1. Introduction: The discourse horizon of Tamil poetics

1If we think of poetics as an ongoing process of reflection on the human undertaking of literature, we shall not be surprised to find it subject to all sorts of changes and adjustments in the course of time. Change will be stimulated by a number of factors, the most tangible among them literary fashion, the development of new theoretical models and influences by sister disciplines. Strategies of dealing with this situation can be different. One possibility is simply to write a new treatise to accommodate new facts. Another is to add a commentary. The third is, in a conservative environment, to enlarge an existing sūtra, either by adding new chapters or by reworking older ones. The Tolkāppiyam Poruḷatikāram (TP) bears traces of all three strategies. Here I will examine the opening section of its Kaḷavu-iyal, in Iḷampūraṇar’s counting the sūtras 89–98 (hereafter, all references to the TP in Iḷampūraṇar’s version will be recorded with the siglum TPi). This appears to be a heavily reworked section. What we see here is an attempt at integrating no less than three theoretical models. I will argue that this process reflects a reaction to all the three factors enumerated above, i.e. literature, poetic theory and outside stimulus.

2The original basic structure of the Kaḷaviyal is, as has been pointed out by Takahashi (1995),1 an enumeration of speech-situations ordered according to their speakers. The concept of kaḷavu as a cluster of situations in premarital love life was enriched early on by putting it into relation with the system of marriages taught in the Sanskrit Dharmaśāstra tradition. On the sūtra level, this influence is shown in the definition sūtras that explain kaḷavu as similar to the Gāndharva marriage. Throughout the commentaries of the poetological tradition this connection is substantiated by sūtras, in prose and in verse, defining the whole set of eight types of marriage.2

3The further development sees a double set of innovations on the theoretical and the practical level, and it is by no means easy to decide on which side the original impetus is to be located. For instance, in the postclassical period, a new genre, the kōvai (or “garland”), develops, a poem where the familiar situations of Akam are depicted in a (more or less) linear narrative sequence. On the theoretical level this means enumerating the situations no longer by a taxonomy of speakers, but as a sequence of events. The first treatise of this type is the second great work on Tamil poetics, the Kaḷaviyal, alias Iṟaiyaṉār Akapporuḷ (IA). I am inclined to believe the IA was one of the factors that triggered the development of the kōvai and not vice versa, for the simple reason that a considerable number of sub-themes of the kōvai are not yet found there. In fact they are cleverly introduced, and inserted in the proper places in the sequence, by the commentator Nakkīraṉ, along with what is probably our earliest specimen of a kōvai, the Pāṇṭikkōvai. We will have occasion to come back to that.

4Let us imagine a situation of intellectual competition. The school of the Tolkāppiyam reacts to the challenge set up by the IA by reworking its own scheme of kaḷavu. A new sequence, organised by event, is grafted onto and into the speaker sūtras enumerating situations with their themes and sub-themes per speaker. We will look at one sūtra in particular, namely the one defining aiyam, doubt: at first glance the talaivaṉ cannot decide whether the talaivi is divine or mortal. This is the beginning of the Akam sequence in a kōvai, but a sub-theme virtually absent from early classical poetry. No commentator (neither Iḷampūraṇār nor his successor Nacciṉārkkiṉiyar) can locate any Caṅkam example. So this makes for two models of depiction in the TP, still in Takahashi’s words, speakeroriented and theme-oriented.

5But the beginning of the Kaḷaviyal underwent a more profound change, and here we come to the main topic of this essay. Perhaps the new orientation of an evolving story for a poetic couple had a natural affinity to drama. The further restructuring of the Kaḷaviyal incurred a dramatic approach of enacted sentiments, i.e. an emphasis not on what speakers do or say, but rather on what they feel. The original definition of kaḷavu via the Gāndharva marriage was replaced by one that can be shown to have clear counterparts in the Sanskrit tradition of alaṃkāraśāstra. Kaḷavu becomes a sequence of stages of passion, kāmāvasthā — remarkably, however, there are nine stages instead of ten, a fact which will have to be discussed.3

6In what follows I want us to have a closer look at the new definition sūtra for kaḷavu in the Kaḷaviyal and at its possible Sanskrit sources. In fact the situation is more complicated than it seems at first sight. Insler’s depiction in his 1988 paper on the daśakāmāvasthā does not seem to take into account the full evidence. Next we will examine the sūtra for aiyam, doubt, which will bring us to the kōvai and Nakkīraṉ’s commentary on the IA. Finally, we must also consider the definition sūtra of the subsequent Kaṟpiyal in the TP, in which we will discover verbal echoes of the marriage sūtras found in the Tamil commentaries, traces of a Dharma tradition in Tamil. Moreover we shall see that this sūtra is the counterpart of the discarded definition sūtra at the beginning of the Kaḷaviyal. This will bring us to the possible reasons why the tradition of kaṟpu remained more stable than that of kaḷavu, as far as the TP is concerned.

