Julia's Fairies

Fairy Chess Classification Project

Fairy Chess Classification Project 

The Fairy Chess Classification (v.0.2, 30.07.2021)

Concept map


In these descriptions of fairy elements, the normal meanings of common orthodox chess terms or concepts such as ‘square’, ‘capture’, ‘promotion’ or ‘castling’ will be assumed to be understood, though these normal meanings will often need modification when applied to fairy elements. However, the term ‘move’ will be used to mean a non-capturing move as opposed to a capture (though ‘move-length’ would be taken as applying to a move or capture), ‘unit’ will be used to mean any piece or pawn, and ‘piece’ to mean any unit apart from a pawn or king. Also, any reference to the royal powers of a king will be taken to apply equally to any other royal piece used instead of a king.

In this classification, the fairy elements divide into the four ‘Main Groups’ of Stipulations, Pieces, Conditions and Boards, these Main groups being divided into Groups and Sub-Groups.

Note: Some elements can be included in more than one group. We have chosen the commonly followed classification. Such cases are indicated when they occur. Generally, the rules applying to one element in a group can be applied to another group to create new elements or sub-groups in that group. For example the rules of the Vertical Cylinder can be applied to particular pieces as an attribute to create “Vertical Cylinder pieces”. The rules of Circe can be applied to specific pieces to create “Circe pieces”. The attributes of certain pieces can be applied to all pieces to create a condition as in “Chameleon Chess”. Certain stipulations can also be interpreted as conditions and vice versa. See also under 1.1: Goals.

All previous posts together with the comments can be found in the Fairy Classification category

Related Articles:


  • Version 1 was the ‘Glossary CT1’ document dated 7 December 2020
  • Version 2 the ‘Glossary CT4.1’ document dated 27 March 2021
  • Version 3. Chris Tylor, 30 March 2021; Outline version by NSR
  • Further revisions by CT/NSR, 8-24 April 2021
  • Further revisions by CT/NSR, 25 April to 21 May 2021
  • Further revisions by CT/NSR, 21-May to 30-July 2021


  1. BCPS Glossary
  2. Die Schwalbe Lexicon
  3. PDB Keyword Search
  4. Bibliography of Chess Problem Lexicons
  5. Jacques Dupin’s Site
  6. Juraj Lörinc’s Site
  7. Strategems
  8. Kobulchess
  9. Phenix
  10. Variantim
  11. Angela & Otto Janko’s Site:
  12. Christian Poisson’s Site:
  13. MatPlus Forum Posts with Fairy Chess Definition and Classification Discussions
Notify of

Newest Most Voted
Inline Feedbacks
View all comments
Marjan Kovačević
Marjan Kovačević
April 22, 2021 23:09

I could only try to imagine the amount of voluntary work to collect, define, redefine, formulate, discriminate, present … and bring us some light … Among all the chess websites, forums and blogs, I couldn’t recall any other case where a small group of enthusiasts from different countries joined their good intentions for the benefit of community. Amazing, hat down to the Team!

April 22, 2021 23:54

Wow! Great job. A cursory look tells me what we can use and probably need:– Locust lion and a Nightrider/Rose locust! A locust with optional capture suggests itself. Now what shall we call it??— Patent claimed for all if not in use already 😀😀

Last edited 3 years ago by Seetharaman
shankar ram
shankar ram
April 23, 2021 11:27
Reply to  Seetharaman

All these: Locust Lion, Nightrider Locust and Rose Locust already exist! There are published problems using them. We haven’t included them here yet!
A Locust with “optional capture”? Not clear.

April 23, 2021 22:13
Reply to  shankar ram

Actually it is a grasshopper with option to capture the hurdle only. Capture beyond the hurdle prohibited. In other words a piece which moves like G but can capture only like Locust.

Last edited 3 years ago by Seetharaman
April 30, 2021 10:33

In what piece category will the Querquisite be? And in 4, should there be a 4.4 modified boards (e.g. (worm)holes and grids) or will that be part of the other categories?

shankar ram
shankar ram
April 30, 2021 11:56
Reply to  Joost

Joost, Thank you for joining the conversation!
Querquisite: Currently this is not yet in our list. We will add it in the future updates. This piece would be in the category 2.6 Other Pieces

Worm holes are currently categorised as a condition under 3.7.6: Effects of all moves, though it is mentioned in the definition that it could also be considered as a board attribute.

Grids are currently categorised as a condition under 3.7.5: Restrictions on all moves, though it is mentioned in the definition that it could also be considered as a board attribute.

