The Consumer Product Safety Commission held a conference on bicycle nighttime protective equipment in November, 1995. Both users and manufacturers made presentations to that conference, and some manufacturers supported their presentations with scientific papers. Therefore, it is necessary that these papers be reviewed to determine the extent to which these papers support the manufacturers' presentations and the extent to which these papers provide useful information about the prevention of nighttime car-bike collisions. The CPSC has sent copies of these papers to me, and perhaps to others.
The following analysis is divided into four parts. The papers are divided into those that deal with the nighttime conspicuity of 2-wheeled vehicles, those that deal with other nighttime conspicuity subjects, and those with other subjects. Following individual reviews of each of these papers is a summary of all of the material, discussed with respect to the question of reduction of nighttime car-bike collisions, and particularly of the equipment that might be used to accomplish this.
By Albert Burg and Jinx Beers, School of Engineering and Applied Science, UCLA, and conducted under contract from the 3M Company.
Phase I of this study determined the detection distance under low "Sealed Beam" headlamp beams of reflectorized motorcycle tires when positioned in 6 different positions relative to the lighted automobile. The six positions were 3 with the tires at right angles (90 degrees) to the headlamp beams, 3 with the tires at 40 degrees. In each set of three, one test was with the tires at the right edge of a 2-lane roadway, one with them in the center of the right lane, and one with them at the left side of the roadway.
The mean detection distances for the tires varied from 1209 feet to 1779 feet, depending on the type of tire and position. The tires were close to the road and far away, not in the position off to one side and closer at which a motorist has to detect the presence of a bicycle with which he would collide. The the maximum angle between straight ahead and the position of the tires was less than 1 degree. This data is absolutely worthless as respects the problem of detecting a bicycle approaching on an orthogonal path.
Phase II of this study used a standard distance of 500 feet from automobile to motorcycle. The scene represented an unlit rural intersection. A motorcycle was sometimes presented, and sometimes at the left margin of the roadway and sometimes at the right margin, as if entering the roadway from the cross street. The subjects were given 3-second glimpses of the scene to see whether they detected the presence of a motorcycle and, if so, its position. The motorcycles with reflectorized tires were nearly always seen, while those without were seen only about half the time. The maximum angle between car's path and the test object was 2 degrees, and therefore does not pertain to the problem of detecting a bicycle approaching on an orthogonal path.
The only thing that test like this demonstrate is that if you are fool enough to start from a stop to pull out in front of a vehicle that has the right of way you are likely to get smashed, night or day makes little difference. If the bicycle were approaching the intersection, typically an urban one, the bicycle would be moving toward the intersection and the car would be going slower, at urban speeds, or waiting at a stop sign. Then the bicycle must be detected at far greater angles than are tested here, say 70 degrees or so.
By Albert Burg, Ph.D., & Slade Hulbert, Ph.D., supported by the 3M Company
This employs much the same method as the study of motorcycle wheels reported above, except with bicycle wheels, and with an additional moving test. In the moving tests, the bicycle wheels were moved either across the roadway, 500 feet in front of the automobile, or diagonally across it at the same distance. The test was to determine the difference in recognition of movement between spoke reflectors and reflectorized tires. Naturally, subjects viewing the reflectorized tires correctly identified their motion more frequently.
However, these tests were made under conditions under which it is impossible for a collision to occur. They have no relevance to the conditions immediately before a collision during which it would be possible for the motorist to avoid the collision, which are the only conditions that apply to preventing car-bike collisions through the use of reflective devices of any type.
This is a study of reflectorized bicycle tires under two conditions, crossing the roadway and bicycle turning left. The crossing study shows that a bicycle with reflectorized tires, positioned crosswise in the roadway, shows nicely at long distances, distances at which the bicycle would be safely across long before the car reached the bicycle. Therefore, this crossing study presents no data useful for preventing accidents in which the bicycle is approaching the intersection. The turning study shows that by the time that the bicycle has turned 30 degrees its reflectorized tires can be seen by the approaching motorist. It this position, the front of the bicycle is in the path of the left front corner of the car. The bicycle will, at a speed of 10 m.p.h., clear the car's path in about 1.5 seconds. After that time there is no danger of collision. Therefore, it is only during this time interval, and the matching distances, that the motorist can avoid a collision that would otherwise occur. The motorist can neither stop nor steer clear in this 1.5 seconds. Therefore, equipment that becomes visible only as late as 30 degrees into the left turn is useless for avoiding a car-bike collision.
By F. Stoovelaar, ANWB, & R. E. Groot, IWACC, 1976
The greater part of this paper discusses different types of rear device, and concludes that a red rear lamp with the rear part of the rear mudguard covered with amber reflective material is the best for Dutch purposes. Only 3 of 26 pages are devoted to side reflectors. The authors don't like spoke reflectors because, so they say, the bouncing image doesn't look like a bicycle. Therefore they like reflective tires, which, so they say, are being adopted by the United States. There is no information about optical performance, performance under road conditions, or effect on car-bike collisions.
