[Home page]

The Cambridge, Massachusetts
Bike Lane Study -- a review

by John S. Allen

First, let me make it clear, I am not ideologically opposed to bike lanes. And there are or have been some in Cambridge, Massachusetts that I think perform reasonably well, at least in part -- on Main Street, Concord Avenue, the Harvard Bridge.

However, bike lanes adjacent to parallel parking have problems -- and Cambridge has many such bike lanes. There also are some serious problems with the research report Cambridge has produced about them. To describe the most serious problem succinctly, the report describes bicyclists as being outside the "door zone" of parked cars when well inside it. I am astonished that this report was published by the Transportation Research Board -- Transportation Research Record #1939, page 3.

Now for a detailed examination of the report. The abstract states:

The analysis examined the percentage of cyclists riding 9 and 10 feet out from the curb. These distances were used as benchmarks for how far cyclists should ride so as to be farther from the "door" zone of a parked car. All three treatments significantly increased the percentage of cyclists riding more than 9 and 10 feet from the curb. There was variation at the measurement sites near the signalized intersection vs. measurement sites near uncontrolled intersections, with higher increases near the signalized locations.

But there's a misleading statement here, and a contradiction. People who only read the report's abstract will not see the following lines, on page 4 of the report:

Data were collected on the following variables: 1) The distance that vehicles parked from the curb for the entire length of the study section on the north and south side of the street (collected once during each condition); 2) The wheel path of bicycles in the roadway at each of the four camera locations; 3) The near side tire path of motor vehicles in close proximity to bicyclists at the same locations.

The words "9 feet from the curb" and "10 feet from the curb" are, then, used in the report to describe the centerline of the bicycle, the wheel path, rather than the rightmost extension of the bicycle and rider. The measurement for overtaking motor vehicles is taken near their right side, at the wheel path nearest the curb, rather than at the rightmost extension (usually a rear-view mirror), and so also is not useable to determine overtaking clearance.

A bicyclist whose wheel path is 9 feet from the curb is almost entirely within the range of opening car doors. A bicyclist whose wheel path is 10 feet from the curb would still collide with the doors of many cars -- the report cites another study from San Francisco which established that 15% of the car doors were opened to more than 9'6" from the curb:

Data gathered in the San Francisco study determined that the 85th percentile of car doors observed opened to 9'6" from the curb (SFDPT data). Giving a 6" clear zone to the bicycle handlebar, the total width of the potential door zone would be 10 feet.

The San Francisco report's main topic is the shared-lane marking, an alternative to bike lanes for streets that are too narrow to accommodate them. Links to the San Francisco report and to other useful references are at the end of this article.

A car door can open too quickly for a bicyclist to avoid it by braking. The only available avoidance maneuver, then, is to swerve to the left, possibly into the path of an overtaking motor vehicle or bicycle. If the right end of the handlebar only nicks a car door, the bicycle will suddenly steer to the right and tilt to the left, and the bicyclist's head will fall to the left into the travel lane, as apparently occurred in a fatal crash on Massachusetts Avenue in Cambridge in 2002.

A bicyclist's wheel track must be approximately 11 feet from the curb -- only 1 foot from the left edge of a standard bike lane -- for the right end of the handlebars to clear nearly all opened car doors. The bicyclist must merge even farther left to overtake the largest vehicles safely (though this need can be anticipated). Additional clearance is advisable at times due to other hazards such as blind spots beyond tall vehicles. These requirements put the bicyclist somewhere close to the left-side of a standard bike lane, and sometimes outside it.

The report states, on page 7:

When one looks simply at an average position, the cyclists did move further away from parked cars in all circumstances, but only by a couple of inches – not as significant as might be hoped. However, the critical evaluation is the effect of the treatments on the distribution of where cyclists rode. Figures 3-6 show histograms of bicycle distance from the curb. Under all test markings, the distributions narrowed, so that there were fewer outliers on either side (which is why the average did not change dramatically). Most importantly, cyclists who were riding the closest to parked cars in the baseline condition moved further away, so the percentage of people riding more than 2 or 3 feet from parked cars went up significantly.

The reduction in the percentage of bicyclists closest to the parked cars is not very important, because those bicyclists are still deeply in the door zone. Note again that the "2 or 3 feet from parked cars" refers to the bicyclist's wheel track, not to the rightmost part of the bicyclist's body or the handlebar. The center of the bike lane is well within the door zone. Bicyclists to the left of center also rode closer to it, so a tightening of the distribution resulted in a slight increase in the percentage of bicyclists riding within the door zone.

