Skip to main content.

RE.C.ORD. - Recursive Classification Ordering.


The aim of RE.C.ORD is that of providing a completely unambiguous ordering for Japanese Kanji characters, and to sort them in a list that can be easily searched by non experienced users, given a minimal instruction on the ordering criterion.

The highest obstacle a student finds in learning the Japanese language is very likely the writing system. Kanjis, or Chinese characters, are a formidable foe for the even willful learner, possibly more than the ones found in the original Chinese language, as their usage and reading can vary under many circumstances.

Many systems has been developed to sort kanjis, so that they can be found in look-up tables or that words composed of kanjis can be sorted in dictionaries. The vast majority of them is based on the recognition and ordering of constituent parts, called radicals. In example, the kanji 「好」- Kou - means "liked, beloved, pleasurable". It is composed of the radical of a woman on the left, and a child on the right. The meaning is quite clear: a woman loves her child. Under radical-based ordering systems, this kanji would be ordered given a numeric reference for the woman and the child (under various systems, they are numbered as 38 and 39). Usually, the more complex radical, that is the one being written with more strokes, "leads" the ordering, the other ones following in ordering relevance.

On a student point on view, or considering the needs of a person casually coming in touch with some Japanese text that needs a minimal interpretation, the systems based on this principle have three main drawbacks:

  • The user must know the vast majority of the radicals, which counts up to around 200 units, and should also have knowledge about their ordering in the radical table that is used by a certain ordering system. Although this is a lesser mnemonic effort with respect to knowing every kanji, it still requires significant time and effort.
  • In many kanjis, radicals are written in a confusing way, often resembling similar ones. It is not unusual to find strokes that are categorized as completely different radicals from the ones they resemble because of historical reasons. There is a non insignificant number of kanjis for which different scholars have different, equally authoritative opinions about the constituent radicals; although a student may decide to stick with one system or another, this fact alone gives an idea of the difficulty of telling radicals.
  • Counting of character and radical strokes is often a not simple task, as there are several conventional rules that must be taken in account. In example, all the squares are counted as three strokes, as the top-right stroke, although angular, is to be executed with a single movement of the hand. Also, provided that the strokes of radicals are correctly counted, kanjis in which there isn't a radical having more strokes the other ones can lead to ambiguities in the character ordering which is often resolved with totally arbitrary conventions.

To make thing worse, many kanji collation systems are based on total stroke count, which defines classes of kanjis under which specific ones are to be searched. In the example of 「好」- Kou - we have a six strokes kanji; once counted all the strokes in the kanji, the desired one is to be found among all the six stroke composed kanjis. The given table is then ordered with radical indexing or other criteria. Reducing the search scope by limiting the search to small subsets may seem an excellent idea at firsts, but the difficulty, or, in many cases, even the ambiguity in counting correctly kanji strokes leads the non-professional user to confusion, forcing him to search into the tables containing the kanjis with one, or even two strokes more or less than the count he has supposed. When a kanji is not found, the inexperienced searcher is never sure about the fault being in his own mis-looking of the target table or in his own counting of the kanji strokes, forcing him back and forth between different tables.

Finally, despise this difficulties can be overcome with practice and despise some ordering system is more helpful for the inexperienced user than others, existing orderings lead to ambiguous results that are resolved using confusing rules, or that at times are not resolved at all. This is a minor problem for lookup tables, as the user would just search the desired kanji in a row of a small number of ambiguously ordered ones, in the order of 10 at maximum. However, this becomes a major problem in dictionary compiling, and especially in computer generated list of items (that must be then searched by human eyes). This is so true that the vast majority of Japanese dictionary sorting is based on either historical orderings or, more often, on reading ordering. In the first case, the searcher must know the number associated with a certain kanji, which is totally impractical for the inexperienced user. In the second case, words are listed under a certain reading index, as in roman alphabet based dictionary. This would seem an elegant solution to the problem, but actually it is highly impractical when applied to the the Japanese language. First of all, knowing the correct reading of a kanji is not an elementary matter, as many kanjis change their reading depending on the contexts, and a non small number of them has also irregular readings in names or in common words. Secondly, Japanese language is rich of homophones that are written with different kanjis, as the word "kouka", which may be written as 「効果」- meaning "result", or 「降下」- meaning "descent" (actually, there are no less than 17 kanji combinations associated with "kouka" on commonly found dictionaries). This makes this ordering absolutely unsuitable both for the non professional user and for compiling lists of lexicographically sorted entries.

