Okay, there's a question we're going to be wrestling with a lot at CityMetric. I'm not saying we're going to answer it any time soon, but the least we can do is ask it, and explain why it constitutes a problem. Here it is:
What, actually, is a city?
At first glance this probably looks like a stupid question, the sort of thing you'd get from a four year old, like "What are unicorns made of?" or “Why is the sky blue?” But (spoilers) it's not.
To explain why, it helps to have a worked example or two. London, official figures tell us, has a population of somewhere around the 8.3 million mark. Paris has a population of roughly 2.3 million. So, one might naturally conclude, London is a lot bigger than Paris.
Except, if you look at a different set of figures, it's very clearly not. Demographia has been publishing a statistical guide to the World's Urban Areas for a decade or so now. Its latest edition, published in May, has Paris on 11 million – four times larger than the official statistics and well ahead of London's 10.1 million.
Then you look at this dataset from the official statistical agency of the European Union and the figures get even bigger but the order's swapped again. Eurostat has Paris on 11.9 million and London on 13.6 million.
You don't have to be a geographer to guess that these three sets of figures are working from different definitions. You can probably even make an educated stab at what each set of figures represents. They are, respectively, the population of the region governed by the city's own authorities, of the continuous built up area, and of the larger metropolitan region.
But which of these is ‘right’? The answer is, infuriatingly, all of them – or, if you prefer, none of them.
To most people, most of the time, none of this is actually a problem: no one ever failed to understand the concept of London because they were unclear on whether you were counting Watford or not. But there are two contexts in which it matters a great deal.
One is its impact on the workings of city governments. If you're the official charged with coming up with a strategy for meeting an area's housing or transport needs, say, then it would help to have some control over the whole of that city.
Very often, though, you don't, so infrastructure investment is likely to focus on the city proper: not only is official power strongest there, but its population get a vote in local elections. Administrative borders can thus end up warping the fabric of the city itself. Just look at the apparently arbitrary places where the New York Subway terminates.
The other reason why which this definitional confetti matters is that it poses a massive barrier to anyone who actually hopes to analyse the urban landscape. Professor Geoffrey West is a theoretical physicist at the Santa Fe Institute, who in recent years has conducted a number of comparative studies on cities. “You'd think by now all the urban planners would have an operational definition of a city that you just could look up,” he says, with some exasperation. “Well that doesn't exist."
As a result, the moment you start trying to quantify any aspect of city life – or worse, compare two cities – you run into difficulties. How many people live in Mumbai? What's the size of New York's economy? Which is more densely populated, London or Paris? All too often, the answer is ‘it depends’. "It's a problem that's plagued a lot of our research," West adds. "And it reflects how little work has been done to quantitatively, analytically, and scientifically understand cities."
The upshot of all this is that what a city is, and where it ends, is a surprisingly subjective matter, and any attempt to nail it down is likely to throw up anomalies. All you can do is keep track of which definition you're working from, and make sure you never, ever, compare apples with larger urban zones.
Here, for future reference, is a quick guide to the main ways of defining cities, which we’ll be trying very hard not to mix up.
The political definition of a city. Paris is 20 arrondissements, New York is five boroughs, London is 32-plus- that-awkward-finance-y-bit-in-the-middle.
The advantage of this definition is that it’s nice and straightforward and easy to exert political power over (or find data on, come to that). The disadvantage is that it tends to throw up anomalies: some municipalities exclude large swathes of suburbia; others include sudden patches of countryside. The official definition of London includes North Ockendon, a tiny Essex village outside the M25, but excludes Buckhurst Hill, a contiguous central line suburb well inside it.
The urban area
This, one might think, is the contiguous urban sprawl: the thing you can point to on a satellite photograph, or from a plane by night.
Life, though, is rarely so simple. For one thing, areas of development separated by less than 200m of open space will generally be bracketed together as a single urban area, on the grounds that anything you can cross in under two minutes doesn’t count as ‘the country’.
For another, there are features like parkland and forests to contend with, so sometimes you’ll notice a distinction between ‘urban area’ (which includes suburban open spaces) and ‘built up’ area (which doesn’t).
There is also no single universal definition of what counts as an ‘urban area’. Most countries define it as one with at least 400 inhabitants per square kilometre. In the US it’s 1,000 people per square mile, which is nearly identical; in Australia, it’s 200 people per square kilometre, which isn’t.
The metropolitan area
A city's effective economic footprint, typically measured by commuting patterns. Satellite towns and ex-urbs may not look like part of a city – but since they wouldn’t exist without it, the thinking goes, they should count.
There are two problems with this idea, however. The smaller one is that you end up counting a lot of people who reject the idea they live in a city at all, if only because they made a conscious decision not to (hello, Surrey).
The bigger problem is that it's largely subjective. What proportion of people need to commute into a city to be part of its travel-to-work area is a matter of judgement. And what if a town between two cities contains large numbers of people who commute to both? Is Warrington a satellite of Manchester, or one of Liverpool? Does Princeton belong to New York or Philly?
The consequence of all this is that different countries use different definitions. While agencies like Eurostat have attempted to harmonise these, there's no current data set that does so worldwide.
Here’s an example of all these various definitions at work. This is Paris, c2008:
That tiny maroon nucleus is the city proper. Surrounding it you have the larger built up area (red), the official urban area (orange) and the metropolitan area (yellow). Suddenly, you can understand why the figures for the city’s population seem to vary by a factor of four.
Just to make things more complicated...
While a metropolitan area will typically contain more than one urban area, the same can be true in reverse. Look down from an aeroplane, and you might think that San Diego and Tijuana are a single city. But there’s a whopping great international border between the two, which most of the area’s population aren’t free to cross. So is it one city, or several? How do you define them? As ever, the answer, awkwardly, is “it depends”.
Lastly, some definitions we won’t be using
The bizarre official phenomenon of British city status, an honorific handed out by the queen, which means that Ely (pop: 20,000) and St David’s (pop: 2,000) have the same official status as Manchester. The equally bizarre habit in some parts of the US, of appending the word ‘city’ to any cluster of four broken down sheds and a dog. What of these, eh?
Well, these are silly and we intend to ignore them. So there.
Images: Photo of Pratt's Bottom courtesy of Tom Tired of London, taken from Flickr under a creative commons license; map of Paris adapted from Wikimedia Commons.