The report indicates that the percentage of bicyclists whose wheel track was more than 10 feet from the curb increased from approximately 25-35% to around 50% with the bike-lane treatment at three of the four measurement sites, and decreased from 45% to 37% at the fourth site. No numbers are given for the percentage of bicyclists riding more than 11 feet from the curb, but graphs (Figures 3-6 in the report) show that they were less than 15% at all measurement sites and under all experimental conditions. The graphs show fewer cyclists riding farther than 11 feet from the curb at all but one site after installation of a bike lane. Interestingly, with the installation of only a lane line (left-side bike-lane stripe, at 12 feet from the curb), the percentage riding more than 11 feet from the curb appears to have increased somewhat -- though motorists also parked a few inches farther from the curb.

Wheel paths below 9 feet from the curb are shown in red in the graphs in the report. Wheel paths between 9 and 10 feet are shown in yellow, and those above 10 feet, in green, making use of the standard traffic-signal colors to strengthen further the impression that 9 feet might be OK, and 10 feet is an acceptable, safe number.

The graph below, from the report, gives the percentages of bicyclists' wheel paths at different distances from the curb for the full bike-lane treatment at one measurement site. Note the misleading use of colors.

I have added a horizontal bar across the bottom  to show wheel paths that actually place the bicyclist outside the door zone as determined in the San Francisco study. As in reality, I don't show sharp boundaries. I show the boundary between marginal and outside as less sharp than the inner one. Why? As I've noted already, a bicyclist determined to stay outside the door zone can see the occasional unusually-wide parked vehicle ahead, and merge farther to the left before passing it. But the bicyclist should not similarly merge right before passing the occasional unusually-narrow vehicle, or empty parking space. There is only time for a motorist to overtake if the bicyclist can merge right into several parking spaces in a row which are empty, or occupied by narrow vehicles.

Graph of bicyclist wheel tracks for bike lane
installation near Columbia Street. I have added the
descriptive comments in small print. The horizontal
bar which I have added at the bottom indicates
the real limits of the "door zone" for bicyclists
overtaking typical passenger vehicles.

hampstrt.gif (6434 bytes)

A statement on page 11 of the report contradicts the others as well as the intent of the graph coloring by telling the uncomfortable truth:

For cyclists to travel completely outside the full door zone, the left handle bar would be in the travel lane.

In other words, the wheel path would be at least 11 feet from the curb for bicyclists to be outside the door zone, just as I have described. Except for this one lapse,  the conclusions of the report do not reflect the data, because of conflation of the wheel track with the rightmost extension of the bicycle and rider.

Bicyclists' proper lane position depends several factors besides the dimensions of the door zone. If traveling as fast as the motor vehicles -- either because these are traveling slowly or because the bicyclist is going fast downhill -- the bicyclist should ride in line with them (explanation). Also, before passing a long truck or bus, the bicyclist should merge several feet away from its side so as to be able to brake and drop back if it begins to merge toward the bicyclist. When passing a long stretch of empty parking spaces, the bicyclist may merge somewhat to the right to make it easier for motorists to overtake. When traveling at 5 miles per hour or less, and especially if traffic to the bicyclist's left is stopped, the bicyclist may safely stay within the door zone. So the distances I have indicated as sufficient to avoid the door zone should not be regarded as establishing fixed limits for the bicyclist's road position. They do, however, reflect typical conditions as bicyclists pass rows of parked passenger vehicles and light trucks.

The Cambridge study included questionnaires, and the answers indicate that bicyclists generally prefer the bike lane over other treatments. This result is consistent with the bicyclists' riding in the bike lane. Most bicyclists are excessively fearful of being struck from behind by a motor vehicle, and excessively unconcerned or fatalistic about the dooring hazard. However, bicyclists' preference has no bearing on whether the door zone actually is safe for riding at normal bicycle speeds. Crash statistics demonstrate that it is not.

Now let me try to take the discussion to a higher level.

Improving bicycling conditions on a street like Hampshire Street in Cambridge, the location of the test sites in this study, is not an easy matter. Hampshire Street is 44 feet wide, two-way, two-lane, with parking on both sides. If bicyclists are riding outside the door zone, motorists will have to merge partway into the oncoming travel lane to overtake safely.

