Chapter 7. Restraint and vacancies¶

The task of this chapter. The first six chapters recorded points along the way. This chapter settles the account. The entries fall into two kinds, places where the system could clearly have taken another step but did not, and places it had to leave unfilled. The first are its restraint; the second, its vacancies. This chapter argues that the two overlap. Where it stopped short, an unfilled place generally remained. Where it had to leave a blank, we can look back and see where it once declined to reach. Chapter 1 set a criterion for evaluating the system. Where all four columns can be filled, there is phenomenology; where they cannot be filled but are filled nonetheless, there is metaphysics. This chapter examines a third case. The four columns cannot be filled, and the system has not filled them.

First, a limit on the word "restraint." Restraint is neither ignorance nor omission. To say that a system is restrained at a given point, we must establish three things. Someone at the time had taken that step; the approach was available and needed no new invention; and the system knew of it and chose not to take it. Only all three together constitute restraint. Whenever this essay speaks of Theravāda's restraint, its point of comparison is Sarvāstivāda, as specified in Chapter 1. Their inventories differ, with eighty-two dhammas in Theravāda and seventy-five dharmas in Sarvāstivāda. More than half the discrepancy lies in the columns for matter and conditioned factors dissociated from mind,¹ precisely where each side is willing to posit dhammas for formal features. Every step Sarvāstivāda took was an option visible to Theravāda at the time. This comparison is what makes the inventory of restraint tenable.

Restraint at the entrance. The first restraint is written into the system's program, in the distinction between ultimate and conceptual dhammas. Anything that can only be spoken of as a whole and cannot be separately identified at any moment is excluded from the table. Persons, chariots, houses, directions, and shapes all belong to the conceptual side. This rule strictly governs entry to the inventory. At one stroke it excludes the great majority of everyday names, including the concept of self. A system claiming to list everything that exists is willing to set such a threshold at its entrance. That is not the norm among the schools. Chapter 3's criterion applies the rule to material dhammas by asking whether something can be taken separately as an object of insight meditation. Chapter 4 uses it too. This restraint provides the strongest criterion in the essay's analysis.

Restraint within the inventory. With the entrance secured, we turn to what the system adds and does not add inside the table.

First, there are no conditioned factors dissociated from mind. The name needs unpacking. "Formations" means conditioned fabrication. Anything conditioned that belongs neither to matter nor to consciousness and mental factors falls in this column. "Association" is a technical term in this system. Two dhammas arise together, cease together, share an object, and share a base. Mental factors have precisely this relation to consciousness, and the later sections of this chapter use the term in this sense. Taken together, conditioned factors dissociated from mind are neither material nor any instance of feeling or cognition in the mind. Yet Sarvāstivāda regards them as real and as arising and ceasing through causes and conditions. In modern terms, this column holds formal items. Which continuity owns a given dharma? How much longer does a life have to run? Two sentient beings belong to one kind; a dharma is now arising or perishing; a string of syllables forms a word. These are the kinds of things it includes. Sarvāstivāda separately posits fourteen dharmas in the formations aggregate, including acquisition, nonacquisition, the life faculty, group homogeneity, arising, abiding, alteration, and cessation.² Theravāda does not take this step. Chapter 1 therefore judged its reification to be milder. It adds nothing, but reads a table of functions as a register of entities.

Second, it posits no unmanifest matter. Sarvāstivāda seeks to explain where the substance of the precepts remains when someone who has undertaken them sleeps. Its answer is a dharma, a kind of matter that does not appear externally, arising when the precepts are undertaken and continuing thereafter. Theravāda does not posit it. Abstinence factors manifest whenever the relevant conditions arise, and these manifestations and the conditioning relations they leave behind carry the persistence of the precepts. Chapter 3 called this a choice we can see clearly. Sarvāstivāda prefers an additional entity to carry persistence; Theravāda prefers conditions.

Third, it posits only one unconditioned dhamma. Sarvāstivāda posits three, cessation through discrimination, cessation without discrimination, and space. Theravāda has only Nibbāna.¹ Space is no ultimate dhamma on this side. Delimiting matter alone accounts for the boundaries between material clusters, and even it is classified as nonconcretely produced matter. Two fewer unconditioned dhammas mean two fewer formal features granted the status of real dhammas.

