,	b2-d2;
26	f5-f7 *,		c6-c5;
27	f7-d7 #,	d2-d4 #;
28	b45-b56 *,	c5-c6 forced;
29	b5-c5 *,	b3-b5xb6 consolation capture;
30	cd5-de5xd6 slide capture mate.		

Sample Game in animated form: Avi-file (178 kB)

Sample Game on 12 by 8 board in animated form: Avi file (84 kB).

Written by David Jagger.
WWW page created: May 11, 2004. Last modified: May 12, 2004.