All 4 books by Edward Tufte now in
paperback editions, $100 for all 4
Visual Display of Quantitative Information
Envisioning Information
Visual Explanations
Beautiful Evidence
Paper/printing = original clothbound books.
Only available through ET's Graphics Press:
catalog + shopping cart
All 4 clothbound books, autographed by the author $150
catalog + shopping cart
Edward Tufte e-books
Immediate download to any computer:
Visual and Statistical Thinking $2
The Cognitive Style of Powerpoint $2
Seeing Around + Feynman Diagrams $2
Data Analysis for Politics and Policy $2
catalog + shopping cart
Edward Tufte one-day course,
Presenting Data and Information
Tampa, March 1
Boston, March 14, 15, 16
Oakland, April 20
San Jose, April 21
Palo Alto, April 24
San Francisco, April 25, 26
Road and exit numbering


-- ET

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

This is a good graphic about an information design issue: identifying exits (names or names/numbers) and roads (numbering systems). The graphic accompanies an article (about the new numbering of California freeway exits, instead of names only) by Patricia Leigh Brown who writes a lot about design for the Times.

Exit numbering that combines ordering of exits along with the mileage between exits is interesting. The potential difficulty is obvious. I'm not experienced enough in practice to be sure about this method, however.

Illustration: NY Times,February 10, 2002

-- Edward Tufte

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Creating a consistent, coast-to-coast numbering system for highway exits is a worthwhile goal. It is extremely convenient for me to tell visitors, for example, that you will get on at Exit 221 and leave at Exit 284, for a distance of 63 miles. At an average speed of 70 mph (the maximum in my part of the country is 75 mph), this means an estimated travel time of 54 minutes between exits.

Many turnpikes in the East (New Jersey, Massachusetts) number exits consecutively and are unlikely to change. Indeed, the distance between exits is usually not great, and this numbering scheme seems to work just fine.

Part of the accompanying article seemed to suggest that by numbering exits, Californians would lose some of the close association of freeway names and locations. Interstate 10 is both the Santa Monica Freeway and the San Bernadino Freeway, depending on where you are. Given that this can be confusing to visitors, exit numbering can be a great help.

Chicago is famous for its "grid" of street numbering. Beginning at the intersection of State and Madison Streets, the city is laid out in 8-block miles, so that someone at the intersection of 800 North and 1600 West is one mile north and two miles west of the origin (the Chicago Transit Authority uses this example on its Website). The grid even extends out into the far suburbs. Locals oftentimes state where they live on the coordinate plane: "We're at 7200 North, 2100 West," for example. Chicago's system might make an interesting addition to the discussion of Micro/Macro Readings in Envisioning Information.

-- Claiborne Booker (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Naming road-exits by distances exemplifies the use of numbers simultaneously as names and as quantitative data. There is a chapter on multifunctioning graphical elements in The Visual Display of Quantitative Information. Also my Envisioning Information has some examples of names/numbers multifunctioning in chapter 2 on micro/macro designs; see, in particular, the conductivity graphics on page 39 which use numbers to identify and also to order hundreds of research papers on electrical and thermal conductivity.

Here is another example of a navigational helper that uses multifunctioning numbers (source: Sporty's Pilot Shop video on how to fly small planes). The convention is:

airport runway number = the magnetic compass heading divided by 10 and rounded to 2 digits.

Thus the same runway has 2 identification numbers, depending upon the direction you're going. Runway 9 means you are going due east; runway 27 at that airport is the same runway but now you're heading due west. Or if a runway heading is 290/110 degrees, the runway will be numbered 29/11.

Such runway names won't distinguish among parallel runways, which could be unfortunate. So, for parallel runways, add letters to the magnetic compass heading divided by 10; letters such as L C R for Left, Center, and Right runways. Thus 27R (heading due west) is 9L (for those heading due east).

Check out the runway numbers the next time you take off and land.

-- Edward Tufte

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I have extensive experience with the interstate/turnpike exit numbering systems as a former truck owner-operator, and vastly preferred the systems that related the exit numbering to the distance from the western or southern border of the state. The states and turnpikes which numbered in sequence seemed interminable to traverse instead of having a clear sense of where I was and how far to the next state. Instead of being able to make driving decisions and exit strategies based on a clear idea about how far the next exit might be relative to the last, and also in relation to the mile markers that simplified planning, the appearance of exits numbered sequentially often seemed surprising. Also, post-original numbering of exit additions could place exit 13B some distance from exit 13 and make its arrival particularly mysterious. Now, if they would only provide the occasional indication for the east and northbound travellers about the remaining miles in the state so that a driver could know how much of the state remained, not just how much of the state had been driven...

