One of the most indicative features of a Kadu Golla hamlet (haTTi) is that it is encircled with a fence (beeli) of boxwood thorns (jaali). The Kadu Gollas are Kannada-speakers and live in the eastern districts of Karnataka. Although today most are agriculturalists and sheep herders, their name literally means "Forest Cow Herders". This paper examines the significance of their fence and what it encompasses.

The term haTTi in Kannada can sometimes mean house, but in the phrase gollara-haTTi refers to the hamlet in which Gollas live. ⁽¹⁾ Such a hamlet contains from a dozen or so to as many as 200 houses (mane). Gollara-haTTis are usually associated with rural villages, sometimes adjacent to, but more usually located a short walk from -- up to as much as a couple of kilometers -- the village proper. The phrase gollara-haTTi is usually attached to the name of the associated village: Bandri-gollara-haTTi ("the Golla hamlet of Bandri village"), Ellagatti-gollara-haTTi ("Ellagatti Golla hamlet"), etc. HaTTi can also mean an animal corral (Telugu, doDDi) and the Gollas call their sheep pens kuri-haTTis (cf also aawu-haTTi, 'cattle pen'; . And within a Gollara-haTTi there is always another haTTi, called a deevar-haTTi, within which Gollas construct shrines (guDi) for their deities.
The fence which surrounds a haTTi can be seen in the two photograhs above.  The one is a kuri-haTTi (sheep corral) and the other is a deevara-haTTi, where the gods are housed.  The fence which surrounds the hamlet (gollara-haTTi) is too extensive to show in a single photograph.
 

Figure 3  represents the general layout of a Gollara-haTTi. Although usually round (suttu) in shape, strictly speaking a gollara-haTTi need not be a circle. The Gollas generally think of it as such, however, and the fence is felt to surround the space, "like a waist band encircles (suttu) the body," as one informant explained. A deevar-haTTi almost always a circle and a kuri-haTTi generally is. Although seldom so today, traditional Golla houses (guDsulu), too, are round (guDasu), like an over-turned bowls. Many shrines (guDi), however, still have this shape. Evidentially, Gollas traditionally thought of their domestic spaces in terms of circles, and one of the momentous changes in their lives over the past fifty years or so is a shift to straight lines and squares: virtually all houses and temples built today are square, and they have largely given up herding in favor of cultivation on crops in fields laid out in straight lines.

As a social entity, Gollas describe the haTTi as a community consisting of people related to one another either as "brothers" (aNNa-tamma) and "affines" (or "in-laws", neNTru). In more anthropological terms, it can be described as consisting of two exogamous moieties. While the Golla description could mean simply "people of my patrilineage and other Gollas (ie. people with whom my lineage will marry or recognize a potential of marriage), in fact, it generally means that it consists of only two intermarrying patrilineages, called beDagus. ⁽²⁾ But even when a haTTi consists more than two patrilineages, the beDagus themselves belong to one of only two sections (or clans), for the Gollas see themselves as decedents of two ancestors, Chittamutti and Chandamutti. ⁽³⁾ [see Appendix 1 for legends] The Chandamutti beDagus characterize their relationship to one another as aNNa-tamma, as do the beDagus of Chittamutti. So, whether in this somewhat more expanded sense, or in a more narrow sense, a haTTi normally consists of two intermarrying, exogamous moieties; or, as the Gollas say, aNNa-tamma and neNTru.
 

Although the haTTi is not endogamous there are typically a large number of intermarriages within the haTTi, between the two moieties, and it is conceived rather like a large (bilateral) family. Following this line of thought, the haTTi appears rather like what other South Indian communities think of in terms of a single house, with rooms (another meaning of mane in Kannada) and a "god room" (deevar-mane in Kannada). Individual Golla houses do not generally have separate "god rooms", and instead, the haTTi community as a whole has a single, common deevar-haTTi, housing the lineal deities of its resident beDagus. ⁽⁴⁾ Like a house, the haTTi (and the deevar-haTTi) has an entrance facing East.
 

