IN THE PREVIOUS CHAPTER, we saw informal relationships of influence form structures that combined horizontal and vertical organization but were different from the “tree” structures common to formal hierarchies. These tree structures have a potential to facilitate collective action in the interests of a single person that influence structures do not. How could such structures arise? In this chapter, we examine relationships of patronage—relationships that are intrinsically antisymmetric and thrive in environments of inequality. We recall that the other structures that we investigated often foundered on problems related to inequality between persons—patronage relationships not only are immune to such problems, but they are likely to arise where other structures have generated inequality. Finally, we will find that such relationships naturally tend toward treelike structures; these spontaneous structures then can serve as the backbone for more deliberate governance structures, as will be explored in chapters 7 and 8.
In chapter 5 we encountered “multihub” communication structures, in which the people in question can be divided into two types, hubs and spokes. Any spoke is only connected to a single hub, while hubs are connected both to spokes and to other hubs. Such structures, we found, reasonably arise when people must establish relationships for the purpose of spreading information. I previously treated communication as an intrinsically symmetric relationship. But some kinds of communication are inherently antisymmetric. An example might be religious-philosophical “schools” in the ancient sense: a charismatic teacher or prophet forms the center of a network of disciples. Such a structure is formally quite close to a hub structure, except composed of antisymmetric relationships.
Accordingly, we might recast the simple hub structure (reproduced in figure 6.1, left) in a way that accentuates the difference between the hub and spokes as in the figure on the right, which, following Scott (1972: 95; also see Clark 1973; Mayer 1966), I shall call a “triangle.” To emphasize the difference between triangles and wheels, I shall call the hub of a triangle the “apex,” and the other members the “base.”
Figure 6.1. Wheel and triangle
The second figure clearly introduces a vertical dimension between the hub and the spokes, while the wheel representation does not, as the triangle spans a directed or vertical space (which is not necessarily true of the wheel).1 It also uses the arrows that indicate an antisymmetric relationship. Thus while the overall arrangement of nodes and edges—people and their relationships—is the same in both figures, they differ according to the nature of the relationship. The bidirectional communication of acquaintances tends toward the former, and the unidirectional communication of master and discipline toward the latter.
This same type of triangular structure is found for other sorts of relationships, which while not primarily communicative in nature, still bear a familiar resemblance to the charismatic teacher. An important set of examples pertains to equally charismatic war chiefs. Tacitus describes the German barbarians as organized into such hubs led by charismatic warriors (principes—big men), who competed with one another to get a following, a personal retinue (a comitatus). As Tacitus (Germany 7, 12, 14, 11) says, the others listen to the leader “more because he has influence to persuade than because he has power to command. If his sentiments displease them, they reject them with murmurs; if they are satisfied, they brandish their spears.”2 This brings us to one of the characteristic features of these triangles, namely that the attachment of the “base” to the “apex” is conditional, even if fierce. (This is also generally true of leaders in hunter-gathering societies: for examples, see Helm [1972: 77] on the Dogrib; Hanser [1985: 61f] on New Guinea in general.) Members are always free to desert a teacher for a more charismatic one, or to switch to a more promising war leader. As Critchley (1978: 43, 32) says regarding the Germans, “If they did not get what they wanted they moved on,”
As a consequence, while the multitude of followers may tickle the apex’s ego, she or he tends to be rather weak structurally. While such apexes appear outstanding, their charisma comes largely from their followers: says Tacitus, “It is the renown and glory of a chief to be distinguished for the number and valour of his followers.” To keep the followers, the chief is forced to keep gifting them with feasts, and to do this, he must keep fighting victoriously, which in turn requires that he keep his followers. This holds for charismatic teachers as well as war chiefs: Jesus evidently was insufficiently attentive to the feelings of at least one member of his base leading to problems serious enough to lead to his temporal dethronement. As Zablocki (1980) argued, there is a serious sense in which the “big man” is not a man at all, but a node in a network. The very charisma that makes this person appear extraordinary may better be seen as a property of the pattern of attributions of charisma, reversing cause and effect. To understand the apex, then, we need to understand the formation of the triangle.
In the previous chapter, we saw multihub communication networks evolving as all actors followed a single heuristic, namely “connect to the ‘hubiest’ person you can.” While we imagined that there might be heterogeneity in willingness to be a hub, this was not actually necessary to produce a multihub structure. One could imagine such a structure arising among interchangeable persons. But with triangles, the antisymmetric nature of the relationship between the apex and base implies a categorical distinction between the two sorts of persons. Of course, in reality, the existence of this categorical distinction is far from obvious until the formation of the structure. While evidently some charismatic religious leaders know themselves to be apexes-in-training shortly after birth (if not before then), others pass through a less clear-cut developmental process. Many of these begin as base members and are appointed as apexes-to-be by the apex as successors. A few, like certain charismatic war leaders, become apexes by displacing (not always, but sometimes, killing) the former apex.
In many other cases, however, it appears that there is a relatively long process whereby one member distinguishes him- or herself from the others and rises to the position of apex. In the intermediate period, this person has ambiguous relations of near-equality to others, and hence the structure is relatively complex, until the rising member achieves a qualitative distinction and a clear triangle emerges. We only have good information on this process for relatively recent charismatic leaders, though there is good reason to think Jesus started as a spoke of John the Baptist (see Matt. 3:13–17, evidently trying to explain away the subordination of Jesus to John implied by his baptism reported more simply by Mark 1:9). Histories written after some person has attained an apex position tend to be silent on or even deny the existence of any preapex life, but for more recent cases we may witness an ascension process.
For one example, we can consider the case of Charles Dederich, the charismatic leader of the psychological/rehabilitation community Synanon. His ascension stretched out over many years, and during this time, it was largely relationships other than leader-follower that held the group together (such as informal friendships). Thus he could avoid relying on weak or incomplete charismatic leadership until his position was crystallized; at this point other relationships were abandoned and the triangular form solidified.3 For another example, Mel Lyman, a now largely forgotten self-proclaimed avatar and God, began as an occasional harmonica player for Jim Kweskin’s famous jug-band, and only became God after reducing Kweskin from hub to spoke (For Synanon, see Yablonsky 1965; Endore 1968; Ofshe et al. 1974; Mitchell, Mitchell, and Ofshe 1980; and Gerstel 1982; on Lyman see Felton 1972.)
Because in such cases the “group” is dominated by the structural imperative of the triangle, it is extremely fragile in response to the removal of the apex. More, we might say that society abhors a vacuum at the apex, and hence former spokes may suddenly find themselves becoming an apex when there is an unanticipated removal of the former apex. As may have been the case for early Christianity, the structure is reconstituted around new apexes who are sucked into the hole (in such cases, frequently rival apexes such as Peter and Paul).4
In these cases, the qualities of a particular person (real or imagined) lead him or her to become differentiated from other persons and so to establish a relation of inequality with others. In other cases, a preexisting material inequality is responsible. One historically important class of examples pertains to “households” in the patriarchal sense of a man’s dependents, which would include not only children and grandchildren, often brothers and sisters and assorted in-laws, but also servants and anyone else who attaches him- or herself to the household. Indeed, as we shall see, this form often seems to be the template by which other homologous structures are given a cultural interpretation.
Previously I emphasized the structural homology between a wheel and a triangle. While the quality of the relationship in question can distinguish them, their structural difference only becomes apparent when multiple structures are concatenated. Consider the unification of four wheels via the introduction of a “hub’s hub”—a second-order hub that connects local hubs, as in figure 6.2. This structure, if properly pulled and stretched, reveals itself to be identical to the “tree” form that we saw characteristic of unilineal descent structures. (Imagine grabbing the central node, pulling upward, and viewing the structure from the side.) The structurally identical tree of the same relations is drawn in figure 6.3.
Figure 6.2. Wheels connected by hub’s hub
Figure 6.3. Figure 6.2 recast as tree structure
This sort of structure, which we can call a “pyramid” (defined below), involves only vertical ties between persons who may be seen as “higher” or “lower.” In stark contrast is a structure that results from the concatenation of wheels via the establishment of horizontal ties between hubs, as in the following example from the previous chapter (see figure 6.4). Here, while there is some differentiation between hubs (two are connected to three other hubs, while the other two are connected only to two other hubs), there is not the same qualitatively different position of a “hub’s hub.”
The pyramid is, I propose, what true triangles are likely to form when aggregated. Accordingly, just as a molecular chemist can determine structure by growing crystals, we understand social structure by examining the larger structures that they tend to form. Now it is not the case that all triangles can be aggregated: charismatic prophets, for example, are notoriously unwilling to work together in any form. But if they do join forces, they will do so either by subordination to a common superprophet, or by one acknowledging the superiority of the other—they will not work together as equals (Benjamin Zablocki, personal communication).
Figure 6.4. Multihub structure
We have thus defined triangles as structural forms that are homologous to wheels but involve antisymmetric relations and that concatenate as pyramids (and not as multihub structures). But in previous chapters we have tried to generate structure by looking at the nature of certain relationships. Let us return, then, to the relationships we have investigated in chapters 4 and 5. We introduced horizontality into vertical relationships by allowing for some persons to decline to establish a relation with others. The resulting structures, although combining horizontal and vertical differentiation, differed from trees in that persons could have ties to more than one superior. This seemed to fit the class of relationships understood as “influence,” in which the underlying inequality might be seen as inherently sociometric—even if people really are rankable in terms of expertise, say, what is crucial for the emerging structure is only the degree to which there is a an architectonic distribution of social beliefs of authoritativeness.
To anticipate: when it comes to exogenous material inequality, however, relationships may arise specifically to mitigate the inequality. The materially advantaged can clearly offer the subaltern material support; the subaltern, however, can only offer their loyalty. Hence such relationships—generally called “patron-client” relationships—tend to concatenate in structures that add an additional heuristic to that of the triset (which was “do not accept influence from someone lower than yourself”). This heuristic is, “do not accept a client who already has a patron.” As in previous chapters, we now begin with the nature of this relationship and draw out its implicit structural potential.
This understanding of patronage is more or less the anthropologists’ use of the word, as opposed to the political scientists’ use (to follow the distinction of Weingrod 1968).5 A patronage relation links a patron at the apex and a set of clients as the base of a triangle. The patron protects or advances the interests of the clients, and the clients rally to the support of the patron when he is challenged. These relationships, as political scientists and anthropologists have noted, tend to arise in settings in which either there is a strong anarchic component to social life, or where preexisting inequalities are severe and must be moderated for the purposes of interaction (Eisenstadt and Roniger 1980: 60; Bloch 1961 [1940]: 160).6
Taking the former point first, it has been noted that widespread patronage never coexists with strong states (Wolf 1977: 174; Weingrod 1968; Scott and Kerkvliet 1977: 443; Critchley 1978: 119).7 Or perhaps it is more accurate to say that they exist only where strong states do not reach, which includes the contained anarchies of the interstitial realms of interaction within societies with strong states, such as illegal activities (for an example see, e.g., Cornelius 1977: 348). Further, as Bloch (1961 [1940]: 142, 148, 225, 443) suggests, patronage structures are more likely where kin groups are unable to fully protect people against violence, perhaps because of the decreased ability of bilineal structures to organize vendettas, as discussed in the last chapter. In violent circumstances, the alternative to organized vendettas may be anarchy. Such anarchy, as Landé (1977a: xxx) points out, leads to the weak being heavily dependent on the mighty for life and limb, but also leads the mighty to need to command followers in order to stand up to other mighties.
In sum, anarchy leads to a need for protection among the less powerful. For example, European feudalism was (as we shall see in more detail) largely a set of patronage relations, and one of the factors supporting the feudal system in the absence of a strong Frankish state was the anarchy resulting from feuds over land (in turn due in part to an absence of primogeniture among the landed).8 Similarly, patronage systems were common in traditional east Africa where refugees would be provided for by patrons who would not technically adopt them but still refer to them as children (and be called “father” in return) (see Mair 1961: 316; Mair 1977: 95 for the case of the Gusii; and cf. Mair 1977: 97f, 141f). Where patronage becomes a normal part of coping with endemic anarchy (as in Europe after the fall of Rome), the very development of the patronage structure increases the need for the powerless to seek security, and the powerful to increase their number of clients (Ganshof 1964: 3, 23).9 Conversely, when states are able to pacify their interiors, we see patronage relationships either weaken or reform into relationships of sponsorship that now focus on control of state bureaucratic careers (cf. Kettering 1986: 213f for seventeenth-century France).
