Chisel and Emptiness: An Interface Analysis between Self-as-an-End and Madhyamaka / the Sanlun School
凿与空：Self-as-an-End 与中观／三论宗的接口分析

0. Introduction

This is the third paper in the Self-as-an-End (hereafter SAE) Buddhist Interface Series. The first two engaged Yogācāra and Chan: the Yogācāra paper treated the ālaya-vijñāna / seed system and the fine-grained construction of the individual mind; the Chan paper treated non-abiding, the breaking of abidable constructs, and the leap within the maturation of the self. This paper engages Madhyamaka and its systematized Chinese form — the Sanlun school, and especially Jizang's system of doctrinal classification.

The nature of the series must be stated at the outset, because it governs how every correspondence below is to be read. What the series does is interface analysis, not translation and not absorption. The task of an interface analysis is to mark, between two frameworks, both where they correspond and where they cannot be translated into one another — the two with equal weight. To state the correspondences in full while concealing the breaks is to let the interface quietly become absorption; to mark only the breaks and refuse the correspondences is to refuse analysis. This paper must especially hold the former line — because the depth of correspondence between Madhyamaka and SAE produces, almost imperceptibly, a pull toward reading Madhyamaka as an "unfinished SAE," and that is precisely what an interface analysis must resist. SAE does not take Madhyamaka to be an early version of itself, nor does this paper take Madhyamaka to be a framework awaiting completion; the two are each complete, with different aims, meeting in some structures and parting in others.

The series' coordinates: from two faces to three

The series reads different Buddhist traditions as high-resolution probes set at different positions in the evolution of mind. By this paper, that coordinate system needs an extension. The first two papers used two faces: Yogācāra illuminates the construct — how the thing built takes fine-grained form at a particular position; Chan illuminates the chisel — the negating act that pares away, that breaks the abidable construct. This paper introduces a third face: Madhyamaka illuminates the law — that no construct and no chiseling can be fixed as self-natured.

The nature of this third face must be made clear at once, or it will be misread as a higher position. It is not. Construct and chisel each fall at some position in the evolution of mind (Yogācāra illuminates the construct at the level of the self's fine-grained building; Chan illuminates the chisel at the seam of the leap in the self's maturation); whereas the law is not another position but a constraint that holds across all positions — at whatever layer, however finely constructed, however thoroughly chiseled, that construct cannot be fixed, without remainder, as something self-existent, self-sufficient, and independent of conditions. In other words, "construct-position," "chisel-seam," and "law" name three faces, not three positions side by side: the construct is the thing built, the chisel is the paring act, the law is the constraint that no position closes. Madhyamaka operates on the layer at which the self grasps itself as an end (why this layer is Madhyamaka's point of application is detailed in the next section), and what it illuminates there is this cross-positional law.

This extension does not revise the first two papers. Yogācāra illuminates the construct, Chan the chisel — both still stand; this paper only advances the series' coordinate system from two faces to three, giving, for the first time, its own probe to the constraint — "no construct and no chiseling can be fixed" — that has been operating in the background all along.

This paper's subject term and point of application

This paper's subject term is "emptiness," and what it engages is Madhyamaka's negation of "self-nature" (svabhāva). But a boundary must be set here, and it runs through the whole paper: on the Buddhist side, what Madhyamaka negates is the grasping of self-nature in all dharmas — emptiness is the universal absence of self-nature in all things; what this paper takes is only the high-resolution interface of that negation at the layer where the self grasps itself as an end. The full range of emptiness belongs to Madhyamaka; this paper's lens illuminates only one segment of it. This boundary is not a modest disclaimer but a necessary statement against narrowing Madhyamaka — wherever the text below speaks of "emptiness operating at some layer," it is to be read under this boundary: that is the point of application SAE has selected, not the whole range of Madhyamaka's emptiness.

Why is the point of application the layer at which the self grasps itself as an end? Because the self-nature-grasping that Madhyamaka negates is, in the sequence of mind's evolution, most acutely condensed here. In the SAE sequence, after self-consciousness emerges comes the layer of "meaning" — where the subject begins to seek reasons for its own actions and takes itself as an end in itself (14DD, the layer of being-toward-death). [4] Every grasping at this layer — seeking a fruit, seeking a station, positing a self-sufficient and complete "I" — is, in essence, the building of "the self" into something self-natured: self-arising, independent, unchanging. What the eightfold negation and the four levels of the two truths take apart, blade by blade and layer by layer, is precisely this grasping that fixes the self as a self-natured end. The point must be stated precisely: what Madhyamaka does is negate — it deflects the grasping that fixes the self as self-natured; and once the grasping loosens under that negation, the subject need no longer stake all its force on guarding this self-natured "I," and so the possibility opens up of turning toward a higher position — acknowledging that the other, too, is an end in itself (15DD, the ethical layer). [4] What this paper tracks is precisely how this negation of Madhyamaka's opens up, within the SAE coordinate system, the possibility of that turn: the through-line is "the possibility the negation opens up," not "negation for the sake of the turn" — the turn is the room left over once grasping loosens, not a goal carried by Madhyamaka's negation. Within its own soteriology, Madhyamaka's negation aims at non-attainment, not at this SAE turn; "opening the possibility of turning toward the other" is a downstream effect read off in the SAE coordinate system, not the intent of Madhyamaka's negation.

The boundary must be restated, because it slips most easily right here: to say that Madhyamaka "operates on the layer at which the self grasps itself as an end" is the point of application SAE has selected for interface analysis, not a claim that Madhyamaka in its own sense is a psychotherapy of the self's ends. Madhyamaka is a total negation of self-nature-grasping in all dharmas; this paper takes only its interface at this layer. When the eightfold negation is discussed below as "total negation," this will be returned to.

The type of probe

In the previous paper, Chan's probe was mixed: it had both a structural-interface face and an engineering-operation face — Chan, with a blow or a shout, a single device in a single situation, directly precipitates the leap; the freedom of the chisel is very high, the precision of the construct very low. [4] Madhyamaka sits at the opposite end of that axis: it approaches the same non-fixable point through the exhaustive enumeration of the eightfold negation, the logic of the tetralemma, and the layer-by-layer advance of the two truths — the freedom of the chisel low, the precision of the construct high. So this paper's probe is "interfacing through the structure of argument, turning apophatic at the terminus" — the form of the argument (the eightfold negation, the tetralemma, the logical architecture of the two truths) can be matched step by step with SAE's structure, but at the terminus (non-attainment, the emptiness of emptiness, words forgotten and thought cut off) nothing can be stated positively any longer; it can only be marked by negation. This mirrors the Chan paper exactly: Chan turns apophatic at the leap, Madhyamaka turns apophatic where the argument exhausts itself.

Texts and citation

This paper's textual spine is the Three Treatises (the Mūlamadhyamakakārikā / Zhonglun, the Dvādaśanikāya-śāstra, the Śata-śāstra) and Jizang's writings (the Dasheng xuanlun, the Erdi yi, the Sanlun xuanyi, the Zhongguanlun shu, the Fahua youyi), in Kumārajīva's translation system. The Zhaolun is brought in as a prehistory-interface to Kumārajīva's translation circle and the early Chinese Madhyamaka milieu, not as an internal criterion of the Sanlun school. Nāgārjuna's Vigrahavyāvartanī serves as a side-witness in the reflexive section, outside the spine. Modern Madhyamaka scholarship (e.g., the relevant svabhāva analyses) is used only to sharpen the accuracy of statements about Madhyamaka, and bears no warrant for any SAE thesis — the SAE-side structure cites only SAE's own published texts.

Below, wherever SAE's structure is cited (the five terms of the chisel-construct cycle, the dimensional sequence, the law at each layer), the reference is to the methodology paper and related applications, marked [4] etc.; wherever Madhyamaka and Sanlun texts are cited, their sources are marked, with line-references to be finalized against the canon.

---

1. Anchoring emptiness

To place "emptiness" into the SAE coordinate system, the first thing is to determine where it falls. The answer is not a point but a line — within the SAE coordinate system, emptiness runs vertically through three layers.

Emptiness as thesis corresponds to a core SAE claim: no construct can close upon itself; every act of construction leaves a remainder not gathered in. [4] When Madhyamaka says "all dharmas are without self-nature," it says that no dharma is self-arising, self-sufficient, independent of conditions; when SAE says "no construct closes," it says that no construct can seal itself off without remainder. The two meet at the conclusion — one "without self-nature," one "non-closing."

Emptiness as operation corresponds to SAE's negating method itself — the act of paring away, breaking, interrogating the points of incompleteness (the chisel). [4] Madhyamaka does not treat "emptiness" as an object to be reached, but takes negation as the operation running throughout: negating self-nature, negating grasping, negating every abidable view. This is isomorphic with the working of the chisel: the chisel posits no terminal construct; it only negates the present construct, again and again, forcing out its remainder.

Emptiness as terminal signature corresponds to that SAE limit which cannot be stated positively and can only be approached by negation — the position of the thing-in-itself. [4] "The emptiness of emptiness," "non-attainment," "words forgotten and thought cut off" — none of these describes an attained state; each is an apophatic mark left where the argument runs to its end. In SAE, this limit can only be trusted, not known; it appears only in the form of a double negation and cannot be positively constructed. [4] Madhyamaka's non-attainment falls precisely at this position.

So "emptiness" in the SAE coordinate system is not an isolated point but a line running from the thesis "no construct closes," through the operation "negation," to the terminal signature "non-attainment."

The three marks of self-nature, and one boundary that must be made clear

For the interface to be precise, the content of "self-nature" must be unpacked. The self-nature (svabhāva) that Madhyamaka negates classically has three marks: not arising from conditions (self-arising), without change (permanent), not dependent on another for its being (independent). The negation of all three SAE accepts wholesale — no construct is self-arising, unchanging, or independent.

But here is one boundary that must be made clear, or an SAE concept will be forced into Madhyamaka's terms. SAE's core concern is whether a construct closes, whether it leaves a remainder — this is a "remainder-perspective" criterion. Svabhāva's second mark, "without change," is a "time-perspective" criterion — it concerns whether a dharma arises and ceases, changes, in time. The two resemble each other but differ at root: SAE asks "does this construct leave an ungathered residue," svabhāva asks "does this dharma change." So when SAE operationalizes "without self-nature" as "no construct closes," it travels its own "remainder-perspective" branch, not svabhāva's "time-perspective" branch of "without change." This must be written out: the three marks (self-arising, without change, independent) are on the Madhyamaka side; SAE reading "closure / remainderlessness" as its own version of "self-nature" is an interface operation, not a claim that svabhāva's original sense was "remainderlessness." Holding this boundary neither distorts Madhyamaka nor lets SAE's main beam (emptiness = no construct closes) lose its footing.

Though the two meet at the conclusion "without self-nature / non-closing," the engines of derivation differ. Madhyamaka derives from dependence: dharmas all arise dependent on conditions, hence without self-nature. SAE derives from occlusion: every construct occludes something, hence must leave a remainder and cannot close. [4] The same conclusion, two engines — this is exactly the "correspondence-plus-boundary" an interface analysis marks: correspondence at the conclusion, boundary at the engine.

A direction, not a terminus

Finally a guardrail must be set, one that recurs throughout. SAE's limit — that non-fixable point — is a direction, not a terminus. The subject cannot penetrate it, only approach its apophatic edge; "keep chiseling" is the discipline, and that limit is the direction the chisel faces, not a terminus that can be occupied. [4]

Strikingly, Madhyamaka runs in the same direction here. Madhyamaka too forbids "arrival" — "non-attainment" says precisely that there is no attained terminus to reach, and that even "emptiness" cannot be grasped as a realized object. The two coincide at the operational layer: both forbid turning that limit into an abidable dwelling.

But this coincidence has its boundary, which must be marked at the same time, or two frameworks of opposite ground will be flattened together. At the operational layer, the two forbid "arrival" in the same direction — this is correspondence. At the ontological layer, the reasons they forbid it are opposite: SAE forbids arrival because that limit really exists yet cannot be penetrated — the remainder of the chisel-construct never reaches zero, and the subject approaches infinitely, along a real direction, a pole that is real yet unreachable; Madhyamaka forbids arrival because that pole is, ontologically, simply empty — a deflected shadow of all proliferation. One is "the road has no end, yet the direction is real"; the other is "both the road and the destination are conventional designation." So this coincidence must not be read as "the two are the same thing at the terminus" — they run in the same direction on "forbidding arrival," and split on the ontological judgment of "whether the thing forbidden is real or empty." Correspondence and break holding at once is exactly the task of an interface analysis here.

This break is not a nuisance for this paper to flatten, but one face of the central untranslatable, to be developed in Section 6. Here its position is marked: the two shake hands at "forbidding arrival" and part at "real or empty."

---

2. Refuting error and revealing what is right, and "establishing"

The banner of the Sanlun school is "refuting error and revealing what is right" (破邪显正). These four characters must first be clarified, because they are most easily read as "refute the wrong, establish the right" — as if "revealing the right" were the setting-up of a correct thesis after the refutation. This is not Sanlun's own sense. Jizang's "right" is not the setting-up of a further correct construct; refuting error is revealing the right — what is called "right" is just the right contemplation of non-attainment itself, the holding that fixes no statement as an abidable place. In SAE's terms: the "right" here is not a terminal construct but the holding of the discipline itself — "no construct can close." Refuting and revealing are not two steps (first refute, then establish) but one thing — refuted to the end, with no abidable "right" set up, that non-setting-up is the right.

This is isomorphic with SAE's chisel: the chisel exists not to set up a terminal construct but is itself the act that negates without cease, that lets no construct be fixed. Sanlun's "refuting error and revealing the right," in the SAE coordinate system, is a thorough carrying-through of the chisel in the dimension of argument — refuting every abidable view to the end, without setting up an abidable right view at the end of the refutation.

The divergence over "establishing"

But it is precisely on this word "establish" that the two have a real divergence, which must be stated precisely.

Sanlun's discipline is refuting without establishing (破而不立). This is not to say Sanlun asserts nothing — it asserts refuting error and revealing the right, asserts the middle way as the substance of the two truths, asserts the various provisional teachings; what it "does not establish" is an abidable right, a terminal view that can be grasped, fixed as self-natured. Sanlun's discipline falls on the act of "setting up an abidable right": to set up an abidable right is to make a new error.

SAE's discipline differs. SAE holds that one should establish — because refutation cannot but produce a new construct. Every negation settles into a new construct bearing a remainder; this is the normal state of the chisel-construct cycle: chisel, and there is sediment; the sediment is a construct. [4] Since refutation necessarily produces a construct, pretending not to establish is itself dishonest; SAE's discipline falls not on "to establish or not" but on the posture of establishing — only an establishing that acknowledges the remainder is a tenable establishing. To set up a construct while acknowledging that it has a remainder, does not close, and will eventually be chiseled by the next blade — such an establishing is sound; to set up a construct while declaring it closed, complete, remainderless — that is what SAE refutes (in SAE's terms, a construct posing as a law). [4]

So the two see the same structural fact — refutation cannot but produce a construct — but their normative responses to that fact are opposite. Sanlun guards the refutation by "claiming no construct as an abidable right": even the teaching itself is only provisional — Jizang states plainly that the two truths are wholly teaching, having nothing to do with object or principle; the teaching is a tool set up to refute grasping, not a sediment forming a new abidable place. SAE guards the establishing by "claiming the construct, but claiming only a construct that acknowledges its remainder": the construct is real, claim it, but in claiming it acknowledge that it does not close.

This divergence is not a matter of who is right, but the appearance, at the disciplinary layer, of the same seam (Section 6 will set this beside its appearances at the ontological and reflexive layers, to see clearly that it is one seam appearing three times). Here it is marked: Sanlun guards "refuting without establishing an abidable right," SAE guards "establishing while claiming its remainder" — both respond to the same fact, "refutation necessarily produces a construct," in opposite directions.

Incidentally, modern Chinese has the idiom "no breaking without establishing" (不破不立), which assumes refutation is for the sake of establishing — first break down the old, then build the new; breaking serves establishing. This is an emergence-intuition, close in direction to SAE's "refutation necessarily produces a construct, the construct bears a remainder." Sanlun's self-description is precisely "refuting without establishing" — refuting not for the sake of an abidable right; the teaching is established but does not abide. The same word "break": the idiom reads it as "breaking for the sake of establishing," Sanlun reads it as "breaking without abiding." This small dislocation between everyday idiom and the school's self-description is itself a miniature of this paper's central tension: the reality of construction, and the thorough non-grasping of construction, meeting here.

---

3. The eightfold negation: total negation of self-nature-grasping

The Zhonglun opens with the eightfold negation: "neither arising nor ceasing, neither permanent nor cut off, neither one nor different, neither coming nor going." Then: "able to expound this dependent arising, [the Buddha] well extinguishes conceptual proliferation." (The last pair, in Kumārajīva's rendering, reads "neither coming nor going-out"; "going-out" here means "going," "departing," and this section's prose, following the sense, also writes "neither coming nor going" — same referent as the original.) These eight "no"s are the core this section interfaces with SAE.

The key must be set first, because here errors are easiest. The eightfold negation is not four blades cutting four different targets. It is a total deflection away from the eight extremes — against the single grasping "dharmas really have self-nature," deflecting from eight directions at once. Pingala's commentary makes the structure plain: stated briefly as eight matters, it is a total negation of all dharmas; with "neither arising nor ceasing" as root, the other six "no"s reduce to it — "if one searches deeply into 'neither permanent nor cut off,' it just is 'neither arising nor ceasing'"; "because there is neither arising nor ceasing, the other six matters likewise." [CBETA, T30, no. 1564, Zhonglun, fasc. 1, "Examination of Conditions," Pingala's commentary, from p. 1c12] That is: the structure of the eightfold negation is "neither arising nor ceasing" as root, with permanent/cut-off, one/different, coming/going reducing to that root from different angles; the eight extremes are the surface enumeration of a total deflection, not eight independent target-domains, still less four blades with four functions.

