Headlight glare

1

Countermeasures for Reducing the Effects of Headlight Glare

Prepared by:

Douglas MaceThe Last Resource

Bellefonte, PA 16823

Philip GarveyThe Pennsylvania Transportation Institute

Richard J. Porter

Richard Schwab

Werner AdrianUniversity of Waterloo

Prepared for:

The AAA Foundation for Traffic Safety1440 New York Avenue, N.W.

Washington, D.C. 20005202-638-5944

www.aaafoundation.org

December 2001

Pho

to b

y S

cott

Osb

erg

This study was funded by the AAA Foundation for Traffic Safety. Founded in 1947, the AAAFoundation is a not-for-profit, publicly supported charitable research and educational organization dedicated to saving lives and reducing injuries by preventing traffic crashes.

This peer-reviewed report was created in response to the increasing public interest in the problemof headlight glare. New headlight technology, an aging population, and innovations in vehicle designhave all contributed to increased problems with glare. This in-depth, comprehensive report examines the factors that contribute to glare problems and looks at a wide range of possible countermeasures.Although the researchers could not identify any single, simple solution, the report does provide anumber of helpful suggestions for countermeasures that can lessen the discomfort and perceived

danger caused by headlight glare.

Funding for this study was provided by voluntary contributions from the American AutomobileAssociation and it affiliated motor clubs; from individual AAA members, and from AAA club-affiliatedinsurance companies.

This publication is distributed by the AAA Foundation for Traffic Safety at no charge, as a publicservice. It may not be resold or used for commercial purposes without the explicit permission of theFoundation. It may, however, be copied in whole or in part and distributed for free via any medium, provided the AAA Foundation is given appropriate credit as the source of the material. The opinions,findings, and conclusions expressed in this publication are those of the authors and are not necessarilythose of the AAA Foundation for Traffic Safety or of any individual who peer-reviewed this report. The AAA Foundation for Traffic Safety assumes no liability for the use or misuse of any information, opinions, findings, or conclusions contained in this report.

If trade or manufacturer's names or products are mentioned, it is only because they are consideredessential to the object of this report. Their mention should not be construed as an endorsement. TheAAA Foundation for Traffic Safety does not endorse products or manufacturers.

3

Foreword

© 2001, AAA Foundation for Traffic Safety

4

The AAA Foundation for Traffic Safety would like to acknowledge the help ofseveral experts who reviewed this document before publication:

Robert OwensNew Holland, PA

Rollin C. BibleApple Valley, MN

Frank SchieberHeimstra Human Factors Laboratories, Vermilioin, SD

Mike PerelNational Highway Transportation Safety Administration, Washington, DC

CONTENTS

FOREWORD ................................................................................................................................3

EXECUTIVE SUMMARY ..........................................................................................................7

CHAPTER 1Introduction ........................................................................................................................11

CHAPTER 2 Factors that Contribute to Headlight Glare at Night ....................................................15

CHAPTER 3 The Problem of Headlight Glare ......................................................................................23

CHAPTER 4 A Comprehensive Review of Glare Countermeasures ....................................................29

CHAPTER 5Countermeasures that Reduce the Intensity of Glare Source ........................................37

CHAPTER 6Countermeasures that Reduce Illumination Reaching the Driver’s Eye......................53

CHAPTER 7Countermeasures that Increase the Glare Angle ............................................................73

CHAPTER 8Countermeasures that Indirectly Minimize the Effects of Glare ..................................77

CHAPTER 9Summary and Recommendations ....................................................................................91

REFERENCES ..........................................................................................................................99

5

ERRATA SHEET for Countermeasures for Reducing the Effects of Headlight Glare Prepared for the AAA Foundation for Traffic Safety by Mace, Garvey, Porter, Schwab, and Adrian, published December 2001 Clarification and correction of misleading statements added 12-18-02 p. 31: This report states that, "Several companies offer halogen bulbs that are coated blue to look like HID bulbs. These bulbs are not legal.…" It also states that the "… expensive HID conversions are illegal in the United States.” The foregoing statements are somewhat misleading because they imply that all aftermarket products are illegal. In fact, if a manufacturer’s aftermarket product, such as a conversion kit, complies with the performance specifications in FMVSS 108, the product is perfectly legal. On the other hand, AAA Foundation research indicates that many aftermarket products that do not meet FMVSS 108 are still being sold, including but not limited to coated bulbs.

p. 31: "Other conversion kits that use true HID bulbs (usually xenon lights) are available, but they are limited in practice to systems that use single-filament bulbs.…" This statement is incorrect. HID lamps do not have filaments. Instead, they generate visible light using a discharge arc. --

Driving an automobile is primarily a visual task,and vision contributes as much as 90% of the infor-mation needed to drive (Alexander and Lunenfeld1990). Night-time driving poses a special chal-lenge, since even at night drivers need to be able tosee traffic control devices, lane lines, vehicles,pedestrians, animals, and other potential hazards.Artificial lighting can illuminate the roadway, buttoo much light or improper lighting may result inglare, which causes visual discomfort and a dimin-ished ability to see the environment.

There are only two practical methods for night-time lighting: fixed overhead lighting and vehicleheadlights. While the number of roads with fixedoverhead lighting increases each year, this form oflighting is expensive and cannot be relied upon asthe only means for providing night visibility.From its inception, the use of headlights on auto-mobiles has involved a compromise between pro-viding enough light for drivers to see the roadahead and avoiding the excessive light that pro-duces glare.

Changes in headlamp designs that affect lightintensity, beam pattern, and aiming have signifi-cantly improved night vision on the highway.Technology has brought changes to headlights,interior surfaces (including mirrors), and the high-way environment that directly reduce glare or indi-rectly reduce the effect of glare on the driver.However, every change has involved a tradeoffwith hidden costs. For example, lowering head-lamps may reduce glare but can result in a loss offorward visibility.

The framework for all subsequent discussions ofglare in roadway lighting was created by Holladay(1926) in the USA and by Stiles (1928 - 29) in theUK. Holladay originated the concept of “disability

glare,” glare that decreases a driver’s ability to seeclearly. Stiles showed that disability glare iscaused by the scattering of light in the optic mediaof the eye, rather than occurring in the opticalnerve, as Holladay had assumed. Holladayacknowledged Stile’s explanation but noted that itdid not explain all of glare’s effects. This exchangeapparently began the present distinction betweendisability glare and discomfort glare.

This report provides the reader with a workingknowledge of glare and the methods used to meas-ure and control glare. It is a rather technical reportaimed at engineers and experts in lighting and traf-fic safety.

Glare occurs when visual field brightness isgreater than the luminance to which the eyes areadapted. It can be caused by direct and indirectlight sources. Discomfort glare causes discomfort,annoyance, fatigue, and pain. Disability glare pro-duces a reduction in the visibility distance of low-contrast objects. The elderly, people with light-col-ored eyes, and those suffering from cataracts areespecially sensitive to disability glare. Glare atnight can be mitigated by design changes in road-ways, automobiles, and vehicle lighting systems.

Countermeasures work in four ways, by:

1) Reducing the intensity of the glare source;2) Reducing the illumination reaching the

driver’s eyes;3) Increasing the glare angle; and 4) Indirectly minimizing the effects of glare.

Some glare countermeasures may require federalrulemaking, while others can be implemented byhighway agencies, vehicle manufacturers, andmotorists.

7

Executive Summary

Countermeasures requiring government involve-ment include:

• Enforcement of headlight aim standards • Changes in beam photometric distribution• Ultraviolet headlights• Polarized headlights

Some 50 percent of vehicles have their headlightsmisaimed and the problem increases as the vehicleages. Because misaim magnifies all other prob-lems with the low-beam photometric pattern, mis-aim is clearly the place to begin finding effectivecountermeasures. In addition, since public com-plaints seem to be mostly about glare from newervehicles, headlamp aim in vehicles with HIDlamps would be a logical place to apply correctivesolutions.

Federal rulemaking is required to make any majormodifications to the low-beam photometric distri-bution. Every effort to reduce headlamp illumina-tion or change the beam pattern has been met withconcern about visibility, and every effort toincrease illumination has been met with concernabout glare. Research does not support change ineither direction or use of the current beam pattern.The present standard is a compromise that hasevolved over time and works reasonably well; therisks of making any major change appear toogreat. Therefore, any effort to develop counter-measures for glare should focus on one or more ofthe other alternatives discussed in this report.

Ultraviolet lighting has some potential to reduceheadlight glare, but only indirectly. This technolo-gy should and will be pursued primarily becauseof its benefits to vision, particularly vision underadverse conditions such as rain, snow, and fog, butit is not likely to replace conventional lighting andso cannot offer a realistic countermeasure to glare.

Of all the countermeasures discussed in this report,polarized lighting is the one that could eliminateglare entirely and make the nighttime road afriendlier place to travel. With the advent of HIDlamps, the very technology that has heightened theglare problem offers the best ultimate solution.

With polarized lighting, the tradeoff between visi-bility and glare is resolved. The only real obstacleto pursuing this countermeasure further is the diffi-culty of implementation.

Countermeasures that can only be initiated by ahighway agency include the following:

• Wide medians and independent alignment• Glare screens• Fixed roadway lighting

Wide medians, independent alignment, and glarescreens are all effective in eliminating glare fromoncoming vehicles, either by increasing the glareangle so that its effect is reduced or by blockingglare illumination completely. Wide medians andindependent alignment are used primarily in ruraland suburban areas, where a wide right-of-way isavailable, while glare screens are used primarily inurban areas, where the right-of-way is narrower.Wide medians and independent alignment are partof the design process and generally cannot beadded after construction. Glare screens are aremedial treatment and can be installed when theinitial design does not allow sufficient right-of-way for a wide median or when traffic volumeincreases and glare problems grow worse.Practical limits on height restrict the use of glarescreens to roadways that are relatively flat and havegentle curves. Since the screens’ effect on crashes isnot documented, the cost must be justified by issuesof comfort and possible safety benefits.

Like wide medians and independent alignment,fixed roadway lighting is often justifiable on thebasis of crash reduction alone. However, fixedlighting requires access to electrical power and isgenerally restricted to urban areas. Installation offixed lighting also depends on budgets and reflectsvarying criteria used by highway agencies. Thequestion of whether low-volume roads should beilluminated to minimize glare is not easilyanswered. Drivers are certainly more comfortabledriving on illuminated roads, and this benefit alonemight justify lighting more roads. However, thisshould be a local decision, made with an under-standing of local resources and priorities.

8

It is unlikely that all roads with glare problemswill ever be illuminated, so other solutions mustbe found. A variety of road improvements, includ-ing alignment, lane width, surface markings, andso on, may reduce the discomfort experiencedfrom headlight glare.

Countermeasures that are primarily the responsi-bility of industry are:

• Headlamp height • Color-corrected headlamps• Headlamp area• Adaptive headlamps

Of these four countermeasures, limiting headlampheight is the only one that could be implementedquickly and with minimal cost. Lowering head-lamp height is a promising “no cost” countermea-sure to glare, but its impact is limited to mirrorglare. With light trucks (pickups, full-size vans,and sport-utility vehicles) representing 50% of allvehicle sales, lowering the headlamp height ofthese vehicles should be pursued. Lower head-lamp height on large trucks may be more problem-atic, given visibility concerns.

Color-corrected headlamps, while offering a low-cost solution, require a significant amount of addi-tional research before adoption.

Headlamp area is related to glare discomfort, withlarger headlamp size causing less discomfort.Projector-style HID headlamps may be contribut-ing to discomfort glare because their surface areais smaller than standard headlamps and so theirluminance is much higher.

Adaptive headlamps, while theoretically promis-ing, have numerous design, cost, and regulatoryobstacles that need to be surmounted. While it isclear that adaptive headlamps offer significantimprovements to visibility, for example on curves,it is not at all certain what their effect would be onglare. Cost and maintenance issues also need to beresolved before adaptive headlighting becomes apractical countermeasure to glare.

Countermeasures that are under the control ofmotorists are:

• Reduced night driving• Night-driving glasses• Anti-glare mirrors• Corrective vision solutions

Drivers who are especially bothered by nighttimeglare should try to reduce their exposure to it bydriving less at night. Scratched or dirty eyeglassesand damaged contact lenses make the problemeven worse. Glare-sensitive drivers who mustdrive at night should learn coping strategies bytaking a driver improvement course, such as thoseoffered by AAA, AARP, and the National SafetyCouncil.

Research shows that, for most individuals, night-driving glasses are not an effective countermea-sure. While discomfort is reduced, so is visibility.This conclusion applies to both full eyeglasses andhalf-glass analyzers that allow the driver to lookthrough the analyzer only on demand. Althoughone study suggested that discomfort glare had littleeffect on driving performance, the measurementsof performance were entirely psychomotor and notvisual. Research is needed to understand the rela-tionship, if any, between discomfort and perform-ance. Although we know how driving affects per-ceptions of discomfort, we do not know how dis-comfort glare affects eye fixations, attention, andfatigue. There is research that suggests that driv-ers’ eyes are attracted to light but are drawn awayfrom glare sources. Additional research is neededto support or reject the assumptions being madeand the conclusion that night-driving glasses(including half-glass analyzers) have no value foranyone driving at night.

Anti-glare mirrors, together with limits on head-lamp height and enforcement of standards forheadlight aiming, are all that is needed to controlmirror glare from passing or following vehicles.Until all vehicles are equipped with some type ofautomatic glare-reduction mirrors in both the rear-view and left side positions, drivers need to be

9

encouraged to use the night setting of their interiormirrors and to aim the left outside mirror so that itdoes not reflect directly into their eyes. The ques-tion of what drivers need to see in their rear-viewmirrors needs further study so that automatic glarereduction mirrors can be properly designed.

Vision correction should be strongly encouragedfor safety reasons, but provides only a minimalreduction in glare.

While one might think that maintenance of head-lamp aim is a countermeasure drivers could imple-ment, there is generally little incentive to do so.Drivers will correct their misaimed headlampsonly when their own ability to see is compro-mised, and this situation usually does not create aglare problem. If misaimed headlamps do causeglare, the driver is likely to think they are fine,because for the driver increased glare also meansimproved visibility. Misaimed headlamps are themost problematic source of headlight glare, sincecorrecting them requires the uniform enforcementof headlamp aiming standards.

Two of the potential countermeasures discussed inthis report are being studied in large ongoingresearch programs. Adaptive headlamp technologyis being developed and strategies for its implemen-tation are being devised with the support of sever-al European countries and manufacturing firms. Inthe U.S., a comprehensive research program forthe development and implementation of UVAheadlamps is underway. Both adaptive headlight-ing and UVA headlamps are more likely to benefitvisibility than to offer any comprehensive solutionto headlight glare.

Topics that should be investigated include polar-ized lighting; the effects of the spectral content oflight on visibility and discomfort at mesopic adap-tation levels; the relative effectiveness of elec-trochromic, photochromic and neodymium mirrorson glare sensation and rearward visibility; theidentification of the population of drivers mostaffected by glare and the reasons for their prob-lems with headlight glare; and a more completedescription of the behavioral effects of discomfortglare.

10

Driving an automobile is primarily a visualtask. By one estimate, as much as 90% of theinformation that drivers gather is received visually(Alexander, G. and Lunenfeld, H. 1990), andwhatever the actual percentage may be, the impor-tance of the visual system to driving can not bedoubted (Sivak, 1996). However, in order for thevisual system to detect, attend to, and recognizeinformation, there must be adequate lighting.Drivers require enough lighting at night to see avariety of objects on the highway, including trafficcontrol devices, lane lines, vehicles, pedestrians,animals, and other potentially hazardous objects.However, too much light or improper lighting canresult in glare, which can be a major problem bothin terms of the ability to see and visual comfort.

There are only two practical methods of light-ing the highway system at night: fixed overheadlighting and vehicle head lighting. While the frac-tion of roads with fixed overhead lighting increas-es significantly each year, this form of lighting isexpensive and can not be relied upon as the onlymeans of providing for night visibility. Head light-ing, from its inception, has involved a compromisebetween providing sufficient lighting for driversto see (with adequate preview time), and avoidingexcessive light that might produce glare. Thesetwo goals have been translated into standards inthe form of minimum requirements to provide vis-ibility and maximum limitations to control glare.

Progressive improvements in headlighting andnew technologies have increased night visibilityand reduced the impact of glare, but any changesshould be carefully considered before implementa-tion. Changes in headlamp designs that affect lightintensity, beam pattern and aiming have signifi-cantly improved night vision on the highway.

Along with improvements in headlight systems,glare resistant interior surfaces, glare-reducingmirrors, and changes to the highway environmenthave either directly reduced glare or indirectlyreduced the effect of glare on drivers. However,every change has involved a tradeoff with hiddencosts. For example, lowering headlamp mountingheights has a minimal monetary cost, but thischange may result in a loss of forward visibility;making other changes, such as installing dynamicheadlight aiming systems, may be very costly. Tobest manage such costs, it is necessary to have agood understanding of the glare problem and ofthe importance of various aspects of the problem.Otherwise, the old adage, “be careful what you askfor, because you might get it,” could apply.

Research on glare has a long history, markedby the competing perspectives of researchers in aninternational community. The seminal work, set-ting the framework for all subsequent discussionsof glare in roadway lighting, was done byHolladay (1926) in the USA and by Stiles (1928–-29) in the UK. Holladay is the acknowledgedoriginator of the concept of what is now known asdisability glare, the glare that results in a loss ofvisibility. Stiles showed that disability glare iscaused by the scattering of light in the optic mediaof the eye, rather than cross-talk in the opticalnerve, as Holladay had assumed. Holladayacknowledged Stile’s explanation to be true, butnoted that such scattering did not explain all ofglare’s effects. This exchange was apparently the beginning of the present distinction in glareresearch between disability glare and discomfortglare. Whereas disability glare impairs the eye’sability to distinguish small changes in brightness,discomfort glare results in discomfort, sometimescauses fatigue, and may even produce pain.

11

CHAPTER 1

INTRODUCTION

The purpose of this report is to provide thereader with a working knowledge of glare and ofthe methods that may be used to measure and con-trol glare’s effects. In Chapter 2, the two types ofglare are defined and key factors are identifiedthat contribute to each, based on the most recentscientific research. This information is essentialfor understanding the potential effects of proposedcountermeasures to glare. Chapter 3 discusseshow headlight glare impacts night driving. Thatdiscussion should help in making an accurateassessment of the extent of the glare problem andso help determine whether the inevitable cost of aproposed countermeasure is worthwhile—are thenegative effects of headlight glare so bad that theirresolution can justify the efforts required? Chapter 4 contains a discussion of the source ofglare in the driving environment and a comprehen-sive review of glare countermeasures. Chapters 5through 8 describe specific types of potentialcountermeasures, explaining why each counter-measure is thought to be effective, giving a reviewof the relevant research (including the extent towhich each countermeasure has been tested orimplemented), and rendering a judgement aboutwhether the countermeasure has been shown to beeffective or could be effective—or, if not effective,what would be necessary to successfully imple-ment the countermeasure. Finally, Chapter 9 pres-ents conclusions concerning what countermeasuresmay be cost-effective and where additionalresearch seems warranted.

The remainder of the introduction providesdefinitions of some basic terms used in the studyof lighting and vision. The key terms to bedefined are:

• Brightness• Point light source• Luminous intensity• Luminance• Illuminance• Reflectance• Glare

Brightness is the attribute of visual sensationaccording to which an area appears to emit moreor less light. Brightness is a relative term which

describes the appearance of an object to an observer. An object of any brightness will appearbrighter if the ambient light levels are lower.Brightness can range from very bright (brilliant) to very dim (dark). In popular usage, the term“brightness” implies higher light intensities,whereas “dimness” implies lower intensities.

Point light source is a light source that subtendsan extremely small angle at the observer’s eye sothat its attributes are not affected by its size, onlyby its luminous intensity. An example of a pointlight source is a star.

Luminous intensity is the light-producing powerof a source, measured as the luminous flux perunit solid angle in a given direction. It is simply ameasure of the strength of the visible light givenoff by a point source of light in a specific direc-tion, and usually expressed in terms of candelas(cd), where one cd equals one lumen/steradian.

Luminance is the amount of luminous fluxreflected or transmitted by a surface into a solidangle per unit of area perpendicular to given direc-tion. More simply, it is a physical measure of theamount of light reflected or emitted from a surfaceand roughly corresponds to the subjective impres-sion of “brightness.” Luminance does not varywith distance. It may be computed by dividing theluminous intensity by the source area, or by multi-plying illuminance and reflectance. The mostcommon units of measurement for luminance arecandelas per square meter (cd/m2), foot-lamberts(fL), and millilamberts (mL).

Illuminance is the photometric flux (or, more sim-ply, the amount of light) incident per unit area of asurface at any given point on the surface. Theilluminance E at a surface is related to the lumi-nous intensity I of a source by the inverse squarelaw E=I/d2, where d is the distance between thesource and the surface. The most commonly usedunits of measurement for illuminance are lux(lumens per square meter) and foot-candles (fc, orlumens per square foot). Retinal illuminance isthe amount of photometric flux that reaches theretina of the eye; it is a function of the diameter ofthe pupil of the eye and the amount of light

12

absorption within the eye. The unit of measure forretinal illuminance is the troland (Td), which isdefined without regard for the absorption in theeye. A nominal retinal illuminance of 1 Td is expe-rienced by an observer looking at a surface ofluminance 1 cd/m2 through a pupil area of 1 mm2.

Reflectance is a measure of the reflected incidentlight (illuminance) that is actually reflected awayfrom a surface. For many surfaces reflectance willdepend on the angle of viewing and the angle fromwhich it is illuminated, as well as the properties ofthe surface (including diffuseness or retroreflectiv-ity of the surface).

Glare can be defined generally as a bright, steady,dazzling light or brilliant reflection that occurswhen the luminous intensity or luminance withinthe visual field is greater than that to which theeyes are accustomed. Glare can cause discomfort,annoyance, or loss in visual performance and visi-bility. Direct glare is caused by light sources inthe field of view whereas reflected glare is causedby bright reflections from polished or glossy sur-

faces that are reflected toward an individual (forexample, a chrome nameplate on a leading vehi-cle). The entire visual field contributes to theglare level, and even a completely uniform field,such as that in a photometric sphere, will producesome glare. Detailed discussion of the factors thatcontribute to glare will be presented in Chapter 2.

Summary Glare occurs when visual field brightness is

greater than the luminance to which the eyes areadapted. Glare is caused by both direct and indi-rect light sources. Discomfort glare producesvisual discomfort, annoyance, and fatigue.Disability glare produces loss in visual perform-ance which is generally defined as a reduction inthe visibility distance of low contrast objects. Theelderly, people with light-colored eyes, and thosesuffering from cataracts are especially sensitive todisability glare. Glare at night can be mitigated byprudent design of the roadway, the automobile,and vehicle lighting systems.

13

To critically evaluate the potential effective-ness of any glare countermeasure, it is essential tounderstand the physiology of glare and itscausative factors. Glare is usually classifiedaccording to its effect on an observer, but eachtype of glare has its own physiological explana-tion.

The two principal types of glare are disabilityglare and discomfort glare. Disability glareimpairs the capability of the eye to perceive smallchanges in brightness (such as the luminance ofthe object to be seen), while discomfort glare, asits name implies, creates an uncomfortable sensa-tion. There is a tradeoff between visibility andglare; for example, research (Hemion 1969a andFlannagan 1996) indicates that visibility isimproved when opposing vehicles both use highbeams, although this creates discomfort glare. Theautomotive headlighting industry has always feltthat discomfort glare was more important than dis-ability glare, because discomfort glare is whatdrivers complain about.1 However, disability glaremay be equally important, because it is directlyrelated to the driver's ability to see objects andthus may be more likely to result in crashes.

Generally, glare results from a bright, steady,dazzling light or its reflection from shiny surfaces;it occurs when the luminous intensity or lumi-nance within the visual field is considerablygreater than that to which the eyes are adapted.Direct glare is caused by light sources in the field

of view (such as headlights, taillights, and lumi-naires). Reflected glare is caused by specularreflections from polished or glossy surfaces suchas the steel or aluminum doors on tractor trailers, arear-view mirror at night, or even bright matte sur-faces, such as vehicle interiors and dashboards,that reflect light toward the driver.

Glare affects both day and night driving per-formance. During the day, sunlight producesdirect glare and gives rise to indirect glare fromsurface reflections. At night, automobile head-lights produce direct glare by shining into the eyesof drivers in approaching cars, and indirect glare isexperienced from rearview mirrors and vehicleinteriors that reflect light from trailing vehicles.The effects of glare on drivers are much greater atnight than during the day, because at night driversare adapted to lower light levels and so require agreater difference in luminance between objectsand background to perceive objects on the road.This luminance difference is reduced by stray lightcoming either directly from a glaring light sourceor indirectly from reflections of headlamps on wetroad surfaces, mirrors, or vehicle interiors. Lightsthat are barely noticeable during daylight can beuncomfortably glaring at night.

Disability and discomfort glare apparentlyhave quite different physiological origins and soare very difficult to compare. Disability glarecomes from light scattering in the ocular media,whereas the sensation of discomfort glare appears

15

CHAPTER 2

FACTORS THAT CONTRIBUTE TO HEADLIGHT GLARE AT NIGHT

1Waltham has expressed the opinion what people sense about glare is discomfort and so it is discomfort on which they basetheir judgements about glare. At low light levels, one can experience vision impairment before experiencing discomfort.(CIBSE meeting in Bristol U.K., 1963)

to be related to neuronal interactions similar tosuch physiological functions as skin resistance orthe pupillary response to light (Fry and King1975). The two types of glare are affected differ-ently by environmental parameters. For example,disability glare does not seem to be affected bysource size or luminance, but it is affected byluminous flux and the angular offset from the line of sight. In contrast, for discomfort glare theapparent source luminance and size are majorparameters. As long as the size and position ofheadlights are kept more or less the same, stepstaken to reduce discomfort glare will also reducedisability glare. However, if the size of the head-light source is reduced while keeping the emittedflux constant, the higher headlight luminance willincrease discomfort while the disability effects willstay the same. This effect may explain the com-plaints about glare with some high-intensity dis-charge (HID) lamps, which, on any specific beamangle, project similar illumination as a halogenlamp, but through a smaller opening.

Another consequence of the different behaviorof disability and discomfort glare is that whenbackground luminance is low, glare sources mayhave a disabling effect on vision without being asource of discomfort. This problem has beenaddressed by using roadway lighting to provideminimum background luminance levels. With suf-ficient ambient illumination, a glare source maycreate a discomforting sensation without a measur-able disabling effect.

Disability Glare Disability glare is created by a light so bright

that its intensity results in a measurable reductionin a driver’s ability to perform visual tasks. Thereduction in visual performance is a direct result ofthe effects of stray light within the eye, which inturn are dependent on the age of the driver.Transient adaptation refers to a temporary reduc-tion in basic visual functions, such as contrast sen-sitivity and form perception, that occurs when theluminances from objects in the visual field changerapidly (Adrian 1991a). The degree of reductionin function is dependent on the change in lumi-

nance to which the eye must adapt. Transientvisual impairments are associated with rapid alter-ations in glare levels and changes in scene lumi-nance, as well as sudden eye movements (or sac-cades). Glaring light scattered in the eye can beexpressed as the superposition of a uniform lumi-nance onto the retinal image. This “veiling” lumi-nance, which is a function of the glare angle (theangle between the line of sight and the glaresource), adds to scene luminance and reduces thecontrast of the target to be seen. This formula forcontrast C without glare is:

1) C = LT – LB / LB

where LB is the background luminance and

LT is the luminance of the object to be seen (the

target).

If disability glare (Lseq, the equivalent lumi-

nance of stray light in the eye) is added to the lumi-nance of both the target and background, the formu-la for contrast in the presence of glare becomes:

2) C = [(LT + Lseq) – (LB + Lseq)] / (LB + Lseq)

= (LT – LB) / (LB + Lseq)

This formula indicates that contrast will bereduced as Lseq is increased.

Visual acuity depends on the contrast betweenthe background and the objects to be seen. Thepresence of glare reduces the ratio of target con-trast to threshold contrast, and the target is lesslikely to be seen or drowns completely in the light veil.

Direct glare in night-driving encounters hasstrong angular or spatial dependence, resulting ineye movements that almost invariably lead totransient changes in the luminance reaching theeye and so to either dark adaptation or lightadaptation. As part of the adaptation process,the retina adjusts to the quantity of light.Although different parts of the retina are exposedto different quantities of light (for normal scenesof non-uniform luminance), it is generallyassumed that an instantaneous state of adaptationof the fovea (the central retinal region) can be

16

described by an equivalent veiling luminancefrom a uniform source superimposed on the visual field. A given state of foveal adaptation,therefore, can be produced by many luminanceconfigurations.

The principle that the effect of a glare sourcecan be represented by a uniform background lumi-nance was developed by Holladay (1926), whoassumed that the glare effect was caused bycrosstalk in the optic nerve. Stiles (1928–29)attributed the glare effect to stray light in the eye.Crawford (1936) showed that stray light in the eyefrom different sources is additive in nature. Theequivalent stray light luminance Lseq varies withage, and can be calculated according to equation 3, below.

Causes of stray light in the eye include scatter-ing, diffraction on the fringe of the pupil, and opti-cal imperfections in the eye, all of which deflectlight from the geometric image on the retina.Added incoherently to the light of the image, straylight reduces the contrast between the image andbackground. The effects of disability glare alsoincrease with the age of the observer due tochanges in the intraocular media and cornea of theaging eye that increase scattering. Adrian (1975)and Ijspeert (1990) found that the glare effectbegins to increase rapidly at between age 35 and 40.

Since the work of Holladay in the 1920s, sev-eral equations have been proposed to account forthe effect of disability glare(see the review byAdrian and Bhanji, 1991). All of the formulasgive veiling luminance Lseq as a function of theilluminance produced by the glare source, meas-ured at the vertical plane of the driver’s eye, andthe glare angle, the angle between the object beingviewed and the center of the glare source. In gen-eral, the formulas are similar, with the primary dif-

ference between them being in the exponent usedfor the glare angle. The formulas have been “finetuned” by Hartman and Moser (1968), Vos andPadmos (1983), and most recently Adrian andBhanji (1991) to include the effects of driver ageand to improve predictions at small glare angles.The three factors now considered to determineveiling luminance are:

1. Illuminance on the eye from the glare source(Egl in lux) in the plane normal to the line of

sight.

2. Angle between the line of sight and the centerof the glare source (θ in degrees)

3. Age (in years)

The equation for veiling luminance (Lseq ) cur-rently accepted by the Illumination EngineeringSociety of North America was first presented byAdrian and Benji (1991):

3) Lseq = kn Egli

, where∑θn

1 i

k = 9.05 [1+ ( age )4]66.4

n = 2.3–0.7 logθ, when 0.2º< θ < 2º

n = 2, when θ >2º

For night driving, Lseq should not exceed 15%of the background luminance LB, which is usuallyapproximated by the luminance of the pavement.2

When Lseq / LB = 0.15, the visual threshold fordetection is increased by approximately 10% for atarget subtending10 minutes of arc at a road sur-face luminance of 1cd/m2. Lseq is most often usedto predict the visibility of a specific target in thepresence of a given background. The impact ofLseq on the visibility of a target subtending10 min-utes of arc can be seen in Figure 1.3 Because dis-ability glare reduces the contrast between target

17

2There is no scientific basis for maintaining the ratio of Lseq / LB at a maximum of 0.15, but it is a generally accepted convention.3At a distance of 410 feet, the stopping distance at 50 mph, an object 14 inches tall will subtend a visual angle of 10 minutes ofarc at the driver’s eye. Smaller objects would require greater contrast and larger objects less contrast than predicted by figure 1in order to maintain visibility.

and background, the target luminance has to beincreased in the presence of glare to render the tar-get visible. The table indicates the amount bywhich glare (represented by Lseq) increases thethreshold luminance, which is the contrast betweentarget and background at which 50% of observerswould see the target when presented for a briefinterval. The threshold increment TI is the percent-age by which the threshold luminance is increased;it can be calculated by:

4) TI (in %) = 60.3 Lseq / L0.86

where L is the average pavement luminance.

We can see from Figure 1 that when the pave-ment luminance is about 0.28 cd/m2, and Lseq is .02 cd/m2 (an unlighted road), the threshold con-trast must be increased by more than 5% to main-tain visibility. However, if the pavement is lightedso that its luminance is more than 0.6 cd/m2 , nosignificant increase in threshold contrast is neces-sary. On the other hand, if Lseq is 0.1 cd/m2 ormore and the pavement luminance is less than 0.35cd/m2, threshold contrast must be increased bymore than15% to maintain visibility.

CataractsDisability glare is caused by the diffusion of

light as it passes through the cornea, lens, vitreoushumor, and retina (Bailey and Sheedy 1986;Winter 1985, Allen 1985). This diffusion is theresult of imperfections and debris in the opticalsystem that are collectively known as opacities.Light that scatters when it encounters an opacityprior to reaching the outer receptor segments in theretina causes a veiling effect which reduces targetcontrast and, therefore, visibility.

A cataract is a lenticular opacity wherein thelens becomes clouded. Cataracts are a widespreadproblem—over 1.5 million cataract surgeries areperformed each year (Daily Apple 1999). Besidesreducing the total amount of light reaching theretina, cataracts also scatter any light that doespass through, causing a reduction of contrast thatcan render objects barely visible. Problems withglare from artificial light sources such as head-

lamps are often, in fact, an individual’s first indi-cation of a developing cataract. In 1988, Schieberconcluded that lenticular changes which occur nat-urally with age, such as cataract, are the primarysource of age-related increases in sensitivity todisability glare. As evidence for this, he cited astriking reduction in glare problems after cataractsurgery.

Discomfort GlareDiscomfort glare refers to a bright light that,

because of its size and luminance, causes a meas-urable level of subjective discomfort or annoy-ance. The most popular scale of discomfort glarewas first used by DeBoer et al. (1967). It is a rank-

18

Figure 1. From W. Adrian, D.A. Schreuder. The assessment of glare in street-lighting, Lightand Lighting, December 1968.

ing of the discomfort caused by a light on a scaleof 1–9, according to the following descriptions:

1. Unbearable3. Disturbing5. Just acceptable7. Satisfactory9. Just noticeable

Discomfort glare is generally influenced bythree major factors.

• Location of glare source relative to the line ofsight

• Luminance of the background• Luminance and size of the glare source lead-

ing to illuminance at the eye

Experiments conducted by Schmidt-Clausenand Bindels (1974) led to the following formulafor the level of discomfort on the DeBoer scale asa function of the three factors listed above:

5) W = 5.0 — 2 log Ei

3.0•10-3 1+ La θi

0.46

4.0•10-2

where:

W = glare sensation on a scale of 1 to 9La = adaptation luminance (cd/m2)

Ei = illumination directed at observer’s eyes from

the ith source (lux).θi = glare angle of the ith source (minutes of arc)

Research has shown that additional factors arerelated to discomfort glare, including apparent sizeof the glare source (or its solid angle subtended atthe eye) and the source luminance (Sivak et al.1990). A common measure of discomfort glareused in research is the “Borderline betweenComfort and Discomfort” (BCD)4. Sivak foundthat there was a very small but significant relation-ship between headlamp area and the DeBoer scale.If all the factors in equation 5 for discomfort glare

are held constant, increasing the size of the glaresource lowers the discomfort level, because thesource luminance drops (for constant illuminanceEi). While the effect of size appears to be small,this reduction in source luminance does reduce thelevel of discomfort.

Other FactorsIn addition to the widely accepted parameters

just discussed, which have the most significant andpredictable impact on discomfort glare, other fac-tors may play a role. The discomfort glare formulapresented above is based on highly controlled lab-oratory studies. Adrian (1991b) has indicated thatthis formula and other published formulas for dis-comfort glare “show the same characteristics andyield comparable results.” However, Boyce andBeckstead (1991) noted the mediocre correlationbetween these formulas and the glare ratingsobtained in less well-controlled field situations.Other factors that may be related to discomfortglare ratings will be discussed below.

Immediate Surround LuminanceHopkinson (1963) introduced the immediate sur-round luminance as an element in his glare formu-lation. In their review, Boyce and Beckstead(1991) found that including this quantity markedlyimproved the accuracy of the glare rating. Mostvision research is based on a two-luminanceworld: The target and the background each have aluminance, with the latter generally having thegreatest effect in determining the eye’s adaptation.Whether we are predicting glare or visibility, theadaptation luminance La represents the multi-luminance world we drive in.

While incorporating the luminance of theimmediate surround may improve glare predic-tions in static situations, under dynamic field con-ditions there are practical difficulties with such anapproach. In the field, the luminance of both theimmediate surround and the target tend to varyrapidly. The luminance of the immediate surroundis actually more relevant to determining the con-

19

4BCD is determined by having a subject adjust the intensity of a glare source until its brightness is between a level that isannoying and a level that is comfortable.

spicuity of an object as defined by the backgroundluminance. In dynamic situations, it seemsimpractical to consider more than one surroundluminance in the calculation of glare. Despite thislimitation, it is important to remember that it is thebackground luminance in its transient state thatbest approximates the adaptation luminance in anyvisual encounter on the road. Fortunately, the eyeresponds rather quickly under normal roadwaylighting levels (Adrian 1991a).

Range EffectThe range effect is a shift in the subjective

assessment of discomfort based on the range ofintensities of the glare sources being evaluated(Lulla and Bennett 1981). If a subject is exposed toa wider ranger of glare intensities, the level of glarethat the subject considers tolerable is increased.Although the work of Lulla and Bennett was basedon relatively short exposures in the laboratory, itmight have consequences for measurements of dis-comfort glare during night driving.