-- Jean Selkirk (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

As a cartographer I find the exit number based on mileage a logical solution to an often random collection of arbitrary numbers and letters. Pennsylvania and Maine are in the process of making such a conversion as a reactionary response to growth, Maine actually has duplicate exit numbers along I-95. Which brings me to another puzzle: the present system of Interstates has followed a logical progression as illustrated in the graphic: Odd numbers I-5 (west coast) to I-95 on the east, I-4 in Florida (south) to I-94 in North Dakota. BUT - The nations most recent Interstate is in Pennsylvania, and it has been designated I-99. It runs parrallel to, and between, 79 and 81. So the most recent interstate has broken the long standing system of logical correlation, which sets a potential precedent to continue. I wonder if #99 was chosen out of sheer desperation - imagine the design confusion as the DOT engineers realize they have run out of 2 digit numbers....

-- Michael Hermann (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

The issue of ordinal exit measurement vs. interval exit measurement also arises in maps of the London Underground. The classic map of ordered stations and a map of geographically accurate stations are both posted in the thread here at

-- Edward Tufte

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

One possible complication to numbering by distance is in cases where exits are significantly less than a mile apart - which I believe does happen in parts of the Northeast. Would one then have exit 23.6, or 23A and 23B? Overall, numbering by distance seems preferable.

-- Heather Lin (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

The numbering of road systems enhances the ease of conceptualizing our position and route in the abstract but comes at the cost of depriving us of much useful information contained within the local names about the geography and relationships of any particular location. So while we have been able to fashion larger systems which are easy to grasp and utilize more easily, there is for every simplifying system a cost to the body of information contained within society about the actual function of a particular road or route. Thus the ways in which we decide to label intersections and routes make the process for any single individual of integrating the information more difficult. As we collaborate on integrating our systems much of the finer grain detail is often lost. Like any large system the ubiquity of a simple conceptual paradigm tends to expunge the myriad more complex individual perceptions of place. And while post WWII there were tremendous gains by our creation of larger systems, we are now on the other side of this equation where the detail obscuring nature of the grid encourages a loss of the complexity of relationship which was the foundation for the discovery of complementary perspectives that provided such rich and comprehensive innovations as so many of us were brought togather within these large standardized systems. So while I appreciate and enjoy the ease with which the grid enhances my ability to move about, I recognize that in my travels if I let this be a predominant conceptual base I will lose the appreciation of the unique character and attributes of any particular location and must take active measures to ensure that I do not begin to regards whole groups of people and economic activitys as simple minded addresses.

The recent anthropology about the mystical religious stories which enabled the south sea islanders to navigate between the islands and how the introduction of the compass sliced up the lore for the new generations is a good example of how a highly integrated system developed over a long time for specific functions can be destroyed by the introduction of a simple reference system. The children are incapable of using the comprehensive lore with confidence because the intricacies of wind and tide, stars, moon, and sun which were previously integrated in the ancient lore have been made untrustworthy by the new knowledge which while correctly illustrating the exact position of themselves in relation to their destination discourages the ability to integrate the lore, much of which uses idiosyncratic and unique identifiers. The names of different types of waves, the descriptions of the wind and smells, and so on. So while one may attempt to get to a location with a map and compass, to get there safely in a small boat may very well be impossible. And to find the particular fish, or anticipate the weather is lost. Similarly, to know that one should take 495 east does not include the information that if there is a traffic advisory, one may also take it west. Let alone the information that depending on which direction one is approaching from will change the choices available. While knowing 495 is the Beltway includes a whole subset of details about ones possibilities.

Old road systems were universally described by the function of the route, or sometimes by the owner of the route. In England they use A and M to designate the type of road, or to get from Nairobi to Mombasa one gets on the Ring road and goes to the Mombasa road. To get from Capetown to Cairo one takes the Great North Road.(Albeit impossible to safely traverse at this time)

So while we enjoy the great improvements in efficiency and knowledge that the larger abstact simplified data points provide us, we must always resist the hubris of believing that data is anything more than a starting point for knowledge because there are always intricate reasons that a location becomes a place.

-- Jon Rock (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

As I recall the road-numbering system runs further: 3 digit roads are either bypasses/loops (if the first digit is even) or spurs (if the first digit is odd); the second two digits identify the national highway to which the road connects. Thus 695 and 495 are the beltways bypassing Baltimore, MD and Washington, DC., respectively, while 195 and 395 are "spur" roads running parallel to I-95 in the same regions.

