The Effects of Bicycle Accommodations on
Bicycle/Motor Vehicle Safety and Traffic Operations
by: W. C. Wilkinson, A. Clarke, B. Epperson, R. Knoblauch
of The Bicycle Federation of America

Review by John Forester

home             facilities

1 Introduction


1.1 Description of Document

This document describes the methods by which the present governmental policy and programs for bikeway design were developed. This document is the basis for the manual for designing bikeway systems, Selecting Roadway Design Treatments to Accommodate Bicycles, FHWA-RD-92-073, and it is quoted from extensively in the document that describes the scientific basis for the government's policy and programs for cycling, Bicycle Safety-Related Research Synthesis, FHWA-RD-94-062.

These documents use the bikeway standards contained in the AASHTO Guide for the Development of Bicycle Facilities. It is assumed that the designs in the Guide have sufficient scientific support, which is an incorrect assumption. The Effects of Bicycle Accommodations supposedly describes the effects of installing different types of facilities and recommends where to install them. Selecting Roadway Treatments is the short field handbook that tells how to do design bikeway systems, copying information from The Effects, without describing the basis for the designs.

I refer to the authors as W&C since Wilkinson and Clarke are the principal ones.

1.2 Importance of Document

Insofar as there is some engineering and scientific basis for the government's cycling programs, the Research Synthesis and The Effects of Bicycle Accommodations on Bicycle/Motor Vehicle Safety and Traffic Operations documents contain it all, at least by reference.

If these documents do not contain adequate scientific support for the government's cycling policy and program, then they do not have such support. I outline the document in section 2, concentrating on the controversial points. My comments are both italicized and are between { } braces, so that readers with either hard copies or ASCII e-mail copies can distinguish them from the descriptions. After the outline I discuss my conclusions in section 3.

2 Outline of Document

2.1 Introduction

"The primary objective of this project was to develop a Manual for selecting roadway treatments to accommodate bicycles. … The report reflects all that has been learned since the bikeway boom of the 1970s."

The intent is to accommodate a "wide range of bicyclists." W&C say that this is a change from the "thinking that has predominated in bicycle facility planning in the United States for the last 15 years," justifying their statement by a footnote reference to Effective Cycling. {This is false. W&C strive to give the impression that bicycle planning in the USA has been devoted to providing the types of facilities that are desired by experienced cyclists, the group A cyclists, who lawfully and competently follow the vehicular-cycling principle as detailed in Effective Cycling. Of all the bicycle plans that I have seen, the only one that came any where near to that was that for New Jersey, which advocated wide curb lanes on all highways. The AASHTO Guide specifically states that "Bicyclists differ widely in their abilities." Every plan I have seen other than New Jersey's is geared to the incompetent cyclists. For that matter, reading Effective Cycling, the book referenced by W&C, discloses a long complaint that governmental cycling policy has always been geared to the incompetent.}

W&C state the policy question as whether to accommodate current users or to increase the number of users. {This is a false dichotomy. The true policy question concerns the best way to increase the number of users and the amount of cycling transportation used.}

W&C quote with approval the FHWA Policy Memo by Dr. Thomas Larson: "Bicyclists and pedestrians are legitimate users of the transportation system." {The trouble with that statement is that the transportation system includes sidewalks, paths, ditches, and lots of places besides roadways. Despite even questioning by Congress, Larson would never admit that cyclists were legitimate users of roadways.}

"Many bicycle advocates … believe promoting bicycling requires better education and training for the many individuals who occasionally ride bicycles, but who are uncomfortable with the idea of `operating in traffic.' … However, apparently the public does not want to buy it. In a Harris poll survey of adult bicyclists, only 1.5 percent of cyclists identified training as a factor that would encourage them to ride more often." {This is a false argument. The purpose of the better education and training is not to persuade people to cycle, but to ensure that those who choose to cycle do so safely, using the only known safe and effective method. The consequences of this argument by W&C are discussed later.}

"The Bicycle Federation of America estimates that less than 5% [of current bicycle owners] would qualify as experienced or highly skilled bicyclists. Therefore, as the goal is to increase bicycle use and as new users will be predominantly novice riders, any plans must meet the needs of both experienced and less experienced riders." {This is a false argument. Almost by definition, almost all new users will be novices. However, that does not mean that the system should be designed for novices. What is needed is an effective procedure for attracting novices to a system that is designed for safe and effective use by those who are experienced.}

W&C then invented the classifications of cyclists that they use thereafter. These groups are:

bulletA Advanced: experienced
bulletB Basic: casual, novice, occasional
bulletC Children: pre-teen

Then by reading some documents, otherwise unspecified, W&C determined the needs of cyclists in these groups.

bulletA: Direct access to destinations; operate at maximum speed with minimum delays; facilities that allow easy overtaking by motorists.
bulletB: "Comfortable" access by low-volume streets or designated bicycle facility; well-defined separation from motor traffic on arterial or collector streets.
bulletC: Access to key destinations near home; well-defined separation from motor traffic; residential streets with low traffic speeds.

W&C then guessed that no more than 5% of the cycling population are now, or ever will become, type A cyclists.

W&C then combine B and C cyclists because both classes want well-defined separation from motor traffic.

{The classifications themselves are trivial; there will always be experienced, inexperienced, and child cyclists, unless all cyclists become experienced during childhood. However, it is incorrect to combine child and adult cyclists. Children will always be inexperienced, while adults acquire experience; adults go to a much wider range of locations than do children. Also important are the other characteristics attached to each class.

W&C describe these characteristics as "needs," which the dictionary defines as necessities. The needs attached to group A cyclists are pretty obvious; they are the facilities that are required to effectively use cycling transportation. Even the need for easy overtaking by motorists can be looked on as a political necessity, because when motorists have found themselves blocked by cyclists they have tried to kick cyclists off the roads. However, why do group B cyclists require access by low-volume streets or designated bicycle facilities with well-defined separation from motor traffic? Why do they require these things while group A cyclists do not? Consider the necessities that W&C have left out; not the trivial ones, in this context, of owning bicycles and having jobs to ride to, but the required skills.

