Muidokan embu: 2-person forms for karate


The concept of 2-person forms as an adjunct to training is not new: in China many schools develop such forms as an additional means of practising their techniques in a contextual environment and packaging their knowledge . In China these types of forms or drills are known as " dui da quan". This tradition is, by contrast, not well established in karate. In order to find 2-person forms in Japan you have to delve further afield into arts such as Doshin So’s Shoriji Kempo or to the Japanese weapons arts such as Tenshin Shoden Katori Shinto Ryu (both of which use 2-person forms as their primary practice method).

[Note that Shorinji Kempo call their more sophisticated 2-person forms “embu” (meaning “demonstration”) and, for want of a better word, I have appropriated this word to describe 2-person forms generally.]

Ippon kumite vs. embu practice

The deficiencies in the standard method of practising kata bunkai (or any other specific technique(s)) in karate has been noted by such eminent researchers as Patrick McCarthy. In his article: "Sometimes you don't know how to fit in until you break out" Hanshi McCarthy writes:1

“During the years I studied swordsmanship [Tenshin Shoden Katori Shinto Ryu], at the Sugino honbu dojo in Kawasaki [Japan], I gained a huge respect for how the style accomplished its combative outcomes through using highly functional two-person training drills. While delving into its origins I became deeply impressed with the way classical attack scenarios had been first identified and studied before being ultimately catalogued into individual and collective learning modules each with prescribed responses and variations on common themes. Never having been terribly satisfied with the incongruous ippon-kumite practices of karate and unable to understand the defensive “effectiveness” of kata [as traditionally taught against modern reverse punch scenarios] or how its abstract mnemonic mechanisms were methodically linked back to actual real-life fight circumstances, I always felt that something was missing in traditional karate and from this blinding flash of the obvious [BFO] I finally realized what it was.”

Hanshi McCarthy’s research, of course, led him to develop his widely respected and highly effective HAPV 2‑person drills.

During the same period my brother and I were making our own exploration the concept of 2‑person drills. This began after seeing a Shorinji Kempo demonstration at the University of Western Australia in 1987, although we didn’t look into the subject seriously until (like Hanshi McCarthy) we were inspired by the methodology employed in the Filipino arts of arnis, escrima and kali and the unarmed Japanese arts of Tenshin Shoden Katori Shinto Ryu (our principal karate instructor, Bob Davies, is a direct lineage student of Otake Risuke Shihan) and, more significantly, aikijo – more on that later.

Ippon kumite (one step sparring) is certainly a valid and appropriate means of practising a technique in an isolated manner. It can be essential for learning “finishing moves” such as locks, holds, throws and other “take-downs”. But, as noted above, it has its limitations, the principal one being the lack of realism. The fact that your opponent: (a) attacks in a known, or at least highly predictable, manner; and (b) is static before and after the attack, militate against the usefulness of ippon kumite in progressing past basic study and practice.

Most ippon kumite in karate works on the principle of committed, "iken hisatsu" attacks - single punch, certain death. The attacks are very committed, and this allows the defender to perform defences suited to committed attacks. But in my experience, fully committed attacks of this kind are not that common in combat.

In other words, while ippon kumite might be useful as a starting point, it is really only the beginning of learning how to apply the particular technique. It is analogous to learning to drive by starting in a deserted car park.

The need for a dynamic context

It is my opinion that techniques need to be practised in a dynamic environment in order for them to be properly inculcated into one’s subconscious — ie. so that they become both reflexive and effective.

In fact, I am of the opinion that many applications of kata were never meant to be applied (and sometimes don’t even work) in a static environment. Consider the examples set out below:

It was aikido’s 13 count (or, in my opinion, 16 count) jo form that first got me thinking about static vs. dynamic bunkai back in 1993:

Count 2 of the kata is a jodan gaeshi (head height deflection) yet, somewhat perplexingly, there is no tenshin (body evasion) to accompany the move – either back or to any angle. I was told that instead one must “keep the forward momentum going” and “bury yourself deep into the attack” in count 1.