Olsen and Sivak (1984) showed that the rangeeffect is also present in real driving scenarios.They found that, except at high glare levels, driv-ers who were exposed to a range of illuminancesfrom headlights gave mean glare ratings that wereone to two scale units above those predicted fromthe Schmidt-Clausen equation. In other words,glare levels predicted by the equation to be “justacceptable” were rated by drivers to be “satisfacto-ry,” and glare levels predicted to be “satisfactory”were rated nearly “just noticeable.” These find-ings suggest that, under certain circumstances, per-missible glare may be much higher than the equa-tions would suggest, and that this increase in toler-able glare may even be extended further by raisingthe extreme levels of glare to which drivers areexposed.

It may be that the more comfortable glare rat-ings obtained in real driving scenarios versus labo-ratory situations are an extension of the rangeeffect. This would be true if the range which sub-jects use in evaluating glare exposures is a range

based upon prior experience in driving situationsand not just the recent experience established in asingle experimental setting. Therefore, even whenthe actual range of glare levels in a laboratory orfield study is restricted, more comfortable judge-ments may be assigned to any given glare levelwhen driving than in the laboratory.

Sivak et al. (1989) reported a practical demon-stration of this effect when comparing Europeanand U.S. driver groups. As will be seen in Chapter4, European headlighting design stresses the mini-mization of discomfort glare as a high priority.Students with recent driving experience inGermany reported higher levels of discomfortwhen exposed to glare than U.S. students. Thestudy tested only a small sample of young driversfrom one country, and the results may be attributa-ble to language differences in interpreting the dis-comfort rating scale. Still, the range effect sug-gests that our experience with glare may well havean influence on how comfortable or disturbing weperceive glare to be.

Task DifficultyA laboratory study by Sivak et al. (1991) and

a field study by Theeuwes and Alferdinck (1996)provide some support for the hypothesis that dis-comfort glare ratings are influenced by the diffi-culty of the task being performed. In the Sivakstudy, subjects performing a gap-detection taskreported that a fixed amount of glare caused morediscomfort as gap size decreased, making the gapmore difficult to detect. Sivak et al. also noted thepotential incongruity in discomfort glare modelslike that underlying equation 5: Conditions suchas fog or a dirty windshield increase veiling lumi-nance and thus disability glare, but according toequation 5 such conditions would reduce discom-fort, because the peak intensity of the glare source(E) would be decreased and the illumination in therest of the retina (La ) would be increased. Yetthese conditions make driving more difficult andmight increase discomfort if task difficulty wereincluded in the equation.

20

5The nine point scale used to measure glare is an ordinal scale which means that the intervals along the scale are not equal. Inpractice it has however turned out to approximate an interval scale.

The field study reported by Theeuwes andAlferdinck (1996) included two sections of roadwith similar levels of background luminance butdifferent road markings and curvature. The dis-comfort glare ratings were much lower (moreannoying) on the winding road with no markings.For the study as a whole (which included nineroad sections), the glare ratings were much higher(more comfortable) than would be predicted byequation 5, a finding which is attributed to the lackof difficulty in driving all but the one section ofwinding road without markings. Although the evi-dence associating task difficulty with the percep-tion of glare is slight and the effect of reducingtask difficulty on discomfort glare is unpre-dictable, most road improvements that reduce thedifficulty of driving would also have safety bene-fits that offset their cost, so that, overall, suchimprovements might be worthwhile.

Physiological EffectsAge — Research into the effects of age on discomfort glare appears to be inconclusive.Theeuwes and Alferdinck (1996) found that olderdrivers were more sensitive to stray light, but rat-ings of glare discomfort on the nine-point ratingscale were not sensitive to age differences. Incontrast, Bennett (1977) reported a small but sig-nificant negative correlation ( r = –0.36) betweenBCD and age. The discrepancy between thesestudies may be explained by the finding ofTheeuwes and Alferdinck that the nine-point ratingscale does not correlate well with BCD and that inactual driving situations subjects rate glare as lessannoying than predicted by the models developedunder laboratory conditions.

Eye Color — The amount of light that enters theeye is determined by pupil size, which is is regu-lated by the dilation and constriction of the iris.Two bands of muscle fibers in the iris, the dilatorand sphincter muscles, cause the opening and clos-ing of the pupil. The iris derives its name fromthe mythological Greek goddess of the rainbow, anindication that its predominant characteristic iscolor. Unlike the sclera, the white portion of the

eye, which is opaque and blocks most incidentlight, the pigmented iris is translucent. For thisreason, the amount of light that enters the eye isdependent on both the pupil size and iris color. Alight-colored iris (blue, for example) transmitsmore light to the retina than a darker-colored iris,where the epithelium contains the same pigmentsas the iris. In fact, when eye color is extremelylight, as with albinos, tinted or colored contactlenses are sometimes prescribed because glare canbecome intolerable (NOAH 1999).6

Corrective Lenses — Lower light levels, such asthose on unlighted roads at night, are associatedwith a state of night myopia. This phenomenon isa result of movement of the accommodative sys-tem toward a resting state, which Leibowitz andOwens (1978) found to peak at 1.52 diopters, or66 cm. If a driver is already myopic, this wouldincrease his myopia.

Many older drivers, while still able to see dis-tant objects well, suffer from presbyopia (farsight-edness of the elderly) and so wear some type ofcorrective lenses to improve their ability to focuson near objects, such as a car’s dashboard. Withincreasing age, the range of accommodationshrinks due to hardening of the crystalline lens inthe eye. As a result, night myopia is less pro-nounced in presbyopic subjects.

It is often observed that corrective lenses suchas glasses and contacts can exacerbate the effectsof glare, particularly if the lenses are scratched ordirty. The scientific data, however, are inconclu-sive. For example, Schieber (1988) found that dis-ability glare can result from damaged contact lens-es or from corneal injury due to prolonged contactlens usage (see also Miller and Lazerby 1977).Sivak, Flannagan, Traube and Kojima (1999),however, failed to find a difference in the DeBoerratings of individuals wearing glasses or contactsand those of individuals without visual corrections.However, the Sivak et al. finding that correctivelenses are not associated with increased discomfortglare does not contradict Schieber’s result thatdamaged lenses may cause disability glare.

21

6The pigment that shows up in the iris is also present in the inner eye. People exposed to high light levels for generationsdeveloped darker eyes to reduce inter-reflection in the eye.

Our awareness of a problem with headlightglare comes primarily from direct observation andfrom the reports of others. Mortimer (1988) citeshis own dissertation, a study of drivers aged 19 to39 in which 65% of the subjects said they werebothered by glare at night, and another study inwhich one-third of the subjects said glare was afrequent problem. The San Francisco Chronicle(June 28, 1999) reported, “Young and old alike,many drivers on the Bay Area’s busy, twisty roadscomplain of blinding lights in the windshield andrear-view mirror.” The article quoted an informalsurvey indicating that drivers are convinced thatglare is worse than ever.

While there seems to be no scientific explana-tion for why glare is becoming a greater problem,contributing factors appear to be the introductionof high intensity discharge lights (HID), the prolif-eration of sports utility vehicles with headlightsmounted higher than passenger vehicles, and theuse of illegal aftermarket devices. As reported bythe San Francisco Chronicle, many complaintsabout HID could actually be about fake HID,“which usually are installed by drivers seeking tolook cool. The fakes cause more glare becausethey diffuse the beam ... posing a safety risk foreveryone on the road.”

This chapter will consider two aspects of theheadlight glare problem: first, what subgroups aremost affected by glare, and second—and most crit-ically— its consequences. Understanding the con-sequences of headlight glare is the basis for evalu-ating the importance of countermeasures.

Subgroups Affected Most by Glare

Based on the factors contributing to glare thatwere identified in the previous chapter, it can beinferred that three subgroups of individuals are athigh risk for nighttime glare: those with light eyecolor, those whose driving is mostly on high-vol-ume, two-lane roads, and the elderly. Anotherpotential high-risk subgroup is composed of drivers with corrective lenses, but this is only aserious factor if the contacts and lenses have been scratched or damaged. Some drivers whohave had vision correction surgery, such as radial keratotomy or LASIK, also complain about glare.

In Chapter 2, a formula for disability glare was presented and introduced a factor to accountfor the effect of age. This formula indicates that,while there is some increase in disability glareamong younger drivers, effects begin to increasesignificantly only after age 40. Results consistentwith this model were obtained by Pulling et al.(1980), who conducted simulator experiments thatdefined threshold glare as brightness from “head-lights on oncoming cars so great that potentialhazards on the highway could not be distinguishedin time...” As discussed in Chapter 2, the increasein disability glare with age is a result of a reduc-tion in the amount and optical clarity of the lighttransmitted in the eye as the lens and cornea age.These changes increase the relative amount ofstray light in the eye and cause this light to scattermore, which increases the veiling luminance andso the effect of glare. Many older people also

23

CHAPTER 3

THE PROBLEM OF HEADLIGHT GLARE

develop cataracts, which cloud the lens andincrease light scattering inside the eye.

There is also evidence that older drivers maybe bothered more than younger drivers by discom-fort glare. Bennett (1977) reported some experi-ments which showed that the level of glare thatcaused discomfort (measured by the “borderlinebetween comfort and discomfort,” or BCD)decreased rapidly with age until age 40, afterwhich it continued to decline, although at a slowerrate. Schwab et al. (1972) found that drivers, par-ticularly older drivers, were willing to reduce theamount of remuneration they received from aresearch study in order to avoid having to partici-pate in trials with high levels of glare. After tak-ing family income into account, Schwab conclud-ed that drivers appeared willing to pay about whata polarized lighting system cost at that time. Thissuggests that, after adjusting for inflation, driverstoday would be willing to spend approximately$100 for a device that would reduce headlightglare to acceptable levels.

Consequences of Headlight Glare

Documentation of the consequences of head-light glare is not readily available. It is far easierto speculate about what these consequences mightbe than to measure them. A controlled field study(Theeuwes and Alferdinck 1996) tested three lev-els of glare illuminance (namely 0.28, 0.55, and1.1 lux, or 350, 690, and 1380 cd, at 500 m) andfound that even the lowest level of glare resultedin reduced detection distances and, in some situa-tions, greater speed reduction and more steeringreversals (thought to be a surrogate measure offatigue-inducing workload). Surprisingly, the dis-comfort glare ratings of these glare levels were notrelated to any of the performance measures.

Very generally, headlight glare has the poten-tial to both increase the frequency of accidents anddecrease the mobility of individuals by discourag-ing them from driving at night. Both of these con-sequences may be mediated by a reduction in visi-

bility or an increase in fatigue or tension—or bythe simple discomfort, and sometimes painfulexperience, which night driving presents. The fol-lowing section will discuss the effects of glare onvisibility, fatigue, accidents, driver behavior, andmobility.

Effects of Glare on VisibilityThe effects of glare on visibility have primari-

ly been studied in the laboratory, in studies thathave produced the formulae discussed in the pre-ceding chapter. Of interest here is what, if any-thing, can be said about the glare-related loss ofvisibility in field situations. One controlled fieldstudy by Cadena and Hemion (1969) demonstrateda reduction in visibility distance that was attrib-uted to the level of headlight glare, although someof the reduction in visibility distance seemed to bedue to the distraction of opposing vehicles bothwith and without headlights. Hare and Hemion(1968) used observations of encounters betweenopposing vehicles in two-lane, open road situa-tions to develop a formula for the reduction in visibility distance.

Hemion (1969) found that while detection dis-tances decreased in the presence of glare fromopposing high-beam headlamps, the distanceswere actually higher when both vehicles used highbeams than when both used low beams, eventhough both discomfort and disability glare werehigher. The additional illumination from the dri-ver’s own high-beam headlights increased the tar-get contrast and so compensated for the loss incontrast that led to disability glare. In terms of vis-ibility distance, more is to be gained from usinghigh beams than low beams.

Hemion (1969) found that the position of theglare car relative to the observer had no effect ondetection distance over a 1,000-foot range. Thisobservation might indicate that the detection dis-tance was primarily due not to glare but to distrac-tion caused by the presence of an opposing vehi-cle. On the other hand, the relative position of theglare car may have had no effect because of thecompeting effects discussed in the previous chap-ter: Glare increases with higher illumination but

24

decreases with a higher glare angle. As theobserver car and the oncoming glare car approacheach other, both illumination and glare angle atfirst increase; then, beyond a certain point, thebeam angle of the opposing vehicle becomes largeenough that illumination begins rapidly todecrease.

Effect of Glare on FatigueOne important consequence of headlight glare

is its potential ability to cause feelings of stressand fatigue. Schiflett, Cadena, and Hemion(1969) defined fatigue as a state of increased dis-comfort and decreased efficiency resulting fromprolonged exertion on a task. The extent to whicha person experiences fatigue is a function of thatperson’s surroundings, including visual condi-tions. Boyce (1981) pointed out that although therelationship between visual conditions and inducedfatigue has been studied for many years, littleprogress has been made in understanding theirrelationship. The slow progress could be due tothe fact that fatigue is very difficult to define andmeasure. However, even without experimentalevidence, personal experience with driving tells usthat fatigue does occur and that lighting condi-tions, including glare from approaching headlights,may influence its occurrence.

Generally, fatigue is divided into two cate-gories: physical fatigue and mental fatigue.Physical fatigue is well understood; it results fromprolonged use of a given set of muscles. Physicalfatigue includes physiological changes in the mus-cles that cause them to simply stop operating;either the muscles become incapable of contract-ing or the central nervous system stops sendingthem signals. Mental fatigue, by contrast, is notwell understood. Boyce theorized that mentalfatigue is nearly impossible to measure directlybecause it involves the entire body as well as themind, not just a single muscle or muscle group.Although difficult to define or measure, there is noquestion that mental fatigue exists, because, asBoyce says, “It is a matter of common experiencethat prolonged and difficult mental work leads tofeelings of tiredness.”

Studies that attempt to determine how lightingconditions affect fatigue usually focus on one ofthe two types of fatigue. Physical fatigue associat-ed with lighting conditions can impair the func-tions of the muscles of the ocular motor system.Binocular vision, according to Weston (1954),involves at least twelve muscles in every move-ment of the eye. In addition to these eye muscles,many different facial muscles are brought into playin situations with bright light: Muscles of thelower eyelid, cheeks, and upper lip cause the eyesto squint and give the face a look of grief or dis-tress. Weston concluded that the prolonged use ofthese “grief” muscles also contributes to thefatigue associated with bright conditions.

Another source of physical fatigue while driv-ing is the effort required to keep the eyes focuseddirectly ahead of the vehicle. Weston describedwhat happens when bright lights, such as passingheadlights from opposing cars, pass into theperipheral field. In this situation, the bright lightacts as a distraction, causing “visual confusion,”and the eyes tend to automatically move towardthe light. An effort must be made to keep the dri-ver’s gaze directed to the near side of the roadinstead of toward the blinding light of anapproaching car.

Because the onset of fatigue is very difficult tomeasure, determining relationships betweenfatigue and variables such as lighting conditionsis a challenge. In the study by Schiflett, Cadena,and Hemion (1969), psychological and physiologi-cal criteria were used as fatigue indicators. Theexperiments investigated whether fatigue wouldcause different driver performance when three dif-ferent lighting systems were used: a glare-produc-ing high-beam system, no opposing headlights,and polarized headlights.

A two-part study by Schiflett et al. (1969)evaluated the development of driver fatigue interms of performance on tests given before, dur-ing, and after sessions of driving. The testsincluded psychological and physiological charac-teristics related to fatigue. In the first part, subjectsdrove around a runway loop approximately 1.75

25

miles long while both real and simulated vehicleswere used as approaching cars that providedopposing-headlight glare. In the second part, simulated driving in a stationary test complex was used.

In summarizing the Schiflett experiments,Schwab and Hemion (1971) pointed out thatalthough effects were seen in some of the tests,they were not consistent among drivers or evenbetween repeated performances of a single driver,and so the observations could not support or refutethe hypothesis that onset of fatigue depended onlighting conditions. The large variance in the datacould be explained by the lack of control overwhat the subjects did during the day before theycame to the test site. However, the methodologyof measuring psychological and physiologicalchanges in the subjects could still be a successfulsolution to the problem of scientifically observingthe fatigue resulting from glare, and it should beemployed in future studies.

Effect of Glare on AccidentFrequency

A total of 42,059 deaths occurred on U.S.roadways in 1998. An analysis of the FatalityAnalysis Reporting System (FARS) of theNational Highway Traffic Safety Administration(NHTSA) showed that, although the number ofvehicle miles driven at night represents only about14% of the total, 45% of deaths took place atnight. The severity of the nighttime accident prob-lem is further indicated by the traffic death rate atnight, which in 1998 was 4.63 deaths per 100 mil-lion vehicle miles, 4.4 times higher than the rateduring the day (National Safety Council 1999).This reflects a notable increase from the 1993 rateof 3.9 per 100 million vehicle miles, or 3.2 timeshigher than the daytime rate.

Nighttime death rates have consistently beenthe highest in rural driving environments. Of the18,874 nighttime traffic fatalities in 1998, over56% took place on rural roads (FARS database).The hazardous nature of rural nighttime drivingbecomes even more apparent when one considers

that only 40% of all vehicle mileage occurs onrural roads (NTS 1999), and less than 15% of thatis during the nighttime hours (National SafetyCouncil 1999). For the past decade, the ruralnighttime death rate has consistently been aboutthree times the rural daytime rate and about two-and-one-half times the urban nighttime rate(National Safety Council 1988, 1990, 1994, FARSDatabase, NTS 1999). According to 1988Pennsylvania Department of Transportation(PennDOT) data, rural accidents are more likely toinvolve fatalities, whereas urban accidents aremore likely to involve property damage. ThePennDOT database also showed that the two mostcommon subgroups for nighttime accidents arerural unlighted and urban lighted areas.

Driving safety on rural roads is compromisedby a number of factors, including heavy use, fre-quent curves and intersections, and decreased sightdistance, all of which combine with high rates ofspeed to produce very dangerous driving condi-tions. NCHRP Report 66 (1979) indicates thatglare from oncoming headlights is most oftenencountered on two-lane rural highways. Glare isalso worse on roadways that curve to the leftbecause opposing headlights are directed into thedriver's eyes in proportion to the degree of curvature.

Despite the known deleterious effects of glareon the visual system, we seldom hear of a trafficaccident caused by glare, and glare is seldom con-sidered a major factor in accident causation.Hemion (1969) found very few states in which“accident reporting forms and procedures madespecific reference to headlight glare as a causativefactor in vehicle accidents.” Seven states out ofthe 25 contacted by Hemion could readily providerelevant statistics; these states reported that onlybetween 0.5% and 4.0% of all night accidentswere attributable to headlight glare. Hemionbelieved glare data to be under-reported becausereporting typically focuses on direct causes andwould tend to not consider a vehicle that is nolonger at the scene. A later report by Mortimer(1988) also stated that glare is rarely reported as afactor in accidents.

26

The situation does not appear to be much dif-ferent today. The FARS database could potential-ly be used to identify glare-related accidents(NHTSA 1987). The categories in FARS closestto glare effects are codes 23 and 61 under “Person Related Factors: “Failure to Dim Lightsor Have Lights on When Required,” and“Reflected Glare, Bright Sunlight, Headlights,”respectively. However, for accidents coded as 23,it is not possible to tell if the accident was relatedto lights being on high beam or not being on at all.There were very few accidents coded as 61—onlyfour in 1997, and of these only one occurred atnight. The PennDOT accident records systemincludes a contributing factor labeled “GlareCondition,” but the records give no indication asto whether the glare is from opposing headlamps,from the sun, or any other light source. The poorcategorization of glare-related accidents could be aresult of the State departments of transportationnot recognizing glare as a major contributing fac-tor in accidents, or it could be related to drivers’inability to identify glare as a contributing factor.There could be a real yet underreported effectfrom glare on accidents that has not yet been rec-ognized because accident victims and those whofill out accident forms don’t know whether theyshould report it or how to do so.

There is a real need to improve accident-reporting systems to account for glare as a con-tributing factor in accidents. Cost-benefit analysescannot be performed to justify an expenditure ofresources on safety improvements without someestimate of the accident-reduction potential. Forexample, the report by Pulling et al. (1980) men-tions the absence of necessary accident data thatmight support the cost of glare screens, which oth-erwise appeared to be a productive countermea-sure on some types of roads.

Owens and Sivak (1996) analyzed data fromFARS for the period 1980–1990 that suggest thatvisibility and alcohol may play different roles infatal traffic accidents. Poor visibility appears to bethe major factor in situations that have inconspicu-ous hazards such as pedestrians and cyclists.However, when the weather is clear and incon-spicious hazards are absent, ambient illumination

does not seem to have a role in fatal accidents;instead, alcohol is the dominant factor. Owensand Sivak suggest that this difference in roles maybe due to the effectiveness of marker lights, reflec-tive materials on vehicles, and delineation. Still,problems with the accident reporting system raisedoubts about these results. The system makes itfar easier to report the involvement of a pedestrianor cyclist than to report the poor visibility of anobject that was not even identified. Accidentsinvolving alcohol may sometimes be avoided withimproved visibility, but again the accident report-ing system does not make the relevant informationavailable.

Nonetheless, on dark roads where disabilityglare may precede discomfort, drivers are unlikelyto be aware of the effect of glare on their vision.Analytically, one must accept the syllogism thatsince driving is primarily a visual task and glarehas a deleterious effect on vision, glare must havea deleterious effect on driving. It should not beassumed that the deleterious effects are necessarilycatastrophic. Drivers may compensate for the lossof vison by driving more slowly or otherwise morecautiously.

Effect of Glare on Driving BehaviorThere has been very little research on the

effects of glare on driving behavior. A study byTheeuwes and Alferdinck (1996) suggests thatglare from oncoming headlights has a minimaleffect on speed and steering. The study concludedthat De Boer discomfort ratings have no predictivevalue with regard to how much drivers will adjusttheir speed. Drivers in the study did reduce speedin the presence of glare (by approximately 2km/h), but not in relation to the amount of glare.The study’s analysis of steering wheel reversalsand gas pedal reversals showed no relationship tothe presence of or the amount of glare.

Effect of Glare on MobilitySince mobility is a primary goal of the trans-

portation system, anything that degrades mobilityis a cause for concern. There is some evidencethat problems with glare during nighttime drivinghave a negative impact on some motorists’ willing-

27

ness to drive at night. Chu (1994), reviewing evi-dence from the National Transportation Survey,concluded that elderly drivers are less likely todrive at night and during peak hours than middle-aged drivers, and are less likely to drive at nightthan during peak hours.

As part of another, as-yet unpublished study1,both an urban (Philadelphia) and a suburban focusgroup were used in 1997 to study the driving expe-rience of older drivers. Focus groups, surveys,and driver logs were used in the study to documentthe problems of older drivers with nighttime driv-ing. When nighttime driving was discussed with agroup of instructors for the 55 Alive course, prob-lems with glare from oncoming headlights werereported to be the greatest nighttime concern forolder drivers. The majority of focus group mem-bers listed glare from oncoming headlights as oneof their major concerns about nighttime driving,and the suburban group said that reducing head-light glare is one of the factors that would help themost with driving at night.

An open-ended question in the driver biogra-phical logs asked, “What is your greatest difficultyor biggest concern about driving at night?” Onethird of the total driver-log study group wrote“headlight glare” or “lights from oncoming traffic”as their primary or secondary answer. Forty per-cent of the urban subjects cited these factors astheir greatest concern and 27% of the suburbansubjects gave them as their most frequentresponse. Only four different concerns were citedby urban drivers, whereas suburban drivers cited11 concerns or difficulties. For both groups, themost frequently cited difficulty regarding night-time driving was headlight glare.

For this group of older drivers, 29% reportedheadlight glare to be their greatest concern andalmost 21% said that poorly lit roads caused themthe most concern. Another 24% of driversresponded with several other difficulties that couldbe related to roadway lighting: difficulty seeing,bright lights, vehicle breakdowns, and seeingpedestrians or bicyclists. Overall, 74% of therespondents indicated that nighttime driving comfort was related to some aspect of roadwaylighting.

SummaryThe extent to which glare is a problem for

night driving is not easily quantified. In theabsence of official statistics or scientific data, evi-dence of a glare problem is based almost entirelyupon subjective reports, most of which are anec-dotal. Without data from well-designed experi-ments, we can only qualitatively assess the delete-rious effects of glare, and the economic and safetyconsequences are left unknown. While there is lit-tle doubt that the number of drivers complainingabout glare is increasing, the age of the drivingpopulation is also increasing. Without good datathere is no way of knowing whether the drivershaving problems with glare are those with themost exposure to glare situations (such as high-volume two-lane roads), or whether they are olderdrivers that have visual problems even in theabsence of glare. If drivers have basic problemswith night vision, solving their problems withglare may increase their risk by giving them afalse sense of security and encouraging them todrive more at night.

28

1Research conducted by The Last Resource Inc. for The Federal Highway Administration under Contract DTFH61-96-C-00023.

The search for and evaluation of countermea-sures for headlight glare has been a subject of dis-cussion from the beginning of automotive lighting.The first attempt at creating a countermeasure wasthe development and regulation of the headlampbeam pattern itself. Were it not for the need tocontrol glare, the intensity of light from headlampscould be made as great as technologically possible.In fact, the intensity of headlamps is restricted asthe result of compromises that determine the head-lamp beam pattern. The design of the beam pat-tern is itself a countermeasure, which is discussedlater in this chapter.

The light output from a headlamp, as meas-ured by its intensity in candelas, is not uniform inall directions. The primary goal of headlampdesign is to provide sufficient light to see objectson the road while at the same time limiting theamount of glare. The variance in the directionallight output of headlamps is illustrated by theheadlamp beam pattern shown in Figure 3 inChapter 5. Although all U.S. headlamps must con-form to Federal Motor Vehicle Safety Standard108 (FMVSS 108), this standard only regulateslight intensity at a finite number of test points. Asa result, different lamps made for different vehi-cles by different manufacturers will have dramati-cally different beam patterns.

Another reason for the variability in beam pat-terns encountered on the highway is that head-lamps are often misaimed, which shifts the beampattern either horizontally or vertically. In addi-tion if the headlamp height is raised without cor-recting the aim of the headlamp with respect to thehorizontal, the beam pattern will be marginallyaffected. Both aiming and reduction in lamp

height are potential countermeasures discussed in Chapter 5.

The horizontal and vertical curvature of theroad changes the relative location of the beam pattern with regard to distant objects. While thebeam pattern remains the same with respect to thefixed axes of the vehicle as it traverses a curvedroad, the location of targets changes, includingboth the eyes of oncoming drivers and signs orobjects on the road. Therefore, the intensity oflight directed at these targets and the amount ofillumination falling on them also varies. Theintroduction of dynamic or “smart” headlights (seeChapter 5) is a countermeasure intended to nullifythis effect.

Finally, variation in the operating voltage ofthe vehicle will affect the output intensity of theheadlamps in all directions. Although the beampattern will remain the same, the overall intensitywill be either increased or decreased.

Other methods of controlling glare have beendocumented by Pulling et al. (1980), who dis-cussed five countermeasures for headlight glare:polarizing headlights, glare screens, road delin-eation, highway lighting, and driving restriction.A more recent report by Duncan (1996) consideredthe application of a number of state-of-the-arttechnologies for glare suppression, includingpolarization, “smart” headlights, modulated head-lights, ultraviolet systems, and infrared systems.

Chapters 5—8 discusses these and otherpotential countermeasures that have been promot-ed to reduce or eliminate the negative effects ofheadlight glare. In general, the causes of both dis-

29

CHAPTER 4

A COMPREHENSIVE REVIEWOF GLARE COUNTERMEASURES

comfort and disability glare produced by head-lighting are either opposing vehicle headlights orindirect light sources such as rear-view and side-view mirrors and other highly specular surfaces onleading vehicles or within one’s own vehicle.Before considering the countermeasures them-selves, it is useful to consider the highway situa-tions that contribute to heightening the effects ofglare from vehicle headlamps.

Causes of Headlight GlareWhile the fundamental factors that determine

disability and discomfort glare were discussed inChapter 2, to better understand the basis of driv-ers’ glare problems one must consider how thesefactors are affected by geometric and vehicleparameters. While it is not possible to attributecausality to any one factor, it is useful to identifythe roadway and vehicle conditions that contributeto glare and to gather evidence to try to isolate themost serious offenders.

Illuminance from the glare source is deter-mined by the photometric intensity distribution ofthe oncoming headlamps, the aiming and height ofthese lamps, whether high beam or low beam is

used, and the distance of the glare source from theobserver. The greater the intensity directed towardan observer, the greater the illuminance reachingthe observer’s eyes. Headlamp intensity is con-trolled by the headlamp design and the beam pat-tern, discussed elsewhere in this chapter. In gener-al, headlamps are designed and aimed to producegreater intensity below the horizontal and to theright of the vehicle center line. While the FederalGovernment’s published standard FMVSS 108 isintended to control glare by limiting the amount oflight above the horizontal axis, roughly half of allvehicles on the road are driven with improperlyaimed headlamps (Copenhaver and Jones 1992).The reasons for this are numerous and are dis-cussed below, under aiming as a countermeasurefor glare.

The closer the observer is to oncoming head-lights, the greater the illuminance and, therefore,the greater the glare. The inverse-square lawdescribes the amount of light reaching an observ-er; it simply states that illumination from a pointlight source is inversely proportional to the squareof the distance from the source. Figure 2, below,shows the illumination resulting from two lightsources of different intensity levels as a function

30

Figure 2. Illumination at increasing distance from two light sources.

of distance from the source. There are two phe-nomena to observe: 1) At near distances, illumina-tion increases rapidly as the light source isapproached; and 2) illumination falls off rapidlywith increasing distance—if the distance isincreased by 50%, the intensity must more thandouble1 to obtain the same level of illumination.In other words, while glare increases dramaticallyas a light source is approached, there is an equallydramatic reduction of illumination as distance isincreased.

While illumination increases with proximity toan oncoming vehicle, in practice the level of glareis often reduced. This happens because the glareangle increases as the glare source gets closer and,if the observer continues to look in a given direc-tion, this increase in glare angle offsets the effectsof increasing illuminance. The dynamic relation-ship between distance and veiling luminance isdiscussed in Chapter 5.

High-intensity discharge (HID) headlights are arecent advance in lighting technology with numer-ous advantages for both auto manufacturers andconsumers. Auto manufacturers like HID head-lights because of the reduced requirements forpower, greater control of the beam pattern, andgreater flexibility in stylistic parameters.Consumers who have them like them for their sty-listic properties and because they offer dramaticimprovements in visibility. Consumers who don’thave them are divided between those who wantthem for stylistic reasons and those who hate thembecause they are perceived as responsible for glare.

The importance of style is shown by the popu-larity of aftermarket conversion kits, which offerthe style without the improvements in visibility.Several companies offer halogen bulbs that arecoated blue to look like HID bulbs. These bulbsare not legal and produce less light than a normalhalogen bulb. While they may be annoying, theyare not contributing to the glare problem. Otherconversion kits that use true HID bulbs (usuallyxenon lights) are available, but they are limited in

practice to systems that use single-filament bulbs(separate bulbs for low and high beams). Theseconversions cost upward of $1,000 and are notguaranteed to have the photometric performancerequired by government safety standards.Although both the cheap blue halogens and theexpensive HID conversions are illegal in theUnited States, they are nevertheless being pur-chased and installed. While there does not appearto be any data about the prevelance of these unitson U.S. roads, their popularity is evidence of theimportance of style to the consumer.

HID headlamps typically have two to threetimes the light flux (volume) of halogen lights,but because the HID filament is smaller, theyallow the light to be controlled more precisely.This causes a sharper cutoff, which results inboth higher intensity on many beam angles belowthe cutoff and in flashing, which is producedwhen roadway undulations cause the cutoff tosweep quickly up and down across oncoming driver’s eyes or the mirrors of preceding vehicles.As a result the design gradients for HID lampstend to be closer to the test points. Because ofthe larger filament size of halogen sources, alamp designer must use more gradual transitionsor risk having a greater proportion of units failphotometric testing. Properly aimed, and whennot flashing, these HID lamps may not appearsignificantly brighter than halogen sources iftheir size is relatively the same. However, if misaimed, and when flashing, these lamps willappear brighter and will produce significantlymore glare than a normal halogen lamp.

Another problem magnified by the greaterintensity of HID is dirt accumulation on its lens.Dirt acts as a diffuser on any headlamp and canresult in additional stray light being directed abovethe horizontal axis into the eyes of oncoming driv-ers. When the total light flux is tripled, as it iswith HID, this diffusion becomes a much greaterproblem. While headlight aim has always been aconcern, HID may force the introduction of coun-termeasures to respond to the misaim problem.

31

1Intensity must increase by a factor of 2.25 to maintain the same illumination when distance is increased by 50 percent underthe inverse square law: (E1 d1

2 = E2 d22).

Whether one encounters legal factory units or ille-gal after market conversions, improperly aimedHID systems are going to exacerbate any problemthat might exist with glare. Even if properlyaimed, the sharp cutoff and the flashing this pro-duces, may require further control of the beam pat-tern. (See the discussion of the photometric distri-bution in Chapter 5.)

The glare angle from oncoming headlights is theangle between the direction of a driver’s gaze andthe direction of the glare source. Glare angledepends on the distance between the opposingvehicle and the observer vehicle, the road geome-try, and the offset of the opposing vehicle paths,which is determined by the number of lanes in onedirection, the lane of occupancy, and the median orshoulder widths. Overall, the glare angle is small-est when the opposing vehicle is furthest away sothe illumination is low. Fortunately, when theopposing vehicle approaches and illuminationincreases significantly, the glare angle becomessufficiently large to minimize the effect of glare.Two-lane roads and freeways without medians orleft shoulders provide environments where theglare problem is exacerbated by small glare anglesat close distances, where there are high levels ofillumination. While the glare angle is increasedduring encounters on right-hand curves, glare isactually worse on left-hand curves becauseapproaching drivers are exposed to the brighterparts of the beam pattern.

Dynamic relationships between beam patternand distance to an opposing vehicle were evalu-ated for several lateral separations by Powers &Solomon (1965). This report showed that, for atypical two-lane road approach, the level of glareincreases gradually as the oncoming car approach-es from a separation of 2,000 feet until it reaches300 to 400 feet; as the car comes closer than this,the glare level drops sharply. At first, the glare iscaused by the portion of the beam pattern that isnear the peak intensity point; although relativelybright, the source is far away. As the vehiclesapproach each other, the distance declines, but thedriver is seeing less candlepower (further to theleft of center on the headlamp beam distribution).The eventual dropoff in glare is so sharp both

because of the large increase in glare angle andbecause at short distances the light comes fromparts of the beam pattern where the output is verylow. With increased lateral separation, the overallveiling luminance decreases and the peak willoccur at somewhat greater distances. At large dis-tances, veiling illuminance is greater on left-handcurves and less on right-hand curves because ofthe portion of the beam pattern to which approach-ing drivers are exposed. This is discussed furtherbelow, under wide medians as a countermeasure.

Background luminance is generally determinedby the reflectance of background (usually thepavement) and the illuminance on the pavement(usually from headlights at near distances andambient light or fixed roadway lighting at far dis-tances). When driving, a driver may look atobjects on or off the road. Some objects, such asoverhead signs, may be seen with the sky as abackground, while objects on the side of the road,such as signs, may be seen with buildings orfoliage as a background. Most objects have thepavement as a background, so the luminance ofthe pavement is thought to have the greatest effecton driver eye adaptation. In general, concretepavements are more reflective and therefore havehigher luminance than asphalt; however, thereflectivity of pavements is affected by wear andother factors. Pavement luminance contributes tothe glare problem primarily when illumination islow, as is the case on dark rural roads. The glareproblem is apparently not as severe on brighterurban roads because the driver’s adaptation level ishigher. This observation is the motivation for theuse of fixed lighting as a countermeasure for glare.However, fixed lighting will produce its ownglare, which must then be controlled.

Size of the glare source is determined by thephysical dimensions of the glare source and thedistance between the driver and the glare source.At close distances, headlamps are no longer pointlight sources; increasing their size while holdingtheir light output constant will reduce discomfort.On the other hand, as an observer approaches areflective surface, the relative size of the surfaceincreases but, as a result of the inverse-square law,the light intensity and the discomfort glare also

32

increase. This apparent disparity in the effect ofsource size on glare is due to the fact that whenheadlamp size is increased, luminance is reduced,whereas the luminance of a reflective surface gen-erally remains constant when its size is increasedunder constant illumination. Headlamp size is gen-erally not a significant factor in glare since there isnot much variability in headlamp size with currentdesigns. However, as we shall see in Chapter 8,the development of larger-sized headlamps is apotential countermeasure to discomfort glare,despite the current trend towards smaller headlamps.

Glare source luminance is determined by boththe intensity and the area of the headlamp:Luminance is increased when either intensity isincreased or area is decreased. As mentioned ear-lier, the intensity of the lamp is modulated by thebeam pattern and luminance will vary dependingon the angle from which it is viewed. As long asheadlamp size is not varied, there does not appearto be any problem with headlamp luminance thatis not related to headlamp intensity, beam pattern,or aiming.

Driver age affects the experience of glare on theroad in the same way it does in the laboratory.Age, as we saw in Chapter 2, has a significanteffect on the magnitude of disability glare: Olderdrivers encounter higher levels of disability glarethan younger drivers under the same lighting con-ditions, and anecdotal reports indicate that olderdrivers complain more about glare and are morerestricted in mobility at night. However, there issome inconsistency in the literature concerningwhether older drivers are more discomforted byglare than younger drivers.