W&C earlier argued that the purpose of developing traffic- cycling skill was for the encouragement of cycling, that it had failed to produce new riders, and that non-cyclists did not think that training would persuade them. With education relegated to the advertising function, this dodge enabled W&C to ignore the need for skill; they pretend that skill is not necessary at all and that the provision of bikeways is a valid substitute for skill because bikeways produce more new cyclists than does skill.

W&C may reply that they specifically recognize group A cyclists as "experienced or highly skilled," although they did not use this characteristic when defining the classes. They may make three arguments to support their point.

  1. They didn't define skill as a need (something lacking) because class A riders already possess it and the others don't need it.
  2. Skill is required only when riding as class A riders ride.
  3. Skill is not needed at all.

Whichever way they choose to make their argument, the fact is that W&C have made the point, without having actually committed themselves to saying it, that traffic-cycling skill is not necessary when cycling on bike-laned streets or on urban bike path systems. Mendacity is the word for such behavior.}


2.2 The Use of Wide Outside Travel Lanes as a Bicycle Facility

W&C give the standard advantages. Disadvantages: doubling-up by motor vehicles, increased motor vehicle speed, costly, don't serve B/C riders, "are a relatively low-visibility improvement with a limited ability to actively encourage increased use of bicycles for transportation." B/C riders prefer striped shoulders to wide outside lanes because of a "greater sense of security and a more clearly defined space."

{Motorists rarely double up. They do so when conditions are very congested and speeds are slow, or when overtaking cars that are waiting to make left turns. I don't see this as a significant problem. Worrying about the increased speed of motor traffic on wide outside lanes is probably more of an anti-motoring argument; such streets are not candidates for traffic calming anyway. W&C are willing to call wide outside lanes costly, while they never call bike lanes costly. In fact, wide outside lanes with a width of 16 feet require 1 foot less width than 12 foot motor lanes plus 5 foot bike lanes. The extent to which the preferences of B/C riders should be taken into account will be discussed later.} 2.3 Bicycle Lanes

"Group A bicyclists prefer wide curb lanes [for not complicating turning movements at intersections]. Group B/C bicyclists, though, prefer bike lanes with their channelized movement of motor vehicles and bicycles and lower level of perceived conflict between the two."

The only agreed-upon "positive effect … achieved by striping a bicycle lane … is improved channelization. … At study sites with bike lanes, 70 percent of vehicles passing a bicyclist showed no change in lateral placement … At study sites with shared roadways, 26 percent of vehicles passing a bicyclist showed no change in lateral placement. …For group B/C riders this effect is particularly important. Less confident riders need to feel that traffic is not going to be driving in the same lane with them and will not be moving about from side to side — with the potential for misjudgment — as they pass." The other advantage is that B/C riders like them.

{The scientific failings of this study are discussed at the appropriate place, under section 6 of the outline of the report.

The psychology of B/C riders is discussed in a later section.}

W&C quote Lott & Lott's Davis bicycle accident study as showing that the Davis bike lanes produced a reduction of car-bike collisions, although they also say that that study showed an increase in car-bike collisions caused by cyclists turning left from bike lanes.

{That study was criticized as long ago as 1977 (in my Handbook of Cycling Traffic Engineering) as using a statistical method never used before or since, and whose results, when confidence intervals are calculated, could not distinguish between reducing car-bike collisions and increasing them.}

W&C list the disadvantages as: adverse effect at intersections, added maintenance, don't work as well adjacent to parking.

2.3.1 W&C's Conclusions and Recommendations

  1. Bicycle lanes work best on urban section (curb and gutter) roads without on-street parking and with a posted speed limit of 40 m.p.h. or less.
  2. Bicycle lanes should be one-way.
  3. 5-ft width, 6-ft where traffic speeds > 45 m.p.h.
  4. Strong channelizing effect. Good "where motorists are not used to sharing the road with bicyclists or where the intention is to serve group B/C riders."
  5. Channelizing effect can complicate turning movements, requires "carefully crafted" striping.
  6. "Group A bicyclists generally prefer unstriped wide curb lanes while group B/C bicyclists prefer bike lanes."

2.4 Effects of Bicycle Use on Shoulders

Although W&C have a very long discussion about the use of shoulders and make 17 conclusions and recommendations, they say nothing that is controversial to a cyclist. The portion discussing freeway shoulders is generally in favor.

2.5 Off-Road Pathways and Trails

While W&C say that bike paths should have only a few intersections with roads or driveways, and they specifically oppose side paths, they do not mention two things. They do not point out how few are the urban locations in which bike paths are suitable. While they say that most paths are multi-use trails, they fail to describe the problems and dangers of multi-use traffic. In one sentence they say, regarding the Burke-Gilman trail in Seattle, that "fast cyclists are encouraged to ride on the nearby street system where their speed is more compatible with other users." W&C fail to distinguish the useful function of short-cut bicycle paths, where because of the shortened distance low speeds are acceptable, from the general transportation function, where distances are generally greater.

2.6 Field Study of the Effects of Bike Lanes on Motor Vehicle Traffic

"A series of field studies was conducted to determine what effect bike lanes and wide curb lanes of various widths have on safety and traffic operations." Studies at intersections were abandoned because "the behaviors observed were a result of the adaptive behavior that bicyclists had learned in response to the unique characteristics of each intersection. … Since the relationship between bicycle facility width and cyclist safety at intersections appeared very difficult to demonstrate, research focused on the relationship between bicycle safety and convenience and facility width at midblock locations."

The cities studied were Madison Wi, Eugene OR, Blacksburg VA. All near universities. Sites: 200-250 ft midblock observation zone, at least 100 ft from stop sign or traffic signal; bike lanes or wide curb lanes signed as bike routes, bikes > 12/hour. Site observation and recording methods: data sheet for each site giving items such as number of lanes, type of bike-lane stripe, etc., coded for keyboard, and detailed drawing with dimensions, not coded for keyboard.

