Original article

Textile Research Journal

82(4) 400–414

Relationships of cotton fiber properties ! The Author(s) 2012

Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
to ring-spun yarn quality on selected DOI: 10.1177/0040517511426613
trj.sagepub.com
High Plains cottons

William Brock Faulkner1, Eric F Hequet2,

John Wanjura3 and Randal Boman4

Abstract
Cotton fiber properties play an important role in determining spinning performance but explain only a portion of the
variability in final yarn quality parameters. This research investigates relationships between ring-spun yarn quality and
fiber properties (measured using the High Volume Instrument (HVI) and Advanced Fiber Information System (AFIS))
given additional information on harvest method and cultivar. Seventy-six samples of commercially grown cotton repre-
senting five cultivars from six locations across the Texas High Plains were collected over three years. Carded 14.5 tex (40
Ne) ring-spun yarns were produced and tested for various yarn quality characteristics. Principal component analysis and
partial least squares regression were used to determine relationships between fiber and yarn properties. Neither harvest
method nor cultivar explained a significant portion of yarn quality variability beyond that captured by HVI and AFIS
results. Yarn work-to-break was highly correlated to fiber bundle elongation, which is not currently reported in official
cotton classing reports.

Keywords
Advanced Fiber Information System, cultivar, harvest, High Volume Instrument, partial least squares, principal component
analysis

Production and use of US cotton has changed dramat- improved to match that of cottons produced in
ically in recent years, with exports surpassing domestic California 10 years ago, and fiber strength has also
use annually since 2001 (Figure 11). markedly improved. However, thus far there has still
The shift to foreign markets has coincided with a been difficulty producing fine yarns of comparable
shift in the primary end use of US cotton from domes- quality to those produced from Acala cottons using
tic open-end mills to foreign ring-spinning mills cottons from West Texas.8
(Figure 2;2–4 Table 14,5).
As the end use of US cotton has changed, so has the
regional distribution of upland cotton production
(Figure 36). Changes in water availability and com- 1
Department of Biological and Agricultural Engineering, Texas A&M
modity prices have reduced production acres in University, College Station, USA.
2
California, which dropped from the second largest Fiber and Biopolymer Research Institute and Department of Plant and
Soil Science, Texas Tech University, Lubbock, Texas, USA.
producing state in 2000 (15% of the US crop) to fifth 3
USDA-ARS Cotton Production and Processing Research Unit, Lubbock,
in 2009 (5% of the US crop). Commensurately, recent Texas, USA.
improvements in irrigation technologies and cotton 4
Southwest Research and Extension Center, Oklahoma State University,
cultivars have increased yields on the High Plains of Altus, Oklahoma, USA.
Texas, increasing the portion of the US crop produced
in this region. Corresponding author:
William Brock Faulkner, Department of Biological and Agricultural
Improved yields on the High Plains have been Engineering, Texas A&M University, 2117 TAMU, College Station, TX
accompanied by improved fiber quality (Figure 47). 77843-2117, USA
Since 2000, the length of fibers produced in Texas has Email: faulkner@tamu.edu
Faulkner et al. 401

