Review of the Cycling Aspects of:
Making Walking & Cycling Safer: Lessons from Europe

by John Pucher & Lewis Dijkstra
Dept. of Urban Planning
Transportation Quarterly V54 N3, summer 2000


Reviewer: John Forester


home                       facilities

1 Scope and Introduction

This analysis concerns only the cycling aspects of Pucher's and Dijkstra's paper. This is because there is an enormous difference between walking and cycling and the present author is an authority only on cycling.

P&D present statistics showing that there is proportionately more cycling in much of northern Europe than in the USA. P&D present other statistics showing that the cyclist accident rates are much lower in those nations than in the USA. This review does not analyze the validity of these statistics.

The purpose of P&D's paper is stated in its title, to: "Mak[e] ... Cycling Safer," and the authors make specific recommendations to accomplish this end. This review analyzes those of the recommendations that are traffic-engineering in nature, while paying only sufficient attention to the other recommendations to provide context.

2 The Failure of General Recommendations For a General Social Program

P&D provide a catalog of recommended actions in a wide variety of fields (urban planning, motoring controls, education, insurance regulations, traffic law, traffic engineering, etc.), while excusing themselves for failing to provide any detailed analysis for any of their recommendations. Indeed, they write: "Data limitations make it impossible to undertake rigorous statistical analysis to isolate the impact each measure has had on safety improvements. Moreover, most of the measures have been implemented simultaneously and in combination with each other, which makes separating out the individual effects even more difficult." In short, the authors not only advance no scientific reason to believe that any individual recommendation either reduces or increases the cyclist accident rate, but they believe it impossible to do so.

The recommendations are derived from a list of differences between Dutch (to use a shorthand word) society, considered in its widest sense, and that of the US, with the recommendations being to copy the Dutch practice. Fair enough: if US society were exactly like the Dutch, it is reasonable to expect similar accident rates. Therefore, P& D ought to be recommending acceptance of their complete package.

However, that is not scientifically acceptable. Any listing of particular aspects of Dutch society that ought to be transferred to convert US society is necessarily incomplete. If you don't know the effect of each difference, then you cannot know whether or not you have listed the truly relevant differences. You may have listed differences that make cycling more dangerous, while ignoring differences that are absolutely critical to reducing cycling accidents, and there is no way to distinguish these.

There is no way to get around this problem without analyzing the evidence for or against particular recommendations, which P&D do not do.

3 Criteria for Analysis

3.1 Safety and Convenience

P&D's paper assumes that the proper criterion for traffic policy is accident rate, sometimes stated on a per capita basis, sometimes on a per distance basis. It is important to remember, without me repeating this for every analysis, that P&D assert that there is no valid statistical evidence to support any of their recommendations. However, if safety is the prime concern, then preventing the activity maximizes safety. The standard criteria are the safety and convenience of the traveling public. Those are the criteria that I will use herein. Convenience has many factors, but a very important one is travel time. Numerous studies have shown that, for necessary daily (or frequent) trips, minimizing travel time is very important.

3.2 Lawful Operation

A very important criterion is the behavior of the cyclist. The standard of behavior used in this analysis is the one of lawful behavior on properly designed roads. These evaluations of P&D's recommendations are based on whether they aid or hinder, protect or endanger, the lawful cyclist using well-designed roadways. That is the current standard, and proposals need to do better than that to earn recommendation.

4 Specific Recommendations

Since bicycle traffic engineering is my specialty, and in which I have some measure of prestige, I analyze several of the bicycle traffic engineering recommendations made by P&D. I quote from their recommendations: "[B]icycle traffic lights, intersection modifications, bicycle streets, bike lanes and bike paths. ... massive and expanding network of bike lane and bike paths ... bicycle streets (the woonerf) which permit car traffic but give bicyclists strict right-of-way priority over the entire width of the street ... streets one-way for cars but two-way for bikes ... bus lanes ... circuitous routing for cars but shortcuts for bikes ... bicyclists to make left and right turns where prohibited for motor vehicles ... advanced stop lines for bikes ... special traffic lights for bicyclists at intersections ... special bicyclist-activated traffic signals at key intersections ... traffic calming ... cycling traffic training for children" (some descriptions condensed).

4.1 Bicycle Paths

The great majority of urban bicycle paths in urban areas, including those in P&D's study area, are sidepaths running alongside major roads. I know of no evidence that these reduce bicycle accidents, car-bike or otherwise. However, the Europeans have started to study these, and several studies have appeared which state that while sidepaths reduce hit-from-behind car-bike collisions, they increase intersection accidents involving turning and crossing. This agrees with my analysis and testing of over twenty years ago that urban sidepaths greatly increase the dangers of crossing and turning movements, which cause the great majority of car-bike collisions. The study by Wachtel and Lewiston shows that, even for the very slow users of sidepaths in Palo Alto, the accident rate exceeds that on the parallel road.

It is possible to use sidepaths safely, and in Europe they may do so. However, the cost of doing so is for the cyclist to ride very slowly and to incur delays at intersections by giving up the right of way to motorists. It is an abuse of the English language to say that a facility is safer when to avoid its greater dangers one must ride at a much slower average speed than is safe on normal facilities.

4.2 Bicycle Lanes

Bike lanes have been officially promoted in the USA for thirty years, and in all that time their advocates have never been able to demonstrate that painting the bike-lane stripe has any significant effect on accidents. P&D assert the same for European bike lanes.

The evidence from the USA shows that bike lanes promote dangerous and unlawful behavior by cyclists and motorists, specifically in motorist-right-turn and cyclist-left-turn errors.