F. Stoovelaar, ANWB, & R. E. Groot, IWACC
This concerns mostly daylight conspicuity of motorcycles and is therefore irrelevant to the subject of bicycles at night.
This is probably the English translation of the text of a Dutch original, that is intended to be used with the pictures in the original document. At any rate, while it shows some line drawings it also refers to pictures that are not present. It is a collection of thoughts about conspicuity and recognition, a part of which applies to bicycles. There are several pages on color perception, the optics of fluorescent materials and of retroreflectors.
There are several pages theorizing about creating recognizable nighttime visual patterns for different types of vehicles. The authors promote the theory that being able to identify the type of vehicle makes it more likely to be seen. Therefore they like reflective tires for 2-wheeled vehicles because these, when seen, clearly identify such vehicles. They show some pictures of bicycles with reflective tires that have come from a film that 3M used to promote such tires. The authors definitely do not like visual clutter at nighttime, such as is produced by merely adding luminous elements without a specific pattern, and they hope that governments will do something to limit it.
There are other musings also. One is about the different views of the world possessed by drivers of four-wheel and two-wheel vehicles. Another is a very dissatisfied description of the Dutch bikeway program, saying that it provides poor and discriminatory service for cyclists.
This collection is a hodge-podge with a little that is useful for explaining the optics of retroreflective and fluorescent devices.
However, despite its advocacy of reflective tires, nowhere does it address the issue of whether front or side reflectors of any type, whether spoke reflectors or reflective tires, are visible to the motorist who may cause the car-bike collision at the time at which that motorist can prevent the collision.
A. Blokpoel, Foundation for Scientific Research on Traffic Safety, 1989?
This study compares the ratios of daytime to nighttime car-bike collisions in the three years before mandatory adoption of reflective tires and in the two years after. This analysis is compared to a similar one for pedestrians, to distinguish between factors that influence accidents to both and factors that influence only cyclists. The reduction in the proportion of nighttime car-bike collisions was greater than the reduction in the proportion of car-ped collisions, leading to the conclusion that the adoption of reflectorized tires reduced nighttime car-bike collisions about 4%.
However, the following cautions need be considered. There is no indication of the year-to-year variation in the proportion of nighttime accidents. Therefore, there is no estimate of the probable error of the estimate (confidence limits). Also, as the author indicates, the effect seems considerably stronger for the types of collisions in which one would expect the least effect from side reflective devices than for the types of collision in which one would expect the most effect from side reflective devices. This may well bear on the stated correlation between the use of side reflectors and the use of headlamps, because the headlamps would reduce not only the collisions from the side but also the collisions from the front. Finally, there is the question that can be answered only by someone with accurate and detailed knowledge of the Dutch cycling scene at the time, which is whether some other change in those years would differentially affect nighttime and daytime cyclists.
Helmut T. Zwahlen, Dept. of Industrial and Systems Engineering, Ohio University.
Zwahlen's tests demonstrated two things. First, the ability to see objects at night falls off rather rapidly as the object moves away from the subject's line of sight. Presumably, that means that peripheral vision falls off more rapidly with increasing angles at night than in the daytime. Second, that drivers tend to look where they are going, both on straight roads and on curved roads. Therefore, objects that are to be seen beside the road, where the driver does not intend to go and therefore where he does not look, must be pretty bright. More importantly, drivers seem to look insufficiently far ahead, so that objects that are further around a curve than they are looking, and are therefore at an angle to their line of sight, need to be very bright to be seen in time to stop before reaching them.
This type of analysis is appropriate to the question of rear lamps versus rear reflectors, but not to the question of side reflectors or reflectorized tires.
Helmut T. Zwahlen, Proceedings of the Human Factors Society, 1986
This is another account of the above research.
Helmut T. Zwahlen, Proceedings of the Human Factors Society, 1988
This study shows that one must be much closer than the initial detection distance before one can determine the shape of an illuminated object, and the more similar are the shapes of the different test objects the closer they must be to be differentiated.
I do not see the relevance of this study to the prevention of nighttime car-bike collisions.
Helmut T. Zwahlen, Dept. of Industrial and Systems Engineering, Ohio University. Transportation Research Record 1327
This study compares the recognition by color against recognition by shape, and concludes that the color of modern highly saturated color reflective surfaces leads to earlier recognition of the import of the sign than does the shape of the sign.
I do not see the relevance of this to the subject of preventing nighttime car-bike collisions.
Helmut T. Zwahlen, Proceedings of the Human Factors Society 1989
In this study Zwahlen concludes that for detection distance, complete reflectorization is best, but for shape recognition, reflectorizing only the borders of the sign is best.
I do not see the relevance of this study to preventing nighttime car-bike collisions.
Theodore J. Szczech and Susan T. Chrysler; Transportation Research Record 1456
This study proposes that a specific pattern of headlamp illumination be adopted for research calculations, based on the 15th percentile of the low beams of current U.S. headlamps. The proposed distribution is useful when comparing the brightness of reflectors at different positions.