Figure 1. Domestic mill use and US cotton exports.1

Table 1. Regional share of yarn rotor and ring-spun4 and 2008/ from a specified lot of fiber under specified conditions)
2009 US cotton exports.5 of a cotton is dependent on fiber properties and spin-
ning method.9 Longer, stronger fibers are better able to
Share of yarn spun (%)
US cotton withstand the large forces placed on them during spin-
Region Rotor-spun Ring-spun exports (%) ning and have more contact between fibers, thus
increasing inter-fiber friction. These fibers are therefore
Africa 20 80 0.6
able to be spun into finer yarns.
North America 44 56 10.2 While fiber length, strength and fineness are most
South America 30 70 3.6 frequently correlated to yarn properties,10 the trash
Asia and Oceania 16 84 76.6 content of cotton can also affect the maximum achiev-
Mainland China 17 83 28.6 able yarn count. Due to the high angular speeds
Turkey 38 62 13.8 encountered by fibers during spinning, trash particles
Western Europe 37 63 0.5 can cause fiber breaks by exerting centrifugal force on
Eastern Europe 76 24 0.01 the forming yarn. Foreign matter and neps increase
Global 20 80 yarn unevenness and ends down (i.e. breaks) in spin-
ning, which decrease production efficiency and increase
imperfections in fabrics. For ring spinning, finer yarns
are particularly susceptible to end breaks due to the
presence of trash in the roving.
Fiber properties and spinning performance
The fiber properties and foreign matter content of
Fiber properties determine performance during pro- US cotton bales are assessed at Cotton Classing Offices
cessing and spinning. To produce high-quality ring- maintained by the United States Department of
spun textiles, fibers must be fine and have sufficient Agriculture (USDA) Agricultural Marketing Service
strength to endure processing (spinning preparation, (AMS). Each bale receives a grade that is available to
spinning, and weaving or knitting). Fiber length and merchants or buyers based on micronaire, fiber length
fineness affect the forces between fibers that dictate (reported as upper half mean length (UHML)), length
the ‘count,’ or fineness, of the final yarn. Fiber maturity uniformity, strength, color and quantity of leaf trash or
and strength affect a fiber’s ability to withstand the extraneous matter. Not included in the data are harvest
forces placed upon it during opening and blending, method and cultivar, both of which may affect spinning
carding, drafting and spinning. The spinning limit (i.e. performance in ways not captured by currently avail-
the finest yarn number that can be spun satisfactorily able classing grades.
402 Textile Research Journal 82(4)

Figure 2. US spinning positions.2,3 Note: rotor positions in the US are able to produce approximately nine times the output (by
mass) of ring positions.4 Therefore, even with fewer rotor positions, the US had the capacity to produce approximately three times
more rotor-spun yarn than ring-spun yarn in 2008.

Figure 3. Percentage of US upland crop produced by state in the top five producing states.6

tight-locked cultivars and short plant heights have

High Plains cotton production led to the use of stripper harvesters rather than the
Production practices in the High Plains are different mechanical picker harvesters used throughout most of
than those of other US regions. Harsh weather the US. However, new germplasm, increased yields and
conditions have historically led to the use of more improved irrigation practices have led to the use of
storm-proof cultivars, and the combination of these picker harvesters on some irrigated cotton on the
Faulkner et al. 403

Figure 4. Average strength (upper) and length (lower) of upland cottons produced in Texas and California from 2000 to 2008.7

High Plains. This moderate shift in harvest method and preliminary research, Krifa and Ethridge11 reported that
the substantial improvement in fiber quality have led to cottons produced on the High Plains with relatively high
questions regarding the impact of these changes on the micronaire (>4.1) performed similarly to California
value of cottons produced on the High Plains. Expecting Acala cottons with similar HVI properties in textile pro-
differences in mill performance beyond those that would duction. However, unlike most of California, the High
be predicted by High Volume Instrument (HVI) classing Plains is characterized by variable weather conditions
data, several cotton marketing pools have offered price that often lead to termination of cotton plants before
premiums for picked cottons with similar classing grades fibers reach full maturity. Experimental work is cur-
to stripped cottons. rently ongoing to characterize the processing character-
With the geographic shifts in US cotton production isitics related to lower micronaire values.12
and the introduction of new cultivars to the High Unlike the HVI, which estimates fiber properties
Plains, there is a question as to whether High Plains from a bundle of fibers, the Advanced Fiber
cottons might be able to fill the demand for fibers suit- Information System (AFIS) measures properties of
able for making high-quality, fine, ring-spun yarns. In individual fibers. For each replication, a 0.5 g sample
404 Textile Research Journal 82(4)

