Defining Cities and Urbanism (2018 version)

I go back and forth about the importance of defining cities and urbanism. Sometimes I think it is crucial to have a valid and useful definition, and I spend time talking or writing about definitions; other times, I get sick of it and try to avoid definitions—I just want to get on with the research. Of course, when I am in the latter mold, reviewers are sure to complain that I don’t define urbanism. So, I drag out some boilerplate text to insert into a manuscript and hope it doesn’t push the paper too far over the journal’s word limit. I can pull handy text and ideas from a bunch of my prior works, including these: (Smith 1989; Smith 2001; Smith 2002; Smith 2007; Smith 2008; Smith 2016).

Wow, my reaction in looking at that list of citations is, “Can’t Smith make up his mind?” I guess not. My 2011 post, “What is a city? Definitions of the Urban,” is still the most widely read post on this blog (probably from students writing papers for archaeology classes). I also discuss issues of definition in a number of other posts as well:

Defining Cities and Urbanism (Again).  2011.

Cities, Semi-Urban Places, and Definitions.  2012

My views on this topic have changed considerably in the past few years, and it just occurred to me that I had not put them down in writing for a while. I started writing a new book this summer at UrbNet in Aarhus, Urban Life in the Distant Past: Archaeology and Comparative Urbanism, and the process of giving lectures and organizing the book outline required rethinking my approach to urbanism. I started writing with chapter 3, leaving the conceptual and definitional material to write later!

So, here are some of my current ideas about definitions of city and urban. I will organize my thought around four statements or propositions.

     1. There are three primary dimensions of premodern settlements (urban and non-urban); institutions and activities in these dimensions are the main drivers of change in settlements. These primary dimensions are:

a.   Social life and society. This is a complex dimension that includes institutions, activities, and processes concerning life and social patterns within settlements. It includes things like social heterogeneity (wealth inequality, ethnicity, division of labor), households and neighborhoods, markets and workshops, and urban services. Here are two interesting things about this domain. First, none of these things are exclusively “urban” traits that occur only in cities and not in villages or rural areas (see more below). Second, the variation among settlements is so great that I can point to only a very small number of urban universals, or features found in all cities, ancient and modern. I’ve talked about these universals on and off in this blog; one of them is neighborhood organization.

b.   City size and population. This domain includes the areal extent of settlements, their population, and density. Large, and/or dense, populations in settlements, have profound social impacts, as analyzed by Roland Fletcher (1995), me (Smith 2018), and others (e.g., Bettencourt 2013).

c.   Urban functions. Urban functions are activities and institutions based in a settlement that affect life and society beyond the boundaries of that settlement (Fox 1977; Trigger 1972). Urban functions address the regional or macro-regional context of settlements (as opposed to the first 2 dimensions, which describe the nature of settlements themselves).

     2. Other dimensions of settlements cut across these three. They play important roles, but they have less causal impact than the three primary dimensions:

a.   The built environment

b.   Urban meanings

c.   Urban growth

     3. Definitions are tools; their usefulness depends on one’s questions, goals, and interests.

a.   There is no single, “best” definition of city, urbanism, or any social phenomenon. Definitions are tools that allow scholars to accomplish intellectual goals; they are not meant to capture the “essence” or most important features of a phenomenon. (The use of definitions in everyday discourse, or in writing for the public, is somewhat different, a topic I will leave for later).

b.   Standard definitions for premodern cities differ in their emphasis on the primary urban dimensions. As I cover in my 2011 blog post (and will not repeat here), two definitions are most common and most useful in discussions of premodern cities:

                                   i.    Louis Wirth’s sociological definition. “For sociological purposes a city may be defined as a relatively large, dense, and permanent settlement of socially heterogeneous individuals.” (Wirth 1938:8)

                                  ii.    Functional definition. A city is a central place; that is, it is a settlement that provides goods and services for a hinterland that extends beyond the city itself. (Fox 1977; Trigger 1972)

                                  **    Although I have argued in various publications that one or the other of these provides the best definition for archaeology, I now realize I was epistemologically mistaken. Read my lips: There is no “best” definition. (A challenge: can you find the publication where I argue for Wirth’s definition and disparage the functional definition? Ah, the follies of youth!)

c.   If you really need a single definition of city/urban, how about this underwhelming concept: An urban settlement is one that has reached a high level on one or more of the three major dimensions of urbanism. Is that vague enough? Do you sense my ambivalence about definitions? Read on.