A guardrail must be set here, and it must be written into the body, not left to a footnote: to read the eightfold negation below as the disassembly of a certain grasping is SAE's interface reading, not a restatement of Pingala's or Jizang's historical doctrinal divisions. The traditional commentaries have their own account of what each pair counters; what this section selects is, under the SAE lens, the eightfold negation as "total negation of self-nature-grasping." The two do not conflict, but this must be marked as an interface re-reading.

Eight extremes cutting into one grasping

In the SAE coordinate system, what the eightfold negation totally negates is the grasping that fixes the self as a self-natured end (established above as Madhyamaka's point of application here). The eight extremes deflect the same "self-natured I" from eight angles:

"Permanent" is "I endure" — fixing the self as something that persists unchanging. "Cut off" is "I am annihilated" — fixing the self as something that is utterly cut off. "One" is "I am identical" — fixing the self as something wholly singular, indivisible. "Different" is "I am sundered" — fixing the self as something utterly divided from its constituents. "Coming/going" is "I as a continuous entity" — fixing the self as an entity coming from somewhere, going to somewhere, continuous across time. "Arising/ceasing" is "I arise and cease self-naturedly" — fixing the self as something that really arises and perishes self-naturedly.

The eight extremes are the surface enumeration of a total deflection; structurally, with "neither arising nor ceasing" as root, permanent, cut-off, one, different, coming, going reduce to that root from different angles; what they deflect is the same "I" fixed as self-natured. This is exactly Pingala's structure of "neither arising nor ceasing as root, the other six reducing": "neither arising nor ceasing" deflects the most fundamental grasping, "self-natured arising, self-natured ceasing," and the other six extremes are that root-grasping unfolded along different dimensions. The eightfold negation is a total negation, not eight equal independent angles.

The relation between "the enumeration of eight extremes" and "one root with six reductions" must be made plain here, lest it be read as eight equal angles: the eight extremes are the surface enumeration of a total deflection from the eight extremes; structurally, with "neither arising nor ceasing" as root, permanent, cut-off, one, different, coming, going reduce to that root from different angles. A reader who takes the eightfold negation as eight blades each striking one spot misses its original sense — "total negation of one grasping."

The blade "neither arising nor ceasing"

Among the eight, "neither arising nor ceasing" must be discussed on its own, because it is the root, and because it shows most force at this paper's point of application.

Traditionally, the range of "neither arising nor ceasing" is all dharmas — no dharma arises self-naturedly or ceases self-naturedly. What this paper takes is its high-resolution application at the point of application: deflecting the grasping "fixing the self as self-natured arising, self-natured ceasing." This must be marked as SAE's lens-selection — narrowing a blade's range onto a high-resolution target, to see clearly how it works at this layer, is not a claim that "neither arising nor ceasing" concerns only the self in its original sense.

At this point of application, "neither arising nor ceasing" deflects the grasping "I really arise self-naturedly, and really perish self-naturedly." To deflect it is not to teach belief that "I do not exist" or "I am eternal and unperishing" — those fall into cut-off, into permanence, which the other two blades of the eightfold negation deflect. To deflect "self-natured arising and ceasing" is to loosen that "I" grasped as self-natured: it neither arises self-naturedly nor ceases self-naturedly; it is dependently arisen, without self-nature, non-fixable.

Under the SAE lens, where this deflection has the most loosening force is here: that layer which grasps the self as a self-natured end stakes all its grasping on guarding a "really existent I"; once the underlying grasping "I really arise and cease self-naturedly" is deflected, the necessity of guarding loosens. The subject need no longer stake all its force on guarding this self-natured I, and so can turn toward a higher position — acknowledging the other, too, as an end in itself. Note the wording here: not "the eightfold negation itself forces one toward the other," but "once the grasping of the self-natured I loosens, the subject need no longer stake all its force on guarding the I, and so can turn toward the other." What the eightfold negation does is deflect that grasping; the turn is a possibility opened up after grasping loosens, not a push the eightfold negation directly applies.

(A doctrinal corroboration may be brought in here: the Dazhidu lun defines the supreme-meaning siddhānta as "past all paths of language, the workings of mind extinguished, resting nowhere, indicating no dharmas," and points to the true reality of dharmas with verses such as "neither arising nor ceasing, the dharma like nirvāṇa." [CBETA, T25, no. 1509, Dazhidu lun, fasc. 1, on the "supreme-meaning siddhānta," line-reference to be finalized] Later tradition may set the eightfold negation alongside this true reality and summarize it as "the eightfold negation is just true reality," but those four characters are a summary, not a continuous original sentence in the treatise. "Resting nowhere, indicating no dharmas" shows precisely the apophatic character of this line — it indicates no positive true reality, only deflects every graspable place. This passage also prepares the apophatic use of the term "true reality" discussed later.)

And the other six extremes

Why the six extremes can "reduce to the root, neither arising nor ceasing" — the transmission must be spoken out, or "reduce" is only a slogan. The mechanism is this: the six extremes — permanent/cut-off, one/different, coming/going — each, to hold at all, covertly presupposes an "I that arises self-naturedly and subsists self-naturedly" as substrate: "permanent" is this I persisting unchanging in time, "cut-off" is its utter annihilation, "one"/"different" are its wholeness or sundering in relation to its constituents, "coming"/"going" are its translation as an entity across time. But once "neither arising nor ceasing" pulls out that root-grasping — establishing that this I neither arises self-naturedly nor perishes self-naturedly — then persistence-or-annihilation in time, identity-or-sundering in relation, coming-or-going in translation all, at the same stroke, lose their substrate: with no self-naturedly arising-and-ceasing "I," there is no such I to persist, to be annihilated, to be identical, to come or go. Pull out the root of self-natured arising-and-ceasing, and the graspings along the other three dimensions disassemble accordingly, with no need to refute each separately. This is the mechanics of "neither arising nor ceasing as root, the other six reducing to it": the six extremes are not six independent targets needing separate deflection, but projections of the same root-grasping along different dimensions — move the root, and the shadows scatter.

The other six extremes below, each deflecting one face, are stated briefly under the transmission of this root.

"Neither permanent nor cut off" deflects the two ends, fixing the self as "enduring" or "annihilated." In the dimension of time, it deflects "I am an unchanging persisting entity" (permanent) and "I am an entity that will be utterly cut off" (cut off) — both fix the self as self-natured. Deflected, what remains is dependent continuity: the self neither abides unchanging nor is annihilated without continuation, but is a dependent continuity — changing without losing its continuation. Pingala speaks of this principle of continuity as "if one, then no condition; if different, then no continuation." [CBETA, T30, no. 1564, Zhonglun, fasc. 1, "Examination of Conditions," Pingala's commentary, from p. 1c12]

"Neither one nor different" deflects fixing the self and its constituents as "absolutely identical" or "absolutely sundered." I and the various things that constitute me are neither wholly one (one) nor utterly two (different); deflecting these two ends presses forth the dependent linkage — the self is a structure linked together out of various dependently-arisen things, yet not fixable as a single self-substance.

"Neither coming nor going" deflects fixing the self as a continuous entity "with a place it comes from, a place it goes to" — as if a self-natured I came from somewhere and went somewhere. To deflect it is to loosen that "I" grasped as continuous across time, self-naturedly persisting.

The six extremes each deflect one face, all disassembling with the pulling-out of the root "neither arising nor ceasing"; the eight extremes together are a total deflection of the same "self-natured I." This is the eightfold negation's interface reading at this paper's point of application — one grasping, eight angles, one total negation.

(Note: wherever this section says the eightfold negation "deflects the grasping of the self," it is all under the boundary set above — this is the point of application SAE has selected; the full range of Madhyamaka's eightfold negation is self-nature-grasping in all dharmas, not the self alone. And the traditional doctrinal divisions have their own account of what each pair counters; this section is an interface re-reading and does not restate those divisions.)

---

4. The four levels of the two truths and the four steps of negation

In the systematization of Madhyamaka in China, Jizang's Dasheng xuanlun is a high point. There the two truths are organized as a layer-by-layer advancing structure: the conventional truth sets up a statement, the ultimate truth negates it; the ultimate truth of the prior layer is demoted to the conventional truth of the next, then negated by a new ultimate truth. In fascicle 1 Jizang unfolds this structure into multiple levels, one strand of which can be clearly read as four steps — and this is what this section interfaces with the chisel-construct cycle.

Two boundaries must first be stated. First, the textual boundary: the Dasheng xuanlun, fascicle 1, divides the meaning of the two truths into ten levels; what this section extracts is one discernible four-step sequence under the "being/non-being" level, neither denying the ten-level organization as a whole nor denying the existence of the three-level formulation in the Erdi yi. To extract one sequence for interface analysis is not to claim that Jizang's text is itself a four-cell table. Second, the boundary of what is interfaced: what this section interfaces is not the full range of Madhyamaka's emptiness (the absence of self-nature in all dharmas), but the step-correspondence between this structure and the four-phase generation of the chisel-construct cycle; the full range of emptiness belongs to Madhyamaka, and this section's lens illuminates only this segment of structure.

Jizang's original four steps are as follows. First level: being as conventional truth, emptiness as ultimate truth — setting up the pair "being" and "emptiness." Second level: being-and-emptiness as conventional truth, neither-being-nor-emptiness as ultimate truth — negating this pair. Third level: "both two and not-two are conventional truth; neither-two-nor-not-two is ultimate truth." Fourth level: "these three kinds of two truths are all teaching; in order to awaken to the not-three, with nothing to rest on, only then is it called principle." (This division into four levels of the two truths appears in Jizang's Dasheng xuanlun, fascicle 1, on "the meaning of the two truths"; the Zhongguanlun shu, end of fascicle 2, also lays out these four levels. [CBETA, T45, no. 1853, Dasheng xuanlun, fasc. 1, line-reference to be finalized])

These four steps correspond, step by step, to a published structure of the chisel-construct cycle. In the cycle's retrospective frame, in the most primordial situation (0D chaos), negation interrogates itself and generates four phases: being, non-being, the negation of (being and non-being), the negation of (neither-being-nor-non-being) — first setting up the pair of affirmation and negation, then negating that pair, then negating "the negation of that pair" without letting it collapse, and finally gathering in as a whole. [4] The correspondence is step-for-step, but at two points one must stop and see clearly, because they are exactly the load-bearing points of this interface analysis.

The first is at the third level. Jizang does not write here a simple repetition of "neither being nor non-being," nor a double negation that would logically collapse back to the start — if the third level were written "neither not-being nor not-emptiness," "neither not-being" would formally fall back to "being," and the whole advance would crumble back to the second level. What Jizang chooses is "neither-two-nor-not-two": what is negated is no longer the predicate at the level of "being" or "emptiness," but the very frame "two / not-two." The focus of negation leaps up one layer. This corresponds exactly to a constraint sedimented at 3DD in the chisel-construct cycle — a double negation does not return to affirmation; between distinct terms it leaves an uneliminable interval, non-coincident and irreversible. [4] Jizang's "neither-two-nor-not-two" is the clearest realization, in the language of argument, of this interval: what it negates is the frame, not a predicate within the frame, hence it does not collapse back; it is the strongest evidence to be found, in a historical text, for this correspondence.

The second is at the fourth level, and the more important. The fourth level differs from the first three: each of the first three sets up an ultimate truth and is itself demoted to a conventional truth to advance; the fourth level says the first three are "all teaching," itself falling on "nothing to rest on." This means the fourth level is not a fifth ultimate truth, not one more negation, but the gathering-in of the whole ladder — the first three sediment, as "teaching," into a construct as a whole, while "words forgotten and thought cut off, nothing to rest on" is the remainder this sedimentation leaves, the remainder that can no longer be gathered up, pointing toward that which cannot be attained by any single statement. A boundary must be marked here, because it is exactly the rehearsal of the seam to be developed in Section 6: on the Madhyamaka side this terminal remainder points toward "nothing to rest on" (empty — both the road and the destination are conventional designation); on the SAE side, the chisel-construct cycle's terminal remainder points toward that limit which is real yet unreachable (the thing-in-itself); the two correspond at the structural position of "a terminal remainder, left after the argument gathers in, pointing toward the non-fixable," yet what they point toward is one empty, one real — they part ontologically. Here only the correspondence at this structural position is marked; the two referents are not prematurely merged — that break is left to Section 6.

Here two kinds of remainder must be distinguished, or it will be misread, and this distinction is exactly the crux of the fourth level. The remainder among the first three levels is propulsive: the truth of each level demoted to conventional is precisely the force compelling the next level's negation to operate — this is the chisel-construct cycle's normal remainder, igniting the next blade. The fourth level's remainder is not of this kind. It does not propel a fifth level (there is no fifth level; to press the question by force yields only "words forgotten and thought cut off" — exactly what the ladder's collapse looks like when "there is no fifth step"). It is the terminal remainder left after the teaching gathers in: it no longer drives this ladder of argument to keep turning. Madhyamaka's ladder of argument has here completed all its work, suspended utterly — words forgotten and thought cut off, nothing to rest on, is prajñā's culmination at the road's end, not a midpoint awaiting continuation. In the SAE atlas of probes, this terminal coordinate of "argument exhausting itself" abuts, back-to-back, the position of Chan's "sudden cut of thought" — the two, each within its own independent and self-sufficient system, arrive by different roads at the same unspeakable boundary. This is the adjacency of two independent termini in the coordinate system, not the handing-over of one party to another to take up.

From this a commonly misunderstood point can be clarified: Jizang's ultimate truth ("the supreme meaning") is not an abidable dwelling. The four characters "nothing to rest on" by themselves deny that it is an arrived-at position. It is the remainder's pointing left after the ladder gathers in, not an occupied height. The supreme meaning is not a dwelling but gathering-in-plus-remainder — and on this point Jizang's own summit accords with the chisel-construct cycle's discipline of "remainder conserved, no construct closes."

It is worth noting that Jizang himself assigns these four levels to different capacities: "the sharp, hearing the first, awaken to the Way, with no need of the latter two; the middling, not awakening at the first, enter the Way only at the second; the dull turn to the third and only then comprehend." This shows that the four levels of the two truths, for Jizang, already carry the sense of "speaking to different people at different depths" — the same structure unfolds to different levels according to the listener. This is of a piece with the "speaking-to-position" discussed later under the three turnings of the Dharma wheel, and the two corroborate each other.

Finally a misreading must be guarded against, because it bears on the honesty of the whole paper. Self-as-an-End is a framework that grants the construct a real standing: every negation settles into a new, remainder-bearing real construct; the ladder is a ladder of constructs ceaselessly emerging. This is the opposite of a Hegelian "synthesis" — synthesis absorbs negation into a higher unity, the remainder eliminated in the synthesis; whereas what the chisel-construct cycle's fourth step leaves is precisely the space of the remainder — gathering-in is not unity. [4] Just for this reason, the fourth step nearly came to be named "synthesis" and was stopped by this distinction: to write the gathering-in as synthesis would be to read it as a higher truth, which is not the sense here. On the Madhyamaka side this distinction has its correspondence: Sanlun's "refuting without establishing" refuses precisely to read the "middle way" as a reconciling higher synthesis — should anyone understand the middle way as a third thing above being and emptiness, a synthesis, Sanlun's discipline would negate that reading. The two sides run in the same direction on "not letting the gathering-in become a higher unity."

(Note: this section's use of "four steps" for interface analysis rearranges Jizang's layer-by-layer two-truths structure into one interfaceable sequence; "step" is the framing of the analysis, not a claim that Jizang's original takes these four steps as its inherent grammar. Likewise, the "elimination" direction — the truth of the lower layer demoted to conventional at the upper — looks opposite to the "stepwise generation" direction of the four phases in the chisel-construct cycle, but the two are unified by the same thing: each step holds only by negating the prior — elimination and generation are two descriptions of the same negating act. And: this section's "interval law" and "construction law" are names of convenience, their content drawn from already-published material; their formal naming awaits a later version of the methodology paper.)

---

5. The two treasuries and the three turnings of the Dharma wheel

In Jizang's doctrinal classification there are two classifications of teaching; this section interfaces each, and the two axes must be separated from the start — they classify different things, and conflated they clash with each other.

The two treasuries: the direction of content

The first is the two treasuries: the Śrāvaka treasury and the Bodhisattva treasury. The Buddhist-side grammar must first be given, then the interface. The Śrāvaka vehicle's mark is self-deliverance — taking one's own liberation as the task, not the deliverance of others; the Bodhisattva vehicle's mark is benefiting others — taking the universal deliverance of beings, the common approach to awakening with all beings, as practice. This is the traditional division of the greater and lesser vehicles by the teachings expounded and the direction of the aspiration.

In the SAE interface reading, these two present as two directions of content: the Śrāvaka treasury presents as a grammar of liberation that "disassembles the self inward" — placing the effort on loosening one's own grasping; the Bodhisattva treasury presents as a grammar of acknowledgment that "opens toward the other" — placing the effort on acknowledging the other, approaching together with the other. This must be marked as an interface re-reading: to read the Śrāvaka vehicle's "not delivering others" as "disassembling the self inward" is SAE's restatement in terms of structural direction, not the original sense of the two treasuries; the two treasuries are originally divided by aspiration and teaching, and this section takes their structural direction. (Per the series' discipline, this section makes no judgment of high or low between the two treasuries, and does not use the term "lesser vehicle," using "Śrāvaka vehicle.")

