Displays 34 (2013) 208–214

Contents lists available at SciVerse ScienceDirect

Displays

journal homepage: www.elsevier .com/locate /d ispla

Reading performance and visual fatigue when using electronic paperdisplays in long-duration reading tasks under various lighting conditions

0141-9382/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.displa.2013.06.001

⇑ Corresponding author. Tel.: +886 988177582; fax: +886 3 318 5522.E-mail addresses: d9301301@mail.ntust.edu.tw, wolf.jamie@msa.hinet.net

(P.-C. Chang).

Po-Chun Chang a,⇑, Shuo-Yan Chou a, Kong-King Shieh b

a Department of Industrial Management, National Taiwan University of Science and Technology, 43, Sec. 4, Keelung Rd., Taipei, Taiwan, ROCb Department of Industrial Management, Oriental Institute of Technology, #58, Sec. 2, Sihchuan Rd., Banciao City, Taipei County, Taiwan, ROC

a r t i c l e i n f o a b s t r a c t

Article history:Received 12 January 2012Received in revised form 9 December 2012Accepted 3 June 2013Available online 12 June 2013

Keywords:Electronic paper displayVisual performanceVisual fatigueAmbient illuminanceLight source

In this study, the effects of ambient illuminance and light source on participants’ reading performanceand visual fatigue during a long reading task were investigated using three electronic paper displays.Reading on electronic paper displays was also compared with reading on paper. In Experiment 1, 100 par-ticipants performed a reading task where the display area for the text was equated for the displays. Theresults indicated that participants’ visual performance and visual fatigue did not differ significantlyamong different electronic paper displays, ambient illuminance conditions, or light sources. In Experi-ment 2, another 60 participants performed the same reading task where the full screen of each electronicpaper display was used to present the text. The results showed that reading speed differed significantlyacross different electronic paper displays and ambient illuminance levels. The reading speed was slowerfor displays with smaller screens and increased as the ambient illuminance increased. Changes in the crit-ical flicker fusion frequency significantly differed across ambient illuminance levels. Implications of theresults for the use of electronic paper displays are discussed.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

The electronic paper display is now the most popular of the newtypes of visual displays, owing to the enormous success of AmazonKindle, which uses a reflective-type display that differs markedlyfrom the conventional transmissive-type liquid–crystal display(LCD). Unlike a transmissive-type LCD, which uses a backlight asits pixel light source, a reflective-type display reflects ambient lightas does ordinary paper and can display text and images indefinitelywithout additional power. The available commercial electronic pa-per devices include Amazon Kindle, Sony PRS-700, and iRex Read-er. Although these electronic paper displays have been in themarket for a number of years, ergonomics studies of such devicesare limited. Isono et al. [1] conducted an experiment in which 13college students read with an electronic paper display and withconventional paper for 90 min. The results showed no significantdifferences in the level of visual fatigue between electronic readingand conventional reading. Jeng et al. [2] reported that the legibilityof text on electronic paper displays depends on the illuminanceintensity but not on the light source, and that conventional paperhad a higher visual comfort rating than electronic paper althoughboth supported similar performance in a letter-search task.

Many factors influence visual performance and visual fatiguewhen using displays, e.g., the display medium, light source, ambi-ent illuminance, polarity, and target-to-background luminancecontrast. Ambient illuminance is an important factor in video dis-play terminal (VDT) design. Many recommendations exist regard-ing the level of ambient illuminance. For cathode ray tube (CRT)workstations, an ambient lighting of 200–500 lx is generally sug-gested. The choice of illuminance level greatly depends on the task[3]. Ostberg [4] reported that a lower ambient illuminance mightbe more appropriate for CRT work. Xu and Zhu [5] studied the ef-fect of ambient illuminance and found that performance deterio-rated as ambient illuminance increased.

Regarding the effect of illuminance on visual performance andfatigue when using an electronic paper display, Lin et al. [6] foundthat illuminance had no significant effect on accuracy of visualtasks on electronic paper. Lee et al. [7] found that ambient illumi-nance had no significant effect on accuracy performance but had asignificant effect on search speed. Shieh and Lee [8] found that dif-ferent illuminances (200, 700, and 2000 lx) resulted in significantlydifferent preferred viewing distances and screen angles. Higherilluminances (700 and 2000 lx) resulted in longer preferred view-ing distances and smaller screen angles. They suggested thatreflective-type displays may require higher illuminance (700 lx orhigher). Wang et al. [9] and Wang et al. [10] pointed out that par-ticipants’ visual performance with electronic paper displays wassignificantly different under various ambient illuminances. In

Table 1Specification of the three electronic paper displays.

