Skin comfort of base layer knitted garments. Part 1: Description and evaluation of wearer test protocol

Abstract

An important role of garments is to provide adequate comfort. A study was undertaken of the sensory scores for perceived comfort of wool base layer long sleeve knitted T shirts. This paper, the first in a series, describes and evaluates the wearer trial protocol in which untrained female wearers scored tactile, thermal, and moisture-based sensations during a controlled series of activities in a range of controlled climatic environments. Wearer scores were sufficiently consistent, that significant differences in aggregate scores between garments were detected that reflected changes in the fiber type (wool, cashmere, and cotton) and fiber specifications. Prickle and discomfort scores responded to different factors. The importance of choosing appropriate test conditions when assessing garments for particular end uses was highlighted as both the environment and activity affected wearer's perception of garment performance. A novel test feature was the use of a ‘link’ garment common to separate trials. This, combined with the observed absence of an effect due to garment washing, enabled the testing to be expanded so that 38 garments were successfully compared over 30 months in nine trials. Finally while the first trial used 43 wearers to obtain good estimates of absolute comfort levels, it was demonstrated that a reduction to 25 wearers was adequate for later trials with minimal loss in sensitivity.

Keywords

next to skin discomfort comfort prickle evaluation test protocol garment wool cotton cashmere wearer sensation environment fiber type fiber diameter

Comfort is a fundamental and universal need for consumers and a significant role of garments is to provide an adequate level of comfort.¹ It is thus a subject of interest both to fabric manufacturers and garment producers as well as consumers. Comfort is both sensory and complex and is indeed very difficult to define accurately.² Kilinc-Balci notes that the perception of clothing involves a very complex interaction of the wearer with clothing in the environment,³ in which numerous physical stimuli and sensory perceptions are integrated and evaluated against past experiences. Thus clearly a comfort value cannot be assigned to a garment per se, i.e. independent of the individual or the environment.

As reported by Li,¹ work by Fuzek and Hatch identified that important aspects of comfort can be grouped into four major areas,^4,5 namely, (a) thermo-physiological components (‘attainment of a comfortable thermal and wetness state’), (b) sensorial comfort (sensations arising when a garment touches the skin), (c) ease of body movement, and (d) aesthetic appeal (‘perception of clothing by the eye, hand, ear, and nose, which contributes to the overall well-being of the wearer’). The breadth of this is perhaps illustrated by the wide range of topics included in the recent book on comfort edited by Song.⁶ It is apparent from this categorization that in a clothing ensemble the base layer may significantly affect comfort as it is this layer which will largely influence the sensorial comfort as well as contributing to other aspects, e.g. thermo-physiological comfort.

An important aspect of the assessment of the comfort of garments is the development of a robust protocol to collect and analyze data from a wearer. This allows an investigation of the links between fiber properties, fabric properties, garment construction attributes, environment, activity, and the comfort sensations experienced by the wearer. Hollies and coworkers were pioneers in this area and identified a range of descriptors that could be usefully used to define key aspects of comfort.⁷⁻⁹ Yoo and Barker adopted this approach to study the comfort properties of heat-resistant protective wear and identified links with objectively measured fabric properties especially surface roughness.¹⁰ Jun et al. studied the sensory thermal comfort of caps during wear by asking subjects to score thermal, moisture, and thermal discomfort sensations on a psychophysical scale of −5 to +5.^11,12 In an examination of textile, physiological, and sensorial parameters in sock comfort, Bertraux et al. obtained ratings of pain, temperature, and dampness during exercise for different parts of the foot.¹³ More recently, Jeon et al. followed a similar approach and linked perceived moisture in shirts during a controlled wearer trial protocol with objectively measured moisture absorption behaviors.¹⁴

A trend observed in recent publications is computer modelling of garment comfort as a function of a range of fabric properties and specifications to predict comfort levels. This incorporates a range of complex analytical tools, including multiple linear regression,¹⁵ principal component analysis,^11,12 and neural networking and artificial intelligence.¹⁶ Prediction of garment responses from fabric properties is of limited value without quantitative measurement of responses to the garments made from these fabrics by wearers. In addition, Wong et al. and others have contended that comfort differs from person to person,¹⁷ and hence any protocol to assess comfort must acknowledge this and evaluate aggregative responses across a large sample of wearers.

Prickle, a specific and sometimes irritating component of the tactile aspect of comfort, has been the focus of a number of studies.¹⁸⁻³³ For garments made from animal fibers, like wool, it has been found that prickle can be an important consumer consideration.¹ Garnsworthy et al. identified the physical stimulus for prickle evoked by fabric to be a mechanical interaction between fiber ends protruding from the surface of the fabric and the skin.²⁰ Westerman et al. showed that the neural basis for the prickle sensation increases with increasing skin temperature, increasing skin moisture content, and exercise induced sweating.¹⁸ Many studies have demonstrated that fiber diameter is a key determinant of prickle sensory scores, and that for research purposes a significant range in prickle responses can be created by varying fiber diameter in knitted wool fabrics.²⁸ For pragmatic/practical reasons, many of these studies have relied on the assessment of prickle using the forearm prickle test as a proxy to simulate the wear experience.^23,26 In a limited trial, Veitch et al. were able to demonstrate a correlation between estimates of prickle of wool garments from the forearm prickle test and wearer trial data.²¹

Despite this significant body of existing research, many questions still remain regarding the influence and interaction of a number of fiber, yarn, and fabric properties on garment wearer comfort. A significant study was undertaken over a period of approximately three years to add to the existing body of knowledge in this area with a strong focus on measured sensory comfort scores of wool base layer garments. This paper, the first in a series covering this body of research, describes and evaluates a wearer trial protocol designed and developed to evaluate the wearer comfort properties of garments constructed of lightweight single jersey fabrics for untrained wearers chosen to be representative of a segment of the market of interest. The results from two trials are analyzed to demonstrate the robustness and sensitivity of the protocol over a range of environmental conditions and a range of garments made from wool, cotton, cashmere, and wool/cashmere blend with characteristics, especially a range of fiber diameter, aimed at evoking a significant range of wearer responses.