4.3 Special Traffic Signals

P&D refer to two types of special traffic signals, those that respond to cyclists and those that provide special signal phases for cyclists. As for the first, every traffic signal ought to respond to the presence of a cyclist; any that doesn't is unlawful and a defective traffic control device. P&D appear not to recognize that principle.

As for the second, the general European practice is to provide traffic signals that time-separate the movements of cyclists and motor vehicles. Wonderful? Well, no, because they require cyclists to wait until the bicycle green appears. The cyclists must wait because of the dangerous design of the intersections with bike paths and bike lanes. With those designs, it is impossible to safely let cyclists and motorists operate at the same time, because the design of the intersection is such that the design forces the two classes of drivers into collisions. In short, the special bicycle traffic signal phases are an attempt to correct the intersection dangers of bike paths and bike lanes. Without the added dangers created by those designs, the special signal phases would not be necessary and both motorists and cyclists would see more green time.

4.4 Held-Back Motorists (Bicycle Waiting Box)

With these, motorists arriving at signalized intersections on the red are not allowed to go up to the normal stop line, but are held back some twenty feet or so by another stop line. This allows those cyclists who arrive on the red in the bike lane to go in front of the motorists.

Now consider the situation for the approaching cyclist. He doesn't know whether he will arrive on the red or the green phase. His decision may depend on the direction he intends to take at that intersection.

If he intends to turn right, he stays right whatever the signal says and the bicyclist waiting box confers no benefit on him.

If he intends to go straight, he is either taking the lane already, or is somewhat to the right, depending on the width of the outside through lane. If he arrives on the green, he just goes on through, and bicyclist waiting box confers no benefit on him. If he arrives on the red, if there is room to his right he may filter up to the normal stop line, but he won't then move left into the space in front of the cars, and the bicyclist waiting box confers no benefit on him.

If he wants to turn left, he has a problem. If he wants to use whatever benefit the bicyclist waiting space could confer on him, he has to arrive on the red, because it would be suicidal to arrive on the green and then cut across the moving traffic to the bicyclist waiting box. But he doesn't know, early enough, which phase of the signal will be showing when he arrives at it. Therefore, he has to plan not to use the bicyclist waiting box and to move left sufficiently early to reach the appropriate position for a left turn regardless of which signal phase will be showing when he arrives.

If the cyclist feels that he wants to turn left but fears going to the appropriate location for the left turn, then he can go straight across the intersection on the green, whichever way the green is showing, and then across again when the green changes direction. This way is quicker, by half a signal cycle, than using the bicyclist waiting box, and just as safe.

I deduce that this invention has to do with the fear of crossing lanes of traffic to make a left turn. However, even in that context it makes no sense.

4.5 Woonerven

These streets where pedestrians and bicyclists have priority over motor traffic may be very safe when operated at their design speeds, which are 8 mph or less. However, they are extremely dangerous for operation at normal road speeds. The cyclist moving at normal road speed runs the danger of being assaulted by the uncontrolled and undisciplined movements of pedestrians of all ages. Ever been thrown headlong by a child who ran across your front wheel? It's not very nice for the child, either. Because there is no order there is no safety. In short, while woonerven may be very safe for their residents, they are dangerously useless for transportation. That is their purpose, but it is totally irrational to say that because they make transportation extremely dangerous they make cycling transportation safer.

4.6 Two-way Cycling on One-Way Streets

I have not heard that these reduce accidents to cyclists. In the US, one-way streets are commonly in pairs a street apart, so that it's very little trouble, if any, to use whichever one is appropriate. In European cities where such layouts are impossible and one-way streets are layed out to allow motor traffic to operate at all, they are a major inconvenience to all, although less inconvenient than the traffic jams that would otherwise exist. In this situation, providing a bicycle lane in the opposite direction may be a major convenience to cyclists without snarling traffic as would opposite-direction motor-vehicle traffic.

4.7 Bicycle Shortcuts

Where a bicycle path can shortcut a lengthy road route, it can be justified. In general, the shortcut should be so significant that the travel time at the safe, low path speeds is still shorter than the normal road time. Then the cyclist can ride on the path as slow as safety demands and still save time.

4.8 Cycling Traffic Training for Children

P&D praise the existence of German and Dutch cycling traffic training while saying very little about what is taught. Children are taught "to anticipate dangerous situations and to react appropriately ... in special `traffic parks' with simulated streets, intersections, traffic signals, and possible dangers." Such things, called Traffic Towns and Bicycle Rodeos when used in the USA, are, strangely enough, unknown to P&D.

Well, consider what is being taught. I have described several very dangerous situations: bike paths and voonerven being the most dangerous. The correct way to handle the dangers of these facilities is to stay well away from them and ride on the normal roadways instead. If that actually was what is being taught, P&D would have had fits. No, the children are being taught how to ride with some safety when using the inherently dangerous facilities that their system provides. To teach them proper and safe cycling would defy the entire system which these nations have built. Doing so is difficult enough in the USA; to do it there would require rebellion. I repeat: it is possible to use such facilities in reasonable safety if you obey all the restrictions and delays that have been built into the system to try to correct for its inherent dangers, but that destroys much of the incentive for cycling by making it take much longer.

4.9 Traffic Training for Motorists

P&D praise Dutch and German training for motorists in these words: "It is assumed that cyclists will make unsafe (and illegal) moves in traffic. Car drivers are required to anticipate such unsafe moves by careful noting the presence of cyclists anywhere along their route so that they can react quickly to avoid hitting them. This ability to anticipate potentially dangerous moves by cyclists is actually tested in the driving portion of the license exam and can easily result in failure."