Device A Device B Device C

Manufacturerinformation

Sony PRS700

Amazon KindleDX

iRex 1000S

Diagonal screen size 6 in. 9.7 in. 10.2 in.Grey scales 8 16 16Resolution 800 * 600 1200 � 824 1024 * 1280Display technology E-Ink E-Ink E-Ink

P.-C. Chang et al. / Displays 34 (2013) 208–214 209

general, the abovementioned studies have found that the requiredambient illuminance for electronic displays is higher than that forCRT or LCD displays and that the proper range may be about 500–1000 lx. The above suggestion was based on short-duration exper-iments using simple word identification tasks. However, an elec-tronic paper display is typically used for reading text, and thereading time can last for hours. Furthermore, visual fatigue occursmostly after long periods of reading. Thus, a short-duration wordidentification task may be inappropriate for evaluating the visualperformance and visual fatigue when using an electronic paper dis-play. Moreover, these studies evaluated an old-type Kolin elec-tronic (Ch–LC) display and Sony e-book, which were notmainstream in the market. Consequently, there is a need to explorethe visual performance and visual fatigue experienced when usingpopular electronic paper displays (e.g., Sony Reader, Amazon Kin-dle, and iRex Reader) in a long-duration reading task and how theycompare with those when performing the same task on ordinarypaper.

In recent years, a number of studies on electronic paper displays[2,8,7] have reported that the effect of the light source is limited;for example, a sunlight lamp (D65) was similar to a fluorescentlamp (TL84) but better than a tungsten lamp (F). However, thesestudies also focused on word identification tasks. Whether the re-sults are similar for a long-duration reading task is unclear.

Many studies have compared reading on displays with readingon paper [11,12]. Slower reading speed, lower accuracy, greater fa-tigue, and less subjective preference were found when reading ondisplays [13,14]. Regarding the effect of the display medium onhigher cognitive ability such as text comprehension, some studies[11] concluded that comprehension was not different for materialpresented on paper or displays. However, Cushman [15] found thatreading speed and comprehension have an inverse relationship; forexample, slower readers show better comprehension. Thus,although the display medium appears to have little effect on com-prehension of reading material, it may affect reading speed andindirectly influence comprehension. However, most of the abovestudies focused on CRT or LCD displays and did not compare paperand electronic displays. Consequently, there is a need to explorethe effects of electronic paper display media, ambient illuminance,and light source on reading performance and visual fatigue duringlong periods of reading.

In summary, electronic paper displays may replace conven-tional paper, books, and magazines in the future. Studies on elec-tronic paper displays are very limited, and existing studies haveused visual search or identification tasks and employed searchspeed or accuracy as dependent measures. Studies that measurehigher cognitive performance and visual fatigue in long-durationreading tasks are required. In this study, we investigated the effectsof ambient illuminance (200, 500, and 1000 lx, with 1500 lx addedin Experiment 2) and light source (TL84 and D65) on participants’reading comprehension and visual fatigue using three electronicpaper displays (Amazon Kindle DX, Sony Reader PRS-505-SC, andiRex 1000S). A comparison was also made between reading onelectronic paper displays and on ordinary paper.

2. Experiment 1

2.1. Method

Two experiments were conducted in this study. The first em-ployed the display medium, light source, and ambient illuminanceas independent variables.

2.1.1. ParticipantsOne hundred college students (53 male and 47 female) between

19 and 28 years old (mean age, 23.3 year; SD, 1.8 year) participated

in the experiment. All had no prior reading experience on elec-tronic paper displays, and had corrected visual acuity higher than0.8 and normal color vision as tested by a Topcon screenscopeSS-3 and standard isochromatic charts. There was an institutionalreview board (IRB) approved by the Oriental Institute of Technol-ogy, and all the participants gave written informed consent. Eachparticipant was paid NT$300 (about 10 USD).