One other problem with beltways and loops: where do you start the numbers? DC's beltway recently underwent a renumbering when Maryland and Virginia finally agreed on a unified system in which numbers correspond to distance traveled.

There is no number one. As far as I can tell, there is no number less than fifteen, and the numbers run into the forties at least. So maybe the system could still use some improvement.

Of course, 495 and 95 are literally the same piece of asphalt in some portions, so that tends to disorient passers-through.

-- Scott Zetlan (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I remember being shocked when the US did not adopt the metric system many years ago when Ronald Reagan was President. It is an outstanding system and as someone who went through the imperial to metric Conversions in the 1980s in Canada in school (which was sometimes painful), however the metric system is a far superior system because it is easy to figure out (being a 10 based system, rather than a 12 or 16 or whatever imperial seems to use). By having your road exits in miles helps entrench the imperial system which is an old system--(how many feet in a mile anyway?) which is anti- progress. It is like still teaching kids on the QWERTY keyboard which is designed to slow down typing.

-- Bill Paton

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

The graphic nicely shows how Interstate highways are numbered.

I find it interesting to note that U.S. highways tend to follow numbering systems in the OPPOSITE directions (compared to the interstate highways). For example, US 1 is along the east coast and US 101 is along the west coast. US 2 is along the north part of the country and US 98 along the southern gulf.

-- Kevin Wright (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Though I agree with Mr. Paton regarding the danger of tying exit numberings to miles instead of kilometers, I must also point out that the QWERTY/Dvorak example is a rather tired example of the so-called market failure to establish standards. See for an explanation of why.

For an interesting typographic example of using numbers simultaneously as names and as quantitative data, take a look at Concrete Mathematics by R. Graham, D. Knuth, and O. Patashnik (Addison Wesley). In this book, a table of data is labeled according to its page number, which makes it much easier to find and refer to later (in fact, this numbering scheme was inspired by analogy to highway exit numberings). The book is otherwise a beautiful example of how appropriate typography can effectively and efficiently convey complicated material.

-- Jim Jowdy

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Table ID numbers = page numbers is the coolest thing I've heard this month!

Wonderful example. Coming from bright authors: Ronald Graham, Don Knuth, and Oren Patashnik. And this design idea plays a bit on one of the main topics of the book, thinking about sequences.

(For those of you from out of town, the book Concrete Mathematics is about mathematics and algorithms, not about concrete.)

Knuth has one of the most inspiring websites around: See his remarks about email ("Email is a wonderful thing for people whose role in life is to be on top of things. But not for me; my role is to be on the bottom of things."), among many other matters.

-- Edward Tufte

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Concrete Mathematics is delightful for another reason. Knuth collected comments written by students on their handouts and included them in the text. So you'll be reading along at a hard bit and find a comment like "this will all be clear in a second" in the margin. And sometimes very helpful comments too. It's a bit like it must have been like to read those medieval books after all your fellow monks had commented on them. Good fun.

-- Josh Mangum

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Two things occurred to me as I read this discussion:

First, progress is possible. When I moved to Georgia in 1994, the interstate exits were numbered sequentially; three or four years ago, the state changed to the mileage-oriented system.

Second, I have seen, in many states which moved from ordinal to mileage systems, exits numbered like this "267 (Old Exit 5) Smith Road" so that old-timers don't get lost. For twenty years, at least, parts of Pennsylvania maintained the old numbers on exits signs in combination with the new ones.

Third, for anyone who is interested in the subject of how, despite logical planning and good intentions, strange things like I-99 and I-238 (there isn't an I-38 for it to be a ring-road of)crop up, this is a good link to begin with:

-- Kathleen Ryan (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Most of the states I have traveled in use the distance related numbering system for exits. When I have encountered exits less than 1 mile apart (which does occur frequently in the northeast) they have always been lettered, using A for the southern or western most of the exits.

One interesting Interstate to take note of is I-19, south of Tucson, AZ. It is numbered entirely in metric.

-- Jarrod Lombardo (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Ken Wright noted that the US highway system counted E-W highways upward going from North to South (US 20 setting out West from Boston, US 40 from Baltimore, US 50 from DC, etc.), the opposite of the Interstate system, which has I-10 as the southern crossing, and I-80 and I-90 in the north. Well, sure! Think of the chaos if they'd overlaid the new system on the old one, using similar numbers for closely parallel roads.

