Modulation of fluorescent light: Flicker rate and light source effects on visual performance and visual comfort

Download (0)

Full text


Publisher’s version / Version de l'éditeur:

Lighting Research & Technology, 29, 1, pp. 54-56, 1997


Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à

Questions? Contact the NRC Publications Archive team at If you wish to email the authors directly, please see the first page of the publication for their contact information.

NRC Publications Archive

Archives des publications du CNRC

This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at

Modulation of fluorescent light: Flicker rate and light source effects on

visual performance and visual comfort

Veitch, J. A.; McColl, S. L.

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

NRC Publications Record / Notice d'Archives des publications de CNRC:


M odula t ion of fluore sc e nt light : Flic k e r ra t e a nd light sourc e e ffe c t s

on visua l pe rform a nc e a nd visua l c om fort

N R C C - 4 0 6 0 3

V e i t c h , J . A . ; M c C o l l , S . L .

J a n u a r y 1 9 9 7

A version of this document is published in / Une version de ce document se trouve dans:

Lighting Research & Technology, 29, (1), pp. 54-56, 1997

The material in this document is covered by the provisions of the Copyright Act, by Canadian laws, policies, regulations and international agreements. Such provisions serve to identify the information source and, in specific instances, to prohibit reproduction of materials without written permission. For more information visit

Les renseignements dans ce document sont protégés par la Loi sur le droit d'auteur, par les lois, les politiques et les règlements du Canada et des accords internationaux. Ces dispositions permettent d'identifier la source de l'information et, dans certains cas, d'interdire la copie de documents sans permission écrite. Pour obtenir de plus amples renseignements :


Corresporuletzu Refere""es

1 Augdal A Equivalent veiling luminance: Different Mathematical approach to calculation LightingRes.TechtwL23(1) 91-93 (1991) 2 Adrian W K Adaptation luminance when approaching a tunnelin

daytimeLightinglW. T«hnol.19(3) 73-79 (l9S7)

RClear Lawrence Berkeley National Laboratory University ofCalifomia Berkeley, California 94720 USA

Evaluations of in-traffic perfonnance of high.intensity dis-charge headlamps

It would have been useful for the authors of the above paper(l) to have included tests on the effect ofHID headlamps on oncoming traffic, and visibility in fog.

How does oncoming traffic perceive the difference in glare between hid and tungsten headlamps in the following cir-cumstances:

dipped headlights on s straightdryroad?

dipped headlights on adryroad when crossing a bridge or road-calming hump

which changes the elevation of the beam to cause maximum discomfort for a short time?

dipped headlights on adryroad when driving around a nearside bend?

the above with wet roads?

allthe abovewithundipped headlamps to simulate poor-ly adjusted headlamps or selfish drivers who do not dip their lights?

How do the different spectral compositions of the lamp out-put affect the back-scatter and visibility in mist, smog and fog of different densities and globular sizes?

ALIsaacs 46Brarnber Road Nonh Finchley LONDON N129NE UK Authors'reply

We agree that possible differences in the glare effects of high-intensity discharge(HlD)and tungsten-halogen(TH) head-lamps are worth a look. We have done some work on the dif· ferential effects of these lamps on the subjective discomfort aspects of glare(3,4). These studies indicate that subjective dis-comfortisequal when light fromTHlamps is about1.5times greater than light fromHIDlamps, in terms of photopic lux at the eye. The possibility of a difference in effects on objective seeing ability (often called 'disability glare') has not yet been studied. Previous investigations of other types of lamp have shown that differences in subjective discomfort do not neces-sarily predict differences in disability glare (most notably in the case of the yellow headlamps that were mandated in France from 1936to 1993;see for example Reference 5. Furthermore, in ,our judgement, there are not strongエィ・ッイ・エゥセ

cal reasons to expect differences in disability glare. Nevertheless, we believe that possibility should beゥョカ・ウエゥァ。エセ



3 Flannagan M, SivakM,GeUady A W and LuomaJ aセャ、ウエオ、ケッヲ、ゥウセ

comfiH! glareftvmhigh-intensitymscllarfle headlamp, ReponUMTRI-92-t6

(Ann ArbOr: University ofMichiganTransportationResearch Institute)


4 Flannagan MJ,Sivak Mj Battle D SI Sato T and Traube E C

Discomfort glare from high-intensity discharge headlamps; Effectsofcontext

and experienceReport UMTRI·93.10 (Ann Arbor: University of MichiganTransportationResearch Institute) (1993)

5 Jehu VJ Acomparison ofyellow and white headlamp beams Lightand Lighring47 2S7-291 (1954)

M Sivak, MJFIannagan,E C Traube, D S Battle andY Sato The University ofMichigan TransportationResearch Institute

2901 Baxter Road, Ann Arbor MichiganTXQPYセRQUP


Modulation of fluorescent light: Flicker rate and light source effects on visual perfonnance and visual comfort The study by Veitch and McColl(6) ('the authors') primarily investigated the consequences of flicker in fluorescent lamps on a visual task, with a secondary evaluation of the effects of light spectrum. Examination of the protocols and data reveals that the primary study shows an effect that is probably greater than that reportedinthe paper, while. the secondary study is weaker than described.