2.1.2. Experiment designThe first experiment evaluated three independent variables: the

display medium, illuminance level, and light source with the textarea controlled. Three electronic paper displays and conventionaloffice paper were used as test visual display units. The two lightsources used were D65 (6500 K) and TL 84 (4000 K). The illumi-nance was set at three different levels: 200, 500, and 1000 lx.Therefore, there were 18 experimental conditions [3(dis-plays) � 2(lightsources) � 3(illuminances)]. This experiment useda between-subject design for all independent variables, and fiveparticipants were randomly assigned to each condition. Thus,ninety participants were used for conducting the tests on elec-tronic paper displays. In addition, a comparison group (10 partici-pants) was used for conducting the tests on A4 paper with D65 asthe light source and an ambient illuminance of 500 lx. The totalnumber of participants in Experiment 1 was 100.

2.1.3. ApparatusThe following three electronic paper displays were used: Sony

PRS-700 with a 6-in. display, 800 � 600 resolution; Amazon KindleDX with a 9.7-in. display, 1200 � 824 resolution; and iRex 1000Swith a 10.2-in. display, 1024 � 1280 resolution. All use E-Ink� elec-tronic paper technology. Table 1 lists the specifications of the threeelectronic paper displays. To avoid brand effects, the three displayswere designated as devices A, B, and C. Fig. 1 shows the three elec-tronic paper displays with text display areas equated. The lightsource and ambient illuminance were controlled using a colorassessment cabinet (VeriVide CAC 120-5). The ambient illumi-nance was measured using a TOPCON Illuminance Meter IM-2D.According to the specification of the color assessment cabinet,the uniformity of the illuminance level was better than 5% andthe screen reflection was controlled. The critical flicker fusion fre-quency (CFF) of a participant was measured with a Lafayette FlickFusion Control 12023.

2.1.4. Experimental conditionsThe experimental task configuration is shown in Fig. 2. The

VDTs were placed inside the color assessment cabinet located ona table 73 cm in height. The front edge of the table was 17 cm fromthe screen center. The inclination angle of the screen was 105� withrespect to the horizontal axis. The viewing distance was controlledto 50 cm by using a chin fixture. These task setup parameters werefixed during the experiment. The participants could adjust theirseating height to make themselves as comfortable as possible.

Fig. 1. The three electronic paper displays. Device A, Device B (middle), and DeviceC (right). In Experiment 1, the three electronic paper displays with text displayareas equated.

Fig. 2. Experimental task configuration in this study.

210 P.-C. Chang et al. / Displays 34 (2013) 208–214

2.1.5. ProcedureA reading task was conducted to evaluate the participants’ vi-

sual performance and visual fatigue. A Chinese article in New Mingstyle font with a character height of 12 pt was used for the task.The article was a report about an earthquake, of which all partici-pants were unfamiliar. The text was presented in black on a whitebackground (positive polarity with a text-to-background lumi-nance contrast of 1:5; hence, the contrast ratio was set to 5). In thisexperiment, the text display area and formats were controlled tobe identical for each display because their screen sizes were differ-ent (see Fig. 1). The text was 30 pages long, with 17 lines per pageand 20 characters per line. The participants were asked to read thetext and answer 10 questions about the content as accurately andquickly as possible at the end. During reading, the participantsoperated the switch button on the device to turn to the next page.The switch speeds of the three devices were the same (around0.6 s). Before the formal experiment, a training session was con-ducted to familiarize the participants with the operation of the de-vice. CFFs were measured before and after the experimental task.In addition, a rating scale comprising the following six items wasused to obtain a subjective rating of visual fatigue [16]: (1) I havedifficulties in seeing; (2) I have a strange feeling around the eyes;(3) My eyes feel tied; (4) I feel numb; (5) I feel a headache; and(6) I feel dizzy looking at the screen. Each item was rated on a

10-point scale for evaluating the severity of discomfort, with 1 rep-resenting ‘‘not at all’’ and 10 representing ‘‘yes, very much.’’