The three turnings of the Dharma wheel: fitting to position

The second is the three turnings of the Dharma wheel: the root turning, the branch turning, and the gathering-the-branch-back-to-root turning. This set classifies not the direction of content but the fitting of teaching to different subjects — the same content of awakening must be spoken, to subjects at different positions, at different depths and in different ways.

A boundary must be set here, the most important in this section: the three turnings are Jizang's division of "how teaching varies with capacity," and what this section interfaces is this structure of fitting of "teaching according to capacity," not a reading of the three turnings as a low-to-high sequence of positions. Although the text below will correspond the three turnings' three moments to subjects' different positions, that is to say "what kind of teaching fits a subject at what position," not to say "Jizang laid out a ladder of subjects ascending level by level." This boundary must be held, or Jizang's teaching-according-to-capacity is misread as a developmental sequence.

Under this boundary, the interface of the three turnings is as follows.

The root turning is the Buddha's first and highest direct statement — the realm shown in the Huayan, of unobstructed interpenetration without end, the full presentation of the Buddha's own inner realization. In the SAE coordinate system, it corresponds to the vision of the layer of "acknowledging the other, too, as an end" (15DD), and points apophatically toward that non-objectifiable limit (which this paper does not develop, leaving it to a dedicated study). [4] This last point must be specially noted: what the root turning shows is not a set of "content" that can be sent and received like a signal — that highest realm, by its nature, can appear only in the present negotiation among subjects who are ends to one another, and cannot be unilaterally received as a piece of information.

The branch turning is the teaching the Buddha then sets up to draw beings in — the provision of the three vehicles, with a fruit to seek, a station to realize, a sequence to follow. In the SAE coordinate system, it corresponds to the grammar of the layer at which the self grasps itself as an end (14DD): seeking a fruit, seeking a vehicle, with a goal to reach.

The gathering-the-branch-back-to-root turning gathers the drawn-in subjects back to the root — uniting the three into one, leading all beings into the Buddha's wisdom alike. In the SAE coordinate system, it corresponds to the layer of "acknowledging the other, too, as an end" (15DD): the subject loosens the grasping at its own fruit and turns instead to the common approach with the other, with beings, as its home.

The difference between the root and the gathering-back must be made clear, because both fall near the layer of "acknowledging the other, too, as an end" and are easily confused. The root turning is the first full direct statement — showing the highest realm from the start, which is why most subjects cannot receive it. The gathering-back turning is a retrieval — gathering those subjects who, drawn in first through the branch, can already loosen the grasping at their own fruit, back to that highest realm. One is the lofty first showing, the other the retrieval after drawing-in; both face the same realm, but one stands at the sequence's beginning, the other at its end — their positions along the path differ.

"As if deaf and mute": an account of a fitting-mismatch

At the Huayan assembly, the two vehicles are "as if deaf and mute" — neither seeing nor hearing the highest realm the Buddha shows. This image is often used to illustrate the loftiness of the root turning. In the SAE interface reading, it is a phenomenon of fitting-mismatch, which can be accounted for thus — but first this must be marked: to read the two vehicles here as subjects at the layer where the self grasps itself as an end is SAE's interface reading, not a claim that the two vehicles in Buddhism are identical to some level of mind.

Under this reading, the two vehicles are subjects whose task is the self-sought realization of fruit — their mind's mode of operation is organized around "what fruit I am to realize, what station to ascend." The realm shown in the Huayan is, by nature, the present negotiation among subjects who are ends to one another; and a mode of operation whose task is the self-sought fruit by itself excludes "being an end for the other" from its operation — it is not that it failed to receive a high-frequency signal, but that its mode of operation has no place for the negotiation of "ends to one another." So "as if deaf and mute" is not a malfunction of the receiver but a non-correspondence of the mode of operation: a mind organized around its own fruit, meeting a realm whose nature is the negotiation of ends to one another, does not correspond, and so neither sees nor hears.

Just because of this non-correspondence, the Buddha withdraws to the grammar a subject whose task is its own fruit can correspond with — the branch turning, providing three vehicles, fruit-stations, a sequence. This is not "lowering the frequency" of that highest realm, but providing a teaching within which that subject's mode of operation can run. The relation here must be stated precisely, not read as a story of growth: when a subject's operation is locked onto the configuration of "unidirectional self-fruit-seeking," the protocol of "being an end for the other" is structurally excluded by that configuration; only when that configuration itself loosens and unloads that absolute grasping does the grammar of "gathering-the-branch-back-to-root" come to correspond with it — this is the un-sealing of a configuration, not the segment-by-segment accumulation of mileage. The gathering-back "corresponds" not because the subject has walked a stretch of road and arrived at some station, but because its configuration has loosened from "unidirectional fruit-seeking" into "able to admit the other"; the configuration changes, correspondence is achieved, with no question of earlier or later, near or far.

In this account, the judgment "the two vehicles are at the layer of their own fruit" is given independently by the subject's mode of operation (organized around its own fruit), not inferred backward from "as if deaf and mute"; "as if deaf and mute" is a corroboration of this independent judgment, not its ground. Thus the account is not one of using an unproven correspondence to explain a phenomenon and then using the phenomenon to confirm that correspondence.

It must close on the boundary set at this section's opening: the above correspondence of the three turnings' moments to subjects' different positions says throughout "what depth of teaching fits a subject of what mode of operation," not "Jizang laid out a sequence of subjects ascending level by level." The three turnings are a structure of teaching fitted to capacity; their directionality (an earlier-and-later among root, branch, gathering-back) is a direction in the provision of teaching, not a developmental ladder of subjects. To read it as the latter is to cross this boundary.

The two axes must not be conflated

Finally, back to the opening: the two treasuries and the three turnings classify different things. The two treasuries classify by the direction of content (disassembling inward / opening toward the other); the three turnings classify by the fitting of teaching to subjects (what depth fits what mode of operation). The two axes cross without coinciding: a Śrāvaka-vehicle teaching (toward self-deliverance, on the two-treasuries axis) can also be a branch turning (a drawing-in provision, on the three-turnings axis); the two fall on different axes and do not conflict. With the two axes explicitly separated, "the Śrāvaka treasury = disassembling inward" and "the two vehicles as if deaf and mute at the Huayan" are not read as a contradiction — the former speaks of this teaching's direction of content, the latter of this mode of operation's fitting-mismatch with a certain teaching; they belong to two axes.

(Also: this section's three turnings of the Dharma wheel and the "three revolvings of the Dharma wheel" discussed in the series' first paper are two different three-part divisions — the former Jizang's classification of "teaching varying with capacity," the latter another three-part division concerning the establishment of subject-position. The two must be explicitly distinguished and not conflated. The textual source of the three turnings is Jizang's Fahua youyi [CBETA, T34, no. 1722, line-reference to be finalized]; see also the Fahua yishu [CBETA, T34, no. 1721] on "concealing the real / raising the provisional / gathering the branch back to root.")

---

6. Where there is no interface

The most important section of an interface analysis is to state clearly where the two frameworks cannot be translated into one another. This is not an appendix to the divergences but the core of the analysis; to state it precisely matters more than to state the correspondences in full.

The divergence is not at "without self-nature." Madhyamaka's negation of self-nature — self-nature meaning independent of another, not arising from conditions, without change — Self-as-an-End accepts wholesale; its claim that "no construct can close upon itself, every one leaves a remainder" meets Madhyamaka's "all dharmas are without self-nature" at the conclusion. The two approach differently: Madhyamaka derives the absence of self-nature from dependent arising, Self-as-an-End derives non-closing from occlusion (every construct occludes, hence the remainder is non-empty); the same conclusion, different engines. The real divergence is one layer deeper: at the second truth, Self-as-an-End holds that the construct is a sediment carrying real constraint — though without self-nature, it is not zero, not a purely conventional designation; it has real effect and remainder. Madhyamaka would ask, from the side of emptiness: this "real constraint" — does it still smuggle in a faint thread of self-nature-grasping? And Self-as-an-End would answer, from the side of operation: that which executes the negation is itself a really operating chisel-construct process — a system that would build continuously and acknowledge that every building leaves a remainder cannot isomorphically translate this "real constraint" into a purely conventional designation, because what states this position and carries through this negation is precisely a real construct. But the proportion of this answer must be stated precisely at once: this is not to say Madhyamaka "cannot do it" or "lacks something" — Madhyamaka does not take continuous building as its aim at all; what it seeks is to negate self-nature to the end and bring about non-attainment, and within its own aim, reducing everything to a self-natureless conventional designation is thorough and self-consistent. The two are not one complete and one deficient, but two aims each complete and untranslatable here: Madhyamaka from the side of emptiness, Self-as-an-End from the side of operation, each with its own complete discipline, sees the other as having a little too much or a little too little. In this exchange, neither can translate the other's words into its own language without distortion. This is the deepest seam of non-interface this paper identifies.

This seam does not appear in only one place. It appears at three layers at once, isomorphic across the three, each corroborating the others.

First, at the ontological layer, the carrier is the famous verse of the Zhonglun: "Dharmas arising from conditions, I declare to be emptiness; this is also conventional designation, and is also the meaning of the middle way." This must first be made clear: in this verse, the four — dependent arising, emptiness, conventional designation, the middle way — are four names for the same matter, four facets of one thing, not a temporal sequence of first arising, then emptiness, then designation, finally the middle way. Self-as-an-End, for the convenience of interface analysis, rearranges these four names into a four-beat reading to set against the structure of the chisel-construct cycle — but this is the framing of the analysis, not the grammar of the original verse. After the rearrangement, the divergence falls at the last: Madhyamaka uses "conventional designation" as a name without self-nature — it points to a matter without lending it the weight of self-nature; the name points to the matter, and when the matter passes the name sets up no self-substance (this is not to say worldly operation has no efficacy: Madhyamaka's two truths still retain, at the worldly layer, the sayable, the doable, causation and instruction; conventional designation is only without self-nature, not without effect). Self-as-an-End reads "construct" as a sediment that, though without self-nature, has real constraining force — though without self-nature, it really constrains what follows. The same dependent arising: Madhyamaka reads out a designation without the weight of self-nature, Self-as-an-End reads out a reality with constraining force — this is the two parting on the same verse.

Second, at the disciplinary layer. Self-as-an-End has a discipline of its own: only an establishing that acknowledges the remainder is a tenable establishing (the "good" or "tenable" here refers only to internal operational soundness in SAE — whether an establishing has acknowledged its own remainder is SAE's own procedural standard, not a moral evaluation, and not some positive "right" Madhyamaka would acknowledge) — because refutation cannot but produce a new construct, and rather than pretend not to establish, one had better claim this remainder-bearing construct. But this discipline cannot hold itself up before Madhyamaka's "tetralemma." "A tenable establishing" itself presupposes a standard for judging "establishing" (whether the remainder is acknowledged), and that standard is again a position, a construct; Madhyamaka would press whether this "tenable" is self-arising, dependent, joint, or causeless — it is exactly the kind of "construct meeting some mark" that Madhyamaka would refute. Here the two cannot be translated, and Self-as-an-End does not retreat here, nor need it; but the ground of its non-retreat must be stated precisely. Its ground is not "I have paid some price, and so have the right to keep this establishing" — that is a way of disguising a mechanical fact as a claim of ownership, and must be avoided. Its ground is: that the construct does not collapse under Madhyamaka's reductio is because it really operates, really leaves a remainder, and not because anyone declared anything. Self-as-an-End claims this construct, and what it claims is something in process, not a right. As for Madhyamaka — seen from Self-as-an-End's operational side, it provides no ground on which to build continuously; but this is not its defect, rather the price of its discipline of "refuting without establishing," and exactly the strength of that discipline. What it seeks is not building at all, and to measure it by "has it a ground" is to use the wrong yardstick.

A further boundary must be added, to prevent a misreading of the temperature: the above talk of "price" and "ground" describes Madhyamaka from Self-as-an-End's operational side; it does not judge Madhyamaka as wrong or deficient. Madhyamaka, refuting without establishing, is thorough and self-consistent within its own aim; Self-as-an-End, acknowledging the remainder-bearing construct, is thorough and self-consistent within its own discipline. The two are not one-right-one-wrong but two walls of the same seam.

Third, at the reflexive layer — this is left to the final section, where Madhyamaka reflects back on Self-as-an-End, to be developed. It asks the form in which the same seam finally turns back upon Self-as-an-End itself: when Madhyamaka carries negation through to the end, does it point at the final footing on which Self-as-an-End argues — that "negativity" adopted as an axiom, that most fundamental negation which is no longer even a negation of itself — and ask, "is this not also one more grasped position, one more self-nature?" This question strikes exactly at Self-as-an-End's mode of self-exemption: under Madhyamaka's criterion, this risk would show thus — if that final negation is made into a point "immune to negation," and "immunity to negation" is, in Madhyamaka's eyes, one of the marks for judging self-nature, then on that yardstick it would appear to have failed at "emptying even emptiness." This is an appearance under Madhyamaka's measure, not necessarily Self-as-an-End's actual condition; the final section will develop it and meet it head-on — here only the seam's position is marked.

The three layers are three appearances of the same seam: the ontological layer is "designation without the weight of self-nature" against "construct with real constraint," the disciplinary layer is "refuting without establishing" against "claiming a remainder-bearing establishing," the reflexive layer is Madhyamaka's interrogation of Self-as-an-End's final position. Laid one on another, this seam of non-interface is not something a soft phrase can smooth over — it has structure, three independent supports. The task of an interface analysis here is to illuminate this seam clearly, to mark its two walls, not to throw across a bridge pretending the two sides are connected.

---

7. The residual-positivity check

Every paper in this series runs a "residual-positivity check": a check of whether, in the tradition being interfaced, there is something that on the surface is apophatic, negating, but at root quietly sets up a positive, abidable terminus — something SAE judges "non-fixable" yet which has still been fixed. This check is part of the honesty of an interface analysis: not to pick at the other's errors, but to mark the places that easily slide toward positivity, and see whether they in fact slid.

A difference from the previous paper must first be noted. The previous paper (Yogācāra) had a check-item, "drawing its own fruit" — how a seed produces a fruit, where a positive structure SAE judged untranslatable was hidden. This paper sets up no such item: "drawing its own fruit" is a concept in the Yogācāra seed doctrine; Sanlun has no such architecture, and this paper does not force a correspondence to align with the previous paper. This paper's residual-positivity check falls on two of Madhyamaka's own terms: quiescence, and true reality.

Quiescence: a boundary discovery

Madhyamaka speaks of "well extinguishing conceptual proliferation" — the extinction of proliferation. Here the check is: does this "extinction" quietly set up a positive terminus? Does it say that at some point everything utterly stops, returning to a quiescent final state?

On checking, this place is not a smuggled positivity but a boundary discovery — it marks the boundary between SAE and Madhyamaka on a fundamental question, and that boundary is itself clean and markable.

The key is to distinguish what "conceptual proliferation" is. Proliferation (prapañca) does not refer broadly to all conceptual activity; it refers specifically to the proliferation arising from dual grasping — being and non-being, arising and ceasing, one and different, coming and going, set up in pairs, grasped in pairs, proliferating layer by layer. This proliferation is dyadic: set up one side, and there is its opposite side; grasp one end, and the other is drawn out. What Madhyamaka means by "extinguishing proliferation" is precisely the extinction of this dyadic proliferation — no longer proliferating concepts in the manner of "setting up one side, grasping one end."

SAE's chisel-construct cycle is not dyadic. Its operation is: chisel a construct, the construct leaves a remainder, the remainder drives the next chiseling — this is an asymmetric succession, not pair-by-pair opposition. [4] What drives it is not the tension of "one side against the other" but the asymmetry between construct and remainder (the remainder is what the construct failed to gather, and the remainder drives the next round, not some opposite of the construct). [4]

So what the two would "stop" are two different machines. What Madhyamaka would stop is dyadic proliferation — this machine can, and should, be stilled. What SAE says does not stop is that asymmetric chisel-construct succession — so long as such a mind operates, the negation will not halt; the remainder is conserved. [4] The machine Madhyamaka stops (dyadic proliferation) and the machine SAE does not stop (asymmetric succession) are not the same machine; Madhyamaka's stilling of dyadic proliferation does not touch the asymmetric chisel-construct succession in the least. So there is no head-on collision here — each speaks of its own machine, one to be stopped, one not.

This boundary is held precisely by Madhyamaka's own text. The Zhonglun's "Examination of Nirvāṇa" defines nirvāṇa: "nothing attained, nowhere reached, neither cut off nor permanent, neither arising nor ceasing — this is called nirvāṇa." [CBETA, T30, no. 1564, Zhonglun, fasc. 4, "Examination of Nirvāṇa," line-reference to be finalized] "Nowhere reached" — no arrival at some other state; "neither cut off" — not annihilation. The difference between nirvāṇa and saṃsāra lies not in arriving at a quiescent final state, but in "receiving conditions" versus "not receiving conditions" — no longer operating in the manner of dual-grasping conditions. [CBETA, T30, no. 1564, Zhonglun, fasc. 4, "Examination of Nirvāṇa," line-reference to be finalized] This accords exactly with "what is extinguished is dyadic proliferation, not a halting of the chisel-construct succession": nirvāṇa is not a machine utterly shut down, but a ceasing to operate in the manner of dual grasping.