The individual houses inside a haTTi are not arranged in any specific fashion. I was told repeatedly that there were no sections of a haTTi belonging to one or the other moiety and that the houses are all mixed together. Nor is there any peripherization of some houses in relation to a center. My own (suspicious) investigation proved this to be entirely true. At best, I thought I was able to discern clusterings of houses of immediate affines, but even this is disrupted over time. The overall impression is that, internally, individual living groups are rather democratically distributed, and this fits very well with many other indications of the remarkably egalitarian relations inside a haTTi. ⁽⁵⁾ The only clear spacial (visual) representation of the beDagus is in the deevar-haTTi, where the beDagu deities are set side by side.
 
 
 

The Golla haTTi and 'Encompassed' Dualism
 

"... encompassment ... means ... the action of surrounding or forming a circle about something, especially in order to protect or attack; to contain."

Rodney Needham, Counterpoints, (1987), p. 130
 

In several obvious senses the fence which encompasses (suttu,) ⁽⁶⁾ a haTTi (gollara-haTTi, kuri-haTTi or deevar-haTTi) is meant to protect that which is within. Sheep need protection from wild animals and human theft; Golla domestic work space needs protection from sheep and other animals; the sacred deevar-haTTi needs protection from defilement. The fence visually demarcates that which it protects, particularly since it consists of a most nasty-looking tangle of thorns.
 

It also contains that which it encompasses: sheep (and especially the lambs) when they are not supposed to wander; but also Gollas (especially Golla women) when they are not supposed to wander. Even a Golla man outside the circle of the haTTi is subject to temptations they had better avoid: most of the Golla legends and songs revolve around the disastrous results of straying from Golla ways when Golla boys meet non-Golla girls and the wily ways of the city-folk (see sample in Appendix 2).
 

I have already noted that the haTTi community consists of Gollas, who are divided into two intermarrying exogamous patrilineal groups. On the whole, the only way to become a member of the community is to be born into it or to marry into it. ⁽⁷⁾ Strictly speaking, even this is not the case, since all Gollas are born outside the haTTi and it is a 40 day period of transition before they appear in their mother's arms before an ancestral god (amaawaasye devaru) and are ritually admitted to the community. Marriages, too, take place outside and in front of the haTTi and there is a remarkably long and drawn out transition before the couple finally settles down in the husband's haTTi. All of this, it would seem, means that the Gollas conceive of the haTTi space as a social-sacred one and the fence serves as its ritual demarcation and that they are extremely particular about the social composition of the haTTi.
 

Collectively, the Kadu Gollas distinguish themselves from other communities (even other Golla communities) by the existence of the fence around each of their hamlets. "Wherever you see a Gollara-haTTi you will see a fence," I was told at the first Golla haTTi I visited, and the prediction held true for virtually all of the several dozen haTTis I later visited. ⁽⁸⁾

One might say, then, that there is a fence around the entire Golla community defining them as a caste separating Gollas from others. When a Golla is excommunicated from one haTTi, he is effectively communicated from all hatttis. ⁽⁹⁾
 

However, notwithstanding its physical and spacial materiality, the fence is primarily a (Golla) representation of the community it demarcates. Many Gollas living away from the haTTi are no less members of it, and there are haTTis in good standing (although few in number) with no fences. The boundary which demarcates the community is an abstract one, a social one, not merely intended as physical protection or containment. The group which it defines is perceived in terms of kinship, a group divided collectively and severally into two intermarrying groups, the descendants of Chandamutti and Chittamutti. ⁽¹⁰⁾
 

Dualism as Evaluation
 
 
 

Although the Gollas use of the concept of 'encompassment' certainly antedates it, the more famous use of this term on the Indian scene is that of Louis Dumont. He uses the phrase "encompassment of the contrary" for a series of dual (or segmentary) oppositions (such as priest and king, purity and power, purity and pollution, good and evil) in which there is a hierachical relationship between the pair, with the superior encompassing (or containing) its inferior contrary at a higher level of totality, by which Dumont seems to mean Society, or Hindu Society (Dumont 1982: 224; 1983: 166). Now, with regard to the Golla encompassment, represented by the encircled haTTi, that which is encompassed is felt to be superior, more virtuous, than those who are excluded from it. Surely this is not what Dumont is talking about. Nor are the Gollas encompassed by others: it is the Gollas who build the fence.

