ABSTRACT
The urban system of most Latin American countries is dominated by a primate city which overwhelms the cultural, economic, political, and social life of the nation. This paper reviews the concept of primacy, the pattern of primacy in Latin America, and the changes in primacy over time in selected countries. It confirms that Latin America, among the world's regions, is most characterized by high primacy. Most Latin American countries not only have a primate city, they exhibit strong or dominant primacy. A disturbing recent trend is the growth of many of the primate cities into giant cities with populations exceeding six million.

INTRODUCTION
The tidal wave of urbanization that has swept through the less developed world in the last half century has been most striking in Latin America. No other region has been urbanizing so strongly and so swiftly (Table 1). Since 1950 the proportion of the population living in places defined as urban in Latin America has gone from 41.0 percent to 65.4 percent in 1980. This percentage is not too far below North America (73.9 percent) or Europe (70.2 percent); it is a bit above the U.S.S.R. (63.1 percent), and far above Asia (26.0 percent), Africa (27.0 percent), or the world (36.6 percent). Within the region, the countries of temperate South America (Argentina, Chile, and Uruguay) have some of the highest levels of urbanization in the world.

It has not been simply the gathering of people into cities in the region, but their increasing concentration in the rapidly growing larger cities. At the top of the urban hierarchy are the dominant cities, or "primate cities," a term Mark Jefferson coined fifty years ago. He called them, "disproportionately large and exceptionally expressive of natural capacity and feeling" (Jefferson 1939, 227). Although Jefferson introduced the term in an almost offhand way with a mainly anecdotal argument, the concept has become a familiar term in the geographic literature (Linsky 1965). Curiosly, the concept of primacy has never received much attention from geographers in the region where it is most dominant, Latin America. Wilkie's massive work on population and urbanization for the region deals only tangentially with primacy (Wilkie 1986).

This paper is an attempt to redress this gap by examining the pattern of primacy in Latin America. A discussion of the concept of primacy and the alternative ways of measuring it is followed by a review of the primacy pattern circa 1980. Its changes over time in selected countries and how Latin America compares with other regions in primacy are also examined. The consequences of primacy complete the paper. The thesis of the paper is that in such a rapidly urbanizing region as Latin America, characterized by so many ,dominant primate cities, the concept of primacy is

[end p. 71]crucial to understanding these increasingly urban countries whose higher order urban functions are concentrated in the leading cities.

THE CONCEPT OF PRIMACY AND ITS MEASUREMENT
Mark Jefferson titled his article, "The Law of the Primate City," an unfortunate choice of the word "law" in the sense that subsequent investigations have been unable to discover any lawful processes. There are seemingly no conditions under which it invariably appears. It occurs in developed and less developed countries, in rich and poor countries, and in small and more populous countries (Browning 1958; Browning 1962; Stetson 1976). It has thwarted most efforts to generalize. One of the few generalizations pertains to those countries where it is not found. None of the countries with very large national territories-Australia, Brazil, Canada, China, United States, and USSR-have a primate city.

Primacy has most often been discussed in connection with the city size distribution question, a topic the economist Harry W. Richardson mainntained was "one of the most fascinating intellectual problems in urban analysis" (Richardson 1973, 139). For almost forty years urbanists have been intrigued by Zipf's ranksize rule expressed as:

where Pr is the population of the r-the largest city (Zipf 1949). If the two axes of the city size distribution (population size and rank) are scaled logarithmically, the distribution becomes a negative sloping straight line. This presumed empirical regularity cried out for an explanation but thus far there have been no theoretically convincing arguments; although there have been numerous atttempts. In their zeal for curve-fitting, many authors assumed the relationship to be the norm, but the geographer Eric Sheppard has correctly pointed out that such abstract quantitative analysis has removed the cities "entirely from their context. No information is retained about the relative location of the cities in space, their economic function, or any other aspects that might explain how they interact together within the system" (Sheppard 1982, 127).