So quiescence, here, is not a smuggled positivity but a boundary discovery: it marks the boundary between "the dyadic proliferation that should be stilled" and "the asymmetric succession that should not be stilled" — Madhyamaka stills the former at the doctrinal layer, Self-as-an-End retains the latter at the mechanical layer; each is thorough within its own system, and the machine they refer to is not the same. A temperature boundary must be held: the above distinction of "two machines" is an SAE-side mechanical analysis; Madhyamaka itself does not speak of proliferation in terms of "symmetric/asymmetric machines." To say that what Madhyamaka extinguishes is dyadic proliferation is SAE's interface description, not a stuffing of SAE's mechanical reading into Madhyamaka's mouth.

True reality: the objectification risk of a noun

The second place is "true reality." "True reality is without mark" — true reality is markless; this is itself a thorough apophasis: true reality is no graspable mark. At root, the claim here is Rho-compliant (in SAE's terms, it sets up no abidable positive terminus). Sengzhao guards it with a double removal: "neither being nor truly being, neither non-being nor truly non-being." [CBETA, T45, no. 1858, Sengzhao, Zhaolun, "On the Emptiness of the Unreal," line-reference to be finalized] The double removal falls to neither side, sets up no terminus, clean.

But the risk lies not in the claim but in the word "true reality" itself. The Chinese term carries a character meaning "real"; "true reality," as a noun, is easily objectified — read as "the real way things are behind dharmas," as if there were a markless, real "true reality" sitting there, an ultimate substrate prajñā is to accord with, to merge with in dark accord. This objectified reading is precisely the positivity-risk SAE marks: it congeals a word that should be apophatic into an abidable real place; more covertly, it implies, epistemically, a "prajñā in dark accord with true reality" — an accord of zero remainder-friction, as if the subject's cognition could correspond seam-tight with that true reality, with no occlusion at all. SAE's discipline is: any cognitive response is itself a construct, with its own occlusion and remainder; there is no zero-friction accord that gathers the object in wholly without residue. [4] Were "prajñā in dark accord with true reality" read as such a zero-friction accord, it would violate "every construct occludes."

This risk, too, is guarded by Madhyamaka's own text. The Dazhidu lun defines the supreme-meaning siddhānta as "past all paths of language, the workings of mind extinguished, resting nowhere, indicating no dharmas," and points to the true reality of dharmas with verses such as "neither arising nor ceasing, the dharma like nirvāṇa"; later tradition may summarize this as "the eightfold negation is just true reality," but this paper does not take those four characters as a continuous original sentence of the Dazhidu lun. [CBETA, T25, no. 1509, Dazhidu lun, fasc. 1, line-reference to be finalized] What matters: true reality is here defined as "resting nowhere, indicating no dharmas" — it rests on no place, indicates no dharma; it is precisely an apophatic mark, not a positive substance. So wherever this paper uses the term "true reality," it takes its apophatic use: true reality is the suchness of the absence of self-nature in all dharmas, the place where the workings of mind and the paths of speech are cut off, not a real substance behind dharmas; and SAE's non-acceptance of a "prajñā in dark accord with true reality" zero-friction cognition must be expressly marked — cognition too is a construct, with its own occlusion.

The check's conclusion

Two places checked. Quiescence is a boundary discovery — it marks the boundary of two machines, not a smuggled positivity. True reality's claim-layer is Rho-compliant; its noun-layer has an objectification risk, guarded by a double guardrail (apophatic use + against zero-friction accord).

The significance of this check must be stated, and the temperature held. Madhyamaka is, among the traditions this series interfaces, the hardest one in which to find "a fixed positivity" — its whole work is to negate every fixable place. If this paper's check finally finds no genuine smuggled positivity in Madhyamaka, this is not a failure of the check but a meaningful result: it supports the placement of Madhyamaka as "the probe of that non-fixable law." But the point must be stated precisely — finding no positivity is a strong support for this placement, not its independent proof. The task of the check is to mark clearly the places that easily slide toward positivity and see whether they slid, not to determine in advance that they must or must not.

---

8. The three off-anchorings of emptiness: Sengzhao's critique of the three houses

Sengzhao's "On the Emptiness of the Unreal" critiques three houses of the time — the Original Non-being house, the Mind's Non-being house, and the Matter-as-such house — pointing out that each, somewhere, anchored "emptiness" awry. In the SAE coordinate system, these three critiques correspond exactly to three typical slides of misplaced negation. This section marks that correspondence as a firewall — it shows that the early Chinese receivers of Madhyamaka were already systematically resisting the various deviations of reifying or half-completing "emptiness." (This section's title uses "off-anchoring" rather than "wrong-anchoring": the "off" is a deviation relative to the "unreal-emptiness" standard Sengzhao himself sets; the "off" and "wrong" below are all read in this sense, and are all Sengzhao's judgment, not SAE's invention — see the next paragraph.)

One thing must first be framed, or the temperature will be misread. That Original Non-being, Mind's Non-being, and Matter-as-such are judged "wrong" is Sengzhao's own judgment — Sengzhao is the one critiquing these three houses. What SAE does here is not itself to judge these three as wrong, but to follow Sengzhao's existing judgment and correspond the places he judged wrong to some structural position in the SAE coordinate system. The "wrong" in "off-anchoring" is Sengzhao's judgment, not SAE's invention; wherever the text below says some house is "wrong" somewhere, it is read in this sense.

The Original Non-being house Sengzhao critiques as "inclining in feeling toward non-being," weighting "non-being" too heavily, as if before and above all transformations there were a primordial "non-being" as substance. In the SAE coordinate system, this corresponds to anchoring the negation in the wrong place — regressing the negation into a primordial, substantial nothingness, as if "emptiness" were some substance-of-non-being prior to everything. [4] This reifies "emptiness" into a substance at the origin, whereas SAE's negation is not a substance of non-being but a continuously operating operation.

The Mind's Non-being house Sengzhao critiques as "no-mind toward the myriad things, yet the myriad things are never non-being" — emptying only at the subject's mind (not arousing grasping), but not emptying at things (the myriad things still really exist). In the SAE coordinate system, this corresponds to a half-completed suspension: cutting off grasping only locally (one's own mind), without carrying it through to the linkage in which mind and things mutually constitute one another — retreating into an isolated observer, plucking oneself out of that mutually constituting linkage while letting the other end of the linkage (things) stay really existent. [4] This is a suspension done only halfway, cutting off the feedback, retreating into isolation.

The Matter-as-such house Sengzhao critiques as "saying matter is not self-matter, yet not grasping matter's non-matter" — knowing "matter is not self-existent" (matter forms dependent on conditions), yet not carrying this negation through to the end, still retaining somewhere some reality of "matter." In the SAE coordinate system, this corresponds to an incomplete negation: the negation is done, but not chiseled to the bottom, holding back at the layer that carries that matter, not also stripping away the grasped carrying-surface. [4] This is chiseling not thoroughly, leaving an un-negated bottom.

Three houses, three wrongs, exactly three slides of misused negation: Original Non-being regresses the negation into a substantial nothingness (anchoring it wrong, as a substance of non-being), Mind's Non-being does the suspension only halfway (cutting off feedback, retreating into isolation), Matter-as-such does the negation incompletely (leaving an un-chiseled bottom). Sengzhao refutes these three with "unreal-emptiness" — dharmas being unreal, hence empty: unreal says dharmas are without self-nature (hence not falling into Original Non-being's substance-of-non-being, nor Matter-as-such's leftover bottom); empty says this absence of self-nature pervades both mind and things (hence not falling into Mind's Non-being's halfway). In the SAE coordinate system, these three refutations hold exactly three lines: the negation does not regress into a substantial nothingness, the suspension must carry through to the end, the chisel must chisel thoroughly.

The significance of this correspondence is a firewall: it shows that reading "emptiness" as that non-fixable law — rather than as a substantial nothingness, a half-completed suspension, a negation with a leftover bottom — was already being systematically guarded by the early Chinese receivers of Madhyamaka. The boundary must be restated: Sengzhao is not an internal criterion of the Sanlun school; he is a figure in Kumārajīva's translation circle and the early Chinese Madhyamaka milieu; this section cites him as a prehistory-interface to how Chinese emptiness-discourse resisted "the reification of emptiness," not as taking his judgment for an internal standard of Jizang's system. And what this section corresponds is how the "wrong places" of three houses Sengzhao already judged wrong fall onto the SAE structure, not SAE itself judging some house wrong (per the framing at this section's opening).

(Linkage with the previous section: this section's guarding of "true reality not reified" and Section 7's guarding of "true reality not objectified" are two faces of the same guardrail — Sengzhao is precisely one of the load-bearers of the term "true reality," so the apophatic boundary set for him here and the apophatic boundary set for "true reality" in Section 7 are to be read together.)

---

9. Reflexivity: Madhyamaka reflecting back on Self-as-an-End

An interface analysis that lets only SAE read Madhyamaka is incomplete. This final section lets Madhyamaka reflect back on SAE — especially upon SAE's most fundamental footing, asking a question SAE must meet head-on.

This section takes up the third face — the reflexive face — of the central untranslatable seam marked in Section 6. It was said there that this seam appears at the ontological layer (designation without weight / construct with constraint) and the disciplinary layer (refuting without establishing / claiming a remainder-bearing establishing); now it is the form in which it turns back upon SAE itself.

Two modes of self-exemption

The question surfaces thus. SAE's final footing in argument is that "negativity" adopted as an axiom — that most fundamental negation which is no longer even a negation of itself. [4] Madhyamaka would point at it and ask: is this not also one more grasped position, one more fixed self-nature? You, SAE, have negated the self-nature of all constructs; then is this final negation, on which you negate, itself a construct, a self-nature?

What makes this question sharp is that it strikes at the difference between the two frameworks' ways of handling "self-exemption" — how each keeps its own final footing from self-dissolving.

Nāgārjuna's way is subtractive. In the Vigrahavyāvartanī, facing the challenge "is your doctrine of emptiness itself a doctrine with self-nature," Nāgārjuna does not simply say the four words "I have no thesis," but says "if I had a thesis, I would then have a fault; because my thesis is no-thing, thus I incur no fault," glossed as "I have no thesis-mark — how can you fault me?" — I have no thesis that can be pointed at, that can be fixed; so your challenge "every thesis needs self-justification" cannot land on me, because I set up nothing. [CBETA, T32, no. 1631, Nāgārjuna, Vigrahavyāvartanī, line-reference to be finalized] The subtractive exemption relies on "emptying even emptiness" — even the statement "emptiness" is not set up as something with self-nature, so there is no target to attack.

SAE's way, at first glance, is performative: would you negate that final negation? The very act of your negating it is that negativity in operation — the more you try to negate it, the more you demonstrate it. This mode of exemption is forceful, but it has a danger: it makes that final negation a point immune to negation — every attempt to negate it conversely vindicates it. And "immunity to negation" is exactly one of Madhyamaka's marks for judging "self-nature": something independent of another, unbreakable, vindicated by the very negation of it, is, under Madhyamaka's criterion, taken as a self-nature. So there is an irony here: if SAE exempts itself purely performatively, that final footing on which it negates all self-nature would, under Madhyamaka's criterion, appear most of all like a self-nature — on that yardstick, it has failed precisely at "emptying even emptiness" as Madhyamaka demands. It must be stressed that this is an appearance under Madhyamaka's measure, the sharpness of Madhyamaka's question, not SAE's having already conceded that it has a self-nature; what follows is exactly how SAE, in its own language, responds to this yardstick.

How SAE meets it

SAE can meet this question, but the way it meets it cannot be carrying the performative through to the end (that would only fix the final negation more firmly). To meet it, SAE must acknowledge that it too has a subtractive face.

Among SAE's resources, what truly meets this question is two things. First, that final negation, by its nature, is not even a negation of itself — it is not an object that can be affirmed or negated; it is prior to that opposition of affirmation/negation. [4] This, structurally, accords with Madhyamaka's "emptying even emptiness": that final negation is not a graspable thing, just as emptiness is not a graspable thing. Second, SAE acknowledges of itself: this paper is a construct, this paper has already said too much. [4] SAE does not take its own exposition as a complete, remainderless ultimate statement — it acknowledges that its own exposition is also a construct, also has a remainder, also will be chiseled by the next blade. This acknowledgment is precisely SAE's version of subtractive self-modesty: it does not fix its own final footing as an unbreakable, complete point of self-nature.

So SAE is not purely performative. It has both a performative face (the act of negating that negation demonstrates that negation) and a subtractive face (that final negation is not even a negation of itself, and this paper acknowledges itself a construct). What meets Madhyamaka's question is the latter face: precisely because SAE acknowledges that its own final footing is not a graspable, complete point of self-nature, precisely because it acknowledges that this paper too is a construct with a remainder, it has not fixed that final negation into a self-nature immune to negation. This echoes the disciplinary layer of Section 6 — SAE claims its own construct, and what it claims is something in operation, not an unshakable ownership; likewise, SAE holds its own final footing, and what it holds is not an unbreakable point of self-nature but a footing that acknowledges itself a construct, acknowledges itself to have a remainder.

By here the three faces of the central untranslatable seam have all appeared: the ontological layer is "designation without the weight of self-nature" against "construct with real constraint," the disciplinary layer is "refuting without establishing" against "claiming a remainder-bearing establishing," the reflexive layer is Madhyamaka's interrogation of SAE's final position and SAE's head-on response with its subtractive face. The three faces, isomorphic, are three appearances of the same seam. The task of an interface analysis here is not to fill in this seam pretending the two sides are connected, but to illuminate all three of its appearances — and, on this reflexive face, to let SAE meet that question head-on, responding in its own language (neither retreating — it still claims its construct and footing; nor fixing its footing into a self-nature Madhyamaka would condemn), rather than turning around to absorb Madhyamaka as a doubt contained within SAE.

Madhyamaka guards its non-establishing by "emptying even emptiness," SAE guards its establishing-without-fixing by "acknowledging itself a construct" — two ways of guarding, yet what they guard is the same posture: the refusal to let negation have an abidable terminus. Few philosophies, arriving at the end of negation, do not stop there, do not set the final insight up as an abidable truth; Madhyamaka insists on emptying even "emptiness," refusing to let that by which all is negated be fixed as an abidable self-nature. This step is extremely hard to hold, and Madhyamaka held it. SAE, on the other side, guards the same posture by "no construct closes, the remainder cannot be eliminated" — no construct can seal itself off as a terminus, no negation fails to leave a remainder driving the next. This "same" must be stated precisely, because the very parting this paper guards throughout hangs on it: what the two share is that operational posture — both refuse to fix the process as an abidable ultimate truth; and what the two refuse to fix is, ontologically, exactly opposite — SAE's end is a limit real yet unreachable (the road has no end, yet the direction is real), Madhyamaka's end is an emptiness that sets up not even an "edge" (both the road and the destination are conventional designation). Isomorphic in posture, parted in referent. So this is not the two meeting at some shared terminus — their termini are ontologically one real, one empty, untranslatable; what they meet at is only the single posture of "refusing to fix a terminus." Madhyamaka states this posture as "the emptiness of emptiness, non-attainment," SAE states it as "the remainder conserved, no construct closes": the same gesture of refusing to abide, reaching toward two incommensurable directions. This seam — open at the referent, yet isomorphic in posture — is exactly the seam this interface analysis has skirted throughout, never, and never daring to, fill in; and precisely because the two guard so thoroughly on the two walls of this seam, this whole comparison is worthy of its object.

---

10. Conclusion

What this paper has done is compare Madhyamaka / the Sanlun school with Self-as-an-End in structure.

On comparison, the correspondences are deep. Madhyamaka's negation of self-nature meets SAE's thesis "no construct closes" at the conclusion; Madhyamaka's "negation" is isomorphic with SAE's "chisel" in operation; Madhyamaka's "non-attainment" answers SAE's non-fixable limit at the terminal signature. Jizang's layer-by-layer advancing structure of the four levels of the two truths can be matched step by step with SAE's steps of negation; the eightfold negation as a total negation of self-nature-grasping, and the three turnings as a fitting of teaching to capacity, both find their corresponding positions in the SAE coordinate system.

But on comparison, the partings are real too, and these partings are not flaws in the correspondence but the most valuable product of this comparison. The deepest is at the second truth: Madhyamaka reads the dependently-arisen dharma as a designation without weight, SAE reads the construct as a sediment with real constraining force — the same dependent arising, one reading out a designation without weight, one a reality with constraint. This seam appears at the ontological, disciplinary, and reflexive layers at once, isomorphic across the three, marking the line at which the two frameworks cannot be translated into one another.

This comparison adds a third face to the series' coordinate system. In the first two papers, Yogācāra illuminated the "construct" — how the mind builds finely; Chan illuminated the "chisel" — the negating act that breaks the abidable construct. In this paper, Madhyamaka illuminates the "law" — that no construct and no chiseling can be fixed as self-natured. The nature of this third face must be restated: it is not a higher position but a constraint that holds across all positions. What Madhyamaka illuminates at its point of application (the layer at which the self grasps itself as an end) is this non-fixable law itself.

A proportion this paper has held throughout must be restated once more in the conclusion: to read Madhyamaka as "the probe of that non-fixable law" is not to read Madhyamaka as "an unfinished SAE." Madhyamaka is not an SAE that lacks some sequence, missing some layer; it is a framework with a different aim from SAE, complete in itself — what it seeks is to negate self-nature to the end and bring about non-attainment, not to build a tower of layers. This paper meets it in structure and parts from it at the second truth; at the meeting it marks the correspondence, at the parting it marks the boundary. This is the task of an interface analysis: neither pretending the two frameworks are the same thing, nor pretending there is no comparable structure between them.

