Summary This paper describes the findings from a major survey of office workers’ views on daylight

and lighting. Sixteen buildings around Britain have been surveyed and around 270 occupants were
questioned. People were generally positive towards daylight. Few felt they had too much daylight;
however, responses showed that people were more likely to be dissatisfied with the daylight when the
design average daylight factor was over 5%. At these high daylight levels there were increased
complaints of sun and sky glare. There was extensive use of venetian blinds in the sample buildings.
Control of electric lighting was a key issue. In some buildings the lamps could not be switched off
when daylight was sufficient.

Occupant reactions to daylight in offices

L Roche MA, E Dewey and P Littlefair MA PhD CEng MCIBSE
Environmental Engineering Centre, Building Research Establishment, Garston, Watford, Herts WD25 9XX, UK

1 Introduction of the project reported here is to provide a detailed evaluation,

using survey data from a range of o~.ce buildings, of the
Recommendations on daylighting in buildings are given in current daylighting recommendations.
BS 8206 Part 2~~~ and the newly revised CIBSE window design
manual<2>. A BRE reportC3) explains the guidance and gives
detailed design advice. 2 Approach
In these guides, the overall amount of daylight in a space is Sixteen office buildings around Britain were identified as
quantified by the average daylight factor DF. A key advantage survey sites. The ages of the buildings ranged from one year
of the average daylight factor is that it can be represented by to nearly a century. Three-quarters of the buildings were
a simple fc~~aul~:
between 20 and 30 years old. Most of the other buildings had
received a major refurbishment in the last three decades. All
_
W/76 the buildings were lit by side windows, and most had had
~~ = ~~i _ ~~~ their electric lighting systems recently refurbished. Some of
the c~~ce~ were high-specification buildings with high ceil-
expressed as a percentage, where W is the glazing area, T is the ings, large windows and a low density of occupants in open-
glass transmittance and 0 is the angle (in degrees) of sky plan vf~~. Others were of low specification, with a high
visible from the centre of the window measured in a vertical density of occupants, lower ceilings and less daylight (owing
section through the window. With no obstruction, 0 is 90° for partly to crowded sites). Photographs of the buildings were
a vertical window. Big obstructions outside can lower 0 signif- taken, but it was agreed not to publish these. The surveys
icantly. r~i is the area of all room surfaces (ceiling, floor, walls involved interviewing the facilities managers, and approxi-
and windows) and .F~ is their average reflectance. mately 20 occupants in each building were asked to complete
a questionnaire comprising 29 questions, almost all multiple
The equation can be inverted to give the required window
choice.
area for a particular av e daylight factor.
The BS code(’) recommends that ‘if electric lighting is not The questions in the questionnaire mostly asked for subjec-
normally to be used during daytime, the average daylight tive, quantitative answers about aspects of the visual environ-
factor should be not less than 50A If electric lighting is to be ment. Seven questions related to satisfaction with daylight
used throughout daytime, the average daylight factor should electric lighting, view, etc. Three questions related to prefer-
ences and frequency of working by daylight and electric
be not less than 2%*. According to the British Standard, these
recommendations should be adopted ’where a predominantly lighting. Six questions asked about perceived levels of
daylit appearance is wanted*. daylight and combined ii~ht~ng. Five questions were about
glare and reflections. Five questions were about attitude to
These recommendations have been widely adopted in other d environmental performance of the windows. One ques-
daylight design guidance, for example CIBSE’S DayhghaW tion asked about the c~f c~ut~i of electric lighting. The
and ~’ design(2), and their Code fcrr interior li,~~ ‘ ~~; the other questions were about the occupation and general atti-
BRE guidance on site layout planning<5) and the environ- tudes of the respondent.
mental labelling scheme BREEAM(6). If used in the development
of a real building, the 5% daylight factors criterion sets An example of three related questions is shown in Table 1.
an important constraint on design. While 2% average DF is
The opinions of the occupants needed to be related to the
’ I~ easy to achieve, 5% often requires unusually large levels of light that they were experiencing. To do this, the type
window areas in side lit rooms, particularly if the space is at
of electric lighting installation and availability of daylight in
all deep. each ~a~c~ had to be determined. As recommendations on
Despite their widespread use, the recommendations on daylight in buildings are quantified by the average daylight
average daylight factor have not, so far, been based on detailed factor, it was necessary to calculate this parameter for each
research, either in the laboratory or in real buildings. The aim room surveyed.

Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016

119
A physical survey of each office area was undertaken, to enable general statistical analysis. In order that numerical techniques
the design daylight factor to be determined. The areas of the could be applied, each of the multiple-choice options was
different surfaces of each room were measured, as was the area converted to an integer scale. For example, answers to a yes/no
of window glass, excluding the framing. The reflectances of question were represented by
the various surfaces in the offices (desks, partitions, floor
No=0
covering, ceilings) were determined using a luminance meter
by comparison with a reference card of known reflectance. Yes= 1
The tr~nsmittan~e of the window glass was determined using
Answers to the question ’Do you prefer working by daylight,
an illuminance meter where possible. In many of the office
electric lighting or a co~zbinatican?~ were represented by
surveyed, the window opening was too small to enable this
measurement to be carried out. Where the transmittance Daylight =
1
could not be measured, it was assumed, and corrected for dirt,
Combination =
2
with the aid of the descriptive tables in reference 3. The loca-
tion and measurements of internal and external obstructions Electric lighting
=3
were also recorded, as well as the location of each respondent
relative to the nearest window. providing a scale with decreasing preference for daylight.
Answers to the question ’In general, how do you rate the year
Facilities managers were interviewed about their building in round daylight at your v~~~’ were represented by
general, and about the performance and reason for installing
the current lighting system in particular. Their comments, Far too little light 1 =

including anecdotal references, were recorded. References was Too little light 2 =

also made to whether and how they explained the use of

lighting controls to the ofice occupants. About right = 3
Selected occupants who had been working in that ofice for at Too much light 4 =

least six months were asked to complete questionnaires asking Far too much light 5 =

their opinions on various aspects of their working environ-

ment, paying particular interest to the daylighting and the providing a scale of increasing quantity.
electric lighting. This approach allowed the application of a wide range of
Illuminance readings were recorded on the desks of the occu- statistical analysis tools.
pants at the time of distributing the questionnaires, and each Of principal interest was how the average daylight factor (ADF)
respondent’s position in relation to the window, and the for each occupied space related to the responses of its occu-
distance to the nearest window, were recorded. pants. The distances of most occupants from the nearest
window were available, and analysis of the effect of distance
It was observed that some people chose to work at very low
on the answers given was also undertaken.
daylight illuminances with no electric lighting; in some cases
this was because they did not know how to operate their elec-
3.1 ADF and individual ~s~~.s
tric lighting. As the initial survey took place during the
winter, a repeat visit to most of the buildings under summer- The ADF was significantly linearly correlated with several of
time conditions was undertaken, with the same occupants, the individual questionnaire answers. Of interest was the fact
where possible, asked to complete the questionnaire again. that several of these correlations only became clear as the data
This was to see whether their opinions varied significantly at set became . It is clear that a much wider survey would
different times of the year. give even more detailed and robust results.
The amountof cloud cover on the day the building was The ADF was found to be a useful predictor of the general level
surveyed was also noted, along with position of the blinds; of daylight, as well as of the general level of combined daylight
and in some of the buildings the ’no sky line’ was also and electric lighting. Questions of the form ’In general how do
determined. you rate the year round daylight at your desk?* were asked
about levels of daylight and combined light in the room, at the
desk, and at the VDU. Of these six questions, five were very
3 Findings significantly correlated with the ADF (~~et~.t~3~~ and the other
was correlated with a lower significance (p =
0.013). Figures 1
The questionnaires and the associated objective data collected to 6 show the mean responses of occupants in spaces with
at the time provided a suitable data set for performing different ADFS, and regression lines for the relationship