Some preliminary descriptions and work using results from this protocol have been published by Stanton,³⁴ Tester,³⁵ and McGregor et al.³⁶

Materials and methods

Nine wearer trials were conducted between April 2008 and July 2011 in Perth, Western Australia, with this protocol. Considerable emphasis was placed on generating a garment evaluation protocol that would enable a reproducible set of responses to the test garments. This protocol incorporated environmental protocols used by Hollies and Barker in which thermal and moisture challenges were introduced to heighten the discomfort sensations of the wearers and thereby improve the discrimination between test garments.^7–9,37 It also incorporated aspects of other methods to ensure representative and responsive wearers were used (e.g. Wang et al.).³⁸

Selection of test wearers

Wearers were drawn from the local urban community. They were untrained, unskilled wearers selected by an independent market research company using a carefully constructed screening process. The wearer group of interest was females, in the 25–35 age group with a family income greater than AUD$35,000 (before tax) who were fit enough to undertake the trial and who had no aversion to wearing wool garments. Based on these criteria, a screening questionnaire (see Appendix 1) was used by the market research company for selection. In accordance with the relevant sections of the Australian National Statement on Ethical Conduct in Human Research 2007, a basic medical questionnaire was used and wearers were restricted to those having a body mass index (BMI) between 20 and 30. Pregnant wearers and those without English as their first language were excluded.

Forty-three wearers were selected for the first trial. The number of wearers in subsequent trials was approximately 25. Over the extended time period of the nine trials, some of the wearers were either not able to participate in some trials or to continue to participate at all due to circumstances external to the trial constraints. This natural progression ensured a large sample of wearers was used for testing and that the trials did not suffer from wearer fatigue. Seventy-seven wearers participated in between one and nine trials (Figure 1).

Figure 1.

Number of trials in which wearers participated.

Garment selection

The protocol focused on understanding the comfort characteristics of base layer knitted long sleeve T shirts. Over various trials, garments were made of fabrics from a range of fiber types (wool, cotton, and wool/cashmere blends) and a range of other yarn and fabric parameters. Fabrics were made into a standard garment design with a round crew neck.

Prior to use in the experiments, each garment was washed three times, after Hollies et al.,⁸ on a gentle wool cycle in a domestic washing machine and flat dried. Garments were washed using the same method after each test session (one wearer on one day) so that they were ready for re-use in subsequent test sessions in the same trial as required. It was assumed that washing subsequent to the third wash would have no effect on garment performance. This assumption is tested and reported in the results section. Choice of garment size for each wearer was undertaken by a technician to achieve a consistent sizing protocol for all wearers in the trials. The technician focused on a level of fit within the normal range, achieving adequate contact between the garment and the body without constricting the movement of the wearer. A limited number of garments were manufactured in each of three different sizes (small, medium and large) per garment type (typically 10 garments at each size). Thus after laundering, most garments were used by multiple wearers.

Test facilities

Two different environmental conditions were utilized during the wearer trials. The ‘cool’ environment was maintained at 23℃ ± 0.5℃ and 45% ± 5% relative humidity (RH) to simulate a normal air conditioned office environment and the ‘hot’ environment was maintained at 40℃ ± 0.1℃ and 24% ± 0.5% RH. The ‘hot’ environment was achieved in a purpose built environmental chamber (4.5 m × 3.8 m) with conditioned air entering through a 10 m² curtain wall at 1 m/s. The temperature and RH were set to achieve a temperature change but not a (absolute) water vapor pressure change between the change room and the environmental chamber.

Sensations and rating scales

Following the original approach developed and reported by Hollies et al.⁸ and reviewed by Li,¹ the sensation of comfort and a suitable subset of sensations were selected to cover tactile, thermal, and moisture attributes of interest.

A total of 11 sensations were rated at defined periods during each test session, as listed in Table 1. The definitions were explained at the familiarization session and wearers were instructed to use these definitions, not any preconceived definition. When responding to the questions in each period, the wearer had the definitions in front of them for reference if needed. The questions were randomized for each test, ie for each garment and each wearer.

Table 1.

Sensations rated and the definition supplied to the wearer in each test session

Sensation	Definition
Absorbent	The garment is able to soak up sweat from your body like absorbent paper.
Clingy	The fabric in the garment is apt to cling or adhere to the body.
Cold	The garment feels cold to the body – feeling no warmth.
Damp	The garment feels wet or moist.
Heavy	The garment makes you feel weighed down and burdened.
Itchy	The fabric evokes a desire to scratch.
Muggy	The air under the garment feels hot and humid.
Prickly	The fabric evokes a sensation like many very gentle pin pricks.
Scratchy	The fabric feels harsh or abrasive (like sandpaper) on the skin.
Sweaty	Your skin under the garment feels wet with sweat.
Uncomfortable	The garment evokes a sensation of discomfort.

In rating the sensations in Table 1, the wearers were asked to score the sensations according to a scale from 1 to 9 using the descriptions found in Table 2. In the second and subsequent trials, when each sensation rated greater than 1, wearers were asked if the level of the sensation was acceptable or not.

Table 2.

Common scale used to describe each of the sensations in each period

Score	Description
1	Sensation not detected
2	Sensation is just detected/threshold level
3	Slightly present sensation
4
5	Moderately felt sensation
6
7	Very aware of the sensation
8
9	Extreme aware of the sensation

Test protocol

Care was given to ensure that the trials were ‘blind’. For example, (a) garments were labelled with an experimental code only, (b) care was taken to ensure that written material (e.g. company logos and promotional material) potentially available to the wearers had no association with a fiber type, (c) the type of fiber or specification was never mentioned or visible on garment labels or test rooms, and (d) the specific purpose of the trial and the fiber type or quality associated with a particular garment was not made known to the wearers. There was no communication between wearers during testing.