In this work both the horizontal illuminance at the task and the task luminance were measured. The luminances indicated in Table 1(6) are generally about15% higher for the low-fre-quency(LF)operation than for the high-frequency(HF) opera-tion, even though the task horizontal illuminance is settothe same value for bothLFandHFconditions. The authors note this discrepancy and srate that it might be due to 'artefacts of the speed of the detector artay' of the luminance instrument (Photo Research, Pritchard Model 703A, Northridge, CAl. However, the instrument maker, in telephone conversations, claims that this model records accurately in bothLFandHF conditions. Likewise, the United States representative for the Hagner illuminance meter (Cooke Corp, Hagner Model S2, Buffalo, NY) claims their meterisaccurate when measuring bothLFandHFsources. A possible explanation for the dis-crepancy between results from the meters could be electroR

magnetic interference(EMI)from theHFballasts. In Figure 2") the illuminance probe is shown resting JUSt above the lamp housing, and might be quite near the ballast. Various degrees ofEMIcan occur, depending on the quality of grounding and the amount of shielding of the fixture, as well as the integrity of the shielding of the probe-to-meter cable (since the electri-cal current carried by the probe cable is very small). Cooke Corp. also indicated to us that meters have malfunctioned under high-EMI conditions associated with high-voltage switching. Since the ta!lk luminance is the psychophysically appropriate measure of stimulus, and the luminance meter was placed much further away from the ballasts with less opportunity for anEMIproblem, we consider its values to be the more representative of the test lighting conditions. In which case, accordingtoTable 1(6), the subjects are provided with 13% to 22% less luminance under HF conditions,


yet they perfonn better on the visual task. Furthermore, the data in Table 2(6) show that the effect is present for each lamp, despite the fact that each lamp's effects were studied with a separate set of16subjects. This result, which might be an 1-irrbif'!!r・セN・Hャイ」ィ '.mel.Techn01oQ"!


unexpected benefit of theHFoperation, deserves furth'er care-fulstudy.

The primary purpose of the study (to determine the effects of ballast operation frequency on threshold size and· contrast) can be inferred because it is theLF/HFcomparison that is studied with the statistically strong within-subjects design. Secondarily, the authors have made some statistically weaker between-group comparisons which they interpret as a test of the 'Berman hypothesis' regarding the effects of the scotopic spectrum on visual performance. The secondary experiment provides no such test, for several reasons.

Firstly, in all our papers(7-10)onthe benefits ofscotopic sensi-tivity, theexperimentswere conducted under full-field view-ing conditions with uniform illumination. However, the test-ing apparatus in the。オエィッイGウセ・イゥュ・ョエ restricted the prima-ryvisual field of view (due


the viewport dimensions) to 0.28 sr (about 5% of the solid angle covered in


field of view). The visual field atlarger angles receives much less illu-mination due to the inter-reflected light from the sides ofthe viewport.

Secondly, lamps that do not differ muchinS/Pratios do not differ much with respect to scotopic outpUt (assuming equal photopic levels.) Given the SIP ratios used in this experiment and our data for the spectral response of pupil size(l!), we would predict only about a7%change in subject pupil area when comparing results from the full-spectrUm lamp with those from thecw lamp. This assumes a near full field of view, not the restricted field used in the authors' study. Thus, there may not have been significant pupil size differences in the authors' experiment. Without a pupil size differenCe, we would not predict a spectral effect on performance.

Thirdly, there is a probability that the contrasts for the visual performance task were not equal for the three spectrally-dif-ferent lamps. The authors found differencesincontrast for the visual comfort reading material under the three test lamps, but did not state whether the contrast ofthe visual per-formance task had been measured. Any differences in task contrast due to lamp spectrum would introduce a bias of unknown magnitude and direction with respect to between-lamp comparisons.

