title: legibility distances of fluorescent traffic...
TRANSCRIPT
Paper Number: 01-2417
Duplication for publication or sale is strictly prohibited without prior written permission of the
Transportation Research Board
Title: LEGIBILITY DISTANCES OF FLUORESCENT TRAFFIC SIGNS AND
THEIR NORMAL COLOR COUNTERPARTS
Authors: Thomas Schnell, Keith Bentley, Elizabeth Hayes, and Martin
Rick
Transportation Research Board 80th Annual Meeting January 7-11, 2001
Washington, DC
Schnell, Bentley, Hayes, and Rick 2
LEGIBILITY DISTANCES OF FLUORESCENT TRAFFIC SIGNS AND THEIR NORMAL
COLOR COUNTERPARTS
Thomas Schnell, Ph.D.
Assistant Professor and Director
Operator Performance Laboratory (OPL)
Department of Industrial Engineering
4135 Seamans Center
The University of Iowa
Iowa City, Iowa 52242-2527
(319) 384 0811
and
Keith Bentley, Elizabeth Hayes, and Martin Rick
Undergraduate Students
Department of Industrial Engineering
The University of Iowa
Iowa City, Iowa 52242-1527
Transportation Research Board 80th Annual Meeting January 7-11, 2001
Washington, DC
Schnell, Bentley, Hayes, and Rick 3
ABSTRACT
It is well known that fluorescent traffic sign have a daytime conspicuity that is far superior to that of normal
color signs. This study was conducted in order to test the hypothesis if adding the property of fluorescence
while leaving all (or in practical terms most) other sign features the same, does in fact increase the
legibility distance. Six diamond shaped signs were fabricated and laminated with the normal colors on one
side and their fluorescent color counterparts on the other side. The investigated sign background colors
included yellow, fluorescent yellow, fluorescent yellow-green, pink, fluorescent pink, purple, and
fluorescent purple. Four of the six signs used Landolt rings to avoid word familiarity issues. A pedestrian
crossing symbol sign and a PLANT ENTRANCE warning sign were used in addition to the Landolt ring
signs. It was found that fluorescence actually did statistically significantly increase the legibility distances.
However, the actual increases were fairly small from a practical point of view, ranging from 5.3% to 15.9%
when adding the property of fluorescence. In general, for negative contrast signs and for positive contrast
signs where both the legend and the background are fluorescent, a slight increase in the legibility distance
can be expected. It should be noted, that the illumination conditions that were present in this study (fair
day) represent somewhat of a worst-case scenario for fluorescent colors. Larger percent increases in
legibility distance may be expected when using fluorescent colors during the time of twilight, during
overcast days, and during daytime in inclement weather.
Schnell, Bentley, Hayes, and Rick 4
INTRODUCTION
Fluorescent colors in traffic sign sheeting materials represent a fairly new method for attracting driver
attention and for enhancing legibility of text or recognition of symbols. Substantial previous research (see
below in literature review) indicated that fluorescent colors are far more attention getting than their normal
color counterparts. The attention getting quality of fluorescent colors is especially pronounced during
overcast or inclement weather daytime hours, and during the time of twilight. Also, fluorescent colors are
far better detected and recognized (color classified) than normal colors at large angles of eccentricity in
the visual periphery. In 1998, the Federal Highway Administration (FHWA) amended The Manual on
Uniform Traffic Control Devices (MUTCD) [1] to approve the optional use of the color fluorescent yellow
green (FYG) for pedestrian, school, and bicycle crossing signs.
The present study was conducted to determine the legibility distances of traffic signs laminated with
selected fluorescent color sheeting materials and their normal color counterparts. The goal was to
determine the legibility distance of traffic signs with and without adding the property of fluorescence. The
investigated sign background colors included yellow, fluorescent yellow, fluorescent yellow-green, pink,
fluorescent pink, purple, and fluorescent purple. Fluorescent yellow-green was investigated because of its
importance in school zones, pedestrian zones, and bicycle lanes. Fluorescent yellow was investigated
because this color gains importance in high hazard warning sign applications. Pink and purple as well as
their fluorescent counterparts were investigated, because these unassigned MUTCD colors may become
important in the near future for emergency vehicle routing and other special purpose signing.
The study results indicate a small increase in the legibility distance when using selected fluorescent
color combinations on traffic signs. We hypothesize that the increase in legibility distance primarily comes
from the increase in luminance contrast when fluorescence is added. This increase in luminance contrast
is particularly notable for negative contrast signs where the legend consists of black paint and the
background is a light fluorescent color. Care must be taken to avoid contrast reduction when adding
fluorescence to the background of positive contrast signs, especially if the legend is non-fluorescent. In
such cases it is actually possible that a reduction in the legibility distances may be observed. In summary,
it is concluded that fluorescent colors outperform their normal color counterparts both in terms of detection
and in terms of recognition.
Schnell, Bentley, Hayes, and Rick 5
REVIEW OF THE TECHNICAL LITERATURE
Studies involving Fluorescent Colors
There are references to fluorescent colors that are technically incorrect. Wortham [2] described
fluorescent colors as reflecting ultraviolet (UV) light, leading to sign reading distances that are 3 to 5 times
greater than what is found with normal colors. We certainly question that one can increase the legibility of
a sign by this stated amount simply by adding fluorescence. Detection distances are likely to be much
longer when fluorescence is used but legibility distances appear to be much less affected by the presence
of fluorescence. Secondly, humans cannot see UV light and thus one would not perceive any visual
stimulation if UV light was actually reflected from a sign. Durable fluorescent traffic sign sheeting materials
do not make use of any UV light, much rather, short wavelength light (in the visible blue range) is
absorbed by the fluorescent pigment, and re-emitted at a longer wavelength, thus giving the overall
appearance of the material to be brighter than would be expected based on reflectance alone.
