This article was downloaded by: [New York University]

On: 18 May 2015, At: 10:19

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of the Illuminating Engineering Society

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uzie20

Illuminance Selection Based on Visual

Performance—and other Fairy Stories
a
P.R. Boyce
a
Lighting Research Center, Rensselaer Polytechnic Institute Troy, NY.
Published online: 19 Sep 2013.

To cite this article: P.R. Boyce (1996) Illuminance Selection Based on Visual Performance—and other Fairy Stories, Journal
of the Illuminating Engineering Society, 25:2, 41-49, DOI: 10.1080/00994480.1996.10748146

To link to this article: http://dx.doi.org/10.1080/00994480.1996.10748146

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of
the Content. Any opinions and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of information. Taylor and Francis shall
not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other
liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

Illuminance Selection Based on Visual Performance—
and Other Fairy Stories
PR Rnvro
Introduction
Once upon a time, there were three illuminating
engineers who lived in a small house on Wall Street.
They were poor but they were honest. They made their
living by providing clear advice on good lighting prac­
tice. But they did not sleep well. Their nights were
haunted by the knowledge that much of what they rec­
ommended was based on accumulated experience and
judgment—it was a matter of consensus. In the darkest
Downloaded by [New York University] at 10:19 18 May 2015
hours of the night they often thought that one day the
wolf of litigation would come to their door and would
huff and puff and blow their house down. But with each
dawn came new hope. There was a solution. It was to
find the magic formula, a formula that accurately
described the relationship between lighting conditions
and the performance of any task. With such a formula,
the illuminating engineers could be objective and
abjure consensus. They could simply state what the rela­
tionship was between lighting conditions and task per­
formance and leave the users to decide what level of task
performance they wanted. Alternatively, they could
make recommendations based on the formula, explicit­
ly stating the conditions they had used. Either way they
would have a defensible basis for their recommenda­
tions, a base strong enough to defy the wolf of litigation.
Year after year they persisted with their search for the
magic formula. After many years and several false
dawns, the magic formula was found and they all lived
happily ever after.
Unfortunately, the above is a fairy story, or rather, the
problem is real but the solution is not. This paper
explains why a magic formula describing the relation­
ship between lighting conditions and task performance
cannot exist in any general form; discusses the differ­
ence between visual wants and visual needs; and con­
cludes that consensus is an inevitable component in all
illuminance recommendations.
Why a magic formula cannot exist
Before discussing the relationship between lighting
conditions and task performance it is necessary to
define some terms. Specifically, the two terms task per­
formance and visual performance need to be clearly dis­
tinguished because they are sometimes used as syn-
Author's affiliation: Lighting Research Center, Rensselaer Polytechnic
Institute Troy, NY.
41
THIS PAPER WAS PRESENTED AT
THE 1995 IESNA ANNUAL CONFERENCE
onyms. Task performance is the performance of the
complete task. Visual performance is the performance
of the visual component of the task. Task performance
is what is needed in order to establish cost-benefit ratios
comparing the costs of providing a lighting installation
with the resulting benefits in terms of better task per­
formance. Visual performance is the only thing that
changing the lighting conditions can affect directly.
The underlying reason why a magic formula cannot
exist is that no task is purely visual. Most apparently visu­
al tasks have three components: visual, cognitive, and
motor. The visual component refers to the process of
extracting information relevant to the performance of
the task using the sense of sight. The cognitive compo­
nent is the process by which sensory stimuli are inter­
preted and the appropriate action determined. The
motor component is the process by which the stimuli
are manipulated to extract information or the actions
decided upon are carried out.
As an example, consider the task of driving a car
along a road. The driver scans the road ahead and its
environs to extract information using the sense of sight.
This is the visual component. The significance of the
information extracted is evaluated by the brain to deter­
mine the appropriate action, which may range from
doing nothing, to braking sharply, to changing lanes.
This is the cognitive component. Actually moving the
steering wheel or applying the brake is the motor com­
ponent. Of course, this is a continuous process where
visual, cognitive, and motor components interact and
overlap. Nonetheless, the only part of the task perfor­
mance that changing the lighting conditions can influ­
ence is the visual component. This implies that the
effect of lighting conditions on task performance
depends on the role of the visual component in the
structure of the task.
The role of the visual component in the structure of
a task can vary in at least three ways. First, the magni­
tude of the visual component can vary between tasks.
For example, the visual component is greater for data
entry from written material than from an audio source.
Second, the significance of the visual component in the
structure of the task can influence the importance of
the lighting conditions, and this significance is not nec­
essarily related to the magnitude of the visual compo­
nent. For example, in the construction of an electronic