5 Analysis and Conclusions

5.1 Unified Analysis

P&D admit that they are unable to demonstrate the value of any particular part of their recommendations. In effect, they are, and have forced themselves to be, recommending the entire Dutch bicycle traffic system. On analysis, each of the important parts of that system (paths, lanes, and the measures to correct their dangers and the training to use them) is shown to be more dangerous than riding properly on normal, well-designed, roadways with the rights and duties of drivers of vehicles. That system, undoubtedly, by their own showing, is built around the societal assumptions that cyclists will operate unlawfully and dangerously, and must therefore be restricted to the kind of facilities that they provide. The training that is provided, both for cyclists and for motorists, is all built around this assumption. P&D are praising the system that is the prime exponent, society wide, of the cyclist-inferiority superstition, yet despite all this evidence, P&D have failed to see that it categorizes cyclists as inferior beings.

5.2 Conclusions with Respect to State of Present Knowledge

Almost all parts of this analysis have been known for almost thirty years, and while it is anathema to many, its opponents have never been able to mount an intellectually respectable challenge. P&D may disagree with this view, as they must in order to have produced the paper that they have, but if so, instead of presenting a position that is directly contrary to present knowledge, they should first have taken up the challenge of demonstrating why their view, for which they have presented no justification whatsoever, should be preferred. For that matter, the referees for this paper should have considered it in the light of present knowledge and sent it back for such justification.

5.3 General Conclusions Regarding Cycling Transportation Engineering

There has been far too much to criticize in this paper to have space to reasonably present the present state of knowledge. I present the following well-supported conclusions. Those who wish to see the supporting evidence should start by reading my book Bicycle Transportation (The MIT Press).

  1. 1: Cyclists fare best when they act and are treated as drivers of vehicles. This is the vehicular-cycling principle and is the basis of most of our traffic law and operating principles. Nobody has yet shown that cyclists who are systematically treated as non-drivers have better transportation (that is, the blend of safety and convenience) than those who act and are treated as drivers.
  2. 2: Cyclists who operate most like drivers have the lowest cycling accident rates in the USA and Britain, by a factor of four or five. Skill in so operating is the prime safety ingredient in the mix of cycling traffic safety.
  3. 3: Cyclists of all ages, even those as young as eight years, can be taught reasonably quickly to operate as competent drivers of vehicles. In fact, by actual bicycle driving tests in real traffic, they ride far better than the average adults around them, provided that in the younger ages they are restricted to two-lane residential roads.
  4. 4: Nobody has yet shown that, in any nation where they have the choice, cyclists who operate according to the cyclist-inferiority superstition ride either better or safer than those who ride properly. In the USA and Britain, they have accident rates about four or five time higher than those who operate competently.
  5. 5: The cyclist-inferiority system, with its facilities, dangerous training and all, was invented by the American motoring organizations to clear the roads of cyclists for the convenience of motorists. They been trying to force this onto cyclists since before 1970. To pretend that this system is something intended and designed for the safety of cyclists contradicts all historic fact.
  6. 6: The first initial official bikeway designs were taken directly from Dutch practice, the same as is praised by P&D in this paper. These were rejected as being so dangerous that the highway departments would be held liable for the deaths and injuries so caused. The fact that the Americans, who wanted to clear the roads of cyclists, rejected the Dutch designs as being far too dangerous for cyclists very strongly questions the motives of the Dutch bikeway creators. Were the Dutch bikeways, with their far greater danger for cyclists, not also created to clear the roads of cyclists? Considering the fact that the Dutch bikeways were created just as Dutch general society became motorized, that question looks as though it has been answered affirmatively.

6 The Real Question

As I have said before, this is no new knowledge. Most of these things have been known for more than twenty years. That brings up the most significant question of all, of special interest to those, like Professor Pucher, who are interested in promoting transportation by bicycle.

Why is it that Dutch people persist in using bicycle transportation when the system is so stacked against them? Why do they still cycle when the system restricts them to slow speed and more and longer delays, disadvantages forced on them by the dangerous design of the system that they are forced to use? That is the question that stares Professor Pucher in the face, yet that is the question that he refuses to investigate. Oh yes, just why does he refuse to investigate that question of so great importance to his urban planning profession?


Second Review of:

Making Walking and Cycling Safer: Lessons from Europe
by John Pucher & Lewis Dijkstra

Transportation Quarterly, Vol 54, No. 3, Summer 2000


Reviewer: John Forester



1 The Bikeway Controversy

Professor Pucher presents the standard, twenty-five year old argument that America ought to install a bikeway system on the Dutch model. The statistics are a bit more up-to-date, but the facts and reasoning are no better than in the past. This bikeway controversy has been studied for almost thirty years, although Pucher writes, in correspondence generated by his release of a preprint of his paper, that he believes that no such studies exist. That is, even though he knows of my work, referring to me as the leader of the opposition and to my book Effective Cycling, in his immediately preceding paper, Bicycle Renaissance.

The bikeway side of the controversy holds that bicyclists are unable to ride properly or safely on normal roads, so that government ought to provide bikeways for their use. Bikeway proponents hold that bikeways can be safely used for urban transportation by persons unable to ride bicycles properly in traffic, a skill they believe is possible only for the elite few. This side is also called the cyclist-inferiority side because it holds that cyclists are inferior to motorists, need protection from them, and must yield to them. Few bikeway advocates admit to any deficiencies in their proposals.