The series will continue. The remaining Buddhist traditions — each, in some problem-domain, having ground a mirror worth setting against SAE — will be interfaced in turn. The task of each interface is the same: not because some tradition "is really expounding SAE," but because every mature system, at the point where it presses hardest, illuminates some structures worth comparing. What this mirror of Madhyamaka illuminates is that law which runs through all positions, suffering no construct and no chiseling to fix itself.

---

References

I. Self-as-an-End works (Zenodo)

Every SAE-side structural thesis cites SAE's own published texts, marked [1]–[4] etc.; the DOI given is the concept DOI.

[1] Han Qin (秦汉). 〔Via Negativa: A Methodology of Negation〕. Zenodo. Concept DOI: 10.5281/zenodo.19544619.

[2] Han Qin (秦汉). 〔Via Rho: The Conservation of the Remainder〕. Zenodo. Concept DOI: 10.5281/zenodo.19657439.

[3] Han Qin (秦汉). 〔The Four Phases of 0DD and the Quadripartite Form〕. Zenodo. DOI: 10.5281/zenodo.20187591.

[4] Han Qin (秦汉). 〔Self-as-an-End: General Methodology (V2)〕. Zenodo. DOI: 10.5281/zenodo.18842449.

[5] Han Qin (秦汉). 〔SAE Buddhist Interface Series I: Yogācāra — The Ālaya-vijñāna/Seed System and the Quadripartite Form〕. Zenodo. DOI: 10.5281/zenodo.20579131.

[6] Han Qin (秦汉). 〔SAE Buddhist Interface Series II: Chan — Non-abiding and the Breaking of the Abidable Construct〕. Zenodo. Concept DOI: 10.5281/zenodo.20635734.

> Note: the exact published titles of [1]–[6] and the DOI of Series II [6] are as given on their Zenodo record pages; on publication, this paper's continues relation points to Series II [6], and its cites relation points to [1]–[5] and to the Buddhist sources below. DOIs are placeholders pending registration.

II. Buddhist canonical sources (CBETA)

Citations follow the Taishō (T) volume and number; the edition is the CBETA electronic canon. Where "line-reference to be finalized" is marked, the fascicle/chapter is fixed and the precise page-line awaits finalization against the CBETA source.

[7] Nāgārjuna, comm. Pingala, trans. Kumārajīva. Mūlamadhyamakakārikā (Zhonglun 中論). CBETA, T30, no. 1564. 〔Eightfold-negation homage verse and Pingala's commentary "stated briefly as eight matters, it is a total negation," "if one, then no condition; if different, then no continuation," fasc. 1, "Examination of Conditions," from p. 1c12; nirvāṇa verse "nothing attained, nowhere reached … this is called nirvāṇa," fasc. 4, "Examination of Nirvāṇa," line-reference to be finalized.〕

[8] Jizang. Dasheng xuanlun 大乘玄論. CBETA, T45, no. 1853. 〔Four levels of the two truths, fasc. 1 on "the meaning of the two truths"; line-reference to be finalized.〕

[9] Jizang. Zhongguanlun shu 中觀論疏. CBETA, T42, no. 1824. 〔End of fasc. 2 also lays out the four levels; line-reference to be finalized.〕

[10] Jizang. Erdi yi 二諦義. CBETA, T45, no. 1854. 〔Three-level formulation of the two truths, consulted.〕

[11] Jizang. Fahua youyi 法華遊意. CBETA, T34, no. 1722. 〔Root / branch / gathering-back three turnings of the Dharma wheel; line-reference to be finalized.〕

[12] Jizang. Fahua yishu 法華義疏. CBETA, T34, no. 1721. 〔"Concealing the real / raising the provisional / gathering the branch back to root," consulted.〕

[13] Sengzhao. Zhaolun 肇論, "On the Emptiness of the Unreal" (不真空論). CBETA, T45, no. 1858. 〔Critique of the three houses (Original Non-being / Mind's Non-being / Matter-as-such), "neither being nor truly being, neither non-being nor truly non-being" double removal; line-reference to be finalized.〕

[14] Nāgārjuna, trans. Kumārajīva. Dazhidu lun 大智度論 (Mahāprajñāpāramitā-śāstra). CBETA, T25, no. 1509. 〔Fasc. 1 on the "supreme-meaning siddhānta," "past all paths of language, the workings of mind extinguished, resting nowhere, indicating no dharmas"; line-reference to be finalized.〕

[15] Nāgārjuna, trans. Vimokṣaprajñā-ṛṣi & Gautama Prajñāruci. Vigrahavyāvartanī 迴諍論. CBETA, T32, no. 1631. 〔"Because my thesis is no-thing, thus I incur no fault," "I have no thesis-mark — how can you fault me?"; line-reference to be finalized.〕

III. Modern scholarship (reference layer)

Per the series' discipline, modern Madhyamaka scholarship may be used only to sharpen the accuracy of statements about Madhyamaka (reference layer), and bears no warrant for any SAE thesis (criterion layer). This paper's body cites no modern scholar directly; should the English rewrite require reference-layer citation on points such as the operationalization of svabhāva (e.g., Garfield, Siderits, Westerhoff, Robinson), these are to be marked separately, without altering their reference-layer standing. 〔Pending author's confirmation.〕

〇、导言

这是 Self-as-an-End（以下简称 SAE）佛学接口系列的第三篇。前两篇分别对接了唯识宗与禅宗：唯识一篇处理阿赖耶识—种子系统与个体心智的精细建构，禅宗一篇处理无住、破除可停留之构、以及自我成熟过程中的那一跃。本篇对接中观，以及它在汉地的系统化形态——三论宗，尤其是吉藏的判教体系。

须在开篇就说明本系列的性质，因为它决定了下文每一处对接的读法。本系列做的是接口分析，不是翻译，也不是吸纳。接口分析的本分是：在两个框架之间，既标出它们对应之处，也标出它们无法互相翻译之处，两者等重。把对应说满而隐去断裂，是把接口悄悄变成吸纳；只标断裂而不认对应，则是拒绝分析。本篇尤其要守住前一条——因为中观与 SAE 的对应之深，会在不经意间产生一种把中观读成"未完成的 SAE"的引力，而那正是接口分析必须抵住的。SAE 不把中观当作自己的一个早期版本，中观也不被本篇当作一套有待补全的框架；两者是各自完整、目标不同的体系，在某些结构上相遇，在某些地方分手。

系列的坐标：从两个面到三个面

本系列把不同佛学传统读作架在心智演化不同位置上的高分辨率探针。到本篇为止，这个坐标系需要一次扩展。前两篇用的是两个面：唯识照亮"构"——所建之物如何在特定位置上精细成形；禅照亮"凿"——那个削除、破除可停留之构的否定动作。本篇引入第三个面：中观照亮"律"——任何构、任何凿，都不能被坐实为自性。

须立刻说清这第三个面的性质，否则会被误读为一个更高的位置。它不是。构与凿各自落在心智演化的某个位置上（唯识在自我精细建构的层级照构，禅在自我成熟那一跃的接缝处照凿）；而"律"不是又一个位置，是一条跨越所有位置都成立的约束——无论在哪一层、无论构得多精细、凿得多彻底，那个构都不能被无余地坐实为自有、自足、不依他的东西。换句话说，"构位""凿缝""律"这三个名指的是三个面，不是三个并列的位置：构是所建之物，凿是削除的动作，律是任何位置都不闭合的约束。中观作用于自我执取自身为目的的那一层（关于这一层为何是中观的着力处，下节详述），而它在那里照亮的，是这条跨位的律。

这一扩展不回改前两篇。唯识照构、禅照凿，依然成立；本篇只是把系列的坐标系从两个面推进到三个面，让"任何构与任何凿皆不可坐实"这条一直在背景里起作用的约束，第一次有了它自己的探针。

本篇的主词与着力处

本篇的主词是"空"，对接的对象是中观对"自性"（svabhāva）的否定。但须在此立一道边界，它贯穿全篇：佛教一侧，中观破的是一切法的自性执，空是诸法普遍无自性；SAE 本篇取的，只是这一否定在"自我执取自身为目的"这一层的高分辨率接口。全幅的空性属于中观，本篇的透镜只照其中一段。这道边界不是谦辞，是防止把中观窄化的必要声明——下文凡涉及"空作用于某一层"之处，都应在这道边界下读：那是 SAE 选取的着力处，不是中观空性的全部射程。

为什么着力处是"自我执取自身为目的"这一层？因为中观破的自性执，在心智演化的序列上，最尖锐地凝结于此。在 SAE 的序列里，自我意识出现之后，紧接着是"意义"层——主体开始为自身的行动寻找理由，把自己当作一个目的本身（14DD，向死而生的那一层）。[4] 这一层的全部抓取——求一个果、求一个位、立一个自足完备的"我"——本质上就是在把"自我"建构成一个带自性的东西：自生、不依他、不变异。中观的八不、四重二谛，逐刀逐层拆的，正是这种把自我坐实为自性之目的的执取。须把话说准：中观做的是"破"——遮拨那个被坐实为自性的自我之执；破到执取松动之后，主体不再须把全部力量押在守护这个自性之我上，遂腾出了转向更高位置的可能——承认他者也是目的本身（15DD，伦理那一层）。[4] 本篇追踪的，正是中观这一"破"如何在 SAE 的坐标里腾出这一转向的可能：主轴是"破所腾出的可能"，不是"破为了转向"——转向是执取松动后腾出的余地，不是中观的破自带的目标。中观的破在它自身的解脱论里指向的是无所得，不是 SAE 的这一转向；"腾出转向他者的可能"是 SAE 坐标里读出的下游，不是中观破的本意。

须再次申明边界，因为它最容易在这里滑掉：说中观"作用于自我执取自身为目的这一层"，是 SAE 为接口分析选定的着力处，不是说中观本义就是一套关于自我目的的心理治疗。中观是对一切法自性执的总破；本篇只取其在这一层的接口。下文论八不为"总破"时，会再回到这一点。

探针的类型

前一篇论禅时，禅的探针是混合的：既有结构对接的一面，也有工程操作的一面——禅以一棒一喝、一机一境直接促成那一跃，凿的自由度极高而构的精确度极低。[4] 中观恰在这条轴的另一端：它以八不的穷举、离四句的逻辑、二谛的层层推进来逼近同一个不可坐实者，凿的自由度低而构的精确度高。所以本篇的探针是"以论证的结构对接、在终点处转为遮诠"——论证的形式（八不、离四句、二谛的逻辑架构）可以与 SAE 的结构步步相照，但走到终点（无所得、空亦复空、言忘虑绝）则不可再正面陈说，只能以否定方式标记。这与禅篇恰成镜像：禅是在跳跃处转为遮诠，中观是在论证自我耗尽处转为遮诠。

文本与引用

本篇的文本主轴是三论（《中论》《十二门论》《百论》）与吉藏的著述（《大乘玄论》《二谛义》《三论玄义》《中观论疏》《法华游意》），依鸠摩罗什的译本系统。《肇论》作为罗什译场与汉地早期中观语境的前史接口引入，不作三论宗内部判据。龙树《回诤论》在反身一节作旁证，不入主轴。现代中观学术（如相关的 svabhāva 分析）仅用于厘清中观陈述的准确性，不承担任何 SAE 论断的依据——SAE 一侧的结构只援引 SAE 自身的已发表文本。

下文凡引 SAE 的结构（凿构循环的五概念、维度序列、各层之律），均指向方法论总论及相关应用论文，记为 [4] 等；凡引中观与三论文本，均标其出处，草稿阶段逐条对勘。

---

一、空的锚定

要把"空"放进 SAE 的坐标，第一件事是确定它落在哪里。答案不是一个点，而是一条线——空在 SAE 的坐标里纵贯三层。

作为论题的空，对应 SAE 的一条核心主张：任何构都不能自我闭合，每一次建构都留下未被收摄的余项。[4] 中观说"诸法无自性"，说的是没有任何法是自生、自足、不依他的；SAE 说"无构闭合"，说的是没有任何构能把自己封死、不留余项。两者在结论上相遇——一个"无自性"，一个"不闭合"。

作为操作的空，对应 SAE 的否定方法本身——那个削除、破除、追问不完备之处的动作（凿）。[4] 中观不是把"空"当作一个要去抵达的对象，而是把"破"当作贯穿始终的操作：破自性、破执、破一切可停留之见。这与凿的运作同构：凿不预设一个终点的构，它只是不断否定当前的构、逼出它的余项。

作为终点签名的空，对应 SAE 那个不可正面陈说、只能以否定方式趋近的极限——物自体的位置。[4] "空亦复空""无所得""言忘虑绝"都不是描述一个被抵达的状态，而是论证走到尽头时留下的遮诠标记。在 SAE 里，这个极限只能信、不能知，只能以双重否定的形式显现，不能被正面构出。[4] 中观的无所得，正落在这个位置。

所以"空"在 SAE 坐标里不是一个孤立的点，是一条从"无构闭合"这条论题，经过"破"这个操作，纵贯到"无所得"这个终点签名的线。

自性的三个标志，与一处必须标清的边界

要让对接精确，须把"自性"的内容拆开。中观所破的自性（svabhāva），经典地有三个标志：不从缘生（自生）、无变异（恒常）、不依他而有（独立）。这三个标志，SAE 全盘接受其否定——没有任何构是自生的、不变的、不依他的。

但这里有一处必须标清的边界，否则就会把 SAE 的概念硬塞进中观的术语。SAE 关心的核心是构是否闭合、是否留余项——这是一个"余项视角"的判准。而 svabhāva 的第二个标志"无变异"是一个"时间视角"的判准——说的是一个法是否在时间中生灭变化。两者貌合而神离：SAE 问的是"这个构留没留下未收摄的剩余"，svabhāva 问的是"这个法变不变"。所以当 SAE 把"无自性"操作化为"无构闭合"时，它走的是自己"余项视角"那一支，而不是 svabhāva"时间视角"的"无变异"那一支。这一点须写明：三个标志（自生、无变异、不依他）是中观一侧的；SAE 把其中"闭合/无余"读作自己版本的"自性"，是接口操作，不是说 svabhāva 的原义本就是"无余"。守住这道边界，既不曲解中观，也不让 SAE 的主梁（空即无构闭合）失去落点。

两家虽在"无自性／不闭合"上结论相遇，推导的引擎却不同。中观从依存推出：法皆待缘而起，故无自性。SAE 从遮蔽推出：凡构必有所蔽，故必留余项、不得闭合。[4] 同一个结论，两条引擎——这正是接口分析要标的"对应加边界"：对应在结论，边界在引擎。

方向，不是终点

最后须立一道护栏，它在全篇反复出现。SAE 的极限——那个不可坐实者——是方向，不是终点。主体不能穿透它，只能逼近它的遮诠边缘；"一直凿"是纪律，那个极限是凿所朝向的方向，不是一个可被占据的终点站。[4]

有意思的是，中观在这一点上与 SAE 同向。中观也禁止"达到"——"无所得"正是说没有一个所得的终点可抵达，连"空"也不可执为一个被证得的对象。两家在操作的层面是相通的：都禁止把那个极限变成一个可停留的住所。

但这道相通有它的边界，须同时标出，否则就把两个底色不同的框架抹平了。在操作的层面，两家同向禁止"达到"——这是对应。在本体的层面，两家禁止的理由相反：SAE 禁止达到，是因为那个极限真实存在却不可穿透，凿构的余项永不归零，主体在一条真实的方向上无限逼近一个实而不可及的极点；中观禁止达到，是因为那个极点在本体上根本为空，是一切戏论的遮影。一个是"路无尽头而方向真实"，一个是"路与目的地皆是假名"。所以这道相通不可被读成"两家在终点处是一回事"——它们在"禁止达到"这个操作上同向，在"那个被禁止达到者是实是空"这个本体判断上分裂。对应与断裂同时成立，正是接口分析此处的本分。

这道断裂不是本篇要抹平的麻烦，而是中心不可译处的一个面向，下文第六节会专门展开。此处先标出它的位置：两家在"禁达到"处握手，在"实抑或空"处分手。

---

二、破邪显正与"立"

三论宗的纲领是"破邪显正"。须先厘清这四个字，因为它最容易被读成"破掉错的、立起对的"——仿佛"显正"是在破之后另立一个正确的主张。三论的本义不是这样。吉藏的"正"不是另立一个正确之构；破邪即是显正，所谓"正"就是无所得的正观本身，是那个不把任何一说坐实为可停留之处的持守。换成 SAE 的话：这里的"正"不是一个终点的构，而是"任何构都不可闭合"这条纪律本身的持守。破与显不是两步（先破后立），是一件事——破到尽处，不立可停留之正，那个"不立"就是正。

这与 SAE 的凿同构：凿不为了立一个终点的构而存在，它就是那个不断否定、不让任何构坐实的动作。三论的"破邪显正"，在 SAE 的坐标里，是凿在论证维度上的一次彻底贯彻——破尽一切可停留之见，而不在破的尽头另立一个可停留的正见。

"立"的分歧

但正是在"立"这个字上，两家有一处真实的分歧，须说准。

三论的纪律是破而不立。这不是说三论什么都不主张——它主张破邪显正、主张中道为二谛之体、主张教门的种种施设；它"不立"的是一个可停留的正，一个可被执取、可被坐实为自性的终点之见。三论的戒律落在"立一个可住之正"这个动作上：凡立可住之正，即成新的邪。

