You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/200550149

The Relationship Between UV Transmittance and Color -- Visual


Description and Instrumental Measurement

Article  in  Textile Research Journal · February 2008


DOI: 10.1177/0040517507081302

CITATIONS READS
21 506

5 authors, including:

Cheryl Anne Wilson Peter Gies


University of Otago Australian Radiation Protection and Nuclear Safety Agency
85 PUBLICATIONS   703 CITATIONS    70 PUBLICATIONS   2,044 CITATIONS   

SEE PROFILE SEE PROFILE

Alan Mclennan Nicola K. Bevin


Australian Radiation Protection and Nuclear Safety Agency
4 PUBLICATIONS   68 CITATIONS   
6 PUBLICATIONS   206 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Sun Protection View project

Cot death and bedding products View project

All content following this page was uploaded by Cheryl Anne Wilson on 31 January 2018.

The user has requested enhancement of the downloaded file.


Textile Research Journal
http://trj.sagepub.com

The Relationship Between UV Transmittance and Color Visual Description and Instrumental
Measurement
C.A. Wilson, P.H. Gies, B.E. Niven, A. McLennan and N.K. Bevin
Textile Research Journal 2008; 78; 128
DOI: 10.1177/0040517507081302

The online version of this article can be found at:


http://trj.sagepub.com/cgi/content/abstract/78/2/128

Published by:

http://www.sagepublications.com

Additional services and information for Textile Research Journal can be found at:

Email Alerts: http://trj.sagepub.com/cgi/alerts

Subscriptions: http://trj.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
Textile Research Journal Article

The Relationship Between UV Transmittance and Color –


Visual Description and Instrumental Measurement
C.A. Wilson1,2, P.H. Gies3, B.E. Niven2,
Abstract This work examines sensory and instru- A. McLennan3 and N.K. Bevin4
mental methods for quantifying color of fabrics 2
The Centre for the Application of Statistics and
and investigates the relationship between fabric Mathematics, University of Otago, Dunedin, New Zealand
color and UV transmittance. The color of 175 sam- 3
Ultraviolet Radiation Section, Non Ionising Radation
ples was visually categorized by human judges Branch, Australian Radation Protection and Nuclear
into color groupings and measured instrumen- Safety Agency, Yallambie, Australia
tally according to the CIE XYZ and CIE LAB sys- 4
Clothing and Textile Sciences, University of Otago
tems using a spectrocolorimeter. The relationship Dunedin, New Zealand
between color and UVA, UVB and UVR transmit-
tance was examined and the most appropriate
method for describing the relationship between
UV protection and color was determined. Color
depth rather than color per se affected mean UV
transmittance with dark fabrics confirmed as being
most effective. L* (lightness), Y (brightness), were
found to be equally effective descriptions of the
‘color’/transmittance relationship. Manufacturers
seeking to color fabrics so as to minimize UV
transmittance should select colors with CIE L* of
approximately < 38 or CIE Y of approximately
< 28. Medium and high transmittance levels were
less clearly related to CIE values.

Key words CIE tristimulus (XYZ), CIE LAB,


color, UV, transmittance, threshold

Introduction red) and light colors (oatmeal and white) differed with
UPF values for the two groups of colors being 50+ and 14–
16 respectively. Differences in protective capacity between
UV protection provided by fabric has been linked to a dark and light colors and also between different colors has
number of fabric properties. Fabric structure, thickness, been attributed to the absorption of radiation by the dye in
weight, extensibility, and finishing treatments such as dye- the part of the dyes absorption spectrum which overlaps
ing modify the proportion of UV light transmitted through the UV spectral region [9].1 The selective absorption of vis-
fabrics. The color and color depth of fabric has been ible light (and hence color) varies according to the chemi-
related to UV transmittance, with light colors transmitting cal structure of the dye [10]. Thus a proportion of UV
more UV than dark colors [1–4]. Dark colors such as dark wavelengths may be absorbed by the dye decreasing trans-
green, red, navy blue and black are reported as providing mittance over part of the spectrum and increasing protec-
good protection (3, 5–7] while darker shades of the same
hue have also been associated with higher ultraviolet pro-
tection factor (UPF) ratings [8]. Gies et al. [2] showed 1
Corresponding author: Clothing and Textile Sciences, Univer-
graphically that measured transmittance through a base sity of Otago, Dunedin, New Zealand. Tel.: +643 4797546; fax: +643
fabric dyed into a range of dark (black, navy blue, green, 4797538; e-mail: c.wilson@ otago.ac.nz