2. Kaḷavu

2.1. Definitions of kaḷavu

7Between them, the TP and the IA share a set of no less than four definitions of kaḷavu. The interrelation and probable sequence among them have been discussed in detail in Wilden (2006). Let me give you a summary of the argument, as this is essential for understanding the position of TPi 97, which contains the latest of these definitions and is clearly at odds with the concept of the Kaḷaviyal. A pre-classical version is found in the TP Ceyyuḷiyal, TPi 487, which still uses the archaic term maṟai (“secret”) instead of kaḷavu (“stealth”) and defines it by an enumeration of four basic situations: 1. union out of passion, 2. meeting [again] at the same place, 3.+ 4. union with the help of HIS friend or HER confidante.4 The familiar connection of kaḷavu with the Gāndharva marriage is first found in IA 1.5 This sūtra has a close phrasal counterpart in TPi 89.6 There, however, the definiendum kaḷavu is replaced by kāmakkūṭṭam, the “coupling out of passion”. This has to be evaluated as a renaming in order to give room to the new definition sūtra TPi 97, which enumerates the stages of passion.

8In Wilden (2006) I reasoned in favour of that sequence of events on the grounds of poetological systematics, but in fact there is a much simpler argument. Both Iḷampūraṇar and Nacciṉārkkiṉiyar (Nacc.) comment on TPi 89 as if it still defined kaḷavu. It is the term they use while silently ignoring kāmakkūṭṭam. This is also the place where they bring in a lengthy digression into the field of Dharma by discussing the whole set of aṣṭavivāha, culminating in Nacc. with a series of veritable Dharma verses. This first area of massive Sanskrit influx is hard to motivate from the point of view of poetic theory. It becomes more comprehensible if we are willing to admit that the commentators on poetics are forever hovering on a thin grate between describing the poetic universe and the real world. It is not lovers’ behaviour in poetry that gives rise to worry, but in real life you may be concerned with the question whether girls should have the right to run away with their chosen young men.7

9The situation with the newly introduced definition sūtra TPi 97 is very different:

vēṭkai orutalai uḷḷutal melital
ākkam ceppal nāṇuvarai iṟattal
nōkkuva ellām avaiyē pōṟal
maṟattal mayakkam cākkāṭ’ eṉṟi
ciṟappuṭai marapiṉ avai kaḷaveṉa moḻipa.

1) desire, 2) thinking [permanently] of a single thing, 3) becoming weak,
4) speaking of increase (?), 5) transgressing the boundary of shame,
6) likening those (?)8 to all the things looked at,
7) forgetting, 8) confusion, 9) death — those according to eminent usage [make up] kaḷavu, they say.

10So here kaḷavu is defined by an — apparently climactic — series of emotive states and/or the related acts. The reader of classical Tamil poetry will be puzzled by most of them. Vēṭkai is a term occasionally occurring in poetical theory, but there it comes at a later point in the sequence of poetic events, namely as the desire for marriage which results in the act of varaivu kaṭātal (mostly the confidante asking the hero to marry). From the kiḷavis we know melital, as one of the possible designations of HER growing anxiety when HE delays marriage. Shame and the loss of it is a topos in poetry. HER imminent death, finally, is something that can be evoked by the woman or her confidante if they want the man to stay or to hurry up marriage arrangements. That makes for four out of (probably) nine. The rest look utterly unfamiliar. The series falls into place, however, if we compare it with one found in different domains of Sanskrit literature, where it is known as the daśa kāmāvasthāḥ, the ten stages of passion.