Table 1 An example of three related questions

120 Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016

 between ADF and the responses given. An important feature is above ’just right’ for any choice of ADF. There were very few
that the mean response to these six questions hardly rises people who thought there was too much daylight.
A comparison with the answers to questions about satisfaction
with daylight provides what appears at first sight to be a
paradox, as satisfaction with daylight is rather higher for lower
ADFS, although the significance is quite low. A comparison
between Figures 7 and 8 is intriguing. Satisfaction with
daylight is rather lower for the rooms with ADF over 5%,
whereas opinions of the quantity of daylight suggest that it is
about as often too much as too little in the rooms with ADF
over 5%, but more frequently too little in the rooms with ADF
less than 5%. A possible explanation is that excessive levels of
daylight are associated with greater dissatisfaction than too
little daylight because of the stronger psychological effects
involved with factors such as glare and overheating. Also,
levels of daylight that are considered too low may easily be
supplemented by electric li~hting, whereas levels that are too
high are associated with problems that are more cult to
deal with.
Of interest was the question of the ideal choice of ADF for
occupant satisfaction. The correspondence between mean
Figure 1 ADF versus mean perceived year-round daylight level on desk satisfaction levels and ADF (Figures 9 and 10) suggests that the
(2 too little, 4 too much)
= =

Figure 2 ADF versus mean perceived year-round daylight level in the room Figure 4 ADF versus mean perceived year-round total light level on desk
(1 far too little, 4 = too much)
=
(2 too little, 4 too much)
= =

Figure 3 ADF versus mean perceived year-round daylight on the vDu Figure 5 ADF versus mean perceived year-round total light in the room
(2 =
too Hide, 4 = too much) (1 far too little, 4 too much)
== =

Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016

121
Figure 6 ADF versus mean perceived year-round total light on the VDU Figure 9 ADF versus mean satisfaction width daylight today (1 =
very satis-
(2 too little, 4 too much)
= =
fieci, ~ 5 = very di,~ssatisfed)

Figure 7 Mean safshfon with daylight over past year for zones with Figure 10 ADF versus mean satisfaction with daylight over past year
ADF ~5°~o and ADF > 5% (1 =
very satisfied, 5 = very di~satisfied)

ADFS between 2% and 5% can give the highest mean levels of

satisfaction, with ADFS outside this range giving lower typical
mean satisfaction. Satisfaction with daylight is a complex
issue depending on many other factors such as ~.~de orien-
tation and obstructions.
The ADF was moderately correlated with the incidence of sun
glare and sky glare (p<O.02) (Figures s I~ and 12), rooms with
higher ADFS having a higher incidence of both types of glare.
It is noticeable that low incidences of sun glare occurred only
when the ADF was low, but high incidences occurred over the
whole range of ADFS, suggesting that this problem bet~~~
dealt with in some of the low-ADF spaces than in others, but
was not solved in any of the high-ADF spaces.

The ADF is also more weakly related to other subjective

parameters such as increasing satisfaction with the view
(Figure 13). In this graph it is noticeable that the two exam-
ples with ADF over 14% are anomalous. In these buildings,
continuous usage of blinds necessitated by the very large
Figure 8 Responses to ’In general how do you rate the year-round daylight windows spoiled the view for much of the time. If these exam-
the desk?’ for ADF <5% and ADF >5%
at
ples were removed from the analysis, a stronger link between
ADF and satisfaction with the view would be apparent.

122
Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016
Figure 11 ADF versus mean frequency of glare from the sky (1 =
~ften, Figure 14 Frequency o~’prefer~nces fc~r daylight, combined lighting or elec-
4 =
never) tric lighting

3.2 Working by daylight

One of the central objectives of the study of daylighting is to
determine how to make it possible for people to use daylight
more and artificial Iighting less. The questionnaire asked
subjects which type of lighting they preferred, finding that a
slight majority of the respondents preferred to work by
daylight alone, with most of the rest preferring a combination
of daylight and electric lighting. Only a small minority
preferred to work by electric lighting (Figure 14). Two ques-
tions were also asked about how often in practice people
worked by daylight alone.
A simplistic analysis would suggest that the more daylight
that is provided, the more people will be able to work by
daylight and the less energy will be used for electric ~ghtin~*
Analysis of the data gathered in this study suggests that this
may not be the case.
The results show that, in the buildings surveyed, people most
Figure 12 ADF versus mean frequency of glare. from the sun (1 =
often, often worked using electric lighting as well as daylight, but
4 = never) that the mean frequency of daylight working was maximised
for some intermediate level of ADF. Quadratic regression on
the two relevant questions suggests that this value is around
8%: a high, but not maximum ADF.
A possible explanation for this is that, in buildings with very
large amounts of glazing, the blinds are frequently pulled
down and the electric lighting is switched on, whereas, under
the same circumstances in a building with more moderate
levels of glazing, daylight can be more effectively harnessed.
This hypothesis certainly needs more thorough testing before
it can be considered to be established, but if it is true the rele-
vance to daylight energy savings is substantial.

