An undertaking undertaking

I answered this question on the Quora question-answering site.

Who is at fault if a driver undertakes you from behind you while you are safely making a left turn on a one-way road?

First, let’ define “undertaking”. That doesn’t mean that the vehicle is a hearse. It means that the vehicle passes between yours and the edge of the road when you are near the edge and turning toward it. Since the question mentions a left turn, the questioner is probably in a country where traffic keeps left by default and normally passes on the right. The term ‘undertake” is more commonly used in the British Isles than in North America in any case. But the question could apply to a left turn from the left side of a one-way street anywhere.

Usually, the driver who undertakes is at fault. Generally there is not room for another dual-track vehicle to pass between the turning vehicle and the edge of the roadway, though there may be room if more than one lane allows left turns, or room for a single-track vehicle (motorcycle, motorized bicycle, bicycle). A driver might also undertake on a highway shoulder or by driving off the road. In any of these cases, it is illegal.

Some countries place a bicycle lane or barrier-separated bikeway in the path of the turning traffic, and if the undertaking driver is a bicyclist, the turning driver is held at fault. The resulting conflict may be not mitigated at all, or mitigated in any of several ways — with warning signs, street geometry, street markings or traffic signals. The intention of this arrangement is to relieve bicyclists of the burden of mixing with motor traffic to travel straight through the intersection, and so, placing all of the responsibility for avoiding collisions on the motorist.

Because of the speed with which bicyclists travel, the turning driver is required to look to the rear to the side toward which he or she is turning, rather than only first merging to the lane position for the turn and then yielding to pedestrians who are standing on the corner or walking in the crosswalk. Looking to the rear imposes an additional task burden on the driver at a time where attention to the front and sides is also in demand, and may even be impossible, depending on the geometry of the intersection and the design of the turning vehicle. Unless drivers know to expect this conflict and mitigating factors are in place, this is a risky situation, often resulting in what is called in the USA a “right hook” collision. It also reduces the throughput of intersections by requiring additional waiting — sometimes by motorists, sometimes by bicyclists and sometimes by both.


Some Dutch roundabouts

Dutch roundabouts have received a lot of publicity, notably here:

Roundabout design in the Netherlands has seen a long process of trial and error. A design used until bicyclists complained strongly enough about it placed the bikeway away from the circular roadway, but cyclists were required to yield. Here is an explanation of Dutch roundabout design developments.

The current preferred design places the bikeway away from the circular roadway, and motorists are required to yield, as shown in this video below. That clears up yielding issues.

Here is a video of a roundabout outside the city of s’Hertogenbosch, put forward as an example of good design.

There is a long discussion of this roundabout, among others, on Facebook.

This is a rather large roundabout at the intersection of major highways, and with moderate deflection on entry or exit.  Looking here in Google Maps,  it’s clear that the highway in the background at the left is a bypass around the city of s’Hertogenbosch — though not a limited-access highway like the one which appears in the distant background in the video.

This roundabout was constructed in connection with the new bypass road around the city. Google Street View from 2009 shows the roundabout under construction. A sidelight on this observation is that Dutch practice does consider motor traffic. Two of the legs of the intersection at the roundabout are new roads being constructed at the same time.

I’ve been told by a knowledgeable person that the  bikeways on either side of the highways are supposed to be one-way, but the only destinations along these bikeways are at intersections — reducing the temptation to ride opposite traffic.

The design requires a lot of space because the circular bikeway is  much larger than the circular roadway. The roundabout  is outside a city, but nonetheless, it appears that several houses had to be demolished or moved to make way for this roundabout.

The installation here  places separate bikeways (red asphalt) and walkways (paver blocks) outside the circular roadway. Bicycle traffic shown in the video is light. If bicycle traffic were heavy, it would result in  congestion of motor traffic because motorists yielding to cyclists could not enter or exit the roundabout. Having a path (or for that matter, crosswalks) around the outside of a roundabout obviates the main advantage of the roundabout, that traffic can keep moving. Only grade separation would avoid this for both bicyclists and pedestrians. Motor vehicles and bicycles sharing the roadway would avoid the bicyclists’ causing congestion, but would not be as attractive for bicyclists lacking in skill and confidence..