Traffic observation and recording methods: visual observation recorded on tally sheet. Data: see sheet page 72. Data Sheet Errors: All cyclists in wide curb lanes were recorded as "In traffic stream," while cyclists in bike lanes were recorded as being in left, center, or right of lane. While there was a space for "No lateral displacement" of car, there was none for No Reduction in Speed, the speed change choices being only Possible Reduction, Obvious Reduction, Applied Brakes, or Increased Speed.

Range of road data. Bike-lane widths 5-10.6 ft; shared-lane widths 10-21 ft; parking lane widths 7-13 ft.

{W&C stated that their intent was to compare the behavior of traffic on roads with bike lanes with that on roads with wide outside lanes, they did not do this. There is no evidence in the report that they even recorded the data necessary to do this; while they give the ranges for the widths of the bike lanes and of the shared lanes, they give no range for the widths of the motor lanes adjacent to the bike lanes.

The real question is whether there is a significant difference between behaviors on streets with equal widths allocated to the outside lane motor traffic and the bicycle traffic, whether or not that width is divided into two lanes. While it might be difficult to match each bike-laned street with a wide-curb-lane street of equal width, it surely is not difficult to compare the behaviors on streets of approximately equal widths, or to compare the behavior as affected by the width of one type of street against the behavior as affected by the width of the other type of street. While the behaviors would show different distributions, the comparison of the distributions for each type of street would be a valid comparison.

Far worse than this error, W&C do not compare what they said they were going to compare and what was the purpose of the study in the first place. They compare the behavior of traffic on streets with bike lanes against the behavior of traffic on streets without bike lanes regardless of the width available, which in some cases is as little as 10 feet according to their own words.}

"Since vehicles tend to track relative to the outside lane line or the bike lane line, when a bicycle is present and there is no marked bike lane the approaching vehicle must move over to pass the bicycle."

{This is erroneous. The majority of motorists driving in wide outside lanes steer their vehicles according to lane line to their left, not the edge stripe or curb to their right. Observation of traffic, and of the position of debris left in the unused spaces discloses this.}

The only table of results that W&C produced is reproduced as Table 1.
Table 1: Lateral Placement Changes by Facility Type, %
Type of Change in Lateral Placements Bike Lanes Shared Use Lanes
No Change


Possible Change


Obvious Change


Crossed Lane Line


Crossed Center Line



Levels of significance. 0.70 > 0.26, p of chance < 0.0005; (0.16 + 0.19 + 0.13) > (0.06 + 0.03 + 0.02), p of chance < 0.01. None other significant.

Speed changes. Shared use, possible + obvious reduction = 0.33, bike lane = 0.08; p by chance < 0.051.

Correlations. Lateral placement changes not correlated with bike lane width, shared use lane width, parking lane width. Negative correlation at shared use sites with width of entire roadway, -0.52. Speed reductions at bike lane sites not correlated with bike lane width, shared use lane width, parking lane width. At wide curb lane sites, no correlation with lane width.

2.6.1 W&C's Conclusions and Discussion

  1. "Motorists tend to slow down and move over when passing bikes on bike lanes and shared use lanes.
  2. There is less slowing down and moving over at locations with marked bike lanes than there is at locations with shared use lanes.
  3. These behaviors are not correlated with bike lane width, shared use lane width, or parking lane width.

These conclusions suggest that wider bike lanes and wider shared use lanes do not necessarily result in less disruption to vehicular traffic and that narrow facilities of either type do not necessarily produce more disruption."

{This conclusion is false; the correlations exist, but are not strong.}

2.7 European Bicycle Facilities: An Overview

Wide diversity in European nations.

Madrid. No cycling, 1% bicycle ownership.

Netherlands. Bikeways are intended to prevent congestion on roadways. [Keep people cycling and get bikes off roadways.] Tilburg & den Hague bikeways (approx. 6 miles) attracted cyclists from other routes and "encouraged a small number of motorists to leave their cars at home." Cycling decrease 6%/year 1960s, 2%/ year 1970s, attributed to bikeways. {It would be equally valid to attribute the decline in cycling rate to the normal logistic curve of replacement. By the 1970s, the enthusiastic motorists had changed over, leaving only those less willing to change. Besides, in the 1970s there were the oil crisis and growing environmental concerns.}

Delft plan provides facilities on 500 m grid, requiring time priority at intersections and shortcut bike paths.

Denmark. Cyclists prefer blue bike lanes carried through intersections. Quote Bracher: "Where cyclists most praised the quality of the new route, Herning, it is least used. In Aarhus the route was especially heavily criticized, but most used." {Yes, Danish cyclists liked the blue bike lanes carried through the intersection. At the Montreal VeloCity conference one of the proponents of this scheme proudly showed the photograph of a cyclist doing what he should, turning left from the curb lane, with a same-direction bus showing close behind his left shoulder. We were all appalled at the danger.}

United Kingdom. Much experimentation. Germany, Belgium and UK have little space, therefore put cyclists with pedestrians.

Germany. Sidewalk-style bike paths.

2.8 The Bicycle Planning Process

2.8.1 Introduction

Provide for group A cyclists on all roads and streets except where prohibited by law. To encourage B/C cyclists, provide a network of bikeways that serve all destinations and routes that are not accessible by low-traffic residential streets.

2.8.2 Establish Performance Criteria for the Bicycle Network

bulletAccessibility. Have bicycle facility within 1 mile of every residence. Every significant place requires reasonable access by bicycle.
bulletDirectness. Reasonably direct as a trade-off with pleasantness.
bulletContinuous network.
bulletFew conflicts with motor vehicle operators.

{The question is not whether the system described above will appeal to B/C riders. The question is whether such a system is the best system for cycling transportation. This is discussed in a later section.}

2.8.3 Inventory the Existing System

Nothing special here.