Reflective surfaces outside the vehicle can be aproblem. For example, reflective surfaces on aleading vehicle stopped at a traffic light can bevery discomforting. Interior surfaces of vehiclescan also become illuminated and cause some glare,but under current Motor Vehicle Safety Standards,adopted in 1966, vehicle surfaces are required tohave glare-reducing matte finishes (this is why theold chrome windshield wipers have disappeared).

This countermeasure is already well implementedand so will not be discussed further. A removableobject in the vehicle can also be a glare source, butthe driver always has the option of moving such an item.

If the glare source is a reflective surface, boththe illuminance and luminance from the source aredependent on the reflectance of the surface as wellas on the photometric properties of the light illumi-nating the surface. If the reflecting surface is insidethe observer’s vehicle, the illuminating source isusually the following vehicle. If the reflecting sur-face is on a lead vehicle, the illuminating sourcemay be the observer’s own headlights.

Glare from a reflecting surface on a leadingvehicle depends on the distance of the observerfrom the glare source. As the headway betweenthe vehicles increases, illumination from the lead-ing vehicle is reduced and the size of the glaresource projected on the retina of the eye becomessmaller. The problem is at its worst when twovehicles are stopped in traffic. Since both vehiclesare stopped, the glare can usually be controlled bylooking in another direction.

Glare from mirrors is the result of the reflectionof headlamps; its magnitude is based on the opti-cal distance to the image of the headlamps. Sincemirrors simply redirect the optical path, it is thedistance between the headlamp and mirror plus thedistance between the mirror and the driver's eyethat determine the illuminance at the driver’s eye.The transmission of the rear window and thereflection characteristics of the mirror (about 4% for a typical day/night interior mirror on itsnight setting, 50% for exterior mirrors) must beconsidered.

Olson and Sivak (1984) observed two distinctdifferences between glare from mirrors and glarefrom oncoming headlamps: First, there is typical-ly more than one mirror, which increases the glareproblem; and second, the following vehicle mayremain in a relatively fixed position for a longperiod of time, which raises questions about thetime-related effects of glare.

33

The principles governing glare from mirrorsare similar to those for oncoming headlights: dis-tance, beam pattern, intensity, and aiming all affectthe amount of glare. The part of the headlightbeam pattern of a trailing vehicle that is directed atthe mirrors of a leading vehicle will change as theheadway changes and depends on the mountingheight of the trailing vehicle’s headlamps. Theglare angle, which ranges from 35 to 55 degrees inpassenger cars (Olson and Sivak 1984), is deter-mined by the location of the fixed mirrors and notby the location of the following vehicle. In gener-al, any countermeasure that can decrease the head-lamp intensity, increase the headway, or reduce theheight of following vehicles will reduce the illumi-nance from the glare source and so also reduce theamount of mirror glare.

Types of CountermeasuresChapters 5 through 8 describe various counter-

measures that have either been implemented orproposed for controlling the effects of glare fromvehicle headlights. Some of these countermea-sures are intuitive and have been deployed to someextent for many years. Others rely on state-of-the-art technology to achieve results that are not at allobvious. Table 1 lists a wide variety of potentialcountermeasures, characterized by the primarysource of implementation, whether it be the driver,industry, or a government agency.

While there are many methods for reducingthe amount of glare attributable to headlights, onlya few countermeasures may be implemented uni-

laterally by the driver. Manycountermeasures must eitherbe implemented by a high-way agency (usually the statedepartment of transportation,or DOT) or by automobilemanufacturers, with or with-out the Federal mandates thatmake specific types of equip-ment available or legitimizechanges in vehicle or head-lamp design. In Europe,headlight glare is limited byaiming the beam downward,whereas in the U.S., the beamis aimed upward. Bothapproaches recognize thetradeoff between glare andvisibility; the European stan-dard accepts reduced visibili-ty while the U. S. standardaccepts more glare. Theimplementation of thesephilosophies is discussed inChapter 5.

Countermeasures in Table1 are grouped according totheir principal method ofattaining a reduction in glare:reducing intensity, reducingillumination, increasing the

34

Chapter 5. Reduce the intensity or luminance of the glare source

Industry Photometric Distribution

Driver Aiming

Industry Adaptive Headlamps

Industry Color Corrected Headlamps

Industry Headlamp Height

Chapter 6. Reduce the illumination that reaches the driver’s eye

Government/Industry Polarized Lighting

Driver Night Driving Glasses

Government Glare Screens

Industry/Driver Anti-Glare Mirrors

Chapter 7. Increase the glare angle

Government Wide Medians

Government Independent Alignment

Chapter 8. Methods that indirectly may result in glare reduction

Industry Ultraviolet Headlights

Government Fixed Roadway Lighting

Government Restricted Night Driving

Government/Driver Corrective Lenses and Ophthalmic Surgery

Industry Headlamp Area

Table 1. Countermeasures grouped by method of implementation

glare angle, or through an indirect effect. A briefdescription of these four categories is presentedbelow. Some countermeasures may fit in morethan one category; we have placed each in the cat-egory that seems most appropriate. Table 1 showsthe countermeasures to be discussed in each cate-gory and whether the driver, industry, or a govern-ment agency should take the lead in implementa-tion.

Reduction of intensity or luminance. Somecountermeasures reduce glare by reducing theluminance of the glare source or the intensity oflight aimed in a driver’s direction. Modificationof the low-beam pattern and independent align-ment of opposing directions of road are examplesof this type of countermeasure.

Reduce illumination at driver’s eye. Anothergroup of countermeasures is intended to block orfilter light, thereby reducing the amount of illumi-

nation reaching a driver’s eye. Anti-glare mirrors,glare screens and some types of night-drivingglasses (those marketed as glare reducers and notfor optical corrections) are prominent examples.

Increase the glare angle. Some countermeasuresreduce glare by increasing the angle between theglare source and the road ahead. Wide mediansare the most common method of implementingthis strategy.

Indirect benefits. The last group of countermea-sures do not themselves directly reduce glare, buthave the potential to indirectly reduce glare byeither reducing the illumination needed for visionor raising the adaptation level of drivers.Ultraviolet lighting and fixed roadway lighting areprominent examples of these countermeasures.

35

• Photometric Distribution• Aiming• Adaptive Headlamps• Color Corrected Headlamps• Headlamp Height

Photometric Distribution Every driver is aware that some headlights are

brighter than others. This variation in brightness isthe result of variation in the beam pattern, aiming,and height of the headlamps. The beam pattern inthe U.S. is governed by Federal Motor VehicleSafety Standard 108 (FMVSS 108), which specifiesminimum and maximum values of luminous inten-sity at approximately 20 test points given by theangles in degrees from the horizontal and verticalaxis of the lamp. The minimum values are intendedto insure the ability of drivers to see critical targets

such as lane lines, signs, and pedestrians, whereasthe maximum values are the countermeasure to con-trol glare. Originally, FMVSS 108 had maximumlimits but no minimum requirement for light outputabove the horizontal; recently, the standard wasmodified to require some minimum levels abovehorizontal so as to maintain the visibility of retrore-flective overhead signs.

The light output from a headlamp, as meas-ured by its intensity in candelas, is not uniform inall directions; rather, it varies according to a pho-tometric distribution that is documented in a head-lamp beam pattern. A beam pattern representingthe median value of 26 sealed-beam and replace-able-bulb lamps conforming to U. S. standards isshown in Figure 3. In this example, the headlightintensity is represented by the shading at all pointsspecified by the angle from the horizontal and ver-tical axes. The upper band in green representsintensities less than 400 cd; the next band repre-sents intensities between 400 and 800 cd; and eachsuccessive band closer to the center represents afurther doubling of intensity, to 1600 cd, 3200 cd,

37

CHAPTER 5

COUNTERMEASURES THAT REDUCE

THE INTENSITY OF GLARE SOURCE(Control of the Lower Beam Pattern)

Figure 3. US headlamp beam pattern mapped from -20 to +20degrees horizontal and -5 to + 5 degrees vertical.

and so on. The inner circle represents a lightintensity greater than 12,800 cd.

The headlamp beam patterns encountered onthe highway vary considerably as a result of sever-al factors. First, headlamp regulations such asFMVSS 108 specify minimum or maximum out-put only at a small number of test points. Lightoutput can always be greater than the minimum orless than the maximum at these test points and cantake any value at all other locations. Lightingtechnologies to date have resulted in beam patternsthat vary gradually throughout the beam pattern,without major reversals in intensity, which effec-tively spreads the control of a few test points overa much larger area and minimizes variability tosome extent. Still, lamps made for different vehi-cles by different manufacturers will have dramati-cally different beam patterns.

The potential of using modifications in thebeam pattern as a countermeasure to control glareis most evident when the U.S. and European head-lamps are compared. European countries havechosen to set the beam pattern to protect oncomingand leading vehicles from glare, to permit visualaiming, and to provide wider and brighter fore-ground illumination than is found in the U.S.Compared with U.S. headlamps, the Europeanlamp has a “low cutoff” without a peak, whichlimits the amount of light emitted above the hori-zontal axis. The absence of a peak in the beampattern makes the lamp easier to aim visually. Anegative consequence of sharp cutoff is the occur-rence of flashing, which occurs when the vehicletraverses roadway undulations and the cutoffsweeps quickly across oncoming drivers’ eyes andmirrors in preceding cars.

A beam pattern representing the median valueof 33 halogen lamps conforming to European stan-dards is shown in Figure 4. The sharp cutoff ofthe European headlamp is apparent from the hori-zontal contour lines, which rise from left to rightuntil the cutoff position is reached. By contrast, inthe U.S. headlamp the contour lines peak just tothe right of zero on the horizontal axis and gradu-ally decline to the right of the peak. By movingthe peak to the right, European headlamps directmore light above the horizontal axis towardobjects on the right side of the road and less lightto the left, limiting the exposure of oncoming driv-ers to glare. The U.S. beam pattern directs consid-erably more light above the horizontal axis than theEuropean pattern, resulting in better visibility ofoverhead signs and other objects on vertical curves.Such improved visibility has been realized evenwithout minimum requirements, which until recent-ly did not exist, to ensure that some light be direct-ed above the horizontal axis.

In addition to fundamental differencesbetween the U.S. and European beam patterns,there is a very significant difference between theway headlamps are aimed in Europe and in theU.S. Headlamps in the U.S. are aimed so thebeam pattern in the real world is aligned identical-ly to the pattern measured in the laboratory. InEurope, headlamps are aimed downward between1.0% and 2.0% when installed, depending onheadlamp height. After installation a 3% down-ward aim is acceptable. The European beam pat-tern and downward aim result in less glare, butalso in shorter seeing distances and much less sign illumination. Another undesirable consequence isthe flashing mentioned previously.

38

Figure 4. European beam pattern mapped from -20 to +20degrees horizontal and -5 to + 5 degrees vertical.

The U.S. has resisted adopting the European beampattern and aiming practice because it severelylimits visibility distance. Limiting the effectiveillumination range of low beams to 200 ft or less isnot acceptable on U.S. roads, where low beams areused for urban and suburban driving at relativelyhigh speeds. In Europe, low beams are used strict-ly for city and heavily urban low-speed driving.To accommodate the European practice in theUnited States, all overhead signs would have to beilluminated and roadside signs would have to beplaced lower than the 7-ft standard on U.S. ruralroads. Signs mounted nearer to the road surfaceare not only more difficult to see because of block-age by other vehicles and by snowbanks innorthen climates, but also are more susceptible tothe accumulation of dirt by splashing from vehi-cles moving past them, which further reduces theirvisibility.

In the United States, users of illegal European(ECE) headlamps sometimes aim the cutoff at thehorizon, following the U.S. practice, which givesmuch better seeing distance but more flashingglare. Users seem to think that the ECE beam issuperior, but this is only true if it is misaimed. Therecent amendment to FMVSS 108 to provide anoption for visual/optical (VOA) headlamp aiminghas resulted in headlamps with a sharper cutoff forvisual aiming purposes, but in their pursuit of aheadlamp harmonized with the ECE beam pattern,vehicle manufacturers sometimes demand a cutoffthat is sharper than necessary for the UnitedStates. Such a sharp cutoff, with the high aimingnecessary to meet the requirements of FMVSS108, make flashing more likely to occur in theUnited States. Setting a limit on the cutoff sharp-ness could help reduce this adverse glare problem.

It is also likely that attempts to develop head-lamps that meet both U.S. and European require-ments will reduce the amount of light availableabove the horizontal axis. Headlamps that wouldonce have come close to the maximum valuesallowed by FMVSS 108 will, if harmonized withECE requirements, produce light much closer tothe minimum values allowed by FMVSS 108.While this may reduce some of the problems driv-

ers encounter with headlight glare, it will alsoreduce the visibility of traffic signs.

Comparing the U.S. and European low beam,Sivak et al. (1992) point out that each approach isadvantageous in certain traffic conditions, but nei-ther appears to be superior overall. They suggestthat this lack of a generally superior alternativeappears to have created a positive attitude towardinternational harmonization. Use of adaptive, or“smart,” headlamps is another approach to resolv-ing this disparity, because such lamps seek tomaintain visibility while controlling glare.

Research on the Low BeamPhotometric Pattern

There have been numerous attempts toimprove low beam headlighting patterns, but,while these attempts have advanced understandingof the problem, a comprehensive solution has notbeen attained. The goal has been to develop abeam pattern that provides adequate visibility,such as of pedestrians, signs, lane lines, and obsta-cles, while at the same time minimizing glare foroncoming traffic and into rear mirrors. Theseefforts have included computer modeling,research, and policy efforts, a few of which arebriefly described here.

In the 1970s, the Ford Motor Company sup-ported the development of computer based visibili-ty models that could be used to evaluate alterna-tive lighting systems. In the 1980s, NHTSAbecame interested in developing a vehicle per-formance standard for headlighting that wouldimprove safety while giving manufacturers morefreedom in lighting design. These efforts con-tributed significantly to knowledge of roadwaylighting, but did not result in any consensus forchanges in FMVSS 108 due to disagreements overassumptions and methodological problems. Facedwith difficulties in analytically evaluating head-lamp performance and in defining a new perform-ance standard, NHTSA turned its efforts toimproving the aim of the beam patterns thatpresently exist. As a result, a revision was made toFMVSS 108 in 1997 that allowed for an optional

39

set of test points to improve objectivity and accu-racy in the aiming of VOA headlamps. Currently,research is being conducted at the University ofMichigan to develop a subjective rating of thephotometric distribution of the forward illumina-tion of different vehicles for different types ofdriving.

Headlamp comparisons with visibility models —Beginning in the 1970s, researchers at the FordMotor Company (Bhise et al. 1977) developed acomputer model named CHESS (ComprehensiveHeadlamp Environment Systems Simulation) toevaluate the performance of low-beam headlight-ing patterns. The model computes a weighted sumof several performance measures—including thefraction of pedestrians and delineation lines detect-ed with and without an opposing glare car, and thefraction of drivers that were discomforted byglare—and so obtains a figure of merit (FOM) foreach headlamp evaluated. The components of theFOM that the model computes are also useful per-formance measures by themselves. Basic inputs tothe model for a headlamp evaluation include theheadlamp beam pattern, height, and aiming; envi-ronmental variables, and driver vision variables.

The FOM reflects the fraction of the total dis-tance traveled on a simulated test route in whichthe visual environment can be considered ade-quate, where “adequate” is defined as being ableto see both pedestrian and delineation targets atsafe distances without unacceptable levels of glare.The simulated route included roads representativeof the various road types and topography in theU.S. Calculations for the FOM are based onencounters with opposing vehicles and pedestrianson a mixture of different road types.

The FOM is highly dependent on severalsomewhat arbitrary assumptions, including wherepedestrians are placed within the road geometryand the importance, or weight, given to discomfortglare. In addition, the results of the model aredependent on the way the performance criteria aredefined. However, the more basic problem is abasic consequence of the inverse square law:Because illumination decreases with the square ofthe distance from the source, very large increases

in candlepower are required to create a usefulincrease in illumination. Comparisons in the studyof prototype headlamps showed that there wasonly a 10% difference in performance betweendesigns, whereas variation in environmental condi-tions produced a performance change of 60%.The authors pointed out that there were “no meansof relating the Figure of Merit to safety and cer-tainly no basis for assuming that a given change inthe Figure of Merit will produce a proportionalchange in accidents.”

The Ford Motor Company studies concludedthat the U.S. low-beam pattern is close to the opti-mum. Several alternate beam patterns were evalu-ated, including a mid-beam design proposed byNHTSA that attempted to improve visibility with-out increasing glare by concentrating the beamintensity in the right lane. The problem with thisdesign was that light was directed where intendedonly if the beams were properly aligned and thecar was on a straight road. While the mid-beampattern produced some improvement in perform-ance, it might be difficult to control the beam pat-tern to the tolerances necessary to produce thedesired result. Internal research by NHTSA withthe CHESS model showed that while some alter-nate beam patterns resulted in a modest improve-ment in performance, improved performance couldalso be obtained by changes in mounting heightand by improved aiming.

In the mid 1980s, NHTSA became less inter-ested in headlamp comparisons and more interest-ed in the development of a headlamp standardrelated to driver needs, particularly with regard tosafety. The important question was not what low-beam pattern was best, but rather how to specify alow-beam pattern that met driver requirements forvisibility and the control of glare, given that thespecification would follow the current method ofspecifying minimum and maximum beam intensityat a limited number of points (with some variationto allow for the necessary manufacturing tolerances).

Vehicle-based performance standard — One ofNHTSA’s primary goals has been to establish a setof vehicle specifications that could replace exist-

40

ing specifications that apply only to headlamps. In1985, as part of its comprehensive review ofFMVSS 108, NHTSA published a Notice ofRequest for Comment asking for suggestions onhow to make the standard more performance-ori-ented and less design-restrictive. The goal was toreduce the burden on the regulated parties whilesimultaneously reducing the burden on NHTSA ofresponding to design-related requests for changesin requirements. Prior to this notice there hadbeen an increasing volume of petitions from vehi-cle and headlighting manufacturers developingnew headlighting systems. Another motivation forthe review was concern that, because there werethen no minimum requirements for headlamp lightintensity above the horizontal, headlamps beingintroduced—mostly by European manufacturers—might not provide sufficient illumination for ade-quate visibility of signs.

In general, the comments received were infavor of requirements that were more perform-ance-oriented, and the process gave rise to a long-term NHTSA effort to develop a vehicle-basedroadway illumination performance requirement.There was also hope that a performance-basedspecification could lead to harmonization betweenU.S. and European (ECE) specifications. However,there was almost universal agreement that any newperformance requirement should not involve test-ing of the vehicle as a whole, only the headlight-ing devices.

One early NHTSA approach to developing avehicle-based specification was to simply convertthe existing lamp specifications to a correspondingset of vehicle specifications. This approach wasnot adopted for the low beam for several reasons:First, there was no basis for relating the specifica-tion to satisfaction of safety requirements; second,it did not address the concern about limited lightabove the horizontal; and finally, it did not providea basis for continuing international discussions onharmonization of headlight specifications.

The vehicle-based approach which was even-tually adopted consisted of identifying a set ofdriving conditions where roadway illumination isneeded on the basis of safety and then determining

the necessary amount of light to provide visibilityfor safe driving in these situations. The drivingsituations that were identified included the follow-ing safety elements: avoidance of pedestrians andother on-the-road objects, staying within laneboundaries, control of oncoming (direct) and fol-lowing (mirror) vehicle glare, and highway signillumination. The driving situations were chosenand prioritized based on careful analysis of fataland non-fatal day and night accident data in whichaccidents were characterized as pedestrian, off-road, roadside, and overturn. According to theunpublished data used in the analysis, which con-sidered both the number of accidents and the acci-dent rate per million vehicle miles, pedestrianaccidents seemed to be of most concern on urbanarterials and off-road accidents seemed to be ofmost concern in rural areas. This approach hadthe advantage that it could provide a basis fordeveloping any of several vehicle-based specifica-tions, including those related to the existingrequirements and compatible with the performanceof existing lamps, or those that provide animprovement in safety by using vehicle lightingsystems.

To help in establishing the vehicle-based per-formance standard, computer models were devel-oped that were based on the original Ford visibili-ty models. These models determined the amountof light needed to see pedestrians, signs and lanelines at distances required for safety, with andwithout a glare car present. Both direct glare andmirror glare were considered in the models. Morethan a thousand individual illumination require-ments were identified and then condensed intobetween 20 and 30 test points, which then becamethe basis for a proposed rulemaking issued in May,1989. Some of the test points were minimumrequirements to provide for visibility of signs(including overhead signs), pedestrians and lanelines; others were maximum values to controlglare. Unlike FMVSS 108, which specifies thelight intensity at specific angles from the beamdirection, the proposed rulemaking specified theminimum and maximum illumination at targetpoints defined in terms of both angle and distance.Specification of the illumination at a distanceinstead of the intensity along a beam angle was

41

necessary because of the variations in headlampplacement, particularly mounting height. A com-puter program was needed to determine whether aparticular headlamp’s photometric pattern couldmeet the requirements when the headlamps weremounted at the height and spacing appropriate to aparticular vehicle.

Proposals for a vehicle-based performancestandard were rejected for a variety of reasons,including disagreements over the assumptionsmade in the modeling, methodological errors, andthe overall conceptual complexity of a vehiclelighting system. Headlamp height, itself a motiva-tion for development of the new standard becauseof the serious problems with mirror glare,increased the complexity of the vehicle standardand so contributed to its demise. Numerous con-troversial assumptions had to be made for suchfactors as pedestrians’ location, activity, and cloth-ing; sign retroreflectivity; drivers’ age and visualabilities; and roadway geometries. Methodologicalproblems included determining weighting factorsthat reflected priorities among the conflictingneeds for visibility and control of glare and estab-lishing the amount of glare to which drivers weresubjected.

Effects of Increasing the Intensity of the LowBeam Pattern — Flannagan et al. (1996) conduct-ed an empirical study of the distance at whichpedestrians could be seen, in which headlampintensity was increased while the light intensityfrom an opposing glare vehicle was also increasedproportionately. The study found that increasingthe intensity of light from both vehicles by a factorof 3.8 increased the visibility distance by 17%.This result is consistent with previous research(Hemion 1969a, Johansson et al. 1963), whichfound that visibility improved when high beamsopposed high beams compared to situations whenlow beams opposed low beams.

These results suggest that, if visibility distanceis the sole criterion, there may not be any upperlimit to how intense low-beam headlamps shouldbe. However, such a conclusion is not completelywarranted, because differences in beam patterns,aiming, and road geometry mean that the condi-

tions found in controlled field experiments, wherethe illumination of two opposing vehicles could beincreased by equal amounts, are unlikely to occurin practice. Further as Flannagan pointed out,“there may be a level at which people simply willnot tolerate the subjectively discomforting effectsof glare, or at which glare indirectly affects objec-tive performance through its effects on subjectivecomfort.” Flannagan added that research is need-ed to understand the consequences of discomfortglare, “including possible effects of discomfortglare on objective behavior.” Theeuwes &Alferdinck (1996) found that subjects were lesswilling to look at a light source when illuminationwas higher. Discomfort glare may affect perform-ance by increasing fatigue or by causing drivers tolook away from the road so that some objects onthe road could only be seen with peripheral vision.

Need for Sign Illumination — Lighting may bemade more efficient by improving sign legibility,using larger, more readable letters and the mostretroreflective sheeting available. Russell et al.(1999) concluded that with signs placed onstraight and level roads and made of high-perform-ance Type III sheeting, “better than 99% of the1,500 vehicles observed would provide sufficientillumination for right-shoulder mounted signs andmore than 90% of these vehicles would providesufficient light for the left-shoulder mounted signs,but only about 50% of them would provide suffi-cient light toward overhead signs.” These figuresdropped to 90%, 45%, and 10% for the more-com-mon Type II sheeting, or if the Type III sheetinghad been degraded for a few years. Sign legibilitywould be still less for older drivers; Russelldefined “sufficient illumination” only in referenceto younger drivers.

Given the desire in the U.S. for harmonizationwith the European beam pattern, it is unlikely thatthere will be any future increase in the amount oflight directed at signs. If the beam pattern were tobe altered to provide less light above the horizontalaxis, overhead signs would have to be externallylighted or replaced with repeated signs on bothsides of the road. Both solutions are costly, and,because of limited sight distance, roadside signs arenot as effective as those mounted overhead.

42

With the states continuing in their quest toeliminate self-illuminated signs because of theirenergy and maintenance costs, it appears that thecurrent preference of government is to continue torely on illumination from low beams for the visi-bility of reflective signs. Some relief from signillumination requirements is offered by new typesof retroreflective sheeting, such as the proposedASTM type IX, which is more efficient at thewide entrance angles created by sign placementoverhead or at extreme offsets. Improvement inretroreflective materials should certainly be seenas a countermeasure to glare. Whether these mate-rials provide sufficient retroreflection for visibilitywith the illumination provided by today’s vehiclesis open to question, but surely the use of thesematerials would limit the need for headlamps togenerate more illumination and thus more glare.

Rating System for Consumer Education —Currently, NHTSA is taking a very differentapproach from that of the past. Instead of tryingto pick the best headlight system or to develop avehicle standard that would enable all headlightsystems to satisfy driver performance require-ments, the focus is now on developing a ratingsystem that would empower consumers to pick theheadlight system that best meets their needs andtype of driving.

A study (UMTRI 1999) was initiated in 1999by the University of Michigan TransportationResearch Institute to assess the feasibility of vehi-cle headlamp ratings of new cars that would pro-vide buyers with information analogous to whatthey might learn from a nighttime test drive. Forexample, NHTSA might rate headlamps as part ofthe New Car Assessment Program (NCAP), whichgives new vehicles one to five stars on the basis ofcrash tests. Headlamp performance might similarlybe rated from one to five stars and the informationposted on the NHTSA web site. The benefits of aheadlamp rating system might include helpingindividual consumers to make choices that betterserve their needs and helping to bring about a gen-eral improvement in the quality of headlighting byincreasing public awareness of differences in head-lamp quality.

Advantages to Altering the Low-Beam Photometric Distribution

Altering the low-beam photometric pattern canreduce glare for on-coming drivers, improve thevisibility of pedestrians, roadside objects, and left-mounted and overhead signs, and make it easier tovisually aim the headlamps. Achieving all threegoals is generally not possible, so priorities mustbe set and tradeoffs made.

Reduced glare. Reducing light above the hori-zontal axis, for example by using the Europeanbeam pattern, can significantly reduce drivers’exposure to glare.

Improved visibility. Requiring a minimumamount of light above the horizontal axis and tothe left can improve the visibility of pedestrians aswell as of signs located to the left or overhead.This requirement has been implemented to someextent by the VOA beam pattern.

Visual aiming. Manipulating the beam pattern tohave a recognizable horizontal cutoff makes it easi-er to visually aim the headlamps and to ensure thatproper aiming is maintained. This can improve vis-ibility as well as limit exposure to glare.

Disadvantages to Altering the LowerBeam Photometric Distribution

The disadvantages of altering the low-beamphotometric pattern are opposite the advantages:

Increased glare. Attempts to provide light in cer-tain locations for left-mounted and overhead signscan increase the level of glare, particularly whenvehicles meet each other on horizontal curves.

Reduced visibility. The most significant disad-vantage to lowering the photometric beam patternor moving it more toward the right edge of theroad is generally reduced sight distance andreduced visibility of left-mounted and overheadguide signs.

Poor visual aiming. Attempts to create a peak inthe beam pattern to improve the visibility of certain

43

objects, such as overhead signs, reduce the accuracyof visual aiming and result in poor illumination ofthe desired targets on horizontal curves.

Summary“In Everyman’s conception of safe and com-

fortable night driving, the headlamps of his vehicleilluminate the roadway far ahead, and the head-lamps of other vehicles never decrease his visionby glaring into his eyes, either through the wind-shield or from the rear-view mirrors” (Haney andMortimer 1974). Attempts to reach a compromiseby allowing some light above the horizontal axis,but not enough to cause a severe glare problem,have proved very difficult. As mentioned above, itis generally accepted that sight distances as low asthose in Europe are not suitable on U.S. roads.Exactly what distance is necessary has never beenagreed on because it depends on vehicle speed,what needs to be seen, and the threat potential ofthe visual information.

The four visual targets generally consideredmost important include road delineation (includingedge lines, pavement, and delineators), pedestrians,signs, and objects on the road. Studies of accidentdata have not been able to rank-order these targets.Many believe that road delineation, which whendeficient can contribute to run-off-the-road acci-dents, is the most important, while others think thatthe focus should be on pedestrians. Without a clear,unequivocal ranking of what needs to be seen basedupon expected accident reduction, the appropriatetradeoffs between seeing distance and glare can notbe made. In designing roadway lighting, theIllumination Engineering Society has used a surro-gate vertical surface 7 inches square, with 18%reflectivity, to represent an object in the road.

Neither changing the current U.S. beam pat-tern nor increasing its overall intensity appears tooffer any meaningful advantage to drivers. Inaddition, because illumination decreases with thesquare of the distance from the source, very largeincreases in candlepower are required in order tocreate a useful increase in illumination. This hasled to the conclusion that it is not practical to pro-vide adequate illumination of pedestrians at speedsgreater than 35 or 45 mph.

Without a means of relating headlamp per-formance to safety, there is no basis for assumingthat headlamp changes would produce a propor-tional change in accidents. Instead of solvingproblems by changing the amount or the distribu-tion of light, existing light could be used moreefficiently by improved maintenance and design ofsigns (including such aspects as retroreflection andangularity), better delineation and other trafficcontrol devices, and by the use of countermeasuresdiscussed later in this report. Another approach isto reduce the need for visibility distance by plac-ing further limits on speed for night driving.Fisher (1970) has suggested that a maximumspeed of 50 mph should be considered for nightdriving. Certainly, differential day/night speedlimits are consistent with the recognized impor-tance of visibility to safety and the reduced visibil-ity that naturally occurs after dark.

Headlight AimIn order to achieve the beam pattern intended

by the lamp designer, headlamps must be properlyaimed. Misaiming headlamps will result inreduced visibility and/or increased glare for otherdrivers. Proper aim and a clean lens, not the beampattern, are the most significant aspects of head-lamp performance. While FMVSS 108 specifieslimits for headlamp intensity at several test pointsabove the horizontal, misaim and road curvaturealter the actual location of the projected beam pat-tern. As a car ages, headlamp alignment changesbecause of road vibration; however, rear loading ofthe vehicle, once thought to have a significanteffect on aiming, appears not to be an importantfactor (Copenhaver and Jones 1992).

The effects of misaim could be minor or dramatic, depending on the amount of misaim.Misaiming shifts the beam pattern horizontally orvertically; it should be apparent from Figures 3and 4 that the effects of horizontal misaim are notas great as those of vertical misaim. Sivak et al.(1993) found that horizontal misaim of 1.5 degreesin either direction had no practical significancewith U.S., European, or Japanese lamps. However,vertical misaim of 1.5 degrees had practical signif-

44

icance for U.S. lamps and even 1 degree miscuingresulted in significant effects for U.S. and practi-cally as significant effects for European lamps.Measurements of misaim are made using mechani-cal aimers that determine horizontal and verticalerror from an ideal target at 25 ft; the error can beexpressed in either inches or degrees. The SAElighting inspection code rejects headlamps withmore than a 4-inch vertical or horizontal error—this is slightly less than one degree.

Before 1983, all headlamps had sealed beamsand were equipped with aiming pads in a standard-ized location, making them all capable of beingmechanically aimed. Mechanical aimers measurethe level of the headlamps relative to the floorslope and to each other. Following the adoption ofreplaceable-bulb lamps in 1983, manufacturersdeveloped on-board mechanical aiming devices tosolve some of the problems of aiming with exter-nal mechanical devices.

In the 1990s, NHTSA began to consider alter-ing the low-beam photometric pattern to make itmore sharply defined. It was felt that the revisedpattern would facilitate visual aiming of head-lamps and might become the basis for a globalbeam pattern. Visual aiming was seen as desirablepartially because of General Motors data thatshowed that most facilities, both private and stateoperated, check and adjust headlamp aim visuallyrather than with the more precise mechanicalaimers. This observation was supported by studiesof headlamp aiming showing that only roughlyone out of every two vehicles have both head-lamps aimed properly (Olson 1985, Copenhaverand Jones 1992). As stated in the Federal Register(1995), “In the most common form, aim in stateinspections is judged subjectively by the eye of aninspector viewing a headlamp beam pattern castupon a distant vertical surface, such as a wall orscreen.” Automatic leveling devices are availableas optional equipment on more-expensive vehicles,and are often standard equipment, along withheadlamp washers, on vehicles equipped with HIDlamps.

In March, 1997, an amendment to FMVSS108 provided both a laboratory specification for

visually/optically aimable headlamps before instal-lation and a field specification for headlamp aimafter installation (Federal Register 1997). TheVOA laboratory specification included tables ofminimum and maximum intensity values similar tothose used for mechanical aiming of sealed-beamand replaceable-bulb lamps. The field specifica-tion intended for visual aiming provides for a cut-off or sharp transition between regions of high andlow luminous intensity. Horizontal aim musteither be fixed or the headlamp must have a built-in vehicle aiming device, because a cutoff forvisually adjusting horizontal aim does not exist—specifications for features in the beam that canresult in accurate visual horizontal aiming havenot been identified.

Research on Aiming Copenhaver and Jones (1992) conducted a

study of the headlamp aim of 768 vehicles, halffrom a periodic motor vehicle inspection (PMVI)state and half from a non-PMVI state. Both head-lamps were aimed correctly in slightly more thanhalf the vehicles from the PMVI state and inslightly less than half of the vehicles from the non-PMVI state. Vertical misaim was more commonthan horizontal misaim, but both were a significantproblem. Usually, vertical misaim moved thebeam pattern upward and horizontal misaimmoved it toward the left. Trucks and vansappeared to be more susceptible to misaim thancars. Variability was very large, with 10% of thevehicles misaimed more than what the equipmentcould record—a 10-inch error at 25 ft.

The study identified many factors that werenot related to misaim, including the type of head-lamp, its height above the ground, vehicle load,and design of the headlamp. There was no signifi-cant relationship between the number of monthssince headlamp inspection and headlamp aim, sug-gesting that shortening the inspection interval maynot reduce misaim.

One of the more interesting findings concernedvehicle age. Although the average amount of mis-aim did not vary significantly as a function of yearof vehicle manufacture, the fraction of vehicles with

45

misaimed headlights was significantly higher forolder vehicles. For example, both headlamps werecorrectly aimed in 68.9% of 1991 vehicles but inonly 35.8% of 1986 vehicles. In other words, oldervehicles were more likely to have their headlampsaimed outside the acceptable SAE tolerance,although the amount of misaim for any one yearwas no different from any other. This artifact ofthe study could be due to the extremely high vari-ance within each set of data, or it could be because10% of the misaim values were labeled “out ofrange,” limiting the misaim values in a way thatcreated a bias toward improving the score of oldervehicles.

The practical effects of misaim on glare fromrear-view mirrors were reported by Miller et al.(1974), who found that reflected glare from themisaimed headlights of following vehicles exceed-ed the just-tolerable glare from oncoming head-lights. With vehicle separations as short as 50 ft,properly aimed low-beam headlights produce lessmirror glare than direct glare from oncoming high-beam headlights at 600 ft. When misaimed, how-ever, the low beam can produce more glare withintercar spacings as close as 150 ft than thisoncoming high-beam level. The study found agreater increase in glare from low beams thanfrom high beams when headlamps were misaimedupward. Low beam headlamps misaimed upwardonly one degree increase the illumination in therearview mirror by a factor of three to four withthe following vehicle in the same lane, and a fac-tor of eight or more when the following vehicle isin the passing lane. With vehicle separations up to400 feet (following vehicle in the same lane),these misaimed headlamps produce more glarethan oncoming high beam headlamps at 600 feet,while vehicle separations of 150 feet or less wouldresult in more glare than from oncoming highbeam headlamps at 1200 feet. When the followingvehicle is in the passing lane, separations less than400 feet will produce more glare from headlampsmisaimed upward only one degree than oncominghigh beam headlamps at 1200 feet. To counteractthe negative effects of a low-beam misaim of onlyone degree, mirror reflectivity would have to bereduced to 10%.

Advantages and Disadvantages ofAiming as a Countermeasure

There is no question that incorrectly aimedheadlamps contribute to glare, and therefore properaiming is a real and effective glare countermeasure.Re-aiming headlamps annually during vehicleinspections is a low-cost procedure that would helpcorrect some problems that contribute to glare. Itremains to be seen what effect the availability ofvisually aimable headlamps will have on headlampaim in aging vehicles. There are no data to suggesthow frequently aiming should be checked, so onlythe more expensive automatic aiming systems canbe said to ensure a lasting improvement.

SummaryBecause misaim magnifies all other problems

with the low-beam photometric pattern, misaim isclearly the place to begin finding effective coun-termeasures for glare. Such measures as changingthe low-beam pattern or the headlamp height willnot be effective if headlamps continue to be mis-aimed. Research is needed to see whether theavailability of visually aimable headlamps hasimproved the aim of vehicles in the fleet, orwhether states should raise the headlamp aimingstandards in their inspection procedures.

Since public complaints seem to be mostlyabout glare from newer vehicles, headlamp aim invehicles with HID lamps seems to be a logicalplace to apply corrective solutions.