Table 2. Sampling site cultivar summary.13

2006 2007 2008

[a]
Site 1 ST 4554 B2RF
Site 3 ST4554 B2RFa, FM 9058 Fb, FM9063 B2Fc, PHY 485 WRFd
Site 4 ST4554 B2RFa, FM 9058 Fb, FM9063 B2Fc, PHY 485 WRFd
Site 5 FM 9180 B2Fe
Site 7 FM 9180 B2Fe
a
ST 4554 B2RF ¼ Stoneville 4554 Bollgard IIÕ Roundup Ready FlexÕ (Bayer CropScience; Research Triangle Park, NC).
b
FM 9058 F ¼ FiberMax 9058 FlexÕ (Bayer CropScience; Research Triangle Park, NC).
c
FM 9063 B2F ¼ FiberMax 9063 Bollgard IIÕ Roundup Ready FlexÕ (Bayer CropScience; Research Triangle Park, NC).
d
PHY 485 WRF ¼ PhytoGenTM 485 WidestrikeTM Roundup Ready FlexÕ (Dow AgroSciences; Indianapolis, IN).
e
FM 9180 B2F ¼ FiberMax 9180 Bollgard IIÕ Roundup Ready Flexß (Bayer CropScience; Research Triangle Park, NC).

is formed into a 30 cm long sliver and placed in the Table 3. Fiber quality properties for all samples from 2006
sampling tube of the AFIS. A pinned cylinder and (Site 1, n ¼ 8).14
fiber individualizer align and separate individual
Minimum Average Maximum
fibers, which then pass through one of two optical sen-
sors: one for trash and dust and the other for length, HVI
maturity and neps. Micronaire 2.9 3.3 3.6
The objective of this research was to investigate the Length (cm) 2.77 2.79 2.84
influence of fiber properties, harvest method and culti- Uniformity (%) 78.8 79.7 81.3
var on properties of ring-spun yarn produced from Strength (g/tex) 24.8 26.6 28.6
High Plains cottons. Fiber properties measured using Elongation (%) 8.1 8.5 9.2
the HVI (i.e. those available to merchants and buyers Reflectance 80.4 81.2 81.8
plus fiber elongation) and the AFIS, along with harvest Yellowness 7.9 8.4 8.9
method and cultivar/harvest location/year interaction,
Leaf grade 2 2.3 3
were analyzed to determine their relative influence on
AFIS
the properties of ring-spun yarns.
Nep size (mm) 707 726 748
Neps (cnt/g) 460 640 780
Methods Length by no. (cm) 1.52 1.60 1.73
Irrigated cotton was harvested using picker and strip- No. length CV (%) 56.9 62.1 65.1
per harvesters from six commercial farms on the High UQLa(cm) 2.79 2.85 2.92
Plains of Texas and ginned at the USDA-ARS Cotton SFCwb(%) 13.7 17.0 18.8
Production and Processing Research Unit in 2006, 2007 Total countc(cnt/g) 252 367 526
and 2008. Sample sites and methods are described in Trash size (mm) 286 304 322
detail by Faulkner et al. (Table 2).13 (Data from Site 6 Dust (cnt/g) 216 321 458
were excluded due to extremely low maturity – average Trash (cnt/g) 34 46 68
micronaire of 2.5 for this site – and non-field-cleaned VFMd (%) 0.83 1.13 1.44
samples were excluded due to the small number of sam- SCN Sized(mm) 995 1091 1202
ples. No fiber quality data was collected at Site 2.) In
SCNe(cnt/g) 13 19 26
summary, 140 kg samples of seed cotton were collected
IFCf(%) 11.1 13.4 15.0
from six commercial farms (sites) in which samples
were harvested using a picker (John Deere 9996) or Maturity ratio 0.75 0.77 0.81
stripper (John Deere 7460) harvester from adjacent Standard fineness 195 197 201
plots in the same field. a
UQL ¼ upper quartile length.
Fiber quality analyses are described in detail by b
SFCw ¼ short fiber content by weight.
Faulkner et al.14 Summary fiber quality data from
c
Total count ¼ Dust + Trash count per gram.
d
each site is shown in Tables 3–5. VFM ¼ visible foreign matter.
e
SCN ¼ seed coat nep.
Twenty three kilogram (50 lbs) samples of lint were f
IFC ¼ immature fiber content.
processed to produce carded ring-spun yarns at the HVI: High Volume Instrument, AFIS: Advanced Fiber Information System,
Fiber and Biopolymer Research Institute. While a CV: coefficient of variation.
Faulkner et al. 405

Table 4. Fiber quality properties for all samples from 2007.14

Site 3 (n ¼ 24) Site 4 (n ¼ 24)