Thus we see that the anarchy that promotes patronage systems is rooted in a more fundamental dynamic, namely a differentiation of persons in terms of power, and a symbiotic need for the more powerful and the less powerful to enter into relatively stable relationships (even if that “need” on the part of the less powerful is induced by threats on the part of the more powerful). Consequently, the precise form of the inequality in question may be crucial for the structural tendency. I begin by examining simple “two-class” systems, and then more continuous inequalities that may or may not be copresent with bifurcations.
In many patronage systems, persons are divided up into two classes, potential patrons and potential clients, on the basis of some preexisting categorical distinction. Relations then may or may not be established that personalize and/or mitigate the inequality associated with this bifurcation. Most obviously, if land ownership is the basis of the division, there may be a clear tendency toward a neat bifurcation: either one owns land or one does not (see Mintz and Wolf 1950; we deal with complications later).
In certain cases, this division is established by colonization: for example, in Peru (as in many other parts of Latin America), the native Quechuan population was divided up (frequently along preexisting lines of community) and “given” to an individual Spaniard, who then became their lord (Guasti 1977: 424; Critchley 1978: 134f). Similarly, Turnbull (1962: 151–54, 203f) describes how Bantu families “owned” the local pygmies (Mbuti). It is worth reemphasizing that the essential point is not the division into two classes, but the establishment of “many-to-one” interpersonal relations between them.
In a few other cases, a similar bifurcation can arise on other bases, such as the division of a number of East African kingdoms into (presumed “original”) ruling and (presumed immigrant) subject classes (Mair 1977: 114, 141), or the division of Rome into plebeians and patricians. The Roman “patron” was paradigmatically a man who had freed a slave but retained rights and obligations with respect to him (Kenny 1960: 14): seeing Rome as divided into patricians and losers, the Romans concluded it was the duty of the former to fatherly guide the latter (Critchley 1978: 101). According to the (historically doubtful) account of Dionysius of Halicarnassus, the Roman system of patronage stems from a decree of Romulus, who allowed “every plebian to choose for his patron any patrician whom he himself wished” (Wallace-Hadrill 1989: 66; Taylor 1968: 41; Drummond 1989: 91f). This case nicely encapsulates the voluntary nature of the patronage relation: the choice of patron is in most cases formally voluntary (though the relationship is also often hereditary), but cannot (at least in a two-class system) be understood without recognizing that there is at least a healthy dollop of compulsion mixed in.10 Thus in the two-class system of Rwanda, every Hutu “chose” to be the client of some Tutsi; those who did not were defenseless against the predation of other Tutsis. But “even if it is true that no Hutu could afford to be without a lord, he could still choose between different lords” (Mair 1961: 315; 1977: 143, 154).
Systems based on land are easily made near universal (as in the famous edict of Charles the Bald of 847: “We also wish that every free man in our kingdom may choose as his lord whomsoever he will” [Ganshof 1964: 30]). Since everyone needs somewhere to stand, declaring that one must be a client of he who holds this land is a rather good way to extend the structure of patronage relationships to encompass more persons and their various interactions. Indeed, this principle of “nulle terre sans seigneur” (no lordless land) was taken in France (though not Germany) by the thirteenth century to imply that there could be no privately held (“allodial”) property whatsoever (Ganshof 1964: 130, 155; Beeler 1971: 151, 214, 217; though Anderson 1974: 148 points out the limits to enforcing this principle). Various changes in economic conditions or taxation policies can force small freeholders into seeking protective alliance of greater lords, in something akin to the proletarianization of the petit bourgeois (this seems to have occurred in France at the turn of the millennium [Fourquin 1976: 51]).
While a basis in ownership of land can lead a patronage system to be quite simple and inclusive, this depends on the specifics of property relations. Multiple rights in land, as seen in feudal Europe, or practices akin to subletting, can lead a property-based structure to be rather complex.11 In such cases, we are more likely to see continuous patronage structures, to which we turn next.
We have seen patronage structures arising when there are two discrete classes. But it is also possible for patronage relations to arise where there are more than two classes. For example, we are likely to see “sponsorship” forms of patronage where there are stratified systems of ranking such as those associated with institutions such as the military or the priesthood that are highly encompassing of their members’ lives and organize these lives into careers that may progress upward in organizational terms. But here there is a preexisting interaction structure that makes this a less central case for the current investigation, though I return to such cases briefly below. Other than such formal structures, then, it seems that the major alternative to the “two-class” basis for patronage relations is when the inequality arises on the basis of some continuous ranking of persons in terms of some resource we will denote as ζ. Then, given any two persons, A and B, if we use the notation from the last chapter, we can say that if a relationship arises between A and B such that A is the patron and B the client, we know that ζA > ζB. (Fourquin 1976: 124 has an informal version of this derivation for the case of feudalism, where ζ is termed “power.”)
Most frequently, patronage relationships grounded in such continuous inequality exist within a stratum that is, if not elite, at least not totally subjugated—for example, European feudalism (see, e.g., Bloch 1961 [1940]:444).12 While serfs might not necessarily have patron-client relationships with their lords, such relations did at times exist; more important, the lords were then arranged in patronage pyramids as in the simplified diagram of figure 6.5 (note the imperfections of the elite structure and the incomplete absorption of the nonelites into the structure).13 Thai feudalism was basically similar, in the bottom level of the pyramid being the subjugated, with less categorical distinctions between higher-level patrons. A third example is Rome in the imperial age—the relationship between patrons and liberti (the freed slaves who took their ex-masters as patrons) was different than that between the patrons and their other clients. The former relationship was fixed by law, the latter only by interest and custom. While there was only one level of liberti, the patron of one client might himself be the client of another (Wallace-Hadrill 1989: 76f; Drummond 1989: 100f).
Figure 6.5. Patronage pyramid
When both discrete and continuous inequalities serve as the basis for patronage relationships, there is a fundamental division in the population, usually based on ownership of land or membership in a governing elite. Every member of the nonelite may be assigned to the domain of one elite. Within the elite, there is a more or less continuous gradation (although many persons are roughly “equivalent”), and elites lower down appeal to those higher up for sponsorship and promise service and support in return. (This is a common Mediterranean pattern we shall examine below.)14 Thus it is possible for a patron to be the client of a second-order patron; following Scott, we will call the highest-order patron, her clients, and her clients’ clients a “pyramid.”15 For example, in addition to the fundamental division between Hutu clients and Tutsi patrons, poorer Tutsi could also be clients of richer Tutsi, leading to a pyramid that connected all adult men and ended with the king as the patron of the greatest patrons (Mair 1961: 322; Mair 1977: 143f). Similar relationships develop between heads of state when one is conquered by the other but left in power as a client (for examples from the Indian subcontinent, see Gunawardana 1992; for the Aztec case see Hassig 1988: 19, 26, 125, 177, 256f).
In sum, patronage structures arise from preexisting inequality, when there is a conscious attempt to ameliorate this inequality by the introduction of diffuse personal relationships (Eisenstadt and Roniger 1980: 49f; Kettering 1986: 38; Bourne 1986: 57, 190). Given our investigations in the previous chapter, we might expect structures of such relationships to form trisets similar to those we uncovered for influence relations. But instead, patronage relations tend to form trees. Why is this?
We will soon explore the structural imperatives that lead to the production of trees, but we can begin by pointing to their cultural correlative, namely a subjective understanding of the relationship as akin to parent-child relations. Thus it is natural for the patronage structure to be assimilated to the unilineal trees seen in the last chapter. How can one be inferior to someone yet have a positive relationship? One answer is, “if the inequality is similar to that between a father (as it turns out) and a child.” The word patron comes from father, the words vassal and thegn from young boy or servant (the former from the Celtic) (Ganshof 1964: 5; Critchley 1978: 104; Bloch 1961 [1940]: 155f, 182); both signeur and Herr come from the “seniority” of the patron (Fourquin 1976: 115; Bloch 1961 [1940]: 145). In Japan since the Tokugawa period patron-client relations (ubiquitous in many fields) are called oyabun-kobun relationships: parent-status / son-status (Critchley 1978: 105).
The connection between patronage structures and patrilineal descent trees is more than a vague affinity: patronage structures differ dramatically according to whether or not an actual patrilineal familial structure underlies social relations—where it does, patronage structures are simply absorbed into lineages.16 It is generally only where patrilineal divisions are not major structural principles for social action (other than inheritance) that classic patronage structures exist (cf. Anderson 1974: 232). Perhaps for this reason, patronage structures are frequently found in bilineal cultures (notably those around the northern shore of the Mediterranean). Indeed, the classic patronage relationship is often that of the “godfather” relation in which a nonkin of generally higher social standing becomes, at christening, a special protector and sponsor of a child, a role that in a patrilineal society might be reserved for an agnate (e.g., Campbell 1964: 222f; compare Duby 1991: 63).
In sum, patronage structures arise as independent social structures to deal with inequality when there is not a preexisting patrilineal structure that is linked to the inequality. Because the amelioration of inequality is so crucial for understanding the structural nature of the relationship, we go on to examine this aspect in some detail.
Given preexisting inequality, patronage systems are understood to reduce the potential arbitrariness of the domination of the superior party by establishing affective quasi-familial ties that imply limits to exploitation (see, e.g, Guasti 1977: 424; Powell 1970: 412). While the inequality may be quite formal (e.g., the case of bifurcations) and/or specific (e.g., the case of landlordism), the relationship of patronage may be quite informal, generalized, or diffuse (Landé 1977a: xxi). What is essential is that the relation of patronage implies a paternalistic bond whereby the patron deigns to take the interests of the client to heart. This bond may merely “soften the terms of an unequal exchange” (Graziano 1977: 363 for the case of Sicily; also see Campbell 1964: 261, cf. 218) and hence refer only to the interaction between the patron and client, or it may involve actions that the patron takes with others on behalf of the client (the client thus being the indirect, not the direct, object of the patron’s actions). Just as the mitigation of the unequal exchange may still be seen as in the ultimate interests of the patron, so such sponsorship can also be seen as a self-interested act whereby a patron pits his client against some other patron’s client so that the former’s success redounds to his own glory.
Since the existence of paternalistic protection is often one of the legitimating arguments claimed by those persons eager to support existing systems of inequality, there has been a tendency for social scientists to discount its very existence, thereby simplifying their implicit or explicit critique of such inequality. But this is a dangerous game of ignoring precisely that aspect of inequality that is its greatest strength, namely its (conditional) weakness. Whether these limitations are enforced or only voluntary, they affect the structural tendencies associated with relationships between unequals. The existence of the bond in no way contradicts the vast inequality between parties; on the contrary, it presupposes it. Similarly, there is nothing particularly warm and wonderful about the fact that the relationship is indeed personalistic—this very quality can be the ground of new forms of power over clients as patrons become free to interfere in their clients’ personal affairs (see, e.g., Pośpiech and Tygielski 1981: 95). Similarly, the pervasive nature of the inequality between patrons and clients should not lead to skepticism about the protection offered by the patron; it is for this very reason that this protection is necessary.
The first feature of the patron-client relationship, then, is that it exists to ameliorate preexisting inequality. Accordingly, the discussion here will not touch hub systems that generate inequality (if there be any); whether these can aggregate into the pyramidal structures that prove characteristic of patronage relations is a separate question that will be dealt with later. The second feature of these relationships is that they are voluntary, by which I mean that at least one of the parties (and not necessarily both) has the option of refusing the relationship. In the Thai case, clients were free to choose other patrons and patrons to reject clients (Englehart 2001: 22). Now given a case in which one portion of a population is expropriated and left at the mercy of the other, we might be hesitant to describe as “voluntary” any relationship in which the former attached themselves to the latter for protection from the most outrageous abuses.