If you look at the video full-screen, you can see a number of details which are not evident in the small window on this page. I am most interested in the interactions and negotiations for right of way, which are the central issue with mobility and safety in any intersection which is not traffic-signal controlled.

Expectation in the Netherlands is that motorists will yield wherever they see shark-tooth markings. The path around the outside of the roundabout is brought out to the entry and exit roads at a right angle and far enough outside the roundabout so that motorists will be able to see approaching bicyclists. Ohio resident Patricia Kovacs has investigated roundabouts in that state and demonstrated that motorists don’t even yield to pedestrians. She has posted some comments about roundabouts on this blog and in the Facebook thread mentioned earlier.

Some cyclists in the s’Hertogenbosch video are shown looking to their right as they pass paths coming in from their right, for example at 0:55 and 2:25, but many are shown not turning their heads to look for conflicting motor traffic. That is to say, they are putting their complete faith and trust in motorists to yield to them, which is a comment on Dutch expectations for motorist conduct. There is an especially stunning example of this at 1:59, where a cyclist powers through an intersection as motorists approach from the left, inside the roundabout, and the right, entering it. However, at 6:07, a motorist stops abruptly at an exit to the roundabout as a fast cyclist comes around from the right.

One cyclist leaves the roundabout on the left side, opposite the intended direction, at 1:38 in the video.  Another is riding around the roundabout clockwise at 2:40 and apparently while talking on a mobile phone.

At 2:34, a motorist is shown slowing to yield to a cyclist who turns right rather than to cross the exit of the roundabout. With no lane changing or negotiation betwen motorists and cyclists, the motorist did not have a way to know which way the cyclist would go.

Cyclists carry various objects in their hands or on the handlebars. At 6:40, a cyclist is carrying something which looks like a hockey stick.

At 7:18 a young woman has a disabled bicycle and is walking.

Now let’s look at some other Dutch roundabouts.

A roundabout inside s’Hertogenbosch, here,  has the bikeway immediately adjacent to the circular roadway, so that cyclists are hidden directly behind — not next to — exiting vehicles. The video shows motorists required to yield to cyclists in spite of this right-hook threat.

Here’s the video of the roundabout. Are the cycling facilities safe, as claimed? Or if safety is achieved here, is it maybe achieved in another way? You decide.

The description of the video indicates that this roundabout is rather new. Its design appears to be restricted by the small available space at an urban intersection.

Some notable interactions:

At 0:20, a car brakes rather abruptly. Shortly thereafter, a motor scooter passes through the roundabout on the roadway.

At 0:30 and again at 0:53, a car blocks the bikeway to allow a pedestrian to cross in a crosswalk which is just outside the bikeway.

Most bicyclists are not paying any attention to the traffic in the roundabout, At 0:45, a bicyclist is looking down at a cell phone, but at 0:50, 1:10, 1:29, 1:53, 2:03 and 2:10,  and a few additional times, bicyclists perform a shoulder check. The one at 2:03 does this while also carrying a cell phone in one hand.

At 1:49 and again at 2:20, there is a motorcycle in the bikeway, waiting along with bicyclists to enter the roundabout, and there is a bicyclist standing over his bicycle, facing opposite the direction of traffic.  It appears that he is having a conversation with the motorcyclist and a couple of pedestrians. They are blocking the crosswalk.

At 2:49, a motorist stops in the roundabout to yield to a bicyclist who does not cross, but instead turns right. The bicyclist gives a right-turn signal, but too late for the motorist to react, and in any case, a prudent motorist would not risk that the bicyclist would go straight even though signaling. The design of the roundabout does not make the bicyclist’s intentions obvious.