circuit mounted on a printed circuit board, the visual
JOURNAL of the Illuminating Engineering Society Summer 1996
 42
inspection of the printed circuit board may take only a
short time, but if a fault is not detected, the consequen­
tial costs in terms of time and money may be substantial.
Third, the emphasis of the visual component can be dif­
ferent for different tasks. For example, reading low con­
trast, small print, of the type found on car rental agree­
ments, places greater emphasis on the illuminance pro­
vided than the spectral content of that light, while dis­
criminating between textiles of different colors places
greater emphasis on the spectral content of the light
than the illuminance above the threshold required for
color vision.
All this should not be taken to mean that the lighting
conditions provided are unimportant for task perfor­
mance. A simple comparison of the difficulty experi­
enced when driving by day, by night, and in dense fog is
sufficient to illustrate the importance of the visual com­
Downloaded by [New York University] at 10:19 18 May 2015
ponent to driving. Rather, what it does mean is that every
task is unique in its balance between visual, cognitive,
and motor components and hence in the effect of light­
ing conditions on task performance. It is this uniqueness
that makes the existence of a magic formula quantifying
the precise relationship between lighting conditions and
task performance for a wide range of tasks, impossible.
Even if a magic formula was possible,
could you use it?
Let us suppose that all the above is false. A magic for­
mula could be found that would quantify the relation­
ship between lighting conditions and task performance,
but could it readily be applied? I suggest the answer is
negative. The reality of much lighting practice is that a
single lighting installation serves to light many different
tasks. A look around any office or workshop will demon­
strate the truth of this statement. Further, most people
do a range of visually different tasks throughout each
day. A consideration of the various materials you look at
each day from which you extract visual information,
which can range from faces to computer screens, will
confirm this assertion. Furthermore, the visual charac­
ters of tasks can be expected to change over the life of a
lighting installation. To appreciate this, it is only neces­
sary to consider how the nature of office work has
changed over the last decade.
This inherent variability in the visual demands in
many workplaces makes the idea of exactly specifying the
characteristics of the lighting to be provided on the basis
of optimizing task performance unrealistic. Indeed, it
reduces the argument about the exact relationship
between lighting conditions and task performance to
one of the "how many angels can dance on the head of a
pin" variety. True, it would be possible to recommend
lighting conditions on a worst case basis, i.e., identify the
task that is most visually difficult and recommend light­
Summer 1996 JOURNAL of the Illuminating Engineering Society
ing conditions that would ensure it would be performed
adequately. But there is no guarantee that this would rep­
resent an appropriate trade-off of cost-benefits or that
the lighting required for the worst case would also be
suitable for easier tasks. Therefore, even if a magic for­
mula did exist, the realities of application require a con­
sensus about the tasks to which it should be applied, for
each application.
So what use are visual performance models?
If a magic formula quantifying the relationship
between lighting conditions and task performance can
never be achieved—and if it could, could only be
applied through a consensus process—what is the value
of the various visual performance models that have been
developed? The answer is that even if they do not predict
task performance, they do tell us what effects lighting
can have on the visual performance of tasks. Such knowl­
edge is valuable because it quantifies the maximum
effect a change in lighting conditions can have as well as
the relative importance of changing the task stimulus by
changing the task materials and changing the lighting
conditions. Indeed, it would be possible to make lighting
recommendations based on achieving a minimum level
of visual performance. Such recommendations would be
justified by the assertion that the business of lighting is to
make things visible, not to maximize task performance.
Maximizing task performance is the business of manage­
ment. Lighting has a role to play in maximizing task per­
formance but it is one role among many, and in many
modern.production facilities, it is a minor role. However,
lighting is a prime mover in making things visible.
Having decided that visual performance models are
valuable, it is necessary to consider which model is most
useful. The development of visual performance models
has an interesting and converging history. Some of die
earliest attempts to determine the effects of lighting con­
ditions on task performance occurred in the 1920s,
among them being some of the famous Hawthorne
experiments. These attempts took the form of field tri­
als seeking a link between the lighting in a factory or part
of a factory, and the output achieved. These studies, and
the others that have occurred intermittently since, did
little more than demonstrate that lighting can improve
task performance but the extent to which an improve­
ment occurs, or if it occurs at all, is different for different
tasks.
It soon became evident tiiat the chances of develop­
ing a comprehensive model of visual performance from
field studies was remote. An alternative approach, based
on modeling the effect of lighting, was proposed by
Beutell and implemented by Weston. Beutell's sugges­
tion was that the visual difficulty of each task could be
characterized by the visual size and luminance contrast
 43