But it may make a great deal of difference for the resulting structure whether or not all members of the expropriated class are linked to a patron, and if so, whether they have any choice in which particular member of the expropriating class is their patron. If the answers to these questions are “yes” and “no” respectively, and hence a specific relationship is imposed on each client, the resulting structures may be quite different from other cases in which a personalistic relationship ameliorates inequality. Thus the case of colonial conquest and privatized government over groups of conquered natives as in Spanish South America that was mentioned above may not constitute a true patron-client system. On the other hand, patronage relations in similar areas between outright slaves and their owners (as in Brazil as described by Hall 1974; also see Downing 1992: 39), may have been good examples of patron-client systems, for the simple reason that not all slaves had relations of clientage with their owners. Thus the “voluntary” nature of the relationship—in that the owner did not necessarily take all slaves as clients—is likely to be related to other aspects of the relationship, both in terms of its content (what actions were involved) and its structural potential. In particular, relationships in which the clients lack any choice as to patron are likely to be relatively flat, as the patrons need less from second-order patrons in order to maintain their clientele.
This spontaneous or voluntary nature of patronage relationships probably has a great deal to do with their durability once constructed: the severing of one relationship may not change the structure, since it is quickly replaced with a similar one. Indeed, it is the very durability of patronage relationships that has led to their neglect by American social scientists: as the heirs of uppermiddle-class reformers dedicated to quash any arisings of such “irrational” relationships in urban machine politics (see, e.g., Hofstadter 1955), for quite some time social scientists evidently thought that the best thing that could be done with patronage relationships was to ignore them and hope (indeed theorize) that they would go away. This meant ignoring a type of relationship that (at least until extremely recently) was of fundamental importance perhaps in most of the world’s societies, including most European democracies, especially those with large agrarian populations.17
One justification for such marginalization was that in the developed world an emphasis on patronage across fields or domains of action seemed related to the particular religious orientations of certain Mediterranean societies, and hence was a cultural aberration as opposed to an inherent aspect of social structure. Certainly in many Mediterranean cultures there is an understanding of religion as patronage, where saints (or minor gods) are treated as simply the higher-level patrons (Whyte 1981; Foster 1977 [1961]: 16; Weingrod 1968: 390; cf. Eisenstadt and Roniger 1980: 68, who add some unfounded assertions), with God perhaps as the ultimate patron, the only one who needs no favors from anyone (Kenny 1960: 15, 22f on Spain).18 Many other theologies also begin with the assumption that people are basically powerless and must petition some divine spirit for favors. But what leads to a patronage structure in the divine realm is two additional features. The first is that the divine insists on service and loyalty in exchange for favors. The second is that there is no clear break between relations between saints or gods and relations between persons.19
It thus not an absurd thought that certain religious traditions are in part responsible for the ubiquity of patronage relations across southern Europe (Hall 1974). But while there may be a cultural predisposition for patronage structures to arise in certain areas, they are too widespread to be understood as simply a specific Catholic, Orthodox, or Mediterranean cultural feature. Instead, this cultural predisposition may be better explained by structural arrangements discussed above: inequality in the absence of either strong states or relevant patrilineal structures. But assuming that we understand where patronage relationships arise, we do not yet understand how they work. To answer this, we must examine the content of the relationship.
The essence of the patronage relationship can be parsimoniously summarized as one that “imposes reciprocal obligations of a different kind on each of the parties” (Silverman 1965: 176; Eisenstadt and Roniger 1980: 49f; Bloch 1961 [1940]: 451; Bourne 1986: 5). Given the mutual contributions, it is tempting for theorists to understand the relationship as one of exchange; exchange, moreover, that (like all exchanges) is assumed to be equal by definition.20
This, however, is a serious misunderstanding. First of all, there is no reason to believe that the “exchange” (assuming it is useful to consider there to be one) is equal (see chapter 3). Indeed, as Landé (1977a: xxvii) has pointed out, frequently the premise of the exchange is that it is vastly unequal, with the client being unable to ever repay the great debt he owes the patron (see Powell 1970: 413; Wolf 1977: 174; Duby 1991: 38f; Campbell 1964: 230; Guasti 1977: 423; Englehart 2001: 22; Kettering 1986: 27; Mair 1977: 143; Scott and Kerkvliet 1977: 445 for other limitations to the idea of equal exchange). The client is generally dependent on the patron, and not vice versa (Barth 1977 [1965]: 218; see Fourquin 1976: 49 for the case of feudal relations between lords and lesser vassals). Further, in most cases the client is easily replaceable, so that even if the clients’ contributions in aggregate are substantial, no one client does much for the patron.
The precise nature of the obligations depends on the substantive context (see Lemarchand 1972: 73), but certain generalizations are easily made, such as that the patron’s main obligation is in the form of intercession or mitigation, and the client’s main obligation is a display of loyalty. In many cases, the services of the client are almost purely symbolic, such as holding the stirrup of the lord as he mounts his horse (Ganshof 1964: 82; Fourquin 1976: 121).21 Both the client and the patron are, paradigmatically, giving a form of support to the other. The patron usually gives material support or uses influence, while the client gives what little he or she can. In settled democracies or oligarchies, this may mean electoral support. (As Cicero said, “Men of slender means have only one way of earning favors from our order or of paying us back and that is by helping us and following us about in our campaigns for office” [Taylor 1968: 42; cf. Wallace-Hadrill 1989: 65].)
When conditions are more anarchic, and violence as opposed to votes determines elite dominance, the chief obligation of the patron is to protect the vassal, and the chief obligation of the client is to support the patron through armed service when fighting breaks out (Fourquin 1976: 125, 107, 122). During the classical age of feudalism, military service was undoubtedly the most important thing brought by the vassal to the relationship, though “counsel” was also often included (Ganshof 1964: 87; Farris 1995: 118, 282f, 297; and Critchley 1978: 35 for the case of Japan; Mair 1977: 141, 143 for the Ankole and Hutu-Tutsi systems; Sharma 1965: 99f, 199, 264, 271 for medieval India; Ikegami 1995: 81 for Japan). Such military service has been a fundamental structure linked to grants of land-use across the globe and throughout history (for examples, see Mair 1977: 98 on East Africa; Guasti 1977: 425 on colonial Peru; and Pośpiech and Tygielski 1981: 83 on early modern Poland).22 The most complete and structurally elegant form of military-based European feudalism came in the Norman conquest, which both imported a complete feudal system and highlighted military service at the same time. Those who conquered England were basically volunteers doing it “on speculation” and their pay could only be the lands they had helped conquer; these in turn would only be granted if the military relationship was solidified. Unlike the previous thegns who had other duties tied to their possession of land pertaining to community service (especially bridge and wall repair), the new lords cared little for their conquered underlings and hence emphasized military service (Bloch 1961 [1940]: 270, 383; Powicke 1996 [1962]: 17f, 26).
In most cases, even where military service is the paradigmatic form of client obligation, other services (in feudal England, known as “sergeanty” service, a distinction we shall explore in the next chapter) could also be exchanged for a benefice of land. Such service was usually minor or degrading or both: in return for the manor of Hemingstone, the tenant was required to “leap, whistle and fart for the king’s amusement” every Christmas day (Prestwich 1996: 67).23
Despite this emphasis on reciprocity (cf. Powell 1970: 412), there is no advantage gained by conceiving of the relationship as one of exchange, for the patronage relationship differs from an exchange relationship in two ways. First, the obligations of each party, while frequently negotiable in degree, are generally fixed in kind and the subject of relatively broad consensus (see, e.g. Critchley 1978: 103f for feudalism). Second, in contrast to other exchanges, the responsibilities are diffuse and protracted (Scott and Kerkvliet 1977: 443; Weissman 1987), and often involve face-to-face interaction.
In a nutshell, just as we do not say that a pie is an exchange of crust for filling, in relation of patronage, the patron does not “trade” protection for subservience; that is what patronage is. While an exchange can be decomposed into two transfers that are in principle separable (and hence forms of generalized exchange may arise), in patronage the two are uniquely fused. In the absence of one half, one either has aggression or toadying, and these are readily recognized as such. When the two action profiles become separable, in that either has a stable existence without the other, then indeed one may find some offering to trade toadying for diminished aggression, but this is seen in different terms from true patronage relationships—in particular, third parties are less likely to imagine that either party is particularly loyal to the other.24
Finally, if the patronage relationship was simply an equal exchange we would expect sets of relationships to take on structures we have seen arising for mutual ties (chapter 2) or reciprocated donations (chapter 3). But there are important structural implications to the difference between patronage relationships and exchange: most important, it is possible for a “rational” patron to commit to relationships in which the clients gain more from the interaction than he does. Indeed, there are good reasons why a patron would want to initiate or cement such a relationship that leads to a net loss. The accumulation of clients may be necessary for prestige (see, e.g., Kettering 1986: 28), or it may be a wise investment in the case of future situations in which favors will be needed, or it may be necessary to be a patron in order to take some other form of action which is restricted to patrons (e.g., a matrimonial alliance); finally, there is no need to deny in principle the possibility of altruism (for the example of Rome, see Taylor 1968: 42f).
Because the avenue of transfer from the clients (as a group) to the patrons (as a group) is not wholly contained within the patronage relationship itself, the patron can give more to the client than is received from these particular clients without the structure being destabilized (as he may get from other subalterns whom he does not treat as clients) (see, for example, Ikegami 1995: 178). We shall later see that the seemingly similar relationship of the “big man” and followers is not so robust in the face of unequal transfers.
In sum, the patronage relationship connects nonequals, and though each has obligations to support the other, it is not expected that these contributions be equal. Indeed, such an insistence would be to deny the patronage of the patron, which is the free gift of ameliorating the (already) disadvantaged position of the client. Now that we understand the nature of the relationship, what does this tell us about the larger structure that sets of relationships form? I go on to argue that these aspects of the relationship answer the question raised above, namely why patronage relationships form trees instead of trisets.
In chapter 2, we first examined the puzzles that confront us when we consider the aggregation of existing structural units. There is frequently more than one way in which units may be joined and which way is chosen may—as Simmel would appreciate—change the content of the relationship in question. In the case of cliques as examined in chapter 2, it turned out to be extremely difficult for cliques to join without such change. What are the options for patronage structures? Is it possible for units to be aggregated in such a way that the essential content of the relationships is preserved?
It might first of all be expected that just as we saw “wheels” aggregating into multihub communication structures in chapter 5, we shall see the same structures arise for the homologous patronage triangles. But there we examined a symmetric relationship of communication; patronage, however, is an inherently antisymmetric relationship because it is based on a preexisting inequality. Hence if relationships were to join patrons, either they would not be patronage relationships (taking us out of the realm of simple structures), or the resulting aggregate would lack the horizontality of the multihub structure. As we shall see, it is the latter that seems to be a recurring structural feature of patronage relationships—they concatenate into large “pyramids” in which any two patrons are connected only by a set of vertical ties, just as any two persons in the unilineal family tree can be so connected.
In our exploration of the formation of patronage structures, we may take for our example feudalism—the relationship in which grants of benefices are tied to vows of fealty. In most cases feudal systems involve the granting of use-rights (in land or in the form of tax farming or both) to those who promise military service; such systems have arisen independently in many societies (see a review in Critchley1978: especially 23, 28; Beeler 1971: 154, 169f; see Fourquin 1976: 134 for examples of nonland based benefices), not only medieval Europe and Japan, but ancient Greece, Babylonia, and the Maya (Adams 1991: 192f). (While specialists frequently like to emphasize the distinctiveness of any particular system, for our purposes, the commonalities are far more interesting than the differences.)