The need of both a criterion that can be employed to compile easy to use lookup table and dictionaries and other lexicographically sorted lists brought to the development of this system.

Working principles of RE.C.ORD.

RE.C.ORD system is based on three principles, whose relevance in the order is decrescent:

  1. Recursive categorization of kanjis on their overall shape
  2. Count of strokes in each part
  3. Order and type of the strokes

Actually, the perfect lexicographic ordering of a kanji under RE.C.ORD rules requires the usage of a reference table indicating stroke order, or an excellent knowledge of kanji writing rules; but this is required for the compilation of the lexicographical order only. Once built, the table, or the dictionary, can be searched by simply knowing the recursive categorization principle. An extremely basic knowledge of kanji writing principles helps, as the ability to count correctly the strokes of a radical, but they are not necessary. The notes contained in this document, together with a small training or self-training in searching the list, should be more than enough to provide the user with all the instruments it needs to consult the table proficiently.

To build the table, the stroke drawing order suggested by O'Neill in "2000 Essential kanjis", as described by Takatsuka Chikudo in Kakikata Jiten, has been used. The stroke writing order as presented by Takatsuka Chikudo is considered one of the most authoritative.

RE.C.ORD categories and recursiveness.

RE.C.ORD subdivides kanjis in seven sets of ordered categories, basing on their overall shape:

  1. Whole
  2. West-East
  3. Cap
  4. North-South
  5. Roadside
  6. Embracing
  7. Encircling

The fact that the categories are ordered means that all the symbols in one category comes before the ones in the next category. This principle is extremely important.

Whole kanjis are the ones that cannot be possibly divided in parts. Kanjis that are in this category are ones as 大 - dai, 日 - nichi, 来 - rai and so on. Notice that some of the whole kanjis may be actually be formed up by different radicals mixed together, but RE.C.ORD is based on their overall look rather than their correct historical composition. Kanjis with small appendices above them are still considered whole, as in 白 - shiro. However, appendices below the kanjis are considered apart, as in 見 - ken (which is a north-south kanji).

West-east ideograms are by far the most common ones. 好 - Kou is one of them; another example is 沖 - oki, where the three small sign on the left are meaning "water", and the part on the right is "middle", giving the meaning of "open sea" (in the middle of the waters).

Cap Kanjis are a set of kanjis with a cap on top of them. The cap may vary widely in shape. In example, 家 - ie has a three strokes cap (the small appendix on top, the vertical sign on the left and the top/right sign drawn as an "hook"). 首 - kubi has a cap made of two small appendices and an horizontal stroke, while 食 - shoku has a roof-like cap. Kanjis with a single stroke above them are considered cap kanjis too. This includes 戸 - to.

North-South kanjis are organized vertically. 台 - dai makes a perfect example. North-south ideograms may also have adjacent parts: in example, 男 - otoko, 早 - sou, 舌 - shita are all north south, despise the two elements they are composed are touching each other. As said, 見 - ken and all the kanjis with lower appendices are categorized as north-south. Also, there are some north-south kanjis having the northern one "invading" the southern one, as in 着 - ki, where the diagonal stroke on the bottom left is actually considered part of the top element, and in 者 where the diagonal stroke starts in the cross-shaped topmost part and touches the bottom part.

Roadside ideograms are the ones presenting the radical of a road, or a similar one, on the left-bottom part. The basic shape of the road radical is 道 - that three strokes part composed of the top-left small quote-like stroke, the left stroke and the bottom one. However there are many variations on the theme, as in 処 - tokoro, 廷 - tei and 疋 - hiki.

Embracing kanjis category covers a very various class of kanjis surrounding some content on two sides. In example, 圧 - atsu, 直 - choku are all in this class. Appendices to any side of the embracing element will be considered as part of the embracing, as in 県 - ken, 危 - i or 病 byou. Embracing kanjis are furthermore subdivided in four categories depending on the side they are surrounding from. Category A surrounds from upper left corner; categories B, C and D follows in clockwise order, respectively upper right, lower right and lower left.