Many of the papers on city size distribution consider a condition of primacy to be simply a variant of the rank-size rule (McGreevay 1971). Perhaps this is because the primate city is simply one dot on the array of cities along the negative slope line. But, that one dot has profound consequence. And, it is argued here that a condition of primacy is fundamentally different from a rank-size condition. Primate cities, especially the more prominenet ones, dominate the higher-order urban functions of their countries. The consequences of high primacy are discussed briefly at the end of this paper.

Although Jefferson in his brief paper discussed the characteristics of primate cities and gave some examples (London, Copenhagen, and Mexico City), he never precisely specified what constituted "dissproportionately large." Subsequent authors have devised various ways to measure primacy. This paper offers three ways (Table 2). The simplest method-used by Jefferson-simply divides the population of the first city by the second to determine how many times larger is the first city. A variation of this approach takes the sum of the second, third, and sometimes the fourth cities as the divisor. An alternative approach uses the percent of the urban population in the leading city (Renaud 1981).

The problem of determining the city ratios has been the availability of comparably defined or delimited cities. If every nation used a standard definition of the city such as the U.S. Census Urbanized Areas or Standard Metropolitan Areas, there would be little difficulty. Unfortunately, the political definition of the city is often the only one available and this may seriously underbound the city and thus underestimate the population. This study utilizes the estimates of the Kingsley Davis group which carefully and painstakingly delimited cities over 100,000 in the world for 1950 and 1960, and the UN and the Statistical Abstract of Latin America which tried to define urban "agglomerations" rather than political cities (Davis 1959 and 1969; UN 1987; Wilkie 1981, 1986). The use of the percent of the urban population in the largest city measure also has its problems. The largest city needs to be delimited on an agglomeration rather than a political basis as the urban population definition varies from country to country. The U.S. has for some time rather arbitrarily considered 2,500 and above to be an urban settlement; whereas some countries consider 500 urban and others 20,000 or more. Given the difficulties of obtaining comparable data, small differences in the measures of primacy should not be taken seriously.

THE PATTERN OF PRIMACY IN LATIN AMERICA
Examining Table 2, it is apparent that the two-city and four-city primacy ratio ranks are quite similar (the Spearman rank order coefficient is .98). The two-city ratio is preferred because it is simpler and more straight forward; it gives a better indication of how far the primate city is above its nearest competition; and it does not depend on comparison with more than one additional city. This can be significant if the smaller cities are politically defined [end p. 72]

rather than defined by urban agglomeration. If the rank-size rule held perfectly for the four-city version, the ratio would be .92, a value exceeded by all countries except Brazil. The two-city ratio is a good proxy for the four-city ratio.

There is much less correlation between the ranks of the two-city ratio and the percent of the urban population in the largest city (.65), and virtually none with the percent urban measure (.06). The writers using the percent of the urban population in the largest city as the primacy index have done so without much explanation as to why it is preferred over the two or four city ratios. Although there is correspondence generally in the high and low ranks, with the two-city ratio the measure can be misleading. Honduras ranks above Mexico even though it has a low two-city ratio and Mexico City is an outstanding example of a primate city. Generally, the smaller countries-in area and population-rank higher on the percent of the urban population in the first city measure.

If the rank-size rule were in effect for the countries, we would expect the two-city ratio to be 2.0 since the first city would be twice the size of the second city. Only two countries are below 2.0 and in this century these two, Brazil and Ecuador, have both seen the replacement of the former leading city by a rival. São Paulo has passed Rio and Guayaquil passed Quito. Striking primacy is the main conclusion from Table 2. It is found in all parts of Latin America, although temperate South America (including Paraguay) is especially characterized by high primacy. Such relatively well-developed countries as Chile, Argentina, and Uruguay have high primacy, but so do less-developed countries such as Paraguay, Guatemala, and Haiti. Brazil fits the world-wide pattern of there being no primacy in countries with very large areas, but the rather small country of Ecuador also lacks primacy.