3.3 Underlying factors

Factor analysis was used as a way of determining the under-
lying modes of variation that led to the entire data set of ques-
tionnaire answers. The factors revealed can be variation in the
physical environment, variation in the individuals themselves
and their specific requirements, and other factors such as the
conditions at the time when the questionnaire was given.
13 ADF versus mean satis~actic~n with the view (1 = very s~tisfed,
5 very dissatisfied)
= Data reduction is one reason why factor analysis is worth
doing. The most important factors explain much more of the
variation in the data set than individual questions. In this
instance, the five most important factors explained 48% of the

Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016

123
variance in a data set with 29 questions. A description follows
of these five factors in the questionnaire data set, and the phys-
ical indicators that were correlated with these factors.

factor d: Daylight adequacy,

Factor 1 is the strongest single factors, accounting for 15% of
the total variance of the questionnaire data set. Factor 1 corre-
lates strongly with increasing perceived daylight levels,
increasing satisfaction with daylight, and increasing perceived
combined light levels. It also correlates with increasing
perceived window size and happiness with view and with
happiness with the general environment.
This factor expresses the large amount of variability in the
amount of daylight provided in the buildings surveyed, and
that this variation was associated with the satisfaction of the
occupants with the daylight and their general happiness. Fi 15 Frequencies of sums of negative ~ers to questions about nega-
tive environmental factors
Unsurprisingly, the season of the year was an indicator of day-
light adequacy, with significantly higher daylight adequacy in
the summer questionnaires. The average daylight factor was individuals are generally more inclined than are others to give
also found to be a good indicator of daylight adequacy, high positive or negative answers to questions about negative envi-
ADFS being associated with higher adequacy. ronmental factors. This variation in individuals could be
quantified by research under controlled conditions. Figure 15
F 2: Choke r~, shows the distribution of the summed responses to all the
f ‘~er~ lighting out of necessity
questions about negative factor. This demonstrates a skew
Factor 2 accounts for nearly 12% of the variance of the data distribution with the median well to the left. The spread of the
set. This hector is associated with increasing frequency of curve indicates that some individuals respond positively to all
working by combined and electric lighting as the perceived questions while others give extreme negative answers to
levels of combined lighting increase, while satisfaction with all questions. The responses to individual negative factors are
daylight and the view decrease. Factor 2 expresses the varying strongly correlated with others. Figure 16 shows that there
need to supplement daylight with electric lighting when was quite limited variation from building to building of the
daylight alone is not sufficient. mean complaint score, all but four buildings being in a narrow

The season had a strong influence on this factor, respondents range. The fact that there were no strong physical indicators
for this factor supports the hypothesis that it is a psycholog-
indicating that combined lighting was chosen over daylight ical one.
more frequently in the winter, showing the seasonal depen-
dence of responses to questions about the whole year. The
distance from the nearest window was also a useful indicator,
Factor 4: G of daylight forking ’lJerSUS sun glaare and daylight
with combined lighting being chosen less with increasing reflections
distance of the respondent from the nearest window. The ADF Factor 4 accounts for 6% of the variance of the data set. This
was not a strong indicator for this factor, showing that the factor is associated with increasing preference for working by
variation within a room has a strong influence on the need for daylight and increasing time spent working by daylight alone.
electric lighting. It is also associated with decreasing perceived glare from the
sun and reflections from the sun in the computer screen, as
Facsar 3: Glare and reflections from sky, sun and a!’X~CM~ RgMMg well as increasing assessment of combined light levels on the
and other negative factors desk and VDU.
Factor 3 accounts for 10% of the variance of the data set. This
factor is associated with increasing glare from sun and sky and
reflections from daylight in computer screens. It is also asso-
ciated with increasing incidence of reflections on computer
n-om ceiling and lighting, and decreasing
happiness with the general environment and satisfaction with
the electric light, ~n ~d incidence of overheating from
sunshine and incidence of cold draughts are also correlated.
This factor is associated with decreasing perceived amount of
daylight and combined light at the VDU.
Factor 3 requires c~re~~ interpretation. It is associated with
increases in all seven negative indicators in the questionnaire,
but it is difficult to imagine a physical mode of variation that
would cause ail of them ? increase at the same time. The asso-
ciation with decreasing perceived daylight and combined
levels at the VDU at the time as glare and reflections are
increasing is also puzzling.
One hypothesis is that factor 3 is principally a mode of varia-
tion in the individual rather than the environment. Some Figure 16 Mean complaint scores four each buildings surveyed