Finally, encircling kanjis are the ones enclosing a content from three sides or in a square, as 区 - ku in 国 - kuni or 図 - zu. There are also some complex encircling kanji as the one formed with the 門 - gate radical, as in 門 - mon (same as radical) and 間 - aida. Encircling kanjis are subdivided in five classes depending on the side they left open: class A kanjis leaves open the right side, B leaves open the bottom, C the left side and D the upper part. E class is formed by the completely surrounding kanjis.

The ambiguity found in the vast majority of the ordering methods depends on the fact that only a small subset of the kanjis are that simple as the one that has been used as examples up to date. The vast majority of them is a composition of many parts, organized in a barely schematic fashion. RE.C.ORD system analyzes the complexity of those kanjis recursively, by considering a "pivot" element then analyzing the remaining part of the kanji. The pivot element is that one that defines the class: for east-West kanjis is the one on the left, for cap kanjis is the cap, for north-south ones is the topmost one, for roadside ones is the road or road-like radical, for the embracing class it's the outer part as for the encircling ones. The kanjis are ordered by categorizing the pivotal element, and then by categorizing the remaining part, till only whole elements remain. In example, 場 - ba is a west-east kanji; the left part is already whole, but the right part is a north south; so this ideogram will come after 坑 - kou, having the same pivot, but whose non pivotal part is a cap. Also, 所 - tokoro will come after all the west-east kanjis whose left part is composed by a whole element, as its left part is a cap. The process is iterative, and can involve many steps: in example, 庭 niwa is an embracing kanji whose internal part is a roadside one; as embracing is the second-last category, and as roadside is the third-last one, you can expect this ideogram will come near the end of the table.

Stroke count

Recursive categorization is an excellent way to parcel the size of the search, but still there is the need to count the strokes a pivotal element is made of to pinpoint it. Actually, it is just necessary to count the strokes the most important part of the pivotal element is made of; if that, combined with the categorization, is not enough, then the second whole element must be counted, if that is not enough the third one must be counted and so on. With this progressive approximation process, a complete and precise count of the strokes of each part is seldom, if not never, needed. It is usually enough to consider the complexity of the whole category element actually being scrutinized to find the kanji.

In example, consider the ideogram 赦 - sha. We'll search it in the west-east kanjis whose left key is north-south. The north element of the pivotal left part (the cross) is three strokes, so we know this item will come before 静 - sei, as the top part of the left element of that kanji has four strokes.

However, we also know that the searched "sha" will come after all the west-east kanjis with a north-south left part with less than three strokes, as 殺 - satsu which has two. Our "sha" will be found before 鼓 - ko as the lower part of the left element of 鼓 - ko is a north-south.

Stroke order and type

At times, even the stroke count applied to the elements of a category is not enough to pinpoint a single kanji. Consider 木 - ki and 心 - kokoro. They both have four strokes; they are easily found in the whole class under the 4 sign kanjis, but to create a sorted list, or to search a wide one, it is necessary to unequivocally determine their order.

RE.C.ORD uses a lexicographic ordering of strokes with the following priority:

  1. Horizontal strokes
  2. Semi horizontal strokes
  3. Diagonal strokes
  4. Semi vertical strokes
  5. Vertical strokes
  6. Hook strokes
  7. Comma or dot-like strokes.

Horizontal and vertical strokes are just what their name suggests. Semi horizontal strokes are the ones that are slightly inclined, as the topmost stroke in 手 - te. Oblique strokes have a visible slope, as the left-top stroke in 生 - sei or the long diagonal stroke in 必 - hitsu. Semi vertical strokes are more or less vertical strokes with some "opening", as the left stroke in 月 - tsuki, or the left stroke in 川 - kawa. Hook strokes are the ones used to close squares, (top/right is hook for all the square-shaped radicals), and also angular strokes as the topmost one in 台. Comma and dot like are the small strokes used on top of caps, as in 学, which has three of them, or on top of some characters as 自 - ji or 単 - tan, and again the top left quote in the roadside kanjis or the three strokes in the water radical as in 沖 - oki.

Given this rules, we can now state that 木 - ki is before 心 - kokoro, as ki is drawn as horizontal, vertical, diagonal and diagonal, while kokoro is dot-like, hook, dot-like and dot-like again.