The vehicular cycling (VC) side of the controversy holds that cyclists fare best when they act and are treated as drivers of vehicles. The cyclist who obeys the rules of the road on normal roadways (preferably well-designed ones), rides in reasonable safety, can use the direct routes, and is subject to no more delays than motorists. VC advocates point to a long list of bikeway deficiencies. First, a constant reminder that bikeways were not built for the welfare of cyclists, but to clear the roads for the convenience of motorists. Since no bikeway system can be ubiquitous, those who use it must also have the ability to ride properly in traffic. Therefore, bikeway cyclists must learn both bikeway and roadway cycling. Because typical urban bikeways create more complicated and more dangerous traffic positions and movements than do normal streets, they require more skill to use safely, they must be used more slowly to avoid these dangers, and they require special traffic signals and restrictive laws that produce otherwise unnecessary delays. Furthermore, all of this is unnecessary because practically all of the population can learn how to drive in traffic, and do so once we let them drive motor vehicles. VC advocates back this up with analyses based on traffic-accident statistics, intersection design with and without bikeways, and standard principles of traffic and human engineering.

Pucher presents the standard, twenty-five year old bikeway argument without knowing anything of the contrary side of the controversy.

2 Pucher's Argument from Correlation

Professor Pucher maintains that his paper presents the way to make American cycling both safer and more useful, therefore more popular. "The necessary technology and methods are already available, with decades of successful experience in Europe." (Pucher uses European to designate the bicycle transportation systems typical of The Netherlands, North Germany, and Denmark, but not of the rest of Europe. I will use the term Dutch, as The Netherlands is the most prominent of these areas.) This is an explicit claim that the Dutch devices that he describes lower the accident rate and increase the utility of bicycle transportation.

Pucher states three facts. Dutch urban cycling is based on bikeways. Dutch cycling has a low death rate. The Dutch use more cycling transportation than do Americans. He claims that the first two conditions have caused the third, without advancing any reason to suppose that this is true.

2.1 Defects in the Correlation Argument

2.1.1 Correlation is not Causation

It is a truism that mere correlation does not demonstrate causation. Without other reasoning, there is no reason to believe that the high volume of Dutch urban cycling is caused by either the low death rate or the bikeway system.

Pucher then claims that creating the first two conditions in America will produce the third. Since no causation has been demonstrated for the Dutch case, the claim is empty. Furthermore, it would likely require a very strong causal link to produce Pucher's desired effect against the force of the other differences between Dutch and American societies that Pucher does not consider.

2.1.2 Selection of Relevant Characteristics

There are other differences between Dutch society and American society. The most obvious set of differences are those of urban history. The places where bicycle transportation is most used are pre-automotive cities, whereas most American cities are largely post-automotive. This difference in history produces a host of social differences that are probably very significant in bicycle transportation. However, Pucher does not consider these.

Pucher provides no justification for selecting the set of specific differences that he chose to ascribe as causes, instead of the sets of other differences that he chose to ignore.

2.1.3 Requirement for Causative Analysis

For Pucher's paper to have been scientifically valid, he would have had to provide causal analyses as to the mechanism by which each device produces the result that he presumes it does. He provides no such analysis. Pucher could have written that this correlation ought to provide fertile ground for investigation as to whether, and, if so, why, each device produces the result that he assumes. However, Pucher has not done so. He assumes that the devices that he describes all produce the result that he desires, but there is no scientific basis for such an assumption.

2.1.4 Scientific Method

As further evidence of Pucher's understanding of the scientific method, he wrote in his public correspondence regarding his paper: "If you are going to make so many claims, then PROVE them!" Proof is not available in scientific questions, only in subjects such as logic and mathematics. In deciding scientific questions one must consider the weight of the evidence on each side. The evidence consists of both raw facts and the reasoning and hypotheses to explain them.

3 Contrary Scientific Evidence

3.1 Reliance on Fatality Rate

Pucher's paper considers only the rate of fatal accidents (probably only fatal bicycle/motor-vehicle collisions) as its criterion for bicycle transportation. Fatal accidents constitute only about 1% of bicycle/motor-vehicle collisions, and only about 0.1% of deaths and injuries to cyclists. Therefore, fatalities do not constitute a reliable measure of danger.

Considering that the Dutch bikeway system operates at much lower conflict speeds than does the American highway system, it is likely that in the Dutch system the proportion of deaths to injuries is lower than in America. This would be all to the good if all things were equal, but an evaluation based only on the fatality rate could mask a greater Dutch total casualty rate than the American rate.

3.2 Failure to Consider Speed

The accepted criteria for transportation are safety and convenience. Pucher considers only safety. Convenience is a combination word that largely refers to total travel time for a trip. Minimizing travel time is a large factor in modal choice, particularly for daily trips. One large portion of travel time is distance divided by speed, plus delays. In Dutch cities, cycling at low speed with many delays may not be at much, if any, competitive disadvantage against motoring, because motoring in Dutch cities also has slow speed and many delays, particularly at the start and end of the trip. In America, where distances are longer and motoring faster, unless the cyclist can travel direct routes as fast as he wants to go with few delays, he will choose to motor instead.

Bikeway advocates do not simply ignore the speed issue; they reject it. A bikeway whose dangers limit the safe speed to a low level cannot be safely used at any higher speed. Its users must travel slowly. On the other hand, the safe speed limit of the normal roadway is generally as high as the cyclist can propel himself, so that the roadway safely accommodates both slow and fast cyclists. To counter this argument, bikeway advocates insist that roadway cycling requires high speed and can be done only by "racing" or "professional" cyclists, whom they despise. They claim that bikeways are necessary for the slow cycling that they admire. Their claim is contrary to known fact.

Pucher's argument ignores the speed factor, both in its effect in The Netherlands and its probably greater effect in America.