The categories are not set in stone and may change in future. Also, some elements like the above two could be considered as belonging to more than one category. We will indicate this in the definition and provide a link(yet to be implemented!) to the alternate category.

Last edited 2 years ago by shankar ram
Andrew Buchanan
Andrew Buchanan
May 2, 2021 05:56

Great work all! Stupendous!
I’ve mostly not yet participated in this, but I’ve spent some time cleaning up PDB fairy piece data (partly in order for me to learn more about them!)
(1) How can we ensure that detailed fairy piece datasets are aligned and stay aligned? Are there any real reasons for differentiation between databases? (E.g. language support might be one?)
(2) There is some terrific scholarship by Christian Poisson at systematizing classes of fairy piece: to take a random example: http://christian.poisson.free.fr/problemesis/categories.html#Bondisseurcoureur. How can you ensure that this structure, which underpins WinChloe engine, forms a part of the FCCP?
(3) Can you embed links to these other fairy piece sites, which often include great examples? E.g. the diagram on the Leaper-Rider page I mentioned. Picture/word = 1000 as they say.
On which note I will shut up 🙂 Thanks again!

Last edited 2 years ago by Andrew Buchanan
shankar ram
shankar ram
May 2, 2021 07:12

Thank you, Andrew!
We welcome your involvement!
To answer your questions:

(1): The present classification was decided by Chris and myself, with some help from others. It is based on the original BCPS glossary by Stephen Emmerson, which we reworked and modified. Obviously some elements may look out of place in their current slots. We will continue reviewing these and make changes as appropriate. Some elements, due to their nature may fit into into more than one category. These, we are trying to indicate by saying “…can also be considered as…” and similar phrases. We are working with Julia to incorporate hyperlinks within the definitions to such alternate categories.

(2): Yes! We are are aware of Christian’s work and have liberally used it in our project. However, in a few cases, there are some inconsistencies between WinChloe and Popeye or a particular element was missing. In these cases we have used our judgement.

(3) We have also used The Die Schwalbe Märchenschachlexikon, PDB keyword search and various other sources on the web. You can see the whole history of the project, from 4-Oct-2020 when it all started, on these pages: https://juliasfairies.com/category/fairy-classification/
I will discuss with Julia about making a place here to list all the sources we have used. We have quite a comprehensive list!

Last edited 2 years ago by shankar ram
Joost de Heer
Joost de Heer
May 27, 2021 08:53

In 1.2.1, I’d replace ‘directmate’ with ‘direct goal’ (White to play and reach the goal in the given number (n) moves or fewer, whatever Black plays.). Idem with all items in this category.
exact- is only mentioned in the chapter description of 1 but should have its own category somewhere in 1.2.

shankar ram
shankar ram
May 28, 2021 14:45
Reply to  Joost de Heer

Thank you, Joost!
You are correct. In fact, all “Plays” could be categorised as either “Direct”, “Help” or “Series”. So you would have “Direct-Goal”, “Help-Goal” and “Series-Goal”.
The existing description was made as an initial and easier way for understanding. See the comment under 1.2: “Note that most of the elements listed under Play are given in the form of a complete stipulation ending in the goal Mate. All of these may be varied to make the goal Stalemate (or one of the other listed goals).
We are discussing alternate ways and the classification may be changed as we progress further.
Some other groupings are “forward” and “backwards” play. “alternating” and “series” play. Within each of these, you would again have the other groupings as sub-groups.
The “exact” qualifier in a stipulation is applicable across almost all categories, with the meaning that even if a shorter length solution is possible, the requirement is to achieve the goal in the stated number of moves. As such, it may not be useful to make a separate category for it.

Last edited 2 years ago by shankar ram
Dmitri Turevski
Dmitri Turevski
May 28, 2021 20:09
Reply to  shankar ram

How about this take on classification:

– there are only two types of play: opposing and helping (in reaching the goal)
– series is not a type of play, but a change to the rules of the game (a side may, and sometimes must, make a “pass” move) – underlined by that the solutions to a series problem can’t happen in a normal game, but can happen in a game with this fairy rule in effect.
– the “exact” qualifier has nothing to do with the type of play, but simply changes the rule by which we determine if the goal is reached.
– backward/forward also has nothing to do with the type of play, but changes the rules how the moves are made

Can this approach be discussed within your group?

Because it makes so much more sense to me compared to:
> all “Plays” could be categorised as either “Direct”, “Help” or “Series”

… and then we have series-direct and series-help. Wait a minute! :)))

shankar ram
shankar ram
May 28, 2021 20:58

The interpretations:
a) there being only two types of play (opposition and help)
b) series being considered as a condition
have been added in the latest release as alternate points of view.