Of course, it is necessary to rely on a determine stroke drawing order for this method to be effective, but in the vast majority of case, the simple principle that top and left strokes should come first is more than enough for the normal user to apply RE.C.ORD correctly, or anyhow precisely enough to find the searched kanji at first glance.

For how precise this method may seem, it is still incomplete. There are still some cases in which the perfect ordering requires more specification. The following rules can be applied to determine the precise placing of a kanji in the order:

  1. T-like strokes come before
  2. Partial strokes come next
  3. Short strokes come next
  4. Long strokes last
  5. Disjoint strokes after joint strokes
  6. Downward hooks before upward hooks
  7. Concave hooks before convex hooks
  8. Simple hooks before complex hooks
  9. Higher strokes before lower strokes

T like strokes are horizontal, semi horizontal or diagonal strokes that are intersected in a T junction by a vertical stroke, while partial strokes are horizontal, semi horizontal or diagonal strokes intersecting a vertical one. So, we can determine that the order of three different pivotal items being all drawn as an horizontal, vertical and diagonal stroke (in this order): 項 - kou comes before 坊 - bou as the t-like topmost horizontal has lower value, and 拡 - kaku comes last as bottom diagonal is not t-like as in bou.

As for partial strokes, the small horizontal stroke in 上 - jou matches the definition (as if it were "blocked" by the vertical line), as the diagonal one in 下 - ka. So, 上 - jou comes before than 土 - tsuchi, as the first stroke in jou is partial.

Long and short strokes are to be determined with respect to the kanji itself, compared with the other kanji having the same stroke order and type, differing only in size. 由 - yuu comes after 田 - ta, as its drawing order and stroke type are the same, but its middle vertical line is longer; 田 - ta comes after 甲 - kou as 田 - ta is drawn vertical, hook, vertical, horizontal, horizontal while 甲 - kou is drawn vertical, hook, horizontal, horizontal, vertical. Again, 甲 - kou will be before 申 - saru because, being their drawing order the same, the longer vertical stroke in saru (spreading above and below) determines saru to come after.

Disjoint strokes have less priority than joint strokes with the same orientation. In example, 入 - nyuu and 八 - hachi have both two diagonal strokes, but as 八 - hachi strokes are disjoint, it comes after 入 - nyuu.

Upward and downward hooks are distinguished by the way their angle is facing. 台 - dai has an upward hook on top, while 甬 - you has a downward hook on top. They are both north south, with the topmost element being two strokes, an hook and a dot-like; the orientation of the hook determines 甬 - you to be first.

Also, the top-right stroke hook in squares is to be considered a downward hook.

Concave hooks are the one bending inside the ideogram, while convex hooks are the ones bending outside. In example, in 周 - shuu the top-right part of the outer frame is a concave hook, while in 風 - kaze it is a convex hook.

Complex hooks are the ones bending more than one time. In example, in 乃 - no the hook stroke (forming the B shape) is a downwards, concave and complex hook; it bends once at the top, goes down a bit and then bends again horizontally, to terminate downwards.

Characters drawn very similarly, except for a small detail (usually a dot-like stroke), as in 犬 - inu and 太 - ta, can be ordered through the last rule. Both the kanjis are written as horizontal, diagonal, diagonal and dot-like strokes, with the diagonal strokes being both long and the horizontal line being the same. The fact that in 犬 - inu the dot-like is above the place it has in 太 - ta determines 犬 - inu to come first.

This group of rules is to be applied in the given order only and exclusively when the kanjis taken into into consideration have the same stroke count and sequence. The reason for this is threefold. First of all, many of the given rules must take into account kanjis as couples; a "short" stroke is only "shorter" in one kanji with respect to another; it would be quite impractical to measure the actual size of kanji strokes in millimeters...

Secondly, using as predominant criterion the presence of T like, short and long strokes, and the other rules given in the second group, would result in moving similar kanjis far one from the other. In example, 土 - tsuchi and 士 - samurai are one right before the other; if short horizontal strokes were to considered before ALL long horizontal strokes, many kanjis may have found space between them.

