There are concepts in Buddhism that open like a door onto a vast courtyard, and there are concepts that work like a needle. 方分 is a needle. It is one short word, but in the Yogācāra (唯识) tradition it is the precise instrument used to take apart one specific claim — the claim that there is a *single, real, indivisible unit* at the bottom of material experience. Once you understand 方分, you have a key for any situation in life where you, or someone around you, insists that some "atomic" thing — a self, a moment, a decision, a fact, an identity — is genuinely one.
这个概念在说什么
The literal meaning. 方分 is two Sanskrit words fused into one technical term. *Dig* means direction; *bhāga* means part, share, portion. So 方分 literally names "the parts a thing has in its directions" — and in classical Indian cosmology, "directions" means exactly six of them: east, south, west, north, above, below. 方分 is therefore the inner directional structure of a piece of matter (色法, rūpa): the parts of it that face one way being distinguishable from the parts that face another way.
The diagnostic role. This is the part most readers skip and shouldn't. 方分 is *not* a category of objects and not a piece of physics on its own. It is a probe. The 《唯识二十论》(Viṃśatikā-vijñaptimātratā, "Twenty Stanzas on Consciousness-Only"), attributed to 世亲 (Vasubandhu), does not introduce 方分 in order to describe the world. It introduces 方分 in order to ask a single yes/no question of one specific claim: that there exists a *single-real atom* (一实极微) — an atomic unit of matter that is genuinely *one* (一) and genuinely *real* (实) and genuinely the smallest indivisible substrate of everything material.
The 论 then sets up what logicians today call a *dilemma* — two horns, both uncomfortable:
- If the atom (极微, paramāṇu) has 方分 (directional parts), it is made of multiple parts, so it is not one. - If the atom has no 方分, then shadow (影), obstruction (障), and the visible size of all material bodies (聚色) become inexplicable — so we cannot get a coherent material world out of it either.
Both horns land on the same conclusion, written in the 论 itself: 「一实极微不成」 — the *single-real-atom* picture is not establishable.
The sub-terms, kept distinct. The 《唯识二十论》 does not have one word for this idea — it has four, used almost interchangeably but with slight differences of emphasis. Preserving these is important, because the argument's force comes from pressing on the same point from slightly different angles:
- 方分 — the umbrella term: directional parts. - 六方分异 — "the six directional parts differ from each other"; emphasizes the difference *between* the parts facing different directions. - 异方分 — abbreviated form of the same thing: "parts in different directions." - 多分为体 — "having multiple parts as its body"; emphasizes that the whole is *constituted by* the parts, not decorated by them.
The 论 uses all four in rotation. They are not four different concepts. They are four ways of saying: what looked like one thing is, on inspection, six things wearing one label.
The supporting vocabulary. Six related terms show up in the argument and need to be precise:
- 极微 (paramāṇu) — the putative *smallest material unit*. The 论's subject. - 聚色 — *aggregate material form*: the visible, tangible bodies we actually encounter. A table is 聚色; the hypothetical atom is 极微. - 六方 — *six directions*: east, south, west, north, above, below. The full spatial frame. - 影 (*shadow*) — what appears on the side of an object light doesn't reach. Used to test the "no 方分" horn. - 障 (*obstruction*) — the way one piece of matter prevents another from occupying the same place. Also tests "no 方分." - 一实极微 — *single-real-atom*; the metaphysical claim the whole argument is aimed at. Not "atoms exist," but "there is an atomic unit that is genuinely one and genuinely real."
The argument, walked in five moves.
*Move 1 — the 合 (co-joining) challenge.* Place one 极微 in the middle. On each of its six faces, suppose it touches one neighboring 极微. At the moment of touching, the two atoms occupy the same place. One place cannot hold two distinct things. So if the central atom is genuinely *one*, it must itself contain six distinguishable parts — one for each face — so it can be "where" each neighbor is. Hence the first verse the 论 produces: 「极微与六合，一应成六分」 — "When one atom meets six, it should have six parts."
The defender of the atomic view, here represented by the Kashmiri Vaibhāṣika masters (迦湿弥罗国毗婆沙师), answers: atoms have *no* 方分, only aggregates do. That is why atoms don't co-join and aggregates do.