We, however, are not interested in such feudalism as a governmental or military scheme, or even a way of organizing property rights, but in terms of interaction. Only to the extent that it implies a patron-client relationship—that is, a diffuse relationship of reciprocated support between lord and vassal, not a specific exchange of rights—will feudalism be of concern to us; for this reason, medieval Europe is probably the most central example. Japan, which is sometimes seen as more feudal than Europe, in that lordship was unambiguously related to military command (Farris 1995: 372, 374f; Critchley 1978: 64, 30), is in our sense a less paradigmatic example of patron-client relations for that very reason.25
Pursuing the example of feudal relationships, then, we hope to investigate the structural properties of patronage relationships by examining the larger structures that are formed when patron-client triangles are aggregated. Doing so reveals two principles of the aggregation of patronage relationships: the suppression of horizontal relationships, and a tendency toward mediation. We investigate each in turn.
The general question, then, is how a set of triangles may be joined: given two triangles, one with apex (patron) A and clients A1, A2, . . . etc., and a second with apex (patron) B and clients B1, B2, . . . , how is one unified structure formed? Is a horizontal relation between A and B created, leading to multihub structures we examined above? In classic patronage systems, the answer is no: the connection occurs by the addition of vertical relationships. Either A can become a client of B (or vice versa), or we may introduce some C who will be a patron’s patron to both A and B—in any of these cases, the triangles have fused into a multilevel “pyramid” (see Lemarchand 1972: 76).
This form of concatenation, then, logically implies the suppression of horizontal relations, and it is gratifying to the analyst that patrons frequently agree and attempt to prevent such relations from arising (at least between their clients), even though this may decrease the communicative efficiency of the structure. For extraordinarily clear-cut example, when telephones were introduced to Swat, the ruler (Aladdin) had all connections pass through his own capital, so that there could be no horizontal communications between provincial lords without his knowledge (Critchley 1978: 88). While self-interest often leads empires that have conquered other states to allow horizontal trading relations between their client states, they generally attempt to suppress other political relations and will also suppress trade relations if these are believed to be political conduits (see, e.g., Hassig 1992: 91f; Randall 1993: 185). Such self-interest, however, can become a general matter of principle: Bloch (1961 [1940]: 354ff, 445; also Anderson 1974: 194; see Duby 1991: xv on other horizontal challenges to medieval verticalism) remarks that the reason the new political form of the commune (the association of burgers) was seen by feudal contemporaries with such loathing was that as a group tied by horizontal bonds of equality, it made hash of a social system predicated on relationships only connecting superiors to inferiors. In other cases, this suppression seems a matter of course: students of traditional Thai society remark that it was taken for granted that no two people confronted each other as equals (Kemp 1984: 59; Englehart 2001: 20).26
This thought experiment does not mean that all patronage structures arise because of the concatenation of preexisting triangles. However, it is important that there be a sociologically consistent answer to the question of how two small structures could be joined if larger structures are to be coherent. In this case, we are able to see how an extension of the principles underlying the smaller structures naturally leads to an aggregate that has a formal homology to the unilineal kinship structures seen in the previous section. As we will see in the following chapter, this solution was in many empirical cases (for example, European feudalism) always an incompletely realized tendency overwritten with other imperatives, yet these other imperatives (for example, horizontal ties or inconsistent rankings) were understood by the participants as troubling—comprehending and navigating the structure leads to an appeal to subjective heuristics compatible with tree structures.
But what does the treelike nature of the resulting structure imply about interaction between clients and “second-order” patrons? We saw in the previous chapter that there has been a tendency in the mathematical analysis of informal social structures to see forms such as patronage pyramids as trees in the algebraic sense of semilattices where all upper bounds are comparable (Friedell 1967); semilattices are a particular type of partial order, and one of the characteristics of a relation in a partial order is that it is transitive: denoting the relation →, transitivity, we recall, means if A → B and B → C, then A → C.
Such transitivity seems natural for patronage structures given their foundation in some system of inequality. That is, patronage relations are antisymmetric (if A → B, then B not → A) because they correspond to some ranking of persons on some attribute (if A → B, there must be some ζ such that ζA>ζB). By extrapolation, this might be understood to imply transitivity, since if A → B and B → C, there must be some ζ such that ζA > ζB > ζC, and hence we might expect that A → C. But this is rarely the case. Instead, the opposite (antitransitivity) is generally true: if A → B and B → C, the one thing we know is that A not → C.
This rule was explicitly formulated for the case of feudalism in fourteenth-century century France: “queritur utrum homo hominis mei sit meus homo. Et dicendum est quod non”: my vassal’s vassal is not my vassal (Ganshof 1964: 97; also see Duby 1991: 28; for Japan, see Ikegami 1995: 82, 235, 237).27 The tendency toward antitransitivity was made clear by Gould (1996) in his analysis of the patronage relations underlying the American Whiskey Rebellion, and he stressed that the reason for this was that it makes no sense for a patron to support a client who can already be supported by one of his other clients. This principle of mediation helps us understand the tendency of patronage structures to form pyramids. If (a) persons are ranked, and (b) very high-rank people do not interact with very low-rank people, and (c) patronage is a fundamentally personalistic relationship with more clients than patrons, the patronage structure must form a multileveled pyramid (Landé 1977b: 508).
The antitransitivity of patron-client ties may lead us to assimilate these relationships to the general case of brokerage. This relationship, discussed by Simmel (1950 [1908]:154ff) as an important case of tertius gaudens, the third who profits by standing in between two others, has inspired much network theory (especially Burt 1992; Gould and Fernandez 1989) because of its richness in explaining organizational power in managerial hierarchies and exchange networks. Furthermore, patronage relations studied by political scientists often involve a great deal of brokerage: most important, the patron brokers relationships between his clients and more cosmopolitan powers such as regional or national governments. In such a setting, the patron is the one who can tell a client how to navigate these unfamiliar waters, who can gather clients together to make a forceful demand of a central power (and take a cut from any resulting grant), or who can smooth the client’s path with a word to someone in power. The assimilation of patronage relations to brokerage relations is doubly plausible because this would explain the tendency toward antitransitivity as a result of broker-entrepreneurs who insert themselves between two previously unconnected parties.
Despite all this, there is reason to suspect that brokerage is not fundamental to the case of mediated patronage triangles. Most important, an emphasis on brokering (or occupying “structural holes”) assumes the existence of a desire among two unconnected echelons to interact. But this very desire is foreign to the nature of mediated hierarchy. In the next chapter, we shall explore what happens when this desire arises (usually by virtue of military need), but it should not be assumed as given.
It is true that there are many examples of patrons brokering between local communities and the state, helping locals navigate bureaucratic intricacies or smoothing their way over hurdles set in place by unsympathetic officers. But such a relationship necessarily preassumes the existence of such a bureaucratic state. Patronage relations, however, are likely to have arisen before this type of state.
As a result, it is easy to reach mistaken conclusions regarding the intrinsic structural tendencies of patronage relationships on the basis of an examination of the roles of brokers. Prior to the massive state penetration of the past two centuries, such functions were not crucial to patronage systems (this point has also been made by Lemarchand and Legg 1972: 154, 158; Parish and Michelson 1996: 1045).28 This emphasis on brokerage, where it has occurred, involves a modification of traditional dyadic patronage relationships to emphasize mediation and brokerage as opposed to reciprocity between unequals (see Kaufman 1977: 114, 118; also Dietz 1977: 441; and Mouzelis 1985: 343 on Latin America). For one, the new ability of patrons to serve also as brokers may have increased their power vis-à-vis their clients (see Scott and Kerkvliet 1977: 453f).29 More important from a structural perspective, a switch to an emphasis on brokerage instead of amelioration leads patronage relationships to become more like the “big man” relationships we shall explore below, and hence to concatenate in multihub structures, in that there are more horizontal connections between big men than in a patronage system, in which all relations are ideally vertical. A classic example here is the “amigocracy” described by Kenny (1960) in Spain—the big man or patron is one with a horizontal connection to a “friend” in the right place. These patron-patron ties, explicitly noted by Kenny (1960: 23; cf. Guasti 1977: 423; Schwartz 1990: 153), are foreign to the purest form of a patronage pyramid, if only because the purer forms are based on land-ownership, and there is little that a peasant needs from a different landowner that he cannot get from his own.30
To illustrate, we can take the rather well studied case of Italy. Because Italy was formed by consolidation of previously separate regions, there were quite understandably many “structural holes”—gaps in the structure of relationships making it difficult for some persons to reach others—that would frustrate the ability of locals to navigate the nebulous national power structure. As a result, “the most valuable patron was neither the wealthiest nor the most generous, but the one with the best connections.” In consequence of the focus on “connections,” the patronage structure found in central Italy tended to be more like a multihub structure than a pyramid—patrons were distinctive in having ties to other patrons outside the community. But Silverman (1965: 179, 180, 182) comments, “Yet this aspect of the patron’s role was elaborated only in the late nineteenth and early twentieth centuries.”31
Brokerage, then, is not intrinsic to patronage (also see Mayer 1966: 113f for a clear distinction between the two). The introduction of cosmopolitan state structures has in many cases led local patrons—in some cases, the local government as a whole—to function as brokers between the community and the national or regional government (e.g., Mair 1977: 228). In such cases, the seeming “broker” may really be simply a first-order patron only having ties to second-order patrons (that is, the broker lacks horizontal ties to brokers of equivalent status).32 Where the introduction of these cosmopolitan political structures allows patrons to become true brokers (who have ties to other brokers, or to other locals independent of ties to second-order patrons), they tend to produce different structures associated with the “big men” discussed above, or with the more exploitative “boss” relation (someone with all the rights and few of the duties of a patron) (see, e.g., Graziano 1977: 363).
We have seen that patronage relationships tend to take on a particular treelike structure. That informal relationships should spontaneously take on such a restricted formal structure seemed quite surprising to some analysts (e.g., Friedell 1967). But it can easily be shown that this structure will occur whenever people follow two simple heuristics.
We have already seen the first: it has to do with the preexisting inequality that is the basis for the relationship. Since the relationship of patronage arises to ameliorate this inequality it stands to reason that A will only take as a patron some B with more of whatever it is (previously denoted ζ, with the amount of this resource possessed by A denoted ζA) than A has. As stressed in the previous chapter, if we assume that ζ is continuously distributed with no perfect ties, a linear order will emerge if people follow the simple rule B → A if and only if ζB > ζA. That is, there is an antisymmetric relationship between A and B if and only if there is a relation of “greater than” between the two.
Assuming no ties, such a structure would lead to N(N – 1)/2 directed relations, since a relationship of inequality will exist for every dyad, whereas a tree structure will involve only N – 1 relations, as every person except the apex is a client of someone, and these are the only relations present. With the exception of a single long line, a tree structure will also include horizontal, as well as vertical, differentiation. We saw in chapter 5 that a horizontal structure will arise if there is some minimum “just noticeable difference” (JND) Δ between any two people before a relationship will be established. (That is, they follow the rule A → B if and only if ζA > ζB + Δ.) But the “JND” rule does not produce mediated trees; instead, as we have seen, it produces structures where those higher up have relationships with all those at the bottom, and we have not found this to be characteristic of patronage structures.
In other words, while some difference in resource is necessary for the establishment of a patronage relationship, it is not sufficient: that is, the rule must really be closer to “A → B only if ζ(A) > ζ(B) + Δ.” But even more important, there seems to be a rule (for the classic structures we are explicating) to the effect that “every client can have only one patron” (Wolf 1977: 174). Such a rule of exclusivity is not found, as Landé (1977a: xx) points out, in structures organized around “stars” or “hubs,” and hence patronage structures are qualitatively different from networks that simply have high variance of degree (number of ties per node). There are exceptions to the “one patron rule” which I shall discuss below, but the rule itself comes out of a fundamental aspect of the relation in question, in which the client gives loyalty above all else. While it is possible to promise loyalty to more than one patron, it seems likely that such loyalty is a bit different from the loyalty promised to one and only one.
This “one patron rule” is reinforced when patronage is tied to ownership of or control over land: tenancy can be the basis for an inherently exclusive relationship of clientship where tenure is defined so that only one person can “own” the land occupied by the client, and where clients only occupy and work a single plot (cf. Hall 1974; also Duby 1991: 139). In such cases, while patronage relations may be an important part of the societal fabric in general, it is possible for these particular relationships to be relatively disconnected from one another: each landlord is patron to his tenants, and is not necessarily the client of anyone else.33 Landlords need not compete with one another for clients.