Last but not least, this rules are given as a residual criterion to create strongly ordered lists; in practice they are actually determining in a very limited set of cases, in the order of the few dozens. Even a fully proficient RE.C.ORD user may safely ignore this rules, also in consideration of the fact that the effect of them is just that of laying very similar kanjis down in a definite order one beside the other. By providing them as a subordinate criterion, the user of the table won't be forced to remember or reference them; their utility is meant mainly for the list compilers.

Subdivision rules a and peculiarities.

RE.C.ORD is a complete ordering system, so it does not need to consider special cases. However, it is necessary to explicate how some part of kanjis are considered, to avoid confusion and clear the field from interpretation ambiguities.

Whole kanjis are the ones:

  • built on a single radical.
  • formed up of single radicals and comma/dot strokes on top.
  • formed of a single radical with appendices on top, as 角

West-east kanjis are made up of a radical or more on the left side, and one or more on the right. The left side is usually easy to be identified; however, the left stroke is often composed of unconnected elements that must be considered as parts of one single whole element:

  • two adjacent strokes as 帰 - ki.
  • two or three comma/dots as in 冷 - rei or 海 - umi.
  • a single stroke near a "standing man" radical as in 修 - shuu.

Cap kanjis are the ones being covered fully with a cap-like stroke or stroke set. Usually, the cap stroke is clearly separated from the rest of the kanji, which comes always below, and it comes in a vertical - hook pattern as in 字 - ji (dot-like, vertical, hook) or as roof-like strokes as in 食 - shoku. Cap are generally composed of two or more strokes. One single horizontal long stroke constitutes a cap only in the following cases:

  • When it doesn't touch any other part of the kanji like in 戸 - to.
  • When the very next stroke is a hook, as in 不 - fu (the left "1" like stroke is a hook), as in 至 - ita or as in the left part in 到 - tou.
  • When the very next stroke is a dot-like, as in 面 - men or 百 - hyaku.
  • When the parts touching the stroke are elements of a west-east subpart as in 死 - shi.

Non horizontal single strokes, as in 乏 - bou are never to be considered caps.

Other than lone horizontal stroke caps, here follows the other specification for cap elements:

  • Caps starting with three small strokes on top of the cap may be either dot-dot-dot or vertical-dot-dot: if the central stroke is to be drawn first, then it's considered vertical. Consider 光 - hikari; it's a cap of four strokes above a two-legs radical, and the middle stroke is to be drawn first. It's visually easy to tell it's vertical. The cap on top of 党 - tou has the same nature, but the one on top of 学 - gaku is different; it may be a bit harder to tell the fact that the three strokes above 党 - tou are vertical, dot-like and dot-like, while the ones above 学 - gaku are all dot-like, but a bit of training should help in this sense.
  • Roof-like caps may have two dot-like strokes above them as in 谷 - tani. The topmost strokes are considered part of the cap, so this cap is a whole of 4 strokes of type dot-dot-oblique-oblique.
  • About roof-likes, there is often a single strokes below them, as in 合 - gou. In this case, we consider the single stroke below the roof as a part of the cap. So, 合 - gou is a cap kanji whose cap is diagonal-diagonal-horizontal.
  • Kanjis having higher part which is composed of other common radicals are never a cap. Consider 益 - eki and 挙 - kyo, in example; they are both north south. As a rule of thumb, caps never have downward strokes except for the extremities. This includes top crosses and X like elements as in 真 - shin.
  • Caps may actually be below an element, forming north south kanjis. In example the hand stylization, that is a semi-horizontal stroke with three dotlike below it, it's a whole on itself. The 妥 - da, ideogram is a north-south with the north part being semi horizontal with three dotlike strokes under it (which, as said, is a stylization of a hand), and the lower part made up by hook-diagonal-horizontal stilyzed shape of a woman. So, when the stilyzed hand is above a cap, like in 受 - ju, the ideogram is a north-south, whose north part is formed by the upper 4 traits, and the south parth is a cap element (two traits cap aboce an hook-diagonal whole).
  • Upper strokes combination of a short diagonal stroke and a long horizontal (usually indicated as one of the form of the "man" radical) is considered a cap and stands always by itself. In example, 気 - ki is a two strokes cap above a single stroke cap, above an encircling.
  • In general, when two or more unconnected horizontal strokes are placed in a row, they are never considered caps, but wholes. This applies for 二 - ni, 三 - san and so on. On this line, 言 - gen is considered as a simple north-south with the north part being a set of four horizontal strokes; it's not a cap. The traditional hand-writing version of this radical uses a dot-like stroke instead of the topmost horizontal stroke, but the printed version has been preferred because casual users are more likely to come in touch with this version first.