LATIN AMERICAN PRIMACY IN THE WORLD CONTEXT
These are the primacy ratios for Latin American countries, but how do they rank in a world conntext? To answer this question a group of countries around the world are included for comparison (Table 3). The countries were selected to represent different geographical regions, different levels of development, and different ideologies. The Table deliberately over represents Europe and the most populous countries. It under represents the smaller, [end p. 73] less developed countries because of problems with data comparability and reliability.

The countries were ranked by their two-city primacy ratios and divided into four classes. Previous discussions of primacy have generally assumed a dichotomous designation-a nation either has or does not have a primate city--or a continuum. I felt our understanding of primacy would be advanced by distinguishing degrees of primacy. The problem was the designation of classes. After considerable thought and experimentation, I abandoned statistical class interval making techniques (i.e., standard deviations) as too mechanistic and adopted the accompanying classification on the basis of the following reasoning. If we accept the viability of the rank size rule, the first city should be twice the size of the second city. A primacy ratio of less than two would certainly not fit Jefferson's "disproportionately large" condition. The other classes were on the basis of twos--2, 4, and 6. There is still a considerable residue of arbitrariness to this classification, so it is offered as a provisional. An alternative would be to take the natural break between 7.8 and 10.0 as the class beginning for the dominant category. This was not done because it seemed appropriate to keep Mexico City and Paris in the dominant primacy class.

Table 3 exposes a weak point of the primacy ratios; they appear to underestimate the primacy role of many of the leading cities. London and Tokyo, for example, are more dominant than their two-city primacy ratios would suggest (3.6 and only 2.0, respectively). The percent of the urban population in the first city measure is no better; London and Tokyo are only about 20 percent. Ideally, a primacy index should be based on the share of selected quaternary (primarily offices) and quinary (control) functions: e.g., selected financial, governmental, transportation, media, company headquarters and cultural activities. However, it is difficult enough to get comparable definitions of cities, let alone the information necessary to construct an ideal index of primacy.

The top ten countries, except one, on the rank of dominant primacy are Latin American. The exception is significant because it is a former colony of Spain, the Philippines. This demonstrates that Latin America is remarkable not only for the frequency of primate cities, but their dominance as well. Paris, a prototype primate city, is well down in rank. In Table 3, a symbol beside the primate city's name indicates if it is also the nation's capital. All of the primate cities in the first two categories are national capitals, and all but two cities in the weak primacy class. The distribution by primacy level for the major regions of the world is given in Table 4. Only 10 percent of the Latin [end p. 74] American countries falls in the no primacy category, a far lower percentage than any other region.

HISTORICAL ASPECTS OF PRIMACY
The pattern of primacy was established early by the Spanish, whose colonial settlement policies (i.e, the Law of the Indies) favored a highly centralized government and the concentration of power--religious as well as secular--in the leading city (Browning 1962). This was especially true of the viceregal cities, Mexico City in Middle America and Lima in South America. The colonial economy was eventually replaced by a capitalist-type economy which maintained the concentrated development in the primate city (Smith 1985; Walters 1985).

High primacy often has been viewed as a condition typical of less-developed countries, especially those which have had a history of colonial rule. As a country becomes more developed and sheds the colonial mold, it is assumed that high primacy decreases and the urban-size distribution will move in the direction of the rank-size distribution. Brian Berry (1970, 1977) has taken this position in a number of publications. Peter Hall (1984) produced a schematic general model of urban growth and change. It had three key measures: the degree of urbanization, the primacy index, and the extent of suburban decentralization. These conditions were taken through five stages; the first three had primacy increasing and the last two had primacy decreasing. The developing nations were in stages one through three, whereas Great Britain and the United States had arrived at stage five.

How has Latin America fared in this regard? Has it gone in the direction posited by the urban theorists? To answer these questions, some historical data have been assembled (Table 5). Because of the difficulty of obtaining data-principally definitions of cities according to their size by urban agglomeration rather than the political city-nine countries in Latin America were selected. Two priimacy measures were included: a two-city primacy index and the percent of the urban population in the largest city.