2.2. The Ten Stages of Passion

11Insler (1988) has collected a great deal of material on these ten stages. Let me just give you the gist of what he says. The basic exponents of the list in theory are Alaṅkāraśāstra and Kāmaśāstra. This has an inherent logic, since the former deals with the depiction of kāma, and the latter with the phenomenon itself. For Alaṅkāra, Insler names the following sources (without giving the exact references): Nāṭyaśāstra, Śṛṅgāratilaka, Kāvyālaṅkāra, Daśarūpa and Sāhityadarpaṇa. For Kāma he mentions Kāmasūtra, Ratirahasya, Anaṅgaraṅga, etc. He acknowledges differences in the various versions, insignificant inside the disciplines, more substantial between Alaṅkāra and Kāma. His lists presumably go back to some editions of the Nāṭyaśāstra (NŚ) and Kāmasūtra, as the oldest exponents of their respective domains. Additionally he has collected literary material from a variety of Kāvya texts in Sanskrit and Prākrit, most notably Mahābhārata and Rāmāyaṇa, along with the Jain Agaḍadatta and Paumacariya. In fact he chooses a lovelorn heroine from the Agaḍadatta story as his starting point, showing how the description of a woman in (unrequited) love corresponds virtually point by point to the theoretical description.

12This material is further contrasted with two paradigmatic female heroines in distress because of love, namely Damayantī and Sītā. The situation of these two, however, is fundamentally different: Damayantī is falling in love for the first time when she hears of Nala. Sītā suffers prolonged separation from her beloved spouse Rāma, as she has been abducted by Rāvaṇa. Insler tries to explain the differences in the lists by tracing them back to two original versions, one for the prathamānurāga, the first passion, one for virahakāma, love in separation.9

13Let us now compare the two fundamental lists from Alaṅkāra and Kāma with our TP sūtra:

14If we first take a look at the two Sanskrit lists, we see that they match each other in number, as well as in their beginning and end, but in between there is quite some fluctuation: either different states are recorded or similar states are found in a different sequence. Abhilāṣa and cakṣuḥprīti might refer to the same thing, namely fixing the attention on a particular object. Similarly, cintana and manaḥsaṅga can both refer to the fact that the said object absorbs the afflicted person’s mind. The next two, anusmṛti and saṃkalpotpatti, are no longer really compatible. One presumably denotes just an intensification of the last stage, i.e. thinking constantly of the attractive object. The other, however, deals with pondering the possibilities of action. The next two are totally different. The Nāṭyaśāstra’s guṇakīrtana still concerns the beloved object, only now the admiration is voiced. So while there the lover is still in the stage of happy infatuation, the one in the Kāmasūtra already stops sleeping at night. For the former udvega, “anxiety”, is the first sign of alarm, while the latter is already in the state of losing weight. With the next step the former begins to lament, whereas the latter is now losing interest in the rest of the world. From stage seven onwards things deteriorate and both developments end in death, but the way to get there is not identical. Unmāda in the Nāṭyaśāstra presumably refers to fits of raving, that is, an intensification of the lamentation state. This becomes a permanent affliction, vyādhi, which leads to jaḍatā, apathy, and finally to death. For the Kāmasūtra, stage seven is lajjāpranāśa, the loss of shame, that is, the inability to keep up appearances. This is followed by unmāda, here obviously already regarded as a permanent state, followed by mūrcchā, unconsciousness, and then death.

15If we try to put our finger on the basic difference between the two lists, we might say that one is concerned with the emotive states and their expression and the other rather with their physical manifestations. This distinction becomes plausible if we look at the motivation for enumerating the ten stages in the two traditions, Alaṅkāra and Kāma. Part 5 of the Kāmasūtra deals with the seduction of other men’s wives. The list is given in the context of the dangers a man has to be warned against before taking up such a hazardous undertaking: if his love is not requited, he risks illness and eventual death. Here we are reminded of Insler’s Suśrutasaṃhitā, where some of the stages occur as symptoms of madness, and one of the reasons for madness is kāma. On the other hand, the Nāṭyaśāstra is concerned with the representation of those symptoms of kāma, be it on stage or in verse. It deals with emotions and utterances.

16If we now compare both with the Tamil list, we shall find that it agrees with both in broad outlines, but that it is no direct counterpart to either of them. In fact it contains cognates to both of them. The first stage, vēṭkai, desire, looks closer to abhilāṣa, but this is not conclusive. The second and third bring us to the most remarkable deviation: they seem to have been compressed into a single stage in Tamil, orutalai uḷḷutal, thinking permanently of the same thing, which again looks more in accordance with the Alaṅkāra list. This brings, however, the number of elements to nine, not ten. This is surprising, because it is well known to students of Indian literature that the most stable feature about classifications and enumerations is the number. One way out was suggested by Jean-Luc Chevillard, namely the possibility of separating orutalai and uḷḷutal. The former alone might mean something like “onesidedness”, the latter simply remembering the beloved object. Alas, this does not look very satisfactory from the syntactic point of view: all states consist either of abstract nouns or of verbal nouns in -al/-tal. The commentators too take orutalai uḷḷutal as a unit. There might be yet another explanation which we shall come back to below.