*Move 2 — why the Vaibhāṣika answer collapses.* The 论 now notices a quiet move in that reply. The defenders treat 聚色 (aggregate form) as a real, separate thing that *can* have 方分, distinct from the atom. But on the atomic theory there is no separate 聚色 — there are only atoms in arrangements. You cannot borrow "the aggregate" to do work the atom can't do, while also claiming aggregates are nothing but atomic arrangements. If you say atoms don't co-join *because they lack 方分*, and then you also say aggregates don't co-join, you have given an explanation that explains everything and therefore nothing. Hence the second verse: 「极微无合，不由无方分」 — "The atom's non-co-joining is *not* because it lacks 方分."
*Move 3 — the "有方分" horn.* Suppose you accept that the atom *does* have 方分. Then it has six directional parts, those parts are different from each other (六方分异), and the atom is therefore a composite — "having multiple parts as its body" (多分为体). A composite cannot be the single, indivisible real entity the atomic theory requires. Hence: 「极微有方分，理不应成一」 — "If the atom has 方分, in principle it cannot be one."
*Move 4 — the "无方分" horn.* Now suppose you bite the bullet and say the atom has *no* 方分 at all — no inner directional structure. The 论 now turns to shadow and obstruction. When sunlight falls on one face of an atom, a shadow appears on the opposite face. For shadow to appear, the light has to be blocked by some *part* of the atom that the light doesn't reach. If the atom has no parts, there is nothing for the light to "not reach," so shadow becomes inexplicable. Likewise, if atom A obstructs atom B from occupying the same place, A must have a part that B cannot enter. With no parts, there is no obstruction. And if atoms don't obstruct each other, then any number of them can sit in the same place — meaning a visible 聚色 should be no larger than a single atom, and the entire material world should be invisible. Hence: 「诸极微便不相碍，可以展转同处」 — "If atoms have no 方分, they don't obstruct each other, and any number can occupy the same place."
*Move 5 — relocating shadow and obstruction to the aggregate.* The defender's last move: shadow and obstruction are properties of 聚色, not of atoms. The 论 replies: you can't have it both ways. You rejected a 聚色 that exists apart from atoms. So shadow-and-obstruction cannot be offloaded to a non-existent second entity. Atoms are distinguished from aggregates only by *arrangement* (安布), not by a difference in kind. Both are composites of parts; neither is the single, real unit the theory needs.
The conclusion is therefore not "atoms don't exist" — it is the much more interesting claim 「一实极微不成」: the *single-real-atom* picture of matter is incoherent. Both horns point at the same target, namely the conviction that there is a *one, real, indivisible unit* underneath everything we can point at. 方分 is the tool that exposes this conviction as unearned.
生活里的走查
The hardest step for a contemporary reader is to map a six-direction metaphysical debate onto a Tuesday afternoon. Let me try to do it carefully with one concrete scene, then briefly mark where the same map applies elsewhere. The scene I'll use is one almost everyone knows: the moment you pick up your phone and decide whether to send a particular message — to a partner, an ex, a parent, a boss, a friend.
Step 0 — the unspoken starting claim. Before you even tap the screen, you are already holding a metaphysical commitment. You are treating this decision as *a single thing* belonging to *a single self*. "I'm going to send it." "I'm not going to send it." One decision, one me, one atom of agency. This is the contemporary equivalent of 「一实极微」 — the claim that there is a *single, real, indivisible unit* (one decision, one self) at the bottom of this moment. 方分 is the question you now put to that claim.
Step 1 — the 合 challenge. Ask: if this decision is genuinely *one*, why does it have six faces? Press the 合 test — does it touch six other things? It does:
- it touches *what just happened* (the face pointing backward, the memory side), - it touches *what might happen next* (the face pointing forward, the anticipation side), - it touches *the other person* (the face pointing outward, the social side), - it touches *your body* (the face pointing inward, the breath-and-heartbeat side), - it touches *your idea of who you are* (the face pointing up, the identity side — "what kind of person am I"), - it touches *what you won't let yourself admit* (the face pointing down, the shadowed side).
If the decision is genuinely one thing occupying one place, how does it sit in all six at once? The 论'sfirst answer applies cleanly: 「极微与六合，一应成六分」 — if it meets all six, it must itself have six parts. Already the "one decision" has become six distinguishable parts. That is the first horn rising.