Minimum Average Maximum Minimum Average Maximum

HVI
Micronaire 3.6 4.1 4.8 3.7 4.1 4.5
Length (cm) 2.69 2.92 3.05 2.82 2.98 3.15
Uniformity (%) 79.0 81.5 83.7 81.3 82.6 84.1
Strength (g/tex) 28.2 29.9 31.8 27.0 29.0 31.3
Elongation (%) 7.1 8.6 10.3 7.3 8.8 10.2
Reflectance 76.2 80.9 84.2 75.4 79.8 83.9
Yellowness 7.2 8.3 9.7 7.5 8.7 10.4
Leaf grade 1 1.6 3 1 1.4 3
AFIS
Nep Size (mm) 651 695 720 670 689 725
Neps (cnt/g) 240 354 506 226 326 426
Length by no. (cm) 1.75 1.87 2.01 1.83 1.95 2.11
No. length CV (%) 50.4 54.9 58.7 50.0 53.3 58.1
UQLa(cm) 2.79 3.03 3.18 2.95 3.09 3.25
SFCwb(%) 9.0 11.1 13.2 7.9 10.0 12.3
Total countc(cnt/g) 294 771 1758 307 945 2849
Trash size (mm) 253 278 294 243 261 298
Dust (cnt/g) 256 682 1528 284 858 2615
Trash (cnt/g) 38 89 230 23 86 234
VFMd(%) 0.66 1.82 3.71 0.65 1.87 4.67
SCN sized(mm) 696 811 923 761 848 955
SCNe(cnt/g) 6 17 30 6 16 28
IFCf(%) 7.5 9.4 11.7 6.8 8.7 10.0
Maturity ratio 0.81 0.84 0.89 0.82 0.85 0.88
Standard fineness 184 194 202 185 194 204
a
UQL ¼ upper quartile length.
b
SFCw ¼ short fiber content by weight.
c
Total count ¼ Dust + Trash count per gram.
d
VFM ¼ visible foreign matter.
e
SCN ¼ seed coat nep.
f
IFC ¼ immature fiber content.
HVI: High Volume Instrument, AFIS: Advanced Fiber Information System, CV: coefficient of variation.

brief description of the spinning process is given here, a (0.51–0.63 turn/cm) was added. Half of the slivers were
more detailed description of the spinning process and spun into 14.5 tex (40 Ne) yarns with a twist multiplier of
yarn quality results is given by Faulkner et al.15 4.2 (weaving twist) on a Seussen Fiomax ring-spinning
Samples were opened using a Rieter Monocylinder B4/ frame. Ten bobbins of yarn were made from each sample
1 and a Reiter ERM B5/5 (Winterthur, Switzerland). using a traveler speed of 32 m/s, a back to middle gauge
Cotton was carded (Model DK-903; Trützschler; of 64 mm, a middle to front gauge of 46 mm and a spin-
Mönchengladbach, Germany) at a production rate of dle speed of 13,500 rpm.
32 kg/h to produce carded slivers with a linear density Yarn count and skein break tests were performed
of 4600 tex. Card slivers were blended and drawn to a with a Scott Tester (Model J-2; Henry L. Scott,
linear density of 3900 tex using an HSR 1000 draw frame Providence, RI) (10 bobbins per sample); yarn elonga-
(Trützschler; Mönchengladbach, Germany). A Reiter tion, tenacity and work to break were measured with an
RSB 851 draw frame (Reiter, Winterthur, Switzerland) Uster Tensorapid (Model 3, Uster Technologies,
was then used to draw the samples to a final linear den- Knoxville, TN) (10 bobbins per sample and ten
sity of 4250 tex. On the roving frame, samples were breaks per bobbin); yarn evenness was tested with an
drawn to a linear density of 490 tex, and a slight twist Uster Tester (Model 3, Uster Technologies, Knoxville,
406 Textile Research Journal 82(4)

Table 5. Fiber quality properties for all samples from 2008.14

Site 5 (n ¼ 8) Site 7 (n ¼ 8)