North-south kanjis are just kanjis above kanjis. It is relevant to notice that:

  • One single stroke, no matter its shape, can't be considered as the northen element of a north-South kanji. However, one stroke below a more complex part can be the south part of a north-south kanji. Given this, 丞 - jou is a north-south strokes whose north part is formed by the hook above (and joining) the water radical, which is the central part; the southern part is just the long horizontal stroke.
  • Exception for the above rule are single non horizontal strokes that are detached from the lower part. Horizontal lone strokes are to be considered cap, so they can't be considered north-south.
  • Long horizontal strokes that are adjacent to both the upper and the lower part are associated with the upper part, as 寺 - tera. The cross in the upper part includes also the "ground" longer strokes, so we have two three-strokes elements.
  • Diagonal strokes touching both the parts are associated with the upper part. In example, 者 - sha and 着 - ki the diagonal stroke is assigned to the upper element.
  • Downward appendices, even if they are considered dot-like, are separated into a different element. In example, 員 - in is a north south element with north part being a whole (square) and the bottom part being again north-south, composed of a top part (five strokes square-like) and set of two dot-like strokes as the bottom element.

At first, some north-south kanjis may seem better fitted in whole category, as 幸 - sachi, which is a north-south, whose top part is considered having three strokes, and the bottom part is considered dot/dot above a four stroke part.

As north-south kanjis are including the ones in which the upper part is partially invading the lower one, ideograms as 馬 - uma, 島 - shima and 鳥 - tori are considered north-south.

Also, some confusion may arise in the case that a north-south part of a Kanji is a west-east being above another perfectly aligned west-east element, like in 範 - han. As there are also kanjis being west-east whose both parts are north-south, as in 能 - nou, first time users may be a bit perplexed. The general rule of thumb is that west-east with north-south elements are aligned vertically, while their left part is horizontally misaligned; the opposite holds true for the north-South having both parts as west-east. Significantly smaller elements in the upper part, as in 符 - fu denotes north-south predominance also when there's substantial alignment both horizontally and vertically. Moreover, in those cases, the existence of well known kanjis as 竹 - take and 付 - tsuki determines that kanji, and similar one, to be just take over tsuki. The advice to the occasional user is to search both possibilities, giving priority to the choice that would seem more fitting.

Some kanjis originally meant and visually arranged as north-south are nowdays drawn with a signle stroke joining the top and the bottom part. In example, 果 - ka and the east part of 裸 - hadaka are a "field" above a "tree", but the vertical middle line of the field is lenghtened to form the tree trunk. This kanjis are considered north/south as long as their visual arrangement of their elements is the same as the original ones taken alone (in this case, 木 - ki and 田 - ta). So, if some of the shapes are changed or visually re-arranged, the connecting line forms a whole. In example, the east part of 様 - sama is actually a "horns" radical above a "water" one, with the middle line of the horns being prolonged into the middle line of 水 - mizu. But the water element changes it's shape; it's shaped as the extension of the vertical line surrounded by four dots, while the water kanji 水 has an hook on the left. As the water on the bottom part it's only approximated, and the stroke is visibly a single vertical one melted with the upper element, the part is considered a whole.

Roadside kanjis are associated with the following rules:

  • In kanjis having both a cap and a roadside element, the roadside element always leads. So, even if graphically 定 - tei is a roadside below a cap, it is classified as a roadside containing a cap.
  • The roadside element never extends to the touching elements. In example, 起 - ki is a roadside containing a west-east element, whose left part has five strokes. The left part of roadside kanjis that, as 起 - ki, is composed of a cross and a lower extension are considered north-south, the north part being the three strokes cross with ground deriving from 土 - tsuchi. Also, the inner part of roadsides elements having a topmost stroke jointed with a lower turned T stroke as in 定 - tei is treated as a whole. This two rules that may seem a bit counter-intuitive to first time users, but they will become immediately natural as soon as the stroke order is understood by writing the elements under the common writing rules.