Fourth, the system neither accepts real existence across the three times nor posits intermediate existence. Book I, Chapter 6 of the Kathāvatthu refutes the claim that past and future are equally real; Book VIII, Chapter 2 refutes intermediate existence.^{3 4} Neither choice makes things easier. Without real existence across the three times, time cannot be a container holding dhammas, and continuity immediately bears more weight. Without intermediate existence, only two adjacent mind-moments remain between lives, with no buffer. Chapter 6 gave the reason for removing intermediate existence. If it contains consciousness and matter, it is already a life. A more important reason follows. Any maintainer can readily be read as a self that crosses over. Removing this position serves precisely to leave no place for a self.

Fifth, an inventory as long as twenty-four conditions is itself a form of restraint. Refusing to let a dhamma's substance persist requires finer distinctions among relations. Four conditions, decisive support, absence, disappearance, and different-moment kamma, specifically address how a dhamma that has ceased can still support a present dhamma. Sarvāstivāda need only say that a dharma's substance exists perpetually. Chapter 5 judged that this accounts for part of the inventory's length. Each side pays a price, Sarvāstivāda in ontology and Theravāda in the length of its inventory.

The system's restraint toward itself. The preceding points concerned restraint toward entries in the table. The following concern restraint toward the system itself, which is harder to achieve.

First, bhavaṅga has not been made into a storehouse. Chapter 6 drew three boundaries. Bhavaṅga adds no entry to the inventory. It is one of nineteen resultant cittas already listed among the eighty-nine; bhavaṅga is only one of its functions. Bhavaṅga receives nothing. Its object is fixed at the moment of rebirth-linking and remains unchanged throughout the life. It records nothing that happens in this life and has nowhere to record it. Nor must it explain kamma and its fruits, since different-moment kamma condition does that. Together, these three boundaries constrain a position that could very easily grow into a bearer. Chapter 6 has compared what fills the same position elsewhere.

Second, no exception is made at the highest point. Liberation is placed in the sequence, in the javana position. It follows the same transitional procedure as jhāna, with the same names, preparation, access, conformity, and lineage transition. No separate mechanism is devised for it. It is not raised above the sequence or passed over as ineffable. Chapter 4 treated this as one of the principal pieces of evidence. A system that makes no exception at its highest point shows confidence in its notation and a willingness to have that point examined under the same rules.

Third, Nibbāna is not described. Of the columns for the four defining aspects, only three are filled for Nibbāna. The proximate-cause column is empty. Chapter 6 judged that the four defining aspects form a machine built for conditioned dhammas. Put an unconditioned dhamma into it, and it stops on its own at the fourth column. It forces no entry and makes no change to itself to accommodate the exception. It leaves the blank in the table. The terms in the three filled columns are all negative or privative. Peace is cessation; deathlessness is the absence of further death; signlessness is the absence of a sign that can be grasped. This is the least that can be done to put an item in the table without describing it.

Fourth, the whole Kathāvatthu is a book of boundaries. Its form is to rule, claim by claim, that something must not be said in a given way, rather than provide a more complete set of claims at every point.³ To include a predominantly negative treatise in the Abhidhamma Piṭaka shows the importance this lineage gives to limits. Its treatment of the Buddha's attributes is likewise relatively restrained. The three levels of omniscience distinguished in Chapter 6 of Essay 13 offer a comparison.

The two restraints differ. Chapter 7 of Essay 13 discussed restraint in the early teaching, which consists in not answering. Certain questions are judged unfit for a declaration, and the Buddha gives no answer. Theravāda Abhidhamma's restraint consists in not positing. It refuses entities. The two differ in nature and in the weight they can bear. Not answering protects the way a question is asked and stops it before it enters. Not positing protects the inventory's purity and stops an item before it enters the table. They differ in what they cannot stop. A violation of not answering is visible, because someone has answered what should not have been answered. A violation of not positing is harder to see. No entry need be added, yet those already in the table can be read as increasingly real. This is where the system's later trouble occurred.

What sustains this restraint? This book judges the restraint to be real, but its support does not come from methodological self-awareness. Throughout the treatises and commentaries, no general rule can be found explaining why further entities should not be posited. Each refusal has a specific reason. These reasons neither invoke one another nor form a stated principle. Two things are actually at work. One is discipline in transmission. The conservatism of compiling the canon makes this lineage prefer duplicate names to merged entries. Nine of the fifty-two kinds of condition are two names for the same thing, simply because each name follows the wording of a discourse. The other is the constant reference to practice. This taxonomy is always used in contemplation. No one wants an added item that contemplation cannot use.