2.8.4 Identify Bicycle Travel Corridors

W&C recommend using existing traffic counts to show where people want to go, plus adjustments for specific institutions and recreational locations that generate cycling traffic.

2.8.5 Evaluate and Select Specific Route Alternatives

"The next step is to select specific routes within these corridors that can be designed or adapted to accommodate group B/ C bicyclists and provide access to and from these locations."

2.8.6 Select Appropriate Design Treatments

"Is the proposed route projected to be used primarily by group A bicyclists, or is it intended to also serve as part of a network of routes for group B/C bicyclists?"

2.8.7 Evaluate the Finished Network Plan

Evaluate and correct.

{While these procedures specify a generally applicable procedure, they do not specify the design that results. That is determined by the methods of designating the types of facilities that result in the finished product. Therefore, comments about this subject are in the next section and in later sections.}

2.9 Design Selection and Specifications

Although the recommendations cover only roadway treatments (shared normal outside lane, shared wide outside lane, shoulder, and bike lane), they also specifically permit bike paths without any significant controls.

Allow for A cyclists to go anywhere on all streets.

Provide for group B/C bicyclists unless it has been determined that the route need be used only by group A bicyclists.

{The result of this process is that the entire area is designed to suit B/C cyclists, because there will be very few locations that only A cyclists would want to use. In some cases the service will be by bike path, but in others it will largely be by bike lane.}

2.9.1 Factors:

Based on five of the following six factors W&C have created tables showing the recommended facility types for groups A and B/ C. Traffic volume.

"Greater potential risks … less comfortable for B/C cyclists." {Not only is there is no evidence that the risks from this factor are greater for B/C cyclists than they are for A cyclists, but there is substantial evidence that the risks must be the same. There is also substantial evidence that these are the minor risks of cycling, rather than the major ones.} Traffic speed.

"Negative impact on risk and comfort." {Not only is there is no evidence that the risks from this factor are greater for B/C cyclists than they are for A cyclists, but there is substantial evidence that the risks must be the same. There is also substantial evidence that these are the minor risks of cycling, rather than the major ones.}

Traffic mix. Large vehicles "can increase risk and have a negative impact on comfort for bicyclists." {Not only is there is no evidence that the risks from this factor are greater for B/C cyclists than they are for A cyclists, but there is substantial evidence that the risks must be the same. There is also substantial evidence that these are the minor risks of cycling, rather than the major ones.} Parking.

Increases the width required. However, while W&C provide different tables of widths for the parking and the no parking conditions, there are practically no differences between them. {Not only is there is no evidence that the risks from this factor are greater for B/C cyclists than they are for A cyclists, but there is substantial evidence that the risks must be the same. There is also substantial evidence that these are the minor risks of cycling, rather than the major ones.} Sight distance.

"Situations … where the sight distance is likely less than that needed by a motor vehicle operator to either change lane positions or slow to the bicyclist's speed." In other words, where motorists are exceeding the speed law. {Not only is there is no evidence that the risks from this factor are greater for B/C cyclists than they are for A cyclists, but there is substantial evidence that the risks must be the same. There is also substantial evidence that these are the minor risks of cycling, rather than the major ones.} Frequency of intersections.

W&C state that they did not use this factor when building their tables, although they admit that this factor ought to be important. They say that "there is some evidence that the disruption in traffic operations associated with bike lanes is temporary."

{Turning and crossing traffic, which is at intersections and driveways, is the cause of approximately 95% of car-bike collisions. Rather than concentrating on the road sections between such locations, a rational cycling policy would have concentrated on what to do about these situations and circumstances. W&C's admission that they ignored these is their admission that their entire program is wrong.

W&C's argument that there is some evidence to show that the disruption caused by bike lanes is temporary is both an admission that they recognize the disruption exists and that they hope that it will disappear. If the design causes disruption, even though the users might at sometime learn how to avoid that disruption, that argument is exactly that the design does not suit B/C cyclists and average motorists, but is suitable only for group A cyclists who have learned how to avoid the disruptions so caused.}

2.10 Appendices

2.10.1 Rumble Strips

Don't use them.

2.10.2 Freeway Shoulder Use, Caltrans

Use of rural freeway shoulders should not be a cause for concern, situations are little different from other rural highways, or are better.

2.10.3 Freeway Shoulder Use, Pima County

A point system for comparing freeway and alternate routes.

2.10.4 Feasibility of Paving Shoulders of Low ADT Roadways

Paved shoulders are economically feasible and beneficial for many reasons.

3 Analysis and Discussion

3.1 The Document is a Fraud

While this document is titled "The Effects of Bicycle Accommodations on Bicycle/Motor Vehicle Safety and Traffic Operations," it contains nothing at all about safety, that is accidents to cyclists and their prevention, and ignores most of what is known about traffic operations involving cyclists. It concentrates on the most minor part of all, motorists overtaking cyclists at midblock locations. The rest of the report is a summary of what government already does about cyclists, good, bad, or indifferent.

The most reasonable explanation for this remarkable concentration on the most minor matter and the avoidance of the important matters specified in the title of this document is that only the most minor matter supported the government's desires for actions concerning cyclists. Practically everything else known about safety and traffic operations leads to conclusions opposing governmental desires about cyclists.

3.2 The Purpose Served By Ignoring Safety

W&C ignored safety because everything that we know about cycling safety contradicts the government's desired program that W&C try to justify. Despite more than 25 years of trying, there is no satisfactory evidence that bikeways reduce the cyclist accident rate or reduce the skill level required for safe cycling. Secondly, there is much evidence that even perfect bikeways cannot significantly reduce the cyclist accident rate because bikeways address only a minute portion of accidents to cyclists, whether one considers all accidents or even only car-bike collisions. Thirdly, there is much evidence that bikeways in urban areas disrupt the traffic patterns and probably cause more car-bike collisions.