Minimum Average Maximum Minimum Average Maximum

HVI
Micronaire 3.4 3.6 3.7 3.0 3.2 3.6
Length (cm) 3.00 3.03 3.07 2.95 3.00 3.05
Uniformity (%) 82.4 82.9 83.6 81.9 82.5 83.3
Strength (g/tex) 28.5 29.2 29.7 28.1 29.3 29.9
Elongation (%) 9.9 10.0 10.2 9.2 9.5 9.7
Reflectance 81.0 82.4 83.1 82.0 82.6 83.7
Yellowness 6.9 7.2 7.5 7.4 7.8 8.1
Leaf grade 1 1.25 2 1 1 1
AFIS
Nep size (mm) 690 699 706 700 708 718
Neps (cnt/g) 368 431 500 391 464 526
Length by no. (cm) 1.93 1.96 2.01 1.88 1.95 2.01
No. length CV (%) 52.4 53.5 54.4 57.8 59.9 62.5
UQLa(cm) 3.07 3.12 3.18 3.07 3.11 3.15
SFCwb(%) 9.3 10.0 10.7 9.3 10.3 11.3
Total countc(cnt/g) 224 331 402 243 384 484
Trash size (mm) 286 315 343 308 319 346
Dust (cnt/g) 190 282 352 207 325 406
Trash (cnt/g) 32 49 70 36 60 78
VFMd(%) 0.60 1.06 1.49 0.78 1.29 1.49
SCN sized(mm) 779 881 952 806 865 964
SCNe(cnt/g) 10 13 17 14 19 22
IFCf(%) 8.3 8.7 9.2 8.3 8.9 9.8
Maturity ratio 0.80 0.81 0.82 0.80 0.82 0.82
Standard fineness 185 187 189 184 187 189
a
UQL ¼ upper quartile length.
b
SFCw ¼ short fiber content by weight.
c
Total count ¼ Dust + Trash count per gram.
d
VFM ¼ visible foreign matter.
e
SCN ¼ seed coat nep.
f
IFC ¼ immature fiber content.
HVI: High Volume Instrument, AFIS: Advanced Fiber Information System, CV: coefficient of variation.

TN) (10 bobbins per sample and 400 m per bobbin) as through the variance of dust size data (in mm) is two
specified by American Society for Testing and orders of magnitude larger than that of fiber length
Materials (ASTM) Standard D1425/D1425M.16 data (in cm); an unstandardized analysis would give
A standardized principal component analysis (PCA) greater weight to variations in dust size than in fiber
was performed on all fiber and yarn quality data to give length). Results of the PCA include a number of prin-
insight to causes of deviation between samples and to cipal components (PCs), which are linear combinations
identify outlying samples. Harvest method, cultivar and of variables (fiber properties, harvest method and cul-
location were treated as categorical variables. All other tivar) that account for the maximum variance within a
data were standardized by centering the mean on zero dataset by describing mutually orthogonal vectors that
and scaling such that the standard deviation of all data most closely fit the n observations in p-dimensional
within each variable was equal to one. This standardi- space, where p is the number of variables measured
zation process was performed to give equal opportunity on each object. Each successive PC explains the maxi-
for all variables to influence the model regardless of mum possible amount of residual variance in the data-
their original variance (e.g. a standardized analysis set. The PCA reveals the effective dimensionality of a
gives equal weight to fiber length and dust size even dataset and eliminates redundancy caused by collinear
Faulkner et al. 407