17Now, the next stage, melital, quite clearly corresponds to neither guṇakīrtana nor nidrāccheda, but rather to the Alaṅkāra’s subsequent udvega, anxiety. In the classical Tamil kiḷavis it is a cognate of aḻital and usually refers to the mental state of mounting anxiety in the heroine who is waiting for her lover. The next, ākkam ceppal, is not very clear in itself. If it means “speaking of an increase” — in attachment or anxiety — it is no direct counterpart of any of the other side, but perhaps it vaguely corresponds to vilāpa, lamentation. The following nāṇuvarai iṟattal for once is a clear counterpart of the Kāmasūtra’s lajjāpranāśa. One further stage, nōkkuva ellām avaiyē pōṟal, is slightly elliptical, with an anaphoric pronoun avai at loose ends, but presumably this avai refers to the remembered features of the beloved which are compared to everything seen here and now. If this is so, it would seem a backward jump in the Alaṅkāra list to guṇakīrttana, praising the features of the beloved. Let us assume that this is no longer done in the presence of the friend or confidante, but in public, since shame is already lost. The next, maṟattal, forgetting, again has no clear counterpart, though it may vaguely correspond to the Alaṅkāra’s jaḍatā. Mayakkam, then, probably is the Tamil counterpart of unmāda in both Sanskrit lists, unless it refers to unconsciousness rather than to confusion, in which case it would agree with mūrcchā in KS, though perhaps slightly weaker. It comes late, in penultimate position. And finally, with cākkāṭu, death, consent among all the three is re-established.

18Summing up, the Tamil list clearly comes from the same background as the Sanskrit ones, but it cannot be termed a translation of either of these. Let me just add in an aside that the situation for TP 6, the Meyppāṭṭiyal, is very similar in this respect. If we look, for example, at TPi 247 and its enumeration of eight rasa-like elements, we realise two facts. Firstly, the terms employed correspond literally not to NŚ 6.15, the adjectival rasas (“tastes”), but to the nominal sthāyibhāvas (“emotive states”) of NŚ 6.17.11 Secondly, the sequence used in the TP, that orders the eight elements into four pairs, makes it clear that its source cannot have been NŚ 6.17, unless we concede a fundamental restructuring on the part of the Tamil translator.12

19The case of the daśa kāmāvasthāḥ is unlike that of the eṇmaṇam/aṣṭavivāha. There, I think, it is possible to make a plausible suggestion regarding the Sanskrit source, namely either Kauṭilya or Nārada (also see Wilden forthcoming). Here, however, it is necessary to carry on with something like textual archaeology. I cannot by any means present definite conclusions, but I want to show you some further passages. All of them come from the Rasādhyāya of the NŚ, a part notorious for its text-historical trickiness, whether we want to describe the matter in the words of Srinivasan (1980) or in those of Gerow (1977). The first of them is NŚ 6.18–21, an enumeration of thirty-three bhāvas, moods. In this enumeration you can recognise no less than six of our ten stages, and it also ends in death. Directly afterwards there is an enumeration of eight sāttvikas, presumably the physical manifestations of the moods already mentioned. Here there is no direct correspondence to the ten stages, but the last one, pralaya, might be seen as a synonym to the ninth stage of the NŚ, mūrcchā, fainting. In a prose passage in NŚ 6.46, finally, we find yet another enumeration, this time of the fifteen anubhāvas of separation, and once again the sequence ends in death. Here there are four of the familiar elements plus one, I would suggest, that we know from the Tamil list only, apasmāra, the loss of memory. This looks like quite a likely source for the unexplained Tamil maṟattal.13

20My main point here, however, is that even as far as the number is concerned, there does not seem to be stability. The NŚ alone (or what little of it I have checked) contains cognate enumerations of mental states ending in unconsciousness and death with eight, ten, fifteen or thirty-three elements. So why should a Tamil author not be allowed to make it nine? To put matters cautiously, we are now faced with several ways of interpreting the evidence. One possibility is that the source for the TP could have been the NŚ in its entirety or as a descriptive system. It is quite probable that a Tamil reader, if he got access to it, might have been impressed by more than one passage when he set out to bring something similar into Tamil poetics. But it is just as possible that texts were even more fluid than we suspect now. The actual source might have been another redaction of the NŚ or one of its parts. It might have been another text altogether. We are only too often reminded of the fact that texts get lost.