14
of the critical detail of die task, any relative movement
nance contrasts, based on the speed with which people
between the observer and the task, and the degree of
could carry out the task. Clear and Berman later provid­
emphasis to be given to the task in its setting. Weston
ed an alternative, visibility level model to fit both the
took this approach and developed a standard task, the
speed and accuracy data obtained on the numerical ver­
Landolt ring task, in which the observer examines an
ification task.
array of Landolt rings and identifies all those with a
While attempts were made to minimize the motor and
specified gap orientation. The Landolt rings forming
cognitive components of the numerical verification task,
the array can easily be varied in visual size and luminous
there remained a lingering doubt that the resulting RVP
contrast. Then the effect of lighting conditions on
model would not apply to other tasks. To overcome this
other tasks can be predicted by measuring the visual
doubt, other experiments were undertaken using simple
size and luminance contrast of the task of interest and
reaction time to the detection of a stimulus as the depen­
using them to identify the matching Landolt ring task.
dent variable. The advantage of using this method is
The effect of lighting conditions on the performance of
that the cognitive and motor components of the task can
the matching Landolt ring task provides an estimate of
be minimized, the visual component maximized, and the
their effects on the task of interest. Despite its crude
question of the trade-off between speed and accuracy
nature, the model that resulted from this approach
rendered moot. The results of these experiments con­
Downloaded by [New York University] at 10:19 18 May 2015

demonstrated the nonlinear nature of the effect of illu­

firmed the RVP model as a comprehensive model of visu­
minance and the relative importance of visual size,
al performance. Bailey, Clear, and Berman produced a
luminance contrast, and illuminance.
7 competing model using reading speed data and based
Yet another approach was developed by Blackwell. on visual size rather than luminance contrast. They con­
Blackwell's approach was to quantify the visibility of a firmed the general form of the contrast-based RVP
stimulus in terms of its visibility level. In its simplest form, model but claim their size-based model is more accurate.
the visibility level of a stimulus is the ratio of the lumi­ Despite the inevitable conflict between these more
nance contrast of the stimulus to the threshold lumi­ recent, competing models, their similarities are more
nance contrast of the same stimulus. The larger the visi­
19 UBMSIDHDWS « MCR0SHMDW6
bility level, the more visible the stimulus. This idea is
inherently attractive in that it introduces the human
being into the measurement of lighting's effects.
Luminance contrast can be measured using a luminance
meter but threshold luminance contrast can only be
measured by a human being.
Having developed equipment for measuring the visi­
bility level of any stationary stimulus, Blackwell then
attempted to use this metric as a means of predicting task
performance, the principle being mat visibility level
would act as a unifying variable to combine the effects of
illuminance, visual size, and luminance contrast. His first
attempt appeared to work with some tasks but not with
BMOWSrfwOWS aOMOWSOAOWS
others. He revised his model but by this time it was col­
lapsing under the exponential growth in the number of
modifying factors that had to be introduced to make the
model fit the experimental results. Such growth is to be
expected if, as discussed earlier, each task has a different
structure.
The next significant figure on the scene was Rea.
Fundamentally disagreeing with the threshold basis of
visibility level used by Blackwell, he returned to what is
essentially Weston's approach. Smith and Rea devel­
oped their own standard task, the numerical verification Figure 1—Relative visual performance plotted as a function of reti­
task, in which observers compare two columns of five- nal illumination (trolands) and luminance contrast for a fixed size
digit numbers for discrepancies between the two of target as measured by the solid angle subtended at the eye
columns. Using this task, a model of visual performance, (microsteradians). A solid angle of 4.8 microsteradians represents
the size of the digits used in the numerical verification task (from
the relative visual performance (RVP) model, was devel­
Rea and Ouellette [1991])."
oped for a range of adaptation luminances and lumi-
JOURNAL of the Illuminating Engineering Society Summer 1996
 44
striking than their differences. Both show a compressive
relationship between some measure of the stimulus to
the visual system and the performance of the task. As an
example of such a compressive relationship, Figure 1
shows the form of the RVP surface for four different visu­
al size tasks, each surface being for a range of contrasts
and retinal illuminances. The overall shape of the RVP
surface has been described as a plateau and an escarp­
ment. In essence what it shows is that the visual system is