A sound technical description of the workings of durable fluorescent traffic sign materials is given by
Burns and Pavelka [3]. Zwahlen and Schnell [4] investigated daytime conspicuity of fluorescent and non-
fluorescent background color targets in the field against a green background as a function of the
peripheral angle and the target size. Zwahlen and Vel [5] conducted a daytime field study to determine the
conspicuity in terms of peripheral visual detection and recognition of different fluorescent and non-
fluorescent color targets against different backgrounds using only one target size (0.152m x 0.305m, 6" x
12"). Burns and Johnson [6] conducted chromaticity measurements and subjective rating experiments in
the field using fluorescent and non-fluorescent color targets. Brekke and Jenssen [7] conducted a study
investigating the visibility distances of fluorescent and non-fluorescent signs.
Studies on Legibility
Schnell and Zwahlen [8][9] hypothesized that for a given available (rectangular) display width and height, it
is possible to enhance the standard highway alphabets (FHWA, 1977) with regard to legibility by reducing
the absolute value of the legibility threshold contrast. Information collected by Hummer and Scheffler [10]
indicated that fluorescent traffic signs affected driver behavior in a positive way. Zwahlen and Schnell [11]
described an exploratory daytime/nighttime sign recognition/legibility field driving experiment involving 11
Schnell, Bentley, Hayes, and Rick 6
new reflectorized right shoulder mounted traffic signs and 10 young, healthy subjects (3 replications). The
experimental procedure used in the present research followed exactly the experimental method used by
Zwahlen and Schnell [11]. Table 1 shows the daytime results of the legibility study performed by Zwahlen
and Schnell [11]. In our present research, we used a PLANT ENTRANCE warning sign with the exact
same dimensions as the one used by Zwahlen and Schnell. Their PLANT ENTRANCE sign provided an
average daytime legibility distance of 125.5m while our non-fluorescent PLANT ENTRANCE sign provided
113.1m. This appears to represent a fairly good correlation on that base line. Sanders [12] provides a
simple mathematical relationship for converting the Snellen visual acuity score of a person into a detection
distance for reading text. Zwahlen, Sunkara, and Schnell [13] conducted a thorough review of the legibility
literature and categorized the research findings of the reviewed studies into three major groups, effects of
width to height ratio (W/H), effects of stroke-width to height ratio (SW/H), and effects of inter character
spacing to height ratio (S/H) on legibility. Their paper represents an extensive collection of relevant
legibility literature.
Schnell, Bentley, Hayes, and Rick 7
METHOD
This study involved daytime viewing, and it was therefore decided that a field setting was the most
appropriate method for the collection of the legibility distances. There are established laboratory methods
[14] to determine the legibility of textual messages, however, to fully capture the potential of fluorescent
color signs, it was felt that a dynamic driver observation task in the field was more appropriate. A total of
six diamond signs were fabricated with one side being the fluorescent color and the other side being the
non-fluorescent color. Landolt Rings were used as legend. Landolt rings (or often called Landolt Cs) are
internationally standardized optotypes established at the 11th International Congress of Ophthalmology,
which was held in Naples, Italy, in 1909, and the inventor was Mr. Landolt, a French ophthalmologist. The
height of a Landolt ring must be exactly five times its stroke with, and the gap opening must be exactly
equal to the stroke width. Like the Snellen E optotypes often used by ophthalmologists, the subject’s job is
to detect the orientation of the Landolt ring by identifying the direction in which the opening is facing (either
in 90o increments, up, down, left, right, or in 45o increments). A PLANT ENTRANCE warning sign was also
used for base line comparison purposes with previous work performed by Zwahlen and Schnell [11]. A
pedestrian symbol sign was used because of ongoing pedestrian safety research at the Operator
Performance Laboratory (OPL) at the University of Iowa.
Subjects
Nine young and healthy subjects (6 males, 3 females) were recruited for this experiment. All subjects had
a valid US driver’s license, normal or better visual acuity, normal or better contrast sensitivity, and showed
no color perception deficiencies. All subjects signed consent forms and instruction forms prior to
beginning their experimental runs. The subjects were instructed to verbally indicate the first point in time
the orientation of the Landolt ring opening was clearly discernible, the PLANT ENTRANCE legend was
clearly legible, or the pedestrian symbol was clearly recognizable during the approach. Guessing was
discouraged. Resulting from this method are functional legibility distances. A discussion as to the
adequacy of our subject sample is given in the discussion section of this paper.
Schnell, Bentley, Hayes, and Rick 8
Apparatus
Experimental Traffic Signs
The six diamond shaped experimental signs are shown in Figure 1 and photometric details are shown in
Table 2a. All Landolt rings had a diameter of 200mm and a stroke width of 40mm (being equivalent to the
gap opening). From Figure 1 and Table 2a, it can be seen that sign 1 had a non-reflective black paint
Landolt ring and a microprismatic normal yellow background on side 1, and a microprismatic fluorescent
yellow-green background on side 2. Sign 2 had a non-reflective black Landolt ring, a microprismatic
normal yellow background on side 1, and a microprismatic fluorescent yellow background on side 2. The
reader may note that both sign 1 and sign 2 had one side that was normal yellow. This duplication by
introduction of the dummy level of normal yellow was necessary to keep the experiment fully balanced, as
we were evaluating three types of yellow on four sign faces. Sign 3 had a non-reflective black Landolt ring,
a non-reflective normal pink background on side 1, and a microprismatic fluorescent pink background on
side 2. Sign 4 had a microprismatic normal yellow Landolt ring on a non-reflective normal purple
background on side 1, and a microprismatic fluorescent yellow Landolt ring on a microprismatic
fluorescent purple background on side 2. Sign 5 was a pedestrian symbol sign with a black non-reflective
paint symbol, a normal yellow background on side 1, and a microprismatic fluorescent yellow background
on side 2. Sign 6 was a PLANT ENTRANCE warning sign with a black non-reflective paint legend (0.15m,
6” series C), a normal yellow background on side 1, and a microprismatic fluorescent yellow background
on side 2.