Textile Research Journal Vol 78(2): 128–137 DOI: 10.1177/0040517507081302 www.trj.sagepub.com © 2008 SAGE Publications
Los Angeles, London, New Delhi and Singapore

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
The Relationship Between UV Transmittance and Color C. A. Wilson et al. 129 TRJ

tion [1, 5, 10]. Color depth will affect both the absorption Thus, the X value is the red, the Y the green, and the Z the
and the reflectivity of UV photons by the fabric with dye blue primary [17]. The Y value is also proportional to the
molecules proposed as having a specific role. Whether luminance of the color and approximately related to the
color depth modifies the effect of other variables known to perception of brightness. A Y value of 100 represents a
affect UPF (such as fabric structural properties), and in color that has 100% luminance and so reflects all light (flu-
both UVA and UVB regions, is currently unclear. orescence may result in Y values exceeding 100 [18]), e.g.
To date, when examining the relationship between color black has a Y value of 0 while white is associated with
and UV transmittance visual classification of fabrics as equal values of X, Y and Z.
dark or light and into color categories (e.g. blue, black, Another system used to describe color measurement is
etc.) has been undertaken [2, 8]. Such sensory descriptions CIE LAB which addresses the problem of unequal color
rely on visual judgment of color by judges using the cone space (i.e. where numerical differences in color do not
sensors in the human eye as the instrument of measure. agree with visual perceptions) arising when the tristimulus
Cones are stimulated by wavelengths of light reflected system is used [11]. CIE LAB utilizes nonlinear transfor-
from the surface of fabrics. Shorter wavelengths result in a mations of the XYZ values to specify color perceptions in
perception of violet and blue colors (400 to 500 nm), and in terms of a three-dimensional space. Colors are repre-
the longer wavelengths (approximately 600–700 nm), red. sented by L*, a* and b* coordinates where L* represents
Classification of color is complex, e.g. an object will appear lightness (from 0 (black) to 100 (white)), a* represents red-
white when incident light is completely reflected and black green (positive a* = red, negative a* = green) and b* yel-
if all light is absorbed with perceived color varying depend- low-blue (positive b* = yellow, negative b* = blue). The
ing on the source of the light. Thus variations in perceived a* and b* coordinates are close to zero for neutral colors
color may occur as a result of differential absorption and such as white, black or grey while more saturated colors
reflection of incident light [11]. Hue, brightness and satu- have higher a* and b* coordinates. CIE LAB has some-
ration are attributes used to describe visually assessed what replaced the use of the CIE chromaticity scale and is
color. Hue refers to the wavelength of light stimulating the scale recommended for use in industry [16].
cones in the eye with all color composed of proportions of
blue, green and red stimuli [12]. Brightness is dependent on
the luminance and reflectivity (the more lumens emitted Method
by an object the brighter the color) while saturation is the
amount of white present (a fully saturated color has no Fabrics with UPF ≥ 40 representing a wide variety of fabric
white) [13–15]. The principal disadvantage of visual assess- colors and depths were retrospectively sampled using a
ment of color is the lack of quantification of the elements stratified random sample from the n = 3980 fabrics sub-
comprising the color, e.g. pink is a less saturated form of mitted to the Ultraviolet Radiation Section, Non Ionising
red yet both are of the same hue. In the absence of instru- Radiation Branch of ARPANSA (Australian Radiation
mental characterization of color, the relationship between Protection and Nuclear Safety Agency) for UPF classifica-
color and UV transmittance is not clear. tion over a three-year period. Stratified sampling was used
The aim of this current work was to (i) identify an instru- to select a sample balanced in terms of structure and
mental method for describing color of fabrics intended to selected fiber contents. Coated, laminated, patterned, fluo-
provide UV protection, (ii) clarify the relationship between rescent-colored and pre-washed fabrics were discarded
color and UV transmittance through fabrics, and (iii) iden- resulting in a final sample (n = 175) of 91 knits and 84
tify characteristics of color associated with lower UV trans- woven fabrics. Woven fabrics were plain weave (n = 56,
mittance. 32%) or twill (n = 28, 16%) structures. Knits were single
Color may be described numerically by measuring the jersey (n = 24, 14%), tricot (n = 32, 18%), interlock (n =
contribution of blue (B), green (G) and red (R) wave- 13, 7%), piqué (n = 9, 5%), or others such as double jer-
lengths using a spectrocolorimeter [14–16]. The Commis- sey, 1 x 1 rib, and eyelet (n = 13, 7%). Fiber types repre-
sion Internationale d’Eclairage (CIE) system of color sented included: 100% polyester (n = 39, 22%), cotton (n
classification is based on description of colors using prima- = 34, 19%), and nylon (n = 22, 13%), and blends of poly-
ries derived from RGB tristimulus values [15]. The CIE ester/cotton (n = 34, 19%), nylon/elastane (n = 22, 13%),
tristimulus values, referred to as X, Y and Z, are always cotton/elastane (n = 10, 6%), polyester/viscose (n = 8,
positive and derive their descriptions from the various pro- 5%) and nylon/cotton (n = 6, 3%).
portions of color contributing to their value according to: The new, non-laundered, fabrics were preconditioned
for a minimum of 24 hours (20°C ± 2°C 65 ± 4% R.H.)
X = 0.49R + 0.31G + 0.20B and, except for UV transmittance, were tested under these
conditions [19]. Fabric structure was described [20], thick-
Y = 0.1769R + 0.812G + 0.01063B
ness [21], and mass per unit area [22] of each specimen was
Z = 0.00R + 0.01G + 0.99B [15] measured. Fabrics ranged in thickness from 0.2 mm to 1.47