The Interstate numberings are pretty good, to my mind. The Lower 48 are more-or-less rectangular. The problem arise, as others pointed out, in the Northeast, which has both dense population and lots of natural boundaries. (I've always visualized the US population as a fan, with the northeast as the handle.)

Politics also has a small hand in the numbering. I live in Princeton, NJ, in the only "hole" in the primary Interstate system. I-95 was originally supposed to cross into NJ from PA about where it does now, but then cut north through the Hopewell valley and eventually connect to what's now called I-287. From there it would have gone east to the Turnpike. The locals protested successfully, and the valley was saved from the superhighway. (Didn't keep the 18-wheelers off Route 31 and US 1, 202 and 206, though--and guess what the current generation complains about?)

Until a few years ago, I-95 ended right around the exit at which it would have headed north under the original plan. There, it took the name I-295. 295 loops around Trenton and then parallels the NJ Turnpike back down to Delaware.

At some point, DOT wanted to declare the primary Interstate system "complete." In celebration of this fact (or perhaps as a gesture of surrender), they moved the point at which 95 becomes 295 to the junction with US-1. Now one can actually give people directions without sounding idiotic: to go north, take 95 to Route 1, then Route 1 to Turnpike/95.

But 95 still has a hole in it! And the Turnpike is 95 down to Exit 10, leaving southbound newcomers baffled as to what suddenly became of the Interstate signs...

-- Roger Lustig (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I beleive that the federal government is making "distance" based numbering of interstate exits mandatory for states to continue receiving federal highway funds. You can see examples of this conversion on the Ohio Turnpike. Can anyone confirm?

-- Tony Keck (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I was so excited a few years ago when Alabama started putting up Km markers. Not that I'm some kind of ISO fanatic, but it smacked of progress in a place where most smacking is of some other variety.

Within the next year, they started coming down. Apparently the Alabama DOT found itself complying with an FHA mandate that had been overturned by subsequent Federal legislation. Given our short history of cooperation with a Federal government and severely overdrawn state budget, bureaucratic failures like these are painful to swallow.

-- John Morse (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

The discussions in the choice of numbering all identify cases in which the approaches fail. They concentrate on the problem of density as a limiting value. They do not consider the more serious case of the realignment of a road with distance based numbering (perhaps a tunnel under a major obstacle) that changes its overall distance. The subsequent exits would then require wholesale renumbering.

These problems are common in information systems and are a result of attempting to consolidate unrelated information. A separation of these concerns can be seen on road in many countries, staring with the Roman practise of placing markers to indicate distance. Road exits in turn are then numbered - or named as desired - for identity.

Maintaining historical content - like "267 (Old Exit 5) Smith Road" - promotes continuity while allowing routine administrative changes to proceed without a public outcry.

A later edition of Constructive Mathematics (expanded, or with a change to paper size) could result in tables with wildly different captions. This makes references edition specific and confusing to users without insight to the 'elegant' solution for locating them in the original.

Embedding information from the underlying data model into references will result in an eventual mismatch after unexpected changes occur in the real world. The only certain way to avoid this is to make identifiers devoid of information.

The accepted wisdom offers an explanation to the problem - not a solution. In practise, it is necessary to choose between the evil of a data-laden key and the lookup required for a value-free identifier.

-- Pekka Pihlajasaari (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I-99 in Pennsylvania, as mentioned above and as you can see from the link above, is numbered outside of the traditional order because a congressman (Bud Shuster) who was chair of the House transportation committee and decided that he wanted a highway in his district, wanted it named after him, and to have a distinctive name. So in 1995 it became the first highway number to be designated by federal law. The road's story says quite a bit about the power individual politicians with the right committee positions can hold over their little domain.

For those interested more information about I-99 or Bud Shuster can be found at or and my apologies for the off-topic comments

-- Patrick McMahon (email)


I just got back from a weekend trip to Maine, where Route 95 has recently been converted from serial numbering to distance numbering. All the exits have little "Formerly Exit ..." tacked to the bottom of them, which looks dumb but remains necessary because as best I can tell there is not yet a single available map that uses the new numbers.

The core argument in favor of distance numbering, I believe, is that it helps you know how far it is until your exit on an unfamiliar road.

On an unfamiliar road, though, how far you have left to go is a second-order question. The first-order questions are "Am I going the right way?" and "Have I made an error since the last time I thought I was going the right way?" The best way to confirm that the driver is on the right road is to put up signs that list the upcoming destinations. These signs also state the mileage until those destinations, thus answering the second-order question as a by-product of answering the first-, without renumbering anything. We already have these signs, thus it seems to me that this renumbering delivers information that the driver already had.