Things are not necessarily this simple, of course: in some cases, there may be multiple forms of ownership of land.34 In other cases, even though land ownership is the basis for patronage relationships, tenants may supplement the patronage received from the landlord by adopting other patrons and may even play these against the landlord (Barth 1977 [1965]: 215). And there are resources other than land that may be stable bases for patronage relations (e.g. Guasti 1977: 429); what is essential is that they can be wholly monopolized, so that there is no chance for enterprising ex-clients to strike out on their own and few opportunities for advancement in a wholly different line of endeavor (Hall 1974).35
In sum, in patronage pyramids, there is a strong tendency for any person to be the client of one and only one patron. But it is not the case that the patron can have only one client; indeed, the opposite is generally true. Since what the client has to offer is of much less value than what the patron has, a number of clients must be assembled before their support is of much use.
These simple rules (accept as a patron only someone of higher ζ than you; have only one patron) lead to the production of the treelike structures that are so familiar to comparative political scientists. The nature of the relationship, then, has clear structural implications that have been noted by analysts in settings widely varying in location and time. These structural implications are clarified by a comparison to those involving somewhat different relationships, such as “big men” and their retinues.
The term big man comes from anthropological investigations of Pacific Island societies in which those of great prestige have neither power over others (as would kings with a permanent staff) nor monopolize any resources (Sahlins 1968: 22, 26f; Hanser 1985: 61, 104, 331). Instead, these “big men” are big precisely because they give away what they have.36 Their prestige can then be used for purposes of garnering resources (others will court the big man’s favor through gifts so as to have a prestigious ally; the big man can then redistribute these gifts to increase his prestige). “Little men” (if they can so be called) tend to attach themselves to a particular big man, forming the wheel structure we previously investigated. Thus the “big man-little man” relationship is like a patronage relationship in that it connects unequals through mutual support, but, as we shall see, it tends to concatenate into different structures.
What is most important about the classic big-man structure in terms of its structural potential is that it does not build on preexisting inequality, but is itself the generation of that inequality. While patrons may be able to choke off the access of others to resources (see Wallace-Hadrill 1989: 73 for the case of Rome), the big man has no special access (though some are born with “better” kin than others). A big man is one who has many followers who will make donations to him; these donations in turn allow him to provide feasts for followers and demonstrate that he is in fact a big man. The bigness of a big man, at least in its purest form, comes wholly from the allegiance of his followers.
Since big men do not arise on the basis of preexisting inequality, they require a jump-start,37 and given the conditional allegiance of their followers, generally face constant competition from other big men for clients (Sahlins 1963: 291; cf. Kenny 1960: 22). The nature of the jump start is various, but fortunately for the analyst, the beginning is not as important as what happens next: the big man becomes the hub of a wheel supported by (potentially unequal) transfers. Indeed, we may derive the hub structure from what we have learned about the interaction-profiles associated with exchange. If the relationship of exchange is to be the basis for obdurate social structure, it must involve not only reciprocity but delay in repayment (otherwise there is no unifying function of the debt) (see Silverman 1965: 176 for the case of patronage relations in Italy; Bourdieu 1977 for a general statement). Given this, it is possible for relational entrepreneurs to quickly amass social capital simply by borrowing from some and giving to others. The “big man” system is basically redistributive, and someone becomes a hub merely by the sleight of hand that makes it appear that the bread you eat today is not the bread you gave to the big man yesterday, but that it is in fact “his” bread (see Keesing 1983: 7; Meggitt 1974: 190, 195). As a consequence, such hub structures are riddled with paradoxes or circularities that can turn vicious.38 Just as we saw in chapter 2 popularity tournaments containing positive feedback that could spin out of control, so big man systems contain an aspect of mimetic speculation (Orléan 1988) whereby all wish to be clients of the important patrons—that is, those with many clients (compare Wallace-Hadrill 1989: 83; we explored such “go with the winner” games in chapter 2 in the form of the popularity tournament).
In particular, “big man” systems are very sensitive to dislocation of the delicate balance between spoke-hub and hub-spoke transfers. Since spokes in such systems are generally free to drop and pick up hubs on the basis of their power and largesse (such as among the Melanesians discussed by Sahlins 1963: 292; or the Mandari of Eastern Africa discussed by Mair 1977: 98; cf. 64), hubs may not be free to desist from ruinous relationships (see Park 1974: 271 for the case of the Anuak headman). What is crucial is not that where the clients have more choice they have a better bargain (though this is in fact the case): it is that where the redistribution is wholly contained within the patronage structure (as opposed to extraction being outside the structure), any intensification of the redistributive schedule can destroy a big man. A wonderful example is given by Barth (1977 [1965]: 218f; 1981: 68) concerning the Swat Pathans. When a patron begins to lose influence and becomes poorer, he must increase his hospitality, even if this means liquidating his capital. One informant made it clear that his relationship with his clients was not an equal exchange but still the lesser of the evils confronting him. While entertaining guests required selling off land, “on the other hand, if people stop sitting in my men’s house, I shall lose the land even faster; only this constant show of force keeps the vultures at bay.”
More generally, as a “big man” gets bigger and has more obligations, there is more pressure on him to squeeze more resources from his entourage (most simply, by increasing the time between when he gets something from them and when he gives it back, but also by increasing his cut from the redistributive process) or even better, if he is able, to squeeze resources from nonfollowers (see Mair 1977: 136f). As he becomes more powerful, this seems a plausible tactic. But if clients then become dissatisfied and leave, his very power is his weaknesses. As Sahlins (1963: 293) says, “Paradoxically[,] the ultimate defense of the [big man’s] position is some slackening of his drive to enlarge the funds of power. The alternative is much worse.”
As a result, hubs may indeed be sociometrically privileged without being materially privileged. The apexes of the Senegalese “clans” that dominated postindependence politics were required to share their profits with their followers to the extent that they frequently were no wealthier than their retinue (Foltz 1977: 245). Similarly, Bashi (1997) finds that hubs who help West Indian immigrants relocate to the United States gain little besides prestige from their efforts. The requirement to redistribute collected surpluses may be the most important factor leading to the preservation of horizontality in hub-spoke systems that follow big-men lines. The big man system generates and destroys inequality at the same time.
True patronage relationships, however, can exist in a stable vertical form as long as there is an exogenous generator or regenerator of inequality. At the same time, patronage relations lack the same endogeneity of the popularity tournament that the big man system possesses. That is, there is no particular limitation to the number of clients that a big man can get—indeed, each additional client only increases the probability that another person will attach himself to the big man. But a patron may only have so much land to give out (cf. Major 1964: 637). Thus despite their formal homology (a central person with clients), big man and patronage relationships produce different larger structures, as we go on to find.
In general, big man structures concatenate into the multihub structure charted above, as opposed to the pyramids of patronage relationships. Put another way, while some big men are bigger than others, there are no “big men’s big men” who have entourages consisting only of (smaller) big men. A perfect opposition between big men and their entourages (on the one hand) and patrons and their clients (on the others) comes from Sahlins’s classic comparison of the multilayered pyramids of Polynesian patronage structures to smaller, unstable, and segmental entourages of Melanesian big men (Sahlins 1963: 287, 292ff).
There are, of course, intermediate cases. In particular, we found that the introduction of cosmopolitan state structures to localities previously organized along patronage lines leads the patrons to assume a new role of brokers or gatekeepers (see Kenny 1960) who can put a client in touch with other patrons. As a result, the relationship tends to be more of one of “communication” than one of amelioration, and consequently there is a tendency toward horizontal connections between patrons and an approximation of the multihub structures that arise in communication networks. Similar are cases where patronage degenerates into mere “sponsorship,” a condition associated with relaxations of the “one-patron” rule examined below.39
But in such cases, we may find that the relationship itself has changed and may be better described as a “boss” relationship. “Boss” (or cacique in Latin America) is a term frequently used in political anthropology to denote someone who has the power of a patron but without the legitimacy (see Scott and Kerkvliet 1977: 443 on differences between patron and boss). Like the big man, the “boss” relation does not build on preexisting inequality; rather, the “boss” is an upstart who creates the inequality of the relationship. In contrast to big men, who have little power except the respect they garner, the boss is generally considered to be important but self-oriented and hence has few followers who have more than a pragmatic loyalty (see, e.g., Cornelius 1977, especially 339, 341).
Further, bosses tend to be brokers: the cacique’s real or pretended ties to central elites are an important reason for his getting support. (Thus in the Caribbean plantations on the coast of South America, Amerindian caciques stood in between plantation owners and African slaves [Whitehead 1992: 141].) More generally, the “boss” may be said to differ from a patron in that there are no clear obligations that the boss has toward his clients, and thus although he may favor them with a great deal of what is commonly considered “patronage,” these seem spontaneous acts of grace.40
Each of the three relationships has something in common with the other two; these commonalities and differences can again be clarified by means of a diagram (figure 6.6). We will find a similarity in the nature of the trade-offs between these dimensions and trade-offs we have investigated in the previous chapters. The first dimension has to do with whether there is a preexisting ranking of persons on some dimension that is ameliorated by the relationship. This is true for patronage structures, but not the other two where the relationship in question must generate the differentiation between the persons. Without payoffs, the boss ceases to be a boss; without esteem, the big man is no longer big, but the noble without fealty is still a landowner.41
The second opposition has to do with the nature of transfers made in the relationship. Both the patronage structure and the big man structure are fundamentally redistributive (though not necessarily to the long-term disadvantage of the superior party): the patron gives some of what he naturally gathers as a superior person back to the lowly (though they may pay a high “price” for this through military service); the big man receives gifts and gives them back as “his” feast. In contrast, the boss takes, and takes more than he receives. There is generally a direct relation between this dimension of redistributivity and perceived legitimacy by clients.42
The third opposition, which distinguishes the big man from the other two, is that the responsibilities of the small man versus the big man are not enforceable via sanction, as are those of the boss and patronage relationships. Perhaps because of this absence of enforcement, more emphasis is placed on impressing upon followers the charisma of the big man, or the accumulation of “symbolic capital,” as Bourdieu (1977) calls it. What leads to charisma of the big man, as Zablocki (1980) has argued, is largely nothing other than the fact that he is seen as charismatic by others.43
Figure 6.6. Three polar relationships
All empirical examples fall somewhere on this space; they are mixtures, but they are not random mixtures. Thus many bosses do some redistribution, which pushes them out of the “boss” corner a bit; many patrons do some exploitation and even feudal kings required some personal charisma to maintain their followers (Reuter 1999: 27). Lords often followed the paradoxical recipe of the “big man” that he who has the most is he who gives all away. An eleventh- century Norman lord, Richard of Aversa, was said by a contemporary to have “carried off everything he could and gave it away, keeping little . . . in this way the land about was plundered and the number of his knights multiplied” (Gillingham 1999: 64; cf. Adams 2005: 146f; Englehart 2001: 25).
Thus the reasonableness of the foregoing scheme depends not on the purity of empirical cases, but on there being an empirical trade-off between movement on any two dimensions and the third: as with our previous conclusions, we find that there are limitations to the number of principles that can simultaneously structure relationships. Taking the first opposition pertaining to preexistence of inequality, it does seem quite reasonable that to the extent that there is a preexisting ranking of inequality, it is both easier for the structure to be enforceable with regard to outside sanction and more of a need for redistribution.
Taking the second opposition pertaining to the degree of exploitation, it seems reasonable that exploitative relationships will be more likely to require the support of external sanctions. It also seems reasonable that it is easier for unequal relationships to avoid exploitation if they build on a preexisting inequality. That is, there is no need for this particular relationship to generate inequality. That does not mean that it is impossible for a patron-client relationship to be exploitative, but it does mean that, in contrast to a boss relationship, it need not be.