The most that W&C can claim, which is what Clarke does claim in the next document (Bicycle Safety-Related Research Synthesis, FHWA-RD-94-062), is that there is no direct evidence (there is a little) that either bikeways significantly reduce accidents or that bikeways significantly increase accidents. That is true; all the evidence is indirect, from analysis of accident types and frequencies to analysis of traffic operations. However, all that indirect evidence contradicts the government's desires for a bikeway program. If we did what was required for safety, we would do many other things than produce bikeways, and we would produce many fewer bikeways of a different style.

By ignoring safety issues W&C were able to publish a pseudo-justification for the government's bikeway program that persuades the uninformed majority.

3.3 The Purpose of Ignoring Most Traffic Operations

The subject of traffic operations covers a wide variety of subjects concerning the movement of traffic. One part of the field concerns the flow characteristics of traffic: the speed and density of flow in different types of facility. Another part of the field concerns the methods for and the effect of the maneuvers that are required when interacting with other users and to distribute the traffic vehicles from their various origins to their various destinations. Another part of the field concerns the accidents that are involved in the movement of traffic.

W&C studied only one subject of this wide field of traffic operations, the subject of the flow effects of motorists overtaking cyclists. Concentrating on this served two purposes. First, this concentration avoided all the other aspects of traffic operations in which bikeways have been shown to be less safe and less effective than normal roads. Second, it was hoped to provide data supporting the reason why the motoring establishment desires a bikeway program. The motoring establishment, which was the original force behind the government's bikeway program and which remains a very powerful proponent of bikeways, wants bikeways to clear the roadway part of the highway system for unimpeded use by motorists. For example, that is why Dr. Larson, Administrator of the FHWA, refused to admit that cyclists are legitimate users of the roadways.

So W&C studied the amount by which motorists changed lateral position and reduced speed when overtaking cyclists. As one would have expected, they showed that motorists made fewer changes when overtaking cyclists who were in bike lanes than when overtaking cyclists who were in shared lanes. W&C should have compared shared lanes that were equal in width to the sum of the motor and the bike lanes. However, by their own admission, they confounded the conditions to make shared lanes look much worse than wide curb lanes are.

While the data show, even if inaccurately, increases in lateral movement and decreases in speed, how important are these? The only real importance of the changes in lateral movement would concern collisions between motor vehicles. Drivers of motor vehicles are allowed to drive at any lateral position within their lane, and before this time nobody, aside from cyclists, has ever shown any concern for where within their lane they cared to drive. Motor vehicles colliding because one changed lateral position within its lane have never been identified as a significant cause of accidents. Even drifting over lane lines (as opposed to purposeful changing of lanes when erroneously thinking the next lane was clear) has not been shown to be the cause of a significant number of collisions.

Slowing down has two significant effects. One is the time lost, another is the number of collisions caused. W&C remark that actual brake applications, easily observed by operation of the brake lights, were so rare the frequency could not be used for analysis. Therefore, the slowing was caused at most by removing the foot from the accelerator. At urban speeds this produces a deceleration of about 2 feet per second per second. In four seconds this reduces 30 m.p.h. to 25 m.p.h. and closes the gap to a car behind that continues to travel at 30 m.p.h. by 20 feet, while traveling 176 feet forward. By then the motorist would be past the cyclist he had slowed for. This does not look like a significant cause of collisions. While rear-end collisions are a significant part of automobile collisions, most of these occur from a stopping car, not a slowing car, and collisions between motorists, one of whom is overtaking a cyclist, have never been shown to be a significant number of collisions, if any at all. (I have never seen such a number advanced by any anti-bike advocate, despite all the efforts made to show that cyclists are a danger to motorists.)

In summary, the differences between the traffic flow effects of bike lanes and wide curb lanes, even when measured by W&C's biased system, are minor.

Other traffic operations may be rated in terms of the total number performed, the number of types, of their difficulty, or of their accidents. While it is true that the most frequent interaction between cyclists and motorists is the maneuver of a motorist overtaking a cyclist, this maneuver is easy to perform and causes very few accidents, 0.2% to 0.3% of accidents to cyclists, 2% of car-bike collisions in urban areas in daylight, and 0% of car-car collisions. (Nighttime car-bike collisions are caused by a different problem, and rural areas won't have bikeway systems.)

Although turning and crossing maneuvers are not nearly as frequent as motorist overtaking cyclist maneuvers, they require far more skill and more carefully designed facilities, and they cause about 20 times more car-bike collisions. This is the subject area of traffic operations that should have been studied and for which facilities and skills should have been evaluated.

3.4 The Types of Cyclists

The classification of cyclists devised by W&C is unremarkable. There will always be child cyclists, and adult cyclists will always vary to some extent in experience and skill. However, W&C err both when applying characteristics to these classes of cyclists and when combining them.

3.4.1 Adult and Child Cyclists Should Not Be Combined

It is appropriate to categorize children as generally inexperienced in any field of adult endeavor. The difference, in the case of children, is not only in lack of experience because of less time in which to have learned, but also because of both insufficient maturity to have learned from experience and from having been protected from exposure to the difficult or dangerous locations that would generate experience. With inexperienced adults only the lack of experience applies. They have the maturity to learn from experience and they already go, by other means, to whatever locations they desire. By the dodge of combining adult and child cyclists into one group for planning purposes, W&C have made it necessary to plan for access by children to wherever adults might desire to go. For example, W&C have made it necessary to design the access to railroad stations to be appropriate for unaccompanied young children, even though we would not allow young children to travel alone by train, to say nothing about the difficulty they would have regarding carrying the bicycle. It is appropriate to design for access by child cyclists only those locations that we would permit unaccompanied young child cyclists to go to.

Designing the whole system to suit child cyclists is undesirable for several reasons. The cost will be greater. There will be a greater tendency to install facilities that are disliked by group A cyclists, who would otherwise become the majority in a few years. Very importantly, a system that is almost everywhere suitable for children will keep adults riding like children; it will dissuade adults from learning how to ride properly and safely.