Both discipline and this reference to practice lie outside the system. If the external constraints loosen, there is no brake within. Chapter 1 recorded a chronology. Sabhāva first becomes a defining term for dhammas in the Paṭisambhidāmagga. In the hands of the sixth-century commentator Mahānāma, dhammas are called real existents (sat), and momentariness enters with them. The Paṭisambhidāmagga has neither.⁵ Not a single entry has been added, but meaning has shifted a level. This is precisely what the restraint of not positing cannot stop.

There was in fact a brake ready within the system, as Chapter 1 noted, in the proximate-cause column of the four defining aspects. No dhamma for which a proximate cause can be entered can be self-sufficient. This column ties each dhamma to conditions and is the least susceptible to ontologization in the system. Karunadasa's counterreading stands on this side too. To treat arising, presence, and dissolution as real dhammas would require secondary characteristics for them, and so on without end. Abhidhamma itself rejects that route.⁶ But the proximate-cause column is a requirement of format, not a declared principle. A format governs those who fill the table, not those who read it.

The same places, seen from the other side, are vacancies. The statement at the end of Chapter 5 can serve as this chapter's thesis. Avoidance works, especially in practice. Viewed positively, it remains the virtue of restraint in metaphysical claims. But what has been avoided has not disappeared; it will eventually be written elsewhere. Restraint and vacancy are the same thing seen from opposite ends. Restraint means that no entity has been posited for something. A vacancy means that this something still works within the system, without recognized status. We now gather the entries recorded in the first six chapters under the three questions set in Chapter 1.

First, the full table. There are twelve entries in the three groups, followed by two questions about how the inventory was made. The table gives two more rows of detail than the discussion below. The substance of the precepts and intermediate existence were recorded separately in Chapters 5 and 6; the discussion combines them under continuity. The second column records what Theravāda actually did, not what it should have done. Three rows read "No treatment given." This means that no corresponding account can be found in the treatises and commentaries, not that this essay has omitted it.

The first group concerns how dhammas are related. The status of conditions is central. If a condition is a dhamma, it should appear among the eighty-two, but the inventory gives it no place. Worse, once it has the status of a dhamma, its connection with either of the two terms it connects becomes a relation between two dhammas. The system requires a condition to account for that relation, and the regress begins. If a condition is no dhamma, yet the whole system depends on it, the inventory cannot claim to list everything that exists. Abhidhamma puts the relation inside one term. Conditioning force is inherent in the conditioning dhamma, as a chili's heat is inherent in the chili.⁷ This stops the regress without adding to the inventory. Chapter 5 identified the cost. The condition fails the criterion set in Chapter 3, which asks whether something can be taken separately as an object of insight meditation. No conditioning force passes it. Yet the second of the sixteen insight knowledges is knowledge discerning conditions, and what it contemplates is conditions.⁸ What is actually contemplated has no recognized status in the inventory.

Another point in this group, recorded in Chapter 5, is easy to miss. Nonultimate dhammas can support ultimate dhammas. Object condition has four kinds of conditioning dhamma, mentality, matter, concepts, and Nibbāna. Concepts are not ultimate dhammas, and Nibbāna is unconditioned. Neither is brought into being by anything, yet both can support other dhammas in arising.⁸ The distinction between ultimate and conceptual dhammas is the strictest restraint at the entrance. But in this column of conditions, what was barred at the door comes in and works all the same.

The second group concerns how dhammas exist in time. It has the most entries, all connected with one another.

The boundary of the moment. A mind-moment divides into three submoments, arising, presence, and dissolution. The treatises say that these are only three aspects of the same episode of arising and ceasing, not three parts. Yet when assigning conditions, they use the phase of presence as an interval with a duration. They do not directly answer whether a mind-moment can be divided further. They offer an account that lets work continue without an answer.

The origin of seventeen. Matter lasts seventeen mind-moments, and a complete five-door process has seventeen slots. The two seventeens define one another. Chapter 6 distinguished the levels of this judgment. Only the successive arising and ceasing of cittas can count as observational evidence. The two steps beyond it do not come from that evidence. One assumes that every mind-moment has the same length; the other fits matter's lifespan exactly to the cognitive process. The number aligns the mental and material lines. It reports no actual duration.