Taking the third opposition, it again seems reasonable that voluntary relationships are more likely to be redistributive and to spontaneously generate the inequality involved as opposed to building on them, for reasons already alluded to. Hence we conclude that this scheme has virtues that are not obvious yet not obscure. The next step is to draw similar conclusions regarding structure.
We have seen that, building on preexisting inequalities, patronage relationships are more stable (less conditional) than big-man relationships. Further, it is easier for them to aggregate into a multitiered pyramid. However, if there is not a single apex to this pyramid, it is still possible for patron-patron conflict to spill over into arenas that require the mobilization of clients. If clients can abandon weak patrons and go to stronger (as among that Pathans discussed above [see Barth 1977 [1965]: 214]), clients become relatively empowered and the system tends to veer off in the direction of a big-man structure. In contrast, “boss” structures may approach simple pyramids due to the tendency of the boss to be an intermediary.
As a result, the type of relationship in question tells us a great deal about the likely structural patterns formed in concatenation of dyadic relationships, but the specifics of the relationship—the degree of underlying inequality, the presence or absence of conflict between higher-ups, etc.—also tell us a great deal about the structural pattern. In fact, we may derive a similar (though not homologous) space of three related structures, namely the web, the pyramid, and the multihub structure, with three similarly related dimensions (see figure 6.7). It is important to note that the corners on the two triangles do not match exactly: while patronage does tend to be associated with pyramids, big manism may form either webs or multihub structures—indeed, in some cases, no concatenation happens at all, since there seems to be a general disinclination on the part of big-man wheels to aggregate to form large structures. Yet the nature of the oppositions and trade-offs between these structures is similar to those we saw for the relationships.
A multihub system, we emphasized before, differs from a pyramid mainly in only having two levels—there are no “hubs’ hubs.” Webs of antisymmetricties, understood as more general structures, also lack the clear gradation of a pyramid: while there may be a continuous variation in quantity of attachments of any node, this undermines any attempt to make a distinction of kind. But in contrast to multihub structures (and patronage structures), webs lack the exclusive connection of any spoke to a hub. Finally, both a patronage system that builds on preexisting inequality and a web that lacks exclusivity defuse the potentially ruinous nature of competition between hubs for spokes.
Figure 6.7. Three polar structures
The difference between the two analytic spaces allows us to understand how two patronage systems may have different structural shapes. For example, all other things being equal, a patronage structure that is based on land will be more likely to fall toward the top of the figure, while one based on less exclusive and more competitive principles may tend more toward a web or multihub structure. Further, we can see that either a multihub or web structure may be pushed toward a pyramid by enforcing exclusivity of patronage and weakening the ability of clients to force patrons to compete. And indeed, this is exactly what we see in cases of feudal organization: in Thailand, just as in feudal Europe, it was declared that all had to have only one patron, and while the lowest ranking clients (phrai) did frequently desert one patron (nai) for a better one, and hence spur the patrons to compete with each other, this was outlawed (Englehart 2001: 43; cf. Ikegami 1995: 130, 334).
But why would anyone wish to do so—to push the structure further toward a pyramid? As we shall see, patronage structures are special in that they can bring a number of persons together in a single structure under a wide range of conditions. We have, we recall, been searching for simple structures that can form the building blocks of larger structures. We have found some structures cannot be concatenated at all—for example, two exclusive dyadic relationships must involve four participants, not three. Other structures make unrealistic interactional demands when aggregated—participants must establish huge numbers of consistent relationships (e.g., the clique, the pecking order). Still other structures may be able to aggregate, but these aggregates may only survive in a relatively narrow range of environments. Thus generalized exchange requires equality among persons or units but tends to generate inequality and hence undermine itself. Like an environmentally sensitive organism that pollutes its surroundings, it necessarily leads a precarious existence.
But with patronage structures, inequality in the environment poses no problem, since it is the fundamental premise of association in the first place. And the antitransitivity means that two structures can be joined at a single place, without examining whether all the implied dyadic relations work out. That is, if two groups with N1 and N2 members each, each group organized into a pecking order, decide to aggregate, then even if the bottom person in the first group successfully dominates the top person in the second, we still have N1N2 – 1 relationships that may possibly go awry and force a reorganization. But if a patron takes on another patron as a client, the “sub” patron’s clients need not acknowledge the sovereignty of their patron’s patron (compare Simon 1962: 476). Accordingly, when there is widespread social disorganization, patronage structures are often the first social organization to move in; the formation of a richer institutional life often depends on the transformation of these patronage structures, or, seen another way, their breakdown. We go on to examine the conditions of such breakdown.
Concatenation of distinct and nontransitive patron-client relations can lead to a sprawling structure which, if weak in many respects, is still able to move support and sometimes money upward and (some) control and protection downward; consequently, patronage pyramids are relatively frequent governance structures that can link the humble local peasant to the central power. Once established, they are surprisingly resilient, for reasons to be explored below. For an elegant example of governance through a patronage structure, we may take the Thai political system of the late Ayudhya period until 1873 (Landé 1973: 111).
At the bottom were phrai (commoners) who belonged to some nai (noble)—while their relationship was personalistic as opposed to being fixed by law or contract, the nai was basically in charge of mobilizing the labor of his phrai for higher ups. An edict of 1356 required that all people were required to register under some patron who in turn would be treated as an official: for quite some time, all phrai had to have the name of their nai tattooed on their body (Terwiel 1984; Englehart 2001: 23, 36–39, 42).44
These nai were then clients of higher patrons, “etcetera etcetera,” progressing steadily upward toward the king. The state was thus the aggregation of these triangles, with almost all governance indirect through clients of clients (Englehart 2001: 55).45 Most important, this joining was based on the principle of mediation, or intransitivity. As Hanks (1977 [1966]: 164) says regarding the Thai structure, “Ordinarily a lowly person must reach the high through one or more intermediaries, and then it would seem ridiculous that anything a woodcutter might need could not be handled as well by a less elevated personage. So the extremes rarely meet, and more effective connections lie a few short steps up or down the hierarchy.”
The story of European state formation is, in broadest outlines, the story of the creation and transformation of such patronage pyramids.46 In ideal-typical form, we may state that the first step is the evolution of hub structures into patronage triangles, as when charismatic warrior bands are the basis of well-developed feudalism. This has been suggested as the origin of European feudalism: two institutions were combined, the first the personal retinue of the war leader’s band, and then second the Roman “precarium,” a form of share-cropping that had become tied to military service (in place of rent) in the fifth and sixth centuries (Beeler 1971: 4; Critchley 1978: 102; Mallett 1999: 211; cf. Anderson 1974: 94, 98, 115, 130). Then the patronage structure is transformed into a command structure. We shall examine this aspect of the creation of governance structures in the next chapter.
The use of patronage pyramids as governance structures is not limited to the political realm, but can easily be found in economic governance. In postwar Japan, many industries were organized according to patron-client principles (oyabunkobun, as said above), an interesting contrast to the accounts of the development of multidivision firms we will examine in the next chapter. Lower level labor bosses controlled workers in a fashion similar to contractors (they were paid and paid their workers in turn); in turn, these oyabun were the kobun of higher oyabun, etc. (Nakane 1970).
In sum, it is now possible to specify why patronage structures are so frequently seen. They are one of the few simple structures that thrive in conditions of inequality and which can be concatenated without necessarily undermining their structural cohesion. That does not mean that there are not structural tensions. I go on to explore these as a way to understanding the reasons why patronage structures may transform into other forms.
Patronage structures tend to be extremely stable. Just as we saw many higher primates able to “fold in” to a pecking order all relationships (e.g., alliance) that might be understood as undermining this order, so people are able to fold in potentially contentious relations into a patronage order. Many of the tendencies toward change that patronage structures do have are ones that actually strengthen the structural tendencies. Most important, there is a tendency toward the “spillover” to new forms of service (that is, patrons take on a new form of patronage and/or clients a new form of clientage [see Lemarchand 1972: 76]).
Political scientists have repeatedly noted—often to their dismay—the resilience of patronage structures which often only release their hold over interactions when an expanded (national or imperial) state deliberately weeds them out. Even in such cases, states frequently prefer to exploit and hence reinforce these existing relations (Lemarchand 1972: 79). Yet any simple opposition between stable patronage structures and the states that destroy them is misleading, for, as I have suggested, the structures associated with the modern state have in large part derived from patronage structures.47 This change was predicated on the few structural instabilities inherent in patronage structures.
Patronage structures, we recall, may be understood as resulting from three rules, the first of which stems from the presence of inequality (form a tie from A to B only if there is some ζ such that ζA > ζB + Δ where Δ indicates a just noticeable difference). The unsaid correlate to this is that all relationships are vertical, and no relationships exist for any A and B where |ζA − ζB| < Δ, which we might consider to be a “horizontal” relationship. The second rule was that any client could only have one patron. The third rule was antitransitivity. Abrogation of any one of these principles weakens the patronage structure, and we consider each in turn. In every case, the patronage structure, durable as it is, does tend to arouse the very forces that would destabilize it.
The patronage pyramid is premised on the inability of clients, especially bottom-level clients, to form relationships among themselves that would allow them to bargain collectively with the patron. Yet the patronage triangle establishes a relation between clients of structural equivalence that can serve as the basis for a relationship of active collusion that may undermine the patronage structure. It may appear that the clients need only perceive that their latent categorical identity can be the basis of actual relationships of solidarity—to go from being a category in themselves to a category for themselves—for the structure to collapse.
Interestingly, this scenario seems to occur only quite rarely, and not only because of the general powerlessness of clients (and the lack of accountability of patrons). It is that it is not enough for clients to collude—to declare their wish to bargain as a corporate group with the patron—they must somehow organize and put forward a representative to bargain with the patron. But this representative of the clients now has a qualitatively different position from the others, and indeed has an antisymmetric relationship with them (they pass up grievances, he passes down information). Accordingly, their relationship can easily slide into one of patronage (Powell 1970: 424; cf. Stokes 1995: 70, 112), and so the challenge to the system is easily absorbed simply by incorporating one new patron. Patronage structures, then, tend not to break down due to the destruction of the position of the patron by uprising clients. Instead, breakdown is more likely occur from what seems to be an enthusiastic endorsement of the principles of patronage than it is from a spirited rejection. If it is so good to have a patron, why not have more than one?
There are a fair number of cases in which clients attempt to better their position by having more than one patron, though these tend to be cases in which the relationship is “weak” in that the obligations on the parties are neither onerous nor inescapable.48 Wolf (1977: 175; cf. Kettering 1986: 75) suggests that such weak patronage is likely to occur where the society’s institutional structure is “far flung and solidly entrenched”—where no coherent patronage structures can exist independent of the formal structure of the society, but neither can that formal structure be totally aligned with the structure of patronage (as can unilineal descent groups, as we recall). In such cases, patronage degenerates to “sponsorship,” for what the patron provides is access to important nodes in the formal structure of the society.49
A good example of this is the Roman republic: the structure of patronage relationships was of vital importance, but there were too many contrasting and overlapping arenas for it to be rationalized. Accordingly, it was possible for clients to have more than one patron, though the ideal was exclusivity (Saller 1989: 53; Taylor 1968: 42; this was also on occasion true in Greece; see Kagan 1987: 299;50 also see the sophisticated discussion of Johnson and Dandeker 1989: 231). This form of patronage as sponsorship was also crucial to the Italian city-states from the Renaissance to the early modern period (on renaissance Florence see McLean 1998). Frequently the “patronage” that is spoken of by political scientists, although an umbrella term for almost all deviations from the universalistic prescriptions of democratic theory, refers to cases of such sponsorship when it does not simply indicate a spoils system.
Such examples of sponsorship may be analytically considered a “degeneration” of the pure form of patronage, but this is not necessarily historically true.51 However, a real breakdown of the one patron rule is seen in our chief example of European feudalism. When feudalism first developed in Europe, it was clear that no vassal could serve more than one lord—“a thing which is not pleasing to God,” in the words of one twelfth-century writer. Yet by the turn of the tenth century, at least in the west of Francia, this had already become possible (Bloch 1961 [1940]: 211; Ganshof 1964: 31, 49; Fourquin 1976: 128),52 and by the twelfth century was common throughout Europe. This was inherently problematic, as any tendency toward factionalization could leave the vassal with divided loyalties. As Fourquin (1976: 128) says, “To be the vassal of several was really to be the vassal of none.”