At 2:58, a bus barely outpaces a bicyclist through the roundabout. The bicyclist turns right, but the bus driver has no way to know that he will. The bus driver is either very highly skilled at judging the bicyclist’s speed, or reckless. The bicyclist would have had to yield to the bus if going slightly faster and continuing around the roundabout.

Starting at 3:00, several bicyclists enter traveling the wrong way on the bikeway or sidewalk. Some turn right but others pass close to a doorway which a pedestrian has just exited, and a blind corner, and cross from right to left in the crosswalk or bikeway. An articulated bus enters the roundabout and these bicyclists pass behind it. Other bicyclist traveling counterclockwise around the roundabout will have to yield to the long bus, though this occurs outside the field of view of the video.

At 3:45, bicyclists share the bikeway around the roundabout with a skateboarder and motor-scooter rider.

Almost all the bicyclists are pedaling about 40 rpm.

Here’s a roundabout where bicyclists go around square corners:

And a little roundabout with advisory bike lanes at some of the entrances:

In the so-called “shared space” roundabout in Drachten, cyclists share space with pedestrians. The meaning of the term “shared space” is very different here from its more usual meaning, that motorists, bicyclists and pedestrians all operate in the same space.  In the Drachten roundabout, bicyclists and pedestrians share space — as on shared-use paths in the USA — but are strictly separated from motor traffic except in crossings, as in the other Dutch roundabouts. The space around the margins of the Drachten roundabout also serves as a pedestrian plaza.

I’m poking around in YouTube and Google maps. Here’s a roundabout in YouTube — — location not given, as is usual in such promotions, but I found it in Google Maps by searching on the name of one of the businesses nearby: A special feature made the roundabout practical: the buildings are set far back at a 45-degree angle on each corner. The circular bikeway around the outside makes it possible for motorists to see cyclists in order to yield (though motorists don’t always, as the video shows) and greatly adds to space requirements, which already are large for a roundabout. There wouldn’t be room for such a roundabout at many urban intersections.

Here’s a blog post which includes the video just described and others of the same roundabout, and describes different types of Dutch roundabouts.

Another roundabout in Amsterdam is of the spiraling Turbo Roundabout design, with a path close around the outside and scary sight lines which place a cyclist too far to the right to be in view of a motorist exiting the roundabout: and street view, . Traffic signals have had to be placed at the exits to mitigate these conflicts. This is a triple roundabout with a tramway going around the inside, also requiring traffic signals.

The left and center roundabouts in this overhead view, also are of the bikeway around the outside type: but the rightmost one, in a wooded area, is of the newer type.

Dutch roundabouts are  of several types for motor traffic, but the major difference for bicyclists is whether they travel around the outside of the roundabout, or there are grade separations. There are no examples like the small modern roundabouts and neighborhood traffic circles in the USA, where bicyclists share the roadway with motor vehicles.

Here is an example of grade separation:

And here is a showcase example of grade separation — replacing an installation much like the one shown in the first video embedded in this post :

Roundabouts are expensive and take up a lot of space.  Many of the promotions we are seeing of Dutch facilities ignore these limitations and the compromises they exact and/or celebrate the newest and most impressive examples.

Brooklyn Bridge air horn blast

The Brooklyn Bridge, one of the engineering wonders of the world, opened in 1883 and is still in full service. (Highly recommended: David McCullough’s book The Great Bridge…) Traffic on the roadways when the bridge opened consisted of horseback riders, animal-drawn vehicles, trolley cars and a few high-wheel bicycles. The elevated, central promenade of the bridge was designed for pedestrians. With the advent of motoring, the trolley line was removed and bicyclists were relegated to the promenade. The bridge had stairs until 1982 when it was renovated and bicycling activists succeeded in getting them replaced with ramps. The activists were able to point to the stairs’ as an accessibility issue and also that a stampede on them had killed people shortly after the bridge opened.