3.3 Observed Causation

Insofar as the correlation argument goes, that argument cannot distinguish cause from effect. The causal link could, equally well, run in the opposite direction from that which Pucher assumes. That is, the high volume of bicycle use caused the motoring and highway organizations to produce bikeways to clear the roadways of cyclists for the convenience of motorists. Such a motive is carefully concealed, for obvious reasons. However, in the two nations in which the records have been preserved, Britain and the USA, the records show that the motoring organizations tried to impose bikeways over the opposition of cyclists for the convenience of motorists. (Oakley 1977; pp 29, 51, 53, 85-86, 129-130. Forester, Bicycle Transportation; Chap 4, Governmental Actions, Chap 13, Bikeway Controversy). Given that the only two documented examples contradict Pucher's thesis, great doubt is cast on that thesis.

3.4 Current Scientific Knowledge Contradicts Pucher's Claim

Most of the devices that Pucher describes have been investigated over the last thirty years, and those investigations have demonstrated that these devices have the opposite effect of that assumed by Pucher. Pucher's papers present no recognition of such investigations, and his correspondence shows that he is ignorant of them. Details follow.

4 Specific Items Argued by Pucher

In previous discussion of Pucher's paper, he stated that there is no scientific support for the vehicular-cycling principle. In particular, he stated that there was no scientific support for a position opposite to his for the following items, using the following words.

4.1 Ratio of Accident Rates

Pucher's words: "Twice you claim that vehicular cyclists have accident rates that are a fourth or a fifth the levels for cyclists on separate facilities; clearly that needs to be clarified, very specifically stating that you mean well-trained vehicular cyclists, not just any cyclists using the regular road network (which is clearly the vast majority of road cyclists in the real world); and of course, you need to cite the exact extent of the sample you choose, timing, and precise reference. If this is just one or two isolated studies, that needs to be noted, since it means they cannot be overgeneralized."

I have never made the claim that Prof. Pucher has stated, neither for the groups he has stated nor for the facilities that he has stated. I do not describe the select group as "well-trained vehicular cyclists," but as a group whose members are more likely to operate in a vehicular manner than is the general cycling public. Fewer than 1% of the select group had ever had training in vehicular cycling. They had acquired vehicular cycling habits from experience and from cycling with their clubmates. Their accident rates are only 20% -25% of those of the general population. The data come from the USA and Britain, areas where very little of the cycling by the general cycling public is done on separate facilities.

The base data are in the following papers: Chlapecka, Schupack, Kaplan, Cross, Watkins.

The analysis and comparison of these data was first presented in: Forester; Cycling Transportation Engineering, Sec 1.4, Accidents; 1977

The discussion was continued in: Forester; Bicycle Transportation, 1983; Chap 5, Accidents

And again, with reference to the later data, in: Forester; Bicycle Transportation, 2nd Ed., 1994; Chap 5, Accidents

4.2 Ages for Vehicular Cycling

Pucher's words: "And also your claim about people of all ages being perfectly able to learn to be competent vehicular cyclists; you need to explain exactly under what conditions this is true and what your evidence is; you just claim it without citing a source; I am NOT talking about your experience or theoretical ideas on the topic, I'm talking about solid, hard, verifiable, generalizable empirical evidence. I also expect you to show that people of all physical abilities, mental abilities, reflex times are just as safe cycling on normal roads together with motor vehicle traffic as they would be on separate facilities and that all these different groups have the same abilities to become highly competent vehicular cyclists. Not just to claim it, but to PROVE it, please."

Pucher's statement confuses ability to learn to operate in a vehicular manner with the safety of separated facilities. In this section I discuss the ability to learn to operate in the vehicular manner.

There should be no question about the ability of most adults to learn to operate in a vehicular manner. That is demonstrated by the very widespread demonstration of that skill when used to operate a motor vehicle in nations such as the USA, where motoring is widespread. Some small portions of the population have been identified as unable to learn that skill: severely mentally incompetent, or physically limited as in blind or grossly physically uncoordinated.

Some other small portions of the population have been identified as unable to practice vehicular cycling. A person who has lost either both arms or both legs is in no position to operate a bicycle, either vehicularly or not. Similarly a person with extreme weakness.

I have identified a small portion of the population, probably less than 1%, who can learn to operate a motor vehicle with automatic transmission in traffic but not a bicycle in traffic. These persons are so poorly coordinated that they have to keep thinking about how to control their bicycle and, therefore, have insufficient attention to direct to the traffic situation.

It is also quite likely that the very elderly who have never cycled, or never cycled in a vehicular manner, have difficulty in learning a new activity and a new skill. Beyond these small portions of the population, the success of the Effective Cycling courses has demonstrated the ability of most adults to learn vehicular cycling without bothering to make a scientific study of it.

The Effective Cycling Program teaches cycling as a common activity, using the same methods as are used for teaching swimming or skiing, or even motoring. That is, the students practice the skills under real conditions of gradually increasing difficulty, while subject to critical evaluation and correction as to how well they are performing the traffic maneuvers. Almost anyone who can ride a bicycle can learn how to do it properly.

The vital question was, "At what age can children learn to operate in a vehicular manner?", and that has been the subject of documented study.

The standard for vehicular-cycling behavior is that given in Forester; Effective Cycling.

The bicycle driving test method and its score sheets are given in Forester; Effective Cycling Instructor's Manual, and in The Effect of Bikelane System Design on Cyclists' Traffic Errors.

For adult students, part of the training and part of the bicycle driving test (for the classes that I taught in my home area) consisted in making repeated vehicular-style left turns from a 6-lane roadway carrying 40,000 vehicles a day with speed limit of 45 mph. With a minimum pass score of 70% on the Cycling Proficiency Test, class average scores on the final exam were about 90%. There seemed no need to conduct special investigations concerning these students.