The thorn fence which separates Gollas from others and, for that matter, that which separates the Gollas from their own ancestral shrines could conceivably be interpreted as representing an valuative division of purity from impurity. ⁽¹¹⁾

It is true that it is a boundary away from which Gollas keep people, animals and behavior which they consider dangerous in some way, but it not overly clear whether this is in anyway directly connected to what one might think of in terms of a means of ranking others in something like a caste hierarchy. Most other communities are allowed to enter the haTTi (but of course not to join the community and live there), and, for annual rituals, they are invited to do so. Some, such as the Malas, enter only during such ritual periods. Only one caste, to my knowledge, is sometimes said to be excluded from the haTTi: the Madigas, whom the Gollas (and most other castes) regard as untouchable. I suspect, however, that in this the Gollas are merely complying with a standard established in the surrounding context of village society in this region of India. The boundary represented by the fence is one which divides and separates, but it is not necessarily one which is intended to project a hierarchical ranking which would encompass the greater society. That is not to say that in some ways the Golla feel superior to other communities -- which they do -- but this can be attributed to a self-esteem which no doubt all communities feel, and it is a superiority they feel toward all other castes, generally. I am uncertain of how Gollas regulate or enforce the exclusion of other communities. I would expect that encroachments of this sort would be brought to the village panchayat if the Gollas could not deal with the matter directly, by physical (including verbal) force.
 

In a number of open-ended, non-directed discussions with Gollas on the purity-pollution nature of the fence-bounded haTTi, Gollas spoke of exclusion only in reference to their own community members. Thus, Golla women must stay outside the haTTi during menstruation and childbirth; Golla men are excluded if they have had relations with non-Golla women; etc. To be re-admitted into the haTTi community one must be given consecrated milk from the priest of the amaawaasye deevaru of their lineage (wombu). ⁽¹²⁾ A Golla woman who has had sexual relations with non-Gollas, however, can never be re-admitted to the community. Gollas who have been excommunicated are describe as keTTa, 'spoiled, fouled, bad, degenerate, ruined, violated, defeated' (DED 1614); and, a settlement which does not enforce Golla custom and/or has no way of re-admitting its members -- ie. is refused the prasaada of an amaawaasye devaru -- is called a keTTa-haTTi. Golla use of the term keTTu in reference to another makes no differentiation as to whether the state is the result of a person's active volition or whether it is a state visited upon them by another. But it certainly implies that it is best if a Golla stays within the haTTi community and follows Golla ways. A person who does so is regarded as satya, a 'true' Golla, 'one who follows Golla custom'. Whether this is thought to be a state of 'purity,' I can't say.
 

The deevara-haTTi, in its relation to the gollara-haTTi replicates the kind of 'inside-outside' opposition the gollara-haTTi itself is: ie. the fence protects that which lies within and contains it. Again, it serves as much as a visual and symbolic boundary as a practical one. And again, in open-ended, unsolicited conversation, the several reasons Gollas gave for the protecting fence around the deevara-haTTi were to create an exclusion zone from which to keep Gollas who are in a censured state and behavior which is inappropriate; and, especially, to keep other Gollas from stealing the deities. Otherwise, Gollas seem to have fairly free access to the deevara-haTTi and are normally quite casual about people of the haTTi entering the devara-haTTi: it is used as a thoroughfare when necessary; children can play there; dogs, sheep and goats go in and out. It must be kept in mind that the group from which that which is inside is protected is already selected by the haTTi fence. So why an inner-inner zone? Because it has to do with the "inner" divisions (internal categories) of Golla society: the local divisions of the clans. When a Golla is excommunicated, it is by his clan at the jurisdictional level of kaTTemane (kaTTe = 'custom, rule; a seat; to bind' + mane = 'house, room') and guDikaTTu (guDi = 'temple' + kaTTu = 'fine, dues, tax'). When a Golla is admitted or readmitted he must be given milk by the priest of the amaawaasye deevaru. But there are constant challanges to the hierarchy of a clan's authority structure. Whoever holds the clan deity holds the power of inclusion and exclusion. In order to have a shrine in their haTTi, a lower order lineal group must be given the amaawaasye deity from a kaTTemane. But a breakaway local lineage may try to get authority through other means. Hence the threat of theft. In protecting the sanctity of a devarahaTTi a group expects the support and cooperation of their affines. Indeed, their affines have a vested interest in being supportive, because the status and legitimacy of the dispersed segments of their own lineal members is dependent on that of their affines (neNTru). So, normally, they share the deevara-haTTi and the responsibilities of protecting the deities. ⁽¹³⁾
 