My confusion as to the application of count 2 quickly evaporated when, in the course of practising bunkai, a student and I inadvertently realised that whoever designed this form created it as a ready-made embu – ie. a “circular” form requiring no modification for 2 person practice.

[This information appears to be unknown to many schools that practice the same solo kata, but that have a 2 person version with a completely different “omote”, but more on this another time.]

Figure 1: a series of moves from the 2 person aikijo 13/16 count form


* count 8 involves an “open door” deflection of count 1;

* count 9 is a straight thrust over the top to the kata side’s throat depicted in the frame 1 of Figure 1);

* having no time to reverse momentum (or even alter it), the kata side’s best bet is to remain committed to the forward moment, using the jodan gaeshi to let the attack pass over the head (count 2 — see frames 1 and 2 of Figure 1);

* count 3 of the kata is then a counter strike to the head which is evaded by count 10 (see frame 3 of Figure 1), and so on...

Figure 2: a series of moves from the Muidokan gekisai dai ichi embu

Consider the bunkai of the penultimate “block/strike” combination in gekisai dai ichi:

The te osae uke (hand depressing block) is generally applied against a punch (leading or reverse) and followed with 2 punches delivered simultaneously to the floating rib or to various vital points on the abdomen.

Note however the artificial nature of the attack and that the attacker remains motionless throughout. The technique simply doesn’t work if the attacker does what he could (and no doubt would) do in actual combat - and that is move away once his attack has been deflected.

By contrast, note Figure 2 depicting moves from the Muidokan gekisai dai ichi embu: the osae uke is effected after a committed mae geri to suppress a counter that you are falling into.

Your forward momentum and the downward motion of the osae uke not only ensure an effective deflection but serve to set you up for an effective counter. This particular sequence can only really be practised in a dynamic, embu-like context. It is unlikely to be even considered via ordinary ippon kumite analysis.

Embu as a forum for learning how to cope with counters

Accordingly I think what you're really learning in embu is that practically every counter can itself be countered. You learn how to "rescue" a situation (as illustrated in Figure 2).

By contrast, if you practise a drill that leads one partner to do such a daft manoeuvre that he or she literally has no chance of defence against your counter you should ask whether you haven’t constructed a “straw man” – just to knock it down.

Incorporating embu practice into an existing syllabus

It occurred to us, as it has with many others, that embu provide a dynamic, yet safe, environment in which to apply and inculcate techniques appropriately and effectively.

However we were faced with a quandary: how could/should we incorporate embu into our existing (primarily goju-based) karate syllabus?

Clearly we could adopt or create a series of new 2-person drills unrelated to our kata (except in a loose sense). However this would not provide (at least directly or comprehensively) a dynamic environment in which to understand and practise existing kata bunkai: it would be a “layer” on the existing material (albeit a useful one). The kata bunkai study/practice would remain rooted primarily in ippon kumite.

Another issue we considered was syllabus complexity:

Kata function to “package” information in a traditional system. However as with any school we realised we already had a multitude of techniques that were not obviously in the kata. These include “omote” (an attack against which the kata defends) and certain “oyo” (extrapolations) which are highly useful and important to student development (and which might not occur anywhere else in kata).

In other words, the kata were not the repository of all the knowledge (at least not obviously). There was/is a significant number of techniques implied in / related to kata that are not actually represented in them.

Once upon a time it might have been seen as necessary to “hide” certain “secret” techniques from all but the most trusted students. However this ideology does not fit within the modern education paradigm and is certainly opposite to our school’s philosophy. Moreover such a syllabus structure is almost guaranteed to result in “information loss”, especially over a longer period of time.

In the case of our dilemma, adding yet a further “layer” of unrelated 2-person drills to this background would simply have exacerbated the existing “problem” of syllabus complexity, such as it was.

Rather, it was (and remains) our view that to be effective in preserving and imparting knowledge a syllabus needs to be structured coherently, logically and, above all, as economically as possible.