The color coordinates that were measured for a D65 illuminant are listed in Table 2a. Using the CIE
1976 (Luv) colorimetric system it is possible to determine the color difference between the backgrounds
and their corresponding legends. The rightmost column in Table 2a lists the color difference ∆E*uv (for
formulae see [15]) of the individual color backgrounds against their respective legends. From this table, it
can be seen that the added property of fluorescence did in fact increase the color difference ∆E*uv,
indicating that the fluorescent color signs have the potential for being more legible in the configuration
used in this study. The yellow, fluorescent yellow, and fluorescent yellow-green sheeting materials were
commercially available microprismatic (type VII) materials. The microprismatic fluorescent pink and
microprismatic fluorescent purple sheeting materials were prototype materials provided by a sign sheeting
manufacturer. An attempt was made to locate a close color match in the corresponding non-reflective
normal pink and purple counterparts. The signs were equipped with a quick mount mechanism for easy
Schnell, Bentley, Hayes, and Rick 9
posting and removal when changing from the fluorescent condition to the non-fluorescent condition was in
order. The sign posts were standard 76mm square posts inserted into sleeves that were driven into the
ground. The bottom tip of each sign was 2.4m (8 ft) above the road surface. The lateral offset from the
right road edge to the center of the sign was 4.2m (wide shoulders are common in Iowa for agricultural
transportation).
Experimental Vehicle
The experimental vehicle was a 1996 Ford Taurus LX Sedan (Figure 2a). The vehicle had a windshield
transmission of 0.71, measured at the observer line of sight. All runs were made with the engine idling and
the vehicle coasting at about 10 – 15 km/h. The following average dimensions apply: Ground to Eye height:
1.14m, lateral distance from longitudinal car axis to the driver's sagital plane, 0.31m. The vehicle was
equipped with a calibrated Distance Measuring Instrument (DMI).
Test Site and Layout
The test site was located on Washington County highway G26, 6 miles north of Washington, Iowa. The
test site map is shown in Figure 2b. The G26 site is ideal for such experiments because of its extremely
low traffic volume, and its straight and level alignment. The highway runs almost exactly in the East-West
direction. During the time of the experiment (8:00am to 4:30pm, April 22, 2000) the sun was always high in
the southern hemisphere, providing a fairly uniform illumination. It should be noted that this experiment
was performed on a fair day with only few clouds. Figure 2c shows the layout of the signs on the test site.
A total of six square posts were driven. Three posts were staggered in the Eastbound direction, and three
posts were staggered in the Westbound direction. Two starting positions were marked on the ground
400m in advance of the first sign in the Eastbound direction and 400m in advance of the Westbound
direction. The pedestrian sign was posted as the first sign in the cluster of signs in the Westbound
direction, so that there was enough approach distance to ensure that no subject could recognize the
symbol from the start. The signs within the test cluster were separated by 121m (with the exception of the
separation between signs 1 and 6, which was 243m). This separation was determined with a small pilot
experiment and turned out to be ideal from an efficiency point of view. A tighter spacing might have led to
a high workload condition for the subject, who would have had to keep track of the signs as they became
legible in rapid succession, a longer spacing would have wasted run time.
Schnell, Bentley, Hayes, and Rick 10
Experimental Design and Procedure
The subjects were screened, instructed, and briefed in the laboratory about one day prior to arriving at the
test site. Two subjects were scheduled to arrive in pairs at the test site in intervals of about 1 hour and 30
minutes. Prior to beginning the experimental runs, the subjects familiarized themselves with the
experimental vehicle and the test course. Each subject drove the instrumented Ford Taurus provided by
the Operator Performance Laboratory (OPL) at approximately 10 - 15 km/h down the course. One practice
loop was allowed, and the subjects were given a chance to look at all traffic signs and post installations
prior to the experiment. Care was taken to ensure that the subjects understood that they were to verbally
indicate the first point in time at which they could clearly discern the orientation of the Landhold rings,
clearly recognize the pedestrian symbol, or clearly read the PLANT ENTRANCE legend. A two-factor
within-subjects repeated measures design was used for the Landolt ring signs. The following independent
within subjects variables were used: 1. Fluorescence (present, absent), 2. Sign color (black on yellow-
green, black on yellow, black on pink, yellow on purple). For clarification, the experimental design used in
this study is called a within-subjects design, because the levels of all factors (fluorescence and color) are
presented to all subjects, thus subjects served as their own control, allowing us to determine the effect of
the factors within the subjects.
The dependent variable was the legibility distance obtained with the experimental signs. Three
replications were performed, resulting in a total of 24 Landolt ring observations per subject. The additional
two signs (pedestrian and PLANT ENTRANCE) increased the total number of observations per subject to
36. Presentation order of the fluorescent vs. normal colors and the approach direction were balanced to
avoid any sequence effects. The time for each subject to complete the total number of 36 observations (in
6 round robin passes through the test site) was approximately 45 minutes. One experimenter was riding
along in the passenger seat. This experimenter instructed the driver subject during the experiment. Prior
to each run, the experimenter in the experimental vehicle instructed the subject to pull the car up to the
start mark so that the longitudinal eye location was exactly abeam the start mark. The experimenter then
reset the DMI, and asked the subject to start the run. The DMI display was paused (but the DMI kept
counting distance in the background) as soon as the subject indicated legibility of the sign closest to
him/her on the right road shoulder. The distance traveled was read off the DMI and logged into a data
collection sheet. The subject immediately concentrated on the next sign ahead as soon as the present
sign was readable. After one pass through the test site in one direction, the vehicle was lined up at the
Schnell, Bentley, Hayes, and Rick 11
start mark on the other end of the test site and the process repeated. The signs were rotated (factor of
fluorescence was switched) after the first three half-replications were complete. The remaining half of the
readings was then completed.