The results are mixed. For some countries the two-city primacy index declined as would be expected from the above discussion (Mexico, Cuba, and Peru). Others remained essentially the same (Brazil, and to some extent Argentina and Ecuador). But three countries showed fairly sharp increases in the two-city ratios: Venezuela, Colombia and Chile. The two measures of primacy generally moved in the same direction. The inconclusive results do not necessarily negate the theoretical asssumptions; the time span is too short for one thing, but neither do the data give a strong confirmation.

Some of the dominant primate cities have slipped over the years, but they are still overrwhelmingly dominant within their countries and they would appear to retain that dominance well into the future. Even in Cuba, where Castro has deliberately not favored Havana for nearly thirty years, Havana is still dominant, with no apparent rival.

POLICY ISSUES OF HIGH PRIMACY
If Latin American countries are generally characcterized by high primacy, what are the consequences which flow from this condition and the policy issues which emerge? These are large and controversial questions; to do full justice to them would take a much longer paper than is appropriate here. Yet to ignore them is to make the same mistake that is so often made in city size distribution studies which treat city systems in very abstract terms. High primacy has a very real impact in the countries in which it is found and on the lives of millions of people. What follows is a [end p. 75]

brief summary of some of the main consequences and policy issues. Many of the statements may be regarded as propositions; I believe they are valid, but do not provide the data and discussion to fully support them because of space limitations.

Many observers, including the author, have a tendency to view high primacy in terms of its negative consequences. Others believe this is an overly emotional reaction, that high primacy has its benefits and that it is difficult to do anything about it in any event. It should be noted at the outset that the consequences of high primacy need to be viewed in the context in which they are found. In Latin America, for example, many of the countries in Caribbean and Central America are so small in area and population that it makes sense to have most of the high order urban functions in one city. Primate cities in these countries can easily serve the entire country and are in no danger of becoming excessively large. In larger countries, however, the concentration of so much of a country's population, political power, wealth, brains and talent often comes at the expense of the regional centers. The siphoning off from the provinces of these able and ambitious people deprives these regions of people with leadership qualities. This was one of the reasons why the French geographer, Gravier (1947), wrote Paris and the French Desert, and why French government policy, especially during the de Gaulle regime, has tried with some success to reduce the dominance of Paris by making the provincial centers more competitive.

One of the telling arguments by those not so alarmed by high primacy is the fact that virtually all attempts to control the growth of the lead city have been unsuccessful. This is especially true in capitalist countries, but state controlled old-style communist countries also have had problems in this regard. In the 1920s Lenin decreed that Moscow should be no larger than four million-it is now more than double that size. Chairman Mao, with his insistence on eliminating differences between the city and countryside, was somewhat more successful in slowing the growth of Shanghai, which was doubly damned in his view as a center of foreign influence and contamination.

The limited success of efforts to control the largest cities has particular meaning in Latin Amerrica where some of the primate cities have grown into giant cities. A recent estimate, by the United Nations for the year 2000 gave Mexico City the dubious distinction of becoming the world's largest city, 25.8 million people. This estimate was scaled down from a 1976 estimate of 31.6 million (U.N. 1987). São Paulo was in second place with 24.0 million. What is striking and disturbing is the projection that the two largest and four of the top twelve cities in the world in the year 2000 will be Latin American cities; this includes Buenos Aires and Rio de Janeiro. Even Lima, Peru, far behind Paris in 1950, will be closing in on the French metropolis by the year 2000. At the turn of the century, twelve cities in Latin America will be over four million and six cities over six million. The growth of many of these cities is out of control and their governments are increasingly unable to service them adequately.

Perhaps one of the reasons why efforts to control the growth of larger cities were not pursued very vigorously was the belief by many that large primate cities were necessary for economic growth. Mera (1973, 324) declared that his research "connfirms the result of earlier studies which tend to support positive correlation between the growth of large cities and economic development." Alonso (1980), in an important paper, hypothesized rising [end p. 76] economic growth rates associated with increasing primacy and subsequently declining rates associiated with decreasing primacy.