The promenade is narrow and often crowded. Though there is a line designating bicycle and pedestrian space, neither space is sufficiently wide. Pedestrians like to stand on both sides, to look out.
Prudence and caution are in order. The bicyclist in the video below is exhibiting neither. Bicycling Magazine has more sympathy for him than I have.

Will bicyclists someday be allowed to ride on the roadway again? A good case could be made for barrier-separated bikeways on the bridge, but removal of lanes for motor traffic would increase congestion and face fierce opposition. At one time, elevated trains ran between the ends of the bridge on rails at the level of the promenade. They were replaced by subway lines which run in tunnels under the river and extend farther. A path could be installed where the trains once ran. In fact, the City is looking into this. Historical status of some elements of the bridge is questionable, as much has changed over the years including removal of much that was original. The superstructure of the bridge evolves to meet current needs, demands, preferences and financing, while its underlying structure stands unchanged, a monument to the genius of its creators and the years of hard work and sacrifices of life and health which went into its construction. Onward to the next stage, with an improved bikeway…

Avoiding the left cross

This post isn’t about pointing the finger of blame. If that is to your taste, you can find endorsement of that position in many of the comments on the video on YouTube. But I think that we would rather avoid crashes in the first place, so this post is about avoiding crashes.

The cyclist could have prevented this crash. He missed three cues that it was about to happen. The motorist missed one cue.

Briefly at 0:09 through 0:10 in the video, the car which was about to turn left is visible with its left-turn signal on. The cyclist’s camera saw the car and so the cyclist also could have seen it and the motorist could have seen the cyclist, but neither was looking at/for the other. The car slowed (note increasing gap between it and the SUV ahead of it). The minivan which the cyclist was passing on the right also slowed, leaving a gap for the car to turn left into the driveway. These were additional cues which the cyclist might have heeded. Following the brief interval when the cyclist and motorist might have noticed each other, the minivan screened the cyclist’s and motorist’s view of one another until too late for either to prevent the collision.

How might cyclists avoid crashes like this? While it is tempting to maintain speed in a bike lane when motor traffic to the left is slow or stopped, do not expect that the bike lane somehow makes you immune to incidents like this. Do not pass on the right any faster than would allow you to avoid a vehicle or pedestrian crossing in front of the vehicle to your left. If you can safely pass motor vehicles on the left (though not here on this two-lane road), do that instead.

Say what?

Click on the photo for a larger view. Your comments are welcome.

Cambridge Street at Sudbury Street, Boston, July 25, 2017

Cambridge Street at Sudbury Street, Boston, July 25, 2017

The location as of September, 2016:

July, 2015

October, 2014

September, 2013


The cyclist’s comment on this Youtube video: “This is why turn signals are important. Had she used a turn signal, I would have stayed back and let her turn. But because she didn’t use one, I assumed she was going straight.”

Let’s take a look into the situation.

The car was initially stopped, second in line at a traffic light. Then the light turned green. The cyclist was approaching in the separated bikeway from the car’s right rear, off to the side. As the motorist initiated her turn, the cyclist wouldn’t be visible in the motorist’s passenger-side rear-view mirror. The motorist would have had to turn her head sharply to the right to see the cyclist, but she needed to look ahead to steer and avoid other potential conflicts. Yes, she should have used her turn signal, but again, she was supposed to yield to the cyclist, not the other way around, and the location of the bikeway made it easy for her not to notice the cyclist.

What are solutions to this problem?

* Well, certainly, drivers should use their signals.

* Bicyclists need to be aware of these conflict situations, and it’s best not to make assumptions.

* Bikeways like this create the appearance of safety because they assuage “fear to the rear” but in urban and suburban areas, most car-bike crashes are due to crossing and turning conflicts, including the one shown in the video, the classic “right hook” — and also the “left cross” (car turns left into the path of an oncoming cyclist). This is a two-way bikeway on one side of a street and so it placed the cyclist farther outside the view of the turning motorist, and can also lead to “Left hooks” and “right crosses”. Germany no longer recommends two-way bikeways like this, as the safety record has proved to be especially poor.