The test protocol consisted of five stages, with each stage (apart from stage 0) segmented into a number of periods.

Stage 0: Pre-test (acclimatization stage, lasting 30 minutes). A stage of thermal acclimatization in a change room type environment at 23℃, sitting in a cotton gown with shirt removed. Temperature and humidity sensors were placed on the hand, shin, neck, and scapula to allow calculation of mean skin temperature and local relative humidity. (From ISO 9886: 1992 (E) Table B1, mean skin temperature = 0.28 × (neck skin T) + 0.28 × (scapula skin T) + 0.16 × (shin skin T) + 0.28 × (hand skin T).) Sensors were also positioned on the neck and scapular to monitor the inside surface of the garment for temperature and RH during the tests. Sensors used to monitor the wearers' physiological condition during the tests (heart rate and blood pressure) were attached. These were measured at the start and at intervals during the test.

Stage 1: Periods 1–4 (cool passive stage, lasting 15 minutes). The cotton gown was removed and the test garment was put on. The wearer was immediately asked to rate all eleven sensations (Table 1) caused by the garment, and the acceptability of the level of the sensation (Period 1). The sensations were rated every 5 minutes to allow the changes in the sensations to be tracked; after 5 minutes standing (Period 2), then after 5 minutes of ‘light motion exercises’ designed to get the garment moving over the skin surface (Period 3), then after 5 minutes sitting in a chair (Period 4).

Stage 2: Periods 5–8 (hot passive stage, lasting 15 minutes). The wearer and the trial supervisor moved to the climate chamber (40℃ and 24% RH). The sensations were immediately rated (Period 5) and again every 5 minutes. As in stage 1, rating occurred after 5 minutes standing (Period 6), then after 5 minutes of ‘light motion exercises’ (Period 7) and finally after 5 minutes sitting (Period 8).

Stage 3: Periods 9–11 (hot active stage, lasting 15 minutes). The wearer stepped onto a treadmill in the climate chamber and was asked to walk at a leisurely pace (5 km/h) for 5 minutes before rating the sensations (Period 9). The treadmill was then inclined 5° and the wearer continued to walk for 5 minutes before again rating the sensations (Period 10). Finally the treadmill was returned to level and after 5 minutes walking the wearer again rated the sensations (Period 11).

Stage 4: Periods 12–15: (cool after hot stage, lasting 15 minutes). The wearer and the trial supervisor then returned to the change room. As in stages 1 and 2, the sensations were rated immediately on arrival in the change room (Period 12) and then every 5 minutes. The wearer rated the garment after standing for 5 minutes (Period 13), then after 5 minutes of ‘light motion exercises’ (Period 14) and then after 5 minutes sitting in a chair (Period 15). The closing question set included the question ‘Did you like the garment?’

Trial design

A maximum of six garments was used in each trial both to minimize the chance of wearer fatigue and to enable each trial to be conducted within a reasonable period of time of two to three months. Within each trial, all wearers tested each garment once. A restricted randomization, similar to a Latin square, was used to specify the order of garment testing for each wearer. The design ensured that, within each trial, the number of times each garment was preceded by every other garment was equal (or approximately equal depending on the number of wearers and garments).

On any particular day a wearer only evaluated one garment in what is referred to as a test session. Wearers returned for testing a further garment when convenient which was on average within 7 days.

Prior to their inclusion in a trial, new wearers completed a full test session which was considered a familiarization test, the results of which were not included in subsequent analyses. Before the first trial all wearers were required to complete a familiarization test. Garments for this test were allocated to ensure that, when looking at the first experimental test session for all wearers, each garment type was preceded by all other garment types the same number of times. Before the start of subsequent trials, when there were only a few ‘new’ wearers who had to complete a familiarization test, garments from the trial were allocated at random for the familiarization test.

Each trial included a link garment which was used in a number of other trials. Using link garment results, and assuming that garment differences would not differ between trials, it was possible to calculate comparable results for garments tested in different trials. This feature was incorporated to facilitate efficient and effective extension of the scope of the experimental design beyond six garments.

The experimental program was run over 3 years and included a linked series of nine trials (S1 to S9) in which 38 garments were compared.

Case study

To illustrate typical results and demonstrate the power and use of the protocol, results from two trials will be presented (S1 and S5). These two trials were separated by 18 months and three other trials. A total of four garments were tested in S1, and five garments in S5. All garments in these trials were of the same fabric construction (single jersey) and similar fabric weight. The fabrics varied in fiber type and fiber specification (Table 3).

Table 3.

Characteristics of single jersey fabrics used in trials S1 and S5

Wearer trial	S1 & S5	S1 & S5	S1 & S5	S1	S5	S5
Fiber type	Wool	Wool	Wool	Cotton	Wool & cashmere	Cashmere
Mean fiber diameter¹, µm	16.3	18.1	20.3	12.8	15.8	15.5
Yarn count, tex	R25/1	R25/1	R25/1	R25/1	R33/2	R33/2
Fabric thickness, mm	0.7	0.7	0.7	0.7	0.7	0.7
Fabric mass per unit area, g/m²	184	173	176	165	169	166
Garment weight, g	135	135	131	159	145	142
Number of tests: Trial S1	43	47	42	40
Number of tests: Trial S5	23	23	23		23	23

¹Fibers extracted from yarns were measured using the OFDA machine.³⁹

For the S1 trial, 43 wearers each tested four garment types, one made from cotton fabric and three made from pure wool fabrics covering a range of mean fiber diameters. Cotton garments were not available in early testing within this trial and so were replaced by one of the other wool garments, resulting in less testing of cotton garments than the other garments (Table 3). In addition, all familiarization tests were carried out using wool garments.