2.3. aiyam — Doubt

21The next instance of remodelling the situation is altogether different. TPi 92 enumerates the indications by which a man realises that the woman impressing him is a human and not a goddess. The word aiyam, doubt, has to be taken over from the preceding sūtra TPi 91, which deals with the social status of the hero and the heroine, a problematic issue in itself but not our concern here.

TPi 92
vaṇṭē iḻaiyē vaḷḷi pūvē
kaṇṇē alamaral imaippē accam eṉṟu
aṉṉavai piṟavum āṅkaṇ nikaḻa
niṉṟavai kaḷaiyum karuvi eṉpa.

Bee, ornament, creeper, flower,
eye, bewilderment, winking, fear -
things like these and others, when they occur there
they call those that continue14 the means to remove [doubt].

22This slightly cryptic way of reference may be in need of some elucidation. vaṇṭu refers to the bees swarming around HER fragrant hair, a topos well attested in poetry. It is not quite clear to me why a goddess should not attract bees, all the more so as her garlands remain ever fresh. As for the ornaments (iḻai), Iḷampūraṇar explains that god-made ornaments are distinguishable from those on earth. As for the floral patterns (vaḷḷi) drawn on HER breast and shoulders, the commentator declares this to be a human custom. The next, pū, is clear: the flowers worn by the girl are fading. Slightly redundant seems kaṇ, her eyes, followed in the same line by imaippu, blinking. Similarly alamaral and accam both seem to refer to the girl being confused or even frightened by the sight of a stranger.

23Now, as for influences, every student of Sanskrit will at once be reminded of one of the texts he first read: the Nalopakhyāna in the Mahābhārata, where Damayantī recognises the gods who have all taken the shape of Nala. Their flowers do not fade, their eyes do not wink, and, piṟavum, there is no trace of perspiration on them and their feet do not touch the ground. So, the topos as such is well known. But is there a Sanskrit sūtra for it? I have not yet been able to locate one. So, as far as possible Sanskrit sources are concerned, this may simply be a case of literary reception. What is interesting here is what has become of the topos in the Tamil tradition.

24In Caṅkam poetry it does not exist yet. As already mentioned, neither commentator is able to give a Caṅkam, or even a Kīḻkkaṇakku quotation for this sūtra. This is all the more remarkable since in fact there would have been an interesting, if isolated, example in Naṟṟiṇai 155. There HE addresses HER and asks her directly whether she is an aṇaṅku (let us say, a deity, in this context) or whether she lives here by the seashore. The kiḷavis are instructive. There are two, often a sign that the message of the poem in question was controversial. One puts the poem down as an instance of iraṇṭām kūṭṭam, in itself an unusual formulation, but evidently concerned with what is later known as iṭantalaippāṭu, the second meeting. In this case, the address has to be understood as a peculiar form of (hyperbolic) flattery. She is as beautiful as an aṇaṅku. The second kiḷavi is a formulaic one about speaking bitter words in times of quarrel (uṇarppuvayiṉ varā ūṭaṟkaṇ), a kaṟpu topos. So, this obviously implies taking aṇaṅku as an invective, calling HER a demoness, perhaps. In sum, the situation of doubt as a prelude to the first encounter is not yet familiar to the kiḷavi authors, who can be counted as the first commentators.

25It is known, however, and has in fact become the second standard situation (after kāṭci, sight) in a very productive post-Caṅkam genre, the kōvai. Unlike the early classical texts with fluctuating topoi and open interpretations, a Kōvai follows a sort of standardised mini-kiḷavi, often reduced to a single word. In this case the word is simply aiyam. On a theoretical level, the IA does not yet know of this. The sequence is fully integrated from the time of Nampi Akapporuḷ onwards. There is one surviving intermediate treatise on the themes of Akam, between the IA and Nampi, namely the Tamiḻneṟiviḷakkam. There too, just as in the IA itself, there is no trace of aiyam. It is Nakkīraṉ, the IA’s famous commentator, who silently and unobtrusively introduces the pre-union phases into the IA. He does this by bringing in the opening stanzas of his illustrative Kōvai, the Pāṇṭikkōvai, as examples for kāmappuṇarcci (IA 2), the first union, the traditional first step of kaḷavu. Then he comments on them, using what would become the established vocabulary of Kōvai exegesis. Thus he manages to create an impression of not being innovative, but merely more explicit than the condensed sūtra. He turns them into sub-phases: kāṭci, aiyam, teḷivu, and so forth.