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
TRJ 130 Textile Research Journal 78(2)

mm (mean 0.70 mm, s.d. 0.28 mm) and in mass per unit standard deviation). Differences in UV transmittance due
area from 92.38 g/m2 to 431.43 g/m2 (mean 216.43 g/m2, s.d. to color and depth were examined using two factor univari-
68.66 g/m2). In all cases, n = 5 replicates (diameter 113 mm) ate analysis of variance (ANOVA) and which of the instru-
were assessed in a randomized order. Fiber content was as mental color measures within each color system best
reported by the supplier. Percent ultraviolet transmittance described UV transmittance through the fabrics was inves-
(UVA, UVB and UVR) was measured at ARPANSA under tigated using regression analysis [25, 26]. F-statistics were
normal ambient conditions according to the method speci- used to determine whether significant differences among
fied in AS/NZS 4399 [23] using a Labsphere UV-1000 variables occurred. Tests of homogeneity of variance and
diode array spectrometer. A fluorescence-reducing filter normality of residuals were performed. No potential out-
was not used as the instrument corrected for the effect of liers were identified but color data was not normally dis-
fluorescence due to dyes by monochromating after the tributed so Spearman’s rank correlation coefficient (non-
sample. The spectral transmittance was weighted with the parametric) was used to determine any relationship
CIE erythemal response and the solar spectral irradiance between each fabric’s transmittance of UVA, UVB and
[24]. Fabrics were then assigned a UPF classification based UVR, and color values. When two factor ANOVA and
on the calculated UPF rating. As fabrics were tested as regression analysis were performed, stabilization of UV
received from the manufacturer, order of testing was not variance (achieved using log transformations and square-
randomized. For each fabric a minimum of eight speci- root transformations respectively) was required to meet
mens were tested under conditions of 20°C ± 5°C and 50 ± the normality and homogeneity assumptions of the tests.
20 R.H. [23]. The need to utilize both log and square-root transforma-
Fabrics were visually assigned into (i) color (black, grey, tions possibly reflected the influence of XYZ and LAB val-
green, blue, red, orange, yellow, taupe beige, white), and ues on the residuals generated during the regression
(ii) depth (light, medium and dark) categories by three analysis. Instrumental color measures were identified using
experienced judges working independently in a grey visual regression analysis. Hierarchical cluster analysis was then
evaluation area under natural light. Where differences in used to form clusters of fabrics and group membership was
judgment occurred, colors were re-evaluated and assigned. used as the basis for exploring the relationship that existed
Color was also measured using a spectrocolorimeter between UV and color measures by (i) describing charac-
(MiniScan XE™ Model 45/0 LAV; 25.4 mm open port and teristics of each group, and (ii) determining the relation-
20.0 mm illuminated area), Hunter Associates Laboratory ship between the group and UV transmittance using
Inc., 11491 Sunset Hills Road, Reston, Virginia 22090-5280, Spearman’s correlation coefficient.
U.S.A.) calibrated in accordance with CIE 51 [16] as
required for measuring daylight color (D65/10) and accord-
ing to the instruments instructions (e.g. calibration was Results and Discussion
performed at the beginning, end, and every four hours
throughout each measurement session). The MiniScan XE Fabric UV and color characteristics are described in Table 1.
was calibrated using black and white tiles where the XYZ UVA, UVB and UVR transmittance through the fabrics
color scale values for the white tile were: X = 79.4, Y = 84.1 ranged from 0.02% to 10.06%, 0.02% to 2.12% and 0.02%
and Z = 90.2. CIE XYZ and LAB values were determined to 8.12% respectively reflecting the high UPF ratings of
for each fabric (n = 5) with specimens placed technical face these fabrics.
upward on a flat, hard, grey surface. Color was measured in
the 0° and 90° orientations and results averaged. Specimen
order was randomized with one scan per specimen simulta- Visual Assessment
neously determining both CIE XYZ and LAB values. Color Fabrics were visually categorized as: blue (n = 46; 26.3%),
was determined both with and without a fluorescence red (n = 25; 14.3%), green (n = 22; 12.6%), white (n = 20;
reducing filter. Use of a filter appeared to mask differences 11.4%), taupe/beige (n = 17; 9.7%), yellow (n = 15; 8.6%),
in color among fabrics, and color measurements recorded black (n = 12; 6.9%), grey (n = 11; 6.3%), and orange (n =
with the filter do not represent the color viewed by the 6; 3.4%). Fabrics were also categorized as dark (n = 62;
human eye. In the absence of relationship between filtered 35.4%), medium (n = 78; 44.6%), and light (n = 35; 20%).
values and UV transmittance this paper details examina- As expected not all shades were represented in each color
tion of unfiltered color measurements. set (Figure 1), e.g. white fabrics were always categorized as
light; beige as light or medium; yellow and orange as
medium only, etc.
Analysis UVA, UVB and UVR transmittance through the fab-
Characteristics of the fabrics, visual classifications, and rics were modified by the color of the fabric (F8,158 = 3.07,
CIE XYZ and LAB measurements of color, and UVA, p ≤ 0.01; F8,158 = 2.24, p ≤ 0.05; F8,158 = 3.02, p ≤ 0.01 respec-
UVB and UVR transmittance were first described (mean, tively). However, when the effect of both color and depth