capable of a high level of visual performance over a wide
range of visual sizes, luminance contrasts, and retinal
illuminations (the plateau), but at some point either
visual size, luminance contrast, or retinal illumination
will become insufficient and visual performance will
rapidly collapse (the escarpment).
It can be argued that the first duty of all illuminating
Downloaded by [New York University] at 10:19 18 May 2015
engineers is to provide lighting that ensures that all tasks
are being performed at the plateau level and far from
the escarpment. The existence of a plateau of visual per­
formance (or rather a near plateau because there is real­
ly a slight improvement in visual performance across the
plateau) implies that for a wide range of visual condi­
tions, visual performance changes very little with
changes in the lighting conditions. To put it bluntly, what
this means is that for many visual tasks, lighting is unim­
portant to visual performance, the visual system being
flexible enough to cope equally well with a wide variety
of visual stimuli.
So what use are visual performance models? The
answer is that they give us a clear understanding of and a
method for calculating die impact of specific lighting
conditions on the visual performance of many tasks.
Such calculations give us a means of knowing when we
are close to the escarpment but, as explained above, this
does not give us an understanding of how the same light­
ing conditions influence task performance.
Visual performance models are valuable but they are
not the magic formulae. The IESNA RQQ committee
recognized this in 1991 when it published a draft report
setting out a method for selecting task illuminance.
What this report suggested is a selection procedure in
which the level of relative visual performance is selected
by consensus and then the RVP model is used to derive
the illuminance required to achieve the specified level of
relative visual performance. Unfortunately, this innova­
tive approach to making illuminance recommendations
appears to have disappeared into limbo.
If a magic formula did exist,
would you want to use it?
A devotee of the conspiracy theory of history might
suggest that one reason why the recommendations of die
RQQ committee may have disappeared into limbo is mat
it is difficult to believe that the RVP model could be used
Summer 1996 JOURNAL of the Illuminating Engineering Society
to justify many of the illuminances recommended in the
IESNA Lighting Handbook. The easiest example to
demonstrate tiiis point is reading. Table 1 shows the illu­
minance required to achieve an RVP of 0.98, for print
sizes ranging from 6 point to 10 point, for print of con­
trast of 0.7 on paper of reflectance 0.7, seen by people of
20 or 60 years of age, at a distance of 40.5 cm. Table 1
Table 1—Illuminances required for a relative visual performance
of 0.98 for 20 and 60 year olds reading 6, 8 and 10 point print of
luminance contrast = 0.7, compared with the illuminance recom-
mendations of the IESNA.
Print size Print contrast Illuminance (lx) IES Recommended
20 yrs 60 yrs illuminance (lx)
6 point 0.7 79 302 500-750-1000
8 point 0.7 38 148 200-300-500
10 point 0.7 27 101 200-300-500
also shows the illuminances recommended by the IESNA
Lighting Handbook for reading 6 point, 8, point, and 10
point type. Given that most reading materials present in
offices today are in 10 point type or larger and are print­
ed in high contrast on white paper, these results suggest
it would be difficult to justify any illuminance for com­
mercial office buildings above 100 lx if die recommen­
dation were to be made on the basis of visual perfor­
mance alone.
The true basis of illuminance recommendations
It would be a brave illuminating engineer who speci­
fied a lighting installation for a commercial office build­
ing which only produced 100 lx. Several studies have
shown that such an illuminance would be considered too
22,18,5
dim, uncomfortable, and hence unacceptable. Yet in
1917, such an illuminance would have been regarded as
excessive. Twenty years ago the average illuminance rec­
ommended for offices was 1000 lx. Today it is 500 lx.
How can this be? Why should the illuminance require­
ments for office work change so much over the years?
The true basis of illuminance recommendations can be
found by considering die possible answers to diis ques­
tion. One possible answer is that peoples' inherent visu­
al performance capabilities have changed over die last
century, but this seems unlikely given that the human
visual system evolved over a much longer time period.
It could be argued that the difficulty of visual tasks has
increased greatly, thereby requiring higher illuminances.
Again, this seems unlikely. The introduction of modern
office technology has, if anything, increased the quality
of printed materials. The day of the fifth-carbon copy is
over. The most likely answer is that illuminance recom­
mendations are not determined by visual performance
alone but rather are subject to many other forces. These
forces are both practical and political. The practical
forces are matters of technology.
There is no point in making illuminance recommen-
 45