The importance of kata-related embu

Accordingly while we, like Hanshi McCarthy and others, became convinced that embu could “bring karate to life”, we wanted something more specific than this:

We wanted our embu to bring each kata to life.

As traditional martial artists, and for the reasons stated above, we have always strived to make our syllabus entirely kata-based - ie. ideally we wanted all our knowledge to be packaged in forms without any "stray" techniques. At the same time, we didn’t want to reinvent the wheel by “redesigning” our traditional kata or replacing them entirely with “new” forms.

We had in the back of our minds that some schools (eg. Tenshin Shoden Katori Shinto Ryu) teach exclusively by means of 2-person forms with no “stray” techniques/elements. Clearly we could try to do the same. The problem is that we would be running the very real risk of “throwing the baby out with the bathwater”. Designing a new martial art from scratch might or might not produce an effective system. However it can’t help but fail to take into account relevant past knowledge. And we, as humans, collectively progress in terms of knowledge by building on the work of previous generations. Even Sir Isaac Newton is said to have attributed his achievements to the fact that he was “standing on the shoulders of giants”.

Could embu and any other “stray” techniques be incorporated into our existing syllabus without creating “unrelated layers”? In other words, could embu serve to explore kata bunkai while simultaneously combining essential “stray” elements/techniques related to the bunkai? We found the answer to be: “yes”.

The anatomy of an embu

The challenge was not insignificant. How does one go about creating a form that meets all the above criteria, is economical in length (18-22 moves so that it is easy to learn) and, above all, is “circular” insofar as it creates only one new sequence (rather than 2)?

More specifically, how do you ensure that the kata’s first count defends against it’s middle count and that all the rest fit in like a jig-saw puzzle? And how do you do this, without compromising the technical base and the “essence” or feeling of the kata bunkai?

For guidance and inspiration we went first to the 13/16 count jo form and spent considerable time analysing its architecture. Later we examined closely the architecture of xingyi’s 5 elements, particularly their “constructive” cycle (which corresponds to their order in solo practice) and their “destructive” cycle (which corresponds to their order in 2 person practice).

What we ended up with was the realisation that while 2 person forms might involve some sophisticated mathematics in terms of their architecture, their manifestation is ultimately a giant, 3-dimensional game of “rock, paper, scissors”. Consider for example that xingyi’s 5 elements have often been represented in Figure 3:

Figure 3: the xingyi pentagram

Each element corresponds to a defence and an attack. The pentagram’s perimeter, moving clockwise (in red), plots the course of the “constructive” cycle (ie. the solo form). On the other hand the pentacle (5 pointed star), moving anti-clockwise plots the 2 person version.

Figure 4: the aikijo 7 point star

Figure 4 shows aikido’s 16 count form as a 7 point star using the same methodology:

Taisabaki and tenshin: the foundations of embu

We discovered that the key to moving from one point in an embu “star” to another lay in understanding the nature of the body movement (taisabaki) and evasion (tenshin) employed in each kata bunkai. In short, body evasion inherent in bunkai is the foundation of embu since it will determine where you are placed relative to your opponent for the next attack or defence.

The subject of taisabaki and tenshin is, in itself, a mammoth one and not one which can be dealt with comprehensively in the context of this article. Suffice it to say that there are 8 or so principal angles of evasion (10 if you include up and down – more if you include compound movements such as weaving). There are also multiple ways in which those angles can be attacked (ie. stepping, lunging, stepping and pivoting, stepping and turning etc.) Then there is also the question of what stances are used, where weight is distributed at a particular point, and so on.

Accordingly we started the embu process by distilling the essential taisabaki or tenshin that forms the base of the bunkai of each kata. We then used this with techniques in the kata and those that are commonly taught alongside the kata in the course of bunkai or “omote”.

Consider the gekisai dai ni embu:

Figure 5: a series of moves from the Muidokan gekisai dai ni embu

The tenshin/taisabaki used in this embu included the standard 45° back into sanchin and 45° back into shiko dachi used with a variety of bunkai for this kata.