As I have said, from the perspective of any individual within the haTTi the community is divided in half, into two moieties, people who are aNNa-tamma and neNTru, 'brothers' and 'affines', terms which are used reciprocally for one another by the Chittamutti and the Chandamutti. No hierarchy is implied in the use of such terms, although the terms do maintain a division; ⁽¹⁴⁾ the whole is divided into to two equal parts. The relationship between these two groups is symmetrical: the Chandamuttinavaru give their daughters to Chittamuttinavaru and the Chittamuttinavaru give their daughters to Chandamuttinavaru. They regard one another as equals: no hypergamous nor hypogamous relations can be discerned within the community: the community is not hierarchically divided in this sense. ⁽¹⁵⁾
 

One of the clearest visual representations of this kind of dual structure of Golla society is in the deevar-haTTi, where we find shrines of the lineage (wombu) deities. Each deity, each lineage, has its own priest who does the puja. The dualism is not always immediately apparent, since the shrines are often supplemented with the shrines of other deities, and it is not essential that each lineage have a shrine. Whether a local lineage wishes to construct and maintain a shrine is largely a matter of convenience and whether they can afford it. If there is a shrine to their amaawaasye deevaru in a nearby haTTi, they may find it more convenient to go there. What we can see, is that when two lineages each have shrine in the same haTTi, they are normally situated side-by-side, in an apparent expression of their equal status in the haTTi.
 

Within these circular shrines dualism also describes the arrangement of many of the items around the central peTTige in which images of the deities are kept. The wooden and metal horses in this shrine are typical of the Pawagada region. Sometimes one is white or blue and the other red; sometimes they are the same color. They stand to the right and left of the peTTige. However, I did not get a consistent explanation for their identity, purpose or placement. Most people said merely they were the horses of the deities whose images were in the peTTige. One informant, however, suggested that they were the horses of Chandamutti and Chittamutti.
 

The names of the Golla ancestors, Chittamutti and Chandamutti, mean the sun and the moon. [photo] The sun and moon motif can be seen on many of the temples (but of course not only of Golla temples).

Dualisms as Opposition
 

Although it may be assumed from a Western point of view that at least some of the dualism we have described above contain features in opposition to one another, it might be best to examine some of the words used for this term (opposition) in Kannada. Besides the concept of 'encircled,' (suttu) which specifically describes the fence which circumscribes the haTTis and separates the community from that which lies outside its boundary, there are at least two terms which are relevant: eduru (for equivalents in other Dravidian languages see DED 680) and maaru/maru (DED 3960, 3903). The term eduru connotes opposition in the sense of "face-to-face; facing; opposed to, confronting; challenging; contradicting." The term maaru connotes opposition in a serial sense: "alternating with; the other, again, repeating, bringing up again, following; changing; reciprocity, exchanging, selling." Both are rather complex terms and often (like the English "oppose") characterize antagonisms, but not necessarily in the cases we are looking at. There is a sense of rivalry in some of the dualities, but conflict is usually under-played.
 

The division created by the haTTi (gollara-haTTi, kuri-haTTi, deevar-haTTi) fence, as we have seen, protects and contains the Golla community and its domestic space in opposition to others in a very specific 'inside/outside' sense. We might ask whether it stands opposed to something else in the sense of the term eduru as well. A haTTi, like most Indian houses and temples, faces East, but it is not obvious that it is in any sense opposed to the East: one might rather say that it is open to the East. The usual explanation is that the East, where the sun rises, is an auspicious direction. This is the direction of the front of the haTTi, where the entrence is, and presumably facing (opposite) the rising sun allows its auspicious darshan and (possibly) lets auspiciousness into the haTTi.
 

The sun and the moon are opposites in the sense of alternates, forever taking turns following one another. While there may be instances in Indian mythology which they confront one another (eduru) in conflict, the sense in which I think they are represented in Golla ideology is in the maaru-sense, alternating with, or following one another. More specifically, shown side by side one another on the temple doors, they are shown as equals, eternally alternating and collaborating with one another in whatever that larger cosmic duty is that the day and night represent.
 

