Average street length

Consider cities around the world, from small to mega. What is the distribution of street lengths, as a function of population and density? How long is a typical street?

I think you’ll need to give some parameters. Do you count every change of name as a new street? What about East Main that continues as West Main on the other side of the center of town.

Where I grew up outside of Cleveland, the main streets tended to keep their names through adjoining suburbs and even the street numbers continued. Here in the East, what looks like a single street often changes it’s name at town lines.

How do you define the endpoints of a street? For instance, around here, Detroit Avenue runs through several western suburbs and exurbs, except that in the more westerly ones, it’s Detroit Road. And even out past the suburbs, it still continues, except then it’s State Route Something-or-other. On the other end, it heads into metropolitan Cleveland, but then in the middle of Cleveland it crosses a bridge over the Cuyahoga River, and turns into Superior Avenue, and I’m not sure what its fate is out on the East Side.

So do you count its length starting from Detroit the city (where it eventually heads), from the Ohio/Michigan border (where it becomes an Ohio state route instead of a Michigan one), from Lorain County (where it takes on the name “Detroit”), from the Lakewood/Rocky River line (where it changes from “Detroit Road” to “Detroit Avenue”), or from the Cleveland city line? And does it end at the Cuyahoga, or is Superior the same street?

Also, how do you do the averages? If a city contains one twenty-mile street and ten one-mile streets, does that make the typical street length one mile (because there are a greater count of them), or 20 miles (because there’s more total length of long street)?

Woo-hoo, Cleveland simulpost!

Good question. I don’t care about name changes or even municipal boundaries, just actual street geometry.

ETA I hadn’t thought about bridges; are they that common that they would skew the statistics? To be consistent, I suppose a street that runs across a bridge is still one street.

In much of the third world,streets don’t have names. In some cases, they are named on google maps, but local people don’t know those names. About half of them cannot accommodate an automotile.

In modern US subdivisions, there are huge numbers of streets that have only a couple of houses on them, connected to the main city by a street fifteen miles long.

In the US, the states have highly variable municipality definitions. In some, every county road has a name, which may or may not be the same as the part of the road that lies within a city or town.

Im going to take a wild guess and say the average named roadway is a mile long, and leave it to someone else to gather and present evidence that that is not a good approximation. There – that was easy.

I mentioned the distribution since that’s more information than a single average, however we define it. In my mind, a typical street should be one that actual residents most frequently encounter (these would be typical residents, of course). I should think that a nice dense city has more than a dozen streets, so maybe when we quantify things the precise averaging method will not make much difference.

What about highways? Is US 80 a 3000 mile long street? If you don’t want to include highways, how do you distinguish them from “streets”?

I hope we can distinguish interstate highways from streets by the fact that they are less numerous. If we must in order to get a well-defined question, we can cut everything off at some agreed-upon urban boundary.

I’m not following that. Given any one particular street, how do we decide whether it is a highway or not? Saying highways are less numerous than streets doesn’t help answer that.

I’m not following this either. Are you saying that a street which crosses city limits doesn’t count as a street by your definition? That would exclude a huge number of things that almost everyone would agree are city streets. And you just said earlier in this thread that you don’t care about municipal boundaries, just street geometry.

As another problematic example, take El Camino Real which runs down the San Francisco peninsula from San Francisco to San Jose, about 50 miles, and passes through over a dozen different cities. Is that one street?

“Well defined urban boundaries” can be difficult, too. There are areas where there is substantial buildup in unincorporated areas of a county, and named streets continuing out into those areas should probably be considered. There are also cities that have official boundaries clear out into the boonies.

As has already been alluded to, something can be labelled a “street” out in the middle of nowhere. If you play the “geoguessr” game, you will find a lot of very rural areas with something that appears to be a county road labelled “5th street”, or something like this. And it may be an extension of a street from a very small town a surprising distance away.

Another one - are you going to include streets with discontinuities? For instance, Mary Avenue in Sunnyvale which ends on one one side of 280, then picks up again about 1/4 mile away on the other side of the high school and 280 off Homestead Road:

Yes, it becomes “South Mary Avenue”, but we’ve already sort of established that directional prefixes don’t count.

Another odd thing to consider. Something like this junction of Mathilda and Saratoga-Sunnyvale, also in Sunnyvale:

By any reasonable measure, Mathilda is a continuation of Saratoga-Sunnyvale. Most of the lanes sweep around a gentle curve and present you with a traffic light that you go straight through as you transition between the streets. Even though there is a much smaller continuation of the street that is labelled “Saratoga-Sunnvale”.
BTW, the Bay Area is one of the worst places for streets changing names whimsically. A few blocks south of there, it becomes “N DeAnza Blvd”, and in another few blocks it becomes “S DeAnza Blvd”.