You’ll notice that directly after an attempted ashi barai, the person on the right in Figure 5 evades a front kick by using the 45° shiko evasion. Other tenshin/taisabaki used include the “open door” movement (forwards and back) and the sideways neko ashi dachi.

Our gekisai embu are fairly basic because the tenshin/taisabaki is basic in that kata. However these are necessary because you are, at that level, just learning the art of evasion. Our gekisai 2 embu is noticeably more advanced than the gekisai 1 embu. If you look at the seiyunchin one you'll see more realistic attacks (like those HAPV addresses) and more realistic defences, so that the embu does look more like fighting with the more sophisticated kata.

What I see embu doing is training you to apply your kata bunkai in sparring/fighting. I think this aspect is sorely missing in karate; most people do their kata and basics, then jump around doing faux boxing moves. There is no "bridge" between pre-arranged and free sparring.

The result and its fidelity to the original kata

Performed by one person an embu looks not dissimilar to the kata it is based on – the techniques are all there but the embusen (pattern) and sometimes the sequence differ (sometimes significantly). However most importantly we have endeavoured to retain the same “feel”.

This is diametrically opposite to many of the “standard” 2-person kata developed by Seikichi Toguchi, one of Chojun Miyagi’s students. The kata side is virtually identical to the kata, while the “omote” side is a completely new sequence. In other words, it is a literal superimposition of attacks on the kata template. In my opinion karate kata were never designed to be 2 person forms. From my perspective a drill that tries to force a kata into this mould is a “flawed” and inefficient training tool.

I say "flawed" but I don't mean this in disrespect. Everything is flawed. We have tried to improve the technology through a systematic and scientific approach based on my experience. I feel our drills are an advancement on Toguchi's, although I appreciate some may view this as controversial. I admire Toguchi's contributions greatly, but to assume that development should cease with the masters of old is, to me, absurd.

Hironori Otsuka, in his book 'Wado-Ryu Karate', wrote:2

"It is obvious that these kata must be trained and practised sufficiently, but one must not be 'stuck' in them. One must withdraw from the kata to produce forms with no limits or else it becomes useless. It is important to alter the form of the trained kata without hesitation to produce countless other forms of training."

Accordingly, while the embu are derived from existing kata they necessarily depart from their structure, retaining only the essence. This is consistent with the theory that new training drills must have an added dimension or else they are simply a pointless variation of an existing sequence.

Embu vs. “omote”

As I have said, an advantage of embu is that they do not comprise "2 further kata" - ie. we don't have 2 additional sequences to learn – the kata movements and the “omote” (attacking/other side). Rather, we have just one in addition to the traditional kata, which remains unaltered.

The aggressive and defensive moves in the embu are all from/related to the kata. Neither side is more "aggressive" or "defensive" - you deflect a technique and counter, your opponent deflects your technique and counters etc. using principally the kata moves. If you look closely many of the kata techniques are responses to other techniques from the same kata: just like xingyi's 5 elements which are capable of being used against each other.

Why yakusoku kumite doesn’t fill the “gap”

Why wouldn’t some form of yakusoku kumite provide the necessary “dynamic” environment? For example many karateka practise a kind of kumite where —

(1) “A” initiates an attack or series of attacks;
(2) “B” defends against those attacks and performs a series of counters;
(3) “A” defends against those counters and performs a series of further counters that finish the sequence.

Some have argued that this type of training is sufficiently dynamic in that it allows for a “responsive” form of kumite.

Yet it is my view that the need for a truly dynamic environment is largely unaddressed:

the first and last moves are not in a dynamic environment (i.e you are not applying a techniques mid-fight except at move (2)). Move (1) starts things from nothing. Move (3) encounters no resistance (in the form of a deflection or further counter). In fact, “B” stops his/her participation at move (2)!

This is not to say that yakusoku drills are not useful. Like ippon kumite, they provide a forum for practising certain moves that cannot be included in continuously flowing embu, eg. finishing moves. However, “B” is undeniably grooving a passive response at point (3). He or she is “learning how to lose” at least to some extent (refer to my earlier comments about learning to deal with counters).