Similarly, Chittamutti and Chandamutti are opposed in the maaru sense, but possibly emphasizing the supportive dimensions of the relationship, over alternation. Their descendents, the Chittamuttinavaru and Chandamuttinavaru, who constitute the two moieties of a Golla village, are also seen in terms of a supportive relationship, as affines. They also reciprocally exchange brides and, from time to time, village to village, change positions with one another in terms of haTTi leadership and dominance. It would probably go too far to suggest alternation -- as the fortunes of one group going up, the fortunes of the other going down -- in the sense of some "limited good," but there is a sense of rivalry between the two groups.
 
 

Opposition Without Hierarchy
 

Clearly Golla society is replete with a series of dualistic divisions. The dualities are expressed in various ways: by fences which encircle and create a division between that which lies inside and that which is outside; by reciprocal terms such as aNNa-tamma and neNTru; by clan names and the kinship behavior of the haTTi residents; and by the arrangement of things.
 

One might display these divisions hierarchically in the sense of genus/species and represent them in a tree diagram, but I'm not sure this would imply that the encompassment of one the by other is the kind of hierachical evaluation Dumont speaks of. We have already examined the question as to whether one member of each pair might be regarded as superior in some way to the other, and have come up with either negative answers or inconclusive ones. Another problem is that the "levels" do not contain similar kinds of oppositions. Thus even though fences similarly define a set, the sets themselves are of a very different nature. What we can say, though is that each circle divides a highly valued inside from a lesser valued outside.
 

There is one aspect of the hierarchical relationship of devarahaTTi to gollarahaTTi that is similar to that raised by Dumont, however. The two halves of the encompassed complimentary opposites represents in one direction by symmetrically opposed descent groups (clans, lineages), or moieties; but, in another direction -- physically represented by the devara haTTi fenced off from the 'profane' world of the Gollas -- it is represented by the temporal power of a yejaman, who protects and maintains order and the spiritual power of the puujaari (priest), who holds the power of inclusion and exclusion. ⁽¹⁶⁾ Now we could maintain that any similarity (linguistic or otherwise) this might have to the relationship of king (Yejaman, ksyatriya) and priest (puujaari; Brahman) in the larger caste system is purely coincidental, that the Kadu Gollas, tribals, are not part of that system, but the coincidence would seem to me to be too great. In any case, the relationship exhibits, at least at first appearances, the same structural principles of what has been called the caste system.
 

Since, as we have already seen, there are many similarities between a haTTi and a house, one might also point out that a similar "inside-outside" relationship that widely exists in South Indian houses and other domestic spaces (see Claus, 1979a, 1979b, 1989, 1991; Ramanujan, 1986, 1989), although in these cases, it is the kitchen and women who are the inner, protected ones and (usually senior) men (Yejaman of the household) who are associated with the outer spaces. We might also note a similarity between the Golla's circles within circles and South Indian temples which often entail a series of walled circles through which one must pass (or may not pass) before reaching the god who resides in the inner sanctum, the garbha guDi. But comparative frames aside for the moment, what we want to know at present is how do the Gollas characterize this relationship.

To be brief, the Gollas assert strongly that the yejaman should always be of the senior lineage and the puujaari of a junior one. ⁽¹⁷⁾ The terms used to distinguish these two lineages of

the same clan, aNNa ('older brother') and tammma ('younger brother'), indicate relative 'precedence' rather than higher and lower rank. And despite the powerful ability to include or exclude the puujaari is not thought to possess a power greater or superior than the yejaman. In fact, if anything, the opposite is true: the senior lineage is regarded as the "owner" of the deity's image and shrine which possesses that power, and the junior lineage that of their employee. It is noteworthy that the yejaman is, within the senior lineage group, an elected position, with considerations of individual personality characteristics such as consensus-type leadership within the haTTi, ability to speak in a larger village forum for the haTTi, etc., while the puujaari, within the junior group, is normally an inherited position, or rather, an inherited right to employment. Should the position fall vacant for any reason, a new puujaari is sought and hired, and given the inheritable right to serve. With regard to relative purity and pollution it is noteworthy that the function of the puujaari can be played by a Helava, a separate caste decidedly inferior to the Kadu Gollas (see Apendix 1, "Origin of the Helavas"). ⁽¹⁸⁾
 

Although there may be nothing of great significance in the opposition of yejaman and puujaari, it remains true that in a guDi at the very center of the deevara-haTTi are the emblems of the deevaru (an ancestor), contained within a box (peTTige) which is only very rarely opened for others to gaze upon. The fear, Gollas tell me, is that these might be stolen, and with them would go the authority of the yejaman. It is in the yejaman's self-interest to proctect the shrine: the possessor of the these emblems has the ability to decide who is a member of the community in good standing.
 