Adaptive HeadlightingFailure to perfect the lower beam pattern has

primarily resulted from inability to meet thedynamic requirements of the road and environ-ment, although variability in the height and spac-ing of headlamps and the difficulty in keepingthem properly aimed are contributing factors.Adaptive forward illumination refers to a tuning ofthe low-beam photometric pattern to meet thedynamic requirements of changing weather andgeometric conditions. It is based on the assump-tion that no static beam pattern is optimal for all

46

driving situations. While the same assumptionwas the basis for allowing manual switchingbetween high and low beams, drivers do not usehigh beams efficiently, if at all. Adaptive head-lighting seeks to satisfy lighting requirements thatchange with speed, road condition, topography,traffic conditions, and so on, by modifying thebeam pattern automatically.

Most development of these systems hasoccurred in Europe. Mannessero et al. (1999)described several functions that adaptive headlight-ing might perform, including bending (for improvedillumination of a curved lane), high-speed freeway(increased visibility distance and control of glare),and other modified beam patterns to meet the spe-cial requirements of town driving, country driving,adverse weather, and driving at dawn and dusk.Proposed systems either combine several lamps,each adding special-purpose beam patterns to abasic pattern, or use optical-mechanical parts thatprovide special lighting functions. Hogrefe andNeumann (1997) described the research and controlfunctions associated with adaptive light pattern(ALP) systems that use a combination of lamps.Kobayashi et al. (1997) reported the constructionand field evaluation of a system that used optical-mechanical parts together with high-beam, low-beam, cornering-beam, and fog lamps.

Honda Motor Corporation has developed anactive headlight system that redirects the beam inresponse to the action of the steering wheel. AnotherJapanese initiative uses activation of the turn signalto change the direction of the beam pattern. Otherdevelopment efforts are reviewed by Wörner (1999)and Rosenhahn (1999), and Duncan (1996) cites asystem that uses an array of headlamps that are selec-tively activated to alter the beam pattern.

Research with Adaptive HeadlightingManassero et al. (1999) reported the road-test

results of a system used to help develop the bestphotometric patterns for different lighting func-tions. Adaptive systems are in the very earlystages of development and additional research isneeded both to improve performance and to devel-op control strategies for determining when and

how the lighting system is switched from onefunction to another. Such research will undoubt-edly lead to the integration of additional sensorsthat can provide the needed data input.

Sivak et al. (1994) reported on three studies ofan active headlight (AH) system developed byHonda and Stanley Electric. One study evaluatedthe effects of the system on pedestrian visibility,another on the effect on discomfort glare, and athird on obtaining a subjective rating of the systemwith respect to several parameters. In the firststudy, the AH system increased visibility of thepedestrian by 14% on left curves and by 2% onright curves. Oncoming traffic was not included,so these results exclude the effects of disabilityglare. The second study found an increase in dis-comfort glare with the AH system on left curves,but an improvement on right curves, where themean DeBoer rating was better than 5 (accept-able). In the third study, the subjective ratingswere not very conclusive, with approximately halfof the subjects preferring the AH system and theother half preferring normal headlighting.Whether or not they liked the system, most sub-jects commented on the wider field of view it pro-vided. Many of the problems identified in thestudies could be addressed in a future redesign ofthe system; however, many of the subjects werebothered by the movement of the headlights, adrawback that might be difficult to overcome.

AdvantagesAdaptive headlighting offers the potential to

provide visibility beyond what is now availablewith the low-beam photometric pattern, and to doso without any increase in discomfort glare. Thelimitations of static systems, which cannot illumi-nate the same sections of the road on both straightand curved sections of highway, can be overcome.

DisadvantagesRegulations. Depending on the particular scheme pro-posed, adaptive headlighting might require new regula-tions to permit nonconforming beam patterns and vari-ations in aiming, as well as to define the switchingstrategies that determine when each is invoked.

47

Human Factors Effects. The negative reaction ofdrivers to moving headlights has been noted.More significant problems may occur as driverslearn that the adaptive headlighting system mayprevent them from seeing objects on a curve thatthey had been looking at along a straight road.

Tradeoffs Needed in Design Parameters. Effortsto choose the “best” beam pattern or define a vehi-cle standard based on the driver’s need for illumi-nation failed, in part, because of disagreementabout what drivers need to see and where theymust see it and the priorities for resolving con-flicts. The AH system does not escape these con-troversies; the system’s design and control strate-gies must answer the same questions. The sameassumptions that defeated efforts to develop soft-ware for the best beam pattern and attempts todevelop a vehicle standard are now being madeabout the design and construction of AH systems.

The skepticism of the preceding paragraph wasalso voiced by Duncan (1996), who suggested thatit may be difficult to dynamically reduce glare foroncoming drivers. A dynamic system would haveto sense oncoming headlights and reduce illumina-tion accordingly, but if both cars had adaptive sys-tems each would extinguish light directed in theother’s direction until the illumination droppedbelow the sensor’s threshold. This could result inan on/off oscillation around the threshold level,which could both be psychologically disturbing andproduce a negative impact on vision.

SummaryAdaptive headlighting is in its infancy, and it is

far too early to assess its effectiveness as a counter-measure to glare. One might ask the obvious ques-tion that, if we have not been able to maintain prop-er aiming of a fairly simple mechanical device suchas the replaceable-bulb headlamp, what new prob-lems will occur with a more complex system thatrelies on multiple sensors and a computer? There islittle doubt that such systems can be designed toimprove visibility, but their effect on glare will notbe known until they are subjected to research andevaluation. Still, there is a lot of interest in thesesystems, and one or more may be marketed soon.

Whether they will be a countermeasure to or a con-tributor to glare is not known at this time.

Color-Corrected MotorVehicle Headlights

An innovative proposal (Karpen 1998, Karpen2001) that might improve visibility and reduceeyestrain is to add color correction to sealed-beam,halogen, and HID headlamps. Like the proposalfor rear-view mirrors to be discussed in Chapter 7,the headlamp proposal (Karpen 1996) wouldincorporate at least 5% by weight of neodymiumoxide doping into the glass used for the head-lamps—including any glass with reflective sur-faces—so as to reduce the amount of yellow lightemitted by the headlights. To reduce the presenceof yellow light in the spectrum of HID head-lamps, neodymiunm oxide doping would be addedto the inner arc tube of the high-intensity lamp inconcentrations up to about 3.0% by weight, or tothe outer glass lens in concentrations between5.0% and 30% by weight (Karpen 1999).

The neodymium oxide doping allows produc-tion of a concentrated light beam with a uniquespectral energy distribution, which, according toKarpen, promotes night vision and visual acuity indarkness by emphasizing the contrast-producingred and green light portions of the visible-lightspectrum. The greatest absorption of yellow light-by the doped glass occurs for wavelengthsbetween 568 to 590 nanometers (Karpen 1996).

In addition to the visibility benefits, Karpenclaims that reducing the emission of yellow lightlessens eyestrain and the reduces the visual dis-comfort caused by the headlights of oncomingvehicles at night. With the yellow part of thespectrum removed, total light output of the head-lamp can be increased without increasing eye-strain. The increased amount of light results in bet-ter contrast and improved nighttime visual acuity.

The characteristic absorption of a neodymiumoxide glass (also called Neophane glass) affectscolor vision in a unique way. Red and green huesare strongly accentuated, and colors containing red

48

stand out especially clearly. Karpen suggests that,under light from Neophane glass headlamps, a redstop sign would appear redder and motoristswould find it easier to see road signs at great dis-tance against a background of green vegetation.

Research on Spectral Content of LightEffects on Visibility. Research on the effects oflamp color can be divided into studies of visibilityand studies of discomfort. Karpen (1996) cites anumber of research papers related to neodymiumlamps. Bouma (1938) describes color shifts underthe influence of Neophane glass, including a shiftof orange and yellow toward red that was experi-enced as an increased “warmth” of the yellow.Green, which under incandescent light became asomewhat dubious yellow-green, was restored togreen under Neophane glass. White and veryunsaturated colors were shifted in the direction ofblue-violet. Karpen believed that Bouma’s resultsshow that Neophane glass has the advantage ofpreserving most colors, albeit in a more saturatedform, and of making orange-yellow warmer.

A physiological explanation of how the eyesees color and the theory behind the hypothesizedeffectiveness of the proposed headlamp design isbeyond the scope of this paper. Some explanationis provided by Karpen, on the basis of a paper byGouras and Zrenner (1981). However, a key con-cept for the discussion of color correction technolo-gy is the distinction between photopic and scotopiclight. Photopic light has a spectral distributionmatching the sensitivity of the bright-light receptorsin the eye, the cones. The peak sensitivity of thecones is for green light with a wavelength of 555nanometers. The spectrum of scotopic light matchesthe sensitivity of the dim-light receptors in the eye,the rods; rod sensitivity is peaked in the blue-green,at a wavelength of 507 nanometers. Light metersare conventionally designed to measure light of dif-ferent wavelengths in proportion to photopic sensi-tivity, so the measured luminance corresponds mostclosely to the brightness that the eye would perceiveat relatively high ambient light levels.

There has been some research into color-cor-rected fixed roadway lighting. Janoff (1996) com-pared the effect on visibility of high-pressure sodi-um, metal halide, and a scotopically rich metalhalide lamp, all at 250 watts power in a cobra-head luminaire with a glass refractor. While thehigh-pressure sodium lamp was more efficient pro-ducing the most lumens per watt, the metal halidelamp and the scotopically rich metal halide lampprovided a higher level of visibility per lumen.

Several studies at interior lighting levels(Berman et al. 1993; Berman et al. 1994, Berman etal. 1996) have shown that visual acuity and contrastsensitivity are determined by pupil size, and thatpupil size and brightness perception are effected byrod activity. Collectively, these studies suggest thatwith lamps rich in scotopic spectral content, lessluminance (as measured with a photopic meter) isneeded for visual performance than with convention-al lamps, which have a higher yellow content. Thisidea was supported by Adrian (1997), who stated,“As the spectral sensitivity of the eye is shifting tothe blue with lower light levels, blue and blue-richpower distribution of the light appear brighter andachieve higher levels of visual performance.”

Janoff (1999) reviewed several research papers(including Janoff and Havard 1997, Rea 1990, andLewis 1997) that evaluated the effects of lampcolor in roadway and outdoor lighting on visualperformance. The review concluded that theeffect of spectral content depends on the level ofambient illumination and the nature of the visualtask. Spectral content is not important for fovealtasks, which involve objects located straightahead, where vision is sharpest, and that are in themesopic range (adaptation approximately .034 to3.4 cd/m2, which includes the range of night driv-ing). However, off-axis tasks at mesopic levelscan be affected by a lamp’s spectral content.Research by Lewis (1998) that was summarizedby Janoff indicates that driver reaction times areshorter at adaptation levels of 1 cd/m2 and belowwith scotopically rich metal halide lamps thanwith non-scotopically rich lamps such as high-pressure sodium Other studies (Boyce et al. 1998,Mehra 1998) showed that color-naming perform-

49

ance was better with metal halide lamps than withhigh-pressure sodium lamps.

Discomfort Glare. While increasing the relativescotopic content of a light source might possiblyimprove visibility, it may well do so only at thecost of increasing discomfort. Fugate and Fry(1956) investigated the role played in producingdiscomfort by the constriction of the pupil afterbrief exposures to light. They found that theamount of pupillary constriction at the borderlinebetween comfort and discomfort varied with thesize of the momentary stimulus, greater constric-tion resulting in greater discomfort. If blue-richlamps produce a smaller pupil size than otherlamps, then discomfort glare might be increased—as has in fact been found by Flannagan, Sivak, andTraube (1994)—even as visual performance isimproved. Karpen’s claim that yellow light is morediscomforting than blue is not easily explained byprior research and should be further investigated.

Sullivan and Flannagan (2001) recently testedcolor-correction technology by comparing the dis-comfort glare produced by neodymium, tungsten-halogen (TH), and blue-tinted replacement lamps.The results were consistent with the earlier find-ings of Flannagan et al. (1994): When all threeheadlamps were adjusted with neutral density fil-ters to have the same photopic illuminance, THheadlamps produced the least discomfort. Thisseems to support the notion that yellow light isless discomforting than blue (see discussion ofnight driving glasses below) and confirms the ear-lier study by Flannagan, Sivak, and Traube (1994).

Advantages and Disadvantages ofNeodymium Headlamps

Neodymium technology, if shown to be effec-tive, should be less costly than almost any othercountermeasure for controlling glare. However,the effect of any change in the rendition of colormust be tested with regard to the recognition ofobjects on the road that are critical to safe driving.With respect to headlighting, such objects includeyellow warning signs, orange construction signs,and various retroreflective hazard delineators andpavement markings. The Maryland Departmentof Transportation study, which found that

neodymium sunglasses with a notch filter between580 and 600 nm completely blocked out yellowLED warning signs and traffic signals, demon-strates the need for thorough testing before thistechnology is accepted. Even if there is someimprovement in the visibility of some objects withneodymium headlamps, such a gain will have to beweighed against any loss in visibility of otherobjects and against any increase in discomfort glare.

SummaryWhile lamps rich in scotopic spectral content

appear to help with some visual tasks under low-level light conditions, the effects on discomfortglare and fatigue are still in question. WhileKarpen suggests that yellow light is more discom-forting and causes more fatigue than light of otherwavelengths, most previous research suggests thatyellow light is actually less discomforting.

Given the lack of certainty in the research,color-corrected headlamps cannot be recommendedas an effective countermeasure for headlight glare atthis time. However, variation in the spectral contentof lighting is a countermeasure worthy of extendedinvestigation. While the issues under debate byresearchers in the field are extremely important,they are beyond the scope of this paper. Clearly,further research is warranted to determine the effectof neodymium (or other scotopically rich) head-lamps on visibility and glare in a night highwayenvironment. One thing is certain: If proven effec-tive, neodymium headlamps should be cost-effec-tive, because very little investment would berequired to install the devices.

Headlight HeightThe FMVSS 108 standard specifies that

motorized vehicles should have both headlampsmounted at the same height, one on each side of thevertical centerline, and not less than 22 in (55.9 cm)or more that 54 in (137.2 cm) above the road sur-face. Sivak et al. (1997) measured 15 of the best-selling cars and 15 of the best-selling light trucksand vans and reported a sales-weighted headlampheight of 24.4 in (.62 m) for cars and 32.7 in (.83m) for light trucks and vans, including SUVs.

50

Allen (1985) concluded that “there was nobasis in highway safety for allowing headlights tobe above the eye level of the driver of a passengervehicle.... All vehicles ... should be required tohave the headlight height and spacing within a fewinches of one another so that headlight aim andlight-output pattern can minimize headlight glarefor everyone.” Mortimer (1988) recommendedthat FMVSS 108 be amended to “limit the mount-ing heights of headlamps to within the range of 22to 30 in. on passenger cars, pickups, vans, trucks,and motor cycles.” Presumably, he would addSUVs to this list today.

Lamp height contributes to glare in the sameway that misaim does: Changing the headlampheight shifts the beam pattern. If the headlampheight is raised for a given beam pattern withoutreaiming the headlamp, a greater intensity will bedirected at greater elevations above the road. Oneconsequence is a modest increase in the illuminationreaching the eyes of drivers of oncoming vehicles,but the greatest effect is an increase in the illumina-tion reaching the mirrors of leading vehicles at theshort inter-vehicle distances of following or passing.

According to the American Association of StateHighway and Transportation Officials (AASHTO,1994), vehicle heights have been decreasing sincethe 1960s. For design purposes, driver eye heighthas now been established at 1.07 m (42 in) for carsand 2.4 m (94.5 in) for trucks. The high-mountedheadlights on SUVs and pickup trucks create visibil-ity hazards for drivers of cars, and the popularity ofSUVs has acted to raise the average vehicle head-lamp height. When a full-size SUV follows a small-er sedan at night, the SUV’s high-mounted lightsshine almost directly into the side and rear-view mir-rors of the leading vehicle. With headlamps goinghigher and with lower driver eye height in manyvehicles, it is not surprising that more people areexperiencing discomfort glare when driving at night.

Recognizing that, to control mirror glare, head-lamp height should not be above the height of sideand rearview mirrors, the SAE task force on head-lamp mounting height is considering the ramifica-tions of reducing the maximum mounting height ofheadlamps on highway vehicles (SAE 1996). Sinceside-view mirrors are generally lower than the inte-

rior rear-view mirror and are often even lower thandriver eye height, headlamp heights as low as 36 to40 in are being considered. However, there is noclear consensus of what the limit should be.

The two concerns delaying any specific recom-mendations by the task force are the reduction invisibility distance, which is expected to be small,and the effect on the legibility of signs, particularlyoverhead signs. The concern is not only about theloss in visibility, but also about what actions driversmight take to compensate for this loss. Still, themajority opinion of the task force is that there is noreason to be concerned about a loss of visibility fordrivers of SUVs. The report noted that “the mar-ginal detection distance loss for some vehicles isoffset by the greater good of reducing glare for thevast majority of passenger vehicle drivers.”Dissenting opinions favored other methods of glarecontrol, including changes in beam distribution,headlamp output, and mirror efficiency.

Research on Headlamp HeightMortimer (1974) reported the results of a com-

puter simulation of rear-view mirror glare with avehicle following at 100 ft that used headlampsmounted at various heights. With a 30-inch mount-ing height, the illumination from rear-view mirrorswas about the same (1.72 lux) as that encounteredfrom oncoming high-beam headlights at 600 ft. Therear-view mirror illumination from all properlyaimed headlights mounted higher than 30 inchesexceeded this tolerance value. With headlampsmounted at 42 inches, mirror illumination fromproperly aimed low beams exceeded the maximum3-lux criterion for long durations suggested byOlson and Sivak (1984). This level of glare could bereduced to below the criterion by reducing the interi-or mirror reflectivity by a factor of 10. The modelresults assumed that interior mirror reflectivity was0.85 and exterior reflectivity was 0.55.

Mortimer’s data make it clear that headlampheight is most important if headlamps are misaimedupward. A one-degree upward misaim of a vehiclefollowing at 100 ft increased the illuminationreceived from rear-view mirrors by a factor of aboutfour. At a height of 42 inches, one degree of upwardmisaim resulted in glare levels as high as 15 lux.

51

Whether this level of glare can be ameliorated by adecrease in mirror reflection is uncertain withoutfurther simulations, but greater amounts of misaimwill certainly create more significant glare problems,and these will be exacerbated by higher headlampmounting heights.

The effect of changes in headlamp height on thevisibility of retroreflective signs was studied bySivak et al. (1993). If headlamp height is reducedwhile keeping the height of the driver’s eyeunchanged, the separation between headlamps andeye will increase. Small increases in this “observa-tion angle” may result in reduced visibility distancesfor retroreflective devices, including signs and roadmarkings, that are more pronounced for large trucksthan lighter vehicles. Some trucks may exhibit poorbraking performance, and any safety problems withthese vehicles might be magnified by reductions invisibility.

Sivak calculated the relative brightness of TypeIII retroreflective signs mounted on the left shoulder,on the right shoulder, and in the center, as a functionof headlamp mounting height. The signs on theshoulders were mounted 4.3 m from the road edgeand at a height of 2.1 m, and the center sign wasmounted 6.1 m overhead. The study found that atfar distances (305 m, or 1000 ft) there was a modest(10%) decrease in detection distance for drivers oflarge trucks compared with drivers of cars. At nearviewing distances (152 m, or 500 ft), if the initialluminance of signs to cars was high, the effect of theincreased observation angle on legibility for driversof heavy trucks was not great, but if luminance waslow (6.8 cd/m2), there was a reduction in legibilityof 22%. While the study did not single out SUVs,the results for light trucks suggest that the effects oflowering SUV headlamp height will be minimal.

AdvantagesLowering headlamp height to below 40 inches

will reduce the glare experienced by the drivers ofleading vehicles from their side and rear-view mir-rors. The greatest reduction in glare will be achievedif headlamps are aimed properly and low-reflectance,preferably automatic, dimming mirrors are usedinside and outside the leading vehicle. While itwould be desirable to place a headlamp height

restriction on all vehicles, the prevalence of SUVsand small trucks suggest that the greatest advantageswould be achieved with this class of vehicles.

DisadvantagesThe reduction in glare from lowering headlight

height or aiming higher-mounted headlampsdownward must be considered in conjunction withthe drop in visibility. If headlamps are aimeddownward to correct for excessive height, the visi-bility of distant objects will be reduced. Lowerheadlight positions are not a problem in passengercars because in these vehicles the driver’s eyeheight is also low, close to the height of the head-lights. For best visibility of retroreflective objects,headlights on heavy trucks need to be high—clos-er to the height of the driver’s eyes—to maintainbrightness equivalent to that obtained by cars. Themagnitude of the loss in visibility from reducedheadlamp height and the consequences for trafficsafety have not been well researched. However,the research by Sivak et al. (1993) suggests thatthe loss may not be great, particularly for SUVs.

SummaryWhile it would be best if all vehicles could

have a similar height and spacing of headlamps(Allen 1985), this might not be practical, because itwould suggest that driver eye height should berestricted to a narrow range for both cars and trucks.In any event, fixing headlamp height and spacing isnot likely to happen, because the consumer is moreinterested in style, versatility, and functionality thanin glare reduction for other drivers.

Reducing the headlamp height on trucks, SUVs,and other large vehicles to a level below the height ofdrivers’ eyes in cars or car side-view mirrors is amore realistic and desirable goal. Because of theirrelatively higher speed and lane selection ability, carsgenerally find it easier to avoid being followed orpassed by large trucks than SUVs and light trucks.Given the prevalence of SUVs and light trucks andthe greater ability of a driver of a car to avoid pro-longed exposure to mirror glare from heavy trucks, itmakes sense to target SUVs and light trucks forimplementation of this countermeasure.

52

• Polarized Lighting• Night Driving Glasses• Glare Screens• Anti-glare Mirrors

Polarized LightingThe technical development of a polarized

headlight system consisting of polarizing filters onheadlamps and a viewer filter on the driver’s eyeswas begun by Edwin Land in the 1940s. Althoughpolarized lighting on automobiles is still not com-mercially available, extensive research was fund-ed by the Federal Highway Administration(FHWA) in the 1960s. In reporting their researchon polarized head lighting, Hemion, Hull, Cadena,and Dial (1971) describe the basic physics ofpolarized light. Ordinary light, including lightfrom an automobile headlight, consists of electro-magnetic waves that vibrate in all directions per-pendicular to the direction of the beam (Figure 5).When ordinary light passes through a polarizing

filter, all of the light waves are absorbed except forthose vibrating in a single plane (Figure 6) andthe light becomes (linearly) polarized. The polariz-ing axis of a filter is the orientation of the electricor magnetic field of the light waves that are able topass through the filter (for example, 45 degreeswith respect to vertical).

The polarizer — The polarizing filter, or polariz-er, can be made in different ways. The originalpolarizers of the 1940s were made from plasticsheets that were stretched to give their molecularstructure a preferred orientation. These polarizersdecreased visibility because they were inefficientand blocked out too much light. In addition, someof the original versions of the sheet polarizer wereprone to overheating and degradation. Since the1940s, more efficient ways of polarizing light havebeen found, some of which are summarized byDuncan (1996).

Scotch optical lighting film (SOLF) — SOLF isconstructed from a pair of plastic sheets, each witha sawtooth cross-section on one surface, so thatthe sawtooth surfaces of the sheets are mated

53

Figure 5. Ordinary Light produced by an automobile headlight.

CHAPTER 6

COUNTERMEASURES THAT REDUCEILLUMINATION REACHING THE

DRIVER’S EYE

Figure 6. Plane polarized light produced by plac-ing a polarizing filer in an ordinary light beam.

together to form grooves. Light polarized perpen-dicular to the grooves is transmitted and lightpolarized parallel to the grooves is reflected.While the original sheet polarizers absorbed lightnot matching the desired polarization axis, polariz-ers like SOLF redirect this light.

Liquid crystals — Liquid crystals with a helicalstructure are stacked into a filter so that the axis ofthe helix is perpendicular to the plane of thedevice. Liquid crystal filters are used to producecircularly polarized light, in which the plane ofpolarization rotates about an axis defined by thedirection of the beam. Looking in the beam direc-tion, the rotation can be either clockwise (left-handcircular polarization) or counterclockwise (right-hand circular polarization). Unpolarized light is acombination of right-hand and left-hand circularpolarized components; when it passes through theliquid crystal filter, the component that is circular-ly polarized consistent with the curl of the crystalhelix is reflected and the other component is trans-mitted. This type of filter will only affect light ofa certain wavelength, letting all other wavelengthspass through unaffected. Therefore, in order tocreate a polarizer that will work for multiple wave-lengths, different liquid crystal filters must bestacked together.

Corning Polarcor – Polarcor is a variation of theoriginal sheet polarizers. It is made of glass thatcontains elongated submicroscopic silver particlesthat are aligned along a common axis. The parti-

cles absorb light that is polarized along the parti-cles’ long axis and that has a particular wave-length. Absorption depends on particle size andshape—more elongated particles absorb longerwavelengths of light. This technology has severaladvantages, including a large acceptance angle forlight not hitting directly on axis and the capabilityto make a polarizer that operates on specific wave-lengths.

Duncan (1996) summarized polarizer tech-nologies (see Table 2, below) and rated them on ascale of 1 (poor) to 5 (excellent), according to thefollowing criteria:

• Manufacturing complexity — A measure direct-ly related to the cost of the product.

• Temperature sensitivity — The ability of thepolarizing material to withstand high tempera-tures.

• Efficiency — How well the polarizer transmitsdesired wavelengths of light in the desiredplane and blocks light of other wavelengths andother orientations. An inefficient polarizer willblock too much light and so decrease visibilityunless a high-intensity headlamp is used.

• Chromatic effects — The ability of the polarizerto be effective for light of all wavelengths.

• Angular effects —How much the degree ofpolarization changes when the angle of inci-dence is varied. Most filters produce maximumpolarization when light strikes them at normalincidence.

54

Manufacturing Temperature Chromatic AngularTechnology Complexity Sensitivity Efficiency Effects Effects Total

Polaroid sheeting 5 1* 2 5 5 18

3M SOLF 2 5 5 3 3 18

Liquid Crystals 2 5 5 3 3 18

Polacor 5 5 2 5 5 22

Table 2. Relative Merits of Various Polarizer Technologies (Duncan 1996)

* Duncan classifies the temperature sensitivity of the sheet polarizer as poor. This was the case whenthe idea was originally suggested in the 1920s, but by the time polarized lighting was tried in automo-biles the heat problem had been resolved. With this marked improvement in temperature sensitivity, thePolaroid sheeting would have a score comparable to that of Polarcor.

The polarized headlight system — A polarizedheadlight system can improve visibility for driversby decreasing the amount of light from oncomingheadlights. For this system to work, both vehiclesmust be equipped with the proper hardware. Inthe following description of the system, the benefi-ciary will be referred to as the “driver” and theopposing vehicle will be the source of the glare.

• A polarizer is placed over the opposing vehicleheadlights with its polarizing axis at 45 degreesfrom vertical.

• Another polarizing filter is placed in front ofthe driver’s eyes with its axis parallel to that ofthe polarizer in front of the headlights. Thispolarizing filter is sometimes referred to as ananalyzer.

• Both cars should have a polarizer and both drivers should have an analyzer.

Two types of analyzers were developed. Onetype, referred to as a visor, was a strip of sheetpolarizer that was hung from the sun visor in sucha way that its bottom edge intersected the line ofsight from the driver to oncoming vehicles. Whenthere was no visible opposing vehicle, the drivercould look under the visor. Another type of ana-lyzer took the form of either full- or half-glassspectacles. The half-glass type operated like clip-on sunglasses and could be used in the same wayas the visor. That is, drivers could look throughthe analyzer when a vehicle was approaching andthrough the bottom, unpolarized portion when the

analyzer was not needed. The half-glass can befitted over prescription glasses or over a pair ofneutral glasses with no optical function (see Figure8 below). The full-glass type, for use by passen-gers, is functionally identical to sunglasses.

If two cars are equipped in exactly the samemanner, then the orientations of the analyzers andpolarizers will be the same when the two cars arefacing in the same direction. However, when thetwo cars approach each other facing in oppositedirections, then the orientation of the analyzer andpolarizer for one car will be perpendicular to thosein the other car. This is illustrated in Figure 7.

The polarizing system reduces glare in the following way:

1. Light from the opposing headlamps of anapproaching car will pass through the polarizerin front of that car’s headlamps. As explainedabove, the orientation of the polarizer in frontof the opposing car’s headlamps is perpendicu-lar to the orientation of the analyzer in front ofthe driver’s eyes, since the cars are facing inopposite directions.

2. Therefore, the light coming from the opposingheadlamps will, upon passing through the polar-izer, become polarized in a direction perpendi-cular to the polarizing plane of the driver's analyzer.

55

Figure 7. Illustration of Polarized Lighting System

OPPOSINGHEADLIGHT

DRIVER’SHEADLIGHT

POLARIZER

POLARIZER

VISOR

PEDESTRIANDRIVER

3. Since the driver’s analyzer is oriented perpendi-cular to the plane of the polarized light comingfrom the opposing car’s headlamps, it will notallow this light to pass through and reach thedriver's eyes.

Note: Light from the driver's own headlightswill be reflected from objects such as pedestriansand road signs and return through the driver’s ana-lyzer, which is oriented to allow light polarized inthe same direction as that from the headlights,making the objects readily visible.

According to Schwab and Hemion (1971), thepolarized headlight system is the most promisingand most likely system to solve the night visibilityproblem. After reviewing studies that tried todetermine means by which night visibility couldbe improved, Schwab and Hemion (1971) con-cluded that:

1. “...the [polarized headlight] system is technical-ly and economically feasible in regard totoday's vehicle population;”

2. “the system would be advantageous in terms ofimproved visibility with less glare formotorists;”

3. “the results of the use of such a system wouldbe increased vehicular control, safety, and com-fort and probably improved traffic flow and uti-lization of highways at night.”

Dynamic Polarization is a modification describedby Duncan (1996). In this system, the headlightsare polarized as described above, but the analyzeris an electro-optical device that is activated when-ever incident polarized light is detected. Unlikestatic polarization systems, this system offersadvantages to those using it even when polariza-tion is not in universal use. Since the analyzerwould normally (in the absence of polarized light)pass all polarizations, objects would be just as vis-ible as at present without any increase in headlightoutput. Duncan points out that the technology thatmost contributes to making this approach possibleis the so-called “polarizer on demand”: “This is

one of several liquid crystal devices that are clearin one state, i.e., they pass all polarizations equallyand pass or block a particular polarization in theother state.”

Because it does not require any increase inheadlamp intensity, use of dynamic polarizationwould cause no increase in rear-view mirror glare.Of course, if the source of the glare were polarizedlighting, it could be filtered by the analyzer aswell. The system could also be designed to reducethe backscatter, which limits visibility in fog. Thissystem might also overcome some of the imple-mentation problems discussed below. However, itappears that there are technological and humanfactors issues which need to be addressed beforedynamic polarization can realistically be comparedwith other countermeasure alternatives.

ResearchThe feasibility of polarized headlighting was

extensively researched by the Federal HighwayAdministration in the 1970s. These studies hopedto answer three questions about nighttime visibili-ty and polarized headlights: 1) Do drivers usingconventional high beams and low beams have aproblem with visibility distance that a polarizedsystem would resolve? 2) How would visibility beaffected during a transition to polarized headlights,when not all vehicles have the polarized system?3) Would the benefits of polarized headlights out-weigh the problems associated with the system ina “real world” situation?

The studies of visibility distance when usingconventional headlights were summarized bySchwab and Hemion (1971). These studiesinvolved determining the distance ahead at whichthree different targets first became visible to sub-jects using four different headlighting systems:conventional low beam, conventional high beam,polarized high beam with analyzer, and polarizedhigh-intensity beam with analyzer. The test wasdone both with one opposing vehicle and with noopposing vehicles. When an opposing vehicle waspresent, both vehicles were equipped with thesame type of headlighting system.

56

The studies found that the visibility availableto the driver in meeting situations with conven-tional high and low beams was unsatisfactorybecause it did not provide enough time to performan evasive action if one were needed. In theopposed mode, use of polarization increased thevisibility distance compared to conventional head-lights, regardless of the target or of the intensity ofthe lamp used with the polarizer.

In order to address the question of visibilityduring the transition period, Hemion (1969a) con-ducted experiments to determine the driver’s abili-ty to see when vehicles equipped with polarizationsystems met vehicles not so equipped. Hemionmeasured disability veiling brightness and targetdetection distance for ten different cases, whichare summarized in Table 3. Each case was rankedin terms of glare and relative visibility, with 1being the best rank and 10 the worst.

Hemion observed veiling brightness levelsthat ranged from 0.0046 foot-lamberts, in case 9,to 4.9 foot-lamberts, in case 1A. In only one casewas the visibility reduced measurably from the

normal low beam-to-low beam meeting, namelywhen the glare car was on low beam and the sub-ject car was on polarized low beam with an ana-lyzer in use. Hemion surmised that this casewould be rare in practice because the subject driv-er would realize that the opposing car did not havepolarized lights and would stop looking throughthe analyzer. Visibility was improved in four ofthe nine cases, all involving a modified vehicle,compared to unmodified high beam-to-high beamor low beam-to-low beam meetings. Hemion con-cluded that, “It appears that improved visibility ofroadside obstacles will be achieved from thebeginning of the period of transition to polarizedlighting with greater and greater improvement asmore vehicles are converted.”

Finally, while planning a large-scale publictest of polarized headlights that was never per-formed, Hemion et al. (1971) conducted a small-scale test, consisting of approximately 120 driversand observers on a 40-mile test loop of two-lane,rural, unlighted roads. The results confirmed thatfor the “average” motorist, the benefits ofimproved vision and nighttime driving comfort

57

Mode Glare Car Subject Car Relative Relative Mean

Glare Visibility Detection Distance, feet

1A HiB* HiB 10 5 305

1B LoB LoB 8 7 255

2 HIP w/A HIP w/A 2 2 469

3 HiB HIP w/A 9 9 208

4 HIP w/A HiB w/A 3 3 442

5 LoB LoBP w/A 5 10 146

6 HIP w/A LoB w/A 4 8 231

7 LoB HIP wo/A 7 1 479

8 LoB HIP w/A 6 4 433

9 LoBP w/A LoB w/A 1 6 293

Table 3. Target Visibility (Hemion 1969a)

*HiB – High Beam, LoB – Low Beam, HI – High intensity, P – Polarized,w/A – with analyzer for driver, wo/A – no analyzer for driver

would outweigh the deficiencies presented by thepolarized headlight system.

AdvantagesThere are numerous advantages to a polarized

headlight system, all of which are just as applica-ble today as they were thirty years ago, when theresearch was performed.

Excellent illumination — While an increase inheadlamp intensity is necessary to offset the atten-uation of the analyzer, even higher intensities thanrequired are permissible because glare fromoncoming drivers is not a concern. Polarizedheadlamps could also be aimed further upward sothat they can intercept obstacles in the roadway ata greater distance than the current system of head-lights, which must be aimed below the horizontalto reduce glare.

No blind driving zone — During a meeting situa-tion, the driver often cannot see beyond the oppos-ing car’s headlights. This is referred to as a blinddriving zone and, according to Land (1948), driv-ers habitually drive into this zone on faith, hopingthere is nothing in the road. With the polarizedsystem, there is no blind driving zone.

Long visibility distances during a meeting situation — Schwab's review of reports foundthat most people overdrive their headlights atnight. This means that, in a meeting situation withanother vehicle, drivers often cannot see farenough to be able to stop for an obstacle in theirpath. Polarized headlights increase the visibilitydistance and therefore increase the detection dis-tance of such targets. (Hemion 1969a)

Insensitive to headlight aiming, vehicle loading,horizontal and vertical curvature — As justdescribed, the polarized headlight system dependson keeping the driver’s analyzer perpendicular tothe polarizer on the opposing vehicle's headlamps.However, during vehicular roll, the analyzer and theheadlamp polarizer will rotate away from being per-pendicular to each other and some light will be ableto leak through. A study by Adams (1971) indicat-ed that this leakage had an insignificant effect ondetection distances during meeting situations.

Proper use is easy to learn — In small-scalestudies of polarized headlights, subjects had noproblems learning how and when to use the ana-lyzer. To look through or past the analyzer, onlyminor head movements were necessary.

DisadvantagesThe disadvantages associated with polarized

headlights seem to be minor, and a solution to eachis presented below. The only major disadvantage ofthe system is the difficulty in implementing it.

Headlamp intensity must be increased —According to Duncan (1996), light intensity reach-ing the eye will decrease by 50%, even with a per-fect analyzer. This reduction also applies to ambi-ent light that is not polarized. The proposed strat-egy for dealing with this problem is to increaseheadlamp output by a factor of two.

Haze of approaching car is not visible — Haze,which warns of the approach of a car over a hill oraround a curve, is not visible if the analyzer isused. To continue to observe haze, the driver canrefrain from using the analyzer unless polarizedheadlights are in view.

Effect on distance perception of approachingcars — The effect of polarized light on a driver’sability to judge the distance to oncoming vehiclesis largely unknown. Hemion (1971) expected thatonce people gain experience with seeing approach-ing cars that are equipped with a polarized head-lamp system, they will adapt to judging speed anddistance with the system and the magnitude of thisdisadvantage will be reduced. All of the drivers inHemion’s (1971) study encountered an oncomingvehicle with polarized headlights, but none ofthem mentioned distance perception as a problem.Still the effect of a polarized system on driverjudgement in passing situations should be studied.

Effect on visibility of Traffic Control Devices —The effect of the analyzer on the visibility of traf-fic signals and hazard warning lights is largelyunknown and should be investigated.