For child students it appeared desirable to document the results. For 7th-grade students, unselected, the course was 15 class sessions of 45 minutes each. Of the 48 students riding the final examination, only 2 failed and the average score of the rest was 97%. The study is reported in Forester; Intermediate-Level Cyclist Training Program.

For 5th-grade and 3rd-grade students, the course was 15 hours in class. For 5th-grade students, the most difficult roadway was 4-lane with commercial traffic. For 3rd-grade students, the most difficult roadway was 2-lane with residential traffic. These students were selected only in that their parents wished them to learn proper cycling. For the 20 grade-3 students who rode the final exam, the class average score was 96% with lowest individual scores of 78% and 93%. The grade-5 students achieved similar scores on their final examination ride. The study is reported in Forester; Elementary-Level Cyclist Training Program.

4.3 Cyclist Training

Pucher's words: "You denigrate the extensive bicycling education and training programs in German and Dutch schools; EVERY, I repeat, EVERY German and Dutch school has extensive required training in bicycling and tests that every student must pass. Perhaps you can find a handful of schools in the USA with such programs. The existence of the occasional traffic town or bicycle rodeo in the U.S. is swamped by their omnipresence in virtually every German and Dutch town. ALSO: Children are NOT taught to disobey the traffic laws, and they are also taught to cycle safely on the normal road network, not just on separate facilities. But evidently you are opposed to any and every policy that you yourself did not propose and take credit for. You are WAY OFF BASE in criticizing bicycling education in Germany and the Netherlands!"

Pucher's own words provide the evidence that my denigration is very probably correct. The Dutch system is largely based on bikeways that physically and legally prevent vehicular cycling, although there are also normal, usually low-traffic, streets on which vehicular cycling ought to be practiced. Pucher believes that no children and few adults can learn vehicular cycling skills. This is also the position taken by the German Auto Association, whose film Children in Traffic has been available in the USA. Pucher praises Dutch training for child cyclists. If Pucher knew that the Dutch system taught vehicular cycling, he would have been outraged. Therefore, Dutch training of child cyclists does not include vehicular cycling, but, obviously, it must include the methods required by the bikeway system. However, Pucher also praises the Dutch system as teaching cycling "on the normal road network." Therefore, the Dutch system teaches some non-vehicular system for cycling on the normal roads, which is against the standard traffic laws and principles and probably is the same system that is used for bikeway cycling. If, indeed, the Dutch system did teach vehicular cycling, that would both contradict the official system and create rebellion among cyclists for being so discriminated against.

4.4 Traffic Calming

Pucher's words: "Woonerven are entire districts where peds, cyclists, playing children have equal rights to use roads along with cars, even priority;... a more advanced, more stringent form of traffic calming.... Your arguments against traffic calming are truly absurd; we cite three different studies from four different countries showing very clearly that TOTAL TRAFFIC FATALITY RATES and accident rates plummeted EVERYWHERE when traffic calming was instituted.Your argument that traffic calming leads to dangerous collisions between peds and cyclists is totally disproved by the overall HUGE reduction in both accidents, serious injuries, and fatalities after the introduction of traffic calming."

Pucher simply misunderstands the argument. The subject is bicycle transportation. Pucher praises voonerven as particularly safe bicycle transportation facilities. That is not so: voonerven are residential or recreational facilities (for playing in the street, as Pucher describes them). They are useless for transportation. Vehicular traffic is limited to about 8 mph because of the danger to the playing children at any greater speed. By the same token, bicycling at any greater speed would be dangerous for both the playing children and the cyclists. The conditions are so very dangerous that speed must be limited to about 1/3 to 1/2 of normal road speed. Very few who want to make purposeful trips by bicycle will choose the route through the voonerven.

I give an American example. One organized 100 mile ride, usually held on a Sunday, goes through the cities of the San Francisco Peninsula, through San Francisco, including Golden Gate Park, across the Golden Gate Bridge to Marin County, and return, all on city streets. However, the streets in Golden Gate Park are, on Sundays, closed to motor traffic, making them similar to a voonerf. When the riders are asked to describe the most dangerous part of their ride, the majority of the answers are "Golden Gate Park," because of the chaotic bicycle and pedestrian traffic there.

4.5 Safety of Bikeways

Pucher's words: "I also expect you to show that people of all physical abilities, mental abilities, reflex times are just as safe cycling on normal roads together with motor vehicle traffic as they would be on separate facilities.. Not just to claim it, but to PROVE it, please."

The standard analysis of the effect of separated, parallel bikeways upon traffic was published in Forester in Cycling Transportation Engineering, in 1977. It has been repeated in the two editions of Bicycle Transportation in 1983 and 1994. That analysis applies standard traffic engineering and human factors knowledge to the situation. It shows that separated, parallel bikeways make both cycling and motoring more difficult and more dangerous than does cycling in the vehicular manner. It shows that bikeway cycling at normal road speeds requires that both cyclist and motorist possess greater abilities than humans have.

I know of no test by bikeway advocates in which they have tested such bikeways for safety when used at normal road speeds in the vehicular manner. So far as I know, I am the only person who has performed such a test. The dangers terminated the test after five miles of cycling along streets that I had, before the bikeways, regularly used in the normal manner and hence knew very well. I rode at the same speed and with the same rules for right-of-way that I had used on the roadways. An average of 0.7 miles apart, I incurred car-bike collision situations that I was able to prevent from becoming collisions only by using extremely high skills of traffic awareness and of bicycle handling. The seventh such encounter would have killed me in a head-on collision with a pack of cars going 40 mph, except for dumb luck. That convinced me to stop the test. On the basis of comparing the collisions on the bikeway avoided only by extreme skill or dumb luck against the absence of any events even faintly similar while regularly using the same roads at the same time of day for several years, I estimated the risk ratio as 1,000:1. This account has been published and republished in Cycling Transportation Engineering and in Bicycle Transportation. I repeat: no bicycle advocate has ever published a repetition of this test to demonstrate the safety of the facilities that he or she recommends.