A shared clock. This extends the preceding point. Chapter 4 recorded it at the junction between lives. A person dies here, while rebirth-linking consciousness and a mass of kamma-produced matter arise together elsewhere. To say that no mind-moment separates these events is to say that this moment here and that moment there are the same present. But mind-moments are only units of sequence. Each continuity counts its own slots, and the system gives no means of conversion. Chapter 6 pursued the point to its end. Time must be explained by continuity. The clock is therefore continuity itself, not something external, and continuity itself is what needs explaining. The same assumption has been used between consciousness and matter, in measuring matter seen in common by many people through the number of slots in one mental continuity.

No words for duration. Absorption is the first place where length is not fixed, and the first where cognitive-process theory, as a pure theory of order, falls short. The other end of the same gap lies in the lifespan table. It gives only sequence at the finest end and only celestial years and great aeons at the coarsest, with no conversion between them. The five hundred great aeons in the heaven of Nonpercipient Beings reveal this gap most clearly. No consciousness occurs there, yet material continuity is still counted on a mind-moment scale.

The identity of continuity. There is no storehouse. Within a life, identity is fixed by rebirth-linking, bhavaṅga, and death sharing one citta and one object. Across lives, the previous life leaves this one only an object's identity and a kammic force. It leaves no dhamma. Chapter 6 therefore concluded that continuity is not a property of mental dhammas. However detailed the inventory of eighty-nine cittas, it cannot determine continuity. The question discarded with the storehouse returns unchanged, except for one word.

The third group: how to draw the boundary between dhammas and what are not dhammas. There are only three entries here, but they are the hardest.

The first is self-referential. Who draws the boundary between ultimate and conceptual, and on which side does the act of drawing fall? The distinction itself is clearly no ultimate dhamma. It is an operation of the system and therefore belongs on the conceptual side. An act on that side thus confers legitimacy on the whole table. Chapter 1 raised this in another form. What is the status of an ultimate dhamma itself? Is it an identifiable quality or a thing established in its own right?

The second is the absence of a firewall between two readings of own-nature. The first treats own-nature as an identifiable quality, the second as an essence established in its own right. Chapter 1 judged that most of the system's restraint depends on the first reading, although the system may not remain on that side throughout. Madhyamaka directs its refutation precisely at the unguarded stretch between the two readings.

The third is Nibbāna in the table, an unconditioned dhamma handed to a machine built for conditioned dhammas. Chapter 6 distinguished two halves. One stays within the boundary because its statements are all relational, made from the conditioned side. The other crosses it because the characteristic column speaks of Nibbāna itself. Format, not doctrine, causes the crossing. The table requires every item to be defined through the four defining aspects. Here uniformity of format overrides the prohibition in Chapter 2 of Essay 13.

Two further questions fall outside the groups and ask how the inventory was made. One asks why matter has no second level. Mental dhammas have two levels, consciousness and mental factors. Association requires all four features, arising together, ceasing together, sharing an object, and sharing a base. Matter appears in a flat table of twenty-eight parallel items. Yet the three kinds of mutable matter, two kinds of intimating matter, and four material characteristics seem to stand in the same relation to concretely produced matter as mental factors do to consciousness. None is an independent quality. All appear only in concretely produced matter, and none can be taken separately as an object. The format of association is available. Why does the material side not use it? The other question is whether adding a slot is legitimate. Smile-producing consciousness gives the clearest example. Someone who produces no kamma still smiles, and the smile is neither wholesome nor unwholesome. It needs a slot. Chapter 2 read the taxonomy's orientation from this. It prefers adding a slot to leaving a phenomenon without a place. Exhaustiveness is its methodological virtue, but the pressure to be exhaustive can itself produce entries.

Three origins of entries. These two questions bring out a distinction that this chapter makes into a criterion. Entries have at least three origins. They are found in contemplation, added to explain something, or added to make the notation itself stand. All three look alike in the table. An entry alone does not reveal its origin; we must look at what it does in the system. Chapter 4 gave examples of each. The exhaustiveness of the fifteen forms is calculated. Matter's lifespan is first set at seventeen, from which at most fifteen past bhavaṅgas are deduced. Assigning three times of maturation to seven javanas looks like starting with the fourfold classification of kamma and then finding places for it. The two-stage pattern of five-door and mind-door processes, the dependence of process length on object strength, and the position of the determining slot all agree with experience and do not look like additions. The twelve items in six pairs in Chapter 2 belong to the third kind. A property that could be stated in one sentence becomes twelve entries because the table's format requires each to be listed. All three origins are present, which constitutes no accusation against the system. It shows only that this inventory combines observation, deduction, and pressure from the format, rather than being purely a record of observation or a result of deduction. To decide which origin an entry has, each must be examined. This essay has offered only a few demonstrations.