Schnell, Bentley, Hayes, and Rick 12
ANALYSIS AND RESULTS
The first step in the data analysis was the offline calculation of the legibility distances. This was
accomplished by subtracting the travel distance from the overall start-to-sign distance (see Figure 2c) for
each sign. The detection distances reported herein represent the longitudinal distance from the observer’s
eye location to the vertical sign face plane. Figure 3 shows the cumulative percentages as a function of
the legibility distance [m] for each of the six signs. Initial visual inspection of these cumulative graphs
immediately reveals that the increase in legibility distance due to fluorescence is not very strong. However,
there is a consistent increase in all cases tested in this experiment. The strongest increase effect is found
in sign 3 (black on pink). Table 2b shows the average legibility or recognition distances for each of the six
signs along with information about the legend dimensions, visual angle subtended by the legend height,
coefficient of variation, the average legibility indices, and the percent increase in legibility when adding the
property of fluorescence. From Table 2b it can be seen that switching from a normal yellow background to
a fluorescent yellow-green background (both with a black legend) increases the average legibility distance
by 5.3%. When switching from a normal yellow background to a fluorescent yellow background (both with
a black legend), the increase is 11.8%. This finding was somewhat surprising as we expected the
fluorescent yellow-green to provide the greater increase of the two. Switching from a normal pink
background to a fluorescent pink background (both with a black legend), increased the legibility distance
by 15.9%. The yellow on purple sign provided an increase in legibility distance of 9.2% when switching
both the legend and the background from non-fluorescent to fluorescent. The fluorescent yellow-green
pedestrian sign showed a recognition distance that was 7.9% longer than that of its normal yellow
counterpart. Finally, a 5.3% increase in the legibility distance was observed for the PLANT ENTRANCE
sign when switching from normal yellow to fluorescent yellow.
Table 3a shows the individual subject legibility distance responses for the six signs. It should be
noted, that the three replications were lumped together, as no significant learning was found. An initial
Analysis of Variance (ANOVA) was performed on the individual replication data, and it was found that
learning was not significant with FREP(2,16)=1.919, p =0.17. Table 3b shows the Analysis of Variance
(ANOVA) table for the Landolt ring signs with all replication data lumped together. The repeated measures
ANOVA was performed only on the Landolt ring signs, because these signs differed only in the within-
subjects independent variables (presence or absence of fluorescence, and color). From the ANOVA table,
it can be seen that the factor fluorescence statistically significantly increased the legibility distances with
Schnell, Bentley, Hayes, and Rick 13
FFLUOR(1,26)=17.778, p = 0.0002. However, the practical gain in legibility distance appears to be only
within the range of 5.3% to 15.9% when adding fluorescence. Box plots for the six signs (12 sign faces)
are shown in Figure 4. The boxes are arranged so that for each color the fluorescent member is always
immediately followed by its non-fluorescent counterpart. Similarly to the cumulative plots in Figure 3, it is
seen from Figure 4 that the property of fluorescence works in the right direction for the signs used in this
study. However, the overall increase in legibility distance is smaller than we expected.
Schnell, Bentley, Hayes, and Rick 14
DISCUSSION AND CONCLUSIONS
This study was conducted in order to test the hypothesis if adding the property of fluorescence while
leaving all (or in practical terms most) other sign features the same, does in fact increase the legibility
distance. Some concerns were voiced in the literature [7], that the increased daytime brightness provided
by fluorescent signs may actually reduce legibility. This field study used six diamond shaped signs that
were fabricated with one side being the fluorescent color and the other side being the non-fluorescent
color. The fluorescent colors were fluorescent yellow-green, fluorescent yellow, fluorescent pink, and
fluorescent purple. The normal colors were yellow, pink, and purple. It was found that adding the property
of fluorescence to the colors used in this study did in fact statistically significantly increase the legibility
distances. However, the actual increases were quite small from a practical point of view, ranging from
5.3% to 15.9% when adding the property of fluorescence. However, all signs showed a consistent
increase in the legibility distance, which is good news for traffic sign designers. We hypothesize that the
small increase in the legibility distance may be attributed to the increase in the color difference ∆E*uv
between the legend and the sign background (see Table 2a), and the increased background luminance,
leading to an increased luminance contrast.
The reader is made aware at this point that our sample of nine subjects consisted only of young,
healthy subjects with normal or better visual acuity. As such, the sample is quite homogenous and with
three replications, we are confident that the sample is very adequate from a statistical point of view, as
evidenced by the high statistical power in the Analysis of Variance (ANOVA) of 0.99 and 1.0 for
fluorescence and color, respectively (see Table 3). These high power values indicate that our statistical
test based on our subject sample was so powerful, that the shift in the mean detection distance due to the
factors of fluorescence and color would be detected virtually at all times. In other words, we are very
confident, that the increases in legibility distance are in fact due to fluorescence and not due to chance
and variation. However, we are aware that our subject sample does not fully represent the licensed driver
population as a whole, since we did not include any older subjects.
Schnell, Bentley, Hayes, and Rick 15
A word of caution may be in place here. The property of fluorescence may potentially reduce
legibility distance in fluorescent positive contrast signs, if the legend is non-fluorescent. In this case it is
well possible that the increased background luminance actually reduces the luminance contrast, because
the non-fluorescent legend will not change in terms of luminance. In such cases, sign designers may have
to trade off the increased detection potential of the fluorescent sign with a slight reduction in legibility. In
most cases, this slight reduction can be expected to be in the range of a few percent. Sign designers may
want to use a larger legend in positive contrast signs with a fluorescent background and a non-fluorescent
legend. It may also be possible to use a thin black border around the outline of the non-reflective positive
contrast legend to somewhat reduce the above-mentioned effect.