       4. Don’t reify “city” and “urban.”

a.   Settlements—places where people live—are empirical phenomena that exist in the world. They have measurable attributes that we can study. In contrast, “city” and “urban” are concepts or categories that depend on one’s goals and definitions. Our understanding will be greatly improved if we focus on settlements, not on cities or urbanism.

b.   When one analyzes settlements using the primary dimensions, definitions of city and urban just get in the way. This is because many “urban” phenomena (e.g., high density, a division of labor, or place-based social complexity) can also be important in small and non-urban settlements. One of the profound and exciting results of the settlement scaling research of the Social Reactors Project (see our website for citations and papers) is that we have shown—quantitatively and with rigor—that this assertion is correct. The same, or very similar, socially transformative attributes of settlement size are fond in contemporary urban systems, ancient urban systems, and village (non-state) settlement systems. These findings do not rely in any way upon defining city or urban; they are empirical results of quantitative analyses of settlements.

I should probably publish an expanded version of this as a journal article.

References

Bettencourt, Luís M. A.  (2013)  The Origins of Scaling in Cities. Science 340:1438-1441.

Fletcher, Roland  (1995)  The Limits of Settlement Growth: A Theoretical Outline. Cambridge University Press, New York.

Fox, Richard G.  (1977)  Urban Anthropology: Cities in their Cultural Settings. Prentice-Hall, Englewood Cliffs.

Smith, Michael E.  (1989)  Cities, Towns, and Urbanism: Comment on Sanders and Webster. American Anthropologist 91:454-461.

Smith, Michael E.  (2001)  Urbanization. In The Oxford Encyclopedia of Mesoamerican Cultures: The Civilizations of Mexico and Central America, edited by Davíd Carrasco, pp. 290-294. vol. 3. Oxford University Press, New York.

Smith, Michael E.  (2002)  The Earliest Cities. In Urban Life: Readings in the Anthropology of the City, edited by George Gmelch and Walter Zenner, pp. 3-19. 4th ed. Waveland Press, Prospect Heights, IL.

Smith, Michael E.  (2007)  Form and Meaning in the Earliest Cities: A New Approach to Ancient Urban Planning. Journal of Planning History 6(1):3-47.

Smith, Michael E.  (2008)  Aztec City-State Capitals. University Press of Florida, Gainesville.

Smith, Michael E.  (2016)  How Can Archaeologists Identify Early Cities: Definitions, Types, and Attributes. In Eurasia at the Dawn of History: Urbanization and Social Change, edited by Manuel Fernández-Götz and Dirk Krausse, pp. 153-168. Cambridge University Press, New York.

Smith, Michael E.  (2018)  The Generative Role of Settlement Aggregation and Urbanization. In Coming Together: Comparative Approaches to Population Aggregation and Early Urbanization, edited by Attila Gyucha. State University of New York Press, Albany. In press.

Trigger, Bruce G.  (1972)  Determinants of Urban Growth in Pre-Industrial Societies. In Man, Settlement, and Urbanism, edited by Peter J. Ucko, Ruth Tringham and G. W. Dimbleby, pp. 575-599. Schenkman, Cambridge.

Wirth, Louis  (1938)  Urbanism as a Way of Life. American Journal of Sociology 44:1-24.

EXTRA CREDIT, #1: Name the artist of the city on a hill image.

EXTRA CREDIT, #2: Name the city of the shantytown image.

Cities, semi-urban places, and definitions

I periodically think about how we define terms like city and urban. There is no final and ultimate definition of terms like this. Definitions are tools. They help us solve intellectual problems and they help us understand the world. A definition that works in one context may not be helpful in another.

As I have discussed here previously (Original post on definitions), there are two major approaches to defining cities and urbanism: the demographic and the functional. Louis Wirth's influential demographic definition uses four features to define cities: permanence, high population, high density, and social diversity or complexity. The alternative functional approach says that any settlement that fulfills urban functions for a hinterland can be called urban. An urban function is an activity or institution in a settlement whose effects extend beyond the settlement. See the original post for more discussion and some references.

Role of 3 factors in definitions of city and urban

In a later post, I discuss how these and other definitions of city and urban vary in the importance they give to three factors: population, social complexity/ diversity, and influence or function. The other day I made up a diagram to illustrate this point. It is a kind of "triangular graph" that shows the relative importance of these three variables (at the 3 points). The two ellipses show how the major urban definitions rely on these three factors. Wirth's (demographic) definition relies entirely on population and complexity, so it lies far away from the functional point on the graph. The functional definition runs from the functional corner up toward the complexity corner. The reason for extending the area toward the top is that urban functions almost always require social complexity in the urban center. For example, a political capital typically requires different occupations and often different social class composition, if only to fulfill the basic operations of the government.