Backward/forward play being different could also be considered as an alternate point of view. This view exists in the current placement of direct mate problems and defensive retractors in different sections in published sources.

Series-direct, Series-help, Series-self stipulations are not really so different. They only differ in the way they are motivated by the final outcome: #, h# or s#. Otherwise, the play in all these has many common features. This point of view is seen, for example, in the separate section for series movers in at least two magazines: Strategems and feenschach.

Last edited 2 years ago by shankar ram
Dmitri Turevski
Dmitri Turevski
May 28, 2021 21:54
Reply to  shankar ram

> Series-direct, Series-help, Series-self stipulations are not really so different. They only differ in the way they are motivated by the final outcome
Yeah, I understand that if we breakdown the play into non-overlapping direct/help/series then the series-help becomes the series-semi-reflex-half-duplex (or “by black”, if you like).
It’s totally valid, but it doesn’t feel economic to me:

  • we need to introduce the concept of semi-reflex beforehand (why?)
  • we need to introduce the concept of “by black”/halfduplex beforehand (and this seems most unnatural to me, but may be it’s just me)
  • we need to maintain not the “alternate point of view” on a single term/concept, but an alternate point of view on a family of stipulations (think about classifying the parry-series-help, it’s possible fit it into the “series” type of play, if we are strict on direct/help/series division, but the definition would be very cumbersome, we can’t just get away with “reflex”).

Thank you for your time answering my questions!

shankar ram
shankar ram
May 28, 2021 23:12

>>> “…if we breakdown the play into non-overlapping direct/help/series then the series-help becomes the series-semi-reflex-half-duplex (or “by black”, if you like). It’s totally valid, but it doesn’t feel economic to me“.
This is a needless complication. And not really required to prove that different types of series movers are similar.

Last edited 2 years ago by shankar ram
Dmitri Turevski
Dmitri Turevski
May 29, 2021 07:21
Reply to  shankar ram

You don’t have to prove that they are similar. You have to prove that they are neither direct nor help.

Dmitri Turevski
Dmitri Turevski
May 28, 2021 22:39
Reply to  shankar ram

> Backward/forward play being different could also be considered as an alternate point of view. This view exists in the current placement of direct mate problems and defensive retractors in different sections in published sources.

This is a logical mistake, I believe. Being placed in a different sections does not imply having different type of play. Twomovers and threemovers are also placed in the different sections in published sources, but they are naturally have the same type of play.

shankar ram
shankar ram
May 28, 2021 23:01

The difference between direct mate two movers and three movers is only one of move length, whereas the difference between direct mates and defensive retractors is one of the move direction. Not really comparable.

Dmitri Turevski
Dmitri Turevski
May 29, 2021 07:22
Reply to  shankar ram

Ergo the reasoning that distinguishes the types of play by sections the problems are placed into in published sources is invalid. 

andrew buchanan
andrew buchanan
May 31, 2021 14:26

Seriesmovers are just one of many manifestations of an operation I’m terming “flip” where we take a position and say it’s the other person to move (others being threats, set play, zugzwang, pass, Codex Art 15, retrostalemate, false initial Position, ortho-reconstruction, tempo loss, Marseille, total amnesia, lese majeste, Thema Null, progressive chess and I’m sure many more). We’ve also consequent seriesmovers and why shouldn’t consequent extend to e.g. threats in direct mates too? What’s important is that we capture the kinds of behaviour which can occur.
No matter the degree to which the innate idea of players moving alternately is attractive (particularly to those who may have written engines which depend upon this 😀 ) it is not useful to try to insist on any one underlying mechanism for this vast array. Flip just flips the move, and is irreducible.
Nor to my mind is it interesting to classify all these things in a fixed hierarchy at this stage. A jungle is not neat: things will belong to multiple classifications. A wooden hierarchy becomes a burden with every additional link that is added.
So please can we move on from this unproductive byway. Wizards of the Coast wouldn’t have got hung up in this way on their journey to producing 6th Edition, which earned them literally billions of US$. We are not going earn money from this, but they are the nearest we have to an antecedent in the space of trying to unify vastly complicated game rules that were never designed to be integrated.

Dmitri Turevski
Dmitri Turevski
May 31, 2021 18:22

I’m totally fine with having multiple alternative classifications.

What I’m not so fine with is classifications that have troubles with putting known objects into one of the established categories. Certainly we should be allowed to critically compare the classification systems?