2.1.6. Data collection and analysisThis study recorded the reading time and comprehension accu-

racy to evaluate the participants’ visual performance, change inCFF, and subjective visual fatigue to assess visual fatigue. The CFFchange was the CFF difference before and after the experiment.Subjective visual fatigue was determined by the total subjectiverating score of visual fatigue. Analysis of variance (ANOVA) wasconducted to evaluate the effects of the independent variables (dis-play medium, illuminance level, and light source) on the depen-dent measures. The level of statistical significance was set at0.05. All statistical analyses were performed using Statistical Anal-ysis System (SAS) software.

2.2. Results

2.2.1. Visual performanceTable 2 shows the participants’ visual performance at each level

of the independent variables. The data were tested and confirmedfor homogeneity of variance.

2.2.1.1. Reading time. The data in Table 2 shows that the readingtimes for the three electronic paper displays were about 1000–1200 s. The ANOVA results indicated that there was no significantdifference among the three displays. The difference between paperand electronic paper displays was also not statistically significant.The effects of ambient illuminance and light source on readingtime were not statistically significant.

2.2.1.2. Comprehension accuracy. Comprehension accuracies rangedbetween about 50% and 60% as a function of display medium,ambient illuminance, and light source. The ANOVA results indi-cated that none of the variables had a statistically significant effecton comprehension accuracy. The difference between paper andelectronic paper displays was also not significant.

2.2.2. Visual fatigueTable 3 lists the means and standard deviations of CFF change

and subjective visual fatigue at each level of the independent vari-ables in Experiment 1.

2.2.2.1. Change in CFF. The data in Table 3 shows that the CFF de-creased by about 2–3 Hz after the reading experiment. The ANOVAresults indicated that there was no significant difference amongthe three displays. The difference between paper and electronic pa-per displays did not reach significance. The effects of ambient illu-minance and light source were also not statistically significant.

2.2.2.2. Subjective visual fatigue. Values of subjective visual fatiguefor the three electronic paper displays were about 17–21; their dif-ferences were not statistically significant. However, the subjectivevisual fatigue for paper was lowest at 14.1. Under the same exper-imental conditions of D65 as the light source and an illuminance of500 lx, a t-test showed a significant difference between paper andelectronic paper displays [t(23) = �2.1, p < 0.05]. Neither the ambi-ent illuminance nor the light source had a statistically significanteffect on subjective visual fatigue.

The results of Experiment 1 indicated that reading speed andcomprehension accuracy were similar for the three electronic pa-per displays. Moreover, the change in the CFF and subjective visualfatigue were not significantly different between the electronic pa-per displays. That is, the major brands of electronic paper displayin the market are comparable in terms of visual performance andvisual fatigue. Two plausible reasons may explain the results.

Table 2Reading time and comprehension accuracy under independent variables in Experiment 1.

Independent variables Level Reading time (s) Comprehension accuracy (%)

Mean SD Mean SD

Display medium Device A 1032.0 252.4 56.0 15.1Device B 1165.7 281.1 57.8 14.1Device C 1117.0 235.5 57.7 16.1Paper 1217.7 289.3 54.0 13.5

Ambient illuminance 200 lx 1186.7 275.2 61.0 14.3500 lx 1057.2 245.0 57.8 13.11000 lx 1070.9 266.4 52.7 16.8

Light source D65 1141.7 278.0 58.2 13.4TL84 1068.2 245.4 56.1 16.5

Table 3CFF change and subjective visual fatigue under independent variables in Experiment 1.

Independent variables Level CFF difference (Hz) Subjective visual fatigue

Mean SD Mean SD

Display medium Device A �2.56 1.98 20.10 9.98Device B �2.77 2.52 20.77 10.32Device C �3.04 2.41 16.67 10.75Paper �2.25 1.59 14.10 6.56

Ambient illuminance 200 lx �2.82 2.35 17.47 9.36500 lx �2.40 1.92 19.13 11.011000 lx �3.10 2.67 19.27 10.01

Light source D65 �2.96 2.38 17.80 10.25TL84 �2.46 2.16 19.73 10.08

P.-C. Chang et al. / Displays 34 (2013) 208–214 211

Firstly, all three electronic paper displays use the same E-Ink�

technology. Secondly, although the display sizes of the three de-vices were different, the text display area and format were ar-ranged to be identical in the experiment (see Fig. 1). Thus, thenumber of text pages was the same for all electronic paper dis-plays. The results may have differed if the full screen had beenused to present the text for all displays.