Fourthly, the statistics used require that each ofthe three dif-ferent groups studied be representative of the general popula-tion, i.e. have the same mean acuity. Since the smallest ring gap size was 0.66 minutes of are, subjects who identified all rings correctly at the high contrast level had at least 6/4

(20/13)vision, which is considerably better than normal. For the cw lamp, 15of the16subjects had at least better than 20/13 vision, suggesting that this group has better-than-aver-age vision. In contrast, for the full-spectrUm(ps)lamp, only 6 of the 16subjects had at least 20/13 vision. Thus the differ-ences in mean scores can just as well be attributed to a chance difference in the mean group acuities, rather than to lamp spectrum. However, because of the experimental design, in which both lamp type and group vary together. itis not possi-ble to disentangle these variapossi-bles.

Fifth and last, even though the authors state that they have not found any statistically significant evidence of a spectral effect on visual performance, they claim the data shows 'trends contrary to Berman's hypothesis'. There are several errors here. A trend should not be taken seriously unless the statistical probability approacheS, but does not reach, signifi-cance (for example,P < 0.10). Thisisnot the case here. In addition, the authors haven chosen to highlight only one



'trend'in.theirdata. Examination of their datawillshow both the fulls:pectrWn(FS)and thefilteredCW (FCW)lamps to have poorerperformance than thecwalone: Since theislamp has


sooropic content, while theFewlamphasless scotopic conrent




lamp, these trendsare in opposite

direc-tions, asmight be thecasefor chanceoccurrences.Why have theauthors chosen to mention only one ofthe two 'trends'? The secondary study is too weak (due, at least,to between-group comparisons, small SIP changes, and a narrow fieldof view)toargue against the hypothesis that spectrally modulat-edchaDgesin pupil size can affect visual acuity.


6 Vtitth J A and McColl S L Modulation of flu!lrescent light: Flicker

rateand light500=effects on visual performance and visllll1 comfort

I.ig1fli1igRes. TedmoL27(4) 24H56 (1995)

7 Berman S M,Fein G,JewettD1.,and Ashford F .Lumina.nre-con-trolledpupilsizeaffects Landolt C taSk performanceJ.Jl1um.Eng.§oc 22(2)15-165 (1993) ... 8 Berman SM, Fein G,JeweriD1.,and AshfOrdF Landolt C


lion in elderly subjects is affected by scotopic: intensity of sllITOlllld illu-minantsJ.Jl1um.Eng. Soc.23(2)123-130 (1994)

9 Berman S M, JewettDL,FeinG, Benson B, Law T, Myers·1<\.','and BuUimore M A Lighting spectral effecton Landolt C perfonnance is enhanced by blur and abolished by mydriasis]. 111um. 211&Soc.25 (I) 42-50 (1996)

10 Bennan S M, Fein G, Jewett D L, Benson B, Law T, and Myers A Luminanre-controlled pupil size affects word-reading accuracyJ,.nJum•.. Eng. Soc.2S (I) 51-59(1996)

II Berman S M, Fein G, Jewett DL,Saika G, and Ashford FSpectnll detenninants of steady-state pupilsizewithfull field of viewJ.mum. E,jg. Soc.21(2) 3-13 (1994)

S M Bennant and Blaine R Benson:\: tEnergyand Environment Division E 0 Lawrence Berkeley National Laboratory Berkeley, CA 94720 USA :j:Abralech Corporation 475 Gate 5 Road, Suite 255 Sausalito, CA 94965 USA


We are pleased that our paper6)provoked the interest of Dr Berman. andMr Benson. Understandably, we disagree with certain of their remarks.

At the time of the experiment, we were unaware of the poten-tial effect of electromagnetic interference from the electronic ballasts on our illuminance meter. We concur with the con-clusion that it is possible that the net effect of this discrepan-cy may be to underestimate the size of the flicker effect on visual performance. Therefore,we can only agree with Dr. Berman andMr. Benson that further careful study of flicker rates on visual processing is warranted.

We disagree with their conclusions regarding the lamp type comparison, which are based on criticisms· of both the meth-ods and the results. First, we point out that the lamp type comparison was not selected solely as a test of Dr. Berman's scotopic sensitivity hypothesis, but as a meanstoinduce chro-matic as well as luminous flicker (Reference6,p 244).



We are unconvinced that the differences in the field ofview between our experiment and Dr Betman's work have any implications for the outcome. In anycase, webelieve that the discussion of differences in methodology begs the question. If the effect of a scotopically enriched lamp on vision is robust, and if it is substantial enough to warrant changes in lighting practice, as Dr Berman has argued(2),then it should occur

reliably, across varying methodologies, and with lamps that are typical of current practice. To date, Dr Berman's work has found the effect consistently, but using an unusual arrang<>-ment that separates task and surround lighting, and using lamps that are extreme in theirSIP ratio (0.24 and 4.31(10».

The ratios of the sources employed in our slady, which were typical of current practice, ranged from 1.04 to 1.68(6'.