Camp meeting

In our attempts to understand complex social phenomena (like urbanism), sometimes examples at the extremes shed light on broader patterns. For this reason, in my class on ancient cities I always cover case studies whose urban status is the subject of great debate. The European Iron Age oppida is one example (functionally urban, but not demographically), or Chaco Canyon (just about nobody besides Steve Lekson considers this settlement to be urban). These cases help students see just what urbanism is all about.

In a paper I am now revising, my coauthors and I follow a parallel logic. In order to support a larger argument that neighborhoods are universal in human settlements, we examine a group of "semi-urban settlements" to see if they have neighborhoods. (the answer is yes in all cases except disaster camps). I talk about this study in a previous post. Our paper got a judgment of "revise and resubmit" from a journal, and one complaint of the reviewers was that we didn't define the term "semi-urban" very well. So I've been thinking about how these settlements relate to the standard definitions of urbanism. I made up a second triangular graph to help me understand these settlements.

Two categories of semi-urban settlement in the definition triangle

Actually, there are two very different kinds of semi-urban settlements, each with different dynamics of change (and different neighborhood processes as well). The first category I call "voluntary camps." These are things like religious camp revival sites, festival sites (like Burning Man), RV camps, and the the various urban "Occupy" campsites from last year. I am fascinated by these settlements, and I am convinced that they can teach us much about cities and urban dynamics. In terms of defining this category, the main traits are that these are limited-purpose settlements, rapidly settled, temporary, and voluntary. On the triangular graph they lie close to the Population corner. They do not have urban functions, and they may or may not have social diversity.

Japanese-American internment camp

The second group of settlements classified as "semi-urban" are those that planner Kevin Lynch called "the city as practical machine." I've talked about these previously. This category includes things like military camps, company towns, internment camps, refugee camps and disaster camps. The key feature of these sites is their top-down design and establishment in order to fulfill a specific activity within the larger society. Their specialized activity, whether confinement, economic exploitation, or something else, can be considered an urban function, so I place these settlements at the function/influence corner of the graph.

So, it turns out what one reason we had trouble coming up with a nice succinct definition of "semi-urban settlement" is that this category actually includes two very different types of settlement. But both are urban-like in some ways but not others, and few would be classified as "cities" or "urban settlements" on their own. But they are all part of the Wide Urban World, and they can all teach us much about cities and urban processes.

Cahokia: "America's Lost City"

"America's Lost City" is the title of a very nice article on Cahokia in the NewsFocus section of Science Magazine (23 Dec, 2012). The article, by science writer Andrew Lawler, describes current archaeological research at Cahokia and related sites, and he interviews many of the relevant archaeologists(Lawler 2011). This is a great supplement and follow-up to my earlier post, "Cahokia: Native American Urban Center on the Mississippi." The experts interviewed by Lawler seem to agree that Cahokia was clearly an urban settlement. The fact that they make a big deal out of this issue suggests that there must still be those who doubt such a classification.

John Clark, a Mesoamericanist, notes, "If you found this [Cahokia] in the Mayan lowlands, there would be no doubt that this was a city. It would be in the top 10 of all Mesoamerican cities."  That is a good way to describe the situation. We know the Mayas and Aztecs built cities, and no one would question the urban status of this site if it were located in Mesoamerica. But since it is in North America, where ancient native societies have generally been viewed as "less complex" than their cousins to the south, people need more convincing that Cahokia was, indeed, a city.

One interesting topic discussed by Lawler is population size. There are debates between archaeologists about whether Cahokia should be considered a single settlement within a larger area of many settlements, or whether the entire zone (east and west of the Mississippi River) should be considered a single dispersed settlement. To me, this is not a productive argument. How one classifies settlements depends on one's goals, and there is no absolute right or wrong answer here. For some purposes it is most useful to view Cahokia as a single settlement, and for other purposes it makes sense to consider the distribution of settlement over the entire area. But in discussing the issue, Lawler paraphrases Timothy Schilling as noting that Cahokia or the region "did not have the concentrated density of European of Mayan cities" (p.1622). I think he is wrong here, and that Cahokia probably had a higher population density than Maya cities.

Now I don't know the demographic data for Cahokia and its hinterland, but I think the population density within the Cahokia urban center was probably HIGHER than within Maya cities, but the population density of the "Greater Cahokia" region was most likely lower than that of the Maya lowlands.