In general, for negative contrast signs and for positive contrast signs where both the legend and the
background are fluorescent, a slight increase in the legibility distance can be expected. It should be noted,
that the illumination conditions that were present in this study (fair day) represent somewhat of a worst-
case scenario for fluorescent colors. Larger percent increases in legibility distance may be expected when
using fluorescent colors during the time of twilight, during overcast days, and during daytime in inclement
weather.
Schnell, Bentley, Hayes, and Rick 16
REFERENCES
[1] US Department of Transportation, Federal Highway Administration, Manual of Uniform Traffic Control
Devices for Streets and Highways, Part VIII, 1988
[2] Wortham, S., “Do Fluorescent Crossing Signs Have a Bright Future ?…”, Traffic Safety, Volume 95,
Pp. 20-3, July-August, 1995
[3] Burns, D. and L. Pavelka. Visibility of Fluorescent Materials for Signing. Color Research and
Application. Vol. 20, No.2, Apr 1995, pp. 108-116.
[4] Zwahlen, H. T., and Schnell, T., “Visual Detection and Recognition of Fluorescent Color Targets
Versus Non-fluorescent Color Targets as a Function of Peripheral Viewing Angle and Target
Size”, Transportation Research Record, Number 1605 , 1997.
[5] Zwahlen H.T., U.D. Vel, Conspicuity in Terms of Peripheral Visual Detection and Recognition of
Fluorescent Color Targets Versus Non-fluorescent Color Targets Against Different Backgrounds
in Daytime, Transportation Research Record 1456, pp. 125-38, 1994.
[6] Burns D.M., Johnson N.L., The Correlation of Measured Spectral Radiance of Fluorescent and Non-
Fluorescent Materials to Perceived Conspicuity under Natural Lighting, Paper Presented at the AIC
Interm Meeting on Colorimetry, Berlin, September 3-6, 1995.
[7] Brekke B and GD Jenssen. A Comparison of the Visibility of Fluorescent and Standard Retroreflective
Traffic Signs Day and Night, in Proceedings of the 1997 Progress in Automobile Lighting (PAL)
Symposium, Darmstadt, Germany, September 23-24
[8] Schnell, T., Zwahlen, H.T., “A Numerical Method For Optimizing Uppercase Alphanumeric Text In
Traffic Sign Applications”, Paper presented at the 15th Biennial Symposium on Visibility,
Transportation Research Board, May 15-16, Washington, DC, 2000
[9] Schnell, T., and Zwahlen, H.T., “Legibility Threshold Contrast of Uppercase Text Seen Against a Dark
Background”, in Proceedings of the Human Factors and Ergonomics Society 43rd Annual
Meeting, September 27 – October 1, Houston, TX, 1999,
[10] Hummer, J. and Scheffler, C., “Driver Performance Comparison of Fluorescent Orange to Standard
Orange Work Zone Traffic Signs”, Transportation Research Record 1657, TRB, National
Research Council, pp. 55-62.
[11] Zwahlen, H.T., and Schnell, T., “Legibility of Traffic Sign Text and Symbols”, Transportation Research
Record 1692, National Academy of Sciences, Washington, DC., 1999
Schnell, Bentley, Hayes, and Rick 17
[12] Sanders M.S., McCormick E.J., Human Factors Engineering and Design, 7th edition, ISBN 0-07-
054901-X, Library of Congress TA 166.S33, McGraw Hill, Publisher, 1993
[13] Zwahlen, Helmut T., Sunkara, Murali, and Schnell, Thomas, "A Review of Legibility Relationships
within the Context of Textual Information Presentation," paper presented at the 74th Annual
Meeting of the Transportation Research Board, Paper No. 950888, January 22-26, 1995,
Washington, DC. Published in Transportation Research Record No. 1485, 1995, pp. 61-70.
[14] Schnell, T., "Legibility Optimization of Uppercase Alphanumeric Text for Displaying Messages in
Traffic Applications", 1998, Ph.D. Dissertation, Department of Industrial and Manufacturing
Systems Engineering, Ohio University, Athens, Ohio 45701-2979, 471 pages.
[15] Wyszecki G., Stiles W.S. Clor Science: Concepts and Methods, Quantitative Data and Formulae, 2nd
edition, John Wiley and Sons, Publisher, New York, 1982.