Some considered resolutions were to simply prohibit a fief-holder from inheriting an additional fief, to favor the patron from whom one received the largest benefice or to have all relationships save the first explicitly undertaken with a proviso that they could not upset earlier pledges (Critchley 1978: 23; Ganshof 1964: 102f).53 But the most successful resolution—and that which became the dominant form of feudalism in France, England, and Italy—was “liegancy”: one lord, the “liege lord,” had priority over the others. From the others, one could accept benefices and give conditional support, but never unconditional support that might lead to conflict with the interests of the liege lord. This resolution more or less simply re-created the patron-client relation under a new name, and since the conditions remained unchanged, it is not surprising that before too long, there were cases in which vassals took more than one liege lord (Bloch 1961 [1940]: 217; Ganshof 1964: 103f; Fourquin 1976: 129); the same attempts to resolve contradictions began again.54
This change probably came from a slow transformation from the basic structural unit being a warlord and his direct dependents in an anarchic situation to being a privileged person in a protostate.55 While one can only serve one master in battle, one can have more than one sponsor. This shift to sponsorship, as we see in the next chapter, is often associated with the monetarization of relations, as in what is often called “bastard feudalism” in England (here also see the discussion of Kettering [1986: 207–10]).
Even where clients may not hold multiple patrons at the same time, it has been suggested that they can play off one patron against another if there is the possibility of exchanging one for another. This may be true (for example, see Fourquin 1976: 132; and Kettering 1986: 21, 28), although it is easy to mistakenly assume that the mere existence of patron-patron conflict will empower clients (e.g., Wolf 1977: 174). Patrons can compete using clients without necessarily competing for clients, and even when clients have the ability to reject one patron in order to take up another, this does not necessarily increase their bargaining power. This increased power of clients arises only if the supply of clients is limited and their support necessary.
Consequently, patrons facing competition for clients can respond by attempting to recruit new members to the system as a whole. There is some reason to believe that the extension of Roman citizenship was spurred by the desire of patrons to enlarge their retinues. If all those in some area granted citizenship can be expected to swear fealty to the patron in charge, this patron will benefit dramatically from their political inclusion, and this might explain the “democratic” leanings of many popular military leaders (see Taylor 1968: 22, 46f; Wallace-Hadrill 1989: 78f).56 In sum, allowing for multiple patrons will push a pyramid toward a web structure; empowering the clients so that an effective competition takes place will probably push a pyramid toward a multihub structure, though the mere fact of competition between patrons does not itself produce the establishment of horizontal patron-patron ties of alliance (which we examine in chapter 8).
The final way that patronage structures can be obliterated is somewhat different, for the structure seems on the surface to be unchanged, and this is by the elimination of the principle of mediation, and its replacement with transitivity. Mediation means that those at the top often have the de facto responsibility for supporting those a few rungs down, but have no control over them. The great strength of patronage structures is this very weakness—because of the lack of transitivity, relationships can be concatenated to make large structures in a way that the other relationships we have examined cannot. But members of large and weak structures have received little in the way of a guarantee that the structural principles involved will not be changed.
In particular, and not surprisingly, a change from mediation to transitivity is often pushed for on the part of those at the top of the structure, and is resisted by those in the middle. If accomplished, the patronage triangle has become what I will call a command tree. This process is historically of the greatest importance, for it may be that what we call government originates in patronage relationships (cf. Mair 1977: 141), but involves the substitution of transitive command relations for antitransitive patronage relations.
Large modern states do many things—they deliver mail, inspect meat, hand out passports, manage forests, and so on. But as sets of social structures they vary from other forms of government most notably in the ways in which we find large numbers of commoners mobilized, namely the army and the party. We go on to examine how these structures emerge via the imposition of transitivity into previously intransitive patronage pyramids.
1 In a word, a hub structure is best seen as composed of symmetric relations if the space is understood as Cartesian, but if it is understood as polar (which would be natural if we were thinking of organization in terms of overall “centrality”), it is basically equivalent to a triangle.
2 See also the description of a Yanomamö headman (Chagnon 1968b: 108) who cannot order followers, but just starts a task and hopes the others will eventually join in.
3 Interestingly, I found after writing this that Dederich described Synanon as a wheel with spokes and hub (Janzen 2001: 32).
4 And probably the only reason people are so familiar with Christianity today is that Paul moved along frequently traveled channels between urban regions, setting up spatially distinct triangles in different city, along the lines recently discussed in formal terms by Hedström et al (2000). For Paul’s work, see Meeks (1983).
5 Political scientists generally use the term to indicate the ways in which government officials may reward supporters. This does not always involve an abiding social relationship, which is what the anthropologists investigate.
6 A further complication is that such relations also are favored where there is a shortage of currency; unable to rely on wages to exchange for services, employers or lords must either grant benefices or take underlings into their households (Bloch 1961 [1940]: 68, 163). (Such a shortage of money at the end of the Roman empire played a role in moving the patronage system to be fused with the system of dependency in access to land [Beeler 1971: 4, 11].) Since expansion of a monetary supply usually requires stable government due to the rareness of silver and gold and the ease of adulteration, anarchy also contributes to a disreliance on money, though this shortage may arise for other reasons.
7 It may be to a nontrivial extent that both a reliance on the personalism of relationships and the small size of polities stem from nonexistence or scarcity of means of administration, most important literacy and other forms of communication (especially paper; van Creveld [1991: 38] points to the simultaneous arrival of paper and gunpowder from the east, both facilitating the expansion of Western governance structures).
8 Indeed, even in the late Roman empire, the constant disruption led the rich to fortify their estates and made them look quite a bit like the noble lords that were to come (Anderson 1974: 94; Contamine 1998 [1980]: 16); it is at this time that the word patron becomes standardized for such protectors (Beeler 1971: 2).
9 The negative character of early vassalage (Ganshof 1964: 84)–a promise not to injure the lord–is a good indicator of the relation between the institution and the anarchy of the time. It must be admitted, however, that such anarchic conditions and a resulting need for protection do not seem fundamental to Indian feudalism (see Sharma 1965: 33).
10 The Domesday Book further suggests that at that time in England it was acknowledged that clients always had the right “to betake themselves to another lord” (Bloch 1961 [1940]: 185).
11 In Italy, the Roman emphasis on a distinction between alienation rights and use rights limited the flexibility with which such multiple ownership could be systematized. In Germany, where the issue was the enjoyment of a property, multiple “enjoyment” was quite easy to comprehend, and, says Critchley (1978: 19f), “in England, the common lawyers lived in a world of their own.” For the tie between grants of land and pseudo-kinship relations among Palauans, see Smith (1983: 40).
12 As will be discussed in more detail below, there is some vagueness in what the word feudalism denotes; a narrow definition sees it as the prevailing structure of interelite relationships in Western Europe in the 10th to 12th centuries, especially in the center of the Carolingian area (Ganshof 1964: xv, xvii; Bloch 1961 [1940]: 448), perhaps calling the earlier system “vassalage” (e.g., Fourquin 1976: 45, 66), while a wider one sees it encompassing practically all European governmental systems between the fall of the western Roman empire and the Protestant reformation (also compare the discussion and typology of Andreski 1968: 143). Here it will be used to describe the connection between (on the one hand) granting of benefices, which involved rights of land utilization and dispensation of justice, and (on the other) vows of fealty, which involved obligations of military or financial support. The mere presence of powerful magnates with armed retainers who usurp governance functions does not deserve the term if there is no linkage of vassalage and benefice (e.g., early Byzantium and Byzantine Egypt, Merovingian Europe; see Bachrach 1967; Andreski 1968: 56; also see Anderson 1974: 98).
On the other hand, even in European feudalism, an important component was often the presence of such retainers (the lord’s “men”). Indeed, for the case of 14th-century England (after the devolution of the strict form arising from the Norman conquest), Bean (1989; cf. Bloch 1961 [1940]: 167–71, 236) has argued that while the relations between the king and his vassals fits this classic model, the most important relationships between most lords and their own underlings was an extension of the household relationship. That is, the lord attempted to treat more and more persons as if they were his personal staff and owed him the same obedience as a dependent who resided wholly within his house. This is different from the construction of a feudal pyramid on the basis of agreements of fealty. Bean argues that such householdlike relations both predated and postdated feudal relations properly understood. For one, an important part of compensation packages was often special wages for the time the retainer appeared in the lord’s household, implying that such presence was a regular part of being a client. This emphasis on the household does not seem to be as important in the 15th century, but in the 16th century we see not only an explosion of livery but arrangements whereby retainers were provided with fees and sponsorship in the court and in suits instead of benefices (Bellamy 1989: 79f, 82f, 91ff, 96).
Accordingly, feudalism was often “purest” in relatively earlier periods where it was less widespread, though early vassalage often was not closely tied to transfer of rights in land (e.g., the well studied case of Burgundy). Certainly “feudalism” here has nothing to do with the mode of production associated with the Middle Ages, namely serfdom (see Fourquin 1976: 13, 43, 46); though the two are often confused they are to a fair degree historically separable (while true feudalism does not coexist with free labor, there are many forms of serfdom that do not involve patronage pyramids within the elite). There are, of course, wide geographical differences: in England feudalism came from the invasion of 1066 in a top-down fashion, in Germany there was less feudal consolidation and the land aspect was emphasized over the bond of vassalage, while in the Carolingian areas (especially in Northern to Eastern France [Fourquin 1976: 71, 73f, 76]) feudalism was a result of the disintegration of a hasty empire built on Roman administrative ruins. In general, Bloch (1961 [1940]: 188) suggests that places where feudalism was imported exhibited a higher degree of structural simplicity and regularity than where it spontaneously developed.
13 Some might distinguish patron-vassal relations from patron-client relations, reserving the former for cases in which the subordinate party is still a member of an elite, and the latter for when the subordinate party is not, but I do not think this is a structurally crucial distinction. Indeed, medieval treatises on law used the word client to refer to vassals (especially subvassals); Tacitus describes a warrior retinue as clients (Critchley 1978: 102f), and Asian and Indian high kings referred to kings who had submitted to them as “slaves” (and in medieval India, such subkings might have to fan the emperor as a humiliating service [Sharma 1965: 26; Hanson 2001: 34]). The issue was not the terminology but the obvious status of the participants–“homage” was at times used to describe both the relation between lords and serfs and higher lords and noble vassals (Bloch 1961 [1940]: 161, 261; though see Duby 1991: 73), though no one would confuse the relationships (for one thing, only the latter had to be renewed each generation).
14 It is possible that there were patron-client relationships within the subjugated stratum, but if so, these have gone undocumented, and probably quickly shaded off into forms of criminality.
15 There are also cases of other pyramids in which the top layers are not necessarily “noble,” such as the magnates’ courts of 17th-century Poland as discussed by Pośpiech and Tygielski (1981: 77), consisting of the magnate (a major lord) as the apex of a pyramid of personally connected clerks organized hierarchically (from chamberlain, secretary, equerry, treasurer) on top of serving men on top of valets and liverymen.
16 In some cases, such kin relations cannot be the framework for a patronage structure because changes (usually associated with colonization and consequent modernization) have disempowered the elders. An example here is found in the “clans” of postcolonial Senegal that are really the political entourages of a single patron and have no real kin basis but fill the power vacuum left by the weakening of previous intercaste patronage relationships (see Foltz 1977: 245).
17 Indeed, residues of feudal property rights (though not obligations for military service) persisted into the 19th century in Sicily and the German states, and into the 20th in Scotland (Graziano 1977: 362; Critchley 1978: 22f), and even after abolition, fundamentally feudal patterns of interaction–tied to the fundamental fact of land ownership–frequently remained for generations afterward (see, e.g., Bell 1994).