In the S5 trial, 23 wearers each tested five garment types of which three were the same wool garments as those used in trial S1. The remaining garments were made from cashmere and a cashmere/wool blend (30%/70%). In this trial, thirteen of the 23 wearers had not taken part in the first trial.

Statistical analysis

Method for comparing wearer sensation scores

Brockhoff has discussed the different attributes of assessors in sensory trials and the need to incorporate these differences in statistical models when analyzing sensory data.^40,41 Each assessor can be expected to have an individual level of assessment, an individual variance between treatments and an individual variance between replicates of the same treatment. These parameters should be included as random terms in a linear mixed model. Treatment effects should be included as fixed terms. When analyzing univariate sensory scores from the garment trials, wearers (assessors) were expected to vary as Brockhoff and Skovgaard described.⁴¹ In addition, it was proposed that variance between scores would change with garment and period and that measurements made in different periods of the same test session (trial by wearer by garment combination) would be correlated with one another. A linear mixed model was fitted to the scores which consisted of a random model including these variances and covariance structures and a fixed model which included garment effects, period effects, trial effects, garment by period interaction effects and trial by period interaction effects (see Appendix 2). Interactions between trials and garments were not present in the fixed model because there was only one garment in common for most pairs of trials and the interaction therefore could not be estimated. In general variances for individual wearers, garments and periods increased as their scores increased, but a logarithmic transformation of the data did not remove the need for individual variances. No transformation of the data was performed. The effect of period was subdivided on the basis of environment and activity (Table 4) and the effect of garment was subdivided according to the factors involved in their selection. All models were fitted using the REML procedure in GenStat.⁴²

Table 4.

Summary of the different stages in the protocol and the factors used to define and analyze the recorded sensory data

Period	Environment	Activity 1	Activity 2
1	Cool	Standing	Not walking
2	Cool	Sitting	Not walking
3	Cool	Range of motion (ROM)¹	Not walking
4	Cool	Sitting after ROM	Not walking
5	Hot	Standing	Not walking
6	Hot	Sitting	Not walking
7	Hot	Range of motion	Not walking
8	Hot	Sitting after ROM	Not walking
9	Hot with walking	Walking (5 kph)	Level
10	Hot with walking	Walking (5 kph)	Elevated (5%)
11	Hot with walking	Walking (5 kph)	Level after elevated
12	Cool after hot	Standing	Not walking
13	Cool after hot	Sitting	Not walking
14	Cool after hot	Range of motion	Not walking
15	Cool after hot	Sitting after ROM	Not walking

¹ROM – Range of motion was a set of gentle stretching exercises undertaken while standing which were designed to generate some movement of the garment over the skin surface with minimal change of physiological state.

For the case study, in which three of the same garments were tested in two trials, the fixed model, was expanded to include a trial by garment interaction in order to test the assumption underlying the use of link garments, that garment differences would not differ between trials. The effect of garment was subdivided into the effects of fiber type (cotton vs 100% wool vs cashmere and cashmere mixtures) and fiber diameter effects within fiber type. The model was used to estimate garment and garment by period means that were comparable because they had been adjusted for trial and wearer effects. Standard errors and 5% least significant differences (LSDs) were calculated to compare garments. Only the sensations of ‘prickly’ and ‘uncomfortable’ (Table 1), hereafter referred to as prickle and discomfort, have been analyzed in this paper.

Method for testing effect of number of wearers

After the first trial (S1), a study was carried out to assess whether the number of wearers could be reduced from 43 and which wearers should be retained. It was decided that the best criteria for choosing wearers for future trials was their ability to discriminate between garments. There are probably many people in the population who cannot discriminate between garments with respect to prickle. While these people will be an important part of consumer testing, where estimating the proportion of the population who like a garment is the goal, they are not useful in terms of identifying differences between garments, when limited resources are available. For each wearer a linear mixed model was fitted to the prickle scores in order to estimate variance components due to garments, periods and the garment by period interaction. Scores from the familiarization test were excluded. Wearers were ranked from most discriminating to least discriminating using their variance component due to garments which represents their ability to discriminate between garments. Some adjustment to the rankings was made for wearers who did not test the cotton garment. Average prickle scores for each garment and average 5% LSDs for comparing garment averages were recalculated using varying numbers of the most discriminating wearers. In addition, average prickle scores and an average 5% LSD were calculated for 25 randomly selected wearers. This random selection process was repeated 100 times and averages were calculated for garment means and the 5% LSD.

Method for comparing LIKE/NOT LIKE responses

A generalized linear mixed model for binomial data was fitted to the binary response (Like/Dislike) that each wearer gave to a garment at the end of each test session. The model included fixed effects for garments and trials and random effects for wearers. The wearer effects were assumed to have a normal distribution.

Method for testing washing effect

A further analysis was carried out across six trials to confirm that repeated washing of garments had not contributed to the performance of link garments in later trials. Results from a single garment type which was assessed by 93 wearers over six trials (i.e. up to 8 additional washing cycles) were used for this analysis. Familiarization tests were excluded. A linear mixed model with the same random terms as described above was fitted to the data. The fixed model included the effect of period, linear and quadratic effects for the number of washes (4–12) and interactions.