The above rule makes kanjis as 題 - dai to be considered roadside, west-east, north-south, cap and north-south. The left part of the inner west-east is just a north-south, having a T with appendix below a 日 - nichi radical; the right part is a cap above a north south composed of a 貝 - kai radical; the radical itself is formed of an upper 5 strokes part (目) and two dot-like strokes in the south element.

The order in which the roadside strokes are drawn in the whole kanji varies among different calligraphy schools. Some draws them before the rest of the kanji, some prefer to draw it last. However, this distinction is not meaningful in RE.C.ORD, as the pivotal element is considered always first, and the drawing order of the other parts of the kanji is not taken into account.

Embracing category is possibly the most various, although not very numerous. It includes all the kanjis that are embracing a content on two sides. The embracing may be constituted by a single hook sign on the top-right as in the lower part of 気 - ki, or it may be a T like hand radical as in 右 - migi and 左 - hidari. It may be also a single hook sign on the low-bottom part of 置 - chi, or it may be partially enclosing on three sides, as in 句 - ku. The two angles from which the embracer guards its content determine the subclass, being ordered clockwise; all the embracers in a class comes before the ones in the next.

  • Class A: left-top embracers as 広 - kou.
  • Class B: right-top embracers as 可 - kou.
  • Class C: right-bottom embracers; currently there isn't any instance of them being considered.
  • Class D: left-bottom embracers as 県 - kou.

The spear radical, 戈 - ka often acts as embracing on the top-right corner (class B). In example, in 戒 - kai, the spear is encircling the "H" like internal part. In this case, we have an encircler composed of 4 strokes, horizontal, oblique, oblique and dot-like.

The spear radical is often used together with a semi-vertical stroke to form up three side encircling, as in the top part of 感 - kan (the encircling nature of this complex element is confirmed also by the writing order, which starts with the outer elements and fills the content afterwards). In the case of 成 - sei, both the graphical look and the drawing order of the elements suggests that the whole kanji is to be considered as an encircler with empty content, as for 門 - mon.

The first stroke of the spear radical is sometimes bound to an upper content, as in 織 - shoku and 職 - shoku (homophones with different meanings). Both the drawing orders and the graphical look of the right part of these kanji suggests taking the 立 - ritsu element above the horizontal stroke of the spear as "melted" in the encircler, and so, to consider it part of the encircling itself.

This principles of "extending" the encircler applies also to other elements as 尾 - bi, where the embracer is the lower part of the door radical 戸 - ko, but notice that 民 - min is considered as a whole as it's shape requires all the strokes to be attached and contained in themselves. For the same reason, 氏 shi is never an embracer.

However, the principle does not apply (normally) when the extension is inside the embracer; in example, 局 - kyoku is a A class three stroke encircler containing a B class one stroke (hook) embracer.

Finally, encircling kanjis are the ones surrounding a content on three or four sides. The 口 - kuchi radical is not considered an encircler, but a whole, as oppositely to embraces, four-way encircles cannot be empty.

As Embracers, encircles are subdivided in classes, ordered progressively in a clockwise fashion; they are namely:

  • Class A: Opening on the right, like 区 - ku.
  • Class B: Opening on the bottom, like 門 - mon or 風 - kaze.
  • Class C: Opening on the left. Currently, no one of them is present in the table.
  • Class D: Opening on the top, like 画 - ga or 鼎 - kanae.
  • Class E: Complete encirclers, as 囚 - shuu.

A kanji being worth be mentioned is 我 - ware. It is an encircling in class B, with empty content, whose pivot (the encircler) is organized east-west. So it precedes 門 - mon and all the kanjis based on this encircler, as mon is a west-east encircler too, but has more strokes. Similarly, 鼎 - kanae has an encircler in class D, so, even if it has less traits than 門 - mon, it comes after.

Complete encirclers (class E) cannot be extended; in example, 鹵 - ro, which has a two traits appendix above a square, is considered a north south, the northen part being just the two traits above the encircler.

The RE.C.ORD Table

RE.C.ORD system has been applied to a selected subset of kanjis to provide a useful table both for the Japanese language student and the translator.