Table 2—Illuminance recommendations for reading in every edi­ Mean detection

tion of the IESNA Lighting Handbook, the dominant lamp technolo­ speed (s"')
gy used in office lighting, and the economic-political state of the
0.06 "
US

IES Visual task; Illuminance Lamp type Economic / 0.05 "

Handbook Reading <lx) Political
State

0.03 "
1947 Regular 300 Incandescent Moderate
Difficult 500 growth
0.02 "
1954 Regular 300 Incandescent Strong
Difficult 500 /Fluorescent growth 0.01 -

1959 Regular 1000 Fluorescent Strong

Difficult 2000 growth u.uu
30 100 300 1000 3000 10,000
1966 Regular 1000 Fluorescent Strong Illuminance (Ix)
Difficult 1500 growth
% lighting
Downloaded by [New York University] at 10:19 18 May 2015

1972 Regular 1000 Fluorescent Growth 'good'

Difficult 1500
100 '
1981 Regular 200-300-500 Fluorescent Post energy
Difficult 500-750-1000 crisis 80 - S "\

60 " / \
1987 Regular 200-300-500 Fluorescent Post energy
Difficult 500-750-1000 crisis

1993 Regular 200-300-500 Fluorescent Environment

Difficult 500-750-1000 concerns
20 " ^ /

dations that cannot be readily achieved in existing build­
ings with existing technology. The political forces are
both financial and emotional. The financial force is the
cost of providing a given illuminance relative to the ben­
efits obtained. The emotional force is the extent to
which the lighting is designed to make people comfort­
able and to meet their expectations. The illuminance
recommendations made at any specific time and in any
specific country will vary with the balance between these
forces. Table 2 shows the illuminance recommended by
the IESNA for general offices in each edition of the
Lighting Handbook, the dominant lighting technology
used in offices at the time, and the economic-political
state of the US. Obviously this is a crude picture, espe­
cially because the IESNA has frequently changed its
descriptions of office tasks, but the pattern of change in
illuminance recommendations with the technical-eco­
nomic-political balance of forces is suggestive of their
importance.
The influence of technology and economics is widely
recognized in many fields, but the difference between
the visual performance and visual comfort is less familiar.
Figure 2 shows some results taken from a study that mea­
sured the performance of a task requiring the subjects to
find a two-digit number among 99 such numbers ran­
domly distributed over a table, for different illuminances
on the table. The speed of performance increases
monotonically as the illuminance increases, but the per-
Ql •--"HIT 1 i i I
30 100 300 1000 3000 10,000
Illuminance (lx)
Figure 2—The mean speed taken to detect a given two-digit num­
ber from 99 two-digit numbers arranged randomly on a table, plot­
ted against the illuminance on the table. Also shown is the percent­
age of subjects who considered the lighting of the task "good",
plotted against the illuminance on the table (after Muck and
Bodmann [1961])."
centage of subjects considering the lighting comfortable
shows a clear optimum. What Figure 2 demonstrates is
that visual comfort and visual performance are not syn­
onymous. It is possible to have an illuminance that allows
a high level of visual performance but that is uncomfort­
able. This, in turn, suggests that visual performance and
visual comfort represent two successive constraints on
illuminance recommendations.
The successive constraints imposed by visual perfor­
mance and visual comfort arise from the differences in
what they measure. Visual performance measures what
can be done. Visual comfort measures what is easy to do.
In modern society, providing lighting is considered to be
technically easy, so only lighting that makes tasks easy to
do is acceptable. Peoples' expectations form another
constraint. Expectations are simply what we expect from
life. Shifting expectations are a part of life. Expectations
as to what constitutes good quality cars, office furniture,
computers, etc., have all changed in recent years. There
is no reason why lighting should be exempt from this
JOURNAL of the Illuminating Engineering Society Summer 1996
 46
process. Put succinctly, illuminances based on visual per­
formance represent visual needs. Illuminances based on
expectations represent visual wants.
The effect of the constraints posed by visual perfor­
mance, visual comfort, and expectations can be illustrat­
ed by the following list of assertions, given in order of
increasing stringency:
• Lighting that limits visual performance will not be
comfortable or meet peoples' expectations.
• Lighting that does not limit visual performance will
not necessarily be considered comfortable or meet peo­
ples' expectations.
• Lighting that does not limit visual performance and
does not cause visual discomfort will not necessarily meet
peoples' expectations.
• Only lighting that does not limit visual performance,
Downloaded by [New York University] at 10:19 18 May 2015
does not cause visual discomfort, and meets peoples'
expectations will be acceptable to users.
What all this means is that lighting practices, and
hence lighting recommendations, are not isolated from
the dynamic flux common to most human activities.
Hence, the desire to base lighting recommendations on
a model of visual performance alone is doomed to fail­
ure. Understanding the effect of lighting conditions on
visual performance is useful because it allows us to
ensure that a recommendation will not fail to meet one
of the constraints, but anyone writing lighting recom­
mendations also has to consider the other forces acting
to determine the acceptability of the recommendations.
As many of these forces are psychological rather than
biophysical, lighting recommendations are inevitably
matters of judgment and hence, in a democratic society,
governed by consensus.
What should the IESNA do?
Given that illuminance recommendations are subject
to many forces—technical, biophysical and psychologi­
cal—the IESNA has a choice to make about how it deter­
mines its illuminance recommendations. I believe the
IESNA has four options:
• Maintain the status quo
• Restrict its role to that of a technical engineering
society, publish information on how lighting conditions
affect visual performance and visual comfort and leave
the determination of the lighting actually installed in
buildings to market and political forces
• Accept that the role of lighting is to make things vis­
ible and base its lighting recommendations on visibility
alone
• Accept the reality of the many forces acting on light­
ing recommendations, work to develop an understanding
of how lighting conditions affect visual performance and
visual comfort, set up an open system for obtaining con­
Summer 1996 JOURNAL of the Illuminating Engineering Society
sensus based on data from the field, and publish lighting
recommendations based on that consensus
The first option is a recipe for decline. Any organiza­
tion that is uncertain about the basis of its most widely
used recommendations is in trouble. Further, for the rea­
sons given earlier, I do not believe there is a magic for­
mula that can be used to quantify the relationship
between illuminance and task performance, therefore,
to continue to search for one is a waste of resources. The
second option is the safe option but it seems to be an
abandonment of responsibility. The third option is intel­
lectually defensible but again is a withdrawal from the
lighting decision process, a process in which the voices of
the IESNA membership should be heard. The fourth is
the most difficult option but recognizes the IESNA's
leadership in knowledge and understanding of lighting.
My vote would be for the fourth option, recognizing
that it would not be easy. Whereas the effect of lighting
conditions on visual performance is, in principle, deter­
mined by the capabilities of the visual system, which are
unlikely to change in the near future, the effect of expec­
tations is much more volatile and is governed by a much
wider range of interests. For example, it is arguable that
anyone wishing to save the planet could be most effective
byworking to change expectations about illuminances. If
this work led to a marked reduction in illuminance rec­
ommendations, both resource depletion and air pollu­
tion would also be reduced.
It is equally arguable that moving expectations to
higher illuminances would result in sales of more lamps
and luminaires. It is the possibility of such attempts to
change illuminance recommendations that necessitates
the setting up of an open process for determining con­
sensus. The process to be used should itself be the sub­
ject of debate.
There are a number of models for determining con­
sensus ranging from a "Supreme Court" approach, in
which a limited number of knowledgeable individuals
hear evidence about the costs and benefits of different
illuminances for each application and then issue their
recommendations, to the simple application of a time-
weighted average of installed illuminances. Whatever the
process chosen, there are three aspects that I believe are
essential requirements: the collection and analysis of
data on current lighting practice and users' opinions of
that practice, the evaluation of the data by informed and
knowledgable mediators, and the process itself to be
open to external scrutiny.
Are we alone?
In case you are thinking that the IESNA is alone in
having a consensus basis for its primary recommenda­
tions, it is worth pointing out that other professional