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
The Relationship Between UV Transmittance and Color C. A. Wilson et al. 131 TRJ

Table 1 Description of UVA, UVB, UVR and CIE tristimulus represented (Table 2). Color depth, not color per se, was
and CIE LAB systems over all fabric types (n = 175). confirmed as the most important variable affecting trans-
mittance, which may explain the lack of an apparent
Variables Mean s.d. Minimum Maximum hierarchy noted when evaluating the effect of color, e.g.
a UV(%) transmittance Zhang et al., [3] found cream colored samples had higher
transmittance than white and maroon lower than black,
UVA 2.11 2.20 0.02 10.06 while Gorensek and Sluga [27] found orange was a more
UVB 0.73 0.55 0.02 2.12
effective block to UV than red dyed fabric suggesting this
UVR 1.82 1.78 0.02 8.12
effect, consistent with the absorption spectrums of orange
b CIE XYZ system and red dyes, was due to the different dye molecules. Pale
X 30.17 28.65 1.29 96.86 blue was also shown to be less protective than pale orange
Y 29.81 29.49 1.34 100.78 which was in turn less protective than dark blue [27].
Z 28.57 32.77 1.52 122.53 While it was not possible to compare UV transmittance
c CIE L* a* b* system
through pale blue and pale orange fabrics with the data
from the current study (as no pale orange fabrics were
L* 56.26 31.07 11.52 115.58 included in the sample), arranging colors in order accord-
a* 8.50 23.96 –68.94 84.70 ing to their UV transmittance identified similar relation-
b* 5.74 31.24 –50.72 116.70
ships for other colors of interest (Figure 1b). Of note is
that medium depth orange fabrics transmitted less UV on
were examined, color did not modify transmittance (F8,158 = average than red fabrics (irrespective of depth), taupe/
1.97, NS; F8,158 = 1.15, NS; F8,158 = 1.83, NS respectively). beige transmitted more than white, and that generally
Instead differences in UVA, UVB and UVR transmittance order of transmittance through the fabric reflected depth
were explained entirely by variations in depth (F2,158 = with most transmittance occurring through light colors
11.23, p ≤ 0.001; F2,158 = 4.57, p ≤ 0.05; F2,158 = 11.14, p ≤ and least through dark.
0.001 respectively). Dark colors were associated with the
lowest, light the highest and medium depth fabrics inter-
mediate levels of UVA, UVB and UVR transmittance (rs =
Instrumental Measurement CIE XYZ Values
0.47, p ≤ 0.001; rs = 0.34, p ≤ 0.001; and rs = 0.46, p ≤ 0.001 Mean XYZ and LAB values are given in Table 1 b and c.
respectively). Such a finding is consistent with reports that Tristimulus XYZ values were related to UVA, UVB and
dark colors (black, navy blue, green, and red) behaved dif- UVR transmittance. As CIE values increased so did per-
ferently to light colors (oatmeal and white) [2, 3], high- cent UV transmittance (Table 3 a). However, while the posi-
lights the need to emphasize depth rather than color when tive relationships between XYZ and UV were not
promoting sun-smart behavior, and supports identification particularly strong they were in the order of (or better
of an instrumental means of (i) detecting differences in than) the relationship found between UV and thickness,
color depth, and (ii) threshold levels. and UV and mass (r = –0.47, p ≤ 0.001 and –0.23, p ≤ 0.001
In terms of interaction, the effect of depth on trans- respectively) (generally accepted as being important indi-
mittance did not vary with the color of the fabric being cators of UV transmittance characteristics [4, 9, 28]) in this
examined (F7,158 = 1.08, NS; F7,158 = 0.57, NS; F7,158 = data set. UV transmittance increased as each of the XYZ
1.07, NS). However, as not all depths were represented values increased. The strongest relationships were between
within each color category (Figure 1) the relationship UVR and UVA, and the X, Y and Z values.
between color and depth was investigated further by exam- With the LAB system only L*, an indicator of lightness,
ining grey, red and blue fabrics in which all depths were was correlated to UVA, UVB and UVR. A slightly more