* To avoid these conflicts, the bikeway needs an exclusive signal phase when other traffic doesn’t turn across it. But that will result in more delay for bicyclists and motorists alike. This bikeway also crosses driveways where the barrier is interrupted.

* A bikeway in a corridor separate from streets, a bike route on lightly-used streets, ordinary striped bike lanes or wide outside lanes avoid the problems with a separated bikeway.

The location, in Seattle, Washington, USA.

PeopleforBikes Interprets Boulder Data

Here’s a quick review of an article by Michael Andersen of the PeopleforBikes Green Lane Project about the City of Boulder, Colorado’s removing what he calls a “protected bike lane”. I prefer to call it at barrier-separated on-street bikeway, avoiding a value judgment. Let’s see what the article in fact establishes.

graph in streetsblog article

Graph in Streetsblog article

According to the graph (copied above) and numbers in the article, the installation achieved a major reduction in collisions between motor vehicles at the expense of a 2.5 time increase in motor-vehicle-bicycle collisions. The article states that bicycle volume went up by 54%, and so the car-bicycle crash rate went up by about 1.6 times. Most car-bike crashes in urban areas involve crossing and turning movements. Forcing motorists to cross a bikeway to enter a travel lane, and forcing bicyclists and motorists to start turns from the wrong side of each other, make these crashes more difficult to avoid.

But the story gets more interesting if you click on the article’s link to city data. The left pie chart at the bottom of the city-data infographic shows crashes per year before the installation and the right pie chart, crashes per week following the installation. There were, on average, 11.3 car-bike crashes per year before the installation and 3 in 8 weeks, about 20 per year, afterward. That comes out to an increase of about 1.7 times, but the afterward sample is very small (3 crashes) and seasonal variation isn’t accounted for. The comparison has no validity.

Now look again at the graphs in the article. They don’t accurately reflect these numbers. The “before” bar reports about 0.15 car-bike crash per week or 8 per year, not the 11.3 per year in the pie chart, and so the graph shows an increase in bicycle crashes even greater than the numbers would suggest .

So, to sum up, the article reports a reduction in car-car crashes, but a large increase in car-bike crashes — while defending the bikeway as “protected” and failing to note that there isn’t enough “after” data to produce any statistically valid comparison.

Oh, and there’s also this, on the second page of the infographic:

“The bicycle volume increase along the corridor is consistent with the increase the city typically sees when school is back in session.”

The cyclist counts, unlike the crash counts, are robust. About half the increase is attributable to the school’s being back in session, not to installation of the separated bikeway — a point which Andersen neglects to mention.

To sum up:

What does the article say about the safety of the Boulder facility? Nothing. No conclusion can be drawn from the data, but despite that the Green Lane Project shot itself in the foot with a graph showing a large increase in bicycle crashes.

What does the say about bicycle use? Maybe an increase of 20% or so due to installation of the bikeway, though some of that may only have been transferred from another street.

What does the article say about the quality of Green Lane Project journalism? I think that I’ve made my point but you can answer that for yourself.

Alice Swanson fatality, a right hook

Here is the intersection in Washington, DC, where cyclist Alice Swanson was killed by a right-turning garbage truck.

The Street View is from 2009, as close as Google gets to the year of the crash (2008). The big cross street is Connecticut Avenue. The little one before it is 20th Street NW. My recollection is that the garbage truck turned right into 20th Street, and Swanson probably assumed she could pass it safely because it would turn right onto Connecticut Avenue and the traffic signal was red. If you open the Street View in Google Maps and click on the clock at the upper left, you can go to Street Views from different times and see the intersection without a bike lane (2007) and with green paint (2014). The dashed bike lane stripes indicate that motorists are supposed to merge into the bike lane, but many do not and it may not even be possible with a large truck. Note also that parking extends close to the intersection — the last 20 feet or so are no parking, with a fire hydrant.