3.4.2 Long-term Demographics of Cyclists

W&C have produced the estimate that only 5% of present bicycle owners are experienced cyclists. Without actually saying so, their logic is based on the assumption that this proportion will remain unchanged. They do not say how long it takes to become experienced. We consider that adults who take up motoring become sufficiently experienced to drive without supervision in about 1/ 4 of a year. Just for the sake of argument, let us assume that learning to cycle takes four times as long, 1 year of practice on working days. With a cycling lifetime of 15 years and an approximately equal division of cyclists over the range of 1 to 5 days per week of cycling (with a slight bias to the lower frequencies), the proportion of inexperienced cyclists on the system would be only 16%. It is foolish to design the system for inexperienced cyclists when the proportion of inexperienced cyclists is only 16% of those using the system at any one time.

It may be argued that at the beginning of a successful program of cycling transportation the majority of the users will be new and inexperienced cyclists, the 95% that W&C promote. That advocacy only shows their bias. No matter how successful any cycling transportation program is, it will never persuade all the nation's bicycle owners to simultaneously take up transportational cycling. The growth will follow something like the logistic curve, and at no time will the majority of users be new and inexperienced.

3.4.3 Preferences versus Fear

W&C give only one reason why B/C cyclists desire bike lanes: "Less confident riders need to feel that traffic is not going to be driving in the same lane with them and will not be moving about from side to side—with the potential for misjudgment—as they pass [they mean, as it passes]." W&C label this a mere preference, no more significant than preferring camembert to stilton but sufficient to direct the market. However, they knew that they ran no risk of having this so-called preference overturned as being an unnecessary frill because they recognize that the market that they serve is actually driven by fear instead of mere preference. That is, fear of being hit from behind on the part of consumers (cyclists) and fear of being delayed by cyclists on the part of suppliers (governments and the motoring establishment). So long as they played upon the fear without actually saying so they had a chance of getting away with this tactic. This is another of their dodges to avoid discussing the real issues of safety and fear.

Inexperienced cyclists demand bike lanes because they believe that being hit from behind is the prime danger of cycling, they are frightened of being hit from behind, and they believe that bike-lane stripes protect them from this danger. They apply the same reasoning to bike paths, which have even greater problems than bike lanes.

Had W&C described this condition as the fear that it is, instead of calling it a mere preference, they would have had to discuss safety. Since the safety evidence is against bike lanes and bike paths, they would have found themselves arguing against the cause and the source of funds to which they are committed.

W&C would also have had to try to justify their division between A and B/C cyclists, their decision to put inexperienced adult cyclists (B class) with children (C class) instead of with other adult cyclists (A class). The decision was based only on the supposed "preference" of both C and B cyclists to be separated from motor traffic by bikeways. (While W&C say that they considered the preferences of children, children have little reason or opportunity to independently consider the effect of bikeways. What W&C actually considered was the opinions of parents, either directly or reflected through their children, and American parents are overwhelmingly either non-cyclists or inexperienced cyclists.)

Experienced cyclists don't like bikeways, inexperienced ones want them. The supposed virtue of bike lanes is that they supposedly protect cyclists from being hit from behind. This is an eventuality that the cyclist cannot protect against. (In daylight; at night the question is one of nighttime equipment.) It is beyond the bounds of all reason to suppose that a motorist traveling along the road looks ahead to distinguish experienced cyclists from inexperienced ones so he can steer around only the experienced ones. Furthermore, there is no evidence whatever for this fantastic hypothesis.

W&C describe the preferences as "needs." They write that group A cyclists don't need bike lanes to protect them from same- direction motor traffic. However, group B cyclists have, by the obvious argument above, no greater need for bike lanes than do group A cyclists.

The reason that experienced cyclists don't worry much about that danger is that experience has shown them that it is not a significant danger while also showing them that traffic from each side and from ahead are far more significant dangers that they have effective means of handling. Experience is the reason why experienced cyclists don't fear motor traffic from behind the way that inexperienced cyclists do.

3.4.4 Marketing versus Safety

W&C argue that we should provide whatever facilities the market desires. Ideally, that would be reasonable with a frictionless market of well-informed independent consumers purchasing non-hazardous products to be used independently.

However, the highway system is not such a market. The highway system is a monopoly; a person either uses it or emigrates. The highway system is an inherently dangerous system that requires both safe design and safe operating practices, both with very critical requirements, for reasonable safety of its users. The consumers being considered are woefully ill-informed, being not merely unskilled but controlled by superstition that is far from any reasonable relationship to reality. The marketing strategy is based on the assumption that future consumers will remain as ill-informed as those who are not yet users. The producers have shown no interest in either the welfare of the consumers or the profits to be made from them, but instead are supplying goods that serve to control the consumers for the benefit of the producers and their constituents.

Therefore, the highway system is an inappropriate place for the kind of marketing strategy that W&C propose. Because the market system proposed by W&C is a grossly imperfect market, the result will not be the mass of satisfied customers that free- market theory assumes. The more likely results are discussed below.

3.5 Safety

Because W&C have ignored safety, one result will be an increase in cyclist accident rates, both in gross numbers and in the rate per bike-mile. Cycling safety has three general areas: facilities, drivers, and equipment. Facilities can be divided into two areas, normal roads and bikeways. Drivers concerns the operating methods and skills of users. Bicycle safety equipment, largely the area of lights and reflectors, is not pertinent to this discussion. W&C have ignored the safety aspect of facilities and have totally ignored drivers (although supposedly giving drivers what they want).

3.5.1 Normal Roads and Cycling Safety

The well-designed modern road allows cyclists to operate safely and effectively under most conditions. Although W&C discuss none of the features that allow well-designed roads to provide this service, this is not the place to instruct the reader in the design of roads. As W&C allow, modern design provides wider outside through lanes than did older design, specifically to allow cyclists to operate with minimum disturbance to motor traffic. Whenever the following discussion mentions roads, it considers roads that are well-designed. Bringing a road up to modern standards is the first step; it is unfair to compare an old road with old design against one that has both new design and a bikeway. As will be shown, any improvement probably was caused by the improvement in normal roadway design rather than the bikeway.