The remaining evidence comes from statistical analyses.

In 1976, Kaplan (p 75) showed that, for club and transportational cyclists, the accident rate on bike paths was 2.6 times greater per bike-mile than their average on roads.

In 1994, Wachtel, Lewiston and Likens showed that even at low cycling speeds the bikeways that I had tested in 1975 produced a car-bike collision rate 1.8 times that for the adjacent roadways.

In 1998 William Moritz showed that sidewalk cycling (which present the same motor-traffic hazards as Dutch sidepaths) have an accident rate 25 times that of major roads without bicycle facilities.

In 1975 the State of California ceased trying to implement its copy of Dutch bikeway designs, Bikeway Planning Criteria and Guidelines, because it feared being held liable for the deaths and injuries that those designs would cause.

About 1978 the Federal Highway Administration gave up trying to implement its bikeway design standard, Safety and Design Criteria for Bicycle Facilities, because the research that it hoped would demonstrate the safety of those designs went awry and showed, if anything at all, the dangers of those designs.

The American Association of State Highway and Transportation Officials, in its Guide for Bicycle Facilities, specifically recommends against bicycle sidepaths of the Dutch type, because of their dangers. AASHTO 1991 gives 8 reasons against sidepaths, and then states: "For the above reasons, bicycle lanes, wide curb lanes or shared roadways may be the best way to accommodate bicycle traffic along highway corridors." AASHTO 1999, after its list of 9 reasons, states: "For the above reasons other types of bikeways are likely to be better suited to accommodate bicycle traffic along highway corridors."

The operational analysis shows that sidepath cycling requires more than human abilities. The test, and the absence of any other test, shows the dangers of exceeding human abilities. The statistics show that paths, particularly sidepaths, have higher accident rates per bike-mile. The governmental decisions to abandon implementations of Dutch-style bikeways show that government has become persuaded of the dangers of such bikeways.

4.6 Efficiency of Bikeways

Pucher's words: "You claim that all the separate facilities and all other pro-bicycling measures in the Netherlands (including traffic calming, bicycling education, motorist education, bicyclist priority measures, special traffic signals, special intersection design, and above all bike paths and bike lanes) make bicycling MUCH, MUCH slower, less convenient, and much more dangerous than your vehicular cycling would be; I repeat, you argue that EVERY individual Dutch cycling policy and measure makes things much, much worse for bicycling in the Netherlands: less attractive in EVERY way, less safe, slower, and less convenient."

Pucher misunderstands the argument. The argument is that the Dutch bikeway system makes cycling much slower because it makes it more dangerous. The special traffic-signal phases that Pucher praises prevent the car-bike collisions that would be produced by the conflicting traffic movements created by the bikeway system. This collision-avoidance benefit is obvious to casual observers such as Pucher. However, they miss the point that if the bikeway system disappeared, so would its conflicting movements and the need for the special signal phases. Then both cyclists and motorists would have more green time and fewer and shorter delays, a benefit to all.

The analysis is straightforward and has been available since 1977 in Cycling Transportation Engineering and developed further in Bicycle Transportation, whose Chapter 14 is titled European Bikeway Engineering and Design.

The discussion of cycling through voonerven similarly demonstrates that because they are extremely dangerous at normal cycling speeds, cyclists must slow down to use them. The discussion of Dutch cyclist training demonstrates that it serves the bikeway system rather than cyclists.

4.7 Weighing the Scientific Evidence

Pucher has repeated the thirty-year-old bikeway argument with no less, but no more, evidence than any other presenter. As with the others, he has presented no causal link to support his conclusion and recommendation. Correlation does not demonstrate causality.

Pucher does not consider the fanciful nature of his reasoning to be a problem. I advance a reason for that. He shows, in his own correspondence about his paper, some of which is quoted above, that he believes that no contrary evidence exists. His attitude is that his arguments are correct because they present all relevant knowledge about the subject. For the claims of non-existence that Pucher specifically stated (and I tried to get him to state more such claims), I have provided summaries of the contrary evidence and the source documents.

I stated years ago that the bikeway controversy is practically unique among scientific controversies, because one side has all the evidence while the other side has none at all. It is a case, purely, of superstition against science. That is not quite correct, because there is some evidence for the concealed argument that bikeways are promoted by the motoring establishment to clear the roadways of cyclists.

That takes us out of the realm of science and into that of politics.

4.8 Popularity and Politics

Pucher's words: "YET the total bicycling modal split in the NL is over THIRTY times higher than in the US, and the total bicycling fatality rate per km cycled is one SIXTH as high. That overall result makes Forester's individual claims about each individual policy utterly impossible. Vehicular cycling, as Forester advocates it, is already possible and legal on most US roads. In short, his beloved vehicular cycling is already possible, and you can see what it has gotten us: less than ONE percent of trips by bike, and a bicycling fatality rate SIX times higher than in the Netherlands. This is the strongest argument of all in favor of the Dutch policies and proof positive that the measures as a whole have been far, far, far, far more successful than vehicular cycling could ever be. Forester CANNOT explain this contradiction of his argument, so he just repeats the question at the end of the critique for someone else to answer, asserting that it must be due to something else. Utterly absurd. HE must explain the contradiction!