Fig. 3. The three electronic paper displays. Device A, Device B (middle), and DeviceC (right). In Experiment 2, the text display areas were filled with experimental text.

3. Experiment 2

One plausible reason for the nonsignificant differences in visualperformance and visual fatigue for the three electronic paper dis-plays is that the text display areas were equated in Experiment1. Ordinarily, most of the display screen area is used to presentinformation; hence, the size of the display screen can make differ-ence. Thus, a more realistic test is to compare the three electronicpaper displays when the full screen of the display is used for textpresentation. Furthermore, because the light sources D65 andTL84 have only limited effects in many studies (including Experi-ment 1 here), the light source was removed as a factor in Experi-ment 2. Moreover, a greater ambient illuminance level, 1500 lx,was added because outdoor illuminance can reach this level. Inaddition, to enhance the motivation to read, the participants weregiven a NT$20 bonus for each correct answer in the 10 questionsregarding the reading material.

3.1. Method

The apparatus and experimental materials, experimental condi-tions and procedures, and data collection and analysis were thesame as in Experiment 1. The major difference was that the fullscreen for each display was used to display the text. Fig. 3 showsthe same experimental devices with full screen text. There were

340, 1040, and 1167 characters per full screen page and 30, 10,and 9 pages of text, respectively, for devices A, B, and C.

3.1.1. ParticipantsSixty college students (30 male and 30 female) between 19 and

26 years old (mean age, 22.7 year; SD, 0.9 year) participated inExperiment 2. These students were different from the 100 partici-pants in Experiment 1. All had corrected visual acuity of higher than0.8 and normal color vision; none had prior reading experience onelectronic paper displays or familiarity with the reading material.

3.1.2. Experimental designExperiment 2 evaluated two independent variables: the display

medium and illuminance level. Therefore, there were 12 experi-mental conditions [3(displays) � 4(illuminances)]. Five partici-pants were randomly assigned to each experimental condition.

Table 5ANOVA for reading time.

Source DF Type I SS Meansquare

F value Pr > F

Display 2 706860.91 353430.46 4.65 0.0143*

Illuminance 3 1480705.85 493568.62 6.49 0.0009*

Display*

illuminance6 21697.626 3616.27 0.05 0.9995

Error 48 3650220.03 76046.25Corrected total 59 5859484.42

* Significant at a = 0.05.

Fig. 4. Boxplot of reading time by three electronic displays.

15001000500200

1500

1400

1300

1200

1100

1000

900

800

Illumination

Mea

n of

Rea

ding

tim

e (S

ec)

Device A

Device B

Device C

Fig. 5. Line plot of reading time under illuminance levels.

212 P.-C. Chang et al. / Displays 34 (2013) 208–214

3.2. Results

3.2.1. Visual performanceTable 4 shows the participants’ visual performance at each level

of the independent variables. The data were first tested and con-firmed for homogeneity of variance.

3.2.1.1. Reading time. The data in Table 4 shows that device A hadthe longest reading time of 1252 s. Devices B and C had similarreading times of 1032 and 1014 s, respectively. The ANOVA results(Table 5) showed that the reading times for different electronic pa-per displays were significantly different [F(3, 48) = 4.65, p < 0.05].Fig. 4 shows a box plot of reading times for the three displays.Compared with the results of Experiment 1, the reading time fordevice A was longer in Experiment 2. The text display formats inthe two experiments were identical for device A. The participantsmay have read more carefully so as to answer the questions moreaccurately and receive incentive bonus, which may explain theirslower reading speed in Experiment 2. The screen sizes for devicesB and C were larger; hence, there were fewer text pages to read forthese devices. This may have led to less page-turning, fewer read-ing interruptions, and thus faster reading speed.

The effect of ambient illuminance on reading time was also sta-tistically significant [F(3, 48) = 6.49, p < 0.05]. The reading times forambient illuminances of 200, 500, 1000, and 1500 lx were 1306,1194, 970, and 926 s, respectively. Fig. 5 shows that the readingtime decreased as the ambient illuminance increased.

Duncan’s multiple range test (Table 4) indicated that the read-ing times for devices B and C were better than those for device A.In addition, the reading times under ambient illuminances of1000 and 1500 lx were shorter than those under 200 and 500 lx.