By way of conclusion, I would suggest a homology between the south Indian house and the Golla haTTi. Both are characterized more by concentric circles of 'inside/outside' (like the Tamil aaham - puram) than by a ladder-like hierarchy of superior/inferior. Both are literally instances of "encompassment", but nothing like Dumont seems to have in mind by that term. What is encircled is vulnerable and thus needs protection. It is also that which contains a link between the past, present and future of the group. In the case of the haTTi, this is the emblems of the ancestors; in the case of a house, it is the women. Divisions of superior/inferior, pure/impure are not what characterizes the relationship; rather, it is matters of authority and responsibility and control of the groups reputation and future status.
 
 
 

Bibliography
 

Claus, Peter J. l978. "Oral Traditions, Royal Cults and Materials for a Reconsideration of the Caste System in South India" Journal of Indian Folklore, 1, (1):1-25.

---- l979. "Mayndala: A Myth and Cult of Tulunad" Asian Folklore Studies, 38 (2):95-129.

---- 1991 "Kinsongs." In Arjun Appadurai, Frank J. Korom and Margaret Mills, eds., Gender, Genre, and Power in South Asian Expressive Traditions. Philadelphia: University of Pennsylvania Press.

Dumont, Louis. 1982. "On Value," Proceedings of the British Academy 66. Pp. 207-41

---- 1983. Affinity as Value. Delhi: Oxford University Press.

Needham, Rodney. 1987. Counterpoints. Berkeley: University of California Press.

Ramanujan, A. K. 1986. Two Realms of Kannada Folklore. In Another Harmony: New Essays on the Folklore of India. Edited by Stuart H. Blackburn and A.S. Ramanujan. pp. 41-75. Berkeley: University of California Press.

Ramanujan, A. K. l989. "Is There an Indian Way of Thinking? An Informal Essay."Contributions to Indian Sociology 23(l):4l-58.

Thangavelu, Kirtana. 1995. "Itinerant Images: Embodiments of Art and Narrative in Telangana." Marg.

Trawick, Margaret. 1990. Notes on Love in a Tamil Family. Berkeley: University of California Press.Basso, Ellen B.

Yalman, Nur. 1967. Under the Bo Tree. Berkeley: University of California Press.
 
 

1. It appears to come from the Dravidian root, paTTi (DED 3199).

2. I shall use the term lineage to translate both beDagu and wombu. Both are patrilineal descent groups; the latter being a sub-section of the former.

3. See Appendix 1. The situation is actually more complex than this, for a third section was admitted into the fold ("given bedagus") at a later time. See Appendix 1 for the story of how the third group was added. People of both of the original sections may marry with members of this added group. While perhaps obscuring the overall moiety-character of the Kadu Golla community, the addition of a third section does not, except in the case of a small percentage of very large haTTis, affect the characterization of a haTTi as consisting of intermarrying moieties.
 

4. Cf. Trawick's highly insightful discussion of walls in Notes on Love in a Tamil Family, Chapter 2, "Postlude": "Households ... are better seen as points of confluence than as 'hold' in any stable sense." (Trawick 1990: 87-88) Within the haTTi, Golla houses are, for Gollas, at least, like Tamil houses, "infinitely permeable"; but then, all Gollas are relatives. And the haTTi itself, is more what we think of as a house "hold".

5. The division of clans is what Dumont (1983: 166) calls equistatutory opposition, an opposition in which the two poles or terms have the same status. As he would predict, "... the hierarchical principle of caste ... does not enter the basic framework of South Indian [kinship] categories ... [these are] something like an island of equality in an ocean of caste."

6. DED 2238. See also 4348, which implies more a sense of surrounding, proscribing, sealing off or besieging.

7. Marital residence is normally patrilocal, a woman going to live in the haTTi of her husband's father. For the many couples who marry within the haTTi this means a woman need only shift into a different house. There are also some instances of uxorilocal marriage residence, men coming to live in the haTTi of their wives. 