As traffic volume increases, the opportunities to overtake become fewer, and bicyclists increasingly delay motorists. Then at even higher traffic volume, the motor traffic becomes congested waiting for traffic signals, and motorists delay bicyclists. Bicyclists can then filter forward slowly on the right of stopped traffic in reasonable safety as long as they do not cross gaps where they may be struck by left-turning vehicles, and do not pass the first vehicle waiting at an intersection. Under these conditions, an edge stripe or bike lane stripe becomes advantageous because it helps keep motorists farther from parked vehicles.

The freest flow of traffic is achieved through a fluid and dynamic interaction among all road users, allowing overtaking whenever it is reasonably safe. Marked lanes regularize the flow of traffic but impose a rigid compartmentalization of space, reducing this fluidity. It is illegal for a motorist to merge across the double yellow line to overtake a bicyclist who is riding outside the door zone, or to merge into an empty bike lane, away from an oncoming vehicle overtaking a bicyclist -- though in fact, both of these practices are common.

The five-foot width of a standard bike lane is intended to be at least marginally sufficient to allow one bicyclist to overtake another, after giving an audible warning. However, when only the leftmost edge of the lane is outside the door zone, then the safe overtaking location is entirely outside the bike lane. Bicyclists may, however, feel pressured to keep right in the bike lane to allow other bicyclists to overtake.

Though crash statistics show that it is preferable to stay outside the door zone, most bicyclists, as the study shows, choose to ride within it. Many if not most bicyclists and non-bicyclists alike tend to take the bike lane at face value, as the one place bicyclists should ride. The City of Cambridge has chosen to accept and formalize this situation by striping a bike lane that is mostly inside the door zone, and in this report, to rationalize the choice by pretending that much of it is not.

The report includes the sentence, on page 3:

Although the Uniform Vehicle Code clearly states that the motorist parking the car is responsible for not opening a car door on the side of moving traffic unless it is safe to do so, the reality is that many motorists have not been well educated about this.

Oddly, the report does not mention that Massachusetts law does not place the responsibility on the motorist, though Cambridge has an ordinance which does. In any case, knowledgeable bicyclists know that riding inside the door zone requires complete trust in the motorists not to open their doors, trust that can't be expected. Knowledgeable bicyclists know that they must ride outside the "door zone" also to avoid other hazards such as pedestrian walk-outs and motorists entering from side streets and driveways (explanation).

The door-zone hazard can be somewhat reduced by educating motorists, but it can be essentially eliminated only by educating bicyclists. Even if motorist education were to succeed in eliminating dooring crashes, motorists would still have to open their doors and walk on their vehicles' street side. Unless bicycle traffic is very light, so motorists can do this entirely within gaps between approaching bicyclists, bicyclists have either to wait or to merge out of the door zone. For this reason, door-zone inconvenience both to bicyclists and to motorists can be avoided only if there is sufficient width for moving motor vehicles to overtake bicyclists who ride outside the door zone.

On a street with marginal width like Hampshire street, I see only three solutions that would accommodate heavy bicycle traffic well, without causing significant delays to motorists. One would be a major reduction in motor traffic. Another would be removal of parking on one side. I'm not ready to answer how either of these would be politically achievable -- though a parking ban at high-traffic times has worked elsewhere. The third solution would be to make the street one-way, so bicyclists mostly travel only on one side. That solution would not work for Hampshire Street because there is no parallel street to carry traffic in the other direction.

Bicyclists' delaying motorists is often described as a major issue with narrow streets, but let's not forget that motorists also delay bicyclists. On a street that has been narrowed by parallel parking, parked motor vehicles reduce the street's carrying capacity and cause the delays for both bicyclists and other motorists. I have read somewhere, and I wish that I could cite the source, that the use of street width for car storage is debatable when it interferes with the main purpose of the street as public travel space. Let's not also forget that bicycle use also benefits motorists, primarily by reducing parking demand in high-density areas.

All in all, I would be much more satisfied if the City would acknowledge, rather than to dismiss and obfuscate,  the problems with its door zone bike lanes. Acknowledging the problems might lead to bicycle-program improvements by turning attention to solutions that work, and to problems that admit of practical solutions.

Ever since I was a member of the Cambridge Bicycle Committee in the mid-1990s, I have been asking for a better alternative to Cambridge's use of one-way signs to keep through traffic out of the residential neighborhoods (see my parting letter to the Bicycle Committee, from 1996). These signs prevent through travel on neighborhood streets, and are more disadvantageous for bicyclists than for motorists. Cambridge has even created more such problems in recent years, with the closing off of Webster Street and the reversal of direction of the east end of Franklin and Green Streets. Other American cities -- Berkeley, Palo Alto, Seattle, Madison, Eugene for example -- use a "bicycle boulevard" approach with bicycle-permeable barriers, diverters, small traffic circles and contraflow bike lanes to reduce through motor vehicle travel on local streets, while still allowing motor vehicle access and through bicycle travel.