Results

Changes in skin temperature and RH%

An example of the temperature and relative humidity changes experienced by the wearers during a test session is shown in Figure 2. After acclimatization in the cool room for 30 minutes, the gown was removed and replaced with the test garment and the experimental testing started with Period 1. The mean skin temperature underwent small changes while the wearer was in the cool room. The temperature then increased over the next 30 minutes while the wearer was in the hot room (Periods 5 to 8) and exercising (Periods 9 to 11) during which time the relative humidity rose to saturation at 100%, and sweat accumulated between the sensor and the skin. Then the last 15 minutes (Periods 12 to 15) showed a decrease in skin temperature associated with standing, then ‘light motion exercises’. The changes in RH largely mimicked the observed changes in skin temperature during the early periods of the test session. However during periods 12 to 15, the RH remained near saturation consistent with the continued presence of sweat and by the reduction in the skin temperature in the cool room.

Figure 2.

Mean skin temperature and mean RH% of the skin of a wearer for each period in a single test. The light gray background covers the time spent in the ‘change room’ environment. The dark gray background covers the time spent in the hot environment.

Protocol and garment effects

Analyses were undertaken to see if the protocol could generate significant changes in sensory scores from a sample of untrained, inexperienced wearers. There were highly significant main effects of environment, activity and fiber diameter (within fiber type) on prickle and discomfort scores (P < 0.001). There was also a main effect of fiber diameter on the percentage of wearers who liked each garment (P < 0.001) and a significant main effect of fiber type for prickle scores (P < 0.001) but not for discomfort scores (P = 0.264).

The environment and activity levels of the test protocol also interacted with the garment effects. There was a significant interaction between fiber diameter, within fiber types, and environment for prickle and discomfort (P < 0.001 and P = 0.009, respectively). This interaction can be seen clearly in Figure 3 as the garment differences varied in different environments. Interactions between Activity 1 and fiber type for both prickle and discomfort (P = 0.029 and P = 0.018, respectively) are also illustrated where the ROM activity can be seen to have more effect on some wool garments. The elevated scores of the 20.3 µm wool garment from the effect of walking in high temperatures is shown as a significant Activity 2 by fiber type interaction for discomfort scores (P = 0.011) but not for prickle scores (P = 0.082).

Figure 3.

Weighted average scores for discomfort and prickle by period for each garment.

Wearer scores

While high scores were less commonly recorded than low scores for both sensations, wearers did use the full range of scores from 1 to 9. Of approximately 18,700 scores recorded for prickle and discomfort over the nine trials, scores of 9 occurred on 44 (0.2%) and 30 (0.2%) occasions for prickle and discomfort, respectively (Table 5). Fourteen percent of wearers used a score of 9 for prickle (13% for discomfort), and 48% of wearers used a score of 7 or more for prickle (53% for discomfort) at least once.

Table 5.

Percentage scores for each prickle and discomfort level

Score	Prickle	Discomfort
1	39.6	40.2
2	26.7	26.7
3	17.6	17.2
4	7.2	6.9
5	4.7	4.7
6	2.0	2.1
7	1.5	1.6
8	0.4	0.4
9	0.2	0.2

Two terms were introduced into the model for the case study to test the assumption, underlying the use of link garments, that garment differences would not differ between two trials which were carried out almost 2 years apart. These were a trial by garment interaction and a trial by garment by period interaction. The trial by garment interaction compared the overall differences between the garments in each trial and the garment by period term compared the difference between garments in each period between trials.

The analysis showed there was no interaction between trial and garment and no interaction between trial, garment and period for prickle (P > 0.361 and P > 0.584, respectively) or for comfort (P > 0.243 and P > 0.512, respectively) indicating that the difference between garments was the same in both trials and for all periods.

Average discomfort scores for each garment type over all periods indicate that the average discomfort score for the 20.3 µm wool garment (2.62) was significantly higher than that for the wool/cashmere garment (1.64), the 100% cashmere garment (1.71), the 16.3 µm wool garment (1.75), the cotton garment (1.92), and the 18.1 µm wool garment (1.94) (P < 0.05; Table 6).

Table 6.

Average discomfort and prickle scores for garments made from different fiber types and fiber specification¹, proportion of wearers who liked each garment (%Like), and standard errors (s.e.)

Garments made of different fiber types and fiber specifications	Average discomfort score	s.e.	Average prickle score	s.e.	%Like	s.e.²
30% cashmere 70% wool, 15.8 µm	1.64	0.13	1.73	0.14	78	10
100% cashmere, 15.5 µm	1.71	0.14	1.56	0.16	67	12
100% wool, 16.3 µm	1.75	0.09	1.77	0.10	79	5
100% wool, 18.1 µm	1.94	0.10	2.02	0.10	78	5
100% cotton	1.92	0.12	1.30	0.12	74	8
100% wool, 20.3 µm	2.62	0.16	2.68	0.14	44	7

¹See Table 3 for specifications.

²Since the s.e. for this parameter is estimated on the log scale, an s.e. equivalent, which is half the 68% confidence interval, is presented.

Average prickle scores for each garment over all periods indicate that the cotton garment was significantly less prickly than the wool/cashmere garment, the 16.3 µm wool garment and the 18.1 µm wool garment (P < 0.05) which were all significantly less prickly than the 20.3 µm wool garment (P < 0.05). There was no difference in the average prickle score between the 100% cashmere garment and the 16.3 µm wool garment (P > 0.05).

Significantly more participants liked the wool/cashmere garment (78%), the 16.3 µm garment (79%), the 18.1 µm garment (78%), and the cotton garment (74%) than the 20.3 µm garment (44%) (P < 0.05). There was no difference between the 100% cashmere garment (67%) and the other garments (P > 0.05).