Helmut T. Zwahlen and Una Devi Vel; Transportation Research Record 1456
This study concludes that fluorescent colors are better than nonfluorescent colors, and recommends fluorescent yellow as best for recognition and fluorescent orange as best for recognition.
Cathy L. Antonakos; Transportation Research Record 1502
This study presents nothing of relevance to prevention of cycling accidents, and little detail that is useful for other cycling programs.
William W. Hunter, Wayne E. Pein, & Jane C. Stutts:L Transportation Research Record 1502
This is the summary of the research that is intended to be a repeat of that done by Cross and Fisher in 1976. The full report will be issued by the FHWA. The authors say that their results are very similar to those of Cross and Fisher. The summary rarely gives sufficient detail to tell the differences from the 1976 results, and the summary lumps together motorist overtaking and motorist left turn car-bike collisions, which are of entirely different types.
However, the summary does give the numbers for the two types that are most important with respect to the facilities controversy. These are Cross's type 13, motorist overtaking who did not see the cyclist, and type 16, motorist overtaking who misjudged the distance. These two types total 76 out of 2,990 collisions, or 2.5%. We know that some of the type 13 were caused by inadequate nighttime equipment, so not all of these are valid indicators of the danger of same-direction traffic, and that these are over-represented in rural conditions where bikeways are not likely to be built. All in all, the study reinforces the principle that we need to work on turning and crossing conflicts rather than same-direction traffic conditions, and that we need to work on improving the skill of cyclists.
The complete study, if it has the appropriate data, may be used in the same way as the Cross and Fisher study to estimate the proportions of the different types of car-bike collisions that are caused by darkness. Cross and Fisher's data show that in about 75% of the car-bike collisions probably caused by darkness the car approaches the bicycle from ahead or from the side. From the comments in the summary, it is likely that this proportion will not change significantly.
National Highway Traffic Safety Administration
There is insufficient detail in this report, and probably in the final one also, to provide information that is useful in deciding how to reduce nighttime car-bike collisions.
U. S. Consumer Product Safety Commission, 1994
This is the CPSC's own study that has been extensively reviewed elsewhere. While the data in this study about types of accidents may be valuable, its data about hours of use appear to have serious errors. That is why a great many of its conclusions that are based on hours of use, such as relative hazard rates or average speed of travel, greatly disagree with other data that is much better supported.
I think that none of the data in this study is useful for evaluating the effectiveness of the all-reflector system, and certainly none of it applies to evaluating the effectiveness of the different parts of the all-reflector system or of competitive equipment such as headlamps.
None of the documents classified as "other subjects" provides information that is usable for the purposes of determining the effectiveness of the all-reflector system, or of its separate parts, or of competing systems such as headlamps.
Zwahlen's papers on the detection and recognition of different types of reflectorized sign have no direct usefulness in determining the effectiveness of the all-reflector system, or of its parts, or of competing systems such as headlamps. The bicycle is not capable of carrying in the appropriate locations the large areas of reflectorized material that would make this work relevant. Zwahlen emphasizes viewing the objects with peripheral vision and concludes that if detection and recognition is to be reliably accomplished these must be much brighter than when the objects are viewed directly. This analysis might be of use in the particular case of a motorist approaching a bicycle from the rear on a curve, and such analysis might conclude that for this situation a rear lamp is better than a rear reflector, but there has been no evidence that this situation is particularly frequent in nighttime car-bike collisions.
The proposal for a standard distribution of headlamp light, based on the 15th percentile of modern U.S. headlamps, for calculations regarding nighttime conspicuity, is good in principle. Whether the particular recommended distribution is most appropriate is a question that is best answered by others.
Of the papers that consider the conspicuity of two-wheeled vehicles, only one attempts to provide information about the effectiveness of side reflectors in preventing nighttime car-bike collisions. While this paper presents a small decrease in nighttime car-bike collisions, it raises disturbing questions about whether this reduction, if real, was caused by the side reflectors.
Of the papers that directly discuss side reflectors, none provides any information about the method by which side reflectors would prevent car-bike collisions or data showing that they do so. These papers are full of praise for the theory that recognizing a bicycle by seeing its rings of light improves the probability of seeing it at all (questionable in itself), but they ignore, either through ignorance or through partiality, the question of whether this view of the bicycle would be presented to the motorist at the time that the motorist has the responsibility and the ability to avoid the collision.
The study by the International Study Group comes nearest to doing so with its test of the conspicuity of the front wheel of a bicycle that is turning left. However, as discussed above, the wheel becomes conspicuous too late to avoid the collision. All the other tests that are reported are tests of conspicuity in conditions under which no collision could occur.
There is no evidence that side reflectors prevent car-bike collisions, either by presenting a reasonable engineering analysis of collision situations or by showing the effect. In addition, there is no evidence, and no effort to produce any, that counters the analysis, given over twenty years ago, that in typical collision situations side reflectors cannot prevent car-bike collisions and reliance must be placed on the headlamp to do so.
Return to: John Forester's Home Page Up: Lights