Step 2 — the Vaibhāṣika escape. The mind instinctively tries the Vaibhāṣika reply: "Sure, the *decision* has no parts — it's just one. The *whole situation* has parts." You split the experience into "the small unit (decision)" and "the aggregate (situation)." The 论'sresponse is sharp and exact: you cannot borrow "the situation" to do the work the decision can't do, while also claiming the situation is nothing more than a bundle of decisions. Once you press on it, the situation has no separate existence — it is just this decision, viewed from six sides. 「极微无合，不由无方分」 — the decision's apparent unity is not because it lacks parts. It looks unified because you have stopped looking.
Step 3 — accept the parts. Now suppose you concede: yes, the decision has six faces; yes, it has 方分. Then it is a composite, 六方分异, 多分为体 — six different directional parts making up what looked like one whole. And a composite cannot be the single, indivisible real thing you wanted it to be. There is no "decider" behind the decision. There is no small, irreducible *me* who owns it. There is the decision, and it is itself a knot of six directions pulling different ways.
This is the first horn's full landing: 「极微有方分，理不应成一」 — once you see the parts, the one-ness was never real.
Step 4 — deny the parts. Now suppose you refuse. You insist: no, this really is one decision, single, indivisible. Then the 影 (shadow) test: if it is genuinely one with no inner structure, why does it cast a shadow? Why does it have consequences it didn't "reach" — a recipient's reaction, a memory it altered, a future it foreclosed? Shadow requires a part of the object the light doesn't touch. If there are no parts, there is nothing for consequences to "not reach," and consequences become inexplicable. The very fact that this moment has a shadow is evidence of its 方分.
And the 障 (obstruction) test: this decision blocks other decisions. While you are deciding *this*, you cannot decide something else with the same attention. Obstruction requires that the object have parts the other object can't enter. No parts, no obstruction. If your decision had no internal directional structure, nothing could get in its way — and a decision that nothing can obstruct is, by the 论's logic, no larger than a point, invisible, unworkable. 「诸极微便不相碍，可以展转同处」 — if the decision has no 方分, all your possible actions collapse into one indistinguishable point.
Step 5 — relocating shadow and obstruction. The last mental escape: "It's not the decision that has a shadow; it's the whole situation that has a shadow." The 论'sreply: you cannot have it both ways. You rejected a separate "situation" distinct from the moment-to-moment decisions. Shadow and obstruction are real, and they need parts to be real. If they appear, the parts are there, even if you don't want to see them. The decision is not one thing in six places; it is six things wearing the label "one decision."
Where the same map runs in other corners of contemporary life. The exact same five moves apply wherever we treat something as a single, real, indivisible unit:
- *A "single self" at work vs. at home vs. in intimacy.* The "self" has 方分: the part that faces colleagues, the part that faces family, the part that faces the lover, the part that faces the mirror. Either it has parts (and so is not one), or it has no parts (and so cannot cast the different "shadows" — work-shadow, family-shadow, intimacy-shadow — that obviously do appear). - *A "single moment" of scrolling.* The moment has six faces: input (what you just read), output (what you'll do next), attention (what it grabbed), body (your shoulders, your breath), social (who you're ignoring), identity (who you're being while scrolling). No parts, no shadow — yet a moment without shadow has no consequences, which is empirically false. - *A "single fact" in an argument.* Once you treat it as one, asking how it touches the other six things in the conversation immediately shows it has parts. The 论's structure is, among other things, a methodological lesson: don't interrogate content first; interrogate the *claim of atomicity* first.
The most common mis-mapping to watch out for. The contemporary reader's habitual mistake is to slide 方分 into "size" or "spatial extent" — to say, "the atom has parts, so it's not the smallest." That misses what the 论 is doing. 方分 is not a measurement argument. It is a *unity* argument. The 论 is not saying "atoms are bigger than we thought." It is saying "atoms are *less one* than we thought." The mistake in daily life is to treat a small enough thing as automatically a single thing. 方分 says: smallness is not oneness.
当代人为什么需要它
Three converging pressures make this ancient needle unexpectedly sharp right now.
1. The "atomic unit" reflex is the default setting of contemporary attention. Modern life is organized around tiny, supposedly indivisible units — a "decision," a "moment," a "self," a "fact," a "task," a "tweet," a "token" in a language model, a "post," a "feeling," a "microaggression," a "data point." The reflex is to take whichever of these is in front of us and treat it as a *one-real-thing* belonging to *one real owner*. The reflex is so automatic that the cost of following it is invisible — until something breaks (a marriage, a project, a decision we can't explain even to ourselves). 方分 is the discipline of *not* accepting the atomic reflex for at least one beat longer than usual, to see whether the unit in question holds together or quietly unfolds into six directions.