58

Implementation — The main factor that hasrestricted adoption of polarized headlights is theabsence of an efficient method to implement thesystem. There are two principal concerns:

(1) How will the change to polarized headlights be motivated?

(2) What operational problems will arise during the transition period, and how will these affect public opinion?

These concerns are discussed in more detailbelow.

How would the change to polarized headlights bemotivated?

When used as a glare countermeasure, polar-ized headlights and analyzers constitute a coopera-tive system, with effective operation dependant onits use by all people. For this reason, the systemwill most likely have to be implemented by man-date and/or promoted by advertising. Advertisingmay be ineffective because, as Duncan (1996)observed, “it is easier to market a product or service to a public that perceives a need than toimplant the idea of need in the minds of the public.”

Another consequence of the cooperative natureof the polarized system is that there is no competi-tive incentive for a car company to be the first toinstall the system. If the system is introduced ononly one make of cars, it will only be effectivewhen it meets cars of that make.

What operational problems will arise during thetransition period and how will these affect publicopinion?

According to Hunt (1948), an additional con-cern with implementing the polarized headlightsystem is that there will not be a smooth transitionperiod from the current system. Obviously, not allcars can be equipped at once, so there may be alengthy period of mixed use of polarized and non-polarized headlights. During this transition period,some situations could arise that may hurt public

opinion of polarized lights. Drivers who paid forthe polarized headlight system will probably bedisappointed in the long delay in receiving fullbenefits from the system. Drivers who had notupgraded to the polarized system could createglare for polarized drivers by misusing highbeams, while protecting themselves against thepolarized beam by purchasing an analyzer. Driverswith polarized headlights could cause glare byusing high beams when meeting cars without thenew system. Public resentment against polarizedlights could build up during the transition.

SummaryThe polarized headlight system appears to be a

cost-efficient means of solving the glare problem.Not only does it eliminate glare from oncomingvehicles, but it provides a solution to the morecritical problem of limited sight distance at night.With one technological stroke, a quantum leap inhighway safety could be accomplished. Whetheror not the costs can be justified on the basis ofglare reduction, the benefits of improved vision fornight driving should be substantial. Still, the prob-lems associated with implementation need to beaddressed.

The major drawback to polarized headlights isthat installing them on a vehicle helps only theoncoming driver, not the person who paid for them(at least, until all cars are so equipped). How cana conversion be accomplished quickly and univer-sally? How can misuse of high beams during theconversion period be controlled? To answer thesequestions a large-scale experiment should be con-ducted similar to the one proposed by Hemion etal. (1971).

Night-Driving GlassesA distinction must be made between night-

driving glasses that are aimed at attenuating nightmyopia and those marketed specifically as road-way glare reducers. The former are untinted andinclude a small diopter lens correction (like that inprescription lenses) to counteract the intraocularlens’s natural inclination to return to its tonus, or

59

resting position, in the absence of accommodativestimuli (i.e., something to focus on). At night,many people with normal vision tend to becomemyopic and many myopes tend to become evenmore nearsighted (Leibowitz and Owens 1976).

Night-driving glasses for glare reduction arestructurally and functionally very similar to sun-glasses and, in fact, some models are merely aremarketing of what was originally a pair of sun-glasses. Like sunglasses, these glare-blockingglasses work by filtering light prior to its reachingthe driver’s eye. Many models use color filters toeliminate short-wavelength or “blue” light and sohave a yellow tint. Some are full-sized, some fitover prescription glasses, and others are smaller,filtering only a thin strip of light in the upper por-tion of the eye. A unique model developed in the1950s (Bryan 1962), called “Nite-Site,” used agreen, ring-shaped filter that was affixed to pre-scription lenses so that the observer looked thoughthe clear center. The filter was purported to“...cause a shadow effect to fall across the pupil toeliminate the oncoming glare of headlights....”Another lens, described by Adrian (1979), had aclear circular center large enough to transmit lightwith a glare angle of less than 5 degrees surround-ed by a shaded area that gradually increased intransmission according to the glare angle squared,so that the wearer could view glare from oncomingtraffic through the shaded area. In most cases, theshading was not uniform over the ring, but onlycovered the sector in which glaring headlampsusually occur.

Another type of night-driving glasses, shown in Figure 8, has been introduced byGlareblockers™. Their product is an ultraviolet-filtering polarizing plastic that attaches to theuser’s own eyeglasses, covering the top half of thelenses, and is adjustable for proper fit. The designis similar to the half-glass analyzer used in thestudies of polarized headlights reviewed above.With a half-glass analyzer, the user can turn filter-ing on and off with a slight tilt of the head, muchlike using bifocals. The product provides glarecontrol on demand, and, when glare control is notneeded, allows the user to normally look throughhis own glasses.

The basic idea behind all glare-reducing night-driving glasses is that a reduction in the amount oflight at the eye will reduce glare. This reductioncould be accomplished by using simple neutraldensity (n.d.) filters, which filter all wavelengthsequally and cause minimal color distortion. Why,then, are most night-driving glasses designed toselectively filter out the shorter electromagneticwavelengths, in the blue and near-ultraviolet spec-tral range? Below, the theory behind this selectivefiltering will be evaluated both from an empiricaland a physiological/perceptual perspective; first,the claims made by the distributors of thesedevices will be explored.

Benefits Claimed by Manufacturers —According to the manufacturers’ or marketers’claims, glare-blocking glasses solve problems asso-ciated with oncoming headlamps (“especially thenew halogens”) and with lack of contrast that caus-es disability glare. They are also said to improvegeneral nighttime vision and night driving safety,reduce glare effects associated with aging andcataract development, reduce eye fatigue, and helpdrivers keep their “eyes on the road.” These glassesare also “cool looking.” The “Blue Max” glasses“block uncomfortable blue light which causes hazyvision.” All these glare blockers are “good for peo-ple with sensitive eyes.” “Confusion is gone.Confidence is restored.” They are also said toimprove vision on roads with overhead lighting.

60

Figure 8. Half-glass analyzer permits normal orfiltered view.

Federal Trade Commission (FTC) RulingNumber 952 3041 — In 1997 the FTC ordered anighttime glare-reducing glasses direct marketer to“cease and desist from representing, in any man-ner, directly or by implication, that such productmakes night driving safe or safer; or such productimproves night vision.” The marketer, NationwideSyndications Inc., of Barrington, IL, reached a set-tlement wherein they were prohibited from usingthe “NightSafe” name and agreed to pay $125,000in consumer compensation. What the FTC essen-tially said was that Nationwide Syndications hadno evidence to back up its claims, not that theglasses did not work. It appears that this rulinghas not stopped other distributers from makingidentical claims for essentially the same product.

Research FindingsGlare Sources — Night-driving glasses are saidto attenuate glare from two sources. Misaimedand high-beam headlamps seem from the promo-tional literature to be the biggest perceived prob-lem. The introduction of high-intensity discharge(HID) lamps and the proliferation of sport utilityvehicles (SUVs) and other vehicles with high-mounted headlamps are seen as increasing theproblems associated with nighttime headlampglare (daytime running light glare is a differentissue entirely). The second glare source of con-cern is luminaires, otherwise known as overheadlamps or street lighting. The impact of overheadlighting on glare (and glare mitigation) has alreadybeen treated in detail in a previous section of thisreport. For this reason, and because overheadlighting is a secondary concern for users of night-driving glasses, the remainder of this section willconcentrate on headlamps.

Headlamps — Headlamp contributions to dis-ability and discomfort glare are treated in detail inother sections of this report, but it may be usefulto revisit the issue of color temperature. With therecent introduction of HID headlamps, drivers areencountering not only the yellow/white of halogenheadlamps but also a bluish-white from the highercolor temperature of the HIDs. The color temper-ature of xenon HID headlamps ranges from 4000to 4500 Kelvin (K), which is much closer to that

of sunlight (4500–5000 K) than are halogen lamps,which fall around 3200 K. Other headlamps looklike HID headlamps but are not; these are the so-called “ion blue” or “diamond blue” halogenbulbs, which are actually halogen bulbs with ablue coating that reduces overall light output.

Do the new bluish-white HID headlampsresult in increased glare problems, as suggested byglare-blocking glasses distributers? There is someevidence that they may. Flannagan, Gellatly,Luoma and Sivak (1992) found that dim HIDsproduce as much glare as bright halogens—morespecifically, that higher levels of light from halo-gen lamps produced no more discomfort thanlower levels from HID headlamps. This finding isconsistent with studies of the effects of wavelengthon discomfort glare discussed later in this section.The effects of the shorter-wavelength HIDs on dis-ability glare will also be discussed below.

Yellow Lenses and The Human Visual System —One possible reason for using yellow filters (bluelight blockers) is that the blue light may be scat-tered more within the ocular media than thelonger-wavelength yellow or red light. Direct evi-dence that longer-wavelength light is less suscepti-ble to absorption and scattering by small particlesis seen at every dusk and dawn, when the sunappears as an orange or deep red ball because ofthe longer path length of those colors through theatmosphere. As sunlight traverses the earth’satmosphere at the oblique angles found at sunsetand sunrise, airborne particles scatter the bluewavelengths, leaving the reds relatively unaffect-ed. The same phenomenon occurs with sound:Higher-pitched sounds die away at shorter dis-tances than lower-pitched (longer-wavelength)sounds when there are obstacles between thesource and the receiver, because the amount ofscattering from atmospheric particles increases asthe wavelength decreases. This phenomenon iswhy foghorns have such a low pitch. However, asBlackwell (1954), citing Fry (1953), stated, “Themajority of scattered light within the eyeball is dueto large particles which scatter nonselectively[with regard to wavelength].” That is, all colorsscatter in the eye to the same degree, so that—withregard to scattering in the optical media, at least—

61

there is no glare-reducing benefit to filtering outcertain wavelengths. Blackwell’s statement is sup-ported by Ucke (1973) who measured scatteredlight of different wavelengths directly at the retina.Smart (1970) compared the effect of blue, red,green and white light sources on disability glare;while he found some small and interesting differ-ences in the laboratory, he concluded that glare inthe real world could not be mitigated by manipu-lating lamp color.

A second theory used to justify the filteringout of blue light is based on chromatic aberration.Focusing of light in the eye depends on refraction,and refraction in the eye’s lens—as in most lens-es—depends on the wavelength of the light. As aresult, light of different colors is focused different-ly on the retina, a phenomenon known as chromat-ic aberration. Blue light is focused “near,” redlight is focused “far” and green light falls some-where in between. Because of chromatic aberra-tion, when red words are printed on a saturatedblue background a three-dimensional effect isseen, a phenomenon known as the chromatic deptheffect. Could chromatic aberration be a problemthat would be ameliorated by using yellow-tintednighttime driving glasses? Probably not. To beginwith, the macula lutea (literally, yellow spot)which covers the retina’s sensitive central visualarea, the fovea, absorbs much of the blue light,helping to alleviate this problem naturally.Furthermore, as pointed out by Blackwell (1954),acuity might be affected by chromatic aberration,but acuity is not as important in nighttime drivingas are target detection and time-to-contact estima-tions (estimations of distance and speed), neitherof which ought to be compromised by chromaticaberration.

If neither blue-light scattering nor chromaticaberration justifies color filtering, why not useneutral density filters, which would maintain colorbalance? The answer might lie in discomfortglare. Blackwell (1954) found that yellow-tintedlenses produce a lower subjective impression ofdiscomfort glare than do neutral density filters,even when both allow equal light transmission.Further evidence is provided by a study on head-lamp color by the Institute for Road Safety

Research (1976). Although it found no percep-tion-based reason to recommend yellow overwhite headlamps, the study concluded that “mini-mal side effects (discomfort glare) were experi-enced which might indicate a preference for yel-low light.” Perhaps the final word on this issuewas an in-depth study (Flannagan, Sivak, andTraube 1994), which found a U-shaped relation-ship between discomfort glare and wavelength,with long-wavelength (650 nm) and short-wave-length (480 nm) light producing the greatest dis-comfort and intermediate wavelengths (centered at577 nm) producing the least. In terms of colors,yellow was the most comfortable and red and bluewere the most uncomfortable, with blue beingeven worse than red.

A convincing argument could be made thatdiscomfort glare might indirectly reduce visibility.That is, if an observer is made uncomfortable by aglare source, his performance on visual tasksmight suffer, even though there is no direct dis-ability glare. This might result because of reducedalertness caused by fatigue or changes in driverfixations caused by glare avoidance. Theeuwesand Alferdinck (1996), for example, found thatsubjects were less willing to look at a light sourcewhen illumination was higher.

Two footnotes are worth reporting here. First,the human eye develops opacities as it ages inwhich the lens progressively yellows. Exposure toultraviolet radiation is considered to be partiallyresponsible for this “natural effect of aging.”Second, tinted glasses can have unintended conse-quences—as discovered by the MarylandDepartment of Transportantion in 1997, when theyfound that neodymium sunglasses with a notch fil-ter in the amber color range between 580 and 600nm completely blocked out yellow LED warningsigns and traffic signals. These kinds of glasseshave since been outlawed by a change in ANSIstandard Z80.3 for sunglasses.

Glare and Reduced Visibility — The idea ofreducing roadway glare by using special glasses isnot new; in fact, fifty years ago Lauer (1951) con-ducted an extensive study of the effect of color fil-ters on visual acuity in the presence of a glare

62

source. Lauer found no difference in performanceamong the 15 color filters he tested (which includ-ed n.d., purple, blue, green, yellow-orange andred); all resulted in significant loss of acuity underglare conditions. The study concluded, “Althoughsome colors reduced glare, they also reduced acu-ity to the same degree.”

A study of two optical filters used in glasses toreduce driving glare in the 1950s (Blackwell 1954)evaluated the effect on threshold contrast for one“pale yellow” filter (luminous transmittance of .87and one “amber” filter (luminous transmittance of.69). The study found that glare reduced targetdetection but so did the tested filters, and the useof filters in the presence of glare did not produce acompensatory result. In fact, it was more difficultto detect targets in the presence of glare whenwearing the glare blocking filters than when not.Translated to the highway, these laboratory find-ings predict that not only will the filters reducedetection distances, but the reduction will be great-est under conditions where detection distanceswould be short even without the glasses. That is,the filters would make a bad situation even worse.Detection distances would be reduced 10% to 20%with a pale yellow filter and 30% to 40% with theamber filter under twilight driving conditions,without headlamps from the observer’s vehicle.The losses would be somewhat less if the observ-er’s headlamps were activated.

DisadvantagesThere are two potential disadvantages to fully

tinted glare-blocking night-driving glasses. First,the user might have a sense of improved nighttimevision disproportionate to any actual gains in visi-bility. Such an exaggerated sense of improvedvision might tempt the user to “override” his visu-al capabilities, driving at speeds in excess of theactual stopping distances given by visibility andnatural driving conditions. Second, wheneverlight is blocked, the visibility of roadway obstaclesis reduced. The glasses lower the overall roadwayluminance seen by the observer (although contrastremains constant), and therefore reduce visibility.It might be argued that this reduction in visibilityis compensated for by the reduction in disabilityglare, but there is no evidence to support this.

AdvantagesThere are several ways that nighttime glare-

blocking glasses might improve driving safety.None of these possible advantages have the empir-ical evidence necessary to support the continueduse of these devices.

First, glare-blocking glasses reduce overallillumination, thereby reducing glare in the sameway as daytime sunglasses. Unfortunately, thelight emitted or reflected from all other roadwaycharacteristics is also reduced, reducing targetdetection over and above the reduction caused bythe glare source itself. Other eyeglasses, such asthe “Nite-Site,” that were designed to filter onlythe small portion of the driving scene that containsthe glare source, will still necessarily obscureobjects outside their target areas, because the glaresource is in constant relative motion and is contin-ually changing in retinal image size as it approach-es the observer’s vehicle.

Second, glare-blocking glasses might selec-tively eliminate those wavelengths of light whichcreate the biggest glare problem for humanobservers and improve visibility by either reducingchromatic aberration or decreasing intraocularscattering. Research shows, however, that chro-matic aberration is not a significant factor in night-time driving and that intraocular light scattering isnot wavelength dependent.

The third possible advantage is perhaps themost plausible. There is good empirical evidencethat discomfort glare is reduced by yellow-tintedglare blockers when an observer is faced with amainly white glare source such as a headlamp.Discomfort glare might have an indirect effect onobject detection, because an observer who isuncomfortable might perform worse on visualtasks. The study by Blackwell (1954) does notsupport this claim, but Blackwell’s data were col-lected in a laboratory setting, and drivers in aroadway environment might behave differently.Blackwell’s subjects, for example, were asked topeer through the glare source to detect the targets,whereas vehicle operators actually on the highwayare encouraged to “look off to the right and downat the edge line” as oncoming vehicle headlamps

63

approach. The issue of how discomfort glareaffects visual performance would benefit fromadditional research.

The introduction of the Glareblockers™product, which provides glare control on demand,brought another dimension to the design of night-driving glasses. This type of half-glass analyzercould provide a real advantage if it is used onlywhen the discomfort is so great that the choice isbetween looking to the side, so the road is seenonly with peripheral vision, and looking straightahead with foveal vision reduced by glasses.Whether drivers would restrict their use of a half-glass analyzer to such worst-case situations isunknown. The drivers most likely to benefit fromthis type of glasses are those with severe sensitivi-ty to glare; these drivers may be candidates insteadfor restricted night driving, as discussed elsewherein this report.

Summary On balance, reasoning based on solid physio-

logical and perceptual concepts and backed up byalmost 60 years of good empirical research yieldsno real support for the use of fully tinted glare-blocking glasses as a means of achieving safernighttime driving. As Lauer firmly stated back in1951, “...any media introduced between the eyeand a stimulus object or situation on the roadwayas a means of reducing glare is not to be recom-mended for night-time or any other conditionswhich lowers acuity when maximum visual effi-ciency is desired.” The half-glass analyzer offersthe possibility of improving safety if properlyused, but the research has not yet been done toevaluate whether this possibility can be realized.

Glare ScreensOn divided highways without independent

alignment or large medians, glare screens placedin the median of the roadway could be a cost-effective way to reduce glare from opposing traf-fic. Screens can improve safety in temporarywork zones, where lanes can be narrow and only aconcrete barrier separates opposing traffic. Under

these circumstances, blocking glare from oncom-ing headlights will dramatically increase the visi-bility of the concrete barrier and the delineation ofthe lane ahead.

The National Cooperative Highway ResearchProgram Report 66 (NCHRP-66 1979), a report ofthe Transportation Research Board, includesdescriptions of the different types of glare screensand the advantages and disadvantages of each.NCHRP-66 defines a glare screen as “. . . adevice placed between opposing streams of trafficto shield drivers’ eyes from the headlights ofoncoming vehicles. . .” (pg. 2). A glare screen canbe any type of object of a certain width and placedat a certain spacing that will prevent glare fromreaching drivers’ eyes. The object may be opaqueor have intermittent openings that allow a view ofthe opposing lanes while at the same time screen-ing out light at angles less than 20 degrees fromthe driver’s eye.

To be effective, glare screens should:

• Reduce a large portion of the glare• Be simple to install• Be resistant to vandalism and vehicle damage• Be repaired quickly and safely• Require minimal cleaning and painting• Accumulate a minimal amount of litter and

snow• Be wind resistant• Have a reasonable installation and mainte-

nance cost• Have a good appearance• Allow for emergency access to opposing

lanes

There are three types of glare screens, labeledType I, Type II, and Type III. A Type I glarescreen is a continuous screen that blocks lightfrom all angles. A Type II screen is a continuousscreen of an open material that is opaque to lightcoming from angles of zero to 20 degrees from thedriver’s eye and becomes increasingly transparentfor angles beyond 20 degrees. A Type III screen ismade up of individual elements that will blocklight coming from angles of zero to 20 degrees

64

from the driver’s eye while providing clear visibil-ity beyond 20 degrees. A plan and elevation viewillustrating the concept of each type of glare screenis shown in Figure 9.

Examples of Type I glare screens include earthmounds and concrete barriers. Earth mounds areused in areas with wide medians and excess cutmaterial. Usually in these areas grades and cross-sections can be modified to leave or build thisexcess earth in the median to block glare. Concreteglare screens are usually provided by simplyextending the height of a standard concrete barrierenough to block light from opposing vehicles.

Type II glare screens include expanded metalmesh, knit polyester fabric, and fencing.Expanded metal mesh is the most widely usedtype of screen. It is usually manufactured fromsteel or aluminum sheets. Parallel slits are cut intothese sheets and then the sheets are stretched untilthe slits form a diamond pattern. The metal

between the slits twists at an angle to form ascreen. Knit polyester fabric also works as a glarescreen by diffusing light rather than blocking it.Chain-link fences are effective Type II glarescreens when the patterns of the wires have a spacing less than one inch.

The most common kind of Type III glarescreen is manufactured by placing individually-supported paddles at intervals so that they willblock light from opposing headlights. An exampleof these paddles is the Glare Gaard, ellipticallyshaped 25-inch blades made out of thermoplasticpolyolefin (TPO) that attach via anchor bolts to aconcrete barrier (see Figure 10).

Other types of glare screens not included inany of the above categories include plants andguardrails. Plants are suitable glare screens oncurves in wider medians as part of a general land-scaping effort. Back-to-back guardrails will blockglare, but they may not be effective in blocking allof the light from an oncoming vehicle since theyare only 27 to 33 in high. Some characteristics ofthe different types of screens are given byNCHRP-66 and are shown in Table 4.

NCHRP-66 (1979) recommends that beforeinstalling a glare screen, “the manner in whichvarious types of screens reduce glare and affect

65

Figure 9. Plans and elevations of glare screentypes I, II, and III (NCHRP 1979).

Figure 10. Paddles at angles less than 20 degreesappear to present a solid screen from the driver’s

view. (Courtesy glaregaard.com)

both visual and physical access to opposing lanesshould be considered. . .” (p. 4). For example,using an opaque screen may prevent drivers frombecoming distracted by happenings in the oppos-ing roadway, whereas using a Type II or Type IIIscreen may provide a limited view of the opposinglanes, which many feel is necessary for lawenforcement and emergency services.

ResearchAs mentioned earlier, we seldom hear of a

traffic accident caused by glare. Therefore, it isdifficult to relate any change in accident rate to theinstallation or use of a countermeasure such as aglare screen. Compounding this problem is therandom nature of accidents and the complexity ofaccident causes.

Despite these problems, accident data in manystates have been collected after the installation of aglare screen. NCHRP-66 summarizes studies thatwere done in California, Indiana, Michigan, NewJersey, Ohio, Pennsylvania and England. TheNew Jersey, Ohio, and Pennsylvania studiesshowed reductions in nighttime accidents in sec-tions where glare screens were added, butNCHRP-66 made no conclusive statements.

Advantages • Effectively reduce glare from oncoming

vehicles

• Simple to install, minimal maintenance, andquick and safe to repair

• Installation can be limited to specific problemareas such as on horizontal curves.

• Reasonable cost

Disadvantages• If the median is narrow, use of a glare screen

requires some type of barrier on which the glarescreen can be installed.

• Unless there is a sufficiently wide shoulder, alane of traffic must usually be closed for instal-lation, repair, or maintenance.

• The screen is effective only where the user'svehicle, the intermediate median and theapproaching vehicles are on the same or near-level plane. Measurements made with an exper-imental glare screen installation on theSchuylkill Expressway (now I -76), nearConshohocken, PA, showed that the glare fromcars passing one vertical curve would haverequired a screen 40 ft tall to block the glarefrom a car at the next crest (Schwab 2000).Glare screens do not work well with rollingalignments, which have a significant amount ofvertical curvature.

SummarySince glare screens have no benefit other than

reducing glare, their application is limited to high-traffic locations where glare is particularly disturb-

66

Characteristic Type I Type II Type III

Prevent gawking accidents yes no no

Prevent pedestrian crossings yes yes no

Prevent slush & other objects from being yes yes nothrown into opposing lane

Permit police surveillance of opposing lanes no yes yes

Permit access to opposing lanes by no no yesemergency personnel

Permit scenic viewing no yes yes

Table 4. Characteristics of Different Types of Glare Screens. (NCHRP 1979)

ing and continual. Unlike wide medians and inde-pendent alignment, glare screens will not reducethe seriousness or the frequency of accidents,except for those accidents directly resulting fromglare itself. However, at locations where glare is aserious or habitual problem (as is often the case intemporary workzones), glare screens provide aninexpensive solution that may be justified on thebasis of expected glare-related accidents alone.

Anti-Glare MirrorsThe purpose of anti-glare mirrors is to achieve

a balance between rearward visibility and protec-tion from glare. Anti-glare mirrors are available ineither of two general designs: the dual-setting,prism-type mirror and the variable-reflectanceelectronic mirror. Two other designs using com-pletely different technology have been patentedand are under development and evaluation; thesewill be discussed briefly at the end of this section.

Prism MirrorsPrism-type rear-view mirrors have been stan-

dard equipment on U.S. cars for more than thirtyyears, and virtually every driver is familiar withtheir use. Prism mirrors change reflectance whenthe driver manipulates a lever mounted on the mir-ror. The reflectance at the anti-glare setting isonly 4%. The two potential dangers with this typeof mirror are that drivers may neglect to changethe mirror back from the anti-glare setting and thatthe anti-glare reflectance may be too low. The rea-son that prism mirrors reflect only 4% when tiltedis that this is the reflectance of a plain sheet ofglass or plastic in front of a dark surface. Theprism mirror has an outside cover glass with amovable mirror behind it. In the normal position,the highly-reflective mirror is parallel to the coverglass and transmits its reflection of the view fromthe rear of the vehicle—a relatively bright scene.In the anti-glare position, the mirror surface isangled so it picks up the image of the inside of thevehicle roof, which is a dark surface; the rear-viewreflection the driver sees comes only from thecover glass, which has 4% reflectivity. It wouldbe very difficult and much more expensive to get a

higher reflectivity from this type of rear-view mir-ror design.

Electrochromic MirrorsElectrochromic mirrors overcome the disad-

vantages of prism mirrors by providing continuouslevels of reflectivity and automatic control. Thesemirrors automatically darken to reduce glare fromthe headlamps of vehicles approaching from therear. The brighter the glare, the darker the mirrorsbecome, without the driver having to take any pos-itive action. These mirrors are often optionalequipment, at a cost from $70 to $100, and arecurrently installed in very few vehicles; however,the market is rapidly expanding. Two manufactur-ers (Gentex and Donnelly) now manufacture bothplanar and convex driver-side and passenger-sideelectrochromic mirrors. The number of vehiclesfor which electrochromic mirrors are availablecontinues to grow.

Photochromic MirrorA new technology, proposed by a company

named Quantics, offers a method of blocking onlyintense light rays, allowing dim rays to go throughunattenuated. This differs from electrochromicmirrors, which reduce the brightness of the entireimage. If successful, the new photochromic tech-nology would allow rear-view mirrors to remainclear while blocking headlight glare.

A photochromic device includes an internalfocal plane onto which a real image is projected. Alight-sensitive “photochromic” layer that darkens inthe presence of intense light and returns to clearwhen the light dims is inserted into the focal plane.Bright objects in the field of view generate a patternof bright spots in the focal plane; in response, a pat-tern of dark spots appears on the photochromicmaterial. Bright rays are self-attenuated becausethey go through the same dark spots that they cre-ate, but dim rays do not darken the photochromiclayer and are not affected. The result is a clear fieldof view with only bright objects attenuated.Variations of the technology have been proposed toimprove response time (to milliseconds or less) andto make the unit more compact by utilizing prismsor mirrors to fold the optical path.

67

Quantics has been funded by a series of smallbusiness grants from NASA, the Department ofTransportation and the National Research Council.The company has secured patent rights for thistechnology and has built a proof-of-concept proto-type that is available for demonstration.1

Neodymium Rear-view MirrorAnother new technology (Karpen 1998) would

construct the rear-view mirror with glass contain-ing neodymium oxide, a rare earth compound thatfilters out yellow light (this technology also under-lies the color-corrected headlamps discussed earli-er in this report). Karpen claims that the elimina-tion of excessive yellow light lessens eyestrain andreduces discomfort from conventional headlightsreflected in the rearview mirror. Contrary to thefindings of Flannagan et al. (1994) discussedabove, Karpen claims that yellow light is thesource of most visual discomfort to a driver.

Pure neodymium oxide has a very narrowabsorption band, but using the compound as adopant in mirror glass results in a wider spectralregion of absorption. Karpen cites a number ofindependent research studies that demonstrate theabsorption by neodymium oxide glass of yellowlight between 568 to 590 nanometers, and docu-ment the effects of this absorption: most colorsbecome more saturated and there are a number ofcolor shifts; for example, orange is shifted towardthe red. A study of low-vision patients foundmore-accurate color rendering and an improve-ment in visual acuity and contrast, as well as areduction of eye fatigue.

In a neodymium mirror, the light from follow-ing vehicles would pass through the neodymiumglass to the silvered back of the mirror and bereflected back to the driver with a unique spectrumthat Karpen claims will promote visual acuity indarkness by emphasizing the contrast-producingred and green light, and, at the same time, reducethe discomfort produced by yellow light.

ResearchOlson and Sivak (1984) conducted studies of

disability glare and discomfort glare, as well as thetransient effects of mirror glare, with the interiormirror in the high-reflectance (not anti-glare) posi-tion. Glare levels under 1 lux, thought to representlow beams at 90 m or high beams at 300 m (basedon data from Adler and Lunenfeld 1973), had aminor effect on visibility, but disability increasedrapidly when illumination was increased above 1lux. Even with the interior mirror set in its anti-glare position, the glare level could be greater than1 lux under some conditions; disability glareeffects could only be avoided by setting the exteri-or mirror to not reflect directly into the driver’seyes. The effects of glare persisted for 1 to 2 kmafter the following car was removed if the driverfailed to use the anti-glare setting, but glare effectsdid not noticeably persist if the anti-glare settingwas used.

The study also sought to relate levels of dis-comfort to levels of disability and to determine theamount of glare that motivated drivers to use theanti-glare position on their mirrors. Two levels ofglare duration were used: 10 seconds and 3 min-utes. Discomfort was rated using the 9-point scalediscussed in Chapter 2. The conditions represent-ed worst-case situations in that both mirrors wereadjusted to reflect the illuminance from the fol-lowing car into the driver’s eyes in a normal driv-ing position. The experiment was conducted on adark, two-lane rural road with both young andolder drivers. The results showed that discomfortwas one scale unit more uncomfortable when driv-ers were exposed to glare for 3 minutes than whenthe exposure was only 10 seconds. Glare wasrated slightly more disturbing than 5 (just admissi-ble) when illuminance was as little as 6 lux forshort durations and 3 lux for long durations; theAdler and Lunenfeld formulas show that these levels of glare approximate those from a low-beam system on a vehicle following at 30 m.Surprisingly, it was not until the level of discom-

68

1For information contact Quantics at 3980 Del Mar Meadows, San Diego, CA 92130, or [email protected].

fort exceeded 3 (disturbing) that all driversswitched their mirrors to the anti-glare setting; atthat level, disability glare would already have sig-nificantly degraded visibility. A subsequent studyby Sivak et al. (1997), which sampled only driversunder age 43, showed similar results for theeffects of glare duration, but the study also foundlarge differences among the drivers.

Given that their studies showed the effects ofdisability, transient, and discomfort glare onlyunder worst-case conditions, Olson and Sivak con-cluded that “if future lighting systems are mademore powerful in an effort to improve night visi-bility, the problem [of mirror glare] will worsen.”This is obviously what has occurred, as HID andhigh-performance halogen headlamps have led tomore complaints from the public. Olson andSivak suggested that, to encourage greater use ofanti-glare mirrors, the reflectance of the anti-glaresetting could be increased so that the setting couldbe used continuously. Alternately, they suggestedlowering the reflectivity of the normal setting toprovide some minimum protection against glare atall times. The automatic variable-reflectance mir-ror appears to satisfy this suggestion.

Misaim can greatly increase glare fromrearview mirrors, as was reported by Miller et al.(1974), who found that reflected glare from themisaimed headlights of following vehicles exceedsthe just-tolerable glare from oncoming headlights.Properly aimed low-beam headlights produced lessmirror glare at intercar spacings as small as 50 ftthan oncoming high-beam headlights at 600 ft, butmisaimed low beams can exceed this glare levelwith intercar spacings as close as 150 ft. Thestudy also found that glare increased by a greateramount when low beams were misaimed upwardthan when high beams were similarly misaimed.Low-beam headlamps misaimed upward by onlyone degree increased the illumination in therearview mirror by a factor of 8 and produce moreglare than encountered from oncoming high-beamheadlights at 600 ft with intercar distances as greatas 400 ft. To counteract the negative effects of alow beam misaim of only one degree, normal mir-ror reflectivity would have to be reduced to 10%,but even a mirror with less reflectivity than 10%

might not counteract the combined effects ofincreased mounting heights, greater misaim, andhigher intensity lamps.

A survey of 424 residents of Ann Arbor,Michigan (Flannagan and Sivak 1990) found thatthe most common way of using prism mirrors wasto switch back and forth when bothered by glare.Twenty percent of the respondents said they neverused the anti-glare setting and 16% left the mirrorpermanently in the anti-glare position. In terms ofthe severity of glare, oncoming headlights andinside rear-view mirrors received equal weight,whereas left outside mirrors were rated lower andright outside mirrors lower still. Both right out-side mirrors and inside mirrors in anti-glare modewere rated close to the “no-problem” end of therating scale. However, the effectiveness of prismmirrors is reduced by the substantial number ofdrivers failing to use them because they either for-get or consider it too much trouble to switch backand forth. The results also indicated that, whilethere appears to be almost no problem with glarewhen the mirror is in the anti-glare mode, the anti-glare reflectivity level of only 4% may provideinadequate visibility to the rear.

These results suggest that there may be atradeoff in controlling discomfort glare and main-taining rearward and forward visibility: Lowerreflectivity reduces discomfort and increases for-ward visibility, but decreases rear visibility. Thereis conflicting evidence concerning whetherimproved rear visibility is actually needed. Driversindicate that a major reason for not using the anti-glare setting is poor rear vision, but, unless thedriver is backing up, the only important things tosee in the rear-view mirror are overtaking vehicles(everything else is receding) and since all vehiclesshould have headlamps on them, they should bereadily visible with a 4% reflective mirror. If thedriver is backing up, he should not be relying onhis rear-view mirror but instead should turn hishead and look back. Flanagan and Sivak (1994)suggest some of the things that drivers may needor want to see to the rear, including empty pave-ment, road markings, and side panels of vehicles,but there is no evidence to support these sugges-tions. If what drivers want to see in the rear is the

69

behavior of their children, safety might be betterpromoted with less, not more, rearward vision. Ifrear visibility is necessary either because driversneed to see or want to see more than the head-lamps of overtaking vehicles, the alternative is theuse of the more costly variable-reflectance mirrors.

A laboratory study by Flannagan, Sivak, andGallatly (1991) evaluated rearward visibility (asmeasured by a visual acuity task), discomfortglare, and the driver’s awareness of changes inability to see rearward while using a variable elec-trochromic mirror. This study suggested that thetradeoff between forward visibility, rearward visi-bility, and discomfort might be found as a linearfunction of the logarithm of reflectivity. The con-tinuous control of reflectivity allowed by electron-ic mirrors results in continuous perceptual changesthat could allow some control of the tradeoffsrequired in night driving; for example, a gain invisibility can be achieved by accepting a corre-sponding loss in discomfort. It is interesting thatthe subjects in this study consistently underesti-mated their ability to see—drivers using these mir-rors apparently do not believe that they see as wellas they actually do. If this result is accurate, itmight undermine the potential success of thisproduct.

While the study suggested the possibility ofoptimizing the variable reflectivity of the mirrorwith respect to the tradeoff between discomfortglare and visibility, the research did not go so faras to develop an optimization algorithm. Towardthis end, Flanagan and Sivak (1994) developed acomputer model to evaluate the effects of mirrorreflectivity in much the same way as Bhise et al.(1977, 1984) developed CHESS for evaluating theperformance of low-beam headlighting patterns(see Chapter 5). The model produced a figure ofmerit that considered forward visibility, rear visi-bility, and discomfort glare. The figure of meritalso reflected the percentage of night driving inwhich both visibility and discomfort criteria weremet. As with CHESS, the results of the mirrorreflectivity model depend on how the criterion lev-els for satisfactory performance are set, and also,as with CHESS, there is room for disagreement asto what drivers must see, particularly toward the

rear. The new study did not offer any new opti-mization algorithms, but did suggest that the vari-able-reflectance mirrors in use today do a good jobof satisfying the tradeoff between visibility anddiscomfort, and an even better job when used inboth the center rear-view and left side positions.

There has not been any independent fieldresearch of either photochromic or neodymiummirrors in nighttime driving, so these technologiescannot be formally evaluated and their advantagesand disadvantages cannot be assessed. While theconcepts appear to be promising, controlled fieldtesting is necessary to show how these deviceswork in various situations and, equally important,how drivers react to them.

Advantages• If used, the prism anti-glare mirror should elim-

inate most problems from mirror glare, includ-ing glare from the high-mounted headlamps onSUVs and the brighter HID headlamps.

• Automatic variable-reflectance mirrors automat-ically regulate their use as a function of thepresence of glare and eliminate the need fordrivers to remember to use them.

Disadvantages• The low level of retroreflectivity with the anti-

glare position on prism mirrors results in poorrear visibility. Because of this, and the need tomanually change the mirror position, these mir-rors are not used by many drivers.

• Automatic variable-reflectance mirrors mayresult in drivers losing confidence in their visu-al performance and making inappropriate deci-sions while driving.