SAE 的纪律不同。SAE 认为应当立——因为破不可能不产生新的构。每一次否定都落成一个新的、带着余项的构，这是凿构循环的常态：凿了就有沉淀，沉淀就是构。[4] 既然破必产构，假装不立反而不诚实；SAE 的戒律不落在"立不立"上，而落在"立的姿态"上——承认余项的立，才是站得住的立。立一个构，同时承认它有余项、不闭合、终将被下一刀凿动，这样的立是合格的；立一个构而宣称它闭合、完备、不留余项，那才是 SAE 要破的（在 SAE 的术语里，这叫构冒充律）。[4]

所以两家共见同一个结构事实——破不可能不产生新的构——但对这个事实的规范回应相反。三论以"不认领任何构为可住之正"来守破：连教门本身也只是施设，吉藏明言二谛唯是教门、不关境理，教门是为破执而设的工具，不沉淀为新的可住之处。SAE 以"认领构、但只认领承认余项的构"来守立：构是真实的，认领它，但认领的同时承认它不闭合。

这处分歧不是谁对谁错，而是同一道缝在戒律层面的现身（第六节会把它与本体层、反身层的现身并置，看清这是同一道缝的三次出现）。此处先标出：三论守的是"破而不立可住之正"，SAE 守的是"立而认领其余项"——两者都在回应"破必产构"这同一个事实，回应的方向相反。

附带一提，现代汉语有"不破不立"的成语，默认破是为了立——先破除旧的，才能建立新的，破服务于立。这是一种涌现的直觉，与 SAE"破必产构、构带余项"的方向相近。而三论的自述恰恰是"破而不立"——破不为立可住之正，教门立而不住。同一个"破"字，成语读出的是"为立而破"，三论读出的是"破而不住"。这个日常成语与宗派自述之间的小小错位，本身就是本篇中心张力的一个微缩版：建构的真实性，与对建构的彻底不执，在这里照面。

---

三、八不：总破自性执

《中论》开篇即是八不："不生亦不灭，不常亦不断，不一亦不异，不来亦不出。"接着是"能说是因缘，善灭诸戏论"。（末一对什译作"不来亦不出"，"出"即"去"、"往"之意，本节行文随义亦作"不来不去"，与原文同指。）这八个"不"，是本节要与 SAE 对接的核心。

须先定调，因为这里最容易出错。八不不是四把砍向四个不同靶子的刀。它是离八边的总遮——对"诸法实有自性"这一个执取，从八个方向同时遮拨。青目的注释把这层结构说得很清楚：略说八事，则为总破一切法；以不生不灭为根，其余六不皆归约于它——"若深求不常不断，即是不生不灭""以无生、无灭故，余六事亦然"。[CBETA, T30, no. 1564,《中论》卷一〈观因缘品第一〉青目释, p. 1c12 起] 也就是说，八不的结构是"不生不灭"为根，常断、一异、来去六不从不同角度归约于这个根；八边是总遮的表面枚举，不是八个各自独立的靶域，更不是四把功能不同的刀。

须在此前置一道护栏，并且要写进正文、不能只放脚注：以下把八不读作针对某种执取的拆解，是 SAE 的接口读法，不是复述青目或吉藏的历史科判。传统注疏对八不各对的对治自有其说；本节选取的是 SAE 透镜下、八不作为"总破自性执"的切入角度。两者不冲突，但须标明这是接口重读。

八边切入同一个执取

在 SAE 的坐标里，八不总破的，是把自我执取为自性之目的的那个执（前文已立此为中观的着力处）。八边是从八个角度遮拨同一个"自性之我"：

"常"是"我恒在"——把自我执为一个不变地持存的东西。"断"是"我灭尽"——把自我执为一个会彻底断灭的东西。"一"是"我同一"——把自我执为一个浑然单一、不可分的东西。"异"是"我割裂"——把自我执为与其构成部分截然两分的东西。"来去"是"我作为连续实体"——把自我执为一个从某处来、往某处去、跨时连续的实体。"生灭"是"我自性生灭"——把自我执为真实地自性生起、自性坏灭的东西。

八边是总遮的表面列举，结构上以不生不灭为根，常、断、一、异、来、去六者从不同角度归约于这个根；它们遮拨的，是同一个被坐实为自性的"我"。这正是青目"以不生不灭为根、余六归约"的结构——不生不灭遮拨的是"自性生、自性灭"这个最根本的执，其余六边是这个根执在不同维度上的展开。八不是总破，不是八个平权的独立角度。

这里须把"八边枚举"与"一根六归约"的关系点明，免得读成八个平权角度：八边是离八边之总遮的表面列举，结构上以不生不灭为根，常、断、一、异、来、去六者从不同角度归约于这个根。读者若把八不看成八把各打一处的刀，就错失了它"总破一个执取"的本义。

不生不灭这一刀

八不里，"不生不灭"这一刀须单独说，因为它是根，也因为它在本篇的着力处上最见力。

传统上，不生不灭的射程是一切法——一切法皆不自性生、不自性灭。本篇取的是它在着力处的高分辨率应用：遮拨"把自我执取为自性生、自性灭"这个执。须标明这是 SAE 的透镜选择——把一刀的射程收窄到一个高分辨率的靶上，是为了看清它在这一层如何起作用，不是说不生不灭本义只关乎自我。

在这个着力处上，不生不灭遮拨的是"我真实地自性生起、又真实地自性坏灭"这个执。遮拨它，并非教人相信"我不存在"或"我永恒不灭"——那是落入断、落入常，正是八不另外两刀要遮的。遮拨"自性生灭"，是让那个被执为自性的"我"松开：它既非自性地生，亦非自性地灭，它是缘起的、无自性的、不可坐实的。

在 SAE 的透镜下，这一遮拨最具松动力的地方在于：那个把自我执为自性之目的的层级，其全部抓取本是为了守护一个"实有的我"；一旦"我真实地自性生灭"这个底层执被遮拨，守护的必要性就松动了。主体不再须把全部力量押在守护这个自性之我上，遂得以转向更高的位置——承认他者也是目的本身。须注意这里的措辞：不是"八不本身逼人转向他者"，而是"自性之我的执取一旦松动，主体不再须把力量全押在守我上，遂得以转向他者"。八不做的是遮拨那个执；转向是执松之后腾出的可能，不是八不直接施加的推力。

（一处教内的印证可在此引入：《大智度论》以"过一切语言道，心行处灭，遍无所依，不示诸法"界定第一义悉檀，并以"不生不灭，法如涅槃"等偈指向诸法实相。[CBETA, T25, no. 1509,《大智度论》卷一释"第一义悉檀", 页行待终校] 后世可将八不与此实相并举、概括为"八不即实相"，但此四字是概括而非论中连续原句。"遍无所依、不示诸法"正可见这一路的遮诠性质——它不指示一个正面的实相，只遮拨一切可执之处。此句亦为后文论"实相"一词的遮诠用法预作铺垫。）

与其余六边

六边何以能"归约于不生不灭"这一根，须把传导说出来，否则"归约"只是一句口号。机制是这样：常断、一异、来去这六边，各自要成立，都暗自预设了一个"自性地生起、自性地存续"的我作为底托——"常"是这个我在时间上不变地持存，"断"是它彻底断灭，"一""异"是它在与构成的关系上浑然或割裂，"来""去"是它作为一个实体跨时位移。可是一旦"不生不灭"把那个根上的执拔掉——确立了这个我并非自性地生起、并非自性地坏灭——那么时间上的持存或断灭、关系上的同一或割裂、位移上的来或去，就在同一刻全部失去了它们的底托：没有一个自性地生灭的"我"，也就没有那个我去持存、去断灭、去同一、去来去。拔掉自性生灭这一根，另外三个维度上的执便随之解体，不必逐一另破。这就是"以不生不灭为根、余六归约于它"的力学：六边不是六个各自独立、需要分别遮拨的靶，而是同一个根执在不同维度上的投影，根一动，影皆散。

以下其余六边各遮一面，皆在此根的传导之下略说。

不常不断这一对，遮拨的是把自我执为"恒在"或"灭尽"的两端。在时间的维度上，它遮拨的是"我是一个不变持存的实体"（常）与"我是一个会彻底断灭的实体"（断）——两端皆是把自我坐实为自性。遮拨之后，剩下的是缘起相续：自我既非常住不变，亦非断灭无续，而是缘起的相续——变而不失其续。青目正以"若一则无缘，若异则无相续"言此相续之理。[CBETA, T30, no. 1564,《中论》卷一〈观因缘品第一〉青目释, p. 1c12 起]

不一不异这一对，遮拨的是把自我与其构成执为"绝对同一"或"绝对割裂"。我与构成我的种种，既非浑然一体（一），亦非截然两分（异）；遮拨这两端，逼显的是缘起的连结——自我是一个由种种缘起连结而成、却不可坐实为单一自体的结构。

不来不出这一对，遮拨的是把自我执为一个"有来处、有去处"的连续实体——仿佛有一个自性之我从某处来、往某处去。遮拨它，是松开那个跨时连续、自性持存的"我"之执。

六边各遮一面，皆随不生不灭这一根的拔除而解体；八边合起来，是对同一个"自性之我"的总遮。这就是八不在本篇着力处上的接口读法——一个执取，八个角度，一次总破。

（按：本节凡言八不"遮拨自我之执"，皆在前文所立边界之下——这是 SAE 选取的着力处，中观八不的全幅射程是一切法自性执，非独自我。又，传统科判对八不各对的对治另有其说，本节为接口重读，不复述科判。）

---

四、四重二谛与否定的四个步位

中观学在汉地的系统化，吉藏《大乘玄论》是一个高点。其中二谛被组织为层层推进的结构：俗谛立一说，真谛否之；前一层的真谛降为后一层的俗谛，再被新的真谛否定。吉藏在卷一将这一结构展开为多重，其中一条可被清晰地读为四个步位——这是本节要与凿构循环对接的对象。

须先声明两重边界。其一，文本边界：《大乘玄论》卷一将二谛义分为十重，本节抽取的是其中"有无"一重之下可辨认的一条四步序列，不否认十重的整体组织，也不否认《二谛义》中三重表述的存在。抽取一条序列用于接口分析，不是宣称吉藏的文本本就是一张四格表。其二，对接对象边界：本节对接的不是中观空性的全幅（一切法无自性），而是这一结构与凿构循环四相生成的步位对应；空性的全幅射程属于中观，本节透镜只照这一段结构。

四步的吉藏原文如下。第一重，以有为俗谛、空为真谛——立起"有"与"空"这一对。第二重，以有空为俗谛、非有非空为真谛——否定这一对。第三重，"二与不二皆是世谛，非二非不二名为真谛"。第四重，"此三种二谛皆是教门，为令悟不三，无所依得，始名为理"。（此四重二谛之分，见吉藏《大乘玄论》卷一论"二谛义"，《中观论疏》卷二末亦开此四重。[CBETA, T45, no. 1853,《大乘玄论》卷一, 页行待终校]）

这四步与凿构循环的一个已发表结构逐步相应。在凿构循环的回溯框架中，最原始的处境（0D 浑沌）下，否定追问自身，生成有、无、非（有和无）、非（非有和非无）四相——先立肯定与否定之对，继而否定此对，再否定"否定此对"而不令其塌缩，最后整体收束。[4] 两者的对应是步对步的，但有两处必须停下来看清，因为它们恰是这篇接口分析的承重点。

第一处在第三重。吉藏在此不写"非有非无"的简单重复，也不写一个会在逻辑上塌回起点的双重否定——若把第三重写成"非不有、非不空"，"非不有"在形式上即退回"有"，整个推进就垮回第二重。吉藏选择的是"非二非不二"：被否定的不再是"有"或"空"这一阶的谓词，而是"二/不二"这个框架本身。否定的焦点跃迁了一层。这正对应凿构循环中一条沉淀于 3DD 的约束——双重否定不返回正面，相异者之间留下一道不可消去的间隔，不重合，亦不可逆。[4] 吉藏"非二非不二"是这道间隔在论证语言里最清晰的一次实现：它否定的是框架而非框架内的谓词，故不塌回；它是这一对应在历史文本中能找到的最强证据。

第二处在第四重，也是更要紧的一处。第四重不同于前三重：前三重各自立一真谛、又各自被降为俗谛而推进；第四重却说前三重"皆是教门"，自身落在"无所依得"。这意味着第四重不是第五个真谛，不是又一次否定，而是整架阶梯的收束——前三重作为"教门"整体沉淀为构，而"言忘虑绝、无所依得"是这一沉淀所留下的、无法再被收编的余项，指向那个不可被任何一说所得者。须就此标一道边界，因为它正是第六节要展开的那道缝的预演：这个终端余项在中观一侧指向的是"无所依得"（空——路与目的地皆假名），在 SAE 一侧，凿构循环的终端余项指向的是那个实而不可及的极限（物自体）；两者在"论证收束后留下一个指向不可坐实者的终端余项"这个结构位置上相应，但所指向者一空一实，本体上分手。此处只标这个结构位置上的对应，不把两个所指预先合并——那道断裂留给第六节。

这里要分清两种余项，否则会误读，而这一分辨正是第四重的关键。前三重之间的余项是推进性的：每一重之真降为俗，正是迫使下一重否定运作的力量——这是凿构循环常态的余项，发动下一刀。第四重的余项不是这一种。它不推进出第五重（没有第五重，强行追问只会得到"言忘虑绝"——这恰是"不存在第五步"时阶梯崩塌的样子）。它是教门收束之后留下的终端余项：它不再驱动这架论证阶梯继续转动。中观的论证阶梯在这里完成了它的全部工作，彻底悬空——言忘虑绝、无所依得，是般若走到尽头处的究竟，不是一个待续的中途。在 SAE 的探针图谱里，这种"论证自我耗尽"的终端坐标，恰与禅那一路"切断思维的顿断"在拓扑位置上背靠背相接——两者各在自己独立、自足的体系内，从不同的路走到了同一种不可言说的边界。这是两个独立终端在坐标上的相邻，不是一方把人交给另一方接手。

由此可以澄清一个常被误解的点：吉藏的真谛（"胜义"）并不是一个可停留的住所。"无所依得"四字本身就否认了它是一个被抵达的位置。它是阶梯收束之后留下的余项之指向，不是一个被占据的高处。胜义不是住所，是收束加余项——在这一点上，吉藏自家的顶点与凿构循环"余项守恒、无构闭合"的纪律相合。

值得一提的是，吉藏自己把这四重配给不同根机："利根闻初即悟道，不须后二；中根闻初不悟，闻第二方得入道；下根转至第三始得领解。"这说明四重二谛在吉藏那里本就带有"对不同的人说不同深度"的适配意味——同一结构，因听者而展开到不同的层。这一点与后文论三法轮时的"对位而说"是同源的，可相互印证。

最后须防一处误读，因为它关系到整篇的诚实。Self-as-an-End 是一个承认"构"有真实地位的框架：每一次否定都落成一个新的、带着余项的真实构，阶梯是构不断涌现的阶梯。这与黑格尔式的"综合"正相反——综合把否定吸收进一个更高的统一，余项在综合中被消去；而凿构循环第四步留出的恰是余项的空间，收束不等于统一。[4] 也正因如此，第四步在命名时险些被叫作"综合"而被这一区别拦下：把收束写成综合，就等于把它读成了一个更高的真理，那不是这里的意思。在中观一侧，这个区别有它的对应：三论"破而不立"恰恰拒绝把"中道"读成一个调和的更高合题——若有人把中道理解为有与空之上的第三者、一个综合，三论的纪律会否掉这种读法。两侧在"不让收束变成更高统一"这一点上同向。

（按：本节将"四步"用于接口分析，是把吉藏可层层推进的二谛结构重排为一条可对接的序列；"步"是分析的取景，不是宣称吉藏原文以此四步为其固有语法。同理，下一层之真在上一层被降为俗的"遣除"方向，与凿构循环中四相"逐步生成"的方向看似相反，但两者由同一件事统一：每一步都靠否定前一步而成立——遣除与生成，是同一个否定动作的两种描述。又：本节"间隔律""构建律"为称引方便所立之名，其内容援引已发表文本，正式命名以方法论后续版本为准。）

---

---

五、二藏与三法轮

吉藏的判教里，有两套对教法的分类，本节分别对接，并须从一开始就把两套轴分开——它们分的是不同的东西，混在一起会彼此打架。

二藏：内容的方向

第一套是二藏：声闻藏与菩萨藏。须先给佛教一侧的语法，再做接口。声闻乘的特征是自度——以自身的解脱为务，不以度他为目标；菩萨乘的特征是利他——以普度众生、与一切众生同趋觉悟为行。这是大小乘在所诠法门与发心方向上的传统分判。

在 SAE 的接口读法里，这两者呈现为内容的两个方向：声闻藏呈现为"拆己向内"的解脱语法——把着力放在松开自身的执；菩萨藏呈现为"向他而开"的承认语法——把着力放在承认他者、与他者同趋。须标明这是接口重读：把声闻乘的"不度他"读作"向内拆己"，是 SAE 从结构方向上的重述，不是二藏的原义；二藏原是依发心与所诠而分，本节取其结构方向。（依系列纪律，本节不判二藏高下，亦不用"小乘"一词，用"声闻乘"。）

三法轮：位置的适配

第二套是三法轮：根本法轮、枝末法轮、摄末归本法轮。这一套分的不是内容方向，而是教法对不同主体的适配——同一个觉悟的内容，对处在不同位置的主体，须以不同的深度与方式来说。