8. Gollas are not the only castes which live in separate hamlets surrounded by thorn fences in this part of India. The migrant LambaDis (also called Banjara or Suukkaliga) live in hamlets called taNDa. The famous forts (durga) of the BeDa Nayakas (Telugu Boyya) Palegars and the Myees BeDas might well be derived from such structures. 

9. Similarly, Yalman, in speaking of the Sinhalese kindred (pavula), insightfully points out, "If the kin group looks after itself, then caste boundaries are also maintained (1967 206)." The Gollas, however, do have an elaborate system of judicial institutions (kaTTemane) for handling, among other things, matters of social boundaries. Although its higher levels are theoretically lineal based, at the root of this system is the haTTi, with its strong bilateral kin ties.

10. See note 1, above, for a qualification of this sentence.

11. For these vague and ambiguous English terms the Gollas often use the term satya (Skt. 'truth') to indicate the "purity" of the community (in relation to others) enclosed in the fence. In Appendix 2, for example, the culture hero Iranna warns away a Beda woman who wishes to come into his cattle pen to sleep with him: "Stop! If you go near the cattle pens the cattle will become impure (satya keTTitu). Go to your fort (durga). If you come near, the cattle will die." He later says her proximity to the haTTi is prohibitted: tappu banda, 'proscribed, taboo'. It should be noted that the Bedas and the Gollas are approximately equal in caste rank, although the Bedas have long held far greater political power.

12. Although the term amaawaase deevaru would appear to be derived from the Sanskrit word amaawaase, meaning "new moon day," in function it seems closer to ame, the term for the ceremonial pollution incurred by childbirth and contact with other matters of feminine sexuality (see DED S25). 

13. The presence of the deities representing the clans resident in the haTTi is also a show of allegiance to one another over time; an expression of their 'familiness'. 
 

14. The phrase anna-tamma, 'elder brother'-'younger brother' it may be noted, does imply a hierarchical relationship. I discuss this at a later point in this paper.

15. At a closer look, too, we see about equal amount of patrilateral cross cousin marriage as matrilateral cross cousin marriage, and often one form will be followed in the next generation with the other. As a traditional practice, when a girl comes of age, her mother should ask her (eldest) brother, "Do you want this girl for yourself, or for your younger brothers, or for your sons (ie. for his immediate local lineal group, the patrilineal joint family.) The net effect would constitute a system of bilateral cross cousin marriage.

16. There are actually usually five offices for a kaTTemane and guDikaTTu: jejaman, pujari, gauda, delavai and heluva. The first two are essential even for a haTTi. The presence of the latter three is variable. The delavai is usually of an affinal clan; the Helava is of a different caste, the Helava caste of Golla "genealogists"; the Gauda is sometimes of the village's dominant caste.

17. Dumont (1983) also notes somewhat similar internal distinctions of junior/senior lineages among the Pramalai Kallar and Maravar of Tamilnadu. Dumont believes the idea of 'junior' generally refers to progeny of illegitimate sexual liasons: "...the basic fact is the distinction between them ... and the full-fledged son, i.e. the eldest son of the first or principal wife..."). This is not the case with Gollas, however. With the concept of "senior/junior" (doDDa/sanna) as a categorical labels for the sblings and spouses of one's parental generation, there is no hint of legitimacy involved. The term sannamma (or cinnamma), for example could be either "mother's younger sister", "father's second wife", or both, in the case where the father took his first wife's younger (and perhaps classificatory) sister. It might also be used for an 'unacceptible' mate of the father, but in this case people are usually quick to point out that it is out of politeness. For those who are siblings (aNNa/tamma, akka/tangi), we cannot distinguish whether the junior/senior distinction refers to children of one mother or different mothers. 

18. This story relflects many themes (junior lineage, defilment and handicap) characteristic of the relationship between a caste and its "dependent" "servants" and "genealogists" found among other herding communities as well: the Kurubas and the Tagari Jogalu (or Kadu Siddolu), the Erra Gollas and the Mandecculu (Thangavelu, 1995). See also Dumont's dicussion of Agambaidyar (aham+adi+yar, 'those who stand within'), a group junior to the Maravar of Tamilnadu.