Advantages and Disadvantages of Photochromic Mirrors

While it is difficult to assess the advantagesand disadvantages of a product not yet on the mar-ket, it is clear that the potential advantage of thistechnology is reduced glare without loss in rearvisibility. Given the lack of consensus on theimportance of rear visibility, the disadvantage is

70

likely to be the tradeoff of whether or not the gainin rear visibility is worth the added cost.

Advantages and Disadvantages of Neodymium Mirrors

Neodymium mirrors are less costly than elec-trochromic and photochromic mirrors. The spec-tral effects of neodymium oxide in lighting andglass are well documented, but the practical effectof any change in the rendition of color must betested with regard to the recognition of highwayobjects critical to safe driving. For rear-view mir-rors such objects might only include flashinglights on maintenance vehicles and snowplows.Given the discovery by the Maryland Departmentof Transportation that a type of neodymium sun-glasses completely blocked out yellow LED warn-ing signs and traffic signals, it is clear that propertesting must be completed before this technologyis accepted.

SummaryAnti-glare mirrors are an effective way to deal

with most of the mirror glare from following orpassing vehicles. However, their benefit is notfully realized because many drivers fail to usethem or don’t use them properly. Unfortunately,drivers do not recognize that their vision may beimpaired by glare before they have any sensationof discomfort, and many drivers who do use anti-glare mirrors do not do so until experiencing alevel of glare that results in disability. This behav-ior argues for the adoption of variable-reflectancemirror technology.

While the automatic variable-reflectance mir-ror is an effective solution to the problem of mir-ror glare, until these devices are available on allmodels in both the inside and left outside positionsdrivers should be encouraged to use their prismaticmirrors. In addition, drivers without a variable-reflectance mirror on the external driver’s side

should set this mirror so that the headlights of fol-lowing vehicles do not reflect directly into theireyes. This is not an ideal solution, since much ofthe information otherwise provided by the exteriormirror is lost.

Since there is no conflict between maximiza-tion of forward visibility and glare reduction (bothgoals are achieved if the mirror is entirelyremoved), rear visibility is the sole reason forhaving a mirror. More attention should be givento determining the minimum reflectivity required;if this were known, the parameters of the variable-reflectance mirror could more easily be optimized.

Photochromic and neodymium mirrors arepromising technologies requiring further researchto demonstrate their function and safety as anti-glare devices. Cost appears to be the critical fac-tor to the viability of the photochromic mirror,although there is also a need to demonstrate relia-bility and effectiveness. For the neodymium mir-ror, whose cost should be minimal, the critical fac-tor is the effect of filtering yellow light on the visi-bility of emergency and maintenance vehicles.

The anti-glare mirrors described in this reportmay not be adequate to control the combined effectsof greater illumination from HID and high-intensityhalogen lamps, misaim, and the higher mountingheights of headlamps on SUVs. Additional researchwould help to define how anti-glare mirrors couldcontrol these variables, but it seems likely that themargin of safety that these devices provide wouldalways be violated under certain circumstances.This limitation is particularly serious with followingvehicles that create a glare situation for sustainedperiods of time. Therefore, while the combinationof inside and left outside automatic variable-reflectance mirrors is a worthwhile addition to saferdriving, other countermeasures, such as improvedheadlamp aiming and lower mounting height, stillneed to be pursued.

71

• Increased Median Width• Independent Alignment

MEDIAN WIDTHThe American Association of State Highway

and Transportation Officials (AASHTO) definesthe median as that portion of a divided highwayseparating the traveled way for traffic in opposingdirections (AASHTO 1994). AASHTO alsodefines the principal functions of medians: “toseparate opposing traffic, provide a recovery areafor out-of-control vehicles, provide a stopping areain case of emergencies, allow space for speedchanges and storage of left-turning and U-turningvehicles, minimize headlight glare, and providewidth for future lanes” (AASHTO, 1994).Medians can be depressed, raised, or flush with thetraveled way surface, and they should be highlyvisible during both day and night. The width ofthe median is defined as the distance between thethrough lane edges, including the left shoulders.

Soaring traffic volumes after World War II cre-ated a need for the immediate design of manyhighway facilities. In response to the urgent need,designers often relied on rules of thumb in theirapproach to median design, some of which haveproven to be incorrect. According to Hutchinsonet al. (1963), “these experiences have greatly stim-ulated research on median performance and factorsof influence in median design.” Among the “fac-tors of influence” are benefits such as safety, com-fort, and convenience. At the time of the report,the extent to which medians could provide thesebenefits was “more a matter of opinion than of

record,” and so it was difficult to use safety orcomfort to justify any cost increases involved invarying the median width. This left the choice ofbasic median width an administrative decision, tobe backed by engineering judgment.

Today, it is known that “reduced frequency ofcrossover accidents and [the] reduction of head-light glare are safety features associated with awide median” (AASHTO 1994). Therefore,AASHTO recommends that, in general, a medianshould be “as wide as practical.” Medians usuallyrange from a minimum of 1.2 m to 24 m or more.The AASHTO report states that medians that are12 m wide or wider provide the driver with a“desired ease and freedom of operation. . .”because there is a “. . . sense of physical and psy-chological separation from opposing traffic”(AASHTO 1994).

Increasing the median width decreases theeffects of glare from opposing headlights byincreasing lateral separation between vehicles. At any distance between vehicles on the road,increasing lateral separation will increase the anglebetween the driver’s line of sight and the head-lights of an opposing vehicle. In addition, increas-ing lateral separation effectively shifts the beampattern of the opposing vehicle, resulting in lowerintensities of light directed at oncoming drivers.The combined effect is a reduction of both disabil-ity and discomfort glare.

Figure 11 was prepared by Powers andSolomon (1965) to show the relationship betweenlateral separation, longitudinal (on-road) distancebetween vehicles, and veiling luminance. The cal-culations are based on the work of Fry (1954).

73

CHAPTER 7

COUNTERMEASURES THAT INCREASETHE GLARE ANGLE

The figure shows that at any given longitudinal dis-tance between an approaching car and a driver'seye, increasing the lateral separation will decreasethe veiling luminance. As vehicles approach eachother for a given lateral separation, the veiling lumi-nance either uniformly decreases (for lateral separa-tion of 32 ft or more) or first increases and thenabruptly falls (for lateral separation less than 20 ft).

ResearchMedian Width vs. Headlight Glare

In a series of experiments, Powers and Solomon(1965) measured the effects of glare as a function ofmedian width. Each of the three experiments simu-lated a meeting of a single vehicle and a singleopposing glare vehicle on a constant-grade, tangentsection of highway where both directions of travelwere at the same elevation. The objective was todetermine how changing the median width, and thusthe glare experienced by the driver, would affect thedetection of a target that was placed ahead of thesubject’s vehicle. The experiments differed withrespect to the positioning of the subject vehicle, glarevehicle, and target. Table 5 summarizes the parame-ters of each experiment and the results that werefound. Sample sizes were small (fewer than sevensubjects in each study), and therefore, according toPowers and Solomon, the experiments do not give“super-reliant” data; nevertheless, they do give someinsight. As expected, the effects of glare on targetvisibility decreased with increased lateral separationbetween the glare car and the opposing vehicle.

Median Width vs. Accident RatesAs mentioned above, the extent to which fac-

tors such as safety, comfort, and convenience areaffected by median width is more a matter of opin-ion than of record. Early studies by Hurd (1957),Telford and Israel (1953), Crosby (1960), andBillion (1962) attempted to correlate accident ratesto median width by studying accident recordsalong different sections of roadways, but theywere not able to establish a definite relationshipbetween these two variables. It was apparent,however, that accidents where vehicles crossed themedian and collided head-on with a vehicle travel-ing in the opposite direction were less frequent

where there were wider medians. As a result ofthis observation, the use of wider medians becamecommonplace (Garner and Deen 1973).

A problem with the early studies was that theydid not recognize and control for several importantvariables that also affect accident rates, such aspavement width, shoulder width, grades, curves,sign locations, and access control on the roadwaysection. A subsequent study by Garner and Deen(1973) controlled for these variables and soobtained more conclusive results.

In the Garner and Deen study, 420 miles ofrural, four-lane, fully controlled-access road sec-tions with medians ranging from 20 ft to 60 ftwere studied using an accident database thatincluded 2,448 accidents recorded between 1965and 1968. The results indeed showed that widermedians are safer medians. Figure 12 shows theaccident rate found by the study as a function ofmedian width.

74

Figure 11. Veiling brightness vs.lateral separation (Powers and Solomon 1965).

10

7

0.1

20

30

50

1070.01

0.0010 500 1,000 1,500

LATERALSEPARATION, FT.

DISTANCE - SUBJECT TO GLARE CAR, FT.

VE

ILIN

G B

RIG

HT

NE

SS

FO

OT

LA

MB

ER

TS

The Garner and Deen results were confirmedby Knuiman, Council, and Reinfurt (1993), in astudy that examined the effect of median width onthe frequency and severity of accidents in homo-geneous highway sections. Data for this studywere obtained from the Highway SafetyInformation System (HSIS) for the states of Utahand Illinois. A total of 3,055 miles of highwaywhere there had been 93,250 accidents between1987– 1990 was used for analysis. The medianwidths along roadways in the study ranged fromzero (no median) to 110 ft. Overall, the studyfound that accident rates do decrease with increas-ing median width.

Independent AlignmentIndependently aligned roadways are those with

horizontal and vertical alignments developed inde-pendently to suit location and design requirements(Peet and Neuzil, 1972). Independent alignment

75

General Set-up of Experiment

The opposing glare car and the targetwere stationary; the subject drovetoward the target and indicated whenhe could detect it. Subjects also indi-cated when they experienced discom-fort glare.

The target and the test subject werestationary and the glare car movedtoward the subject. The subject report-ed at what point he could and couldnot see the target.

The target and the subject car were stationary and the glare car movedtoward the subject. As the glare carapproached the subject, the subjectvaried the brightness of the target sothat it remained at the threshold of visibility.

Table 5. Summary of Studies by Powers and Solomon (1965)

Results

The distance at which the target was detectedincreased with lateral separation andapproached those of a no-glare condition at alateral separation of approximately 80 ft. Thedistance from the opposing car at which thesubjects reported discomfort glare generallyincreased as lateral separation was decreased.

For some runs with small lateral separations,the target was not visible until after the glarecar passed the subject car. For some largerlateral separations, the target was visible dur-ing the entire run.

At a given distance from the subject car to theglare car, the brightness necessary to maintainthreshold visibility decreased as the lateralseparation increased.

Study 1

Study 2

Study 3

Figure 12. Total accident rate versus medianwidth (Garner and Deen 1973).

places each roadway at a different elevation, with avariable amount of separation between the differentdirections of travel, in contrast to a narrow-mediandesign, where each roadway is at or near the sameelevation and a constant distance apart. With inde-pendent alignment, the driver on one roadwaymight not even see an approaching vehicle on theother roadway—which, of course, eliminates glareentirely. Since independent alignment is used whenthere is a large median width, it provides all of theadvantages of a wide median and more. Peet andNeuzil (1972) compared narrow median design withindependent alignment and listed several advantagesfor independent alignment, which are shown below.

AdvantagesIn addition to decreasing the effects of both

discomfort and disability glare, wide medians pro-vide numerous other benefits that increase thereturn on investment:

Accident Reduction. Wide medians have thepotential to decrease both the frequency and sever-ity of accidents.

Safety Zone. Wide medians provide greater stop-ping or recovery abilities and space for vehiclesthat run off of the left edge of the pavement.

Driver Comfort. Psychological discomfort andstress from driving are greatly reduced when awide median is available.

Visibility. Increasing the median width may alsoimprove visibility. Less light from opposing head-lights will fall on the pavement, thereby decreasingthe background brightness for some objects on theroad. This will increase the contrast between pedes-trians and other objects which are seen in positivecontrast (that is, object brighter than pavement).

Additional advantages of independent align-ment over narrow-median designs include:

Less earthwork required. Independent align-ment permits the designer to avoid areas of diffi-cult topographic, soil or drainage conditions.

More Environmentally Friendly. With inde-pendent alignment there is less likelihood of slope

failures and erosion, groundwater problems, andbedrock excavation. Also, the highway may belocated and designed in a manner that produces aminimum of visual and environmental disruption.

DisadvantagesThere are few disadvantages associated with

wide medians. The major disadvantages are theadditional cost involved.

Increased costs. Wider medians increase con-struction costs due to the following factors: (1)extra right-of-way must be purchased; (2) addi-tional gratings, seeding, mulching, and earthworkwould be required; and (3) more maintenance(mowing and shrubbery) would be required.

Poor traffic signal operation. A less-apparentdisadvantage of wide medians is found when at-grade intersections are required: The increasedtime needed for vehicles to cross the wide medianmay lead to inefficient traffic signal operation.

The disadvantages of independent alignment are:

Increased construction time. There are morenonproductive equipment movements requiredduring the construction of independently alignedroadways than with narrow-median roadways.

Right-of-way Requirement. Independent align-ment requires a larger amount of right-of-way thannarrow median designs.

SummaryWide medians and independent alignment are

effective means of controlling glare from oncom-ing vehicles and can often be justified by accidentreduction alone. However, this countermeasure isnot a practical solution to the glare problem onmany roads, particularly two-lane rural roads,where the traffic volume is low, and urban arterialroads, where land is not available to increase theright-of-way. As stated earlier in this report, glarefrom oncoming vehicles is primarily a problem ontwo-lane roads, and many of these will continue tobe such for some time. An additional solution toglare is needed for these situations.

76

• Ultraviolet Headlights• Fixed Roadway Lighting• Restricted Night Driving• Corrective Lenses and

Ophthalmic Surgery• Headlamp Area

Ultraviolet Headlights Ultraviolet (UV) headlighting has the potential

to reduce glare on the highway by reducing the needfor visible light. Although not itself visible, UVradiation (UVR) can improve nighttime visibility ofpedestrians, lane lines, signs, and other objects onthe road. Either replacing high-beam headlampswith UV headlamps or modifying the low-beamphotometrics in a UV-enhanced lamp (perhaps usinga European beam pattern) could eliminate much ofthe glare experienced by drivers of oncoming vehi-cles and still maintain visibility. Even without anyreduction in glare from visible light, disability glareeffects may be reduced by the improved visibility.

Ultraviolet radiation cannot be seen by mosthuman observers, but, when it is absorbed by cer-tain materials, UV is converted to longer-wave-length visible light. This phenomenon, called fluo-rescence, makes objects more easily seen and isthe reason that UV headlamps have the potential toimprove highway safety. Unlike standard automo-bile headlamps, UV headlamps’ intensity andalignment can be adjusted to optimize visibilitywithout concern for glare.

UVR ranges from approximately 4 nm to 400nm wavelength, but UV headlamps operate in therelatively long-wavelength UVA band, whichranges from 320 nm to 400 nm. The cornea andaqueous humor of the human eye absorb all UVRbelow 300 nm (Kinsey 1948) and approximately50% up to 365 nm (Zigman 1983); the healthyadult lens absorbs nearly all of the remaining UVR(Ham, Mueller, Ruffolo, and Guerry 1980, Zigman1983). However, two groups of individuals doexperience significant transmission of UVA andperceive it as light: youths under the age of ten(Lerman 1983) and the elderly who have under-gone eye surgery to remove and/or replace a lens(Anderson 1983, Sliney et al. 1994, Wald 1945,Zigman 1983).

ResearchMost of the research on using UVA headlamps

for enhancing highway visibility has been con-ducted in Europe, in particular by the SwedishNational Road Administration. In the U.S., theFederal Highway Administration (FHWA) becameactive in the mid-1990s, sponsoring a study enti-tled “Safety Evaluation of Ultraviolet (UVA)-Activated Fluorescent Roadway Delineation”(Nitzburg et al. 1998). This study included a com-prehensive literature review, which, together withdocumentation of the study, can be obtained on theInternet. More recently, the FHWA has funded aconsortium to undertake a comprehensive look atall aspects of implementing this technology.

Visibility — In general, the Swedish research hasshown that UV headlighting significantly improvesthe visibility of anything on the highway that con-

77

CHAPTER 8

COUNTERMEASURES THAT INDIRECTLYMINIMIZE THE EFFECTS OF GLARE

tains fluorescent pigments, without any increase inglare. While UV light alone did not enable driversto see where they were going, it did increase the vis-ibility distance to objects such as lane lines andpedestrians. UV headlights were able to highlightanything that contained fluorescent pigments,including road signs, road markings, and clothing.Pedestrians were seen more clearly and at furtherdistances and the path of the roadway could be seenfar beyond oncoming vehicles, thanks to the fluores-cent road markings that were used. While UV lightmay highlight the fluorescent pigments in some roadsigns, there is not likely to be a noticeable effectfrom visible light, as fluorescent materials are muchless effective in producing luminance than retrore-flective materials.

Clothes vary in their degree of fluorescence,depending on material and color. Studies havefound that even fabrics of relatively low fluores-cent efficiency, such as jeans, could be seen atapproximately 100 m (328 ft) in the presence ofglare and at more than 150 m (492 ft) withoutglare; in the absence of UV headlighting, the low-beam visibility distances were 60 m and 70 m (197ft to 230 ft). The improvement in visibility waseven greater for white cotton clothes and syntheticfabrics, but dark clothes, such as black wool, wereno more visible with UV light than with normallow beams. Washing can improve the fluorescentproperties of garments because many detergentscontain optical whiteners, which convert ultravio-let light into visible light.

A report from Ultralux (1994), cited inNitzburg et al. (1998), states that fluorescent roadmarkings can be seen at a distance of 150 m (492ft) with UVA light, compared with 60 m to 70 m(197 ft to 230 ft) with ordinary low beams. Theimprovement in visibility distance for roadsideposts was even better: they could be seen at morethan 200 m (656 ft) with UVA light.

Helmers et al. (1993), in a study cited byNitzburg et al. (1998), evaluated detection dis-tances for black, gray, and white targets as well asfor clothing under low-beam illumination supple-mented by UVA. The study was a full-scale simu-lation of two opposing vehicles on a straight,

level, two-lane road closed to traffic. The addi-tional UVA illumination resulted in only a slightincrease in detection distance for the black andlight gray targets but caused a doubling in detec-tion distance for the white targets. The ability ofclothing to emit visible light in response to UVAlighting was found to depend on reflectance: Whenthe reflectance of the clothes increased, the visiblelight resulting from UVA increased much morerapidly than the reflected ordinary light.

Nitzburg et al. (1998) found similar improve-ments in visibility from UVA in a series of teststhat included static tests to determine detectionand recognition distances for pedestrians, bicycles,disabled vehicles, traffic cones, and delineation,and dynamic tests to measure driving speed andlateral lane position, as well as subjective compar-isons of different headlighting configurations. Thestudy found that using UVA in addition to conven-tional U.S. low-beam headlights increased recog-nition distance by between 14% (right curve) andnearly 50% (no-passing zone and crosswalk). Stillgreater improvements were observed in the pres-ence of oncoming headlight glare, suggesting thatUV could be a direct countermeasure for disabilityglare. Objects that have fluorescent but no retrore-flective components (cones and bicycles) experi-enced improvements in detection and recognitiondistance of greater than 250%, whereas objectsthat have only retroreflective components (dis-abled vehicle) showed no significant improvementin visibility. Detection distances for five differentpedestrian cut-outs (child, adult, tall adult, kneel-ing adult, and jogger) ranged from 60–250 m withUVA, representing increases of 20–140 m overconventional headlighting.

When asked what headlights they liked bestand what headlights enabled them to best steer thecar, drivers clearly favored the U.S. low beam andUVA combination. Still, performance comparisonsin speed and lateral lane position showed no sig-nificant differences between the headlighting sys-tems that were tested, with or without UVA.

The studies indicated one potential danger ofUVA headlighting: Some of the fluorescentobjects were so highly visible that they may have

78

distracted drivers from attending to other objects.This is one of many human-factor issues that willneed to be addressed before UVA can be acceptedas a headlighting alternative.

The FHWA is currently sponsoring a compre-hensive research and development program to sup-port a universal UVA headlighting system. Theprogram, which is being carried out by theVirginia Institute of Technology Center forTransportation Research and the University ofIowa, is intended to do the following:

• Develop a UVA headlamp specification• Evaluate fluorescent infrastructure materials• Quantify the glare and photo-biological risks• Perform a cost/benefit analysis• Conduct a demonstration and implementation

of the system

Without headlamp specifications (includingthe photometric pattern for both visible light andUVA), the effectiveness of UVA as a countermea-sure will remain unknown. The project will con-sider coupling UVA with current low-beam,improved halogen, HID, and European headlampsin an effort to determine the ideal combination foradvancing this technology. Although HID lampsproduce a considerable amount of UV radiation,very little of this is UVA and most UVB and UVC is filtered for safety reasons (Schoon andSchreuder 1993). To use UVA with HID wouldprobably necessitate separate light sources.

To some extent, the optimum beam patternscannot be identified until there is an assessment ofthe fluorescent infrastructure materials. It is stillunknown, for example, whether the beam patternshould be designed to emphasize the visibility ofroad delineation, hazards, or pedestrians. Manyadditional questions need to be addressed. The via-bility of the concept depends on the assessment ofrisk and a realistic assessment of costs and bene-fits, and so these issues are also being addressed.Only after an effective system is defined andunderstood can the process of marketing the ideato the public and industry begin.

Health - The basic premise that UV is impercepti-ble to human observers is not quite true; somepeople actually do see near-UV radiation as light.Infants and children are one such group. Duringthe first year of life, more than 80% of radiation inthe UVA range is transmitted through the lens ofthe eye, and less than 25% of UVA radiation isabsorbed by the lens by age ten. It is not until 25years of age that lenticular UVA absorption reach-es its highest level—about 80% (Lerman 1983).

The 1990 U.S. census recorded over 37 millionchildren under the age of ten. These youths willbe substantially more sensitive to radiation fromUV headlamps than normal adult observers. Whendetecting a young pedestrian or bicyclist, courte-ous motorists turn off their standard high beams toreduce glare for these individuals. If the drivingpopulation is led to believe that UV headlamps areinvisible to the human eye, they will not be sensi-tive to any discomfort and disability they mayinflict. The extent of the glare effect is empha-sized by Sliney (1994), who stated that “anyyoung child, aphakic or pseudophakic person [nolens or artificial lens] whose retina is exposed to asignificant level of UVA will have a strong aver-sion response....” (Sliney et al. 1984, p. 15).

Intraocular lens implantation (IOL). Both the nor-mal aging process and constant exposure to naturalUVA radiation lead to an increase in lenticular flu-orogens (UV-absorbing chromophores) and toopacification (lens yellowing) so that eventuallythe lens absorbs nearly all UV radiation in the 300to 400 nm range (Lerman 1983). It is thought that,as an individual ages, this constant absorption ofUV radiation plays a significant role in the forma-tion of cataracts (Zigman 1983). Currently, thesurgical treatment of choice for cataracts is lensremoval and implantation of an artificial lens. Ifthe artificial lens is not treated with UV-absorbingmonomers and polymers—and many olderimplants were not—then all of the UVA that istransmitted through the cornea and aqueous humorwill reach the retina (Pitts 1990) and be perceivedas visible light (Anderson 1983, Wald 1945).

79

Ultraviolet-absorbing IOLs. If an IOL is notspecifically treated to absorb UVR, the retina ofthe wearer will be exposed to radiation in the UVArange (Werner and Hardenbergh 1983, Zigman1983). Research studies report that aphakic indi-viduals and pseudophakic individuals without UV-absorbing IOLs have 47 times the sensitivity ofnormal individuals to UVR at 380 nm (Werner andHardenbergh 1983) and 1000 times the normalsensitivity at 365 nm (Anderson 1983).Furthermore, some lenses that purport to be UV-absorbing have been found to transmit up to 70%of UVA radiation (Pitts 1990).

In 1983, the first UV-absorbing IOLs weredesigned (Zigman 1983). While it is impossible tosay exactly how many IOLs are in use today, it wasnot until 1985 that a substantial proportion (48%) oflens implants used this design (Stark et al. 1986).The government still does not require UVR absorp-tion treatments in IOLs (USHHS 2001).

Chromatic Aberration: Like most optical sys-tems, the human eye focuses different wavelengthsof light at different spatial locations; in a phenom-enon called chromatic aberration. If the eye isfocused on a yellow object, blue objects will focusin front of the retina, and red will focus behind theretina. An example of chromatic aberration is thedepth effect that occurs when a highly saturatedblue and a highly saturated red are juxtaposed.Chromatic aberration is not a noticeable problemunless the eye attempts to focus simultaneously ontwo divergent wavelengths. Unfortunately, this isjust what will happen when individuals with UV-transmitting lens implants try to see objects illumi-nated by UV headlamps.

Artificial IOLs are designed to focus lightoptimally in the spectral range of highest daytimeretinal sensitivity, which is centered around 555nm wavelength. If external objects are illuminatedpredominantly by light of wavelength 320–400nm, a blurring effect will occur. Although youngerUV-sensitive individuals can compensate for theUVA underfocus by accommodation, the artificiallens exacerbates chromatic aberration and removesthe eye’s ability to control focal length.

Two studies investigated the effects of chromaticaberration in the presence of UVA illumination:one in which monochromatic monitors were usedto test contrast sensitivity (Hammer, Yap, andWeatherill 1986) and another where wall chartswere used to test visual acuity (Rog, White, andWilliams 1986). The former study found visualperformance to be unaffected by IOL absorption ofUV; individuals with UV-transmitting lensimplants performed as well as those with UV-absorbing lenses. The latter study found a signifi-cant reduction in acuity for individuals fitted withUV-transmitting lenses when the test charts werebathed in light containing a UVA component;optotype visual acuity was reduced from 20/46 to20/57 and vernier acuity was reduced from 20/65to 20/78.

The observed reduction in visual acuity in thepresence of UVA could directly affect a pedestri-an's ability to read signs, and the perceived colorof traffic signs will likely change for these individ-uals (White and Wolbarsht 1975). Furthermore, iflight from UV headlamps is transmitted throughautomobile rear and side windows, a UV-sensitivedriver's ability to read in-vehicle instrumentationcould be compromised. UV-sensitive individualsmay not be aware of the loss in visual function andso might engage in risky behavior.

Aside from the possible drawback of adverseeffects on UV-sensitive individuals, the benefitsof a UV headlamp system could be considerable.A vehicle equipped with UV headlamps could beoperated with the lamps on continuously, conceiv-ably increasing pedestrian detection by 50 m to100 m over standard low-beam headlamps (Fastand Ricksand 1984). Replacing current highbeams with UV headlamps would dramaticallyreduce glare to oncoming vehicles and increasevisibility by not forcing drivers to dim their lightsfor oncoming traffic. Furthermore, because UVheadlamps do not cause glare, they could be aimedsolely to optimize visibility.

Further study is necessary to determine whetherthe effects of UVA headlighting on UV-sensitiveindividuals would result in real safety hazards ormerely increased discomfort for a subpopulation.

80

The literature supports the idea that UV-sensi-tive individuals will experience reduced visibilitydue to glare or chromatic aberration resulting fromUV headlamps. This reduced visibility could causea safety hazard for both pedestrians and drivers.Future research should study glare, acuity, andcontrast sensitivity using illuminants that matchthe spectral output of UV headlamps at the intensi-ties expected under normal roadway conditions.

Other health risks - The dangers and problems of ultraviolet light have been investigated by theSwedish Road and Traffic Research Institute andare currently under investigation in the FHWAstudy mentioned earlier. One proposal to remediatethe problems is to use filters to remove most of theharmful rays. Filtering the lower UV wavelengthsreduces biohazards (which includes prematureaging of the skin and skin cancer) and filtering theupper end (near the visible range) limits the poten-tial to generate glare. With such filters, the UVlamps appear black in daylight and glow faintlyblue in the dark. In addition to the use of filters,the Swedish Road and Traffic Research Institutehas proposed that sensors be used to turn off theUVA component when speed is reduced below 48km/h (30 mph). This practice, which is also underconsideration by the FHWA, is intended to mini-mize the health threat to pedestrians.

Materials - Widespread implementation of UVAheadlighting will require the development of newtraffic control devices. Kozak (1996), cited inNitzburg et al. (1998), identified materials thatwere likely to produce stable, long-life traffic con-trol devices and identified specific problems thatmust be addressed before any implementation isundertaken. The FHWA study in progress willevaluate alternative lighting systems in connectionwith fluorescent pavement markings, pedestrians,cyclists, and workers with fluorescent vests. Thisevaluation could be hampered by poor availabilityof materials; for example, not all colors are avail-able in fluorescent material. Measuring an object’sluminance under UVA relative to its irradiancegives an indication of the fluorescent efficiency ofits components. Fluorescent efficiency will varywith the spectral composition of the UVA source,and so is dependent on the actual filters chosen for

the system. The largest obstacle to the develop-ment of new materials is that industry has littleincentive to develop them without the implementa-tion of UVA headlighting, but UVA headlightingcannot flourish without the required materials.

Advantages• UVA offers the potential to offset the negative

effects of disability glare by improving the visi-bility of objects in the presence of glare.

• Improvements in visibility raise the possibilitythat high beams could be eliminated and thesharp-cutoff low beam pattern used in Europemight become acceptable.

• The greatest improvement in visibility isexpected in places without street lighting, whichare currently the places with the greatest prob-lems from glare.

• Since UV light is not normally visible, itsbackscatter under adverse weather conditions isnot visible either. Therefore, glare frombackscatter could potentially be reduced andvisibility improved in such conditions.

Disadvantages• The potential health threat to children and the

elderly needs to be quantified and the corre-sponding liability issues must be addressed.

• Longer UVA wavelengths might produce glare.Filters to eliminate such glare may reduce UVAoutput, reducing its efficiency with respect topower consumption.

• The introduction of UV headlamps does noteliminate the need for low-beam headlamps,because visible light is still required to illumi-nate retroreflective signs; this visible light willcontinue to cause a glare problem.

• Perhaps the biggest threat to the viability ofUVA headlighting is the limited durability offluorescent traffic control devices. There issome evidence that suggests that the life cycleof these products is limited. If they require fre-

81

quent replacement, costs will increase and cost-benefit ratios will be less favorable. Less obvi-ous is the threat that this problem poses to real-izing any benefits at all. If fluorescent devicesare not maintained, then not only would thebenefits of accident reduction be lost, but theaccident rate may very well be higher thanbefore the introduction of UVA. This problemwould be still worse if glare countermeasuressuch as elimination of the high beam and modi-fication of the low beam are implemented.

• The apparent improvement in visibility mightlead to higher speed and more risk-taking.

• The above-threshold levels of visibility that canbe achieved with UVA may prove distractingand result in some critical targets going unno-ticed. Whether drivers will adapt and eventual-ly ignore these distractions is not known.

• The costs of implementing this technology arestill uncertain.

SummaryGiven the extent and magnitude of the

research needed before UVA could be implement-ed at the national level, it is impossible to assessthe viability of this technology. Clearly, UVA hasthe potential for developing headlight systems thatattain the longstanding goal of improved visibilitywithout added glare. It is unlikely that UVA canreduce the level of glare generated by low beamheadlamps, because UV headlamps would noteliminate the need for low-beam headlamps—visi-ble light would still be required to illuminateretroreflective signs, since fluorescent materialsare much less effective in producing luminancethan retroreflective materials. Clearly UVA hasmany problems, including health issues and theavailability, durability, and cost of materials.

Fixed Roadway Lighting In the U.S., fixed roadway lighting is general-

ly designed according to one of the three methodsdescribed in the American National StandardPractice for Roadway Lighting (RP-8 2000), eachbased on illuminance, pavement luminance, or vis-ibility. Fixed roadway lighting is installed formany reasons, including reduction in nighttimeaccidents (primarily on major arterial roads andfreeways), pedestrian safety, crime reduction, andarea ambience (on urban streets and areas with sig-nificant retail activity at night). In addition to thesespecific purposes, a secondary benefit of fixedroadway lighting is the mitigation of the effects ofheadlight glare.

During daylight, headlight glare is generallynot a problem because of high visual adaptation inbright ambient light. At night, a driver’s adapta-tion level is much lower and headlight glarebecomes a problem. Adaptation level is deter-mined by the entire visual environment, which caninclude the surrounding ambient lighting in visual-ly complex areas, the steady stream of oncomingheadlights, or the luminance of the pavement inareas without much traffic or ambient lighting.Fixed roadway lighting is a countermeasureagainst headlight glare whenever pavement lumi-nance is a significant factor in determining the driver’s adaptation level.

ResearchThe theoretical concepts used in evaluating

fixed lighting and pavement luminance as a coun-termeasure for glare were discussed in Chapter 2.Pavement luminance has a direct effect on bothdiscomfort and disability glare. As shown in equa-tion 5, discomfort glare is determined partly byadaptation luminance, and insofar as pavementluminance determines adaptation, it also affectsdiscomfort glare. With regard to disability glare, itwas suggested in Chapter 2 that veiling luminancebe limited to below 15% of background luminance(generally dominated by the pavement luminance)in order to minimize the effects of glare.

82

Each of the three design methods in RP-8attempts to maintain minimum pavement lumi-nance levels. The illuminance method (RP-8,Table 2) does this by requiring greater illuminationon those roads with less reflective pavements andless illumination on more-reflective pavementssuch as concrete. The luminance method estab-lishes minimum levels of pavement luminance fordifferent types of roads (RP-8, Table 3). The visi-bility method (RP-8 Table 4), in addition to havingminimum visibility levels for a standardized target,also sets minimum pavement luminance levels tocontrol disability glare from an oncoming vehicle85 m away, both with and without a median sepa-rator. The resulting pavement luminance require-ments for the visibility method are somewhatlower than those for the luminance method, butthey are still thought sufficient to control disabilityglare whenever pavement luminance is prominentin establishing driver adaptation.

The effectiveness of fixed lighting systems inreducing night accidents has been evaluated byusing daytime accidents as the basis of comparisonbetween sites with different lighting systems andby comparing the accident rate at a given a sitebefore and after a change in lighting. These stud-ies have invariably indicated that the night acci-dent rate is reduced when supplementary fixedlighting systems are installed. According to RP-8:

The nighttime fatal accident rate is about threetimes the daytime rate based on proportionalvehicular kilometers/miles of travel. This ratiocan be reduced when proper fixed lighting isinstalled because these lighting systems revealthe environment beyond the range of the vehi-cle headlights and ameliorate glare fromoncoming vehicles by increasing the eye’sadaptation level...the IESNA Roadway LightingCommittee is of the opinion that the lighting ofstreets and highways generally is economicallypractical. These preventive measures can cost acommunity less than the accidents caused byinadequate visibility.

Cost-benefit analysis of lighting on the basisof accidents is not possible since the research isinconclusive concerning how much light is neces-

sary for maximum accident reduction. Severalstudies from the 1970s indicated that safety bene-fits are not always seen after additional investmentin lighting. In his analysis of 22 freeway sites,Box (1973) found that, although lighted freewayshad a lower ratio of night to day accidents thanunlighted freeways, 11 sites with horizontal illumi-nation greater than 6.4 lux had a higher night/dayaccident ratio than sites with illumination of lessthan 5.4 lux. A graph of night/day accident ratesas a function of illumination level had a U shapewith high ratios both for low and high levels ofillumination. Box concluded that the bestnight/day accident ratios were found on lightedfreeways with illumination of about 6.0 lux.While other studies have supported this conclu-sion, the relationship between the amount andquality of lighting has not been well established.

According to the National Safety Council(1999) the nighttime traffic death rate in 1998 was4.4 times the day rate. Of the 18,874 nighttimetraffic fatalities in 1998, over 56% took place onrural roads. Surprisingly, although vehicle mileageis less on rural than on urban roads and even lessat night, the nighttime rural death rate for the pastdecade has consistently been three times the ruralday rate and 2.5 times the urban nighttime deathrate. According to Keck, Wortman concluded thatfixed lighting was warranted on a cost-benefitbasis whenever the night-day ratio exceeds 3.0.Using this standard, there are likely to be manyrural locations that would benefit from lighting;once in place, the lighting would also help miti-gate the effects of glare.

AdvantagesThe benefits of fixed roadway lighting as a

countermeasure for headlight glare are twofold.First, fixed lighting increases the visibility ofobjects and pedestrians on the road, making it pos-sible to reduce the headlight illumination required.Second, fixed roadway lighting raises the adapta-tion level of the driver, thus reducing the effects ofheadlight glare.

Improved visibility with less headlamp illumi-nation. Given that motorists use high beams

83

infrequently in meeting situations, if at all (Hareand Hemion 1968), there is little to be gained fromfixed lighting in terms of reducing high beamusage. If fixed lighting were to result in a reductionof headlight illumination it would be due to the useof adaptive headlighting or the elimination of low-beam headlamps in some situations. For example,some European cities once encouraged motorists toswitch to parking lights within the city limits, wherefixed overhead lighting is prevalent and good.While this practice has been largely discontinuedbecause of the difficulty it created for pedestrians tosee an approaching vehicle in advance, daytimerunning lights might prove to be an adequate substi-tute. In either case, implementing such a policy inthe U.S. would require significant improvements inthe fixed-lighting infrastructure and a clear defini-tion of what constitutes good lighting.

Reduction of discomfort glare. Adequate pave-ment luminance from fixed lighting would miti-gate the effects of headlight glare in environmentsthat are otherwise relatively dark and where thereis not a constant stream of traffic.

Reduction in nighttime accidents. Numerousstudies show that fixed lighting lowers the night-time accident rate. With nighttime accident ratesoften four times the daytime rate, fixed lightingsystems can pay for themselves in accident reduc-tion alone.

Disadvantages Fixed roadway lighting has only two disadvan-

tages: cost and hazards from the poles supportingthe luminaires, which become a significant factorin fixed-object accidents.