Why these vacancies are inevitable. The three groups are not arbitrary. Analysis at the level of ultimate dhammas must answer how dhammas are related, how they exist in time, and how a boundary is drawn between dhammas and what are not dhammas. None of these questions can be solved within the analysis. They concern the conditions that make this tool possible, not its objects. To use an inventory requires relations first; to order a sequence requires time first; to establish an inventory requires a boundary first. What makes analysis possible cannot also be one of its entries, or it would both support the inventory and have to be included within it. The vacancies are thus the shape of the method's boundaries, not omissions. More diligent commentaries and cleverer theorists would still leave these three places empty. If they were filled, what filled them would necessarily fall outside analysis.

Ultimately, it is the same place. Chapter 6 gathered the three costs into one statement. Wherever a persisting bearer is needed, the system posits a condition instead, without giving the condition itself a place. This explains both halves of this chapter. Restraint is possible precisely because conditions offer a general substitute. Different-moment kamma condition replaces a storehouse; abstinence factors and conditioning relations replace the substance of the precepts; immediacy condition replaces intermediate existence. The more readily this substitution works, the fewer entities are added and the cleaner the inventory remains. Vacancies are inevitable because the substitute has no place of its own. The system's restraint and its vacancies are two sides of one account. It transfers the entire ontological burden to relations, then gives relations no ontological place.

An echo of Chapter 7 of Essay 13. That chapter concerned leakage. Questions the early teaching explicitly declined to answer were later expressed in other ways. This chapter concerns vacancies. What the analysis explicitly declines to posit still works within the system. These are two stages of the same mechanism. First, a question is judged unfit to be asked and so receives no answer. Second, a system needs something in order to work but refuses to grant it recognized status. It has a place it must use but cannot articulate. What remains unsaid does not disappear. In the next generation's hands, it will appear in another form. The places where Madhyamaka, Yogācāra, and tathāgatagarbha thought each emerge all have a corresponding slot in these two tables.

Data and grounds. This chapter cites no new material. The grounds for each entry in the inventory of restraint are in the text and notes of Chapters 1 through 6; the relevant Kathāvatthu discussions are cited in Chapters 4 and 6. The Sarvāstivāda comparisons follow the Kośa's five-category classification, with its reasons for positing dharmas left to Essay 15. Two passages are also taken from the Charts, Chapter 7, "Compendium of Categories," and Chapter 9, "Compendium of Meditation Subjects."⁸ The former shows how the system gathers diverse names into groups; the latter distinguishes tranquility from insight meditation. This chapter develops the concerns listed in Chapter 4 of Essay 3 in Part I. That section's preliminary judgment was that Theravāda Abhidhamma's phenomenology is excellent, while its problems lie in ontologization and fixing things in place. This chapter keeps the first half and revises the second. Ontologization does occur, but through cataloging procedures and pressure from the format, rather than in any declared doctrine. As for fixing things in place, this book regards it as a boundary of the analytical approach itself, not a failing unique to this lineage. That point will be established only when the next essay's entries are complete.

Handing over. The next essay concerns Sarvāstivāda, another set of answers to the same questions. Most places judged vacant here are filled there by an entity. Real existence across the three times fills the time group; the perpetual existence of dharma-substance fills the identity of continuity; acquisition and unmanifest matter fill persistence; the six causes and four conditions fill relation. Only comparison can show whether these vacancies are omissions peculiar to Theravāda or boundaries of the analytical approach itself. Later, Madhyamaka directs its refutation at own-nature and relation, Yogācāra fills gaps in continuity and seeds, and tathāgatagarbha thought fills them from another end. Which vacancies each approach addresses must be identified one by one after both essays' entries are complete. This essay and the next are therefore prerequisites for those that follow.

Passing the discussion to the next essay. When the same analysis is followed to its end, what will the other route posit?

References¶