No nearby streets, however, parallel Hampshire Street and a couple of other main east-west arterials in Cambridge, and so any solutions must be found on these streets rather than by finding alternative routes.

Stepping back even further in the discussion, is the motivation and intent of bicycle programs to provide the best possible conditions for an existing or anticipated bicycling population, or is it to attract more people to ride bicycles, at the expense of turning a blind eye to safety issues? Cambridge's bicycle program has tilted heavily toward the promotional approach, with a number of facilities that it promotes as attractive to less skillful bicyclists, but that have practical problems which have been described in the engineering and safety literature.

Are we afraid to admit that bicyclists sometimes delay motorists, or to point out that motorists also delay bicyclists? Will we foster the excessive fear of being struck from behind, and ignore the real hazard of car-door collisions, because this misconception plays well with the uninformed public? Do we assume that motorists' convenience comes first, so we will settle for second-class bicyclist accommodation right from the start?  Do we accept door-zone bike lanes rather than to point to the real -- though politically difficult -- solution of moving parking off the street?

And all in all, the proof of the pudding is in the eating. Is bicycle use increasing in Cambridge? Are crash rates decreasing? Is congestion decreasing? Crash data is collected by the police department. I assisted with bicycle traffic counts when I was a member of the Cambridge Bicycle Committee in the early- and mid-1990s, and Cambridge has continued to conduct such counts.

I still see bicycles used for transportation primarily by the same type of college and university student and recent ex-student population as 30 years ago. There are a few older adult holdovers like me, and a small but growing population of low-income people who can not afford other transportation. I have seen little movement toward bicycle use by the general population, either among children -- who would benefit most from traffic calming on the residential streets -- or among adults. Promotion of bicycling has been much more successful in several other American cities. A Cambridge city official told me twice, most recently in August of 2006, that bicycle use in the city had not increased in the more than ten years since the first bike lanes were installed. (Update as of 2007: Cambridge claims a 70% increase. I haven't seen the data on which this claim is based).

I have no information yet as to crash rates, but I can't imagine that they have declined singificantly, as nothing has occurred that would have a major effect in that direction. Data collected by the Massachusetts Executive Office of Transportation rates Cambridge as having the highest bicycle crash rate of any city or town on a population basis, but on the other hand, bicycle use in Cambridge is heayy, thanks to the high percentage of college and university students.

A "visionary" social-engineering approach which stresses encouragement while it sets aside real issues of safety and efficiency, and turns its back on scientific veracity, can fairly be criticized as "end justifies the means" politics, window dressing that exposes citizens to real hazards without their informed consent, in pursuit of what is seen as the greater good. If real improvement is not achieved, all this also is in vain.

I am concerned, in the long run, as to what will happen in Cambridge as the rising cost of motor fuel either slowly or suddenly leads more people to take up bicycling. Will the city be ready? Will the measures the city is putting in place now be seen as having prepared well, or will they be seen as mostly irrelevant?  Cambridge has done certain things such as installing bicycle parking which are unquestionably beneficial for bicyclists. But has Cambridge, with its active bicycle program, actually done more to improve bicycling conditions on its streets than Boston, which has done nearly nothing at all? Time will tell.


How Pavement Markings Influence Bicycle and Motor Vehicle Positioning: A Case Study in Cambridge, MA, by Ron van Houten and Cara Seiderman

San Francisco Shared Lane Marking Study -- includes data on car-door opening widths and describes an alternative to bike lanes

Two papers, Bicycling and On-Street Parallel Parking and AASHTO and Door Zone Bike Lanes, by Wayne Pein.

A survey of car-door opening widths, which corroborates the San Francisco data.

The City of Chicago Bike Lane Design Guide, a widely-used but controversial reference for design of streets with bike lanes.

My review of The Bike Lane Design Guide. Find out why it is controversial. The design drawings shown are of a street with the same geometrics as Hampshire Street.

Two papers, Bicycling and On-Street Parallel Parking and AASHTO and Door Zone Bike Lanes, by Wayne Pein.

A survey of research into car-door collisions by the author of this page.

Examination of bike lanes and other facilities in Cambridge, Massachusetts, by the author of this page.

Hampshire Street photos and comments from two stages of the study period.

The Dana Laird fatal crash on Massachusetts Avenue.

[Home page]