The average prickle and discomfort scores at each period in the test protocol showed that the 20.3 µm wool garment had significantly higher average discomfort and prickle scores in all periods than other garments (P < 0.05), with prickle scores for this garment showing effects of activity and environment (periods 2 vs 3, 3 vs 4, 7 vs 8, 8 vs 9, 14 vs 15 significant; P < 0.05) (Figure 3). Other wool and cashmere garments showed smaller effects of environment and some effects of activity on prickle and discomfort scores. The cotton garment had significantly lower average prickle scores than other garments in all periods apart from the 100% cashmere garment when walking in a hot environment (P < 0.05). It showed no effects of activity or environment (P > 0.05). However, discomfort scores for the cotton garment were not significantly different to most other garments (excluding the 20.3 µm wool garment) in all periods (P > 0.05). It had significantly higher discomfort scores than the wool/cashmere garment in periods 3, 5, 7, and 8, significantly higher discomfort scores than the 100% cashmere garment in periods 5, 7, and 9, and significantly higher discomfort scores than the 16.3 µm garment in period 7 (P < 0.05).

Effect of washing on garment score

There were no effects of washing (linear and quadratic) on prickle (P = 0.214 and P = 0.970, respectively) or discomfort (P = 0.260 and P = 0.148, respectively) scores.

Determination of the required numbers of wearers to detect differences between fabrics

Using the data from Trial S1, the results in Table 7 show the effect on the garment means and 5% LSD values of successively removing the scores associated with the least discriminating people. The absolute values of the scores and the 5% LSD values progressively increased as the number of wearers decreased. However when the 5% LSD values were expressed as a percentage of the difference between the cotton and the coarsest wool garment, the percentage fell initially as the number of wearers was reduced and then rose, passing through a minimum value with approximately 30 wearers. The average 5% LSD based on the 25 most discriminating wearers was 0.25 which was 11% of the difference in mean prickle score between the cotton garment and the 20.3 µm garment. A group of 25 wearers randomly selected from the full set of wearers had an average 5% LSD of 0.31, which was 17% of the difference in mean prickle score between the cotton garment and the 20.3 µm garment. Based on this information it was decided to reduce the number of wearers to 25 in trials S2 to S9, starting with the most discriminating wearers in trial S2. This decision took into account that as discriminating wearers were replaced by randomly selected wearers in S3 to S9 the precision of the trials would be expected to decrease as indicated by the randomly selected group in Table 7.

Table 7.

Garment prickle means and 5% LSDs between garment means as the number of most discriminating wearers increases from 20 to the whole group of 43 wearers. The 25R group (bold) represents average values when 25 wearers are randomly resampled 100 times

Number of wearers	Garment means				Average 5% LSD	5% LSD, %¹
Number of wearers	Cotton	Wool 16.3 µm	Wool 18.1 µm	Wool 20.3 µm	Average 5% LSD	5% LSD, %¹
20	1.4	2.3	2.5	4.0	0.29	11
25	1.4	2.2	2.4	3.7	0.25	11
25R	1.3	1.9	2.1	3.0	0.31	17
30	1.3	2.1	2.3	3.5	0.22	10
35	1.4	2.1	2.2	3.3	0.23	12
40	1.3	2.0	2.1	3.1	0.28	16
43	1.3	1.9	2.1	3.0	0.27	16

¹5% LSD as a percentage of the difference in prickle means of the cotton and coarse wool garments.

Questions	Potential answer	Decision
Q1. Which of these following age groups do you fit into? Q2. How often would you say you shop for clothing for yourself? Q3. Which of these categories does your combined household income fall into, before tax? Q4. Which of the following fabric types will you avoid buying when shopping for clothing for yourself? Q5. How many times per week would you say you exercise? Q6. Do you believe that you are able to walk for 30 minutes without taking a break?	Under 25 years25 to 29 years30 to 35 yearsover 35 yearsWeeklyFortnightly/Every 2 weeksMonthlyLess frequently than once a monthDon't know (This option not read out)Less than $35,000$35,001 - $65,000Over $65,000Refused to answerLycraCottonWoolPolyesterSilkNylonAcrylicNeverLess than once a weekOnce a weekTwice a weekThree times a weekEvery second dayEvery dayYesNo	Reject and close interviewAccept provided no more than50% in each categoryReject and close interviewContinue to next questionContinue to next questionContinue to next questionReject and close interviewReject and close interviewReject and close interviewAccept provided no more than50% in each categoryReject and close interviewContinue to next questionContinue to next questionReject and close interviewContinue to next questionContinue to next questionContinue to next questionContinue to next questionContinue to next questionContinue to next questionAcceptAcceptAcceptAcceptAcceptAcceptReject and close interview

Discussion

The performance of the garment test protocol outlined in this paper was tested using single jersey, base-layer garments which differed only in the fiber type and fiber specifications using a group of unskilled and untrained women between the ages of 25 and 35 years. All garments were tested under the same environment and activity conditions so that garment effects could be compared under specific conditions or averaged over all conditions in the experimental protocol. Sensations of prickle and discomfort were recorded by wearers on a scale from 1 to 9 for all garments in all conditions.

The analyses showed that the sample of wearers was able to use the prickle and discomfort scores to describe their sensory responses to the garments they were wearing, their activity level and the environmental conditions. In general, when there was a difference in the sensory score between garments, the sensory profiles for each garment across periods did not cross over indicating that the ranking of the garments on that scale remained the same in all environments. However the size of the difference was generally largest under the most stressful conditions.

The analysis showed that there was no effect on the relative scores of garments when they were tested in another trial at a later date with a different group of wearers. This validated the use of a set of link garments that allowed the results from different trials to be combined. Furthermore, the ability to combine trial results and the absence of a washing effect extended the scope of the testing so that 38 garments could be compared over 30 months in one or more of nine trials. Five garments were used as link garments over the nine trials. This is possibly the first use of link garments in this type of consumer testing.

The significant interaction between garments and environment and activity in the case study presented indicates that environment and activity changes in the protocol are justified, especially when testing the effects of using different fiber types. However this interaction relates to the specific environmental conditions used in the test protocol which used high temperatures in an attempt to mimic sporting conditions and/or living in hotter climates. There is no assumption being made that the garment by environment interaction observed here would carry over into low temperature test conditions. However, the results confirm the importance of testing garments under a range of environment /activity conditions as garments can perform very differently under different conditions.