2. Modern physics has, in its own language, been saying something very close to 方分 for a century. The classical atom turned out to have a nucleus and an electron cloud. The nucleus turned out to be made of quarks. The quarks turned out to be excitations of a quantum field with no clear "edges" in the classical sense. Solid matter turned out to be mostly empty space; the "smallest unit" of chemistry turned out not to be a unit of physics. None of this *proves* the Yogācāra claim — the Buddhist claim is ontological, not physical, and conflating the two is a category error — but the convergence in spirit is striking. Both traditions, on their own terms, are saying: the thing you are pointing at is not as single as your pointing assumes. At minimum, 方分 inoculates the contemporary reader against the naive metaphysics that the word "atom" smuggles in.
3. The contemporary self is the place this matters most. When someone says "I decided to do it," "I am sad," "I am fine," "this is my boundary," they are, philosophically, making a 一实 claim — there is a single, real me, and this property is *of* that me. The reflex is so deep that the very *feeling* of being a person is the feeling of being a 极微. 方分 gives the contemporary reader a way to question that feeling without losing it. It does not say "you don't exist." It says "the one-ness you are pointing at, when you say 'I,' has parts you have been calling the whole." That is a much more useful diagnosis than either flat denial (which produces dissociation) or flat acceptance (which produces rigidity). It is the middle path that the very structure of the 《唯识二十论》 is pointing toward: don't inflate the parts into a real single owner; don't deny the parts to keep the owner; *see the structure* by which "one" was assembled.
常见误读与澄清
Misreading 1: "方分 is just a fancy word for 'size' or 'spatial extension.'" No. 方分 is a question about *parts and their differences*, not about *how big something is*. Something small can still be 一 (one) if it really has no inner structure; the argument is that it cannot both have no inner structure and do the work that real material things do (cast shadows, obstruct each other, take up the space of a body). The 论 is operating on the category of *unity*, not the category of *size*.
Misreading 2: "The 论 is denying that atoms exist." No. The 论 is denying that *a single-real atom* can be established. This is a logical-ontological point, not an empirical one. Even classical Indian physics, which the 论 does not dispute at the level of "atoms do their work in experience," is being shown to lack a foundation in a *single-real-indivisible-unit*. Atoms may be postulated; a *一实* atom cannot.
Misreading 3: "Modern physics has proven Buddhism right about 方分." No. Modern physics has shown that classical atoms have internal structure; it has not shown, and cannot show, the specifically Yogācāra thesis that the *category* of a single-real-indivisible-material-unit is incoherent. The spirit is similar; the metaphysical commitments are not the same. Conflating them is sloppy on both sides.
Misreading 4: "方分 proves that 'I' don't exist." Also no. 方分 shows that the *one-ness* of "I" cannot be cashed out as a single, real, indivisible unit. It does not deny experience, agency, or continuity. It denies only the ontological inflation of those into a *一实* owner. This is closer to the Buddhist middle path than to either materialist denial or naïve reification.
Misreading 5: "方分 is just a logical game, irrelevant to practice." This is the most important misreading to clear. The 论'sstructure — *don't accept the atomic reflex; let the unit unfold into its six directions; notice that shadow and obstruction require parts; notice that relocating the parts to a separate "aggregate" doesn't work* — is *exactly* the structure used in any deep moment of contemporary life: a "single decision" that turns out to be six, a "single self" that turns out to be six, a "single fact" that turns out to be six. The 论 is a logic *and* a practice. To work through a real decision using 方分 is not to intellectualize the decision; it is to stop pretending the decision is one thing when it is not. That is the beginning of freedom from being driven by the apparent oneness of the next thing that comes through the door.
A short coda. The《唯识二十论》 does something subtle at the end. It does not give you a new doctrine. It does not say, "and therefore the real unit is X." It simply shows you that *X* — whether X is the atom, the self, the moment, the fact — was never the single, real thing you took it to be. The unit that survived your questioning was the questioning itself. 方分 is not a description of the world; it is a refusal to inherit the world's claim that its smallest pieces are single. Once that refusal becomes habitual, the world doesn't go away. It just stops insisting that any of its pieces are more solid than they actually are.