W&C admit both that experienced cyclists ride on the roads that W&C consider most dangerous and that experienced cyclists don't like bikeways. What W&C do not say is that experienced cyclists have an accident rate only about 20% of that for other adult cyclists. This has been shown by studies in both the USA (1) and in Britain (2), and there have been no studies leading to opposite conclusions. In short, cycling skill, merely developed by experience of cycling on normal roads, reduces the accident rate enormously. As will be shown, this is a far greater effect than any that is produced by varying facilities.

Most cycling has been on normal roads. Car-bike collisions constitute only about 12% of accidents to cyclists.(4),(1) All transportational bikeway systems will be in urban areas, and will be largely used in daylight. (Also, nighttime accident statistics are confounded by the inadequate reflector systems commonly used.) Only 2% of car-bike collisions in urban areas in daylight, which are only 0.3% of accidents to cyclists, are caused by same- direction motor traffic hitting lawful cyclists.(5),(3) About 95% of car-bike collisions are caused by turning and crossing movements. (3)

3.5.2 Bike Lanes and Cycling Safety

Bike-lane stripes are intended to protect cyclists from same-direction motor traffic. It is obvious that this is their sole function, for all other hazards remain. It is therefore obvious that bike-lane stripes cannot make cycling significantly safer. If they worked perfectly, the most that they could do is to reduce the accident rate to cyclists by 0.3%. It is obvious that those who demand bike-lane stripes, those who, according to W&C will not cycle unless they are protected by bike-lane stripes, must have a completely inaccurate concept of the hazards of cycling and therefore of the skills required for safe cycling.

Recognizing this, some have argued that bike-lane stripes have no effect on safety; while the cannot reduce accidents significantly, neither do they cause any. If that is so, they argue, we should give the public the bike lanes that it wants. W&C implement this conclusion merely by ignoring safety. Then Clarke, one author of this document, makes this specific argument in his later document, Bicycle Safety-Related Research Synthesis,</ EM> FHWA-RD-94-062.

However, bike-lane stripes are not neutral towards safety. They produce both a direct and several indirect adverse safety and political effects. The bike-lane stripe enforces the rule that motorists stay to its left and cyclists to its right. That is the principle on which it was installed and by which people believe it should operate. However, that contradicts the proper relationship when cyclists go faster than motorists (often easy in congested traffic), when motorists turn right, and when cyclists turn left. In each of these situations, the cyclist should be to the left of the motorist. Observation has shown that cyclists in cities with bike-lane systems make more errors than those in cities without bike-lane systems, and that the typical errors are related to the design of the bike-lane system of each city.(6) Motorists make their similar errors also. As a result, cyclists make left turns from the bike lane without yielding to motor traffic, motorists make right turns from the motor lane without yielding to cycle traffic, and cyclists get caught by trying to overtake cars that turn right.(7) Observation also shows that expert cyclists don't make these mistakes around bike lanes. Therefore, some argue that bike-lane stripes are harmless because all cyclists ought to ride like experts do. That immediately leads to the conclusion that we should not have bike lanes because W&C classify expert cyclists as neither liking nor needing bike-lane stripes. More to the point, since it takes time for average cyclists to become experts, there will be more accidents while they are learning. However, a policy of bike lanes has further adverse effects that are discussed later.

W&C try to get around these facts by presenting the designs for bike-lane intersections from the AASHTO Guide, and by making the recommendation that "Intersection striping should be carefully crafted to reduce turning-conflict accidents." However, no set of designs that has been devised prevents all such turning conflicts, and there is good reason to believe that such a set of designs is impossible. Bike lanes will always produce more turning conflicts than normal roads possess.(8)

3.5.3 Urban Bike Paths and Cycling Safety

It is often thought that bike paths eliminate the danger of car-bike collisions because there are no cars on them. This is erroneous, and there are the other 88% of accidents to consider also. All bicycle transportation systems will be in urban areas. In urban areas there are very few locations where a lengthy bike path will not cross the same amount of traffic that a road that served the same areas crosses. The exception are largely waterfronts (except industrial waterfronts). Cyclists on typical urban paths cannot avoid either the crossing traffic or much of the turning traffic, and typical intersections between roads and paths make the conflicts much more difficult and dangerous. In addition, there are no urban bike paths; they are all multi-use trails. As a result, while urban bike paths are acceptable for slow cycling, when used at transportational speeds they are very dangerous.

3.5.4 Operating Skill and Cycling Safety

The only way to ride in reasonable safety is to operate as the driver of a vehicle, preferably on well-designed streets. Cyclists who do so have an accident rate less than 20% of that of the typical adult bicycle rider. Clearly, we should be promoting this style of cycling instead of bikeways; W&C have managed to dodge this issue only by ignoring all discussion of safety. W&C manage to deflect this discussion by pretending that proper cycling is a sales program instead of a safety program. "This approach can be thought of as a sales approach. Bicycle advocates have a product (training to give casual bicyclists the skills they need to share most roadways with motor vehicles) and have been selling it to the public as something essential. However, apparently the public does not want to buy it." We will consider safety first and popularity later.

The cyclist who wishes to ride safely on streets with bike lanes must have all of the "skills needed to share most roadways with motor vehicles," to quote W&C's own words expressing the opposite concept. Every study considering the specific skills that are required has reached that conclusion, and nobody, neither W&C nor anybody else, has ever specified the normal cycling skills that are not needed when cycling on streets with bike lanes. Experienced cyclists know this, and that is one reason that they do not like bike lanes and bike paths. Whether or not the non-cycling public wants to ignore this knowledge, there is no doubt that any cycling transportation program must be based on a program for teaching cyclists how to ride properly and safely with all the skills that are required on normal streets.