Table 2 Differences in UV transmission through fabrics visually categorized by color (grey, red and blue) and depth (light,
medium and dark).
UVA UVB UVR
Variables F Sig F Sig F Sig
Color (df = 2,74) 2.13 NS 0.76 NS 2.03 NS
Depth (df = 2,74) 10.69 p ≤ 0.001 4.13 p ≤ 0.001 10.49 p ≤ 0.001
Color by depth (df = 4,74) 0.62 NS 0.69 NS 0.66 NS
Two factor univariate analysis of variance

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
TRJ 132 Textile Research Journal 78(2)

Figure 1 (a) Percent of total fabric


sample visually assigned specific
color and depth (D = dark, M =
medium and L = light) categories,
and (b) Mean percent UV transmit-
tance through each fabric and
depth category (n = 175).

dependable relationship with UVR and UVA was observed Lightness (CIE L*) differed according to the fabric’s
than with UVB (Table 3 b). color and depth coding (F8,158 = 12.25, p ≤ 0.001; F2,158 =
68.79, p ≤ 0.01 respectively) as did brightness (CIE Y)(F8,158
Instrumental CIE Measurements vs Visually = 23.41, p ≤ 0.001;F2,158 = 60.98, p ≤ 0.01 respectively).
Five subsets of CIE L* and six subsets of CIE Y were iden-
Categorized Color and Depth
tified within which lightness and brightness respectively did
Mean lightness (L*) and brightness (Y) of each color not vary significantly (Figure 2). Examination of CIE Y
group and depth are shown in Figure 2a and b and CIE values revealed more overlap among colors. The effect of
color descriptions of each visually categorized fabric color visually classified color on brightness (Y) and lightness (L)
are shown in Figure 3a and b. did not vary according to the depth of the color being

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
The Relationship Between UV Transmittance and Color C. A. Wilson et al. 133 TRJ

Table 3 Relationship between UVA, UVB, UVR, and (a) CIE tristimulus (XYZ), (b) CIE LAB (L*a*b*) systems over all fabric
type, and (c) and (d) group membership determined using hierarchical cluster analysis (n = 175).
UVA UVB UVR
Variables rs Sig rs Sig rs Sig
a CIE tristimulus system
X 0.47 p ≤ 0.001 0.36 p ≤ 0.001 0.47 p ≤ 0.01
Y 0.50 p ≤ 0.001 0.38 p ≤ 0.001 0.50 p ≤ 0.01
Z 0.54 p ≤ 0.001 0.41 p ≤ 0.001 0.54 p ≤ 0.01
b CIE LAB system
L* 0.50 p ≤ 0.001 0.37 p ≤ 0.001 0.51 p ≤ 0.001
a* –0.19 p ≤ 0.05 –0.14 NS –0.19 NS
b* 0.00 NS –0.01 NS 0.00 NS
c CIE tristimulus group membership
X 0.36 p ≤ 0.001 0.29 p ≤ 0.001 0.37 p ≤ 0.001
Y 0.44 p ≤ 0.001 0.34 p ≤ 0.001 0.45 p ≤ 0.001
Z 0.40 p ≤ 0.001 0.30 p ≤ 0.001 0.42 p ≤ 0.001
d CIE LAB group membership
L* 0.43 p ≤ 0.001 0.34 p ≤ 0.001 0.44 p ≤ 0.001
rs = Spearman correlation coefficient