Richardson and Schwartz (1988) recently re-examined and redid some of the earlier studies of EI-Shakhs (1972), "Development, primacy and system of cities," and Mera (1973), "On the urban agglomeration and economic efficiency." These authors were two of the strongest proponents of the link between economic development and primacy. After making what they felt were some modificaations and improvements in El-Shakh's study, they found little support for his conclusions. They found serious flaws in Mera's influential paper. The index of primacy used was the ratio of the primate city to the total population which they argued made little sense; and this is confirmed in this paper (Table 2). After correcting for this and other dubious procedures, they found a very weak relationship between primacy and economic growth. Richardson and Schwartz (1988) seriously questioned the prevailing conventional economic wisdom that primacy was associated with economic growth. They found a closer relationship between primacy and demographic variables rather than economic ones.

Growth of the larger primate cities has worsened already severe urban problems: traffic, polllution (air and water), the provision of water and waste disposal, and increases in land prices and crime levels. Politicians and political parties are offten particularly sensitive to the needs of the primate city, traditionally a symbol of national pride and achievement. The political authorities also view these giant cities as potential political tinderboxes of discontent. Thus, there is a tendency to favor the primate city at the expense of the smaller towns and rural areas who are left to muddle through because they pose less of a threat to political stability. Urban services (i. e., transportation) and food are provided at highly .subsidized prices (Wall Street Journal 1985). The federal government becomes locked into providing highly subsidized services at increasingly high costs to prop up and maintain an inefficient, uneconomic urban infrastructure at the expense of more productive national investments in other sectors of the economy or more equitable spending in other parts of the country.

As some primate cities have increased greatly in size in just the last decade or two, they may no longer be the economic engines of their countries but economic burdens as well.

SUMMARY AND CONCLUSIONS
The population explosion in Latin America, propelled by high birth rates and lowering death rates, is seen most acutely in the rapid urbanization of the region. For most countries, the apex of the urban system is the primate city which dominates the cultural, economic, political, and social life of the nation. Most Latin American countries have a primate city. Considering that it was a geographer who originated the concept, and its importance to our understanding of urban development in the region, it is curious that the topic has been given only lip service by geographers.

The findings of this study do not confirm the widely held belief that primacy is a stage of the urban-size distribution associated with less developed countries and that, with higher economic development, the city system evolves into a rank-size distribution. However, a longer time span, more cities and better measures of primacy are needed before a more definitive conclusion can be drawn.

A number of Latin American countries like Mexico have moved belatedly to reduce the growth and dominance of their primate cities (Riding 1985). Population planning is now more widely accepted, so that lower birth rates and slower population growth may result in less pressure being put on the cities to accommodate future growth. Mexico has tried to divert some of the growth out of the Valley of Mexico to such satellite cities as Puebla, Toluca and Cuernavaca. The Argentine government has announced plans to eventually move the capital from Buenos Aires to Viedma in northern Patagonia. These moves may be too little and too late. Many Latin American countries are so burdened with debt and beset with low commodity prices that their weakened economies are unable to make much investment to implement these policies.

More research needs to be done on how primate cities actually function, the extent to which specific activities or urban functions are dominated by the primate cities, and what changes are evident (Stetson 1976). More conceptually sound and precise primacy indices are needed. The current indicators--primacy ratios and percent of the urban population in the largest cities--are crude surrogates and may be misleading especially the latter measure. The concept of primacy is a topic rich in unanswered questions, controversies, and major policy decisions. Thus far, the concept has drawn the most attention from anthropologists, demographers, economists, and sociologists. The perspective of the geographer has been missing, especially in Latin America.

REFERENCES CITED
Alonso, W. 1980. Five bell shapes in development. Papers Regional Science Association 45:5-16.

Berry, B. J. L. and F. Horton. 1970. Geographic perspectives on urban systems. Englewood [end p. 77] Cliffs, NJ: Prentice-Hall.

Berry, B. J. L. and J. Kasarda. 1977. Contemporary human ecology. New York, NY: Macmilllan.

Browning, C. E. 1962. Primate cities and related concepts. In Urban systems and economic devellopment, Conference on Urban Systems Research in Underdeveloped and Advanced Economies; 16-28. Eugene, OR: University of Oregon School of Business.