Increased costs. Lighting systems involve fixedcosts for poles and luminaires and maintenancecosts for lamp replacement and electricity. Highmaintenance costs have led some agencies tomove toward high-mast lighting, which allowslamp replacement without the use of a buckettruck and disruption of traffic.

Effect on accidents. Thirty years ago, there wereestimates that lighting poles were involved in 5%

to 35% of fixed-object accidents (Farber et al1971, Cassel and Medville 1969). This problemhas been recently addressed by requiring greaterpole setbacks and by the use of breakaway poles;as a result, the role of poles in accidents has likelybeen reduced.

SummaryFixed lighting appears to be an effective coun-

termeasure to the negative effects of headlampglare that also improves visibility and reducesaccidents. Rural two-lane roads would apparentlybenefit most from fixed lighting, because in theseareas pavement luminance is more likely to deter-mine visual adaptation than any other source oflight, glare from oncoming headlamps is at itsworst, and the night-day accident ratio is highest.

Restricted Night DrivingThroughout this report, countermeasures

aimed at reducing or eliminating glare have beeninventoried and evaluated. These remedies havedealt mainly with changing the glare source (forexample, by intensity reduction or wavelengthmodification) or altering the conditions throughwhich the light passes (such as those in wind-shields and ocular media). However, there is noglare problem that does not involve the perform-ance loss or increased discomfort of a human oper-ator. Discomfort and disability glare are not theinherent characteristics of a light source, but ratherare the result of an interaction between lightsource, environment, and observer—and withregard to glare sensitivity, not all observers arecreated equal. Therefore, one countermeasurewhich would alleviate at least part of the glareproblem is to enforce or encourage night drivingrestrictions on those most sensitive to glare. Thiscould be done through law or by means of self-imposed behavioral changes, as advocated by driv-ing improvement programs aimed at older adults.

ResearchResearch on individual differences in and age-

related changes to glare sensitivity and to glarerecovery time have been treated elsewhere in this

84

report. The effectiveness of restricting night driv-ing as a glare countermeasure, however, hinges onthe availability of an accepted measure of glaresensitivity and the establishment of a causal rela-tionship between glare sensitivity and traffic safe-ty. The literature on these two topics is notencouraging. The reviews of vision screening byBailey and Sheedy (1988) and Hu, Lu, and Young(1993) found no standardized, clinically acceptablemeasure of glare sensitivity. Hu et al. (1993) addedthat, although glare tolerance is important, there isno statistical link between highway accidents andglare recovery. Shinar (1977) similarly found thatglare tolerance and recovery time had little rela-tionship with driving safety.

Even in the absence of hard scientific data,however, some researchers have made the judge-ment call that persons with glare problems shouldrestrict their night driving. For example, Hu et al.(1993) stated that, “Individuals with problemsadjusting to glare should limit their driving to day-light hours,” and Bailey and Sheedy (1988) wrote,“Individuals who are known to perform poorly ontests of glare or night vision should be consideredfor night driving restrictions.” Staplin (1994)noted, “Older drivers’ self-awareness of decliningvision, together with the high face validity ofvision testing to safe driving performance, makeslicense restriction on this basis socially accept-able....” This insight reduces the urgency for asso-ciating glare with traffic safety, but still leavesunresolved the issue of identifying an appropriatescreening device.

Most driver refresher courses encourage olderdrivers to avoid night driving and warn that condi-tions such as cataracts and glaucoma magnifyproblems with headlight glare.

AdvantagesTo the extent that disability and discomfort

glare negatively impact both traffic flow and safe-ty, the benefit of restricting night driving for thosemost effected by glare would be to improve thesetwo critical transportation indices. However, theadvantages may extend beyond the control ofglare. Those whose licenses would be restricted or

who would limit their own night driving on thebasis of glare sensitivity are likely to be thosewhose nighttime visibility requirements were notadequately met in other areas of performance,because increased glare sensitivity is closely corre-lated with other visual deficits (Schieber 1988).The advantages might therefore also include anoverall reduction in visibility-related nighttimecrashes. Furthermore, if older drivers would bethe most likely to have restricted licenses or torestrict their own driving on the basis of glare,then the reduction in crashes could result in anoverall decrease in fatalities and injury severity,because older drivers are disproportionally repre-sented in these two areas.

Disadvantages The principal disadvantage of restricting night

driving is the loss of nighttime mobility of therestricted drivers, most of whom would be olderindividuals who are already experiencing reducedmobility. Without community planning andinvolvement (including such measures as shuttlebus transportation), the costs associated with anyreduction in mobility would be distributed amongfamily and friends, who would be required totransport those who chose to avoid night driving.

An argument against license restriction is thecost of adding another component to driver’slicense screening. The actual costs associated with adding glare screening to driver licensure isnot known, in part because no mechanism is nowavailable, but Bailey and Sheedy (1988) list thefollowing cost-related issues that must be considered:

• Record keeping• Equipment costs and maintenance• Staffing• Staff training• Overhead (room costs and so on)• Applicant time

If there is a safety benefit, however, thesecosts could be offset by a reduction in the costs ofglare-related crashes.

85

SummaryWhile license restriction appears to have the

potential to reduce some of the reported problemswith glare, more research is required before it canbe recommended. A cost-benefit tradeoff is com-plicated, for a number of reasons:

1. The cost of administering the screening isunknown.

2. The social and personal costs associated withthe loss of nighttime mobility may be incal-cuable.

3. The benefits of reducing the number of glar-sensitive drivers on the road at night areunknown.

Before the costs and benefits of license restric-tion can be identified, standardized, clinicallyacceptable instrumentation must be available formeasuring glare sensitivity, and a sound, empiri-cally-based criterion for glare sensitivity must bedetermined.

Without a sound cost-benefit argument, alicense restriction based on glare sensitivity wouldbe politically unpopular and indefensible. Self-imposed restrictions on night driving such as thoseencouraged by driver refresher courses might bethe most reasonable method to restrict the nightdriving of those bothered the most by headlightglare. Family members and family physicians canalso play a role in encouraging older drivers tolimit night driving.

Corrective Lenses and Ophthalmic Surgery

In a vacuum there is no glare. The perceptionof glare is the result of aberrant light transmissionthrough some medium. Light emanating fromheadlamps must pass through air, windshield, anycorrective lenses the driver is wearing, and variousoptical media (including cornea, aqueous humor,lens, and vitreous humor) before reaching the reti-na. Any changes in refractive index from one

medium to the next will cause light to changedirection (refraction), and any small particles inthe medium will cause light to scatter (glare). Theeffects on glare of atmospheric conditions, wind-shield treatment, and optical media are mentionedelsewhere in this report. This section will bedevoted to a treatment of glare effects when lightpasses through corrective lenses (eyeglasses andcontact lenses), corneal modifications made duringsurgical procedures to correct refractive error, andartificial intraocular lenses implanted duringcataract surgery.

ResearchCorrective Lenses

The departments of motor vehicles (DMVs) inall states require that myopic (nearsighted) driversof any age wear corrective lenses if their visualacuity falls below a set criterion (such as 20/40).(Despite this requirement, USA Today inSeptember, 1999 reported that 10 states do notrequire vision testing for license renewal.) Manyolder drivers have presbyopia, and although theycan see distant objects clearly they must wear cor-rective lenses to read in-vehicle displays. In thelate 1980s, the Iowa Department of Transportationestimated that as many as 75% of drivers over 75years of age were wearing corrective lenses (IowaDOT 1989).

It has been proposed that scratched or dirtyeyeglasses and contact lenses might exacerbate theeffects of glare by scattering incoming light. Forexample, Schieber (1988) stated that glare canresult from damaged contact lenses or, as found byMiller and Lazenby (1977), from corneal injurydue to prolonged contact lens usage. However,there has been very little research conducted onthis topic.

An exhaustive keyword search of theTransportation Research Information Services(TRIS) database, which contains over 400,000records, together with queries of professionals inthe field resulted in the discovery of only oneresearch study that directly evaluated the impact ofcorrective lenses on glare. In that study, Sivak,

86

Flannagan, Traube, and Kojima (1997) asked agroup of myopic subjects to compare discomfortglare with contact lenses to that with eyeglasses;subjects reported no difference between the two.The study also found no significant difference indiscomfort glare between subjects who wore cor-rective lenses and those who did not. This singlestudy, involving only sixteen subjects, certainlydoes not provide a definitive answer to whethercorrective lenses affect glare, and it does not con-tradict Scheiber’s (1988) statement that damagedlenses may cause disability glare. Study subjectscould well have had corrective lenses that wereclean and free of scratches, and in any event thislaboratory study may not directly apply to thenighttime highway environment. However, in theabsence of any conclusive empirical data on theeffects of corrective lenses on glare, it would beprudent for drivers to keep their glasses and con-tact lenses, like their windshields, clean and free ofscratches and abrasions.

Ophthalmic SurgeryRefractive Error — Radial keratotomy (RK),photorefractive keratectomy (PRK), and LASIK(laser in situ keratomileusis) are surgical proce-dures to correct refractive error. In RK surgery, ablade is used to make radial incisions in thecornea, whereas in the other techniques, the frontportion of the cornea is removed (PRK) or pulledback (LASIK), and a layer of stroma, or cornealtissue, is shaved off with lasers. All three tech-niques involve reshaping the cornea in an effort tochange its refractive index so that light is betterfocused on the retina.

Most recipients of these procedures havereported increased sensitivity to glare and haveexperienced “flaring” (streaks of light emanatingradially from a light source), and/or “haloing” (theappearance of an annulus of light around a lightsource) from oncoming headlights, at least in theinitial three to six post-operative months(Consumer Reports 1999). This increased glaresensitivity occurs when the pupil dilates at night toa size larger than the portion of the cornea thatwas modified by surgery. Currently, refractive

surgery in the U.S. occurs within a 6 mm zone.Light passing through the edge of the zone willscatter. If pupil size is larger than 6 mm—a com-mon occurrence for young people at night—therewill be glare. This is a permanent condition,though it may become less prominent after thefirst six months following surgery when the edgebecomes smoother. Because of this phenomenon,measurement of patient maximum pupil size isrecommended as a precursor to this type of surgi-cal procedure (Applegate and Gansel 1990,Applegate 1991).

Pupil size, however is not the only contribut-ing factor to glare. Glare may also be the result ofslight irregularities (optical aberrations) in themodified area of the cornea. Proponents of refrac-tive surgery assert that most, if not all, glareeffects dissipate within the first six months of sur-gery, but a recent Consumer Reports (1999) articlestated that perhaps as many as 10% of patientshave lasting problems with “glare, ghosting, orfuzziness.” Other studies reported in the medicalliterature have found reduced contrast sensitivity,poorer low-contrast visual acuity, and glare dis-ability for up to two years after surgery(Applegate, Hilmantel, and Howland 1996,Applegate, Trick, Meade, and Hartstein 1987,Ghaith, Stulting, Thompson, and Lynn 1998).

Cataracts — It is hypothesized that, as an individ-ual ages, absorption of UV radiation by the lenscauses the formation of cataracts (Zigman 1983).Currently, the treatment of choice for cataract sur-gery is lens removal and implantation of an artifi-cial lens (intraocular lens, or IOL). The popularityof lens-replacement surgery skyrocketed in the1980s and early 1990s; currently, nearly 1.5 mil-lion IOL implant procedures are conducted annual-ly in the U.S. (Daily Apple 1999). By the mid-1990s, approximately 10 million IOLs had beenimplanted (USHHS 1994). A recent report by theU.S. Department of Health and Human Services(USHHS 2000) reported that, “Cataract extractionwith prosthetic IOL insertion is the most commonprocedure paid for by Medicare.... In 1991,Medicare paid for an estimated 1.14 millionIOLs.”

87

There are separate glare problems associatedwith cataracts and with cataract removal. First, asmentioned in Chapter 2, a cataract both reducestransmission of light into the eye and scatters lightas it passes through the lens, creating a veilingluminance effect that reduces contrast sensitivityand increases disability glare. Indeed, headlightglare is often the first sign of cataract develop-ment. Removal of the affected lens allows light toreach the retina undisturbed, but the cataract (orlens) removal presents two potential glare sideeffects.

Edge Glare - Cataract surgery involves theremoval of the lens and either IOL implantation ora prescription for contact lenses or eyeglasses.The possible glare effects of corrective lenses thatwere discussed above apply here as well. In addi-tion, a separate phenomenon has been reported,known as “edge glare” because it results from theedges of IOL implants. This phenomenon is simi-lar to the glare effects resulting from refractivesurgery. In the case of cataract surgery, “The glareresults from the smaller optical zone and exposureof the IOL edge to the light that will pass throughthe aphakic capsule peripheral to the optic.”(Steinert 1997). In other words, light hitting theedge of the implant often scatters out of the “cor-ner of the eye.” Edge glare is a growing concern,as is reflected in comments made by TheAmerican Society for Corrective and RefractiveSurgery (ASCRS) on the Food and DrugAdministration’s Draft Intraocular Lens GuidanceDocument. ASCRS (2000) expressed concern,stating that “unwanted optical images, such asglare, halos, etc,... [have] become an increasinglyimportant visual characteristic associated with theperformance of an IOL,” and “...in addition tovisual acuity, a statement should be made withregard to the overall optical performance of theimplant. In particular, the occurrence of unwantedoptical images such as glare and halos should beincluded.”

A clinical study on IOL satisfaction (Masket2000) indicates that there could be additionalproblems associated with IOL implants, particular-ly multifocal lenses. Masket asked his patients,“How satisfied are you with your ability to see at

night?” Eight of the 20 multifocal lens patientsinterviewed complained of glare, halos, and star-bursts, while only one of the 20 monofocalpatients reported such problems.

UV Glare — The second potential problem asso-ciated with cataract removal is a largely unad-dressed, and at present hypothetical, conflictbetween the use of UV headlamps as a glare coun-termeasure and individuals who have undergonecataract surgery. This issue is discussed in somedetail above, in the discussion of UV headlamps asa glare countermeasure.

The extent of the potential UV glare problemis difficult to estimate. However, given that wind-shields absorb most UV radiation and that the pop-ulation of affected individuals consists mainly ofchildren under 10 years of age and the elderly,disability or discomfort glare from UV headlampsis most likely to have its greatest effect on pedes-trians rather than drivers. However, in September2000 there were approximately 39 million U.S.residents under the age of 10 and, while the num-ber of non-UV-absorbing IOLs in use today isunknown, an unpublished survey (Garvey 1994)estimated it to be around one million.

Advantages• Correcting refractive error and cataracts may be

necessary to meet state requirements to legallyoperate a motor vehicle and has overall safetybenefits, such as improved visual acuity and con-trast sensitivity, that also improve a driver’s abili-ty to read traffic signs and/or in-vehicle displays.

• For corrective lenses, keeping them clean andfree of scratches may minimize the effects ofglare—although there is no definitive researchto confirm this. Scratch-resistant glasses areavailable at very little extra cost to the con-sumer (about $20). Soft contact lenses areavailable that can be cleaned and rarely scratch,and when they are scratched they can be dis-carded. Hard contact lenses can be scratched,but the abrasions can be polished out inexpen-sively (about $10).

88

• Ophthalmic surgery could grant freedom fromcorrective lenses, although this is not guaranteed.

• Cataract surgical costs are covered by insuranceand Medicare.

• Recent research (Owsley, Sloane, Stalvey, andWells 1999) found a significant correlationbetween cataracts and at-fault crash involve-ment. In other words, drivers with cataractswere more likely to cause an accident than sim-ilar drivers who did not have cataracts.

Disadvantages• Corrective lenses can be lost or scratched.

Given their importance to safety, owning a second pair is recommended.

• Refractive eye surgery costs between $1,500and $5,000 for both eyes. There are risksinvolved with this surgery besides those relatedto glare sensitivity. The costs, in addition to theinitial surgical expense (which is considered cos-metic and not covered by most insurance plans),include subsequent surgery for further correc-tions, possible temporary or permanent reductionin contrast sensitivity, and impaired night vision,as well as increased glare sensitivity.

• With cataract surgery, as with any surgical tech-nique, there are risks. If an IOL implant isselected, problems with edge glare can be sub-stantial. However, newly developed lenses withmodified edges might help reduce glare occur-rence in the future (Holladay, Lang, andPortney 1999).

• At present, UV glare issues are unresolved andhypothetical, as UV headlamps have yet tobecome a reality.

SummaryA driver must have visual acuity that meets

criteria set by the state DMV to legally operate amotor vehicle. In the absence of other contribut-ing pathologies, visual acuity at a distance can

usually be corrected inexpensively and safely tomeet state requirements by using eyeglasses orcontact lenses. Given the importance of vision tosafe driving, maintaining these lenses clean andscratch-free is worthwhile, even though the contri-bution to a reduction in glare may be minimal.

Maintain MinimumHeadlamp Area

As mentioned in Chapter 2, a larger glaresource will generate a lower discomfort level (allother things being equal), because luminance mustbe reduced if illuminance at the eye is to be heldconstant. Although all U.S. headlamps must meetstringent photometric criteria, size and shape aregenerally unregulated. One consequence of this isthat projector-style HID lamps may be contributingto discomfort glare because their surface area issmaller than standard headlamps and so their lumi-nance is much higher. As a countermeasure, somevehicles with HID lamps now incorporate reflectorsinto the lamps that have a larger area than the lamp.

Discomfort caused by small lamps with largeluminance may occur when meeting another vehicleor from mirrors reflecting the image of the head-lamp of a following vehicle. Glare from smalllamps might also be accentuated when cars arestopped on both sides of an intersection waiting fora light to change. In this situation, there is shortviewing distance and long exposure, and, given theidle time, there may be a tendency for drivers toinspect the oncoming lamps more closely.

ResearchSivak et al. (1990) reviewed a paper by Lindae

(1970) that concluded headlamp area was impor-tant in determining discomfort glare. Lindae rec-ommended a minimum area of 150 cm2 for theEuropean type low-beam headlamp.

Sivak et al. also conducted a laboratory studyof the effects of glare illuminance and size on dis-comfort. Subjects performed a tracking task withthe glare source located 15.25 m away and at aglare angle of 3.6 degrees. For a given glare size

89

(sources were either 45.4 or 181.5 cm2), discom-fort as measured by the DeBoer scale was a linearfunction of the logarithm of glare illuminance.This was the expected result, already given by theformula in Chapter 2. The comparison of the twoglare sizes revealed a small (0.2 units on the 9point scale), but significant effect of size on dis-comfort when illuminance was held constant. Thesmall glare source having greater luminance wasmore discomforting.

On an actual two-lane road, the glare angle oftwo opposing vehicles is about 12 degrees, not the3.6 degrees simulated in the Sivak et al. study.The effect of headlamp size on discomfort glarewould therefore be even smaller than that suggest-ed by this study unless the driver chose to look inthe direction of the oncoming headlights.

As headlamp size is reduced at any given dis-tance, the lamp becomes more like a point source,for which luminance is not a relevant parameter.Sivak et al. considered the transition between pointand extended source to be in a zone the size of atarget subtending a visual angle of between 10minutes and 1 degree. As lamps become smaller,they approach a size at which the eye does notrespond to luminance; to maintain a sufficient sizefor the eyes to respond to luminance, the glaresource must be moved closer. With two opposingvehicles, moving the glare source closer willincrease the glare angle, which will also reduce thediscomfort glare and decrease the likelihood thatthe driver will look in the direction of the source.

Advantages• The primary advantage: reduction in discomfort

for drivers who choose to look toward the glaresource at close distances.

• Reduction in discomfort when looking in therear-view mirror, when not on the anti-glare setting.

Disadvantages• Maintaining a minimum headlamp size would

constrain the flexibility that designers wouldlike to have to appeal to consumer taste.

• Any physical restrictions on the headlamp mayrestrict future capability to control the headlampbeam pattern. This restriction could be particu-larly important for adaptive headlights.

SummaryGiven the expected modest impact on glare,

there does not appear to be any need to regulateheadlamp size. As lamps are made smaller, theybecome more like a point source and closer dis-tances are required for the eye to respond to lumi-nance; but at closer distances, the glare angle foron-coming vehicles increases, reducing the dis-comfort glare. Discomfort glare from small head-lamps is a potential problem when vehicles meetat intersections; this situation requires further con-sideration.

Headlamp area is most likely to be an issuewith respect to mirror glare, because the distancesinvolved are typically shorter and the visual anglesubtended larger than for oncoming vehicles.Although the glare angle is large in the mirror,drivers are likely to look directly at the glaresource. Although the luminance as viewed in mir-rors has not been studied, anti-glare mirrors couldprobably be an effective countermeasure. Thereare few headlamps that are both very small andmounted very high (such as projector lamps inSUVs and pickups), and so, presumably, the prob-lem of small lamps in rear-view mirrors is notsevere (Flannagan 2000). However, some of themost recent SUVs are equipped with projector-style lamps, and this could be cause for concern.

90

This report has discussed a wide range ofcountermeasures that could help mitigate theeffects of headlight glare on the vision and dis-comfort of drivers at night. Many of these coun-termeasures are effectively used today, but someproposed solutions have regulatory and technolog-ical hurdles to surmount before they will be able toresolve the most intractable problems with head-light glare. Within this report, countermeasureswere grouped according to method of operation,including such approaches as reducing intensity,reducing illumination, increasing the glare angle,and providing an indirect benefit by some othermeans. This section will provide a summary ofthese countermeasures, grouped by who shouldand who can take a particular action. This sectionwill conclude with a discussion of the researchneeded to develop and/or justify the use of someof these countermeasures.

Summary of CountermeasuresCountermeasures that can only be initiated by

a highway agency include the following:

• Wide medians and independent alignment• Glare screens• Fixed roadway lighting

Wide medians, independent alignment, andglare screens are all effective in eliminating glarefrom oncoming vehicles, either by increasing theglare angle so that the effect of glare is reduced orby blocking glare illumination completely. Widemedians and independent alignment operate pri-marily in rural and suburban areas, where right-of-way is available, whereas glare screens operateprimarily in urban areas, where right-of-way is not

available. Wide medians and independent align-ment are part of the design process and are notgenerally remedial treatments that can be subse-quently used to counteract glare. While the cost ofsuch measures can be justified by accident reduc-tion alone, the accidents avoided may only be evi-dent when volume has grown much higher thanwhen the road is first designed and built.

Glare screens are a remedial treatment and canbe installed when the initial design could not pro-vide sufficient right-of-way for a wide median orwhen volumes unexpectedly increase over timeand worsen the glare and safety problem.Practical limits on height restrict the use of glarescreens to relatively flat topography and gentlecurvature. Since their effect on the incidence ofaccidents is not documented, their cost must bejustified on the basis of comfort and the expectedsafety benefits. Glare screens, while effective, arenot a cure-all for every glare situation. They aremost appropriate where both traffic volume andthe demands of the driving task are high, as isoften the case on urban arterial roads and in con-struction work zones with narrow pavement widthand concrete barriers that are sometimes difficultto see.

Like wide medians and independent align-ment, fixed roadway lighting is often justifiable onthe basis of accident reduction alone. However,there is wide variation in the U.S. in the fraction ofroads that are lighted. Fixed lighting requiresaccess to electrical power and is generally restrict-ed to urban areas. Installation of fixed lighting alsodepends on budgets and reflects varying criteriaused by highway agencies. The question ofwhether low-volume roads with low accident rates

91

CHAPTER 9

SUMMARY AND RECOMMENDATIONS

should be illuminated to minimize the effects ofglare is not easily answered. Drivers certainly aremore comfortable driving on illuminated roads,and this benefit alone might justify lighting moreroads. However, this should be a local decision,made with an understanding of local resources andpriorities. Therefore, it is unlikely that all roadswith glare problems will ever be illuminated, anduntil this happens, another solution to the prob-lems introduced by glare must be found.

To the extent that task difficulty is associatedwith discomfort glare, a variety of road improve-ments, including alignment, lane width, surfaceand markings, and so on, may reduce the discom-fort experienced from headlight glare.

Countermeasures that are primarily the respon-sibility of industry are:

• Adaptive headlamps• Headlamp height• Headlamp area• Color-corrected headlamps

Of these four countermeasures, limiting head-lamp height is the only one that could be imple-mented quickly and with little if any cost. Color-corrected headlamps, while offering a low-costsolution, require a significant amount of additionalresearch before adoption. Headlamp area wasshown to have little practical consequence, withthe possible exception of when drivers are waitingat an intersection. Adaptive headlamps, while the-oretically promising, have numerous design, cost,and regulatory obstacles to surmount. While it isclear that adaptive headlamps offer significantimprovements to visibility, for example on curves,it is not at all certain what their effect would be onglare. Cost and maintenance issues also need tobe resolved before adaptive headlighting becomesa practical countermeasure to glare.

Lowering headlamp height is a promising “nocost” countermeasure to glare, but its impact islimited to mirror glare. With light trucks (includ-ing pickups, full-size vans, and sport-utility vehi-cles) representing 50% of all light vehicle sales,lowering the headlamp height of these vehicles

should be pursued and, if appropriate, the conclu-sions of the SAE task force should be investigatedfurther to establish the impact of lower headlampheight on visibility.

Countermeasures that are under the control ofindividual drivers are:

• Night-driving glasses• Anti-glare mirrors• Corrective vision solutions

Research appears to show that, for most indi-viduals, night-driving glasses are not an effectivesolution to the glare problem. What is gained inthe reduction of discomfort is lost in visibility.This conclusion applies to both full-eye glassesand half-glass analyzers that allow the driver tolook through the analyzer only on demand.Although one study suggested that discomfortglare had little effect on driving performance, themeasurements of performance were entirely psy-chomotor and not visual. Research is needed tobetter understand the relationship, if any, betweendiscomfort and eye fixations, attention, andfatigue.

The conclusion that the loss in visibility fromwearing night-driving glasses is not offset by thereduction in discomfort is grounded in the assump-tion that discomfort results in no immediate per-formance deficit other than its effect on visibilityand the annoyance it causes. This assumption orig-inated in the laboratories and may not be valid onthe highway, where people behave differently thanthey do when sitting in a research laboratory. Inaddition, although we know how driving affectstheir rating of discomfort, we do not know howdiscomfort glare affects eye fixations, attention,and fatigue. There is research that suggests thatdrivers’ eyes are attracted to light but are drawnaway from glare sources. Additional research isneeded to support or deny the assumptions beingmade and the conclusion that night-driving glasses(including half-glass analyzers) have no value foranyone driving at night.

Anti-glare mirrors, together with limits onheadlamp height and enforcement of standards for

92

headlight aiming are all that is needed to controlmirror glare created from passing or followingvehicles. Until all vehicles are equipped withsome type of automatic glare-reduction mirrors inboth the rear-view and left side positions—and wedo not doubt that this will happen—drivers need tobe encouraged to use the night setting of theirprism mirrors and to aim the left outside mirror sothat it does not reflect directly in their eyes. Thequestion of what drivers need to see in their rear-view mirrors needs further study so that automaticglare reduction mirrors can be properly designedand drivers believe that they are seeing everythingthat is necessary.

Vision correction should be encouraged tobenefit visibility but reduction in glare would beminimal.

Countermeasures that will require governmentinvolvement include:

• Changing beam photometric distribution• Maintenance of headlight aim• License restriction• Ultraviolet headlights• Polarized headlighting

Any major modification to the low-beam pho-tometric distribution must be made by the federalgovernment by means of FMVSS 108. As dis-cussed in this report, the last modification, in1997, included some minor changes to allow asharper cutoff for VOA headlamps. This modifi-cation has resulted in development of some head-lamps that produce less light above the horizontal,threatening a potential reduction in the illumina-tion for signs. Attempts to achieve a standardwhich harmonizes ECE and U.S. requirementshave not proven successful. Every effort to reduceheadlamp illumination or change the beam patternhas been met with concern about visibility, andevery effort to increase illumination has been metwith concern about glare. Research does not sup-port any change in either direction, any more thanthe research supports the use of the current beampattern. The present standard is a compromisethat has evolved over time and that works reason-ably well; the risks of making any major change

appear too great. Therefore, any effort to developcountermeasures for glare should focus on one ormore of the other alternatives discussed in thisreport.

While one might think that maintenance ofheadlight aim is a countermeasure that could beimplemented by the driver, there is generally littleincentive for drivers to do so. A driver is likely tocorrect his own misaimed headlamps only whenthey are not providing adequate visibility of theroad ahead, and in this situation they are usuallynot creating a glare problem. If misaimed head-lamps are creating a glare problem, the driver islikely to think they are fine, because for the driverincreased glare production is associated withimproved visibility.

Misaimed headlamps are the most problem-atic source of headlight glare. Without anenforced limit to the amount of misaim, it isimpossible to define the worst case that any glarecountermeasure must remedy. The introduction ofVOA headlamps has made it easier to detect mis-aim, but does not provide any incentive to correctproblems. While regulation of vehicle inspectionis thought to be a state issue, federal involvementmay be appropriate for vehicles that cross statelines.

Driving restriction through license restricture,while not a viable alternative from either a politi-cal or practical perspective, may be effective ifself-enforced. Older drivers can be sensitized tounderstanding their limitations by relatives andfamily physicians as well as through older drivereducation courses.

Ultraviolet lighting has some potential toreduce headlight glare, but only indirectly. Thistechnology should and will be pursued primarilybecause of its benefits to vision, particularly visionunder adverse conditions such as rain, snow, andfog, but it is not likely to replace conventionallighting and so cannot offer a realistic countermea-sure to glare.

Of all the countermeasures discussed in thisreport, polarized lighting is the one that could

93

eliminate glare entirely and make the nighttimeroad a friendlier place to travel. With the adventof HID lamps, the very technology that has height-ened the glare problem offers the ultimate finalsolution. With polarized lighting, the tradeoffbetween visibility and glare is resolved, and wecan have our cake and eat it without recrimination.It appears that the only real obstacle to pursuingthis countermeasure further is the difficulty ofimplementation.

Research NeededIn addition to the countermeasures discussed

in this report, there are a vast number of patentedproducts in various stages of development. Manyare for products that offer enhancements to thecountermeasures already discussed.Unfortunately, it is beyond the scope of this reportto explore the potential of all of these concepts.Hopefully, those ideas that offer useful solutions tothe glare problem will gradually find their wayinto commercial products.

Two of the potential countermeasures dis-cussed in this report are being studied in largeongoing research programs. Adaptive headlamptechnology is being developed and strategies forits implementation are being devised with the sup-port of several European countries and manufac-turing firms. In the U.S., a comprehensiveresearch program for the development and imple-mentation of UVA headlamps is underway. Bothadaptive headlighting and UVA headlamps aremore likely to benefit visibility than to offer anycomprehensive solution to headlight glare.

Topics that are not currently being investigatedbut which should be include one countermeasure(polarized lighting); the effects of the spectral con-tent of light on visibility and discomfort atmesopic adaptation levels; the relative effective-ness of electrochromic, photochromic andneodymium mirrors on glare sensation and rear-ward visibility; the identification of the populationof drivers most affected by glare and the reasonsfor their problems with headlight glare; and a morecomplete description of the behavioral effects ofdiscomfort glare.

Polarized HeadlightingIt seems that polarized lighting is the only

countermeasure for headlight glare which has thepotential to be used in all situations and resolve allproblems. Research has shown that headlampintensity must be doubled before polarized lightingwould be feasible; however, HID lamps offer threetimes the intensity of ordinary headlamps and caneasily meet this requirement. Thus, the onlyremaining challenge is developing a strategy forimplementation.

Hemion, Hull, Cadena, and Dial (1971) sug-gested that a large-scale public test of polarizedlighting should be conducted to observe first-handthe public’s reaction, as well as to determine thebenefits, side effects, and operational problemsthat would result from a change to polarized head-lights. This test should take place at a site with agood cross-section of the country’s motoring pub-lic but which can also be kept essentially isolatedfrom the intrusion of vehicles having normal head-lights (Hemion 1969b). An ideal site would be anisland that could only be accessed by ferry or boatand where incoming, unmodified vehicles could bemodified before interacting with traffic.

During a delay while developing the plans forsuch a test, Hemion (1971) conducted a small-scale test of 120 drivers. As explained in Chapter6, the results of this small-scale study showed thatthe benefits outweighed the deficiencies of thesystem. After completing the small-scale study,the authors recommended that, in order to achieveclarification in the areas of system development,operation, and performance, the goals of a large-scale test should include the following:

• To determine the most effective mode of polar-ization during and after the transition period,from the following possibilities:

(1) Polarized high beams with unpolarized low beams;

(2) High and low beams polarized;(3) Polarized high beam for use on rural,

unlighted roadways with unpolarized, low-intensity low beam for urban, lighted roadway use;

94

(4) Unpolarized high beam for open road operation with polarized high beam for meeting other vehicles on rural, unlighted roads and unpolarized, low-intensity low beam for urban, lighted roads.

• To compare the consequences of an extendedtransition period for conversion to polarizationwith the consequences of a simultaneous, “com-mon-date” conversion.

• To determine the extent to which headlightpolarization beneficially or adversely affectstraffic factors such as traffic flow, accidents,vehicle utilization, and vehicle maneuvers.

• To determine the extent to which polarizationbeneficially or adversely affects vehicle opera-tion and maintenance requirements, includingsuch factors as battery and generator life, bat-tery charge, system voltage, lamp life, polarizerand analyzer life, windshield depolarization,and headlamp aim and reaim requirements.

Exploration of the Spectral Content of Light

As indicated in Chaptert 5, more study is warranted on the effects that varying the spectralcontent of light would have on visibility, driverdiscomfort, and fatigue. Tradeoffs between gainsand losses in visibility need to be compared withchanges in discomfort glare before any significantchanges are made to the spectral content of head-lighting.

Effectiveness of Electrochromic,Photochromic, and NeodymiumMirrors on Glare Sensation andRearward Visibility

Additional research is needed to evaluate thetradeoffs between rearward visibility, forward visi-bility, and discomfort glare for the three competingautomatic control systems for mirror glare dis-cussed in Chapter 6: electrochromic, pho-tochromic, and neodymium mirrors. Studiesshould help to determine the appropriate levels of

reflectivity, what rearward visual information driv-ers need, and the nature of any luminous or spec-tral filtering under various conditions.Optimization algorithms are needed for each newtechnology to ensure that efforts to minimize dis-comfort do not result in the obscuring of criticalstimulus elements, such as original color or thecontent of a changeable message sign (CMS).

Identification of the Population ofDrivers Affected and the Basis ofComplaints about Headlight Glare

Although glare from automotive headlampshas been a concern from the time they were firstused, only in the past few years have public com-plaints become highly vocal. What fundamentalchange has occurred to make the problems withheadlight glare so serious? Higher headlampmounting heights, which are associated with anincrease in the number of SUVs, and HID head-lamps, along with their cosmetic after-marketderivatives, are most often the recipients of blame.However, other factors may make significant con-tributions to the problem, particularly increasedtraffic volume, headlamp misalignment, and anaging population. Research identifying the contri-bution of each factor to the increase in complaintsabout headlight glare would help direct efforts tofind appropriate countermeasures.

In addition to learning the basis for complaintsabout headlight glare, research should seek toidentify who is complaining. At a minimum, theinformation obtained would include the age, sex,and urban/rural driving patterns of those voicingcomplaints. One very plausible reason for theincreasing complaints about glare is the aging ofthe population—the number of older drivers isincreasing, as is the average age of all drivers.While the relationship between age and discomfortglare is inconclusive, age is related to the inci-dence of cataracts and bears a relationship to avariety of other measures of visual performance.If an analysis of complaints were to show a dispro-portionate representation among older drivers, theappropriate countermeasure might be to pursue

95

voluntary restrictions on night driving. While an understanding of complaints about

headlight glare is important, it must be remem-bered that disability glare problems may exist evenin the absence of complaints about discomfort.Many drivers who are not significantly botheredby glare from headlamps may nonetheless suffersignificant decrements in visibility and not beaware of the loss.

Description of the Behavioral Effectsof Discomfort Glare

A detailed discussion of discomfort glare waspresented in Chapter 2. For some time the accept-ed definition of discomfort glare has been the 9-point DeBoer scale, where discomfort is ratedfrom 1 (unbearable) to 9 (just noticeable). Thecriterion typically used in making judgements ofdiscomfort glare is that the rating should be 5 (justacceptable) or greater. The vocal complaints aboutheadlight glare are presumed to reflect experiencesof glare rated 4 or lower.

While it may be worthwhile to attempt toreduce the number of complaints about glare, anycost-benefit analysis must identify how the reduc-tion in complaints will be accompanied by behav-ioral changes that would improve safety.Whenever discomfort glare is used as the criterionfor evaluating a countermeasure, it would be valu-able to know how the level of discomfort glarerelates to driving performance, for example bycausing drivers to look away from the road or byproducing a reduction in alertness mediated byfatigue. Theeuwes and Alferdinck (1996) foundno relationship between discomfort glare andspeed reductions, but they did not consider theeffect of discomfort glare on fatigue or on thedirection of the driver’s attention.

In the discussion of night-driving glasses inChapter 6, the possibility was raised that the half-glass analyzer might be used only when the dis-comfort is so great that the choice is between look-ing off the road, seeing the road peripherally with-out glasses, or looking straight ahead with foveal

vision reduced by the glasses. Without the knowl-edge of how discomfort relates to visual acuity, thepotential usefulness of night-driving glasses can-not be dismissed. Research is needed to quantifythe effects of discomfort glare on driver fixationsand overall alertness, and specifically to determinewhether the half-glass night driving analyzer canoffer a net benefit in reducing fatigue and/orenabling drivers to keep their eyes on the road.