3.5.5 Superstitions of the Cycling Public

Several studies have been made of the way that typical cyclists rate the safety of roads. Regardless of the initial intent to consider the actual causes of accidents, the studies have all degraded into consideration of only the condition of same-direction traffic: its volume, speed, and the space available. I have analyzed these studies.(9) The six studies (one of which is by one of the authors of this study, Epperson) are listed in the endnotes.(10), (11), (12), (13),(14), (15)

4 The Result of Ignoring Safety and Promoting Bikeways

The direct result of ignoring safety has been to recommend a program that attracts the general public to cycling under the false promise of cycling safety. To the extent that that program does attract new cyclists, it will directly increase the cyclist accident numbers and the cyclist accident rate per mile.

The indirect effect aggravates that adverse result. The public, to the extent that it considers cycling at all, is greatly concerned about the dangers of riding on streets, giving this as its most important reason for not cycling. The Harris poll taken for Bicycling Magazine(16) is only one bit of the widespread evidence for this. The public thinks that the facilities program makes cycling safe. That is what the public thinks; that is why the public demanded such a program. (The experts employed by the motoring establishment do not believe this, as evidenced by Wilkinson's own words; they have their own reasons for expressing this opinion.) However, despite its demand for bike safety, the public rejects skills training, the only known method of significantly reducing accidents to cyclists, a method that has produced reductions of 80%. The public does so because it believes that cycling skills are dangerous. Instead, the public places its faith in bikeway programs, a method of accident reduction for whose efficacy there is no evidence whatsoever, and considerable evidence against.

Given this complete misconception by the public about cycling safety, this document and its companion documents that are used to justify the government's bikeway program, and that bikeway program itself, must produce serious adverse effects beyond the bikeway program itself. By ignoring the subject of cyclist safety, these documents and the bikeways program they attempt to justify allow the public's superstition about cycling safety to control the government's cycling policy and program, which suits the purposes of the motoring establishment. There is nothing in these documents to question the validity of the public's superstition. The provision of funds for the bikeway program, the political activities of the bicycle advisory committees that these funds call into being, the construction of the bikeways, all these events demonstrate to the public the validity of its belief. If these things were not true, would the government build these bikeways for us?

Strengthening the bikeway superstition opposes cycling safety because it opposes cyclist training, the only bicycle safety program that works. It does so in many ways.

  1. First and obvious is that cyclist training no longer appears necessary, since bikeways make cycling safe.
  2. Second, because safe cycling training must teach that bikeways do not make cycling safe, and indeed create more dangers, those who are committed to bikeways and those who want bikeways for their own purposes oppose safe cycling training, denigrating it as arduous, necessary only for risk-taking experts, and the like.
  3. Since the bikeway program demonstrates to the public that the prime danger of cycling is same-direction motor traffic, the public is even more likely to believe these statements. That is because safe cycling training pays far more attention to the known causes of accidents, including crossing and turning maneuvers by both cyclists and motorists, than to the largely imaginary dangers of same-direction motor traffic. To the public, this seems the height of dangerous foolishness.

Thus these documents and the bikeway program that they supposedly justify make it practically impossible to have any program of cycling safety except among the very few who have the ability to see through the falsity, the courage to oppose society's opinions, and the desire to cycle regardless.

End notes

1Comparison of Kaplan's statistics with those from the National Safety Council's two studies of Bicycle Accidents and Usage Among Elementary-School Children and Among Young Adults; Authors are Chlapeka, Schupack, Planek, Klecka, & Driessen

2Watkins, S. M.; Cycling Accidents; 1984; Cyclists' Touring Club; Godalming, UK

3Calculated from Cross's and Kaplan's statistics. See Bicycle Transportation Chapter 5

4Kaplan, Jerrold A.; Characteristics of the Regular Adult Bicycle Rider; 1975; Masters Thesis, University of Maryland.

5Cross, Kenneth D., & Gary Fisher; A Study of Bicycle/Motor- Vehicle Accidents: Identification of Problem Types and Countermeasure Approaches; 1978; National Highway Transportation Safety Administration; Washington DC

6Forester, John; The Effect of Bike-lane System Design on Cyclist Errors; 1978, 1982; Custom Cycle Fitments, Sunnyvale CA

7While the Forester study shows increased dangerous behavior around bike lanes, no study so far shows increased accidents. Errors such as are described cause about 30% of car-bike collisions on normal streets. Comparing accidents on normal and bike-laned streets would be very difficult to do, considering all the confounding factors, but it is reasonable to conclude that where there is a higher frequency of dangerous maneuvers that are known to cause many accidents, there is also a higher frequency of the accidents so caused.

8For further discussion, see Forester, John; Bicycle Transportation, 2nd ed., 1994, The MIT Press

9Forester, John; Studies on the opinions of cyclists about different street conditions. Noland on Perceived Risk; Antonakos on Preferences; Sorton on Stress Levels (1st); Sorton on Stress Levels (2nd); Davis on Evaluation of Conditions; Epperson on Suitability; Landis on Hazard Estimation; Sunnyvale, CA; 1996

10Noland, Robert B.; Perceived Risk and Modal Choice: Risk Compensation in Transportation Systems; Accident Analysis and Prevention, Vol 27 No 4 pp 503-521, 1995

11Antonakos, Cathy L.; Environmental and Travel Preferences of Cyclists; TRB Transportation Research Record 1438

12Sorton, Alex, and Thomas Walsh: Bicycle Stress Level as a Tool To Evaluate Urban and Suburban Bicycle Compatibility; Transportation Research Record 1438

13Sorton, Alex; Measuring the Bicyclist Stress Level of Streets; ASCE HPT Committee, November 1995

14Davis, W. Jeffrey; Bicycle Test Route Evaluation for Urban Road Conditions; ASCE HPT Committee, November 1995

15Epperson, Bruce; Evaluating Suitability of Roadways for Bicycle Use: Toward a Cycling Level-of-Service Standard; Transportation Research Record 1438

16The Cycling Consumer of the 90s; Bicycling, Rodale Press, Emmaus PA; 1991

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