examined (F6,158 = 0.93, NS; F6,158 = 1.00, NS respectively). p ≤ 0.001, R2 = 28.3%). UVB was best described by L*
CIE Y, brightness, exhibited less overlap among groups of (F1,173 = 23.85, p ≤ 0.001, R2 = 12.1%), and UVR by L*
visually categorized colors than L* (Figure 2). However, (F1,173 = 53.26, p ≤ 0.001, R2 = 23.5%) with the description
given the most important relationship with visually catego- further improved using L* and b* (F2,172 = 34.14, p ≤ 0.001,
rized color was between depth and UV transmittance and R2 = 28.4%). When instrumentally describing color, ideally
that: (i) for both the L* and Y depth categories groups did the best description of fabric would describe the relation-
not overlap, and (ii) measurements within each depth ship between color and UVA, UVB and UVR. Thus, instru-
group were significantly different from each other, both L* mental characterization of color should detail both Z and Y
and Y were considered effective predictors of visually cate- values if using the tristimulus XYZ system. However, the
gorized depth and its relationship with UV transmittance. best description of the relationship between ‘color’ and UV
was using the L* (23.5%), and L* and b* (28.4%) compo-
nents of the LAB system. In all cases the percentage of the
CIE Measurements and UV Transmittance variance explained (R2) was low suggesting that when
As color measured using both the CIE systems is described designing fabrics intended to reduce UV transmittance the
by the relative proportions of three measurements (Figure 3), color, while generally accepted as an important variable,
whether a single value or combinations of values best must be considered in combination with other properties
described UV transmittance was explored using multiple such as fabric structure, thickness, mass, etc. known to mod-
regression methods. Using the XYZ system, UVA transmit- ify transmittance.
tance was best described by brightness of the fabric (Y;
F1,173 = 34.79, p ≤ 0.001, R2 = 16.3%) with only a small
improvement obtained when Y and Z were both used to
Clusters of Like Fabrics
describe color (F2,172 = 19.83, p ≤ 0.001, R2 = 17.8%). Fabrics were clustered into groups using (i) L* lightness,
UVB was best described by Z (F1,173 = 19.64, p ≤ 0.001, R2 = and also (ii) Y brightness within which color descriptors for
10.2%) and UVR by Z only (F1,173 = 19.64, p ≤ 0.001, R2 = L and Y varied only slightly. Three groups of fabrics within
16.7%), or Y and Z (F2,172 = 20.46, p ≤ 0.001, R2 = 18.3%). which L* values and three in which Y values did not differ,
For the LAB system UVA transmittance was best were identified. In both cases these groups were all signifi-
described by brightness of the fabric (L*; F1,173 = 52.86, cantly different from each other (F2,172 = 640.40, p ≤ 0.001;
p ≤ 0.001, R2 = 23.4%). A further small improvement in F2,172 = 963.13, p ≤ 0.001 respectively). Group membership
description was obtained when both L* and b* (yellow; was examined and the relationship between CIE Y, and L*
blue) were included in the predictive model (F2,172 = 33.94, clusters and UVA, UVB and UVR determined. Lowest

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
TRJ 134 Textile Research Journal 78(2)

Figure 2 CIE L* (lightness) and Y


(brightness) values for fabrics vis-
ually categorized into color groups
(lines connect colors for which the
L* or Y values do not differ signifi-
cantly).