124
Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016
Factor 4 can best be described as the inclination to work by

daylight out of choice, where combined light levels are too

high and glare and reflections from the sun are not a problem.
The ADF had a borderline level of correlation with this factor
with ADF increasing as preference for daylight working
decreased and perceived glare problems increased. This sug-
gests that the choice involved in this factor was influenced to
a large degree by the glare problems, which are correlated with
the ADF.

F 5: 1’re~’~rc~ce fcrr ~t seat

Factor 5 accounts for 5% of the variance of the data set. Factor
5 is strongly related with increasing importance of being near
a window, and is also associated with increasing perceived
window size and with increasing happiness with the window
size. However, this factor is also correlated with increased
occurrence of overheating from the sun and with having
Figure 17 ’How important is it to you to have a window in your room or
particular dislikes about the workplace. It is also associated immediate working environment?’
with decreasing problems with reflections from ceiling
lighting.
Factor 5 can be described as working near a window out of
choice, with the downside of more overheating problems.
The distance from the nearest window is an excellent indi-
cator for this factor, with distance decreasing as the impor-
tance of being near a window increases. This that for
the respondents in our survey the distance at which people
work from a window is to a significant extent a matter of
personal choice. An alternative explanation is that proximity
to a window increases its perceived importance. Both possi-
bilities could help to explain the results.

4 Conclusions

The importance of providing daylight in the workplace has

been supported by the results. The single strongest factor in
the questionnaire data was the concept of how adequate the
daylight was perceived to be. In answer to the question of Figure 18 Perceived d of control of electric lighting
which combination of lighting people preferred to work by,
only 4% preferred electric lighting alone. It was also noticeable The importance of the preferences of the individual was very
from the factor analysis that, whereas many people worked by visible in the responses. The preferences for daylight working
daylight out of choice, they generally used electric lighting out and for working near a window have considerable influenced
of necessity.
on responses. The location of respondents in a room strongly

The alsu made clear the balance between advantage affect~d their need for electric lighting, showing how impor-
and disadvantage in the choices made by respondents. There tant local controls are for ~ig~at.~ng in deep spaces. Despite this,
was a trade-off between the glare, reflections, overheating and 4()O,r(¡ of respondents had very little or no control over the elec-
even cold draughts associated with being near a window and tric lighting (Figure 18). The importance of ~e~v~ lighting
the reflections from electric lighting. Nevertheless, most controls(7) was emphasised by the responses from occupants
people (73%) considered having a window in their work area and facilities managers. Several buildings visited had daylight
very important (Figure 17). linking systems, which had been disconnected after occupants
The suggested t the u t u~’ time that respon- had complained about lamps switching off suddenly.
dents spent working by daylight alone was maximised for Occupancy ~~zsia~g systems were also not appreciated in small
some high but intermediate level of ADF, estimated at 8%. The areas, or areas with several partitions. Several respondents
explanation for this seems to be die phenomenon of pulling complained of having to wave at the lamps to get them to
down the blinds and switching on the lights when there were come on again, particularly unappreciated in toilet cubicles.
problems with large areas of glazing. This tentative result has Some buildings ran purely on occupancy sensing, with no
implications for daylight energy saving calculations. daylight linking or easily accessible manual override, which
The average daylight factor was found to be a good indicator would not ensure the optimum use of daylight. Some occu-
of daylight adequacy, and was also incidentally significantly pants wished to have individual remote sensors to control
correlated with the incidence of sun and sky glare. It is likely their lighting. However, when hand-held remote controllers
that a better indicator for these phenomena could be derived, were provided, they were often lost or not easily accessible.