Maya cities had very low URBAN population densities (even compared to a sprawling modern city like Phoenix):

• Tikal (Maya):    600 persons per square kilometer
• New York City:  9,400
• Phoenix:    1,900

But the Maya lowlands had a very high  REGIONAL population densities:

• Maya lowlands:  180 persons per sq. km
• New York State:   150
• Illinois:   80
• Arizona:  17

The high density of Maya regional populations (how many people lived on the landscape, whether in large or small settlements) is one of the remarkable features of ancient Maya society. I'd be interested to see how Cahokia fits in comparison with these figures. But in any case, there is no doubt that Cahokia was an urban settlement.

For more discussion of some of the issues of how archaeologists (and others) define cities, see some of my prior posts, What is a City? Definitions of the Urban, or Defining Cities and Urbanism (again).

Reference:

Lawler, Andrew  (2011)  America's Lost City. Science 334:1618-1623.

Defining cities and urbanism (again)

I just found out that my post "What is a city? Definitions of the urban," is the most popular post in this blog. Since my views on this topic have been changing slightly, perhaps it is time for more consideration of the topic. The earlier posts contrasts two definitions: the demographic definition (cities are places with lots of people and social complexity) and the functional definition (cities are places whose activities affect a larger hinterland). In Mesoamerica, these opposing definitions have been most commonly invoked in comparisons of Teotihuacan and the low-density Maya cities. This iconic comparison, from Sanders and Price (1968) is informative:

Are these both cities? Teo and Tikal at the same scale

My thinking these days has shifted slightly. I am less concerned now with coming up with complete definitions of city and urban than with exploring the different kinds of features that make up the concept of urban. The different definitions of urbanism (the two I have discussed, and others as well) vary in the weight given to three main features: Population, complexity, and influence. Settlements can be urban-like on one, two, or all three of these dimensions. The one we choose to emphasize depends on our goals.

Population

In the traditional (demographic) definition of urbanism, population is of primary importance -- both the number of people and the density per unit of area. For the functional definition of cities, the population doesn't matter much. Right now, in our project on semi-urban settlements, population is the most important attribute. These are places like refugee camps and internment camps that are formed rapidly, and we are looking to see whether they have neighborhood organization. For our purposes, it doesn't really matter whether these places exhibit social complexity, or influence on a hinterland. What makes them "semi-urban" or city-like is their aggregation of people in one place. Similarly, Roland Fletcher's important work on settlement size (Fletcher 1995) is about the role of population size and population density on human settlement dynamics.

Complexity

My university campus

Social complexity or variation is part of Louis Wirth's (1938) demographic definition of urbanism. This refers to occupational specialization, social classes or wealth variation, ethnic or cultural differences. Large population concentrations do not necessarily exhibit social complexity; large villages are an example. Settlements with urban functions--that is, settlements that influence a hinterland--almost always have some kind of social complexity. If a settlement has administrative functions, then it probably has government officials, bureaucrats of various types, perhaps military personnel--which means it would have social complexity. The same holds for economic or religious urban functions. But can a settlement be socially complex but NOT have a large population or urban functions? This would have to be some kind of self-contained highly specialized installation, perhaps a university campus or a large medieval monastery in a rural area.

Influence

Urban influence: capital city (Addis Ababa)

Urban functions are activities and institutions in a settlement that affect or influence a larger hinterland. This is what I mean by influence. A settlement can be large but have little complexity and little hinterland influence (e.g., a large agricultural village), or it can be complex with little influence (e.g., the college campus mentioned above). This dimension of "urban-ness" is important because it addresses the roles of cities in their societies. Cities are important nodes in a regional landscape, and the concept of influence points to the varying roles they play. So from a general perspective, when I need to define cities or urbanism, I usually point to the functional definition (as in my 2008 book, Aztec City-State Capitals).


Population, complexity and influence capture much of what we usually mean when we talk about concepts of the city or urban settlement. Most definitions of city and urban can be constructed from variations in these three factors. But sometimes we learn more by focusing less on such definitions and more on the individual dimensions. These three factors, and the ways they vary across time and space, are crucial components of the wide urban world

Fletcher, Roland
1995    The Limits of Settlement Growth: A Theoretical Outline. Cambridge University Press, New York.
Smith, Michael E.
2008    Aztec City-State Capitals. University Press of Florida, Gainesville.
Wirth, Louis
1938    Urbanism as a Way of Life. American Journal of Sociology 44:1-24.