3.2.1.2. Comprehension accuracy. Comprehension accuracies werearound 60%, which is slightly greater than those of Experiment 1(about 57%). This is presumably because of the incentive bonusfor correct answers in Experiment 2. The ANOVA results indicatedthat neither the display medium nor ambient illuminance had astatistically significant effect on comprehension accuracy.

3.2.2. Visual fatigueTable 6 lists the means and standard deviations of the CFF

change and subjective visual fatigue at each level of the indepen-dent variables in Experiment 2.

3.2.2.1. Change in CFF. The CFF decreased by about 1–3 Hz after thereading experiment. The ANOVA results (Table 7) showed thatthere was no significant difference among the three displays. Theeffect of ambient illuminance on the change in CFF was statisticallysignificant [F(3, 48) = 3.89, p < 0.05]. Duncan’s multiple range test(Table 6) indicated that the changes in the CFF under ambient illu-minances of 500 and 1000 lx were smaller than the change under

Table 4Reading time and comprehension accuracy under independent variables in Experiment 2.

Independent variables Level Reading time (s) Duncan’s multiple range testa Comprehension accuracy (%)

Mean SD Mean SD

Display medium Device A 1252.8 261.6 A 64.5 15.0Device B 1032.2 307.2 B 62.5 13.7Device C 1014.0 192.1 B 60.3 13.5

Ambient illuminance 200 lx 1306.9 296.0 A 61.8 12.9500 lx 1194.9 279.5 A 62.0 16.21000 lx 970.7 242.0 B 59.7 12.31500 lx 926.2 150.2 B 66.2 11.6

a Duncan’s multiple range test, values with the same letter are not significantly different.

Table 6CFF change and subjective visual fatigue under independent variables in Experiment 2.

Independent variables Level CFF difference (Hz) Duncan’s multiple range testa Subjective visual fatigue

Mean SD Mean SD

Display medium Device A �2.16 1.96 16.10 8.37Device B �1.76 1.70 15.17 8.32Device C �1.38 1.90 18.12 10.10

Ambient illuminance 200 lx �2.96 1.33 B 16.57 7.58500 lx �1.38 1.90 A 15.14 9.001000 lx �1.12 1.70 A 16.40 9.851500 lx �2.04 1.99 A B 17.67 11.03

a Duncan’s multiple range test, values with the same letter are not significantly different.

Table 7ANOVA for CFF change.

Source DF Type I SS Mean square F value Pr > F

Display 2 5.56 2.78 1.63 0.2175Illuminance 3 19.97 6.66 3.89 0.0144*

Display* illuminance 6 13.05 2.18 1.27 0.2880Error 48 95.14 1.76Corrected total 59 120.68

* Significant at a = 0.05.

15001000500200

-1.0

-1.5

-2.0

-2.5

-3.0

-3.5

-4.0

Illuminance (lx)

Mea

n of

CFF

cha

nge

Device A

Device B

Device C

Fig. 6. Line plot of CFF change by illuminance and displays.

P.-C. Chang et al. / Displays 34 (2013) 208–214 213

200 lx. Fig. 6 shows that an illuminance of 200 lx resulted in great-er CFF reduction, i.e., greater visual fatigue than did 500 or 1000 lx.

3.2.2.2. Subjective visual fatigue. Values for subjective visual fatiguewere about 15–19. The ANOVA results indicated that the displaymedium and ambient illuminance had no significant effects on par-ticipants’ subjective visual fatigue.

4. Discussion

This study was designed to investigate the effects of ambientilluminance and light source on users’ reading performance and vi-sual fatigue with electronic paper displays.