providing a design parameter for architects. Occupants were sometimes unaware of the location of
125
Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016
switches, either because these locations were illogical or Did the survey consider aspects of environmental comfort
because the switches were obscured by furnishings. other than the lighting, namely thermal and acoustic comfort?
Satisfaction with daylight is a complex phenomenon. The fact In-depth interviews have shown other factors, such as the
that problems with glare and reflections have a strong influ- ’likeability’ of the person at the adjacent work station, signifi-
ence means that satisfaction is highest for lower levels of cantly affect a person’s perceived comfort level and that if they
do not get on well with that other person then they complain
daylight, despite the fact that mean assessments of the per- of discomfort fram lighting, heating, etc. in attempt to have
ceived level of daylight in a space rarely rise much above the
’about right* level. This study can only provide a rough figure,
themselves moved from that position in the o~c~, thus
but it appears that satisfaction with daylight can be maximised masking their unbiased opinion of visual comfort.
for some level of ADF between 2% and 5%. The mean level of R S W ebb (Heriot-Watt University)
satisfaction with daylight varied considerably between spaces
In the survey, were occupants asked to rate the importance of
with similar ADFs in this range, indicating that other design
factors such as orientation and the effectiveness of blinds are daylight in comparison with other environmental factors?
important. Stephen Cannon-Brookes (Cannon-Brookes Lighting &
Design)
Acknowledgements Is there any evidence of satisfaction declining at high ADFS
relative to VDT luminance?
The authors thank Dr D. Carter and Mr T. Moore of the
University of Liverpool for their invaluable assistance prior to Authc~rs‘ response to discussion
and during the surveys, and Mr A. Motin for collecting many
of the physical measurements and assisting with transport.
The results of the experiment suggest that an average daylight
Thanks are also due to the c~~ce occupants and facilities 1&dquo;actor of 109% would be likely to lead to low levels of satisfac-
managers for their enthusiastic cooperation with the surveys. tion due to problems of glare and overheating, particularly in
This paper was produced as part of the research programme a workspace where computers are used. An average daylight
of the Construction Directorate of the Department of the factors of no more than 5% will generally result in significantly
Environment, Transport and the Regions. higher levels of satisfaction.
The second and third questions address a similar issue. The
References survey did ask questions about other environmental problems
such as overheating and draughts, as well as about the occu-
1 British Standards Institution BS8206 Part 2: Code of practice for pants&dquo; attitude to the general environment, but the respon-
daylighting (London: BSI) (1992) dents were not explicitly asked to rate the importance of
2 Chartered Institution of Building Services Engineers Lighting Guide 9: daylight in comparison with other environmental factors.t
Daylighting and window design (London: CIBSE) (1999) The answers to questions asked about their preference for
3 Bell J and Burt W BREDesigning buildings for daylight Report BR 288
(Garston: Building Research Establishment) (1995)
being near a window and for working by daylight in combi-
4 Chartered Institution of Building Services Engineers Code for interior
nation with those about overheating, draughts and glare
(London: CIBSE) (1994)
lighting would, however, provide some insight into this issue. The
5 Littlefair P J Site layout planning for daylight and sunlight; a guide to good sizeable sample should ensure that random noise factors such
practice BRE Report BR 209 (Garston: Building Research Establishment) as whether respondents get on with their co-workers should
(1992) not cause any systematic distortion of the results. Respon-
6 Baldwin R, Yates A, Howard N and Rao S BREEAM 98 for offices BRE dents were aware that the questionnaire responses would be
Report BR 350 (Garston: Building Research Establishment) (1998) anonymous for statistical purposes and would have no effect
7 Slater A I, Bordass W T, Heasman T A People and lighting controls on future omce moves.
Information Paper IP 6/96 (Garston: Building Research Establishment)
(1996) In response to Dr Cannon-Brookes, as the study did not
collect data on VDT luminances it is not possible to draw any
Discussion conclusions about the dependence of satisfaction on the rela-
tionship between ADFS and this parameter.
G Philips (Consultant)
Would it be far to regard an daylight factor of lOOk as °~ Note: Research on the perceived relative importance of different environ-
mental factors is bemg conducted by Geoff Levennore at UMIST, see Build.
being an absolute, or desirable, upper limit to design? Sm. Eng. Rim Te . I~2} 113-118 (1994) and 17(1) (1996) (S A Fotios)
PJ (University of the West of’England)

126
Downloaded from lrt.sagepub.com at University of Sheffield on March 29, 2016