26There is every reason to believe Nakkīraṉ did not invent these. He is quite obviously trying to account for a fait accompli in poetry. Now, what about the TP? Here it becomes difficult to argue in a convincing manner. The impulse of theoretical innovation certainly comes from the side of the IA. For the first time kaḷavu is depicted as a fairly coherent sequence of events, and this seems to be the central idea that has informed the restructuring of the TP’s Kaḷaviyal. Apart from the sūtra TPi 92 on aiyam, doubt, we also find, for example, TPi 98. Before the extensive sūtra dealing with the situations of speech for the hero in TPi 99, it brings up a smaller number of sub-situations that can be variously counted as part of iyaṟkai puṇarcci or iṭantalaippāṭu, depending on the commentator.15 But, the TP does not arrive at the full kōvai series. Is that simply an indication of conservatism? Why aiyam but not kāṭci and teḷivu? And is it plausible that the TP was still being reworked in a phase when the IA already had a commentary? One possible scenario is that both Nakkīraṉ and the person (or persons) in charge of the TP tradition reacted in different ways to the challenge of the kōvai, i.e. the Pāṇṭikkōvai and perhaps even the Tirukkōvaiyār.16 Nakkīraṉ, with the greater textual freedom and flexibility of a commentator, chose the systematic way of sub-classification. The TP author, however, restricted by the existing text and its tradition, chose a punctual insertion, a concession to changing literary fashion. He wrote one extensive sūtra about what might have been felt as the most spectacular of the new sub-situations.

3. kaṟpu

27Let us move over now from kaḷavu to kaṟpu. Here again the situation is quite different and raises yet another set of questions. As with kaḷavu, there is more than one definition to be found in the TP. The counterpart to TPi 487 is 488, which is just as clear and straightforward. Similarly it appears to represent a slightly earlier (or at any rate different) stage of the theory. According to the classical model, the phase of kaṟpu begins with marriage. For some theoreticians, however, the starting point is when the secret love affair becomes public. It is in this frame that TPi 488 enumerates seven basic situations: 1) the revelation of the secret, 2) the elders’ consent to the marriage, 3) joy at the prospect, and then post-marital 4) vexation, 5) quarrel, 6) reconciliation and 7) separation.17

28Completely different in context and intention is TPi 140, the kaṟpu definition of the Kaṟpuyiyal, placed at the very beginning of the section as is to be expected:

kaṟpu eṉappaṭuvatu karaṇamoṭu puṇara
koḷaṟku uri marapiṉ kiḻavaṉ kiḻattiyai
koṭaikku uri marapiṉōr koṭuppa koḷvatuvē.

‘Kaṟpu’ is called HIS, who is traditionally entitled to take,
taking HER, when those traditionally entitled to give,
give [her], so that they unite along with ceremonies.

29Here, from the point of view of classical poetry, the reader is totally at a loss. The ceremonies (karaṇam) referred to are obviously the marriage rites, a topic basically of no concern at all to Caṅkam poets (with very few exceptions). Similarly, the vocabulary of giving and taking is unfamiliar. In fact the sūtra makes sense only when it is regarded as a counterpart of the discarded definition sutra for kaḷavu at the beginning of the Kaḷaviyal, where the term kaḷavu has been exchanged for kāma-kūṭṭam (see p. 98 above and Wilden 2006: 97 ff.). There kaḷavu was compared to the Gāndharva form of marriage. Kaṟpu, then, is a form of marriage that is NOT like Gāndharva, in that the marriage is performed in public. However, we find definitions of the full set of eight marriages not in the text of the TP itself, but only in the commentaries. And indeed it is there too that the vocabulary connected with giving and taking and with suitability reappears. One example should suffice to illustrate the point in question. Iḷampūraṇar on TPi 89 describes the second of eight marriages, that of Prajāpati, in the following words:

(2) piracāpattiyamāvatu, makaṭ kōṭāṟk’ uriya kōttirattār makaḷ vēṇṭiyavaḻi irumutukuravarum iyaintu koṭuppatu.