Education May ReduceComplaints of Glare

Several organizations, including AAA, AARPand the National Safety Council offer refreshercourses for older drivers. While not presented asa legitimate countermeasure, these programs makea number of recommendations for reducing theproblem of glare from headlights. The most basicsolution suggested by this course is to simply

• Avoid night driving

and, if you must drive at night,

• Limit night driving to well-lit roadways,• Increase following distance,• Avoid looking into headlights—look slightly to

the right of glaring headlights.

These courses also recommend some actionsthat relate to countermeasures discussed in thisreport.

• Position outside mirrors so the headlights offollowing cars are not directed into your eyes.

• Do not wear colored lenses, anti-glare glasses,or sunglasses for night driving, unless directedto do so by an eye doctor.

• Keep headlights, taillights, windshields (insideand out), and your eyeglasses clean.

• Keep headlights properly adjusted.• Have eye examinations, frequently, by a

licensed ophthalmologist or optometrist.

96

ConclusionsOne countermeasure that has been proposed is

the elimination of HID lamps. The application ofHID in both the UVA and polarized headlamp sys-tems is a reason to remain active in this technolo-gy, but the elimination of low-beam HID lamps bygovernment regulation may be an appropriate stepto control glare, if these lamps are indeed thesource of glare problems. Although HID lampsoffer an enormous increase in visibility, they canalso result in a comparable increase in glare.

Properly aimed and cleaned, they should beable to deliver improved visibility without exces-sive glare on roads without curvature; however, ifthese lamps become dirty, are misaimed, or areencountered on vertical or horizontal curves, theamount of glare can exceed levels considered tol-erable. Such conditions result in glare with allheadlamps, but the greater flux from HID sourceswill produce more glare when the lamps are dirty,and their greater illumination on some beamangles will result in more glare if they are mis-aimed. If HID lamps are kept aimed and clean,they are likely to produce a net gain in perform-ance and visibility with minimum discomfort; butif such steps are not taken, their removal fromU.S. highways may be an appropriate countermea-sure. This point was clearly made by Schoon &Schreuder (1993).

Still, the proper operation or removal of HIDlamps will not solve all problems of headlightglare, nor would universal participation in olderdriver training courses. The design and selectionof countermeasures for headlight glare are includ-ed in the basic tradeoffs that are incorporated inthe design of headlamps. As discussed in Chapter5, the U.S. standard design, compared to theEuropean standard, favors visibility over glare.The U.S. standard, which began as a conceptualexercise, has, over time, determined the design oftraffic control devices. Many devices have beendesigned under the assumption that they willreceive minimum levels of illuminance from head-

lights at the same locations in the beam patternthat also illuminate mirrors or the eyes of driversin oncoming vehicles. In this way, the highwayinfrastructure places restrictions on any manipula-tion of the beam pattern to reduce glare. Eachcountermeasure discussed in this report offers lim-ited benefits in special situations, but, as statedearlier, only polarization provides a global solu-tion. We hope that research in that area will goforward with as much dedication as is beingdevoted to the research and development of UVAand of adaptive headlighting.

It is not possible to recommend one counter-measure that would eliminate discomfort glare foreveryone in all situations without unacceptablenegative consequences for visibility and safety.However, a number of the countermeasures dis-cussed can be implemented with minimal cost orwith costs offset by benefits other than glarereduction. Fixed roadway lighting, glare screens,and wide medians are, in certain situations, cost-effective methods of accident reduction regardlessof their effect on adaptation and glare. It wouldmake the most sense to start simply by initiatingthose strategies that can be implemented with min-imal or no cost. Maintenance of corrective lensesor ophthalmic surgery, remedial driver education,and self-imposed restrictions on night driving offermany safety advantages beyond their effects onglare. Some limitation on headlamp height, cou-pled with the use of anti-glare mirrors and bettermaintenance of headlamp aim, would eliminatemost, if not all, complaints of mirror glare.Finally, UVA and adaptive headlighting still needto be pursued, but their benefits and costs willnot be known until these approaches are fullydeveloped. At the present time, UVA and adap-tive headlighting appear to offer greater promisefor improving visibility than for reducing glare,but more development and further evaluation willgive a better picture of the benefits. Polarizedlighting seems to offer the most promise for elim-inating glare, but the obstacles to implementationmay prevent this technology from becoming reality.

97

REFERENCES

AASHTO (1994). A policy on geometric design of highways and streets, American Association of StateHighway and Transportation Officials, Washington, DC pp. 136–137.

Adler, B. and Lunenfeld, H. (1973). “Three-beam headlight evaluation.” Report no. HS-238-411-1,Airborne Instruments Laboratory, Deer Park, NY.

Adrian, W. K. (1975). “Grundlagen der physiologischen und psychologischen Blendung und ihrenumerische Darstellung” (Fundamentals of disability and discomfort glare and their numerical represen-tation). Lichttechnik 27, 312–319.

Adrian, W. K. (1979). “Ueber eine neuartige Blendschutzbrille” (On a new anti-glare spectacle). 30.Sonderderuck der WVAO, Mainz. Die Fachvortrage des WVAO - Jahres Kongresser 1979 in Karlsruhe(ISSN 0171-6417), pp. 243–251.

Adrian, W. K. (1991a). “Transient adaptation process: A model to predict its effects on vision.” CIE22nd Session, 2, 121–133. CIE transactions of the International Commission on Illumination, 22nd ses-sion. Melbourne 1991, Vol 2, pp 121-133.

Adrian, W. K. (1991b). “Comparison between the CBE and the CIE glare mark formula and earlier dis-comfort glare descriptions.” Proceedings of the First International Symposium on Glare, Orlando,Florida, pp. 155–164.

Adrian, W. (1997). “A method to predict visual function and visual performance in mesopic lighting lev-els.” University of Waterloo, Ontario, Canada.

Adrian, W. and Bhanji, A. (1991). “Fundamentals of disability glare. A formula to describe stray light inthe eye as a function of the glare angle and age.” Proceedings of the First International Symposium onGlare, Orlando, Florida, pp. 185–194.

Alexander, G. J. and Lunenfeld, H. (1990). “A users’ guide to positive guidance, third ed.” Report No.FHWA-SA-90-017, U.S. Department of Transportation, Federal Highway Administration, Washington,DC pp. 1–39.

Allen, M. J. (1985). “Vision of the older driver: implications for vehicle and highway design and fordriver testing.” 55+, 1–7.

Anderson, R. M. (1983). “Visual perceptions and observations of an aphakic surgeon.” Perceptual andMotor Skills, 57, 1211–1218.

Applegate, R. A. (1991). “Acuities through annular and central pupils after radial keratotomy (RK).”Optometry and Vision Science, 68, 584–590.

Applegate, R. A. and Gansel, K. A. (1990). “The importance of pupil size in optical quality measure-ments following radial keratotomy (RK).” Refractive and Corneal Surgery, 6, 47–54.

99

Applegate R. A., Hilmantel, G., and Howland, H. C. (1996). “Corneal aberrations increase with the mag-nitude of radial keratotomy (RK) refractive correction.” Optometry and Vision Science, 73, 585-589.

Applegate, R. A., Trick, L. R., Meade, D. L., and Hartstein, J. (1987). “Radial keratotomy (RK) increas-es the effects of disability glare: Initial results.” Annals of Ophthalmology, 19, 293–297.

ASCRS (2000). American Society of Cataract and Refractive Surgery, “Comments on FDA’s draftintraocular lens guidance document,” available at http://www.ascrs.org/advocacy/fdaioljan00.htm.

Bailey, I. L. and Sheedy, J. E., (1986). “Vision screening for driver licensure.” Transportation in anAging Society, pp. 294–323. Transportation Research Board, National Research Council, Washington,DC 1988.

Bennett, C. A. (1977). “The demographic variables of discomfort glare.” Lighting Design andApplication, 22–24. January, 1977.

Berman, S. M. (1992). “Energy efficiency consequences of scotopic sensitivity.” Journal of theIlluminating Engineering Society, 21(1). Volume 21, 3-14.

Berman, S. M., Fein, G., Jewett, D. L., and Ashford, F. (1993). “Luminance controlled pupil size affectsLandolt G test performance.” Journal of the Illuminating Engineering Society, 22, 150–165.

Berman, S. M., Fein, G., Jewett, D. L., and Ashford, F. (1994). “Landolt C recognition in elderly sub-jects is affected by scotopic intensity of surround illuminants.” Journal of the Illuminating Society, 23,123–130.

Berman, S. M., Fein, G., Jewett, D. L., Benson, B. R., Law, T. M., and Myers, A. W. (1996). “Lightingspectral effects on Landolt C performance is abolished by mydriasis.” Journal of the IlluminatingSociety, 25, 42-50.

Berman, S. M., Jewett, D. L., Benson, B. R., and Law, T. M. (1997). “Despite different wall colors, ver-tical scotopic illuminance predicts pupil size.” Journal of the Illuminating Engineering Society, 26,59–68. Volume 26, Summer, 1997.

Bhise, V. D., Farber, E. I., Saunby, C. S., Troell, G.M., Walunas, J. B., and Bernstein, A. (1977).“Modeling vision with headlights in a systems context.” Society of Automotive Engineers, InternationalAutomotive Engineering Congress and Exposition. Cobo Hall, Detroit. February 28-March 4, 1977.

Bhise, V. D., Matle, C. C., and Hoffmeister, D. H. (1984). “CHESS model application in headlamp sys-tems evaluation.” SAE Technical Paper Series #84046, Society of Automotive Engineers, Warrendale,PA.

Billion, C. E. and Parsons, N. C. (1962). “Median Accident Study—Long Island, New York.” Bulletin308, Highway Research Board, National Research Council, Washington, DC, pp. 64–79.

Blackwell, H. R. (1954). “Visual detection at low luminance through optical filters.” Highway ResearchBoard Bulletin No. 89, National Academy of Sciences, Washington, DC, pp. 43–61.

100

Bouma, P. J. (1938). “The colour reproduction of incandescent lamps and ‘Philiphane Glass’.” PhilipsTechnical Review, 3, 27–29.

Box, P. C. (1973). “Freeway accidents and illumination.” Highway Research Record, 416, 10– 20.

Boyce, P. R. (1981). Human Factors In Lighting. MacMillan, New York, NY.

Boyce, P. R. and Beckstead, J. W. (1991). “Influences on discomfort glare ratings.” Proceedings of theFirst International Symposium on Glare, Orlando, Florida,pp. 93–100.

Boyce, P. and Bruno, L. (1998). “An evaluation of high pressure sodium and metal halide light sourcesfor parking lot lighting.” Proceedings of the IESNA, San Antonio. Illumination Engineering Society ofNorth America

Bryan, W. E. (1962). “Lenses for night driving.” Highway Research Board Bulletin No. 336, NationalAcademy of Sciences, Washington, DC, pg. 104.

Caden, D. G. and Hemion, R. H. (1969). “Disability glare effects from vehicle headlights and fixedoverhead street lighting.” Final Report on Phase IV-B of FHWA Contract CPR 11-4126, Report No. AR-700, Southwest Research Institute, San Antonio, TX. p. 1-23.

Chu, X. (1994). “The Effects of age on the driving habits of the elderly. Evidence from the 1990National Transportation Study” Contract No. DOT-T-95-12, U.S. Department of Transportation Researchand Special Programs Administration, Washington, D.C.

Consumer Reports (1999). “Zap your myopic eyes? Laser surgery is booming—but pitfalls persist.”Consumer Reports, June 1999, pp. 56–57.

Copenhaver, M. J. and Jones Jr., R. E. (1992). “Measurement of headlamp aim and the electrical andphotometric performance characteristics of rear lighting systems.” DOT HS 807 930, U.S. Departmentof Transportation, National Highway Traffic Safety Administration, Washington, DC.

Crawford, B. H. (1936). “The integration of the effects from a number of glare sources.” Proc. Phys.Soc. London 48. p. 35.

Crosby. J. R. (1960). “Cross-median accident experience on the New Jersey Turnpike.” Bulletin 266,National Research Council, Highway Research Board, Washington, DC, pp. 63–77.

Daily Apple (1999). Available at http://thedailyapple.com/.

DeBoer, J. B. and Schreuder, D. A. (1967). Transportation Illuminating Engineering Society (London),32, 117.

DHHS (2000). “Acquisition cost of prosthetic intraocular lenses.” U.S. Department of Health andHuman Services, Office of Evaluation and Inspections, Washington, DC. Available athttp.//www.dhhs.gov/.

Duncan, D. D. (1996). “Nighttime glare reduction and illumination enhancement.” Report No. FS-96-037, Johns Hopkins Applied Physics Laboratory, Fleet Systems Department, pp. 1–35.

101

Farber, E., Gallagher, V., and Cassel, A. (1971). “Interaction between fixed and vehicular illuminationsystems—Phase I Report,” FHWA Contract No. FH11-7599, U.S. Department of Transportation, FederalHighway Administration, Washington, DC.

Fatality Analysis Reporting System (FARS). National Highway Traffic Safety Administration, NationalCenter for Statistics and Analysis. Available at http://www.nhtsa.dot.gov/people/ncsa/.

Fast, P. T. and Ricksand, A. (1994). “Safer nighttime driving, UV headlamps improve visibility of pedes-trians.” Paper No. 94-0908, Presentation at the Transportation Reseach Board 73rd Annual Meeting,,Washington, DC. January, 1994. Transportation Research Board, Washington, D.C.

Federal Register (1995). “Lamps, reflective devices and associated equipment; establishment of negoti-ated rulemaking advisory committee.” 49 CFR Part 571, National Highway Traffic SafetyAdministration, Washington, DC. Vol. 60, No. 111, Friday, June 9, 1995.

Federal Register (1997). “Federal motor vehicle safety standards; lamps, reflective devices and associat-ed equipment.” 49 CFR Part 571, National Highway Traffic Safety Administration, Washington, DC.Vol. 62, No. 46, Monday, March 10, 1997.

Finsterwalder, Josef (1991). Untersuchungen zur Physiologischen blendung bei sehr KleinenBlendwinkeln. (“Investigations on Disability Glare at Very Small Glare Angles”). Ph.D. thesis at theFaculty of Physics of the Ludwig-Maxililian University, Munich, Germany.

Flannagan, M. (2000). Michael Flannagan, Personal Communication, October 2000.

Flannagan, M. J., Gellatly, M. J., Luoma, J., and Sivak, M. (1992). “A field study of discomfort glarefrom high-intensity discharge headlamps.” Report No. HS-041 319, UMTRI-92-16, University ofMichigan Transportation Research Institute, Ann Arbor, MI.

Flannagan, M. and Sivak, M. (1990). “Nighttime effectiveness of rearview mirrors: Driver attitudes andbehaviors.” SAE Technical Paper Series #900567, Society of Automotive Engineers, Warrendale, PA.

Flannagan, M., Sivak, M., and Gellatly, A. W. (1991). “Rearward vision, driver confidence, and discom-fort glare using an electrochromic rearview mirror.” SAE Technical Paper Series #910822, Society ofAutomotive Engineers, Warrendale, PA.

Flannagan, M.J., Gellatly, M.J., Luoma, J., and Sivak, M. (1992). A field study of discomfort glare fromhigh-intensity discharge headlamps. University of Michigan Transportation Research Institute, AnnArbor. Industry Affiliation Program for Human Factors in Transportation Safety. Report No. UMTRI-92-16. UMTRI-82468.

Flannagan, M. J., Sivak, M., and Traube, E. C. (1994). “Discomfort glare and brightness as a function ofwavelength.” Report UMTRI-94-29, University of Michigan Transportation Research Institute, AnnArbor, MI.

Flannagan, M. J., Sivak, M., Traube, E. C., and Kojima, S. (1996). “Effects of overall low-beam intensi-ty on seeing distance in the presence of glare.” Report UMTRI-96-26, University of MichiganTransportation Research Institute, Ann Arbor, MI., pp. 1–26.

102

Fisher, A. J. (1970). “A basis for a vehicle headlighting specification.” IES Lighting Review 32, 1–9.

Fry, G. A. (1953). Presented to the light and vision conference, June 1953, University of Michigan.Journal of the Optical Society of America 43(3), 189-195.

Fry, G. A. (1954). “Evaluating disabling effects of approaching automobile headlights.” HighwayResearch Bulletin 89, Night Visibility, pp. 38–42. Highway Research Board Bulletin, 89, 38-42.

Fry, G. A. and King, V. M. (1975). “The pupillary response and discomfort glare.” Journal of theIlluminating Engineering Society 4, 307.

Fugate, J. M. and Fry, G. A. (1956). “Relation of changes in pupil size to visual discomfort.”Illumination Engineering L1, 537-549.

Garner, G. R. and Deen, R. C. (1973). “Elements of median design in relation to accident occurrence.”In Highway Research Record 432, National Research Council, Highway Research Board, Washington,DC, pp. 1–11.

Garvey, P. M. (1994). “Potential adverse effects of UV automobile headlamps on children, aphakes, andnon-UV absorbing pseudophakes.” Grant Submission to the National Institutes of Health, National EyeInstitute, Bethesda, MD (personal communication).

Ghaith, A. A., Stulting, D. J., Thompson, K. P., and Lynn, M. (1998). “Contrast sensitivity and glare dis-ability after radial keratotomy and photorefractive keratectomy.” Archives of Ophthalmology 116, 12-18.

Gouras, P. and Zrenner, E. (1981). “Color Vision: A review from a neurophysiological perspective.” InProgress in Sensory Physiology 1, Springer, NY, pp. 139–179.

Ham, W. T., Mueller, H. A., Ruffolo, J. J., and Guerry, D. (1980). “Solar retinopathy as a function ofwavelength: Its significance for protective eyewear.” In The effects of constant light on visual processes,T. P. Willams and B. N. Baker, eds., Plenum, NY, pp. 319–346.

Hammer, H. M., Yap, M., and Weatherill, J. R. (1986). “Visual performance in pseudophakia with standard ultraviolet-absorbing intraocular lenses: A preliminary report.” Transactions of theOphthalmological Society of the United Kingdom, 105, 441–446.

Haney, J. E. and Mortimer, R. G. (1974). “Toward more effective headlighting.” Report No. UM-HSRI-RI-74-2, University of Michigan Highway Safety Research Institute, Ann Arbor, MI.

Hare, C. T. and Hemion, R. H. (1968). “Headlamp beam usage on U.S. highways.” Report No. AR-666,Contract No. CPR11-4126, U.S. Department of Transportation, Federal Highway Administration Bureauof Public Roads, pp. 1–82.

Hartmann, E. and Moser, E. A. (1968). “Das Gesetz der physiologischen Blendung bei sehr kleinenBlendwinkeln.” Lichttechnik 10, 67A.

Helmers, G., Ytterbom, U., and Lundkvist, S. O. (1993). “Detection distances to obstacles on the roadwhen using UV headlights.” Statens Vaeg-och Transportforskningsinstitut, Stockholm, Sweden.

103

Hemion, R. H. (1969a). “A preliminary cost-benefit study of headlight glare reduction. Final Report onPhase V of a Study for the Bureau of Public Roads, Federal Highway Administration, Contract No. CPR11-4126, SWRI Project No. 11-1908.” Southwest Research Institute, San Antonio, TX.

Hemion, R. G. (1969b). “Preliminary site selection for public test of polarized headlighting. Report onPhase VI of a Study for the U.S. Bureau of Public Roads.” Report No. AR-692, Southwest ResearchInstitute, San Antonio, TX.

Hemion, R. H., Hull, R. W., Cadena, D. G., and Dial, B. C. (1971). “Development of plans for publicevaluation of polarized headlights.” Final Report No. AR-815, Contract No. FH 11-7349, U.S.Department of Transportation, Federal Highway Administration, Washington, DC, pp. 1–110.

Hogrefe, H. and Neumann, R. (1997). “Adaptive light pattern—A new way to improve light quality.”SAE Technical Paper Series #970644, Society of Automotive Engineers, Warrendale, PA.

Holladay, L. L. (1926). “The fundamentals of glare and visibility.” Journal of the Optical Society ofAmerica, 12, 271–319.

Holladay, J. T., Lang, A., and Partney, V. (1999). “Analysis of edge glare phenomena in intraocular lensedge designs.” Journal of Cataract and Refractive Surgery, 25. Available at http.//www.ascrs.org/publi-cations/jcrs/absjun9.htm.

Hopkinson, R. G. (1963). Architectural Physics: Lighting. Her Majesty’s Stationary Office, London, UK.

Hunt, J. H., Land, E. H., and Roper, V. J. (1948). “The polarized headlight system.” HRB Bulletin 11,National Research Council, Highway Research Board, Washington, DC.

Hurd, F. W. (1957). “Accident experience with traversable medians of different widths.” Bulletin 137,National Research Council, Highway Research Board, Washington, D.C., pp. 18–26.

Hutchinson, J. W., Kennedy, T. W., and Scott, W. A. (1963). “Medians of divided highways: Current sta-tus of knowledge and annotated bibliography.” Highway Research Board Bibliography, 34, 2–5.

Ijspeert, J. K., et al. (1990). “The intraocular straylight function in 129 healthy volunteers; Dependenceon angle, age and pigmentation.” Vision Research, 30, 699-707.

Institute for Road Safety Research (1976). “White or yellow lights for vehicle headlamps?” Institute forRoad Safety Research Report, Leidschendam, Netherlands.

Janoff, M. S. (1999). “The effect of lamp color on driver visual performance: A summary of recentresearch.” Roadway Lighting Committee Papers Symposium, Birmingham, AL.

Janoff, M. and Havard, J. (1997). “The effect of color on the visibility of small targets.” Journal of theIlluminating Society, 26. Vol. 26, No. 1, 173-181.

Johansson, G., Bergstörm, S.-S., Jansson, G., Ottander, C., Rumar, K., and Ornbergt, G. (1963). “Visible distances in simulated night driving conditions with full and dipped headlights.” Ergonomics 6,171–179.

104

Karpen, D. (1996). “Color corrected motor vehicle headlight.” U.S. Patent No. 5,548,491.

Karpen, D. (1998a). “Motor vehicle rearview mirror.” U.S. Patent No. 5,844,721.

Karpen, D. (1998b). “The road less traveled.” LD+A, 47–50. Journal, July 1998.

Karpen, D. (1999). “ Color corrected high intensity discharge motor vehicle headlight.” U. S. Patent#5,961,208, pp. 1–14.

Karpen, D. (2001). “Neodymium oxide doped headlight lamps.” Report No. 2001-01-0319, Society ofAutomotive Engineers, Detroit, MI, pp. 1–7.

Kinsey, V. E. (1948). “Spectral transmission of the eye to ultraviolet radiation.” Archives ofOphthalmology 39, 508– 513.

Knuiman M. W., Council, F. M., and Reinfurt, D.W. (1993). “Association of median width and highwayaccident rates.” In Transportation Research Record 1401, National Research Council TransportationResearch Board, Washington, DC, pp. 70–-80.

Kobayashi, S., Takahashi, K., and Yagi, S. (1997). “Development of new forward lighting systems withcontrollable beams.” SAE Technical Paper Series #970646, Society of Automotive Engineers,Warrendale, PA.

Kozak, E. M. (1996). “Feasibility of UV-activated fluorescent materials.” Draft Final Report, FederalHighway Administration, McLean, VA.

Lauer, A. R. (1951). “Filter study of the effect of certain transmission filters on visual acuity with andwithout glare.” Highway Research Board Bulletin No. 43, National Academy of Sciences, Washington,DC, pp. 45–51

Leibowitz, H. W. and Owens, D .A. (1978). “New evidence for the intermediate position of relaxedaccommodation.” Documenta Opnthamologica, 46, 133–147.

Lerman, S. (1983). “An experimental and clinical evaluation of lens transparency and aging.” Journal ofGerontology, 38(3), 293–301.

Lewis, A. (1997). “Visual performance as a function of spectral distribution of light sources at illumi-nances used for outdoor lighting.” Journal of the Illuminating Society, in press.

Lewis, A. (1998). “Equating light sources for visual performance at low luminances.” Journal of theIlluminating Society, 27. p. 80-84.

Lindae, G. (1970). “Optimale Scheinwerfergrosse fur europaisches Abblendlicht” (Optimum size ofheadlamps for the European low beam). ATZ Automobiltechnische Zeitschrift, 72, 427– 431.

Lulla, A. B. and Bennett, C. A. (1981). “Discomfort glare: Range effects.” Journal of IlluminatingEngineering Society, 10, 74-80. January, 1981.

105

Manassero, G., Paolini, A., Battaglino, S., Bigliati, C., and Sinesi, S. (1999). “Adaptive lighting systems:Technical solutions and a methodological approach for photometric specifications.” progress in automo-bile lighting. Department of Lighting Technology. 514–525. Volume 6, Progression AutomobileLighting, Department of Lighting Technology, Published by Prof. Dr. Ing. H.-J. Schmidt-Clausen,Darmstadt University of Technology.

Mehra, M. et al. (1998). “A field study of visual acuity, color naming and visual search under mesopiclighting.” Proceedings of the IESNA, San Antonio, TX.

Miller, D. and Lazenby, G.W. (1977). “Glare sensitivity in corrected aphakes.” Ophthalmic Surgery, 8,54–57.

Miller, N. D., Baumgardner, D., and Mortimer, R. G. (1974). “An evaluation of glare in nighttime driv-ing caused by headlights reflected from rearview mirrors.” Report No. 740962, Automobile EngineeringMeeting, Society of Automotive Engineers, Warrendale, Automobile Engineering Meeting, Oct. 21-25,1974.

Mortimer, R G. (1974). “Some effects of road, truck and headlamp characteristics on visibility and glarein night driving.” Report No. 740615, Society of Automotive Engineers, Warrendale, PA.

Mortimer, R. G. (1988). “Headlamp performance factors affecting the visibility of older drivers in nightdriving.” In Transportation in an Aging Society: Improving Mobility and Safety for Older Persons,Special Report 218, Vol. 2, pp. 379–403. Committee for the Study on Improving Mobility and Safetyfor Older Persons, Transportation Research Board, Washington, DC.

National Safety Council (1988). Accident Facts, National Safety Council, Washington, DC pp. 49–58.

National Safety Council (1990). Accident Facts, National Safety Council, Washington, DC, pg. 60.

National Safety Council (1994). Accident Facts, National Safety Council, Washington, DC, pg. 64.

National Safety Council (1999). Injury Facts, National Safety Council, Washington, DC, pg. 90.

National Transportation Statistics (1999). “National Transportation Statistics.” U.S. Department ofTransportation. Available at http://www.bts.gov/ntda/nts/NTS99/data/Chapter1/1-29.

NCHRP Report 77 (1979). Page 19.

NHTSA (1987). Fatal accident reporting system, HS 807 360. U.S. Department of Transportation,National Highway Traffic Safety Administration, Washington, DC

Nitzburg, M., Seifert, R., Knoblauch, R., and Turner, D. (1998). “A safety evaluation of UVA vehicleheadlights.” Report No. FHWA-RD-99-079, Center for Applied Research, Turner Fairbank HighwayResearch Center, McLean, VA. Available at http://www.tfhrc.gov/library/UVA/introset.htm.

NOAH (1999). Available at http://www.albinism.org/.

Olson, P. L. (1985). “A survey of the condition of lighting equipment on vehicles in the United States.”SAE Technical Paper Series #850229, Society of Automotive Engineers, Warrendale, PA., pp. 1–14.

106

Olson, P. L. and Sivak, M. (1984a). “Glare from automobile rear-vision mirrors.” Human Factors 26,269–282.

Olson, P. L. and Sivak, M. (1984b). “Glare and headlighting design.” Report No. 840047, Society ofAutomotive Engineers, Warrendale, PA., pp. 1-9.

Owens, D. A. and Sivak, M. (1996). “Differentiation of visibility and alcohol as contributors to twilightroad fatalities.” Human Factors 38, 680–689, page 20 under ‘Effect of Glare on Accident Frequency.’

Owsley, C, Sloane, M. E., Stalvey, B, and Wells, J. ([YEAR?]). “Older drivers and cataract: drivinghabits and crash risk.” Journals of Gerontology A: Biological and Medical Sciences. 54, 203–211.

Peet, J. S. and Neuzil, D. (1972). “Independent versus narrow-median alignment: Comparative economy,safety, and aesthetics.” Highway Research Record 390, National Research Council, TransportationResearch Board, Washington, DC, pp. 1–14.

Pitts, D. G. (1990). “Ultraviolet-absorbing spectacle lenses, contact lenses, and intraocular lenses.”Optometry and Vision Science, 67, 435–440.

Powers, L. D. and Solomon, D. (1965). “Headlight glare and median width—Three exploratory studies.”Public Roads, 33(7), 125–242.

Pulling, N. H., Wolf, E., Sturgis, S. P., Vaillancourt, D. R., and Dolliver, J .J. (1980). “Headlight glareresistance and driver age.” Human Factors, 22, 103–112.

Rea, M. (1990). “Essay by invitation.” LD&A, October.

Rog, S. J., White, C. W., and Williams, T. T. (1986). “Ultraviolet effects on visual acuity inpseudophakia.” American Journal of Optometry and Physiological Optics, 63, 867–872.

Rösenhahn, E. O. (1999). “Headlamp components for adaptive frontlighting—Usability of a lightingfunction for adverse weather conditions.” Progress in Automobile Lighting, Department of LightingTechnology. Proceedings of the Conference, pp. 677–688.

RP-8 (2000). American National Standard Practice for Roadway Lighting, ANSI/IESNA RP-8-00(ISBN#0-87995-160-5), pp. 1–61. The Standard Practice Subcommittee of the IESNA RoadwayLighting Committee, 120 Wall Street, New York, NY 10005.

Russell, E. R., Rys, M., Rys, A., and Keck, M. (1999). “Characteristics and needs for overhead guidesign illumination from vehicular headlamps.” Report No. FHWA-RD-98-135, Kansas State UniversityDepartment of Civil Engineering, pp. 1–146.

SAE Draft Technical Report (1996). “Recommendations of the SAE task force on headlamp mountingheight, SAE J 2338.” SAE Ground Vehicle Lighting Standards Manual, 2001 Edition, Society ofAutomotive Engineers, Warrendale, PA, pp. 113-116.

San Francisco Chronicle (June 28, 1999). “Bright headlights A glaring problem on roadways.”

107

Schieber, F. (1986). “Vision assessment technology and screening older drivers: past practices andemerging techniques.” Transportation in an Aging Society, pp. 325–378. Special Report 218,Committee for the study on Improving Mobility and Safety for Older Persons, Transportation ResearchBoard, Washington, DC.

Schiflett, S. G., Cadena, D.G., and Hemion, R. H. (1969). “Headlight glare effects on driver fatigue.Report on Phase II of a Study for U.S. Bureau of Public Roads.” Southwest Research Institute, SanAntonio, TX.

Schmidt-Clausen, H. J. and Bindels, J. T. H. (1974). “Assessment of discomfort glare in motor vehiclelighting.” Lighting Research and Technology, 6(2), 79–88.

Schoon, C. C. and Schreuder, D. A. (1993). “HID car headlights and road safety.” SWOV Institute forRoad Safety Research, pp. 1–33. SWOV Institute for Road Safety Research, Leidschendam, TheNetherlands.

Schwab, R. N. (2000). Personnal communication, October 2000.

Schwab, R. N. and Hemion, R. H. (1971). “ Improvement of visibility for night driving.” HRR No. 377,Visibility 11 Report, pp. 1-18. Transportation Research Board, Washington DC.

Schwab, R. N., Solomon, D., and Lyons, J. F. (1972). “Monetary value drivers place on comfort in nightdriving.” Report No. FHWA-RD-72-14, Environmental Design and Control Division, U.S. Departmentof Transportation, Federal Highway Administration, Washington, DC pp. 1–28.Sivak, M. (1996). “The information that drivers use: Is it indeed 90% visual?” Perception 25,1081–1089.

Sivak, M., Flannagan, M. J., Budnik, E. A., Flannagan, E. C., and Kojima, S. (1997). “The location ofheadlamps and driver eye positions in vehicles sold in the USA.” Ergonomics, 40, 872–878.

Sivak, M., Flannagan, M., Ensing, M., and Simmons, C. J. (1991) “Discomfort glare is task dependent.”International Journal of Vehicle Design, 12(2), 152-159.

Sivak, M., Flannagan, M., and Gellatly, A. W. (1993). “Influence of truck driver eye position on effec-tiveness of retroreflective traffic signs.” Lighting Research Technology, 25, 31-36.

Sivak, M., Flannagan, M. J., and Sato, T. (1993). “The relative importance of horizontal and vertical aimof low-beam headlamps.” Report No. UMTRI-93-26 University of Michigan Transportation ResearchInstitute, Ann Arbor, MI, pp. 1–36.

Sivak, M., Flannagan, M. J., Traube, E. C., Aoki, M., and Sayer, J. R. (1994). “Evaluation of an activeheadlight system.” Report No. UMTRI-94-17, University of Michigan Transportation Research Institute,pp. 1–30.

Sivak, M., Flannagan, M. J., Traube, E. C., and Kojima, S. (1999). “The influence of stimulus durationon discomfort glare for persons with and without visual correction.” Transportation Human Factors, 1,147–158.

108

Sivak, M., Flannagan, M. J., Traube, E. C., Kojima, S. (1997). “The influence of stimulus duration ondiscomfort glare for persons with and without visual correction.” Report No. UMTRI-97-23, Universityof Michigan Transportation Research Institute.

Sivak, M., Helmers, G., Owens, D.A., Flannagan, M. (1992). Evaluation of proposed low-beam head-lighting patterns. University of Michigan Transportation Research Institute, Ann Arbor. Sponsored bythe Motor Vehicle Manufacturers Association, Detroit, MI. Report No. UMTRI-92-14.

Sivak, M., Olson, P. L., and Zeltner, K. A. (1989). “Effect of prior headlighting experience on ratings ofdiscomfort glare.” Human Factors, 31, 391– 395.

Sivak, M., Simmons, C. J., and Flannagan, M. (1990). “Effect of headlamp area on discomfort glare.”Lighting Research Technology, 22, 49–52.

Sliney, D. H., et al. (1984). Ultraviolet Countermeasures.

Sliney, D. H., Fast, P., and Ricksand, A. (1994). “Analysis of potential health hazards of ultravioletheadlamps.” Presented at the Transportation Research Board 73rd Annual Meeting, Washington, DC.

Smart, A. E. (1970). “An investigation of chromatic glare with particular reference to night driving.”Lighting Research and Technology, 2, 33–37.

Stark, W. J., Whitney, C. E., Chandler, J. W., and Worthen, D. M. (1986). “Trends in intraocular lensimplantation in the United States.” Archives of Ophthalmology, 104, 1769–1770.

Steinert, R. F. (1997). “Consultation section: cataract surgical problem.” Journal of Cataract andRefractive Surgery, 23. Available at http://www.ascrs.org/publications/jcrs/ccsoct7.htm1#steinert.

Stiles, W. S. (1928–29). “The effect of glare on the brightness difference threshold.” Proceedings of theRoyal Society (London) B104, 322.

Sullivan, J. M. and Flannagan, M. J. (2001). “Visual effects of blue-tinted tungsten-halogen headlampbulbs.” Report No. UMTRI-2001-9, University of Michigan Transportation Research Institute, pp. 1–24.

Telford, E. T. and Israel, R.J. (1953). “Median study (California).” Proceedings of the 32nd AnnualMeeting, Highway Research Board, pp. 208–231.

Theeuwes, J. and Alferdiock, J. W. A. M. (1996). “The relation between discomfort glare and drivingbehavior.” Report No. DOT HS 808 452, U.S. Department of Transportation, National Highway TrafficSafety Administration, Washington, DC, pp. 1–64.

Transportation Research Board, 1988. “Transportation in an aging society. Improving mobility and safe-ty for older persons.” Special Report 218, Volume 2, chapter on headlamp performance factors affectingthe visibility of older drivers in night driving. National Research Council, Transportation ResearchBoard, pp. 379–403.

Ucke, C. (1973). Streulicht im Auge (Stray light in the eye). Ph.D. thesis, University of Munich, October1973, Germany.

109

Ultralux (1994). Ultralux AB brochure, Swedish National Road Administration.

USHHS (1994). U.S. Department of Health and Human Services, personal communication, February,1994.

USHHS (2001). U.S. Department of Health and Human Services, personal communication, May, 2001.

Vos, J. J. and Padmos, P. (1983). “Stray light, contrast sensitivity and the critical object in relation totunnel entrance lighting.” DIE transactions of the International Commission on Illumination, 20th ses-sion.

Wald, G. (1945). “Human vision and the spectrum.” Science 101, 653–658.

Werner, J. S. and Hardenbergh, F. E. (1983). “Spectral sensitivity of the pseudophakic eye.” Archives ofOphthalmology, 101, 758–760.

Weston. H. C. (1954). “Visual fatigue.” Illuminating Engineering, 49, 63–76. Also in Selected Papers onArchitectural Lighting, M. S. Rea, ed. (1992), pp. 286–297.

White, C. W. and Wolbarsht, M. L. (1975). “Color vision in the ultraviolet” (abstract). Journal of theOptical Society of America, 65, 1119.

Winter, D. J. (1985). “Learning and motivational characteristics of older people pertaining to traffic safe-ty.” 55+, 77–85.

Wörner, B. (1999). “Adaptive frontlighting—Experimental system development and functional evalua-tion.” Progress in Automobile Lighting, Department of Lighting Technology, Proceedings of theConference.

Zigman, S. (1983). “Effects of near ultraviolet radiation on the lens and retina.” DocumentaOphthalmologica, 55, 375–391.

110

Headlight glare

Documents

Transcript of Headlight glare