organizations face the same dilemma. The American
Society of Heating, Refrigerating and Air-Conditioning
Engineers (ASHRAE) is in a similar position. For many
years ASHRAE has published recommendations for a
comfort envelope based on a measure of temperature
and a measure of relative humidity. These recommenda­
tions are clearly based on human thermal comfort and
carry with them considerable economic impact. The rec­
ommended conditions are much more limited than
needed simply to ensure that task performance is unlim­
ited. The recommended conditions have changed over
time as expectations have changed. There is even an
argument about the thermal comfort envelope, namely,
should it be based on the thermal comfort of people
wearing clothes with the same level of thermal insulation
or should it be relaxed to allow for the possibility that
Downloaded by [New York University] at 10:19 18 May 2015
people can be expected to change their dress according
to the thermal conditions? This may seem an arcane
argument, but in practice, considerable sums of money
are riding on the answer. Any professional society mak­
ing recommendations where people are die consumers
of diose recommendations is operating in a field subject
to psychology as much as biophysics. It is time for the
IESNA to recognize this fact and organize an appropri­
ate process to regularly monitor the consensus on illu­
minance recommendations.
Postscript
This paper is not the usual conference paper. It does
not describe research or application. Some of the state­
ments made are opinion rather than fact. Few are origi­
nal. My basis for making them rests on nearly 30 years
spent in and around lighting research and application.
What I have put forward is an argument, written in order
to generate an argument, about the nature of illumi­
nance recommendations and the responsibilities of pro­
fessional organizations. My hope is that these observa­
tions will clarify the basis of illuminance recommenda­
tions and lead to a more open process of determining
what illuminances are recommended.
References
1. Audel and Co. 1917. Hawkins electrical guide,
Number Eight. New York: Theo Audel and Co.
2. Bailey, I.; Clear, R.; and Berman, S. 1993. Size as a
determinant of reading speed. / of the IES 22 (no.
2):102-117.
3. Beutell, A.W. 1934. An analytical basis for a lighting
code. Ilium Eng 27:5.
4. Boyce, P.R., and Rea, M.S. 1987. Plateau and escarp­
ment: the shape of visual performance. Proc CIE 21st
Session, Venice.
5. Boyce, P.R., and Rea, M.S. 1994. A field evaluation
of full-spectrum, polarized lighting. / of the IES
47
23(no.2):86-107.
6. Clear, R. and Berman, S. 1990. Speed, accuracy and
VL. / of the IES 19 (no. 2):124-131.
7. Commission Internationale de l'Eclairage. 1972. A
unified framework for evaluating the visual performance aspects
of lighting. CIE Publication No. 19. Paris: CIE.
8. Commission Internationale de l'Eclairage. 1980. An
analyatical model for describing the influence of lighting para­
meters on visual performance. CIE Publication 19/2. Paris:
CIE.
9. Elton, P.M. 1920. A study of output in silk weaving dur­
ing winter months. Industrial Fatigue Research Board
Report no. 9. London: His Majesty's Stationary Office.
10. Illuminating Engineering Society of North
America. 1991. RQQ Report #7. Recommended method for
selection of task illuminance. New York: IESNA.
11. Illuminating Engineering Society of North
America. 1993. Lighting Handbook, eighth edition. New
York: IESNA.
12. Mclntyre, D.A. 1981. Indoor climate. London:
Applied Science Publishers.
13. Muck, E. and Bodmann, H.W. 1961. Die bedeu-
tung de beleuchtungsniveaus bei praktische sehtatigkeit.
Lichttechnik 13:502.
14. Rea, M.S. 1986. Toward a model of visual perfor­
mance: foundations and data. J of the IES 15 (no.
2):41-58.
15. Rea, M.S.; Boyce, PR.; and Ouellette, M.J. 1987.
On time to see. Lighting Research and Technology 19:101.
16. Rea, M.S. and Ouellette, M.J. 1988. Visual perfor­
mance using reaction times. Lighting Research and
Technology 20:139.
17. Rea, M.S. and Ouellette, M.J. 1991. Relative visual
performance: A basis for application. Lighting Research
and Technology 23:135.
18. Saunders, J.E. 1969. The role of the level and
diversity of horizontal illumination in an appraisal of a
simple office task. Lighting Research and Technology 1:37.
19. Smith, S.W. and Rea, M.S. 1979. Relationships
between office task performance and ratings of feelings
and task evaluation under different light sources and lev­
els. Proc CIE 19th Session, Kyoto.
20. Stenzel, A.G. 1962. Experience with 1000 lx in a
leather factory. Lichttechnik 14:16.
21. Urwick, L. and Brech, E.F.L. 1965. The making of
scientific management: vol 3, The Hawthorne investagations.
London: Pitmans.
22. Van Ierland, J.F.A.A. 1967. Two thousand Dutch
office workers evaluate lighting. Research Institute for
Environmental Hygiene Publication 283. Delft, The
Netherlands: TNO.
23. Weston, H.C. 1922. A study of efficiency in fine linen
weaving. Industrial Fatigue Research Board Report no.
20. London: His Majesty's Stationary Office.
JOURNAL of the Illuminating Engineering Society Summer 1996
 48
24. Weston, H.C. 1945. The relation between illuminance
and visual performance. Industrial Health Research Board
and the Illumination Research Committee (DSIR).