What is a city? Definitions of the urban

What is a city? How do we distinguish urban from non-urban settlements in ways that make sense? Does it matter how we define cities, or is definition just a sterile typological exercise? Definition is crucial for comparison. If we are going to compare cities, or parts of cities, then it helps to have a clear and explicit notion of what we are talking about. In my field, archaeology, there are two main kinds of definitions of the city, based on two of the important characteristics of cities. The demographic definition is based on the idea that cities are big places with lots of people, while the functional definition flows from the notion that cities have an impact on their surroundings. Neither definition is “correct” or “best.” Rather, they are more or less useful for various kinds of purposes.

The Demographic Definition

This definition was first codified by sociologist Louis Wirth in his influential 1938 paper, “Urbanism as a way of life” (Wirth 1938).  Cities, according to Wirth, are defined by four characteristics:

1.      Permanence
2.      Large population size
3.      High population density
4.      Social heterogeneity

This sounds pretty good to most modern ears. It certainly fits contemporary cities, although there is always room for quibbling with quantitative definitions (How many people? How much heterogeneity?). To use the demographic definition, one looks at a settlement, makes some measurements, and decides whether or not it is a city.

The Functional Definition of Urban

Although there may be precursors, most modern functional definitions of cities derive from mid-20^th century economic geography, where central place theory focused on the regional distribution of retail market centers. Market centers provisioned a hinterland, and the larger the hinterland (and the more goods and services provided), the more important the center. In these models retail marketing is an urban function—an activity or institution located within a settlement that affects people and places beyond the settlement. Later developments in anthropology and geography expanded the notion of urban function beyond economics to include politics and religion (Fox 1977). From this perspective, the Classic Maya jungle cities can be considered urban because their kings ruled city-states larger than the individual settlement, and their temples were the focus of worship for peasants as well as urban dwellers. From the demographic perspective, on the other hand, the Maya centers were not big enough to be called cities. To use the functional definition, one cannot simply look at a settlement and decide whether it is urban; one has to look at the entire regional context, including the hinterland and other nearby settlements. If the settlement in question was the setting for people and institutions that impacted a larger realm, it can be considered an urban settlement.

I generally prefer the functional definition, and I’ve argued in its favor on and off over the years (Smith 1989, 2008). I don’t feel that it is inherently better than the demographic definition, only that it contributes more to advancing research and understanding in some of the topics I work on. A good example is the Classic Maya, the urban status of whose capitals was the subject of a lengthy debate. In a series of publications starting in the 1960s, William Sanders applied Louis Wirth’s demographic definition of urbanism to ancient Mesoamerica. In his perspective, Teotihuacan (over 100,000 population; density ca. 5,000/km sq) was a clearly a city, but Tikal and the other Maya centers were not sufficiently big or dense or socially complex to be called cities. If the goal is to analyze how large dense cities differ from smaller more dispersed settlements, then the demographic definition makes sense. The urban experience must have been radically different for the residents of Teotihuacan compared to the residents of Tikal or Copan.

On the other hand, if one is interested in how the Maya centers affected their hinterlands, or the nature of their political, religious or economic influence, or how the inhabitants of the Maya landscape were linked together by polities or market systems or religious communities, then the demographic definition is of little help. The functional definition of urbanism acknowledges Tikal and Copan as urban centers, and this helps us understand how Maya society worked on a regional scale.

One of the things I like about the functional definition of urbanism is that it allows us to talk about different kinds of cities: political capitals, religious centers, economic centers, and other types. Sometimes the type of city differs among urban traditions. Swahili cities were trade centers, whereas most Maya cities were political-religious centers. And sometimes a single society has different functional types of cities. Just as the United States today has such diverse cities as Washington, DC, New York City, and Miami, so too did Aztec central Mexico have an imperial capital (Tenochtitlan), ritual centers (Malinalco), and city-state capitals (many examples).

Definitions help orient us in a field, they aid comparison, and they help us understand urban societies and their transformations. There is more to say about the topic, which I will return to in a future post.

References

Fox, Richard G.

1977    Urban Anthropology: Cities in their Cultural Settings. Prentice-Hall, Englewood Cliffs.

Smith, Michael E.

1989    Cities, Towns, and Urbanism: Comment on Sanders and Webster. American Anthropologist 91:454-461.

2008    Aztec City-State Capitals. University Press of Florida, Gainesville.

Wirth, Louis

1938    Urbanism as a Way of Life. American Journal of Sociology 44:1-24.