With respect to the differences among the three popular elec-tronic paper displays used in this study, the results of Experiment1 indicated that reading speed and comprehension accuracy weresimilar for the three electronic paper displays. Moreover, the CFFchange and subjective visual fatigue were also similar for the elec-tronic paper displays. That is, the major brands of electronic paper

display appear comparable in terms of visual performance and vi-sual fatigue. A few reasons may explain the results. Firstly, all thethree electronic paper displays use the same E-Ink� technology.The screen resolutions for the three displays are greater than800 � 600 pixels. The ANSI/HFS standard [20] suggests that imagequality does not improve substantially for screens with resolutionsabove this range. The text-to-background luminance contrast ofthe three displays was equated at 1:5; and the text display areaand format were controlled to be identical on each display. If theseconditions are set differently according to the screen characteris-tics of the three displays, significant differences between electronicpaper displays may appear. Experiment 2 tested one such factor—the screen size. The full screen was used to present the text foreach individual display (Fig. 5). The number of text pages were30, 10, and 9, respectively, for devices A, B, and C. Device A hadthe smallest screen size and most text pages (the same as in Exper-iment 1); hence, page-turning and reading interruptions may havebeen more frequent than for devices B and C, which in turn mayhave led to the slowest reading speed. Thus, the balance betweenportability and reading efficiency should be considered in design-ing electronic paper displays. Smaller screens may be good for por-tability but disadvantageous for reading efficiency.

Compared with reading on paper, reading on electronic paperdisplays supported similar reading speed and comprehensionaccuracy. That is, electronic paper displays, especially those of E-Ink�, are acceptable media for reading. The suggestions [11,12]that reading on VDTs is slower and less accurate than reading onpaper may not hold true for electronic paper displays. Further-more, objective CFF decrements were similar for reading on paperand electronic paper displays. However, the results of Experiment1 indicated that participants experienced less subjective visual fa-tigue when reading on paper than on electronic paper displays.These results are consistent with the findings of other studies[17,2,18]. A plausible reason is that people have more experiencewith paper reading. Another possibility is that screen reflectionscommon to VDTs [19] also play a role in inducing visual fatigue.

With the portability of electronic paper displays, the ambientilluminance for electronic paper reading may vary because the usercan bring it to various places to read. The results of Experiment 2showed that ambient illuminance had a significant effect on visualperformance and visual fatigue. The reading time decreased as theambient illuminance increased. Illuminances of 1000 and 1500 lxsupported faster reading speeds than did those of 200 and 500 lx(Table 4). However, the change in the CFF indicated that a lowambient illuminance of around 200 lx may cause greater visual fa-tigue (Fig. 6). When visual performance and visual fatigue are con-sidered together, ambient illuminances around 1000 lx appearappropriate for reading on E-Ink� electronic papers. Overall, the re-sults of this study support the suggestion of many studies [8–10]that the ambient illuminance for electronic paper displays mustbe higher than that for a CRT or LCD display. Moreover, the effect

214 P.-C. Chang et al. / Displays 34 (2013) 208–214

of ambient illumination appears to be greater in a long-durationreading task (e.g., this study) than in a short-duration word identi-fication task [6,7], where the effect of ambient illuminance is usu-ally nonsignificant.

Fluorescent light (TL84) and sunlight (D65) support similar lev-els of visual performance and visual fatigue when reading on elec-tronic paper displays. Both may be proper light sources forelectronic paper displays. These results are similar to the findingsof many studies [7,17,1,2]. Electronic paper displays appear to beappropriate reading media both in general offices where fluores-cent light is the main light source; as well as outdoors where sun-light is the light source.

5. Conclusions

In this study, users’ visual performance and visual fatigue wereinvestigated when reading using different electronic paper dis-plays under different ambient illuminances and light sources. Thefindings and suggestions are as follows:

(1) When the text display areas are identical, the three elec-tronic paper displays are similar and comparable to paperin terms of visual performance and visual fatigue.

(2) With the full screen used for text presentation in each indi-vidual display, the three electronic paper displays are differ-ent, and displays with larger screens and hence fewer textpages support faster reading speeds.

(3) Both sunlight (D65) and fluorescent light (TL84) are properlight sources for reading on electronic paper displays.

(4) Ambient illuminances of 1000 and 1500 lx support bettervisual performance, and those of 500 and 1000 lx cause lessvisual fatigue than an illuminance of 200 lx when reading onelectronic paper displays.

Acknowledgment

This study was supported by a research grant from the NationalScience Council of the ROC, Grant No. NSC 98-2221-E-161-009-MY2.

References

[1] H. Isono, S. Takahashi, Y. Takiguchi, C. Yamada, Measurement of visual fatiguefrom reading on electronic paper, IDW’04 (2004) 1647–1648.

[2] S.C. Jeng, Y.R. Lin, C.C. Liao, C.H. Wen, C.Y. Chao, D.S. Lee, K.K. Shieh, Influencesof ambient illumination on legibility of electronic paper, in: The 5thInternational Meeting on Information Display, COEX Seoul, Korea, 2005.

[3] M.G. Helander, B.A. Rupp, An overview of standards and guidelines for visualdisplay terminals, Applied Ergonomics 15 (1984) 185–195.

[4] O. Ostberg, Accommodation and visual fatigue in display word, in: E.Grandjean, E. Vigliani (Eds.), Ergonomic Aspects Of Visual Display Terminals,Taylor and Francis, London, 1980, pp. 41–52.

[5] W. Xu, Z. Zhu, The effects of ambient illumination and target luminance oncolour coding in a CRT display, Ergonomics 33 (1990) 933–944.

[6] Y.T. Lin, P.H. Lin, S.L. Hwang, S.C. Jeng, Y.R. Lin, Ergonomic evaluation ofelectronic papers: influences of anti-reflection surface treatment, illumination,and curvature on legibility and visual fatigue, Journal of the Society forInformation Display 16 (2008) 91–99.

[7] D.S. Lee, K.K. Shieh, S.C. Jeng, I.H. Shen, Effect of character size and lighting onlegibility of electronic papers, Displays 29 (2008) 10–17.

[8] K.K. Shieh, D.S. Lee, Preferred viewing distance and screen angle of electronicpaper displays, Applied Ergonomics 38 (2007) 601–608.

[9] A.H. Wang, H.T. Kuo, S.C. Jeng, Effects of ambient illuminance on users’ visualperformance using various electronic displays, Journal of the Society forInformation Display 17 (2009) 665–669.

[10] A.H. Wang, C.C. Tseng, S.C. Jeng, K.I. Huang, Effects of electronic displaymedium and inclination on users’ comprehension under various ambientilluminance conditions, Journal of the Society for Information Display 16(2008) 101–106.

[11] A. Dillion, Reading from paper versus screen: a critical review of empiricalliterature, Ergonomics 35 (1992) 1297–1326.

[12] J.D. Gould, L. Alfaro, V. Barnes, R. Finn, N. Grischkowsky, A. Minuto, Reading isslower from CRT display than from paper: attempts to isolate single variableexplanation, Human Factor 29 (1987) 269–299.

[13] J.M. Noyes, K.J. Garland, VDT versus paper-based reading: reply to Mayes,Sims, and Koonce, International Journal of Industrial Ergonomics 31 (2003)411–423.

[14] D.K. Mayes, V.K. Sims, J.M. Koonce, Comprehension and workload differencesfor VDT and paper-based reading, International Journal of IndustrialErgonomics 28 (2001) 367–378.

[15] W.H. Cushman, Reading from microfiche, VDT and the printed page:subjectivefatigue and performance, Human Factors 28 (1986) 63–73.

[16] H. Heuer, G. Hollendiek, H. Kröger, T. Römer, Die Ruhelage der Augen und ihrEinflß auf Beobachtungsabatandund visuelle Ermüdung bei Bildschirmarbeit,Zeitschrift für experimentelle und angewandte psychologie 36 (1989) 538–566.

[17] I.H. Shen, K.K. Shieh, C.Y. Chao, D.S. Lee, Lighting, font style, and polarity onvisual performance and visual fatigue with electronic paper displays, Displays30 (2009) 53–58.

[18] D.S. Lee, Y.H. Ko, I.H. Shen, C.Y. Chao, Effect of light source, ambientillumination, character size and interline spacing on visual performance andvisual fatigue with electronic paper displays, Displays 32 (2011) 1–7.

[19] T. Läubli, W. Hünting, E. Grandjean, Visual impairments related toenvironmental conditions in VDU operators, in: E. Grandjean, E. Vigliani(Eds.), Ergonomic Aspects of Visual Display Terminals, Taylor & Francis,London, 1982, pp. 85–94.

[20] American National Standard for Human Factors Engineering of Visual DisplayTerminal Workstations (ANSI/HFS Standard No. 100-1988), Human FactorsSociety, Inc., Santa Monica, CA, February 4, 1988.