London: His Majesty's Stationary Office.
25. Weston, H.C. and Taylor, A.K 1926. The relation
between illumination and efficiency in fine work, (typesetting
by hand) Final Report of the Industrial Fatigue Research
Board and The Illumination Research Committee.
(DSIR). London: His Majesty's Stationary Office.
Discussion
Bravo to an excellent, insightful, and delightful paper.
The author's perspective deserves serious consideration.
Only two comments are offered. First, regarding what
we, the IESNA, should do, I vote for the fourth option.
Second, let's finally put this illuminance thing in per­
Downloaded by [New York University] at 10:19 18 May 2015
spective and move on to other important criteria.
G.R. Steffy
Gary Steffy Lighting Design Inc.
P. Boyce is to be commended for his delineation of
the issues facing the lighting community in the develop­
ment of illuminance selection procedures. As he points
out, it is time for the IESNA to understand that there is
no magic formula for determining illuminance levels
and that the issues surrounding visual performance are
many and complicated. His recommendation of option
four—further study and development of procedures for
obtaining real consensus—is one that the Society should
heed.
To reinforce P. Boyce's viewpoint, it is important to
realize that the use of consensus in standards develop­
ment is a hallmark of all engineering and other stan­
dards-writing bodies, such as ASTM, IEEE, ASME, etc., in
the United States. This refers to the way in which agree­
ment is reached, not failure to use and understand tech­
nical data correctly. Thus, use of rigorous procedural
requirements and diverse representation are central to
the consensus process. The American National
Standards Institute (ANSI) defines consensus as follows:
"substantial agreement has been reached...by directly
and materially affected interest categories. Substantial
agreement means much more than a simple majority,
but not necessarily unanimity. Consensus requires that
all views and objections be considered, and that an effort
be made toward their resolution." These procedures are
replicated at the international level by both the
International Standards Organization and the
International Electrotechnical Commission (IEC). Each
of these entities uses consensus and due process as key
elements of their standards development process.
ISO/IEC defines consensus in the following manner:
"General agreement, characterized by the absence of sus­
tained opposition to substantial issues by any important
Summer 1996 JOURNAL of the Illuminating Engineering Society
part of the concerned interests and by a process that
involves seeking to take into account the views of all par­
ties concerned and to reconcile any conflicting argu­
ments."
To equate the use of consensus with a failure to use
sound technical judgment is misleading and hampers
our efforts to find a reasonable basis on which to make
lighting decisions. Failure to use consensus means that
standards are set by only one person or interest group
that may not have properly examined all the technical
evidence.
Boyce correctly points out that the IESNA had been
hindered by the desire to find a single number or equa­
tion to predict visual performance and has failed to con­
sider the various nonvisual components of task perfor­
mance, comfort, and productivity. Our consensus deci­
sions in the future must consider all the parameters and
involve research of the complex tasks actually performed
in offices and the ways in which lighting affects these
tasks. Now that hardware improvements have made it
possible to reduce energy use without sacrificing visual
comfort, it is possible for the IESNA to determine and
study all the parameters needed for "good" lighting,
whatever that may be. At that time, I believe that we can
reach a true consensus for lighting recommendations.
B. Collins
NIST
The author has presented many important and
thought provoking issues of concern to our society, often
in the form of opinions. Although, as he states, this paper
does not present research or application, there is mater­
ial of value provided to better understand the purpose
and future directions of IESNA. However, for every opin­
ion, Newton's fourth law states there is most likely a
counter opinion. A few of these are given here.
To argue that lighting professionals are in some way
responsible for total task performance, i.e., not just the
visibility and visual comfort portions, but also the other
aspects (cognitive, motor, etc.) is misleading. This issue
arises in interpreting visual performance or subjective
response studies, but should not be confused in forming
recommendations for lighting practice. To take the case
to extremes, illuminance levels were not a factor when
the surgeon amputated the wrong leg.
Because the nonvisual components may be unwar­
ranted in practical applications, perhaps the present
level of knowledge of how lighting influences the visual
components is adequate for general lighting practice. P.
Boyce's discussion of visual performance models never
directly answers this question.
Currently, the primary and perhaps the only analytic
procedure available to advise illuminance levels is
through visual performance relationships. Boyce would