By contrast, every technique in an embu can be performed in a continuum. You can cycle through the entire flow drill once or twice or more so that all the techniques have been performed in a dynamic environment at least once - including the "opening" move.

In yakusoku drills the first and last moves are always just that - they are never "in the middle".

How to make yakusoku kumite “continuous”

However what if your yakusoku drill were cleverly designed so that the counter in point (3) were capable of being defended against by the techniques comprising point (2)?

(1) attacks (3)
(2) attacks (1)
(3) attacks (2)
(1) attacks (3)
and so on...

Both sides could then go “hell for leather”. Same techniques, same emphasis, same sequence, but one side doesn’t “stop” or “give in”. One side isn’t being trained to “lose”.

Figure 6: a diagrammatic depiction of 3 point embu

What we have above is a basic 3 point embu that can be set out diagrammatically in Figure 6. You will see that just like the 5 and 7 point stars, it has both a “constructive” (red) and a “destructive” (blue) cycle. This is in fact a visual representation of “rock, paper, scissors”.

3 point embu are very common in, say, arnis/escrima/kali (eg. box pattern single stick, de cadena trapping drills, etc.). I have found these to be refreshingly pragmatic, if not essential in weapons. Why not karate?

By comparison our embu comprise 5 to 10 point drills – mostly to try to get all of the “essential” bunkai into one package.

Does the defence of each attack condition people to “be thwarted”

No. Both partners are doing their best to win. You don’t “let” your partner thwart you in the embu: if he or she doesn’t thwart you, you can follow through with your own (successful) attack. Nor is it the case that the techniques are designed to "allow” you to be thwarted: every technique is responded to with the best available defence offered by the kata (or failing that, the simplest defence).

Again, you should not assume that some counters have no defence. In my opinion this is not true, unless your tactics are absolutely useless and you have put yourself in a real “fix”. Embu teach you good tactics and, just as importantly, how to deal with having your own attacks thwarted.

Figure 7: “Pao quan” – one of the 5 elements demonstrated by
Master Chen Yun-Ching and Dan Djurdjevic

I first appreciated that no counters are “invincible” in the context of xingyiquan’s 5 elements: I once performed a counter with one of the 5 elements and thought my position was unassailable, only to find my opponent “pulling a rabbit out of the hat” with another element. Before I had knew it I was facing an “invincible” counter myself.

The “too many kata” dilemma

But aren’t the embu just “extra kata”? Don’t we have enough already? The answer to the first question is yes, if you define kata as a sequence of moves that one person may practise (a sensible definition).

But while we are, in a sense, "adding kata", the sequences are directly derived from the classical kata, so they are more an exploration or variation of the existing forms. This is an attempt to ensure not only that knowledge is not lost, but that development (and knowledge dissemination) is enhanced.

Embu for grappling techniques?

We call our 2-person striking/kicking drills “embu”, however we also have our grappling methods packaged in the form of what we call “kata tuide” drills. These can be practised both standing and on the ground. By contrast our tuide drills are “lock flows” rather than mutual exchanges. The examination of these drills is however outside the scope of this article and requires separate analysis.


Accordingly each Muidokan embu is really a collection of bunkai practised in a continuous, dynamic context.

It is not exhaustive of bunkai. Rather it is a circular sequence that can be practised solo or as a 2 person form and that –

(a) is built on the essential principles of evasion and body shifting inherent in the bunkai; and

(b) applies the principal techniques that are in the kata, implied by the kata or against which the kata defends; and

(c) teaches the student the principles of counterattack and dealing with counterattacks.

Embu practice does not replace ippon kumite, yakusoku kumite, or any other pre-arranged or free sparring. It should be ancillary to these activities.


1. See:
2. "Wado-ryu karate” by Hironori Otsuka Rising Sun Productions 1997 ISBN: 0-920129-18-8

Copyright © 2008 Dejan Djurdjevic