Browning, H. L. 1958. Recent trends in Latin American urbanization. Annals of the American Academy of Political and Social Science 316:111-12

Davis, K. 1959. The world's metropolitan areas. Berkeley, CA: University of California.

__________. 1969. World urbanization 1950-70, vol. 1. Berkeley, CA: University of California, Berkeeley, Institute of International Studies.

El-Shakhs, S. 1972. Development, primacy and systems of cities. Journal of Developing Areas 7(1): 11-36.

Gravier, J. F. 1947. Paris et desert Françias [Paris and the French desert]. Paris: Le Portulan.

Hall, P. 1984. The world cities, 3rd Ed. London: Weidenfeld and Nicolson.

Jefferson, M. 1939. The law of the primate city. Geographical Review 29 (2): 226-232.

Linsky, A. S. 1965. Some generalizations concerning primate cities. Annals of the Association of American Geographers 55(3): 506-13.

McGreevay, W. P. 1971. A statistical analysis of primacy and log normality in the size distribuution of Latin American cities. In The urban development of Latin America 1750-1920, ed. R.M. Morse, 116-130. Palo Alto, CA: Stanford Center for Latin American Studies.

Mera, K. 1973. On the urban agglomeration and economic efficiency. Economic Development and Cultural Change 21(2):309-324.

Renaud, B. 1981. National urbanization policy in developing countries. Oxford: Oxford University Press.

Richardson, H.W. 1973. The economics of urban size. Lexington, MA: Lexington Books.

Richardson, H. W. and Schwartz, G. 1988. Ecoonomic development, population and primacy. Regional Studies 22(6): 467-75.

Riding, A. 1985. Distant neighbors. New York, NY: Knopf.

Sheppard, E. 1982. City size distributions and spatial change. International Regional Science Review 7(2): 127-51.

Smith, C. 1985. Theories and measures of urban primacy: A critique (Ch. 6) and Class relations and urbanization in Guatemala: Toward an alternative theory of urban primacy (Ch. 7). In Urbanization in the world economy, ed. M. Timberlake, 871-69. New York, NY: Academic Press.

Stetson, G. 1976. San Juan, Puerto Rico: A case study of the evolution and functional role of a primate city. Ph.D. dissertation, University of North Carolina, Chapel Hill.

United Nations. Department of International Ecoonomic and Social Affairs. 1987. The prospects of world urbanization (revised as of 1984-85). New York, NY.

Wall Street Journal. 1985. Mexico City's growth, once fostered, turns into economic burden, October 4, pp. 1 and 8.

Walters, P. B. 1985. Systems of cities and urrban primacy: Problems of definition and measurement. In Urbanization in the world econnomy, ed. M. Timberlake, 63-87. New York, NY:Academic Press.

Wilkie, J. ed. 1981 and 1986. Statistical abstract of Latin America, vols. 19 and 24. Los Angeles, CA: UCLA Latin American Center.

Wilkie, R. W. 1986. Latin American population and urbanization analysis: Maps and statistics. Los Angeles, CA: UCLA Latin American Cennter.

Zipf, G. K. 1949. Human behavior and the prinnciple of least effort Cambridge, MA: Addison Wesley.

RESUMEN
El sistema urbano de la mayoría de los paises de América Latina está dominada por una sola ciudad,la cual tiene una población desproporcionadamente grande en comparación a los otro centros urbanos y representa el eje principal de la vida cultural, económica. política, y social de la nación. Este estudio analiza el concepto de primacía urbana, el modelo de primacía en América Latina, y los cambios que han ocurrido en la primacía urbana de algunos paises de la región. Se demuestra que América Latina, entre las otras regiones mundiales, se caracteriza por un nivel muy alto de primacía urbana. La mayoría de los países no solamente tienen una ciudad dominante, pero también tienen un nivel de primacía muy alto. Una tendencia muy preocupante es el crecimiento de muchas de estas ciudades, en centros urbanos gigantes, con poblaciones en exceso de seis millones. [end p. 78]