The table contains:

  1. The 1945 kanjis officially taught in Japanese schools, commonly known as Jouyou.
  2. The extra 51 kanjis added by Patrick Geoffrey O'Neill as "essential kanjis".
  3. The most common among the "non common" kanjis listed by Kenneth G. Henshall in the appendix of "A Guide to Remembering Japanese Characters" .
  4. A short list of selected kanjis from literature sources, still very common in nowdays poetry, prose, song lyrics and mangas.

The total count of characters in this version of the table sums up to 2176.

The choice has fallen on this subset as it contains all the kanjis a student or a professional is likely to encounter in nowdays Japanese live culture. The 1945 Jouyou kanjis are the official list that the Education Ministry has related in the late '40s, that are officially taught to every Japanese in the ordinary school course (in about 12 years). O'Neill's "essential kanjis", if not actually essential, are extremely common; they have very high frequency ratings. The "Non common characters" listed by Henshall are not very common, except for some of them, but they are relevant both because they are likely to be found in literature (also contemporary), poetry, songs or otherwise because they have an high didactic value, as they are usually roots for more complex, but also more known, kanjis. Knowing them, or just having them for reference, helps to understand better the kanji listing.

Finally, some kanjis that are almost obsessively used by contemporary writers, and that are reported even without reading indication (meaning that the publisher expects the reader to know them very well), were left out of the cited lists, so they have been added here. They are namely 寐 - bi, 頬 - hou, 拭 - shoku, 馳 - chi and 崖 - gake.

The table lists the RE.C.ORD ID of a given Kanji, it's writing and it's name. Mind the fact that kanji names may not correspond with the main pronunciation of the kanji, or even with one of them. You may consider it as occidental "letter names". The name of the "L" letter spells "el", but there's no employ of "L" letter in which its name comes to play an active role.

Associated meanings and kana readings are given next. Katakana readings are ON readings, while hiragana are KUN readings. Both the readings and the meanings columns are by no mean complete, nor punctual. Some of the meanings may be associated with just some of the readings, and there may be alternative readings for irregular Kanji groups or person/place names that are not listed in the column. They are meant only for a fast reference directly while scanning the list of kanjis.

Stroke count is given in the STK column. While having just a marginal role in RE.C.ORD, it HAS some role, and it is still very important to find the kanji in other ordering systems.

ASCII-JIS code and UNICODE codes, both in hexadecimal format, are given as they are direct keys for many on-line and computer based lookup systems. I.e, by entering the Unicode entry on the character table of a Unicode enabled font, it is possible to copy and then paste the character.

The O'Neill index, where available, is given next, so that having the "Essential kanji" pocket guide at hand, a more complete reference to the kanji can be immediately retrieved.

Last, the entry index in the "The New Nelson Japanese-English Character Dictionary" is given. This is not what goes under the common name of "Nelson Index".

Data to complete the given fields (except for RE.C.ORD ID, of course) have been extracted from KanjiDic by Jim Breen, to which I send all my thanks and respect for the awesome work, without which nothing of this would have been possible.

Actually, without Prof. Breen's work, I would have probably find extremely more difficult to study the Japanese language.

The table is actually a long, uninterrupted list of flowing by kanjis. However, for user convenience, it has been divided into sections indicating the class of the included kanjis. Subsections have been created at approximately 20 to 100 kanji intervals, where possible, indicating the class of the pivotal element and the size, in strokes, of the pivot.

The on-line version of the table contains also an hyperlink to an external site: is a bit flashy site, but it has possibly the greatest online interface to kanjidic ever.

PDF version

A PDF version of the table, printer and eye friendly, is available here.


The algorithm presented as RE.C.ORD, as presented in this document, and the order deriving from the application of the algorithm, that can be expressed as a pair of an integer number and a Japanese Kanji characters, are both intellectual property of Giancarlo Niccolai.

You are granted the right to use any material described in this document for non-commercial purposes to any extent, provided that you cite the original author and you provide a reference to the original source; that is, this document.

You are granted the right to modify the contents of this document as per your own will, and to distribute your modified version, provided that you still cite the author and the original source; moreover, in case you modify this document and you distribute your modified copy, you must indicate what modifies you have brought.

You can use the contents of this document and the RE.C.ORD ordering system for commercial purposes under the written permission of the author, that can be contacted at the site

The table

The complete online table is here