须在此立一道边界，它是本节最要紧的一道：三法轮是吉藏对"教法如何因机而异"的判分，本节对接的是这个"因机施教"的适配结构，不是把三法轮读成一条从低到高的位置序列。下文虽会把三法轮的三个环节对应到主体的不同位置，但那是在说"什么样的教法适配什么位置的主体"，不是在说"吉藏排出了一条主体逐级上升的阶梯"。这道边界须守住，否则就把吉藏的因机施教，误读成了一套发展序列。

在这道边界下，三法轮的接口如下。

根本法轮，是佛最初、最高的直陈——华严所示的、事事无碍重重无尽的境界，是佛自内证的全幅呈现。在 SAE 的坐标里，它对应的是"承认他者亦为目的"那一层（15DD）的视野，并以遮诠方式提示那个不可对象化的极限（这一极限本篇不展开，留待专论）。[4] 须特别说明这一点：根本法轮所示的不是一套可以像信号一样发送、接收的"内容"——那个最高的境界，本性上只能在主体之间互为目的的现行交涉中显现，无法被单向地接收为一份信息。

枝末法轮，是佛随后为接引而设的教法——三乘的施设，有果可求、有位可证、有次第可循。在 SAE 的坐标里，它对应"自我执取自身为目的"那一层的语法（14DD）：求果、求乘、有目标可达。

摄末归本法轮，是把前面接引来的主体收归于本——会三归一，令一切众生同入佛慧。在 SAE 的坐标里，它对应"承认他者亦为目的"那一层（15DD）：主体松开自我之果的执，转而以他者、以众生的同趋为归。

须把根本与摄末归本的差别说清，因为两者都落在"承认他者亦为目的"那一层附近，容易混。根本法轮是最初的全幅直陈——一上来就示以最高的境界，因此大多数主体接不住。摄末归本法轮是回摄——把那些先经枝末接引、已能松开自我之果的主体，重新收归于那个最高的境界。一个是高广的初示，一个是接引之后的回收；两者朝向同一个境界，但一个在序列之初、一个在序列之末，路径上的位置不同。

"如聋如哑"：一个适配错位的解释

华严会上，二乘"如聋如哑"——对佛所示的最高境界不见不闻。这个意象常被用来说明根本法轮之高。在 SAE 的接口读法里，它是一个适配错位的现象，可以这样解释——但须先标明：此处把二乘读作处在"自我执取自身为目的"那一层的主体，是 SAE 的接口读法，不是说二乘在佛教里就等同于某个心智层级。

在这个读法下，二乘是以自求果证为务的主体——其心智的运作方式，是围绕"我要证得什么果、登什么位"而组织的。华严所示的那个境界，本性上是主体之间互为目的的现行交涉；而以自求果为务的运作方式，本身就把"以他者为目的"这件事排除在它的运作之外——它不是没接收到一个高频的信号，而是它的运作方式里没有"互为目的"这个交涉的位置。所以"如聋如哑"不是接收装置的故障，是运作方式的不相应：一个围绕自我之果组织起来的心智，遇到一个本性为互为目的之交涉的境界，二者不相应，故不见不闻。

正因这种不相应，佛退回到那个以自我之果为务的主体能够相应的语法——枝末法轮，给出三乘、果位、次第。这不是把那个最高境界"调低频率"，而是给出一套那个主体的运作方式能够运转其中的教法。须把这里的关系说准，不让它读成一部成长史：当一个主体的运作锁定在"单向求自果"的构型上时，"以他者为目的"的协议就被这个构型从结构上排除在外；唯有这个构型本身发生松动、卸下那种绝对的抓取，"摄末归本"的语法才与它达成相应——这是构型的解除封闭，不是里程的逐段积累。摄末归本之所以"相应"，不是因为主体走完了一段路程、抵达了某一站，而是因为它的构型从"单向求果"松动成了"可纳他者"；构型一变，相应即成，无所谓先后远近。

这个解释里，"二乘处在自我之果那一层"这个判定，是由主体的运作方式（围绕自我之果组织）独立给出的，不靠"如聋如哑"来反推；"如聋如哑"是这个独立判定的一个印证，不是它的依据。这样，解释就不是用一个待证的对应去解释现象、再用现象回证那个对应。

须收束于本节开篇的边界：以上把三法轮的三个环节对应到主体的不同位置，说的始终是"什么深度的教法适配什么运作方式的主体"，不是"吉藏排出了一条主体逐级攀升的序列"。三法轮是因机施教的适配结构；它的方向性（根本、枝末、摄末归本之间有先后）是教法施设上的方向，不是一套主体发展阶梯。把它读成后者，就越过了这道边界。

两套轴不可混

最后须回到开篇：二藏与三法轮分的是不同的东西。二藏按内容的方向分（拆己向内／向他而开），三法轮按教法对主体的适配分（什么深度适配什么运作方式）。两套轴交叉而不重合：一个声闻乘的教法（二藏里偏自度）也可以是枝末法轮（三法轮里的接引施设），二者分在不同的轴上，不冲突。把两套轴显式分开，"声闻藏＝拆己向内"与"二乘于华严如聋如哑"才不会被读成矛盾——前者说的是这套教法的内容方向，后者说的是这套运作方式与某种教法的适配错位，分属两轴。

（又：本节三法轮与本系列第一篇所论的"三转法轮"是两套不同的三分——前者是吉藏判教里"教法因机而异"的判分，后者是另一套关于主体位置确立的三分。两者须显式区分，不可混用。三法轮的文本出处为吉藏《法华游意》（CBETA, T34, no. 1722, 页行待终校）；另参《法华义疏》（CBETA, T34, no. 1721）"隐实／起权／收末归本"的展开。）

---

---

六、不可对接之处

接口分析最要紧的一节，是说清两个框架在哪里无法互相翻译。这一处不是分歧的附录，而是分析的核心；把它说准，比把对应说满更重要。

分歧不在"无自性"。中观对自性的否定——自性意谓不依他、不从缘、无变异——Self-as-an-End 是全盘接受的；它对"任何构都不能自我闭合、总留余项"的主张，与中观"诸法无自性"在结论上相遇。两者进路不同：中观从缘起的依存推出无自性，Self-as-an-End 从构的遮蔽推出不闭合（凡构必有所蔽，故余项不空）；同一结论，不同引擎。真正的分歧在更里面一层：在第二谛上，Self-as-an-End 主张构是带着真实牵制的沉淀物——构虽无自性，却不是零、不是纯然的假名施设，它有真实的作用与余项。中观会从空性一侧问：这个"真实牵制"，是不是仍然偷藏了一缕细微的自性执？而 Self-as-an-End 会从操作一侧回应：执行否定的，本身就是一个真实运作的凿构过程——一个要持续地建构、并且承认每一次建构都留下余项的体系，无法把这个"真实牵制"也同构地翻译成纯然的假名，因为说出这个立场、贯彻这个否定的，正是一次真实的构。但须立刻说准这道回应的分寸：这不是说中观"做不到"或"缺了什么"——中观本就不以持续建构为目标，它要的是破尽自性、令无所得，在它自己的目标下，把一切都还原为无自性的假名是彻底而自洽的。两家不是一方完整、一方残缺，而是两套目标各自完整、在此不可互译：中观从空性侧、Self-as-an-End 从操作侧，各以自己一套完整的纪律，看对方多出了或少了一点东西。这一来一回，双方都无法把对方的话翻译成自己的语言而不失真。这是本篇判定的、最深的一道不可对接之缝。

这道缝不是只在一个地方现身。它在三个层面同时出现，三处同构，互为印证。

其一，在本体的层面，载体是《中论》的著名一颂："众因缘生法，我说即是空，亦为是假名，亦是中道义。"须先说清：此颂中"缘起、空、假名、中道"四者是同一桩事的四个名相，是对同一件事的四面之说，不是先缘起、再空、再假名、最后中道的时间次第。Self-as-an-End 为接口分析的方便，把这四名重排为一个四拍的读法来与凿构循环的结构对照——但这是分析的取景，不是原颂的语法。重排之后，分歧落在最后：中观把"假名"作为无自性的施设之名来用——它指事而不赋它自性的重量，名以指事、事过而名不立自体（这并不是说世俗的运作没有效力：中观的二谛仍保留世俗层面的可说、可行、因果与教化，假名只是无自性，不是无作用）；Self-as-an-End 则把"构"读作一个虽无自性、却有真实牵制之力的沉淀物——它虽无自性，却真实地约束着后续。同一桩缘起，中观读出的是无自性重量的施设，Self-as-an-End 读出的是有牵制力的真实——这是两家在同一颂上的分手。

其二，在持守的层面。Self-as-an-End 有一条自己的纪律：承认余项的"立"才是站得住的立（这里说的"好"或"站得住"，仅指 SAE 内部操作上的合格——一个立是否承认了自己的余项，是 SAE 自家的工序标准，不是善恶评价，也不是中观会承认的某种正面之"正"）——因为破不可能不产生新的构，与其假装不立，不如认下这个带着余项的构。但这条纪律在中观的"离四句"面前无法自持。"站得住的立"本身预设了一个评判"立"的标准（是否承认余项），而这个标准又是一个立场、一个构；中观会追问这个"站得住"是自生的、依他的、共的、还是无因的——它正是中观要破的那种"满足某个标志的构"。这里两家不能互译，而 Self-as-an-End 在此不退让，也不必退让；但要把不退让的根据说准。它的根据不是"我已付过某种代价、所以有权保留这个立"——那是一种把力学事实伪装成产权声明的说法，须避开。它的根据是：那个构之所以能在中观的归谬下不垮，是因为它确实在运作、确实留下余项，而不是因为谁声明了什么。Self-as-an-End 认下这个构，认下的是一件正在发生的事，不是一项权利。至于中观——从 Self-as-an-End 的操作一侧看，它不提供一个可供持续建构的根基；但这不是它的缺陷，而是它"破而不立"的纪律所付的代价，也正是这一纪律的力量所在。它要的本就不是建构，用"有没有根基"去衡量它，是拿错了尺。

须再补一层边界，以免误读温度：以上"代价""根基"之说，是从 Self-as-an-End 的操作侧描述中观，不是判中观为错或缺。中观破而不立，在其自身的目标下是彻底而自洽的；Self-as-an-End 承认余项之构，在其自身的纪律下也是彻底而自洽的。两者不是一对一错，是同一道缝的两壁。

其三，在反身的层面，这一处留到末节谈中观如何反照 Self-as-an-End 时再展开。它问的是同一道缝最后转回到 Self-as-an-End 自身的形态：当中观把否定贯彻到底，它会不会指着 Self-as-an-End 据以立论的那个最终否定——那个被采用为公理、连自身之否定都不再是的"否定性"——问一句"这难道不也是又一个被执取的立场、又一个自性"。这一问恰恰打在 Self-as-an-End 自我豁免的方式上：在中观的判准下，这个风险会这样显现——若那个最终否定被做成一个"免疫于否定"的点，而"免疫于否定"在中观看来正是判定自性的标志之一，那么它在中观尺度下就显得像是未能做到"连空都空"。这是中观尺度下的一种显现，未必是 Self-as-an-End 的实情；末节将就此展开，并正面接住它——此处只标缝的位置。

三层是同一道缝的三次现身：本体层是"假名无自性重量"对"构有真实牵制"，持守层是"破而不立"对"认下余项之立"，反身层是中观对 Self-as-an-End 最终立场的反诘。把它们叠在一起看，这道不可对接之缝就不是一句软话所能抹平的——它有结构、有三个独立的支点。接口分析到此处的本分，是把这道缝照清楚、标出它的两壁，而不是架一座桥假装两边是通的。

---

七、残余正性检查

本系列每一篇，都要做一道"残余正性检查"：查一查被对接的传统里，有没有什么东西，表面上是遮诠的、否定的，骨子里却偷偷立了一个正面的、可停留的终点——一个被 SAE 判为"不可坐实"却仍被坐实了的东西。这道检查是接口分析诚实性的一部分：不是去挑对方的错，而是把那些容易滑向正性的地方标出来，看它们究竟滑了没有。

须先说明本篇与前篇的一个不同。前篇（唯识）里有一个"引自果"的检查项——种子如何生果，那里藏着一个被 SAE 判为不可译的正性结构。本篇不设这一项：引自果是唯识种子学说里的概念，三论没有这个架构，本篇不为了与前篇对齐而硬找一个对应。本篇的残余正性检查，落在两处中观自己的语汇上：寂灭，与实相。

寂灭：一道边界发现

中观讲"善灭诸戏论"——戏论的息灭。这里要查的是：这个"灭"，是不是偷偷立了一个正面的终点？是不是说，到了某处，一切就彻底停了、归于一个寂灭的终态？

查下来，这一处不是偷渡正性，而是一道边界发现——它标出了 SAE 与中观在一个根本问题上的分界，而这道分界本身是干净的、可标的。

关键在于分清"戏论"是什么。戏论（prapañca）不是泛指一切概念活动，它特指那种依二取而起的增殖——有与无、生与灭、一与异、来与去，成对地立、成对地执，由此层层增殖。这种增殖是对偶式的：立一边，就有它的对边；执一端，就牵出另一端。中观说的"灭戏论"，灭的正是这种对偶式的增殖——不再以"立一边执一端"的方式去增殖概念。

而 SAE 的凿构循环不是对偶式的。它的运作是：凿一个构，构留下余项，余项驱动下一次凿——这是一个不对称的接续，不是一对一对的对立。[4] 驱动它的不是"一边对一边"的张力，而是构与余项之间的不对称（余项是构所未能收摄者，余项驱动下一轮，而不是构的某个对立面驱动下一轮）。[4]

于是两者所要"停"的，是两台不同的机器。中观要停的，是对偶式的戏论增殖——这台机器可以、也应当被止息。SAE 说不停的，是那个不对称的凿构接续——只要有这样的心智在运作，否定就不会停机，余项守恒。[4] 中观停的那台（对偶增殖）与 SAE 不停的那台（不对称接续），不是同一台机器；中观把对偶式增殖止息掉，丝毫不触及那个不对称的凿构接续。所以这里没有正面对撞——两家各说各的那台机器，一个该停，一个不停。

这道分界，恰由中观自己的文本守住。《中论·观涅槃品》界定涅槃："无得亦无至，不断亦不常，不生亦不灭，是说名涅槃。"[CBETA, T30, no. 1564,《中论》卷四〈观涅槃品第二十五〉, 页行待终校]"无至"——没有抵达一个别的状态；"不断"——不是断灭。涅槃与生死的差别，不在于到了一个寂灭的终态，而在"受诸因缘"与"不受诸因缘"——不再以二取因缘的方式运转。[CBETA, T30, no. 1564,《中论》卷四〈观涅槃品第二十五〉, 页行待终校] 这正合"灭的是对偶式的戏论，不是凿构接续的停机"：涅槃不是一台机器彻底关停，是不再以对偶二取的方式运转。

故寂灭这一处，不是偷渡正性，是一道边界发现：它标出"该止息的对偶增殖"与"不该止息的不对称接续"之间的分界——中观在教理层止息前者，Self-as-an-End 在力学层保留后者；两家各在自己的体系内彻底，而所指的那台机器并非同一台。须守一道温度边界：以上"两台机器"之分，是 SAE 一侧的力学解析；中观自身并不以"对称/不对称的机器"来说戏论。说中观所灭者为对偶增殖，是 SAE 的接口描述，不是把 SAE 的力学读法塞进中观的口中。

实相：一个名词的对象化风险

第二处是"实相"。"实相无相"——实相是无相的，这本是彻底的遮诠：实相不是任何一个可执的相。骨子里，这一处的主张是 Rho-合规的（用 SAE 的话说，它不立一个可停留的正面终点）。僧肇以双遣守之："非有非真有，非无非真无。"[CBETA, T45, no. 1858, 僧肇《肇论·不真空论》, 页行待终校] 双遣不落两边，不立终点，干净。

但风险不在主张，在"实相"这个词本身。汉语的"实相"带一个"实"字；"实相"作为一个名词，极易被对象化——被读成"诸法背后那个真实的样子"，仿佛有一个无相的、真实的"实相"在那里，是般若所要去契合、冥符的终极底座。这一被对象化的读法，才是 SAE 要标的正性风险：它把一个本应是遮诠的词，凝固成了一个可停靠的真实之处；更隐蔽的是，它在认识论上暗示了一种"般若冥符实相"的、零余项摩擦的契合——仿佛主体的认识可以与那个实相严丝合缝、毫无遮蔽地相应。而 SAE 的纪律是：任何认识的响应，本身也是一个构，也有所蔽、有余项；没有一种零摩擦的、把对象完全收摄无遗的契合。[4] "般若冥符实相"若被读成这种零摩擦的契合，就违了"凡构必有所蔽"。

这一风险，同样由中观自己的文本可守。《大智度论》以"过一切语言道，心行处灭，遍无所依，不示诸法"界定第一义悉檀，并以"不生不灭，法如涅槃"等偈指向诸法实相；后世可概括为"八不即实相"，但本文不把这四字当作《大智度论》的连续原句。[CBETA, T25, no. 1509,《大智度论》卷一, 页行待终校] 要紧的是：实相在此被界定为"遍无所依、不示诸法"——它不依任何处、不指示任何法，正是一个遮诠的标记，不是一个正面的本体。所以本篇凡用"实相"一词，皆取其遮诠用法：实相即诸法无自性的如实、即心行言语断处，不是诸法背后的一个真实之体；并须明标 SAE 不接受"般若冥符实相"式的零摩擦认识——认识亦是构，亦有所蔽。