have us encompass the complexities of human nature by
offering the importance of user expectations in lighting
considerations. But in any consensus forum, other exi­
gencies of human nature will also be manifest—such as
aggressiveness—with a particular self interest, as well as
other modes of single or even hidden agendas. Analytic
procedures become an important and unique tool in
dealing with this kind of human nature in a forum envi­
ronment.
The Society's illuminance selection values will contin­
ue to require some help from consensus because it must
perform its responsibilities with imperfect or incomplete
information. But this feature should not obscure the
need for the continual search for better knowledge
which is absolutely essential for the Society's survival.
Our situation is somewhat analogous to the attending
Downloaded by [New York University] at 10:19 18 May 2015
physician of a cancer patient. In the absence of a cure,
the physician treats as best possible with present knowl­
edge, but that doesn't mean eliminating the search for
better understanding even though the research is com­
plex. The reason that "consensus does not appear to be
a comfortable posture" is that we know, as the physician
knows, that it is possible to do better. Let us not forget
this responsibility.
S. Berman
Lawrence Berkeley Laboratory
Author's response
To G. Steffy
First, I thank G. Steffy for his kind remarks. Second, I
should say that I agree with both his comments. The
fourth option is what I would vote for too. As for "putting
this illuminance thing in perspective," I agree but I sus­
pect that my perspective may be rather different from
his. Specifically, he implies that illuminance is not that
important and that we should move on to other impor­
tant criteria. There undoubtedly are other important
lighting criteria, but for functional lighting at least, I
believe that illuminance remains an important criterion.
This is not because it has a major effect on visual perfor­
mance, but because it is closely linked to peoples' expec­
tations.
To B. Collins
I thank B. Collins for providing an authoritative defi­
nition of consensus and for explaining that the use of
consensus to achieve an engineering recommendation is
common practice in many standards-writing bodies, both
nationally and internationally. The processes by which
consensus is reached in these various bodies should be
studied by the IESNA before adopting a process of its
own.
49
To S. Berman
I should like to thank S. Berman for his thoughtful
counteropinion. If the IESNA is ever to reach consensus,
it will do so through clearly stated arguments such as
those presented by S. Berman. His first point is that light­
ing professionals should not be held responsible for total
task performance, but only for the visibility and visual
comfort parts. I agree, but the corollary of accepting this
limitation on responsibility is that the lighting profes­
sional can no longer claim to have changed total task
performance, only the visibility or visual comfort of the
task. This may be important. For every task that has a
visual component, there are critical lighting conditions,
and if these conditions are not met, task performance
will deteriorate. Berman stated, "Illuminance levels were
not a factor when the surgeon amputated the wrong leg."
Illuminance levels could have been a factor if the reason
the surgeon amputated the wrong leg was that he could
not read the patient's chart in the lighting provided.
S. Berman then raises the question of the precision
required in our knowledge of how lighting influences
visual performance. I did not discuss this question
because I do not know the answer. Further, I doubt if this
question can be answered until some research on the
application of visual performance models to practical sit­
uations is undertaken.
Berman then asserts that analytical procedures are a
unique and important tool in dealing with self-interest in
the consensus process. This is correct and it has two
implications. The first is that the study of analytical pro­
cedures should continue. The greater the understanding
of how lighting conditions affect visual performance, the
less there is to reach consensus about. The second is that
the consensus process itself has to be open. Only then
can the operation of self-interest be exposed.
Finally, S. Berman makes a plea for continued
research so that in the future we may do better. Again, I
agree, but I also consider it essential that the objectives
of research be redirected. Specifically, it seems to me that
it is about time attention was paid to how some of the
existing models of visual performance could be applied
in practice and what would be representative values of
the input variables required for each model. If this were
to be done, we could not only develop a useful consen­
sus process for selecting illuminances, but we could also
identify some important questions that those intent on
refining visual performance models should address.
JOURNAL of the Illuminating Engineering Society Summer 1996