18 And since, as we shall see, these patronage structures are the core of factions, it is possible for patron saints to be the “leaders” of factions, as in the case of the Maltese studied by Boissevain (1964).
19 A counterexample might be ancient Greece, where it was also assumed that the gods were the masters of men, and that men were dependent on gods as a client on a patron, and that this dependence could never be fully paid off. Yet the “free” Greek citizen was not similarly beholden to some mortal (Vernant 1995: 7f), and thus an integrated structure did not develop in the same way as in later Catholic countries.
20 Funnily enough, even though clearly mutual relations such as spending time together obviously satisfy the frequent demand that exchanges be equal, theorists have not attempted to analyze these relations in terms of exchange, probably because of the very incorrect assumption that there is no reason to exchange one thing for an identical one, or perhaps to avoid the paradox that this generates (for how could two unequal people exchange the same amount of time with one another?). Instead, they have attempted to analyze obviously unequal relations as involving equal exchange.
21 Holding a stirrup does make it easier to get on a horse, especially if one has on armor, and it can be embarrassing to be unable to mount elegantly. That this aid was symbolic of the servile and supportive nature of the vassal, however, is indicated by the fact that this role is found outside of the case of armored knights: see the example of a Mexican “big man” whose apprentice had this job in Friedrich (1977: 269).
22 Indeed, this link was so strong that in a number of cases it led to ladies taking up arms: if they held the land, it was natural that they should fight (Contamine 1998 [1980]: 241).
23 As discussed in chapter 7, administrative service was often part of the obligations of a vassal in medieval India (Sharma 1965: 100).
24 Thus historically, feudalism as an institution developed in accordance with the fusion of two different institutions, that of vassalage in which a vassal transferred his self-direction to a lord, and the granting of benefices, in which a lord granted the use rights to certain property to someone below cost (Ganshof 1964: 15). When they became separated again later, allowing for the de facto sale of benefices, the relationship became one of exchange and the pure structural qualities we are about to examine dissolved.
25 Relations between samurai were stronger and more militaristic than the patronage relations of the court. While the lord-vassal relation was assimilated to father-son relations in the ideal, the reality was generally that relations were tenuous and self-interested (see Farris 1995: 312; Varley 1999: 55f; Ikegami 1995: 78).
26 A partial exception is the importance of samurai organizations of equals (ikki) and other horizontal associations of nonsamurai that gained importance in the late medieval period. Still, comments Ikegami (1995: 128f, also 122, 337), “Japanese history in the 16th century is the record of a process of violent suppression of horizontal forms of social organization by a strictly hierarchical system of samurai vassalage.”
27 At the same time, there was an understanding that the transitively implied relationship has some content (Montgomery 2007): thus my lord’s lord, while not my lord, is my “lord suzerain,” though this probably only appeared as the structure of vassalage was breaking down in the 14th century (and transitively implied relationships became common in the 17th [Kettering 1986: 21]). Further, there were some exceptions to the suppression of relationships with the lord’s lord. One was that when a lord died without an heir, his vassals would temporarily be regarded as vassals of the lord’s lord until an heir was chosen. A second was when a tie of vassalage was broken due to the improper conduct of the lord, the lord’s lord could assume the relationship with the client, to avoid the latter being left unprotected (Bloch 1961 [1940]: 228f, 146). Third, kings were at times able to make sure that those swearing fealty to one of their vassals maintained primary allegiance to the crown (Duby 1991: 214).
28 Thus Weingrod (1968: 382) equates patronage with brokerage, emphasizing as a condition for the development of patronage the existence of “gaps” in the structure of society due to primitive transportation systems or markets, linguistic diversity, laws against migration, etc., also. But his examples (the classic Chinese state, for one, and his own study of Sardinia in response to increasing national integration) demonstrate that these pertain to situations in which a state already exists. For another study of such patronage, see Stokes (1995: 16) on Peru.
29 In other cases, the increased emphasis on brokerage can disempower those who previously held semiformal positions of leadership; see, e.g., Helm (1972: 78) on chiefs of the Dogrib Indians now expected to be brokers: “My people come to me for help,” says one, “and I have nothing to give them but my fingernails.”
30 Or rather, little he was likely to get. This is a simplification of actual peasant life, but as life approaches this point of simplicity, patronage relations become more purely vertical.
31 Similarly, the process of state-building in 17th-century France involved the state transforming local patrons into brokers who could be used to extend central control into the provinces; indeed, the state intendants were arranged in some places to form an alternative pyramid through which patronage could flow, starving the great nobles and transforming patrons into brokers (Kettering 1986: 91, 159, 234ff); we return to this case in chapter 9.
32 For the case of Peru, which seems to have such vertical relations, as opposed to an “amigocracy,” see Stokes (1995: 55).
33 To take the case of the Polish magnates of the 17th century, there was clearly a well understood relationship between rights to land and clientship. Thus the lord Tomasz Zamoyski assumed he had the right to terminate the lease of a man who had taken a different patron: “For whoever takes a lease from me becomes my serving man, and . . . as he was serving another man, he could not lease property from me” (Pośpiech and Tygielski 1981: 84; cf. 87).
34 When there is a distinction between use rights and alienation rights over the land, patronage is likely to flow from the former as opposed to the latter (see Silverman [1965: 176f] and Weingrod [1968:391] on the mezzadria [sharecropping] relation in Italy).
35 Even a land-based system can be unstable if there is unclaimed arable land (Scott and Kerkvliet 1977: 448).
36 Outside of this context, “big men” may not necessarily be men, but I retain the gendered term for consistency with earlier accounts.
37 As one big man recalled, it was once he had some valuables that “I could start doing things in feasts, and making a name for myself” (Keesing 1983: 67)
38 It must be conceded that it is possible for a patronage structure to arise without clear claim to the resource in question if a similar sleight of hand can be used to grant what is not yours to grant. In a word, if A has X but is not strong enough to keep it, while B is not strong enough to take it unilaterally, C may benefit from “giving” X to B. Since B will naturally then defend the legitimacy of C’s right to allocate X and other resources, C’s act is something akin to a self-fulfilling prophecy. A tolerable approximation of this process may in part underlie the development of feudalism in Western Europe (alternatives are discussed below). Having previously expropriated church property to give to supporters in the constant civil wars of the 7th-8th centuries (Bachrach 1972: 98), the Frankish monarchs needed to pacify relations with the church and to retrieve some of the power that had devolved downward. The solution was to claim de jure ownership over the property held de facto by the lords, and thus to allow these powerful military leaders to continue to hold that which they were unlikely to give up, while asserting an ultimate right to control that over which they had no real control . . . at the time (Ganshof 1964: 16f). The efficacy of this move, however, presupposed the other factors we have already examined.
39 For example, in 17th-century France, Kettering (1986: 56, 59, 85, 99, 140ff) argues, most clients tried to secure ties to more than one patron, and brokers (both those who had resources of their own and those who were mere go-betweens) were crucial in mediating ties between patrons. But already this is a form of patronage that is heavily connected to an emerging national state, with parallel clienteles based in the nobility and in the administration; further, these brokers played a role in pushing forward the national political integration of a still poorly coordinated state. They managed to cement the provincial nobility to the crown by centralizing the source of all patronage, but at a cost: to the extent that patronage was given with no strings attached, the patrons allowed their clients to build up an independent power source; to the extent that the center retained control, patronage was politicized. In chapter 8 we shall see how this tends to provoke civil war.
40 For an example of irregular patronage in a “boss” system of patronage in New York, see Johnston (1979: 395); this irregularity of obligation–and indeed ambiguity of the clients’ own obligations–is also pointed to in the case of the Sicilian capitalist broker-patrons by Graziano (1977: 363).
41 This corresponds to the dimension of dependence on some external scaffolding for establishing a preexisting vertical ordering of persons.
42 To the extent that the system of relationships is redistributive, this lessens the differentiation of persons though there may be other relationships that increase this differentiation, if this structure is dependent on another structure. We recall that this parallels an emphasis on equality as opposed to inequality; while all these relationships are vertical, it is the boss relationship that heightens the perception of inequality in its negative sense.
43 This may correspond (albeit quite roughly) to the involution of relationships–their spontaneous organization and feedback.
44 While Englehart (2001: 20, 27) emphasizes that in the Siamese context, it was labor, and not land, that was in short supply and was to be mobilized through the pyramid, and that “land was not a source of power because it was abundant and because anyone could hold it,” his own work suggests that this is an oversimplification. While it is true that even villages (meuang) were defined as assemblages of rights over persons, as opposed to rights over land, there was clearly some relation of patronage to land: there was an attempt to designate rank in the patronage pyramid in terms of sakdina (or sagdinaa), which means “power of the [rice] fields” and was originally measured in land area (Terwiel 1984: 22; Englehart 2001: 26, 59).
45 Kemp (1984) stresses the simplicity of this view of the Thai structure, which emphasizes the most formal forms of patronage at the expense of more affectual, ambiguous, and flexible relations often described using kinship terminology.
46 This is also true of Japan, as argued by Ikegami (1995: 37f), who stresses that the early modern Japanese state was created via the destruction of horizontal relationships and the consolidation of a particular structure of vertical relationships of vassalage. Ikegami also notes the duality between the cultural understanding of the nature of the relationship in question and the structural form evolving.
47 Much of the early work on patronage was motivated by the modernization theories of comparative political science, which took universalism in political systems as some sort of timeless benchmark according to which other systems would be understood, and hence it was usually concluded that patron-client structures were vestigial at best or deliberately obstructionist at worst. The actual breakdown in patronage systems that might arise with economic development or world system political integration was therefore misunderstood as successful evolution; such interpretations are not reviewed here.
48 One strange exception is the Hutu-Tutsi case, where clients supposedly could have more than one patron by assigning different sons to perform the required duties (Mair 1977: 144).
49 The shift from patronage to sponsorship is associated with an increased ambiguity in the nature of the relationship and, dually, a decreased structural clarity (see Silver [2002]). There is also decreased clarity when there are two semi-independent hierarchies, as in the case of France discussed in note 39, or in medieval Japan in which there were both ties of vassalage and a second set of relationships pertaining to landholding (the shoen system) and tied to the imperial court (Ikegami 1995: 80, 121f)
50 While the citizens of the democratic city-states such as Athens in the classical age generally did not have formal patrons, they often had sponsors, and in Athens, the metics–free noncitizens, often immigrants from other Greek cities or freed slaves–were required to have a citizen as a prostates, who would mediate between them and political institutions (Millett 1989: 34).
51 For example, in Japan the samurai at first had multiple samurai patrons, a relation of vassalage separate from the relation some had as clients of the aristocracy (Ikegami 1995: 78f, also 123).
52 Of course, as Fourquin (1976: 116) points out, serf-master relations by nature remained “one patron.”
53 Such careful agreements were found where a lord pledged support of a prince with reservations so that this agreement did not contradict previously existing vassalage. Thus when Robert II, count of Flanders, made a treaty with Henry I of England in 1101, he agreed to support Henry except against the king of France (Philip), to whom he had pledged fealty. He would attempt to dissuade Philip from attacking England, but only by “supplication” and not by undue influence or bad council (Contamine 1998 [1980]: 49).
54 Similarly, the particular relationships underlying feudalism (the linkage of the granting of a benefice to the vow of fealty) also broke down when vassals were able to alienate fiefs that in some sense were not theirs to alienate, most profitably by subenfoeffment but also by sale (Fourquin 1976: 157). While such acts were outlawed in England in 1290, the new law simply led to more complex ways of allowing multiple claims on the product of certain property (Critchley 1978: 20). The underlying tendency, probably a result of the increasing power of vassals due to the continuous appropriation of benefices, could not be suppressed by legislation.
55 For example, Bean (1989: 236) argues that in late medieval England we can see a shift from “lordship” to “patronage,” meaning a shift from what the lord offers the client is his own wealth to one in which it is his ability to sponsor.
56 For example, Caesar enfranchised the Transpadane region, inflating the number of citizens by as much as 50%, “and making the new citizens his personal clients” (Taylor 1968: 173).