Results show that the protocol was capable of detecting change in garment scores that were a reflection of changes in the fiber type (wool, cashmere and cotton) and fiber specification (wool with different mean fiber diameter). In the case study average scores over all periods indicated that the cotton garment was significantly less prickly than all the wool garments and that the 16.3 µm and 18.1 µm wool garments were less prickly and more comfortable than the 20.3 µm garment. These garments did not differ in appearance and the wearers were given no information about the garments so the significant differences in sensory scores were based only on the sensory and physiological responses to the garment. The fact that differences in fiber type and specification were detectable in the sensory scores is indicative of the sensitivity of the protocol.

Analysis of the statistical model showed that the prickle scores and the discomfort scores were responding to different factors. For example, there was a main effect of fiber type for prickle scores (P < 0.001) but not for discomfort. It is clear from Figure 3 that, while the cotton garment showed little change in prickle scores between periods, it showed an increase in discomfort scores as temperature and activity increase compared to both prickle and discomfort scores for the wool garments. This result indicates that prickle and discomfort, as reported by the untrained and unskilled wearers, are different and that the relationship between comfort and prickle can change depending on the garment. As noted in the introduction there are many components which together make up comfort. It may be that the two fiber types differ in one or more of the other sensations tested (Table 1), e.g. clingy.

While the cotton garment was less prickly than all other garments it was not liked by more wearers than any other garment apart from the 20.3 µm garment. Wearers' dislike for a garment appeared to be associated with their perception of overall comfort as measured by the discomfort score. As in the previous paragraph this may indicate that one of the other comfort components (i.e. not prickle) is contributing to reducing the overall percentage of people who liked the garment. This is an interesting topic which will be explored in a later paper utilizing the data now available from the wider range of sensations recorded.

The effect of the number of wearers in the trial was examined. While larger trials would be expected to produce results with lower variance, the large trial is also associated with increased resource allocation and longer time frames. Using results from the first wearer trial the 25 most discriminating wearers were expected to detect a difference of approximately 0.25 in prickle score between garments. As wearers were replaced for various reasons by new untried wearers, this 5% LSD could be expected to increase to approximately 0.30. With due consideration to time, cost and the ability to find sufficient suitable wearers in the local urban community it was concluded that this level of precision was acceptable in the series of trials described. Hence, all trials after the first used 25 wearers.

Future work will look at the effect of a broad range of fiber types and fiber specifications on a range of the comfort components. Changes in the yarn structure and fabric structure and finishing will be investigated. The relationship between the scores for the individual sensations and the approval rating of the garment will also be analyzed.

Conclusions

A garment test protocol for assessing sensory responses was evaluated and shown to be useful to identify and characterize changes in discomfort and prickle sensory scores as a function of fiber characteristics, activity level, and environmental conditions.

A novel feature of the protocol was the use of link garments to combine the results of different trials over an extended time period. The efficacy of this approach was underpinned by the observations (a) that relative scores of garments did not change when tested in another trial at a later date with a different group of wearers and (b) a lack of an effect of subsequent garment washing on recorded scores.

In general, wearer scores were affected by the environment and activity highlighting the importance of choosing appropriate test conditions when assessing garments for particular end uses.

Funding

This work was supported by the Western Australian Department of Agriculture and Food and, in part, by CRC for Sheep Industry Innovation Ltd.

Footnotes

Acknowledgments

The authors thank Drs M Dolling and S Rankin for their input to the development of aspects of the project and the testing protocol. The authors also acknowledge the technical support provided by Sharon Pamment, Amanda Murphy, and Padmaja Ramankutty.

Appendix 1

Appendix 2

References

. The science of clothing comfort. Text Prog 2001; 31: 1–138.

Roy-Choudhury

Majumdar

Datta

Factors affecting comfort: human physiology and the role of clothing. In: Song

(ed). Improving comfort in clothing, Cambridge, UK: Woodhead Publishing Series in Textiles No. 106, 2011, pp. 3–60.

Kilinc-Balci

Testing, analyzing and predicting the comfort properties of textiles. In: Song

(ed). Improving comfort in clothing, Cambridge, UK: Woodhead Publishing, 2011, pp. 138–158. Woodhead Publishing Series in Textiles No. 106.

Fuzek

. Some factors affecting the comfort assessment of knit T-shirts. Ind Eng Chem Prod Res Dev 1981; 20: 254–259.

Hatch

. Textile science, New York: West Publishing Company, 1993.

Song

. Improving comfort in clothing 2011. Woodhead Publishing Series in Textiles No 106Cambridge, UK: Woodhead Publishing.

Hollies

NRS

Psychological scaling in comfort assessment. In: Hollies

NRS

Goldman

(eds). Clothing comfort, Ann Arbor, MI: Ann Arbor Science Publishers Inc, 1977, pp. 107–120.

Hollies

NRS

Custer

Morin

. A human perception analysis approach to clothing comfort. Text Res J 1979; 49: 557–564.

Hollies

NRS

. Improved comfort polyester. Part IV: Analysis of the four wearer trials. Text Res J 1984; 54: 544–548.

10.

Yoo

Barker

. Comfort properties of heat resistant protective workwear in varying conditions of physical activity and environment. Part II: Perceived comfort response to garments and its relationship to fabric properties. Text Res J 2005; 75: 531–541.

11.

Jun

Park

Shim

. Thermal comfort properties of wearing caps from various textiles. Text Res J 2009; 79: 179–189.

12.

Jun

Park

Kang

. Effect of heat and moisture transfer properties on microclimate and subjective thermal comfort of caps. Text Res J 2010; 80: 2195–2203.

13.