Schnell, Bentley, Hayes, and Rick 18
a. Sign 1: Black Paint Landolt C on Yellow Micro prismatic, Side 1 Normal Yellow, Side 2 Fluorescent Yellow Green (Figure shows Fluorescent Yellow Green as Example)
b. Sign 2: Black Paint Landolt C on Yellow Micro prismatic, Side 1 Normal Yellow, Side 2 Fluorescent Yellow (Figure shows Normal Yellow as Example)
c. Sign 3: Black Paint Landolt C on Pink, Side 1 Non-Reflective Pink Film, Side 2 Microprismatic Fluorescent Pink, (Figure shows Fluorescent Pink as Example)
d. Sign 4: Yellow Microprismatic Landolt C on Purple, Side 1 Background is Non-Reflective Purple Film, Legend is Microprismatic Normal Yellow, Side 2 Background is Microprismatic Fluorescent Purple, Legend is Fluorescent Yellow, (Figure shows Fluorescent Purple as Example)
0.76
0.6
0.30.05
e. Sign 5: Black Paint Pedestrian on Yellow Micro prismatic, Side 1 Normal Yellow, Side 2 Fluorescent Yellow Green (Figure shows Fluorescent Yellow Green as Example)
e. Sign 6: Black Paint PLANT ENTRANCE (Series C) on Yellow Microprismatic, Side 1 Normal Yellow, Side 2 Fluorescent Yellow (Figure shows Normal Yellow as Example)
Figure 1. Experimental Signs Used in the Present Study
Schnell, Bentley, Hayes, and Rick 19
Test Section
a. Rear View of the Experimental Vehicle Performing a b. Test Site used in Present Study is Located Run in the Westbound Direction, Signposts Staggered on Washington County Highway G26, 6 Miles along Shoulder North of Washington, Iowa, Sign Approaches
Exactly Along East West Axis
c. Layout of the Six Experimental Signs used in the Present Study Figure 2. Experimental Test Site and Layout
Schnell, Bentley, Hayes, and Rick 20
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 50 100 150 200 250
Detection Distance [m]
Cum
ulat
ive
Per
cent
age
LC, Fl. Yellow Green Rep 1to 3, Average= 122.8 m,Std.dev= 22.4 m, N= 27
LC, Yellow Rep 1 to 3,Average= 116.6 m, Std.dev=25.7 m, N= 27
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 50 100 150 200 250
Detection Distance [m]
Cum
ulat
ive
Per
cent
age
LC, Fl. Yellow Rep 1 to 3,Average= 125.3 m, Std.dev=23.8 m, N= 27
LC, Yellow Rep 1 to 3,Average= 112.1 m, Std.dev=24.7 m, N= 27
a. Sign 1: Black LC, Yellow or Fluorescent b. Sign 2: Black LC, Yellow or Fluorescent Yellow Green Background Yellow Background
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 50 100 150 200 250
Detection Distance [m]
Cum
ulat
ive
Per
cent
age
LC, Fl Pink Rep 1 to 3,Average= 109.9 m, Std.dev=28.2 m, N= 27
LC, Pink Rep 1 to 3,Average= 94.8 m, Std.dev=20.7 m, N= 27
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 50 100 150 200 250
Detection Distance [m]
Cum
ulat
ive
Per
cent
age
LC, Fl. Purple, Rep 1 to 3,Average= 119.0 m, Std.dev=20.7 m, N= 27
LC, Purple, Rep 1 to 3,Average= 109.0 m, Std.dev=18.5 m, N= 27
c. Sign 3: Black LC, Pink or Fluorescent d. Sign 4: Yellow/Fluorescent Yellow LC, Purple Pink Background or Fluorescent Purple Background
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 50 100 150 200 250
Detection Distance [m]
Cum
ulat
ive
Per
cent
age
PED, Fluorescent YellowGreen Rep 1 to 3, Average=194.7 m, Std.dev= 21.1 m,N= 27PED, Yellow Rep 1 to 3,Average= 180.5 m, Std.dev=25.2 m, N= 27
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 50 100 150 200 250
Detection Distance [m]
Cum
ulat
ive
Per
cent
age
PE Fluorescent Yellow GreenRep 1 to 3, Average= 119.1m, Std.dev= 18.8 m, N= 27
PE Yellow Rep 1 to 3,Average= 113.1 m, Std.dev=20.3 m, N= 27
e. Sign 5: Black Pedestrian Symbol, Yellow f. Sign 6: Black PLANT ENTRANCE, Yellow or Fluorescent Yellow Green Background or Fluorescent Yellow Background
FL=Fluorescent, LC=Landolt C, PED=Pedestrian Symbol on diamond shaped sign, PE=Plant Entrance legend on diamond shaped sign
Figure 3. Cumulative Percentage as a Function of Recognition Distance for Traffic Signs used in the
Current Study
Schnell, Bentley, Hayes, and Rick 21
0
50
100
150
200
250
Legi
bilit
y D
ista
nce
[m]
LC F
L P
urpl
e
LC P
urpl
e
LC F
luor
. Yel
low
LC F
luor
. Yel
low
Gre
en
LC Y
ello
w
LC Y
ello
w.2
LC F
luor
Pin
k
LC P
ink
Flu
or. Y
ello
w P
ed
Yel
low
Ped
Flu
or. Y
ello
w P
LAN
T E
NT
RA
NC
E
Yel
low
PLA
NT
EN
TR
AN
CE
Note: LC = Landolt C, Ped = Pedestrian Symbol Sign, FL=Fluorescent, LC Yellow and Yellow 2 are both ordinary yellow but were presented on two separate signs (signs 1 and 2, see Figure 1) at separate locations (see Figure 2c)
Figure 4. Boxplot of Legibility Distances [m] Obtained in Present Study
Schnell, Bentley, Hayes, and Rick 22
Table 1. Daytime Legibility Results of the Study Performed by Zwahlen and Schnell [11]
Table reproduced from [11]
Stimulus Daytime
Sign
Symbol/ Letter
Height H [m]
Symbol Width,
Average Letter Width W [m]
Typical Symbol Stroke
Width or Letter StrokeWidth
SW [m]
Contrast Polarity
Average Legibility/
Recognition Distance D
[m]
Visual Angle [min]
subtended by H at D
StdDev. [m]
Coeff. Of Variation
COV
Average Legibility
Index [m]/[cm ]
Average Legibility
Index [ft]/[in]
N
Sign1, Curve Arrow (W182) 0.46 0.46 0.1 Negative 318.9 4.96 62.2 0.20 6.9 57.8 30Sign 2, Large Arrow (IM24) 0.56 0.3 0.11 Positive 291.2 6.61 59.4 0.20 5.2 43.3 30Sign 3, Right Angle (IM19) 0.21 0.33 0.06 Positive 182.5 3.95 48.9 0.27 8.7 72.4 30Sign 4, Small Arrow (IM26) 0.25 0.17 0.07 Positive 149.6 5.74 36.4 0.24 6.0 49.9 30Sign 5, DO NOT PASS (R33) 0.15 0.0825 0.02 Negative 125.7 4.10 28.5 0.23 8.4 69.9 30Sign 6, NO EDGE LINES (OW 167) 0.2 0.134 0.03 Negative 165.7 4.15 45.8 0.28 8.3 69.1 30Sign 7, RUN CREEK (D4-A) 0.33 0.264 0.06 Positive 259.9 4.37 54.7 0.21 7.9 65.6 60Sign 8, PLANT ENTRANCE (W75) 0.15 0.0825 0.022 Negative 125.5 4.11 28.6 0.23 8.4 69.7 30Sign 9, Landholt C Blue, Day 0.2 0.2 0.04 Positive 157.1 4.38 45.8 0.29 7.9 65.5 30Sign 10, Landholt C Green, Day 0.2 0.2 0.04 Positive 141.1 4.87 40.4 0.29 7.1 58.8 30Sign 11, Landholt C White, Day 0.2 0.2 0.04 Negative 148.0 4.64 49.9 0.34 7.4 61.7 30Average 0.25