"Forester won't directly admit it, but he advocates MAINTAINING cycling as an elitist activity for very well-trained vehicular cyclists who place the main emphasis on speed; he thus excludes the vast majority of the population from cycling: especially anyone who physically or mentally does not have the skills or even the capability or the desire to cycle in mixed traffic on roads; he wants to keep cycling a marginal mode in the USA to maintain the elitist status. I instead advocate cycling for everyone, or at least for a much of the population as possible, as in the Netherlands, where even the elderly make a fourth of their trips by bike (compared to one tenth of one percent 0.1% in the USA). I strongly oppose this ELITIST view of cycling, and think that many policies should be implemented to encourage bicycling by everyone, young and old, rich and poor, men and women, children and grandparents, students and businessmen, etc. Forester's policies would guarantee that cycling remains a marginal mode here in the US.

"Don't think for a moment that I or anyone else will fall for your lazy trick of simply claiming your VC views are correct unless others can prove them wrong. You'll need to do better than that to convince anyone outside your circle of VC devotees."

Pucher has provided no criticism of the documents that demonstrate the falsity of his positions regarding the safety and efficiency of the Dutch bikeway system. (Indeed, he believes that such do not exist.) Neither has he advanced documents demonstrating the falsity of the vehicular-cycling system. As for the elitist aspect of the vehicular cycling method, my work has demonstrated that it is an activity available to all and is within the ability of all but a very small portion of the population.

As for Pucher's assertions about the effect of vehicular cycling upon American cycling, their inaccuracy is easily demonstrated. While vehicular cycling has in most ways been the American legal standard (because the traffic-law experts know that violating the standard driving rules causes collisions), few American cyclists operate in the vehicular manner. Measurements of the operation of cyclists in typical American cycling cities show that the city-average scores on the Cycling Proficiency Test usually are flunking. The studies are discussed under Cycling Proficiency in Bicycle Transportation. One study, The Effect of Bikelane System Design Upon Cyclists' Traffic Errors, is available at The reason that American cyclists have a high accident rate is not that they operate in the vehicular manner, but that they don't, as shown by the low accident rates of the groups whose members are most likely to operate in the vehicular manner.

American cyclists fail to operate in the vehicular manner because American society and government, mobilized by the motoring organizations, for at least 60 years, has been teaching them that operating in the vehicular manner is dangerous and they ought to be operating as if they had bikeways. For the last thirty years, American society and government have been trying to impose Dutch-style bikeways upon American cyclists, for the convenience of motorists. The resistance of American cyclists to this imposition has instigated the scientific work described herein.

5 Bibliography

American Automobile Association; Children in Traffic, a film originally produced by the German Automobile Association

American Association of State Highway and Transportation Officials; Guide for the Development of New Bicycle Facilities; Washington; 1981 to present

California, State of; Bikeway Planning Criteria and Guidelines; Sacramento; April 1972

California, State of; Planning and Design Criteria for Bikeways in California; Sacramento; 1976 to present

Chlapecka, T. W., S. A. Schupack, T. W. Planek, H. Klecka & J. G. Driessen; Bicycle Accidents and Usage Among Elementary School Children in the United States; Chicago, National Safety Council; 1975

Cross, Kenneth D. & Gary Fisher; A Study of Bicycle/Motor-vehicle Accidents: Identification of Problem Types and Countermeasure Approaches; National Highway Traffic Safety Administration; Washington; 1976

Federal Highway Administration; National Bicycling and Walking Studies 1-24; Washington; 1994. Forester's review available at

Federal Highway Administration; Safety and Locational Criteria for Bicycle Facilities; 3 vols. FHWA-RD-75-112 is the research report. Washington; Dated 1975, not issued until 1976

Forester, John; Bicycle Transportation; Cambridge Mass; The MIT Press, 1983, 1994

Forester, John; Cycling Transportation Engineering; Palo Alto; Custom Cycle Fitments; 1977

Forester, John; Effective Cycling; Custom Cycle Fitments, 1976; Cambridge Mass; The MIT Press, 1984, 6th ed 1993.

Forester, John; Effective Cycling Instructor's Manual; Palo Alto; Custom Cycle Fitments; 1977, 1980

Forester, John: "Elementary-Level Cyclist Training Program: Objectives, Techniques and Results"; 1981. Now available at

Forester, John & Diana Lewiston: "Intermediate-Level Cyclist Training Program: Objectives, Techniques and Results". Bicycling Committee, Transportation Research Board; Washington; 1980. Now available at

Oakley, William; The Winged Wheel; The Cyclists' Touring Club; Godalming UK; 1977

Kaplan, Jerrold A.; Characteristics of the Regular Adult Bicycle User; MS thesis, U. of Maryland; FHWA; 1976; Springfield VA; National Technical Information Service

Moritz, William E.; "Adult Bicyclists in the United States: Characteristics and Riding Experience in 1996"; Bicycling Committee, Transportation Research Board; Washington; 1998

Pucher, John, Charles Komanoff & Paul Schimek: "Bicycle Renaissance in North America? Recent Trends and Alternative Policies to Promote Bicycling"; Transportation Research Part A, Vol 33, Nos 7/8, 1999, pp 625-654

Schupack, S. A & G. J. Driessen; Bicycle Accidents and Usage Among Young Adults: Preliminary Study; Chicago, National Safety Council; 1976

Wachtel, Alan, Diana Lewiston & Gayle Likens; "Risk Factors for Bicycle-Motor Vehicle Collisions at Intersections"; ITE Journal, Sept. 1994

Watkins, S. M.; Cycling Accidents; Cyclists' Touring Club; Godalming UK; 1984

Return to: John Forester's Home Page                                          Up: Facilities