And if there was no reason to renumber, to begin with, there's a really good reason *not* too, which is that distance-based numbering is significantly more error-prone. It produces a larger, more sparsely populated exit-number space that requires additional precision in all communications, and in which there is no way for anyone to deduce which values are valid. (If you think you're supposed to be going to exit 147, and you see exit 141, you have to ask yourself whether you got the number right, because the next exit might be exit 174. And if there's no exit 147 at all, you can't even get off and ask for directions!)

I strongly suspect that a dirt-simple usability test (sit a person in a chair, tell them some directions, then show them paper "highway signs" in sequence and have them say what they'd do) would quickly confirm that serial numbering produces more *accurate* travel.

-- glenn mcdonald (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

It was always my understanding that the exit numbers ( that correlate to distance), are also numbered according to the mile markers that dot the highways. In the case of multiple exits that exist between mile markers letters were added. These are also helpful in the case of an accident, or breakdown of some sort. By using the mile marker that you are closest to with the last exit number you saw, you can determine how far you will have to travel to reach help, and how far an emergency vehicle will have to travel to reach you.

-- Phil Davis (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I remember travelling through France a few years back. We took secondary roads through many smaller towns. They were all directly connected, but most of them had a two-lane ring around them to connect the medieval star system and help traffic congestion on the center. If you entered these rings from the outside, they were always two signs: "Toutes directions" (all directions) and "Autres directions" (other directions). Once you were on the ring, you could look for the city and find your way. It really did not matter whether you went with toute or autre. Saves a lot of money instead of listing all cities on every sign...

-- Matthias Ritzkowski (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Here in the San Francisco Bay area, the exits on highways 101 and 280 are finally being numbered. Geographical naming gets rather silly around here, since it has lead to inconsistent labelling of identical exits. The same exit off 280 in Cupertino is "De Anza" going North but "Sunnvale-Saratoga" going South. I'm just grateful that exits are numbered at all; I don't care if they are using sequence or mileage!

It doesn't stop there: in the San Jose area, one needs to know that "Highway 280 North" goes west but that "680 North" goes east, then north-west, then east. Increasing or decreasing exit numbers can at least tell you if you're heading in the right direction.


-- Mathew Lodge (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Glenn McDonald makes an excellent point regarding the potential for confusion with mileage-based exit numbering systems. That's why I think that exits always should, in addition to a number, have an identifier that is based on a road, place name, or landmark. The signage may be more expensive, but "Exit 147 - Metropolis" is far more reassuring to an unfamiliar driver than "Exit 147."

Placement of exit and other road signs is equally important, however. There's nothing more frustrating than seeing the sign for your exit too late to take it safely.

-- Franklin Tessler (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I have always been amused by the sign on a stretch of highway between Berkeley and the Bay that has both an 880 East and a 580 West escutcheon. The road travels roughly north.

-- Richard C Haven (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

A sketch of an old New Yorker cartoon version of exit signs:

-- Edward Tufte

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

I strongly agree with Franklin Tessler's comment on the placement of signs. If signs are not placed right, they are of little use and may even be dangerous. The problem is particularly acute in cities where information overload is severe.

Sign visibility is also an issue. Faded signs are hard to read unless you are close to them. Have you tried to find an unfamiliar house at night? Many houses have unlighted signs that are invisible at night.

-- Ralph Gillmann (email)

Response to Graphic of the Day: Road and exit numbering, New York Times, February 10, 2002, page 4-2

Franklin also mentioned the inadequacy of signs that give the exit number without the name of the street. A similar problem occurs when the sign gives the *number* of the road, but not the name. For example, a sign that says "Exit 7 - U.S. 23" is not much help to someone who is looking for Parker Street. I presume this happens because the people who choose the text for the signs are used to thinking in terms of route numbers, not street names.

-- Tom Snider-Lotz (email)

Universal Principles of Design and Hicks Law

A good theoretical approach to the design of road/exit numbering signage is presented in many of the theories that are explained in Universal Principles of Design. This particular text takes a look at some of the most well known principles and then applies them appropriately in the design world. Specifically, Hicks Law, which states that the number of choices someone has is directly related to the length of time that it will take them to make a decision, is one problem that drivers face when they see signs with too many options. Interestingly, no matter how you design the sign, it has more to do with how many times the driver is is exposed to the information, for how long, and then, exactly how long they have to make a decisions and execute the actions needed to complete the task.

-- Matthew Gluskin (email)

Threads relevant to design case studies:

Threads relevant to road signage: