Summary This paper complements the data presented in phase I of this study. This paper looks at three new escape lighting systems, and compares them with three systems tested previously. In addition to testing their suitability for 30 people with different visual impairments, results are presented for 12 normally sighted people. Using the facilities at the Building Research Establishment, it has been possible to examine the performance of ceiling-mounted emergency luminaires providing two levels of illuminance, LED strips in two configurations, LED strips with overhead luminaires and miniature incandescent wayfinding strips. Speeds of movement are presented here, along with subjective responses. Comparisons are made with findings from normally sighted people. There are no significant differences in the walking speeds under the lighting systems, but the visually impaired subjects had significantly differentopinions on the ease with which they could see where to go. Emergency lighting and wayfinding provision systems for visually impaired people: Phase II of a study G K Cook&dagger; BSc PhD, M S Wfiot&dagger; BSc PhD, G M B Webber&Dagger;&sect; BSc PhD and K T Bright&dagger; MSc &dagger;University of Reading RG6 6AW, UK &Dagger;BRE, Garston, Watford, Herts. WD27JR, UK &sect;Consultant, Bricket Wood, St. Albans. List of symbols 2 I L The test facility at BRE is as described in the previous paper(’). Six lighting systems were tested, as follows. Miniature incandescent wayfinding system LED wayfinding system L2 LED wayfinding system variant 2 LO LED wayfinding system with overhead emergency lighting 0 Basic overhead emergency lighting 02 Enhanced overhead emergency lighting 1 Introduction Following our previous papery which answered general questions about the suitability of wayfinding systems and highmounted emergency luminaires for visually impaired people, it was decided that there was a need to look at specific design details of lighting systems. This was prompted by some of the conclusions from phase I, particularly that visually impaired people found it easier to see where to go under the three powered wayfinding than under the basic highmounted emergency luminaire system that slightly exceeded the illuminance recommendations of British Standard BS5266(l). One issue we wanted to address was whether significanuy increasing the light output of the overhead emergency lighting system would significantly improve visually impaired people’s opinion of the traditionally high-mounted luminaires. We also wanted to investigate different design attributes of a powered wayfinding to see how important some design are to people’s In addition we needed to assess the effect combining wayfinding systems and British Standard emergency lighting on the performance of visually impaired subjects. This paper presents of a study carried out by the University of Reading in collaboration with the Building Research Establishment (BRE) using the BRE facility. This paper details the from the second set of 30 visually impaired people, looking at six lighting systems: three common to tested in phase I, and three new systems. This describes the experimental facility, the procedure and the characteristics of the volunteers. The findings are presented for visually impaired and normally sighted volunteers, and the conclusions of the study are discussed. Experimental facility 21 jBaMc oc~~K~ ~M~giMcy %~~ (Q) This system consisted of five 4 W fluorescent luminaires mounted on the ceiling, and one high on the wall at the bottom of the stair. There was a fluorescent pictogram exit sign (160 mm in height, BS5499 Part 1 format<3), luminance 935 cd m-2) on the mock exit door, at a height of about 1800 mm. This system was the as phase I, but with new lamps that increased the illuminances by about 40%. 2.2 LED wayfinding ~~M (L) Identical to that in phase I, this system consisted of upwarddirected LEDs providing a line of markers along the route, and downward directed LEDs to illuminate the floor. 2.3 Miniature MesM&K~Kt nM~M~ system (1) This was identical to that in phase I. 2.4 JEMMMc~ (02~ This system consisted of five 8 W fluorescent luminaires mounted on the and one high on the wall at the of the stair. same exit was used as with emergency lighting system 0. 2.5 LED wayfinding Vst--m ’Daritmt 2 (L2) The full LED system, L, was modified to the provision of upward- and downward-directed LED strips on one side of the long straight corridor compared to both sides. A second modification was to exclude the LEDs in the nosings of the stair. Elsewhere the L2 system was the same as L, including the directional arrows on both sides of the corridor. The sign on the exit door was the LED pictogram (at height 1380 mm, size 80 mm high) consisting of LEDs developing 140 mcd. Downloaded from lrt.sagepub.com by JC CONCEPCION on May 14, 2015 43 Table 1 Mean horizontal illuminance (lu~) on the floor at the centre line along the escape route LEDzvqyfihdilqwsum zvimwomdmneqa*yIi#hGng (LO) The LED system variant 2, L2, was modified to investigate the combination of marker LEDs (excluding illuminator LEDs) with high-mounted emergency lighting to illuminate the floor. In the long corridor the marker LEDs were provided only on one side, elsewhere the marker LEDs were on both 2.6 sides of the route. The mean illuminances along the route are shown in Table 1. 3 the chart for phase I, or 2.65 Ix for the re-lamped 11. The second test was under normal lighting, with a mean illuminance of 92.1 Ix on the chart. The rationale behind allowing the volunteer to get as close to the chart as 1 m, compared to the conventional 6 m used by optometrists, is that in a building people would move as clc~s~ as possible to read a sign that was unclear if they thought it contained useful information. The cumulative percentages of subjects who could read an entire line are plotted against the visual angle subtended by the height of the letter in Figure 1, for both sets on Experimental procedure The experimental procedure was identical to that described in phase I. Thirty subjects with visual impairments carried out the tests. There were 14 men and 16 women, of average age 53.2 years, with a range of 36-73 years. There was only one person older than 66 years, and she was not noticeably slower than the other subjects. Ten people used a white cane; one of these had a long cane. A simple vision test using a Snellen chart was carried out under lighting conditions that provided vertical illuminance of 92lx. It was found that 12 of the 30 volunteers had vision worse than 3/60. Six people had tunnel vision, five had peripheral vision and three people had sporadic vision. The of eye conditions exhibited by these 30 volunteers was wider than for phase I, with more of multiple eye problems. People with macular degeneration formed the largest group, with seven of the volunteers having this eye condition; two of them also had other complications. The volunteers also had glaucoma, retinitis pigmenrosa, cataracts, optic nerve damage, detached retina as well as other eye conditions. Of the 30 people, 23 were registered blind, six were partially sighted (one of whom could be registered blind), and there was one person who could legally be registered partially sighted. Comparing the volunteers from n with those in phase I, there are two main attributes to look at - age and visual acuity. The mean age of the participants for phase II is 53.2 compared to 51.6 for phase this is not a significant difference under a t-tesL are different gender mixes in the two groups, but not radically so (18 men in phase 1,14 for phase W. The noticeable difference is that 18 people used a cane in phase I, but only 10 in n. When comparare using the data from the vision the results are different. Two vision tests carried out, using different Snellen visual acuity charts. The volunteer stood at 1.5 m to read the chart; a successful letter acuity was achieved if all the letters on a line were correctly read After the line with the letters had been read, the volunteer was allowed to stand at 1 m to see which line could be completed at that distance. The volunteer was allowed 3 minutes to complete this test. The first eye test was carried out under the British Standard emergency lighting system (0) with 1.43 Ix incident 44 1 The relationship between the cumulative percentage of subjects who could read an entire line of a Sndlen chart and the angle subtended at the eye by the height of a letter on that line. The top figure presents results measured under emergency lighting (1.43 or 2.65 Ix on chart), the lower figure presents results measured under normal lighting (92.1 Ix on chart). Figure Downloaded from lrt.sagepub.com by JC CONCEPCION on May 14, 2015 of volunteers. It can be seen that people from phase I had, on average, worse visual acuity than those from phase n. This is also shown by the fact that five (eleven) people could read the top line of the chart under normal (emergency) lighting in phase I compared to one (two) people from phase U. The top letter on the chart is 87 mm in height, and subtends an angle of 5° at a distance of 1 m. Because there are no data for the visual angle that people can see, it is impossible to carry out t-tests to compare the two groups. However, when carrying out a rank sum test (the Wilcoxon-Mann-Whitney tesd4», the visual acuity of the people from the two groups is found to be significantly ~nt. ~ II people have better visual acuity than phase I people, with p < 0.025 for emer0.05 for normal gency lighting illuminating the chart, andp < lighting. Results from 12 normally sighted people are included in this paper. These sighted subjects were 4 men and 8 women, of average age 51.25 years, with a range of 42-61 years, which is not significantly t from the age of the visually impaired volunteers. The results of the vision tests for the normally sighted subjects are also plotted in Figure 1, down to the minimum angle of letter height that was measured, 0.28°. 4 Findings The results that are presented here come from two sources. The objective results are speeds that are calculated from times taken froan video recordings from the low-light-sensitive cameras. The subjective analysis comes from the answers to the questionnaire, which is described in Section 4.3. Throughout the following sections, the lighting systems are referred to by the code letter used in the headings in Section 2. The statistical tests that were carried out to compare the results from different lighting conditions depended on whether the data was normal or skewed. Whenever the data sets that being compared were skew, non-parametric tests were carried out. For skew pair comparisons, a sum test, the ’~i.lcc~~c~ir-.t~fann ’tn~ tesf4), was performed instead of the more usual parametric ~-t~t. The results of the tests are reported up to significance levels of 0.05. Further statistical tests were carried out - for non-skewed data the .~ te.st, and for skewed data the Kruskal-Wallis one-way analysis of variance by ranks test’4) - to ensure that the multiple t-tests were not including insignificant results. Results are presented when the relevant tests suggest that there are differences between the data sets at a significance better than 0.05. 4.1 speeds measured for the different lighting systems shown in Figure 2. These are shown for three sections of The mean are the route: - - - Getting up from the chair to leave the office ~o~ce outward) Start of the landing to the bottom of the stair (stairway outward) From the bottom of the stair to the mock exit door (corridor outward) The speeds were also measured when the subject returned. These sections are identical to those included in phase 1. None of the speed data sets was skewed in phase l3, so parametric tests could be carried out. o~ctwctrd and office return As can be seen in Figure 2, the slowest part of the route is leaving the o~ce, as was found for phase I. Again this might not be a function of the lighting systems, merely the effect of the sudden change from 400 Ix to an illuminance in the range £D.7~-6.5 l~ disabling peoples eyes. The claim that it is not the particular lighting system that is at fault, but the sudden reduction in illuminance, is confirmed by only six of the 208 runs being slower returning through the office than leaving it, , and that 47.6% of the runs are at least twice as fst on returning in the office. This occurs despite the fact that people are heading into a relatively dark room, instead of heading to an illuminated doorway. It is interesting that these results are similar to those from phase I, despite the range of eye conditions being very different. The F-test showed no significant differences between the speeds for the lighting conditions, either outward or return. On carrying out a one-tailed related t-test to compare the speeds in the o~ce on the outward and return parts of the route, it is clear that the subjects are quicker when returning (p < 0.0005 for all lighting conditions), when they know the route. However, it is not just a route learning process, as the above results are found if only runs 4 to 7 are examined. At this stage the test subjects ought to be familiar Speed ~, f e~r in different sections off ~r~e speeds are calculated from all completed runs. If any incomplete, the data from the whole run were ignored. Findings for the visually impaired suhj are presented for 208 completed runs; there were two incomplete runs by two subjects involving 0 on both occasions. In both the incomplete runs occurred the first time the volunteers attempted the route, and because the volunteers thoughtt they had finished the route and returned before they had reached the mock exit door. As mentioned earlier(’)., this is unlikely to happen in a real situation where there is likely to be less ambiguity about what constitutes the end of the escape route, as this will generally be outside the building. Obviously, in a real-world situation people would not return along an escape route, once having reached the (supposed) exit door, but would use it to leave the building. Mean runs was 2 A graph of mean walking speeds of people who are visually impaired passing through different parts of an escape route under six different lighting conditions: 0, basic overhead emergency lighting; U LED wayfinding system; I, incandescent wayfinding system; 02:, enhanced overhead emergency lighting; L2, LED wayfinding system variant 2; LO, LED wayfinding system with basic overhead emergency lighting. Figure Downloaded from lrt.sagepub.com by JC CONCEPCION on May 14, 2015 45 with the route, which indicates how slowly visually impaired people start this route, even when they have completed the route three times. Stairway ~ru~d and st~ir~ay r and ~~midar ~~c~rd and corridor return The F.test showed no significant differences between the speeds for the lighting conditions, either outward or return. 4.2 Comparisons of arr~urznce with volunteers ta normally sighted collected from 12 normally sighted people in the way as for the 30 visually impaired volunteers. The lighting systems are identical for the two groups. As for phase I, the comparison between normally and visually impaired people is achieved by finding the ratio of the mean speeds in t parts of the route; this is shown in Table 2. were same Generally, the visually impaired subjects walk at 0.54 to 0.86 the rate of a normally sighted person; this compares with 0.43 to 0.87 from phase I. The faster relative speeds on the stair that occurred in phase I did not happen in phase II. When unrelated t-tests were carried out to compare normally sighted people and visually impaired people on all sections of the route for all lighting conditions, it was found that the normally sighted walk significantly faster, with a significance of, at worst, p < 0.05. However, there were only five occasion out of the 60 comparisons when the significance was worse than 0.005, twice for L (stairs out and corridor return, p < 0.01), twice for 02 (corridor return, p < 0.05, and stair return,p < 0.025) and once for LO (corridor return,.? < 0.01). Two measures are used to compare the lighting systems. First, if the number of times that the speeds on the ten parts of the route are not as significantly different as < 0.0005 is counted, it is found that the system with the highest score, 6 times out of a possible 10, is for 02. The next system that causes the visually impaired subjects’ walking speeds to be less significantly different to normally sighted speeds is L, with four parts of the route being not as significant as < 0.0005. Another way of assessing which lighting system helps visually impaired people die best in comparison to normally sighted people is to count the number of times that a ~ght~ng system has the highest ratio in Table 2 for er~nt parts of the route. There are only two lighting systems that have the highest ratio: 02, seven times, and L, three times. These findings suggest that these two help visually impaired people to a greater extent in comparison to normally sighted people. It should be remembered that there are no significant differences in the mean walking speeds under any of the lighting conditions for the visually impaired volunteers. Referring again to the ratios in Table 2, there are two parts of the route where visually impaired people perform relatively badly in comparison with normally sighted volunteers. As in phase I, the landing outwards part of the route causes visually impaired people to slow down more than on other parts of the route; this is true for all lighting conditions and may indicate the problems in identifying the top of the stair when the eye is still undergoing changes in dark adaptation. The other part of the route that causes problems is returning into the c~~.c~, which is the worst for all lighting systems on die return part of the journey. 43 ~’z~rj~c ° responses As for phase I, two questions on di~culty were asked The questions were asked and responded to orally: they were How difficult did you find it to see where you were going on .1..;> the stair? How difficult did you it to see where you were going in the ~~~ with a response on a seven-point scale: 1 cult. = very easy, 7 = very Some of the responses to these questions were skewed. The non-normality of the data is due to people having limits to their choice, so for systems that score averages close to the top or bottom limit the distributions cannot be symmetrical. Rank sum and Kruskal- W allis tests were therefore used for these comparisons. Results for both questions were shown to have significant differences between lighting conditions when tested using die Kruskal-Wallis metric. The mean scores for culty on the stair, against the mean illuminances on the stair, are shown in Figure 3. Lower scores, below 4, indicate the lighting system made it easier for people to see where they were going; the error bars indicate the standard error. Figure 3 shows that the visually impaired volunteers thought it hardest to see where they were going using lighting system 0; it is significantly worse than four of the other systems, with significances ofp < 0.0005 for 02 and LO, p < 0.005 for L, ~.nd~ < 0.025 for L2. The next highest mean score is for I and is significantly worse than three of the other systems, 02 (p < 0.0005), LO (~ ~ 0.005) and L (p < 0.01). The third highest (and so fourth best of six) is L2, which is Table 2 The mean ~~r visually impaired subjects divided by that for no y ’ t~d subjects, for different lighting syss Figure 3 The relationship between the mean degree of difficulty of seeing where to go against mean illuminances on the stair for the six lighting condi1ions. The error bars show the standard error. 46 Downloaded from lrt.sagepub.com by JC CONCEPCION on May 14, 2015 Table 3 Mean scores of the answers to ’How uncomfortable did you find the brightness of the lighting system?*, 1 = uncomfortably dim, 7 uncomfortably bright, for visually im and normally sighted volunteers. Final column shows the mean illuminances along the entire escape route significantly worse than 02 (p < 0.005), LO (p < 0.025) and L ~p ~ 0.05). Hence, the basic British Standard overhead emergency lighting system, 0, is regarded as making it significantly = more difficult to see where to go than four of the other five syswhether the systems are regarded tems tested. When easier or mare ’ cult in as being where to assessing significantly seeing go than a neutral score of 4, the results are that L, 02 and LO are d~~nitel~ ~ier ~ ~ 0.0005), as is L2 (p < 0.005). The differences in the scores for the stair between L and L2 is significant at p < 0.05, which seems to indicate that the six in the stair nosings are preferred to some extent. When comparing systems L2 and LO, the system with the overhead lighting (LO) is pr~f d, even though there is less iii nance on the treads of die stairs. System 02 scores significantly lower than 0, indicating the visually impaired subjects preferred the increased light from the enhanced emergency luminaries, even if there is no significant difference in mean walking speeds. The minimum illuminances along the stair are system O. The minimum illuminances along this part of the mute are 2.92 lx and 0.73 Ix, respectively. The subjects were also asked about the brightness of the systems; they were asked Hozv r~r~c~mfi~bl~ did you find the brig of the lighting 3.93 Ix for 02 and 1.08 Ix for 0. system? with a response The findings for the mean scores for difficulty in the corridor are shown in Figure 4, plotted against the mean illuminance in the long corridor. As for the stair, system 0 scores highest (worst), and is significantly worse than 02 (p < 0.0005), and LO and L (p < 0.005). System L2 scores significantly worse than 4~2 ~r ~ 0.005), LO (p < 0.025) and L (p < 0.05). System I also scores significantly more difficult in seeing where to go than do ~2 ~ ~ 0.01), LQ ~p ~ 0.025) and L (p < 0.05). Tests of whether the mean scores sign~ t1y erent from a neutral score of 4 show t systems L, 02 and LO are definitely easier (p < 0.0005); as for the stair, I is now easier (p < 0.005), as is L2 (p < 0.01). seven-point scale: 1 uncomfortably dim, uncomfortably bright. The mean scores are presented in Table 3, along with the mean illuminances along the route. For the visually impaired, none of the mean scores is above 4, which is the ’happy medium* score, neither too dim nor too bright for ~or~c~rt. For the visually impaired subjects, 7 Kruskal- W allis test indicates that there are significant erences between the scores for erent lighting conditions. Four of the systems have a mean score that is significantly lower than 4 (0 and I, p < 0.0005; L2, p < 0.005; and LO, p < 0.025), so these systems are regarded as more dim than is comfortable. Comparing the ~r~nt lighting systems, system 0 is regarded as significantly dim compared to all other systems (p < 0.0005, apart ~m I for which ,p < 0.025). The system that is ranked as being the next dim (although it has a mean illuminance less than 0) is I, and this is significantly er~nt from L (p < 0.01), 02 (p < 0.025) and LO (p < 0.05). The only other system that is regarded as t is L2, which scores significantly lower dm L (p < 0.05). These results indicate that L is reported as the least dim system (closest to 4), despite having a mean illuminance along the centre line of the floor less than half that of 02. It should be noted that this system, along with most other wayfinding provision systems, produces light near floor level that is directed up towards the eye, so the measure of its illuminance along the centre line of the escape path on the floor may not be the best measure of the brightness perceived by the eye. ’1 4.4 * vtr~ crf ~~~r~~ Figure 4 The relationship between the mean degree of difficulty of seeing where to go against mean illuminance in the long corridor for the six lighting conditions. The error bars show the standard error = a In the long corridor system, L2 had marker and illuminator LED tracks on one side of the corridor only, while system LO had a marker track on one side together with the basic overhead emergency lighting system 0. System L had marker and illuminator LED tracks on both sides of the corridor. The lighting arrangements presented by systems LO and L are preferred to that of system L2. In part, the preferences towards systems LO and L may be attributable to their higher mean illuminances on the floor, 2.19 Ix and 0.91 Ix, respectively, compared to system L2, 0.69 bL The increased illumination from the enhanced emergency luminaires, 02, is preferred to the basic British Standard on a = *~e responses with normally sighted When comparing the results of the visually impaired volunwith those of 12 normally sighted people, it is found, surprisingly, t there are no significant differences between the mean scores by the n groups for five of the six lighting conditions. The only lighting system that has better scores from the normally sighted volunteers than from the visually impaired group is L2. For the stair, the significance isp < 0.005: for the corridor it is p < 0.025. This is in contrastt to the results from phase I, where systems L and 0 are scored worse by the visually impaired subjects. This seems to indicate that the scoring its not absolute, but relative. Coinddentally, when F-tests are carried out, the result is that there are no significant erences between the scores given to the six lighting conditions by the 12 normally sighted people. This seems to indicate that normally sighted people are less demanding in their requirements for emergency lighting systems than are visually impaired people. Downloaded from lrt.sagepub.com by JC CONCEPCION on May 14, 2015 47 For the question about the brightness of the systems, as for the visually impaired, none of the mean scores for the differentt lighting systems was above 4. The mean scores of the normally sighted people are presented in Table 3. When making comparisons with the data from the visually impaired subjects, there are no statistically significant differences between the mean scores. The rank order of the lighting systems is the same for the two groups, with the one slight ~ that 02 and LO are given the same mean score by the normally sighted group. It should not be surprising that the subjective brightness rank order of the systems is unchanged between the two groups; what is perhaps surprising is the similarity of the mean scores. When the data from the normally sighted group are tested using an F-test, there is no suggestion of significant differences between the scores of the lighting systems. There are four systems that score significantly less than 4, indicating they are too dim for comfort. These systems are 0 (,~ < 0.005), I and L2 (p < 0.01) and 02 (p < 0.025). 4.5 Comparisons cr~’ ~lun ~ ,~ with phase I visually impaired Comparisons with normal lighting are a useful measure of how successful emergency lighting is: if there is no significant difference, the lighting system is thought to be as good as possible. ~nsequ~ntl~, it was hoped that the results frum phase I could be used. To see whether this is possible, it is necessary to check that the volunteers ~m phase I and phase II are directly comparable. It is seen from Section 3 that there are differences in visual capabilities and cane use between the two groups of people who carried out phase I and phase II of this study. However, appropriate pair comparison tests between the data frorn phase I and n revealed that there are no significant differences in the speeds for the o~ce, stairway or corridor parts of the route for the three systems (0, L, I) in common in phases I and II. ’ If the subjective views for the three questions in section 4.3 are compared, there are no significant differences for 0 or L, but for the miniature incandescent system two of the questions are scored significantly better for phase I than for phase I~l. The ~ty of seeing where to go on the stair is scored dfferendy (p < 0.025), as is the uncomfonableness due to the brightness of the system (p < 0.005). The differences in subj we scores also mean that comparisons with absolute values (for example, the neutral position of 4) are dif5erenL- It is to be noted that although people who were tested in phase I had worse visual acuity than those of phase H, they scored one system better. Although the difference between the two phases are small, there are some between the two g when comparisons are drawn between the lighting conditions. For this it was decided not to compare iigltting systems between phases, as my possible differences could be to using a group of 30 people father than to innate differences in the lighting systems. 5 Condusions The subjective results for phase II indicate that visually impaired people find that traditional basic emergency luminaires do not make it as easy to see where they are going as do the brighter overhead (enhanced) emergency luminaires, or as is the case with a combination system of marker track wayfinding provision and basic emergency luminaries, or a ~ waynnding system with marker and illuminator s. This finding is similar to those when when a ia incandescent wayfinding system is compared with the same three systems, although with less significant differences. A conclusion is that visually impaired people prefer an enhanced overhead lighting system that produces a minimum of 2.5 Ix the level parts of the route and 3.9 Ix on the stair, to a basic system that has a minimum of 0.73 Ix on the level parts of the route and 1.1 Ix on the stair. They also prefer the full LED wayfinding system to the basic overhead emergency lighting. Adding a low-mounted track of upward-pointing LEDs to this basic emergency lighting system is seen as an improvement. It is found there is a slight pr~f’er~nce (at a significance of p ~ 0.05) for the LEDs in the nosings of the stair. There is a slight preference for the light tracks on both sides of an escape path compared to on one side only (at a significancep < 0.05). The LED wayfinding system with upward-pointing markers combined with overhead emergency lighting is preferred to the LED wayfinding system with upward-pointing markers and downward-pointing illuminators. A BRE Information for recommendations on erPaper has been published<5> and gency lighting wayguidance provision four people with impaired vision based on this research. on The walking speeds of visually impaired people under powered emergency escape route lighting conditions are generally about 55-85% of the speed of the normally sighted. The higher illuminance produced by the enhanced overhead emergency lighting system gave greater assistance to the visually impaired than to the normally sighted people. Both the normally sighted and visually impaired subjects placed the basic overhead lighting system last in terms of their subjective opinions. Acknowledgements ’ This work has been funded by the Sciences Research Council. Engineering and Physical References 1 2 3 4 The mean walking speeds that visually impaired people achieve under different powered emergency lighting condi- 48 tions have been shown to be generally very similar, as reported in phase I. 5 Wright M S. Cook G K and Webber G M B Emergency lighting and wayfinding provision systems for visually impaired people: phase I of a Lighting Res. Technol. 31(2) 35-42 (1999) study BS 5266 Part 1 Code of practice for the emergency lighting of premises (London: British Standards Institution) (1988) BS 5499 Part 1 Specification for fire safety signs (London: British Standards Institution) (1990) Seigel S and Castellan NJ Nonparametric statistics for the behavioural sciences (New York: McGraw-Hill) pp128-136, 206-212 (1988) Webber G M B, Wright M S and Cook G K Emergency lighting and wayfinding systems for visually impaired people BRE IP9/97 (1997) Downloaded from lrt.sagepub.com by JC CONCEPCION on May 14, 2015