UV transmittance values were associated with CIE Y and b and d). Fabrics assigned to the lowest transmission group,
L* values of less than approximately 28 and 38 respectively i.e. group 1 were associated with the lowest spread of the
(Table 4 a and c). As the descriptions of UV transmission three groups. Groups of Z and b* were not mutually exclu-
using Y and L* were improved with addition of Z and b* sive but examination of the group plots suggested (i) that
respectively, the clusters of Y and L* were used to examine using the tristimulus system lowest transmission was asso-
the characteristics of Z and b* values respectively (Table 4 ciated with L values of less than 38 and low Z values of less

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
The Relationship Between UV Transmittance and Color C. A. Wilson et al. 135 TRJ

Figure 3 Mean CIE tristimulus


(XYZ) and CIELAB (L*a*b*) values
of fabrics in visually categorized
color groups.

than approximately 20 (Table 4 b). When using the LAB


system low transmission fabrics were generally negative
Conclusions
b* values ideally of between 0 and –20, i.e. ‘blue’ rather
It is depth of color, rather than color per se, that is the prin-
than ‘yellow’ in character with the ‘blue’ light part of the
cipal aspect of color affecting UV transmittance, and not all
spectrum falling close to the UV range (400–500nm)
depths are able to be achieved for all colors i.e. the only
(Table 4 d). This is consistent with observations that
appropriate classification of black is ‘dark’ and white is ‘light’.
blue fabric is associated with lower transmission [10, 29].
The best description of the relationship between color
and UV was provided by the L*, and the L* and b* compo-

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
TRJ 136 Textile Research Journal 78(2)

Table 4 Description of CIE Y and L*, UVA, UVB and UVR transmission in groups of fabrics in which CIE Y and L* values do
not differ significantly and Z and b* for each Y and L* group respectively
Cluster CIE UVA (%) UVB (%) UVR (%)
Group Mean s.d. Minimum Maximum Mean s.d. Mean s.d. Mean s.d.
a CIE Y
1 9.17 7.71 1.34 27.67 1.24 1.28 0.61 0.54 1.10 1.08
2 44.02 8.85 29.48 59.40 3.88 2.85 0.92 0.49 3.22 2.27
3 81.75 10.40 64.49 100.78 2.91 2.28 0.97 0.52 2.47 1.81
b CIE Z for each Y group
1 10.95 10.48
2 36.36 22.08
3 77.74 37.96
c CIE L*
1 24.09 7.63 11.57 38.25 1.00 1.00 0.56 0.53 0.90 0.88
2 62.55 13.48 40.73 84.35 2.60 2.49 0.78 0.56 2.19 2.01
3 100.41 9.84 86.13 115.58 3.14 2.38 0.97 0.46 2.66 1.88
d CIE b* for each L* group
1 –5.8 13.99
2 8.15 29.91
3 21.44 45.02
Hierarchical cluster analysissss

nents of the LAB system. When developing fabrics with the Irradiance through Clothing Materials, Health Phys., 72(2),
intention of enhancing UV protection characteristics or 256–260 (1997).
choosing fabrics for UV protective garments, selection of 4. Sarkar, A. K., An Evaluation of UV Protection Imparted by
dyes that generate colors with CIE Y, or L* values of less Cotton Fabrics Dyed with Natural Colorants, BMC Dermatol-
ogy, 4(15), DOI: 10.1186/1471-5945-4-15, www.pubmedcentral.
than approximately 28, or 38 respectively, is recommended.
nil.qov/tocrender.fcqi?iid=12315(2004).
Color must be considered in combination with other 5. Curiskis, J. I., Postle, R. and Norton, A. H., “Fabric Engineer-
properties known to modify transmission. ing – Present Status and Future Potential,” in Objective Evalu-
ation of Apparel Fabrics, The Textile Machinery Society of
Japan, Osaka, 1983, 539–550.
Acknowledgements 6. Gies, P. H., Roy, C. R., McLennan, A. and Toomey, S., Clothing
and Protection against Solar UVR, J. HEIA, 4(3), 2–6 (1997).
We gratefully acknowledge the support of the University 7. Pailthorpe, M., Sun Protective Clothing, Textile Horiz., 16(5),
of Otago in the form of an Otago Research Grant; 11–14 (1996).
ARPANSA for access to the fabric samples; Dr D. J. Carr, 8. Reinert, G., Fuso, F., Hilfiker, R. and Schmidt, E., UV-Pro-
C. Smith and V. Paquin for assistance with selected meas- tecting Properties of Textile Fabrics and Their Improvement,
urements. Text. Chem. Color, 29(12), 36–43 (1997).
9. Srinivasan, M. and Gatewood, B. M., Relationship of Dye
Characteristics to UV Protection Provided by Cotton Fabric,
Text. Chem. Color. & Am. Dyestuff Reporter, 32(4), 36–43
Literature Cited (2000).
10. Palacin, F., Textile Finish Protects against UV Radiation, Mel-
1. Pailthorpe, M. T., Textile and Sun Protection: The Current Sit- liand Textibrichte, 7–8 519–522, E113–E115 (1997).
uation, Australasian Text, 14(6), 54, 56, 59, 61–62, 64, 66 11. Chamberlin, G. and Chamberlin, D., “Colour: Its Measure-
(1994). ment, Computation and Application,” Heyden and Sons Ltd.,
2. Gies, P. H., Roy, C. R., Toomey, S. and McLennan, A., Protec- London, 1980.
tion against Solar Ultra Violet Radiation, Mutation Research/ 12. Bray, J. J., Cragg, P. A., MacKnight, A. D. C. and Mills, R. G.,
Fundamental and Molecular Mechanisms of Mutagenesis, Lecture Notes on Human Physiology, Oxford, Blackwell Sci-
422(1), 45–22 (1998). ence, 1999.
3. Zhang, Z., Thomas, B. W., Wong, C. F. and Fleming, R. A., 13. Nave, C., “Light and Vision”. Web page, http://hyperphysics.
Fast Measurements of Transmission of Erythema Effective phy-astr.gsu.edu/hbase/vision/colper.html; http://hyperphysics.