Schnell, Bentley, Hayes, and Rick 23
Table 2. Characteristics and Legibility Results of the Experimental Traffic Signs
a. Measured CIE 1931 (x,y) and CIE(Luv) 1976 Chromaticity Coordinates (2 degrees, D65 Illuminant) Note: MP = Microprismatic, PT = Nonreflective Paint, NRF, Nonreflective Film
Chromaticity Coordinates CIE 1931 Chromaticity Coordinates CIELUV 1976
Color Difference Legend to
Backgroundx y Y X Z u' v' L* u* v* ∆E*uv
Yellow Background MP 0.529 0.461 20.270 23.260 0.440 0.283 0.555 52.141 57.835 58.803Black Landholt C PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949
Fl. Yellow.Green Background MP 0.412 0.573 83.980 60.384 2.198 0.182 0.570 93.442 -19.111 123.096Black Landholt C PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949Yellow Background MP 0.529 0.461 20.270 23.260 0.440 0.283 0.555 52.141 57.835 58.803Black Landholt C PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949
Fl. Yellow Background MP 0.533 0.460 45.860 53.138 0.698 0.286 0.555 73.455 84.251 83.109Black Landholt C PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949Pink Background NRF 0.510 0.280 10.750 19.580 8.063 0.382 0.472 39.156 93.779 1.800Black Landholt C PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949
Fl. Pink Background MP 0.538 0.274 19.690 38.661 13.510 0.413 0.473 51.485 143.968 3.190Black Landholt C PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949Purple Background NRF 0.380 0.170 5.200 11.624 13.765 0.355 0.357 27.297 55.837 -39.353Yellow Landholt C MP 0.529 0.461 20.270 23.260 0.505 0.283 0.555 52.141 57.721 58.580
Fl. Purple Background MP 0.432 0.185 6.180 14.431 12.794 0.397 0.382 29.862 77.215 -33.441Fl. Yellow Landholt C MP 0.533 0.460 45.860 53.138 0.809 0.286 0.555 73.455 84.129 82.872
Yellow Background MP 0.529 0.461 20.270 23.260 0.440 0.283 0.555 52.141 57.835 58.803Black Pedestrian PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949
Fl. Yellow.Green Background MP 0.412 0.573 83.980 60.384 2.198 0.182 0.570 93.442 -19.111 123.096Black Pedestrian PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949Yellow Background MP 0.529 0.461 20.270 23.260 0.440 0.283 0.555 52.141 57.835 58.803
BlackPLANT
ENTRANCE PT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949
Fl. Yellow Background MP 0.533 0.460 45.860 53.138 0.698 0.286 0.555 73.455 84.251 83.109
BlackPLANT
ENTRANCEPT 0.302 0.326 4.500 4.169 4.757 0.194 0.471 25.260 -1.235 0.949
6 Side 1 86.943
6, Side 2 127.988
Sign Number Color Item Optics
5, Side 1 86.943
5, Side 2 141.026
4, Side 1 101.053
4, Side 2 124.406
3, Side 1 96.029
3, Side 2 147.570
2, Side 1 86.943
2, Side 2 127.988
86.9431, Side 1
141.0261, Side 2
b. Average Legibility/Recognition Distance, Visual Angle Subtended by Symbol/Letter Height, Standard
Deviation, Legibility Index, Number of Observations, and Percent Increase in Legibility Distance When Using Fluorescence
Stimulus
Symbol/ Letter
Height H [m]
Symbol Width,
Average Letter
Width W [m]
Typical Symbol Stroke Width
or Letter StrokeWidth
SW [m]
Contrast Polarity
Average Legibility/ Recogniti
on Distance
D [m]
Visual Angle [min]
subtended by H at
D
StdDev. [m]
Coeff. Of Variation
COV
Average Legibility
Index [m]/[cm]
Average Legibility
Index [ft]/[in]
N
% Increase in Legibility Distance
When Fluorescent
1, Side 1 Yellow Background MPBlack Landholt C PT
1, Side 2 Fl. Yellow.Green Background MPBlack Landholt C PT
2, Side 1 Yellow Background MPBlack Landholt C PT
2, Side 2 Fl. Yellow Background MPBlack Landholt C PT
3, Side 1 Pink Background NRFBlack Landholt C PT
3, Side 2 Fl. Pink Background MPBlack Landholt C PT
4, Side 1 Purple Background NRFYellow Landholt C MP
4, Side 2 Fl. Purple Background MPFl. Yellow Landholt C MP
5, Side 1 Yellow Background MPBlack Pedestrian PT
5, Side 2 Fl. Yellow.Green Background MPBlack Pedestrian PT
6 Side 1 Yellow Background MP
BlackPLANT
ENTRANCEPT
6, Side 2 Fl. Yellow Background MP
BlackPLANT
ENTRANCEPT
7.9
5.3
5.3
11.8
15.9
9.2
6.0 49.6 27
Sign Number Color Item Optics
119.1 5.77 18.8 0.16
0.18 5.7 47.1 27
0.11 9.7 81.1 27
0.14 9.0 75.2 27
0.022
180.5 3.81 25.2
194.7 3.53 21.1
113.1 6.08 20.3Negative
0.6 0.3 0.05
0.15 0.0825
5.5 45.4 27
119.0 5.78 20.7 0.17 6.0 49.6 27
109.0 6.31 18.5 0.17
4.7 39.5 27
109.9 6.26 28.2 0.26 5.5 45.8 27
5.6 46.7 27
125.3 5.49 23.8 0.19 6.3 52.2 27
112.1 6.13 24.7 0.22
0.2 0.2 0.04
Negative
Positive
5.8 48.6 27
122.8 5.60 22.4 0.18 6.1 51.2 27
116.6 5.90 25.7 0.22
94.8 7.25 20.7 0.22
Schnell, Bentley, Hayes, and Rick 24
Table 3. Individual Legibility Distance Data and ANOVA Table
a. Individual Subject Responses, 9 Subjects x 3 Replications = 27 Data Points Per Sign Condition
Note: There were three types of yellow: 1. Fluorescent Yellow-Green, 2. Fluorescent Yellow, 3 Normal Yellow. In order to completely remove the effect of sign post location, we used signs that were florescent on one side and normal on the other side. The signs were simply turned over on the same post to switch from fluorescent to normal. However, since there were three yellow colors, we had to use normal yellow as a dummy variable twice (see also Table 2a). Therefore, in the table below, there are two non-fluorescent (normal) yellow conditions. The signs were identical and the variation between the two signs is small.