Suburban streets are usually just a few blocks long. The longest street is the main traffic artery leading in/out of the development. It may be a half mile.

Hah, now you are in my part of NE Ohio, this is where I grew up. Let’s try for a Cleveland hat trick. Detroit Rd becomes SR 254 until in Sheffield Township at Rt 57 where it becomes North Ridge Rd. When it crosses into the city of Lorain it becomes Cooper Foster Park Rd. C-F-P Rd jogs at Baumhart Rd in Brownhelm Township and ends at Vermilion Rd, but historically it dropped down the valley and crossed the Vermilion River then petered out into Gore Orphanage Rd. - a twisting country lane. Never made it to Detroit. Rt 6 (just north of there) does however and the entire journal may have been called the road to Detroit. Looks like going north from C-F-P Rd to Rt 6 via Vermilion Rd would have been the best route as Rt 6 has always had a good bridge over the river.

The end point of SR 254 has been debated and changed several times over the past century as local jurisdictions and the state have battled it out. Presumably to make the other guy pay for maintenance. It is currently under discussion again.


Around here, the street that’s most frequently encountered would probably be I-90, at a few thousand miles long, because while most folks don’t usually go on it, everyone goes on it sometimes. The street any given person encounters most often will be the one they live on, but that’ll be a different street for everyone.

I did specify “city”, so I hope to avoid small settlements and towns. They would have to be substantially built up. As for cities extending into the boonies, it is anticipated that there will be variations in density: Berlin has 4100 persons/km[sup]2[/sup] and Lagos has 7941.

I make no attempt (yet, anyway) to decide; the fact that they are less numerous simply means that if there are a couple of “streets” that are 3000 miles long, they will not screw up our reckoning since most streets will not be so long.

I meant, imagine you have got satellite photos and you cut out with scissors a New York region, a Manila region, etc., and mark the roads on each one. That does violate cutting things off at boundaries, but it is at least a definite procedure that would effectively produce some data without having to worry about which streets are highways and what the names are. In fact, a computer could do this, starting from digital photographs and producing a tangle of lines which can be automatically counted (there is still the issue of whether a Y-shaped feature represents the intersection of two streets or three…)

It is one street according to my original definition, but I hadn’t thought too much about the fact that the same street would feature in multiple cities, except to reason that such streets are less numerous (like highways).

I think you underestimate the problem of cities with non-regular plans. Check out Dublin, for instance; it’s a knot of thoroughfares that developed organically and that run into one another at all kinds of angles and in all kinds of combinations. Or London. Or Rome. Or Nanking. Or Tokyo. Or most cities that are much more than about a hundred and fifty years old. I don’t think you can look at a city plan and separate the thoroughfares into clearly identifiable distinct streets. Does a particular street stop at a particular junction or continue through the junction along one of the other thoroughfares that meet at that junction? You can say that fairly easily if the city has regular grid pattern, but for other cities you will need information not shown on the map about the social geography of the city to say how the lines on the map function as streets in relation to one another.

Even for cities with a regular grid pattern, how long is a typical street is going to be largely a function of the physical geography of the city. The length of the streets named “Street” in Manhattan is substantially determined by the east-west dimension of the island; the length of the streets named “Avenue” by its north-south dimension. As between cities, I’m not sure that “average” or “typical” street length have any great signficance.

The original question is a bit like asking how long is a piece of string?

However, from an urban morphology standpoint, I feel confident in saying that cities with the longest average will be ones like Chicago, Denver, and Los Angeles, that were laid out on uninterrupted plains, with a gridiron pattern that was added onto in a regular fashion. Medieval cities will have shorter streets that were added as needed, only a couple hundred feet at a time. Subdivisions from the mid-20th century will also have shorter streets, often in a dendritic network, as that became the favored practice in the auto age. Long continuous arterials, whether called Karl-Marx Allee or Yonge Street or I-90, are present in most cities in about the same amounts, and don’t really change the average much.

Figure an urban parameter. Assume, for instance, that a “block” is a twentieth of a mile (4 chains). Based on that, establish a lower limit: say, if there are less than 3 doors (main entrances to homes or businesses) per block for more than 5 blocks, you have crossed the urban boundary. Even if the next town is half a mile away and a suburb of the last one (or vice versa), you would still call that a boundary. The likelihood of a single road bed maintaining an urban condition for a hundred miles would be pretty low.

Another idea I was toying with was to ignore political borders and define metro regions by population density, so ≥40000 people/km² defines a (possibly disconnected) region within which we could say something about the snarl of streets, then see how that changes (longer roads) as we extend to 20000, 10000, 5000, and so on.

This might go to the density of the road network, but the length of the roads will be determined by the size of the area defined by your population density criteria. (As in, a road which goes from one side to the other of an area 100km across it itself 100km long - more, if it does not take a straight line route, but certainly not less.)