It is true that the statistical power of a betw""!!.-groups com-parison with 16 participants per group (total sariiple size of48 in three groups) is lower than that ofa within-groups compar-ison, and it is possible that with either a larger sample size (as we noted in Reference 6, p 252) or a within-subjects design, a lamp type effect on visual performance might have been observed. However, as we noted in our earlier replytoDrM S Rea (Reference 6, p 255), the VALiD task provided only.two paraliel forms ofthe test To have used the same task fur more than one experimental condition, as a within-subjects design with six combinations of lamp and ballast type would have required, would have risked a confounding practice effect We judged this to be more serious than the alternative, whichwas

a less-than-optimal level ofstatistical power.

Differences in task contrast between the lamps were small enough that they were unlikely to have affected between-groups visual performance (cf. Reference 13).Inanycase,if lamp specttai qualities have this differential. effect on contrast of commonly read materials (such as the laser printer copy used for the contrast measurements), then any associated dif· ferences in visual performance would apply to real·world con-ditions. We note, moreover, that the order of task contrast, from highest to lowest(FS,FCW, cw), is opposite to the order of visual performance scores (cw, FCW,FS).

Dr Berman andMr Benson make a logical error when they state that a priori differences in acuity between the lamp type groups confound the contrast, in that they base their state-mentabout acuity on the dependent measure ofvisual perfor-mance, which could have been influenced by the lamp type. To compare the baseline acuities of participants would require a pretest (initial) measure of acuity based on identical test conditions. We are confident that the groupswere

Com-parable in this sense because all participants passed the

Titmus vision test with at least 20120 vision and were subs<>-quently assigned randomly to one of the three groups. This is the standard procedure for a true between-groups experimen-tal design(14,IS).We quote Cook and Campbell, authors of one

of the most widely-cited books on research design in the

behavioural sciences: '...a properly implemented random assignment procedure will usually result in initially compara-ble experimental groups...' (Reference. 14, p 342). It allows us to infer that differences in the outcome measures between the groups were caused by the treatment (lamp type) rather than by pre-existing differences between the groups.

In their fifth point, Dr Berman andMrBenson ask why we highlighted only one trend. We tested two specific hypoth<>-ses using planned comparisons. Neitherwasstatistically sig· niftcant, but we highlighted the contrast between theFS group and the combined FCWlcw group because its multivari-ate F(6,40)= 1.92 hadp = 0.10, leading us to suggest that the 56

trend might have been statistically sigrtiticant with a larger sample size.Ourinterest in that contrastwasrelated in part to thefact that the


of variance assiJciated with it(7%)

wasas large as that explained by the statistically significant ballast effect (6%). We did not discuss the mean difference between the results fur the FCWand


groups, which we also tested, because its F(6,40)=1.20hadp=0.32. We rejected the hypothesis that these means differed.

In conclusion, we thank these researchers for their comments and we hope that their interest is reflected in independent replications ofthiswork.


12 Berman SM· Energy efficiency consequences of scotopic sensitivityJ.

l1Ium. Eng. So<.21(1) 3-14(1992)

13 ReaM Sand Ouellene MI Relativevisual performance: A basis for

,ppUOItionLWluingRa. TedowJ. 23135--144(1991)

14 Cook T D and Campbell D T qiオオゥセZ Dengnand a..myri$

forfi81d,.ningt(Boston: HoughtonMiftlin)(1979)

15 Kerlinger F N FoundalWns ofbeMvioral research(3rd edn) (New York: Holt, Rinehan&Winston) (1986)

Jennifer A Veitch§ and Shelley Lm」cッャャセ

§NationalResearchCouncil ofCanada

Insti[Ute forResearthin Construction

Ottawa, ON KIA OR6 Canada

'IIDept ofPsychology McGill University

120SDrPenfield Ave'l Montreai,QCIDA IBI


International Daylighting Measurement Programme -Singapore data

In the author's response todiscussion(l6) on the above

papers(17-19', Dr Ullah says that·the international community should come forward with guidelines tegarding the author-ship of data.

Protocol on data exchange was agreed before the 1991 start of the IDMP. A summary of the agteement is printed in Reference 20.

The procedures have been discussed in further detail in Division III of the Commission Internationale de l'Eclairage (CIE) during the last two years; research workers in CIE member states should be kept informed through their nation-al representatives.

The main points ofthe agreement are:

1 Participating stations in the lDMP shall exchange data freely with other participating stations, subject only to these conditions:

(a) There may be an interval of one year after theend of a survey periodtoenable the collector's own research on his or her data.

(b) There shall be a further interval of two years durin8 which the data may only be disseminated to participating organisations.

(c) A station supplying data may require the receiving organisationstopay the cost of the physical data transfer.




Related subjects : Visual performance