PurpleFluor
Yellow
Fluor. Yellow Green
PinkFluor. Yellow
Ped
Fluor. Yellow PLANT
ENTRANCE
Purple Yellow Yellow PinkYellow
Ped
Yellow PLANT
ENTRANCE
101.8 108.2 99.1 64.3 175.3 123.4 107.6 80.2 93.6 80.5 207.6 87.888.1 134.1 158.5 121.9 199.3 89.9 87.8 118.3 128.3 112.5 166.7 94.889.6 97.8 138.1 95.1 187.1 107.0 91.7 108.8 124.1 84.7 195.1 68.9
146.9 133.2 134.7 156.1 202.7 84.1 127.7 130.5 126.8 112.5 197.5 85.6120.7 105.2 103.6 110.0 184.7 115.5 109.7 92.0 73.5 74.7 166.7 93.6122.5 124.4 128.0 112.8 178.6 104.6 136.6 142.6 87.5 64.3 122.8 89.0141.1 183.2 122.5 162.5 195.7 110.3 127.4 169.8 150.3 120.7 170.7 107.688.4 109.7 78.0 84.1 197.8 86.9 110.3 107.9 150.9 93.0 209.4 101.8
134.7 133.2 137.8 139.9 202.7 100.0 134.4 141.1 146.6 110.9 153.3 99.4104.9 93.6 96.0 49.7 224.6 132.0 83.8 69.2 76.8 66.8 215.2 118.3129.5 152.4 150.0 86.0 173.1 124.1 90.2 117.0 118.6 97.8 178.0 115.894.2 110.0 113.1 91.1 158.2 123.1 93.0 78.3 112.5 103.3 190.5 119.2
124.1 150.6 143.0 129.5 194.5 99.1 143.0 113.4 128.9 91.4 179.2 90.2133.5 112.2 100.0 110.3 131.7 123.4 121.9 94.5 73.2 74.4 178.3 103.6121.6 119.5 141.4 110.0 180.4 126.8 114.0 120.1 108.5 76.5 167.6 104.2133.5 167.0 149.7 148.4 199.9 118.3 129.2 143.9 121.9 142.3 219.2 119.5121.6 136.2 78.9 83.5 175.3 111.9 89.0 109.4 130.1 96.0 191.4 114.9133.8 124.1 129.2 136.9 195.1 111.3 113.1 125.9 146.3 117.0 208.5 117.097.8 75.3 99.7 78.9 211.5 145.4 74.7 75.9 79.2 69.5 226.5 131.1
103.6 119.8 133.8 109.7 193.2 143.9 85.0 118.9 115.8 104.5 182.3 134.7108.2 111.6 124.7 100.0 205.1 135.9 100.0 100.9 107.6 84.7 176.2 136.6134.7 153.9 141.4 146.6 201.2 146.3 92.7 79.9 121.9 87.8 153.3 141.7131.1 102.4 104.9 86.9 227.4 153.3 112.5 95.4 92.0 78.9 187.5 145.1117.7 140.8 120.1 111.6 222.8 119.5 112.8 102.7 99.1 65.8 142.3 143.0171.9 138.1 147.8 137.2 216.4 105.2 131.7 145.1 166.4 126.5 141.7 130.189.6 110.3 102.4 93.0 208.8 128.6 110.6 111.3 131.7 111.3 175.3 132.3
128.3 137.5 140.5 112.2 214.0 145.4 113.1 132.6 137.5 112.5 170.4 127.7Average [m] 119.0 125.3 122.8 109.9 194.7 119.1 109.0 112.1 116.6 94.8 180.5 113.1Std.Dev [m] 20.7 23.8 22.4 28.2 21.1 18.8 18.5 24.7 25.7 20.7 25.2 20.3LC Average [m]LC Stdev. [m]
119.324.3
108.123.7
Landholt C
Fluorescent Non Fluorescent
Other Landholt C Other
b. Landolt C ANOVA Table for Replication 1 – 3 Data Lumped together
26 66046.147 2540.236
1 6703.945 6703.945 17.778 .0003 17.778 .990
26 9804.135 377.082
3 10235.510 3411.837 12.637 <.0001 37.912 1.000
78 21058.693 269.983
3 619.261 206.420 1.016 .3904 3.047 .259
78 15853.795 203.254
DF Sum of Squares Mean Square F-Value P-Value Lambda Pow er
Subject
Fluor
Fluor * Subject
Color
Color * Subject
Fluor * Color
Fluor * Color * Subject
ANOVA Table for Legi Dist [m]