Take phser-#, for example. “Opposing/Helping” classification puts it into the “helping” bucket (when black parries they are helping to reach the goal, as opposed to pser-#).
Where does “Direct/Help/Series” put it? Probably, it can be argued that it belongs to “series” and not to “help”, but that is not a simple argument and the burden of proof lies on whoever has proposed this classification:
> In fact, all “Plays” could be categorised as either “Direct”, “Help” or “Series”.

It has nothing to do with writing engines, really, just basic common sense.

Andrew Buchanan
Andrew Buchanan
June 1, 2021 05:04

I agree.
If one ends up in a classification debate, then normally to me this is a sign that it’s wrong to be trying to classify in this way. I always think: if a classification decision is narrow, then why should others come down on the same side? What’s the value of a forced narrow classification?
I think any underlying classification is being incorrectly conflated with the navigation through the browser to find terms. These are separate concerns. And navigation these days can have very general solutions compared to what existed before.

Andrew Buchanan
Andrew Buchanan
June 2, 2021 04:56

Shankar Ram kindly commented in response to https://juliasfairies.com/problems/no-1615/ “Also interesting – A PG with Retro content!”
PGs always feel to me retro and are also clearly non-adversarial. This is why a single procrustean* hierarchy although it may give illusions of quick win can be counter-productive. “Retro” like “fairy”, is a cross-cutting property. There are many direct, help & self problems with retro and/or fairy content, for example. Unlike fairy, retro can appear by itself with its own characteristic stipulations.
Although FIDE genres exist for album purposes, this organizational decomposition is not fundamental. It is a famous “don’t” in Information Architecture practice to base a taxonomy on an organizational structure. So please don’t do this “don’t”.
Retros are defined at the Retro Corner website https://www.janko.at/Retros/index.htm#whatis as problems where position legality is a key element. That seems to me as good a provisional definition as anything. Legal position = there’s a “proof game”. So you can’t get more retro than that. Proof games in this sense can’t be unique, so we are asked in a PG to find the *shortest* proof game out of those which are available. The notion of shortest, most elegant proof, “the Proof from the Book”, is a common mathematical topic of interest. In practice, we are often *given* the number of moves as a hint, but in most cases it’s equivalent to looking for the Proof from the Book. So all PGs are retros.
But in my opinion we should not be classifying right now. If a classification decision is narrow (normally indicated by vocal disagreement) that’s usually a good sign that it’s wrong to force it. Collecting the objects in the jungle and dropping them in some very provisional marshalling area is enough.
I wince whenever I read here that oh yes well this we classify in bucket etc. I don’t see this as progress. Rather, there are a bunch of orthogonal dimensions, and some of them will have indeed independent mini-hierarchies which naturally grow.
Onwards and upwards! 😀
* “Procrustean” = enforcing strict conformity to an arbitrary standard, disregarding individual differences. Procrustes was one of many villains defeated by the Greek hero Theseus. According to Greek mythology, Procrustes was a robber who killed his victims in a most cruel and unusual way. He made them lie on an iron bed and would force them to fit the bed by cutting off the parts that hung off the ends or by stretching those people who were too short!

shankar ram
shankar ram
June 2, 2021 06:50

Alternative procrustean viewpoints which were suggested have been included in the latest release without wincing. 😉

Dmitri Turevski
Dmitri Turevski
June 2, 2021 09:34

Funny thing, Andrew, as you wrote elsewhere: “A proof game is often like a jigsaw…”, in that comment “a proof game” can be replaced with “a [long] helpmate” with pretty much no loss of accuracy. In fact the “Intelligent” mode in the solving software works exactly like that: it starts with the possible finals and tries to figure out if they are legally reachable from the starting position. 
Helpmates are retros to confuse things further! 🙂
Another (slightly) funny thing is that I knew the word “procrustean”, but had to look up “conflated/conflation”.

shankar ram
shankar ram
February 25, 2023 14:46

Kostas Prentos has kindly informed us that he is not the inventor of the fairy condition “half-check“, as mentioned in the project.
He further points out that this condition was (first?) introduced in the Sake tourney of 2010 by Tadashi Wakashima, as part of the 53rd WCCC in Crete, Greece. See the tourney announcement and award (contained in the final bulletin).
The actual inventor’s identity is uncertain. It could be Tadashi Wakashima or Kohey Yamada or both. Perhaps our Japanese friends could confirm?

shankar ram
shankar ram
March 4, 2023 13:37
Reply to  shankar ram

Tadashi Wakashima has confirmed over Twitter that he is indeed the inventor!
Thanks to “antilles” for posting the question!

Would love your thoughts, please comment.x