The [marriage of] Prajāpati [is] giving the girl, in accord with both parents, when someone of a gotra suitable to give the girl to wants [her].

30So here, for the first time, we are faced with a clear phrasal correspondence not between different sūtras or treatises, but between sūtra and commentary. Three explanations seem possible. Firstly, the definition of kaṟpu could be a very late intrusion into the text of the TP, in fact based on the discussion in the commentaries on the kaḷavu sūtra. This looks very unsatisfactory as a temporal scheme and moreover badly motivated from a systematic point of view. The point to make in the given context is that kaṟpu begins with marriage — neither with a secret consent between the parties (which would be Gāndharva or kaḷavu) nor with the revelation of the secret, as some other theoreticians would have it. Secondly, the commentator could have grafted his exposition of marriage on the wording found in the Kaṟpiyal. This seems very unlikely because Iḷampūraṇar is not an isolated instance, but, throughout the schools of Akam poetics, commentaries share the same idiom, be it Nakkīraṉ, Nacciṉārkkiṉiyar, Nampi or the commentator of the Tamiḻneṟiviḷakkam and the author of the Yāpparuṅkalavirutti.

31The third possibility is that the author of TPi 140 had something to refer to that was still familiar to the commentators: a Tamil tradition of Dharmaśāstra. No such texts have survived, except for a number of anonymous verses in the context of the eight marriages. The existence of TP 140 could thus be understood as an indication that those verses were no ad hoc creations in the context of the commentaries. They might truly represent the legacy of a Tamilised Dharma tradition on a Sanskrit substratum.

4. Conclusion: The mutual influence of theories and practices

32I have just presented you with three different varieties of influence that have been exercised from several Sanskrit domains of knowledge on the field of Tamil poetics and most notably a series of aphorisms in the Poruḷatikāram. The first and foremost conclusion is that such influences cannot be imagined as simple and straightforward processes of borrowing. In the case of kaḷavu, we have to distinguish two layers of reception. In the case of the first definition, and that does not concern the Tolkāppiyam alone but the whole tradition of Akam poetics, a certain lack of differentiation between the poetic and the real world have led to the infiltration of a poetic concept with one belonging to the domain of Sanskrit Dharma: kaḷavu was explained as something similar to the Gāndharva marriage, as one of the eight customary types of marriages. Here for once an educated guess as to the probable sources seems possible. The variety of numbers, designations and sequences in Sanskrit treatises have been standardised in Tamil into a series like those found in Kauṭilya’s Arthaśāstra and in the Nāradasmṛti. The subsequent definition, only to be found in the TP Kaḷaviyal, proves to be an adaptation of the Sanskrit list of the ten stages of passion (daśa kāmāvasthāḥ), to be found in various treatises of Alaṅkāra- and Kāmaśāstra. Here, however, it does not seem possible to identify a concrete source. The motivation seems to be a general interest in a dramatic representation of literature.

33In the case of the Akam topos of aiyam, doubt, poetic theory seems to be in the first place a reaction to a literary development, the creation of the Kōvai as an Akam genre. Topic similarity makes it probable that the authors of the first Kōvais, which make doubt a sub-situation of Akam, have been inspired by literary examples from a Northern tradition, perhaps by the Mahābhārata and Damayanti’s plight. In the case of kaṟpu, finally, the influence seems to be, so to speak, of the second generation. What is presupposed in the formulation of the definition of kaṟpu is perhaps no longer a source from Sanskrit Dharmaśāstra, but a Tamil reformulation of Dharma rules concerning marriage. Here I assume the situation to be different for kaḷavu and kaṟpu, because the notion of kaḷavu as a Gāndharva type of marriage has permeated the whole Tamil tradition, whereas the TP definition of kaṟpu appears to have in its turn influenced only part of the later theoreticians. There is no counterpart in the IA.

34I do not claim to have solved even half of the text-critical problems that come with the opening section of the TP Kaḷaviyal. Speaking globally, one might ask now why kaḷavu seems to have been subjected to much more extensive reformulation that kaṟpu. I see two reasons, and one of them is mere chance: kaḷavu comes first in the system, and it is in the beginning where you experiment with models of presentation. The other reason is that the thematic development of kaṟpu seems to have been much more of an additive type. The main growth in the Kaṟpiyal is found with the diversification of the mediators in marital quarrels. These could easily be fitted in without causing a disruption of the overall structure.