Bertraux

Derler

Rossi

. Textile, physiological, and sensorial parameters in sock comfort. Text Res J 2010; 80: 1803–1810.

14.

Jeon

Yoo

Kim

. Psychophysical determination of moisture perception in high-performance shirt fabrics in relation to sweating level. Ergonomics 2011; 54: 576–586.

15.

Sztandera

Cardeloo

Winterhalter

. Identification of the most significant comfort factors for textiles from processing mechanical, handfeel, fabrics construction, and perceived tactile comfort data. Text Res J 2013; 83: 34–43.

16.

Singleton

Yao

. Prediction of clothing sensory comfort from fabric properties using hybrid neural-fuzzy model. In: Textile bioengineering and informatics symposium proceedings 2008, Hong Kong on 14–16 August 2008, 1051–1056.

17.

Wong

ASW

Yeung

PKW

. Neural network predictions of human psychological perceptions of clothing sensory comfort. Text Res J 2003; 73: 31–37.

18.

Westerman

Garnsworthy

Walker

Aspects of human cutaneous small nerve function: sensations of prickle and itch. In: Chahl

Szolksanyi

Lembeck

(eds). Proceedings of the 29th IUPS satellite symposium, Newcastle, Australia. Budapest: Hungarian Academy of Science Press, 1985, pp. 329–345.

19.

Garnsworthy

Gully

Kandiah

. Understanding the causes of prickle and itch from skin contact of fabrics. Australas Text 1988; 8(4): 26–29.

20.

Garnsworthy

Gully

Kenins

. Identification of the physical stimulus and neural basis of fabric-evoked prickle. J Neurophysiol 1988; 59: 1083–1097.

21.

Veitch

Kenins

Naylor

GRS

. Comparison between laboratory protocol and simulated wear conditions for the evaluation of prickle. In: Proceedings of the 8th international wool textile research conference 1990; Vol, 5, Christchurch, 286–295.

22.

Veitch

Naylor

GRS

. The mechanisms of fiber buckling in relation to fabric-evoked prickle. Wool Technol Sheep Breed 1992; 40: 31–34.

23.

Naylor

GRS

. The role of coarse fibers on fabric prickle using blended acrylic fibers of different diameters. Wool Technol Sheep Breed 1992; 40: 14–18.

24.

Naylor

GRS

. The relationship between the fiber diameter distribution of wool top, fiber ends and yarn surface fiber. Wool Technol Sheep Breed 1992; 40: 40–43.

25.

Dolling

Marland

Naylor

GRS

. Knitted fabric made from 23.2 µm wool can be less prickly than fabric made from finer 21.5 µm wool. Wool Technol Sheep Breed 1992; 40: 69–71.

26.

Naylor

GRS

Veitch

Mayfield

. Fabric evoked prickle of some wool fabrics. Text Res J 1992; 62: 487–493.

27.

Naylor

GRS

Phillips

Veitch

. The relative importance of mean diameter and coefficient of variation of sale lots in determining the potential skin comfort of wool fabrics. Wool Technol Sheep Breed 1995; 43: 69–82.

28.

Naylor

GRS

Phillips

Veitch

. Fabric-evoked prickle in worsted spun single jersey fabrics. Part 1: The role of fiber end diameter characteristics. Text Res J 1997; 67: 288–295.

29.

Naylor

GRS

Phillips

. Fabric-evoked prickle in worsted spun single jersey fabrics. Part 2: The role of fiber length, yarn count and fabric cover factor. Text Res J 1997; 67: 354–359.

30.

Naylor

GRS

Phillips

. Fabric-evoked prickle in worsted spun single jersey fabrics. Part 3: Wear trial studies of absolute fabric acceptability. Text Res J 1997; 67: 413–416.

31.

Naylor

GRS

Stanton

. Time of shearing and the diameter characteristics of fiber ends in processed tops: an opportunity for improved skin comfort in garments. Wool Technol Sheep Breed 1997; 45: 243–255.

32.

Naylor

GRS

Oldham

Stanton

. Shearing time of Mediterranean wools and fabric skin comfort. Text Res J 2004; 74: 322–328.

33.

Naylor

GRS

. Fabric-evoked prickle in worsted spun single jersey fabrics. Part 4: Extension from wool to Optim™ fine fiber. Text Res J 2010; 80: 537–547.

34.

Stanton J. Wool and garment In: Song G (ed.) Improving comfort in clothing Ed G. Song. Woodhead Publishing Series in Textiles No. 106. Cambridge, UK: Woodhead Publishing Series in Textiles No. 106, 2011, pp.79–96.

35.

Tester

. Relationship between comfort meter values and the prickle rating of garments in wearer trials. Anim Prod Sci 2010; 50: 1077–1081.

36.

McGregor

Naebe

Stanton

. Relationship between wearer prickle response with fibre and garment properties and Wool ComfortMeter assessment. J Text Inst 2012; 104: 618–627.

37.

Barker

. From fabric hand to thermal comfort: the evolving role of objective measurements in explaining human comfort response to textiles. Int J Clothing Sci Technol 2002; 14: 181–200.

38.

Wang

Zhang

Postle

. Evaluating wool shirt comfort with wear trials and the forearm test. Text Res J 2003; 73: 113–119.

39.

IWTO Test Method IWTO-47-00:2000. Measurement of the mean and distribution of fibre diameter of wool using an optical fibre diameter analyser (OFDA).

40.

Brockhoff

. Statistical testing of individual differences in sensory profiling. Food Qual Preference 2003; 14: 387–395.

41.

Brockhoff

Skovgaard

. Modelling individual differences between assessors in sensory evaluation. Food Qual Preference 1994; 5: 215–224.

42.

Payne

Harding

Murray

. The guide to GenStat release 14. Part 2: Statistics, Hemel Hempstead, UK: VSN International, 2011.