Daytime running, nighttime blinding

Here’s a sign I just noticed for the first time, on the Minuteman bikeway, outside Boston.

Sing on Minuteman Commuter Bikeway

Until LED lights went on steroids, who would have ever thought that this would be a problem? But then, unregulated industry is ever willing to find an answer to a perceived need.

Do you think that major bicycle manufacturers are straight about safety, any more than, say, auto manufacturers were until their feet were held to the fire by Ralph Nader’s book Unsafe at any Speed and the Federal regulations it led to? (Not that auto manufacturers are straight about safety even now, witness the commercials of cars being driven at insane speeds, but that’s another story).

Latest example: Trek’s campaign for daytime running lights — not a bad idea, but I have some real problems with Trek’s advice.

A light that is visible from a long distance in daylight is blindingly bright at night. Light-emitting diodes have become so efficient that this is entirely possible with a small, battery-powered headlight. Trek markets and sells lights designed for daytime running and improperly designed for nighttime use. Trek claims that these lights are visible at distances up to a mile in daylight. These lights shine into people’s eyes and blind them at night (as with many other brands as well).

This trend is abetted by the “more is better” syndrome. Brighter must always be better, right? There is even a light  marketed under the name “blinder”. I kid you not. Look it up on the Web — you’ll find it.

The round beam pattern of Trek’s headlights is inappropriate for nighttime use. A headlight used at night should have a flat-topped beam pattern. Automobile headlights have one. German standards for bicycle headlights require a flat-topped pattern  (not surprisingly, as a lot of bicycling in Germany is on paths, and German regulators are sticklers about detail). You may read about very fine headlamps from Germany on Peter White’s web site. In the Boston, Massachusetts area, Harris Cyclery sells them, as do other discerning bike shops. (Disclaimer: a write for a Web site which has a business relationship with Harris Cyclery). Lights with a flat-topped beam pattern are available from some Asian manufacturers, too.

Trek’s advice doesn’t avoid mentioning blinding people, and also doesn’t mention the pedal reflectors or ankle bands which are required by law in Massachusetts, or the rear reflector which still alerts overtaking drivers if a taillight quits or becomes dislodged. Trek taillights do not include an integral rear reflector. Laws vary from place to place, but they generally require a steady taillight or rear reflector at night. A blinking taillight alone is hard to track with the eyes.

To use a headlight properly designed for nighttime use as a daytime running light, just tilt it upward a little so it shines into people’s eyes!

Above all though, safety needs to be active, not only passive. “Being visible is key to your safety on the road,” says Trek. Well, yeah, but lights protect you only when there is a clear line of sight between you and the person who needs to see you. Ride to be visible, and that means, among other things, pass on the left, not on the right, and on a narrow, winding country road, merge away from the right edge as you enter a blind right-hand curve.

A more detailed discussion of bicycle headlights is at

A more thorough understanding of how to be safe can be gained by taking a CyclingSavvy course, or a League of American bicyclists Smart Cycling course, Can Bike course (Canada), ROSPA Cycling Proficiency course (U.K.) reading the booklet Bicycling Street Smarts, John Franklin’s book Cyclecraft, John Forester’s Effective Cycling… Disclaimer: I am a CyclingSavvy and League of American Bicyclists instructor, and I wrote Bicycling Street Smarts.

But why do i do these things? Because I want you to be safe, to give a quick answer.






Everyone thinks of changing the world…

“Everyone thinks of changing the world, nobody thinks of changing himself.” —

Leo Tolstoy

3 PM yesterday, I was bicycling on a 2-lane street in Wellesley, Massachusetts when a school bus coming the opposite way stopped a few hundred feet ahead of me, its blinkers flashing. A kid got out and ran across the street past the front of the bus. I rode up to him.

“Kid, people are supposed to stop for school buses, but not everyone does.  So look before you cross the next lane. I was a passenger in a car once where the driver didn’t.”

“Thank you.”

Do you see an analogy to the expectations of cyclists for “infra” to prevent them from having collisions?