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
The Relationship Between UV Transmittance and Color C. A. Wilson et al. 137 TRJ

phy-astr.gsu.edu/hbase/vision/colhist.html; http://hyperphysics. 22. International Organization for Standardization, “ISO 3801–


phy-astr.gsu.edu/hbase/vision/colchar.html (2000). 1977. Methods of Test for Textiles – Woven Fabrics – Determi-
14. Denton, M. J. and Daniels, P. N., “Textile Terms and Defini- nation of Mass Per Unit Length and Mass Per Unit Area,”
tions,” Manchester, The Textile Institute, 2002. Genève, International Organization for Standardization, 1977.
15. Hunt, R. W. G., “Measuring Colour,” Ellis Norwood Ltd, 23. Standards Australia/Standards New Zealand, “AS/NZS 4399:
Chichester, 1987. Sun Protective Clothing—Evaluation and Classification,” Home-
16. Hunter Associates Laboratory Inc., “Miniscan XEtm User’s bush, Australia/Wellington, New Zealand, Standards Australia/
Guide. Manual Version 1.2”, Hunter Associates Laboratory, Standards New Zealand, 1996.
Inc., Reston, Virginia, 1995. 24. Roy, C. R., Gies, P. H. and McLennan, A., Sun Protective
17. Majunder, A., “PDS Consulting”. Webpage, http://www.pds- Clothing: 5 Years of Experience in Australia, Recent Results in
consulting.co.uk/glossary/XYZ.htm (1999). Cancer Research, 160, 26–34 (2002).
18. HunterLab, Applications Note: Ultraviolet-Absorbing Filters 25. Harraway, J, “Introductory Statistical Methods and the Analy-
and Fluorescent Whites, Insight on Color, 6(4), 1–2 (1995). sis of Variance,” University of Otago, Dunedin, 1993.
19. International Organization for Standardization, “BS EN ISO 26. SPSS Inc, “Statistical Package for Social Sciences, Version
139: 2005 Textiles. Standard Atmospheres for Conditioning 11”, SPSS Inc, Chicago, 2001.
and Testing,” Geneva, International Organization for Stand- 27. Gorensek, M. and Sluga, F., Modifying the UV Blocking
ardization, 2005. Effect of Polyester Fabric, Text. Res. J., 74(6), 469–474 (2004).
20. British Standards Institution, “BS 5441:1988. British Standard 28. Crews, P., Beyer, A. and Kachman, S., Influences on UVR
Method of Test for Knitted Fabrics,” London, British Stand- Transmission of Undyed Woven Fabrics, Text. Chem. Color.,
ards Institution, 1988. 31(6), 17–26 (1999).
21. International Organization for Standardization, “ISO 5084:1996. 29. Wang, S. Q., Kopf, A. W., Marx, J., Bogdan, A., Polsky, D. and
Textiles –Determination of the Thickness of Textiles and Tex- Bart, R. S., Reduction of Ultraviolet Transmission through
tile Products,” Geneva, International Organization for Stand- Cotton T-Shirt Fabrics with Low Ultraviolet Protection by
ardization, 1996. Various Laundering Methods and Dyeing: Clinical Implica-
tions, J. Am. Acad. Dermatol., 44(5), 767–774 (2001).

Downloaded from http://trj.sagepub.com at UNIV OF OTAGO on April 5, 2008


© 2008 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

View publication stats

You might also like