故实相这一处的结论是：其主张层（双遣、无相）不偷渡正性，是 Rho-合规的；其风险在"实相"作为承重名词在汉语接受史与认识论操作中的对象化倾向。本篇以遮诠用法的标注、以反"零摩擦契合"的护栏，守住这一处。

检查的结论

两处查毕。寂灭是一道边界发现——它标出两台机器的分界，不是正性偷渡。实相的主张层 Rho-合规，名词层有对象化风险，已以双重护栏（遮诠用法＋反零摩擦契合）守住。

须说明这道检查的意义，并守住温度。中观是本系列对接的传统里，最难在其中找到"被坐实的正性"的一个——它的整个工夫就是破尽可坐实之处。本篇这道检查，若最终在中观处查不出真正的正性偷渡，这不是检查的失败，而是一个有意义的结果：它支持了把中观读作"那条不可坐实之律的探针"这一定位。但须把话说准——查不出正性，是这一定位的一项强支持，不是它的独立证明。检查的本分是把容易滑向正性的地方标清楚、看它滑没滑，而不是预先断定它必滑或必不滑。

---

八、空的三种偏锚：僧肇的三家批判

僧肇《不真空论》批判了当时的三家——本无宗、心无宗、即色宗——指出它们各自在某处把"空"安放偏了。这三家批判，在 SAE 的坐标里，恰好对应着误置否定的三种典型滑坡。本节把这层对应标出，作为一道防火墙——它显示，汉地早期的中观接受者，已经在系统地抵抗把"空"实体化、或半截化的种种偏差。（本节标题用"偏锚"而非"错锚"：所谓"偏"，是相对僧肇自己所立的"不真空"准星而言的偏离，下文"偏""错"皆在此意义下读，且皆是僧肇的判、非 SAE 的发明——详见下段。）

须先框清一处，否则会误读温度。本无、心无、即色之被判为"错"，是僧肇自己的判——僧肇本就是在批判这三家。SAE 在这里做的，不是自己去判这三家为错，而是跟着僧肇已有的判，把这三家被判错的地方，对应到 SAE 坐标里的某个结构位置上。"错锚"之"错"，是僧肇的判，不是 SAE 的发明；下文凡言某家"错"在某处，皆在此意义下读。

本无宗，僧肇批其"情尚于无"，把"无"看得太重，仿佛在万化之前、之上，有一个原初的"无"作为本体。在 SAE 的坐标里，这对应着把否定锚错了地方——把否定倒退成一个原初的、实体性的虚无，仿佛"空"是某个先于一切的本体之无。[4] 这是把"空"实体化为一个起点的本体，而 SAE 的否定不是一个本体的无，是一个不断运作的操作。

心无宗，僧肇批其"无心于万物，万物未尝无"——只在主体心上做空（不起执著），却不在物上做空（万物仍实有）。在 SAE 的坐标里，这对应着一种半截的悬置：只在局部（自身的心）切断了执，却没有贯穿到那个心与物相互构成的连结——退回成一个孤立的观察者，把自己从那个相互构成的连结里摘出来，却让连结的另一端（物）保持实有。[4] 这是悬置只做了一半，切断了反馈，退入孤立。

即色宗，僧肇批其"言色不自色，未领色之非色"——知道"色不自有"（色待缘而成），却没有把这一否定贯彻到底，仍在某处保留了"色"的某种实。在 SAE 的坐标里，这对应着一种不彻底的否定：否定做了，却没有凿到底，在承载那个色的层面上留了一手，没有把那个被执的承载面也一并剥离。[4] 这是凿得不彻底，留了一个未被否定的底。

三家三种错，恰是误用否定的三种滑坡：本无是把否定退成实体的虚无（锚错成一个本体之无），心无是把悬置只做一半（切断反馈、退入孤立），即色是把否定做得不彻底（留了一个未凿的底）。僧肇以"不真空"——诸法不真，故空——破此三家：不真，是说诸法无自性（故不落本无的实体之无、不落即色的留底）；空，是说这无自性遍于心物（故不落心无的半截）。在 SAE 的坐标里，这三破恰好守住了三条线：否定不退成实体的虚无、悬置要贯穿到底、凿要凿彻。

这道对应的意义，是一道防火墙：它显示，把"空"读作那条不可坐实之律——而非读作一个实体的虚无、一个半截的悬置、一个留底的否定——这件事，汉地早期的中观接受者已经在系统地守护。须再次申明边界：僧肇不作三论宗的内部判据，他是罗什译场、汉地早期中观语境里的人物；本节引他，是作为汉地空论如何抵抗"空之实体化"的前史接口，不是把他的判当作吉藏体系的内部标准。又，本节所对应者，是僧肇已判为错的三家之"错处"如何落在 SAE 的结构上，非 SAE 自行判某家为错（承本节开篇所框）。

（与上节的联动：本节守的"实相不实体化"与第七节守的"实相不对象化"是同一道护栏的两面——僧肇正是"实相"一词的承重者之一，故本节为他立的遮诠边界，与第七节为"实相"立的遮诠边界，须一并读。）

---

九、反身：中观反照 Self-as-an-End

接口分析若只让 SAE 去读中观，是不完整的。这最后一节，让中观反过来照 SAE——尤其照向 SAE 那个最根本的立足点，问一个 SAE 必须正面接住的问题。

这一节接的是第六节标出的、那道中心不可译之缝的第三个面向——反身的面向。前面已说，这道缝在本体层（假名无分量／构有牵制）、戒律层（破而不立／认领余项之立）各有现身；现在是它转回到 SAE 自身的形态。

两种自我豁免

问题是这样浮现的。SAE 立论的最终立足点，是那个被采用为公理的"否定性"——那个连自身之否定都不再是的、最根本的否定。[4] 中观会指着它问：这难道不也是又一个被执取的立场、又一个被坐实的自性？你 SAE 破了一切构的自性，那么你据以破的这个最终否定，它自己是不是一个构、一个自性？

这一问之所以锋利，在于它打中了两个框架处理"自我豁免"的方式之别——它们各自如何使自己的最终立足点免于自我消解。

龙树的方式是减法式的。《回诤论》里，龙树面对"你的空论本身是不是一个有自性的论"这一诘难，并不简单说"我无宗"四字，而是说"若我宗有者，我则是有过；我宗无物故，如是不得过"，并释为"我无宗相，何得咎我"——我没有一个可被指认、可被坐实的宗，故你那套"凡主张皆需自证"的责难，落不到我身上，因为我无所立。[CBETA, T32, no. 1631, 龙树《回诤论》, 页行待终校] 减法式的豁免，是靠"连空也空"——连"空"这个说法本身也不被立为一个有自性的东西，故没有一个可被攻击的靶。

SAE 的方式，初看是表演式的：你要否定那个最终否定吗？你否定它的这个动作，本身就是那个否定性在运作——你越是试图否定它，越是演示了它。这个豁免方式很有力，但它有一个危险：它把那个最终否定，做成了一个免疫于否定的点——任何否定它的尝试都反过来证成它。而"免疫于否定"恰恰是中观判定"自性"的标志之一：一个不待他、不可破、否定它反而成立它的东西，在中观的判准下，就被认作一个自性。于是这里有一个反讽：SAE 若纯以表演式自我豁免，它据以破一切自性的那个最终立足点，在中观的判准下反而最显得像一个自性——在那把尺子上，它恰恰未能做到中观所要的"连空都空"。须强调这是中观尺度下的显现，是中观这一问的锋利所在，而非 SAE 已然承认自己有一个自性；下文要看的，正是 SAE 在自己的语言里如何回应这把尺子。

SAE 如何接住

SAE 接得住这一问，但接的方式，不能是把表演式贯彻到底（那只会把那个最终否定坐得更实）。接住它，须承认 SAE 自己也有减法式的一面。

SAE 的资源里，真正接得住这一问的，是两条。其一，那个最终否定，按其本性，连它自身的否定都不是——它不是一个可被肯定、也可被否定的对象，它先于这种肯定/否定的对立。[4] 这一条，结构上正与中观的"连空也空"相通：那个最终否定不是一个可执的东西，正如空不是一个可执的东西。其二，SAE 自认：本文是构，本文已经说得太多了。[4] SAE 不把自己的论述当作一个完备的、不留余项的终极之说——它承认自己的论述也是构、也有余项、也终将被下一刀凿动。这一自认，正是 SAE 版的减法式自谦：它不把自己的最终立足点坐实为一个不可破的、完备的自性之点。

所以 SAE 不是纯表演式的。它既有表演式的一面（否定那个否定的动作演示了那个否定），也有减法式的一面（那个最终否定连自身之否定都不是，且本文自认是构）。接住中观这一问的，是后一面：正因 SAE 承认自己的最终立足点不是一个可执的、完备的自性之点，正因它自认本文也是构、也有余项，它才没有把那个最终否定坐成一个免疫于否定的自性。这与第六节戒律层的那一处呼应——SAE 认领自己的构，认领的是一件正在运作的事，不是一项不可动摇的产权；同样，SAE 持守自己的最终立足点，持守的也不是一个不可破的自性之点，而是一个自认其为构、自认其有余项的立足点。

到这里，中心不可译之缝的三个面向就都现身了：本体层是"假名无自性重量"对"构有真实牵制"，戒律层是"破而不立"对"认领余项之立"，反身层是中观对 SAE 最终立足点的反诘、与 SAE 以减法式一面对它的正面回应。三面同构，是同一道缝的三次出现。接口分析到此处的本分，不是把这道缝填平假装两边相通，而是把它的三次现身都照清楚——并在反身这一面，让 SAE 正面面对那一问、以自己的语言回应（既不退让，它仍认领自己的构与立足点；也不把自己的立足点坐成一个中观所诃的自性），而不是反过来把中观收编为一个被 SAE 包容的质疑。

中观以"连空也空"守其不立，SAE 以"自认是构"守其立而不固——两种守法，守的却是同一种姿态：拒绝让否定有一个可安顿的终点。极少有哲学走到否定的尽头而不停下来，不把那个最终的洞见立成一处可安顿的真理；中观偏要连"空"都空掉，不许那个用以破一切者被坐实为可停留的自性。这一步极难守，而中观守住了。SAE 在另一侧，以"构不可闭合、余项不可消灭"守同一种姿态——没有一个构能封死自己成为终点，没有一次否定不留下推动下一次的余项。须把这"同一"说准，因为本篇自始至终守的那道分手就系于此：两家同一的是那个操作姿态——都拒绝把过程坐实为一个可停留的终极真理；而两家所拒绝坐实者，本体上恰恰相反——SAE 那一端是实而不可及的极限（路无尽头而方向真实），中观那一端是连"边"都不立的空（路与目的地皆假名）。同构在姿态，分手在所指。所以这不是两家在某个共同的终点处相逢——它们的终点本体上一实一空、不可互译；它们相逢的，只是"拒绝坐实终点"这一个姿态。中观把这姿态说成"空亦复空、无所得"，SAE 把它说成"余项守恒、无构闭合"：同一种不肯停留的手势，伸向两个不可通约的方向。这道在所指上敞开、却在姿态上同形的缝，正是这篇接口分析自始至终绕着走、不曾也不敢填平的那道缝——而正因两家在这道缝两壁上各自守得如此彻底，这场对照才配得上它的对象。

---

十、结语

本篇做的，是把中观／三论宗，与 Self-as-an-End，在结构上对照了一遍。

对照下来，对应之处是深的。中观对自性的否定，与 SAE"无构闭合"的论题，在结论上相遇；中观的"破"，与 SAE 的"凿"，在操作上同构；中观的"无所得"，与 SAE 那个不可坐实的极限，在终点的签名上相应。吉藏四重二谛层层推进的结构，与 SAE 否定的步位，可以步步相照；八不作为对自性执的总破，三法轮作为因机施教的适配，都能在 SAE 的坐标里找到它们的对应位置。

但对照下来，分手之处也是真的，而且这些分手不是对应的瑕疵，是这次对照最有价值的产出。最深的一道，在第二谛上：中观把缘起之法看作无分量的假名，SAE 把构看作有真实牵制之力的沉淀——同一桩缘起，一个读出无分量的施设，一个读出有牵制的真实。这道缝在本体、戒律、反身三个层面同时现身，三面同构，标出了两个框架不可互相翻译的那条界线。

这次对照，给本系列的坐标添了第三个面。前两篇里，唯识照亮了"构"——心智如何精细地建构；禅照亮了"凿"——那个破除可停留之构的否定动作。本篇里，中观照亮了"律"——任何构、任何凿，都不能被坐实为自性。须再次说明这第三个面的性质：它不是一个更高的位置，是一条跨越所有位置都成立的约束。中观在它的着力处（自我执取自身为目的的那一层）所照亮的，是这条不可坐实的律本身。

须守住本篇自始至终的一道分寸，在结语处再申明一次：把中观读作"那条不可坐实之律的探针"，不是把中观读作"一套未完成的 SAE"。中观不是差了什么序列、缺了什么层级的 SAE；它是一个目标与 SAE 不同的、自身完整的框架——它要的是破尽自性、令人无所得，不是建一座层级的塔。本篇与它在结构上相遇，在第二谛上分手；相遇之处标出对应，分手之处标出边界。这就是接口分析的本分：既不假装两个框架是一回事，也不假装它们之间没有可对照的结构。

本系列还会继续。其余的佛学传统——各自在某个问题域里，磨出过一面值得与 SAE 对照的镜子——会在后续逐一接口。每一次接口的本分都一样：不是因为某个传统"其实在讲 SAE"，而是因为每一个成熟的体系，都在它最用力的地方，照出了一些值得对照的结构。中观这一面镜子照出的，是那条贯穿一切位置、不容任何构与凿坐实自身的律。

---

参考文献

一、Self-as-an-End 著作（Zenodo）

凡 SAE 一侧的结构论断，均援引 SAE 自身已发表文本，记为 [1]–[4] 等；DOI 取概念 DOI（concept DOI）。

[1] 秦汉（Han Qin）：〔Via Negativa：否定方法论〕，Zenodo，concept DOI: 10.5281/zenodo.19544619。

[2] 秦汉（Han Qin）：〔Via Rho：余项守恒〕，Zenodo，concept DOI: 10.5281/zenodo.19657439。

[3] 秦汉（Han Qin）：〔0DD 四相与四分形〕，Zenodo，DOI: 10.5281/zenodo.20187591。

[4] 秦汉（Han Qin）：〔Self-as-an-End 方法论总论（V2）〕，Zenodo，DOI: 10.5281/zenodo.18842449。

[5] 秦汉（Han Qin）：〔SAE 佛学接口系列 I：唯识宗——阿赖耶识—种子系统与四分形〕，Zenodo，DOI: 10.5281/zenodo.20579131。

[6] 秦汉（Han Qin）：〔SAE 佛学接口系列 II：禅宗——无住与可停留之构的破除〕，Zenodo，DOI: 〔待补〕。

> 注：[1]–[6] 之确切发表标题与系列 II 之 DOI 以 Zenodo 记录页为准；本篇于 Zenodo 发布时，continues 关系指向系列 II（[6]），cites 关系指向 [1]–[5] 及下列佛教典籍。

二、佛教典籍（CBETA）

引用依《大正新修大藏经》（T）卷次与编号；版本以 CBETA 电子佛典集成为据。凡标"页行待终校"者，卷次／品次已定，逐字页行待于 CBETA 原典终校时补足。

[7] 龙树菩萨造，青目释，姚秦·鸠摩罗什译：《中论》（一名《中观论》），CBETA, T30, no. 1564。〔八不归敬偈及青目释"略说八事则为总破""若一则无缘，若异则无相续"，卷一〈观因缘品第一〉，p. 1c12 起；涅槃"无得亦无至……是说名涅槃"，卷四〈观涅槃品第二十五〉，页行待终校〕

[8] 隋·吉藏：《大乘玄论》，CBETA, T45, no. 1853。〔四重二谛，卷一论"二谛义"；页行待终校〕

[9] 隋·吉藏：《中观论疏》，CBETA, T42, no. 1824。〔卷二末亦开四重二谛；页行待终校〕

[10] 隋·吉藏：《二谛义》，CBETA, T45, no. 1854。〔二谛三重表述，旁参〕

[11] 隋·吉藏：《法华游意》，CBETA, T34, no. 1722。〔根本／枝末／摄末归本三法轮；页行待终校〕

[12] 隋·吉藏：《法华义疏》，CBETA, T34, no. 1721。〔"隐实／起权／收末归本"，旁参〕

[13] 后秦·僧肇：《肇论·不真空论》，CBETA, T45, no. 1858。〔本无／心无／即色三家批判，"非有非真有，非无非真无"双遣；页行待终校〕

[14] 龙树菩萨造，后秦·鸠摩罗什译：《大智度论》，CBETA, T25, no. 1509。〔卷一释"第一义悉檀"，"过一切语言道，心行处灭，遍无所依，不示诸法"；页行待终校〕

[15] 龙树菩萨造，后魏·毗目智仙、瞿昙流支共译：《回诤论》，CBETA, T32, no. 1631。〔"我宗无物故，如是不得过""我无宗相，何得咎我"；页行待终校〕

三、现代学术（参考层）

按本系列纪律，现代中观研究仅可用于厘清中观陈述之准确性（参考层），不承担任何 SAE 论断之依据（判据层）。本篇正文未直接援引现代学者；如于英文重写阶段需就 svabhāva 之操作化分析等处补入参考层引用（如 Garfield、Siderits、Westerhoff、Robinson 等），另行标注，并不改其参考层定位。〔待汉确认〕