variables, such as those expected in the type of fiber properties were conducted using the General Linear
analyses conducted. Model function in SPSS (ver. 14.0; SPSS, Inc.;
Following PCAs, partial least squares (PLS) regres- Chicago, Ill.).
sion analyses were performed to relate variations in
fiber properties and harvest method to relevant yarn
quality parameters. A PLS analysis is well suited for Results and discussion
relating the host of fiber quality parameters from HVI
and AFIS to yarn quality parameters because, unlike
Fiber quality PCA
standard regression methods that fail under conditions The PCA was performed using all data from HVI and
of collinearity among independent variables, PLS AFIS analyses. The first two PCs explained 70% of the
regression handles collinearity well.17 PLS regression variance in the fiber quality dataset. The score plot of
is akin to the PCA in that linear combinations of var- PC1 and PC2 shows clear grouping by cultivar, with all
iables (fiber properties, harvest method and cultivar) FiberMax cultivars clustered together and distinct
are constructed that account for the maximum variance groupings for PHY 485 and ST 4554 (Figure 5). PC1
within the response variables (yarn quality parameters). explained 40% of the variance in the fiber property
Given the limited number of samples, full cross valida- data and primarily differentiated between samples
tion was used to test model predicitivity after each suc- based on micronaire, length and length variation. PC2
cessive PC was added. Because full cross validation explained 30% of the variance in fiber quality dataset
does not utilize a truly independent dataset, it tends and primarily differentiated samples based on reflec-
to lead to over-estimation of the predictive ability of tance (Rd) and foreign matter content.
a model, but it is the best option when sample size is The loading plot of PC1 and PC2 (Figure 6) shows
limited relative to the model dimensionality.17 Loadings that length parameters were highly correlated, as were
and regression coefficients from PLS analyses were used length uniformity and foreign matter parameters.
to identify those fiber properties that most influenced Correlations between all fiber length parameters were
yarn quality. PCA and PLS analyses were conducted significant, with AFIS fiber length by weight (Lw)
using The Unscrambler (v. 9.8; CAMO Software AS; being most highly correlated to all other parameters
Oslo, Norway) software. Analyses of variance investi- (Figure 7). Correlations between trash count, dust
gations into select differences in fiber and yarn count and visible foreign matter were also significant,

Figure 5. Score plot for first two fiber quality principle components.
408 Textile Research Journal 82(4)

Figure 6. Loading plot for fiber quality principal component analysis.

Figure 7. Correlation matricies for length parameters.

with all correlation coefficients greater than 0.94. While between HVI and AFIS length uniformity parameters
AFIS length uniformity parameters (LnCV and LwCV) is likely the result of AFIS characterizing the entire
were highly correlated (r ¼ 0.969), correlation coeffi- distribution of measured fiber lengths, whereas HVI
cients between AFIS uniformity parameters and HVI length parameters focus on the longest half of the
uniformity were less than 0.8. This reduced correlation fiber distribution.
Faulkner et al. 409

Figure 8. Score plot (a) and loading plot (b) for yarn quality principal component analysis.

greater than 200% of the yarn cross-sectional area)

Yarn quality PCA explain most of the variations in the sampled yarns
The PCA was performed on all carded yarn quality followed by variations in yarn strength parameters
parameters. The first two PCs explained 92% of the (tenacity and count-strength-product (CSP));
variance in the yarn quality data. The score plot of – yarn strength parameters (CSP and tenacity)
PC1 and PC2 also showed clear groupings by cultivar were highly correlated to each other and nega-
(Figure 8). The loading plot (Figure 8) showed that: tively correlated to yarn evenness parameters
(Figure 9);
– variation in yarn evenness parameters (thin places, – yarn elongation and work-to-break were corre-
thick places, coefficient of variation (CV) and neps lated (r ¼ 0.838; p < 0.0005); and
410 Textile Research Journal 82(4)

Figure 9. Correlation matrix for yarn strength and evenness parameters.

– yarn hairiness was not correlated with any of

these parameters (maximum jrj ¼ 0.578). Table 6. Average yarn evenness parameters

FiberMax Non-FiberMax
Yarn evenness regression Parameter samples samples p-value

Because all yarn evenness parameters were well corre- CV (%) 14.5 14.7 <0.0005
lated to the number of neps (+200%) per km of yarn, a Thin placesa (cnt/km) 9 28 <0.0005
PLS1 regression was performed in which the number of Thick placesb (cnt/km) 90 170 <0.0005
yarn neps was regressed against HVI, AFIS, harvest Neps (+200%) (cnt/km) 65 104 0.001
method and cultivar data. Seven outliers having low a
Thin places ¼ points in the yarn less than 50% of the average thickness.
maturity (maturity ratio  0.78) were removed from b
Thick places ¼ points in the yarn greater that 150% of the average
this and all subsequent regression. These seven samples thickness.
formed an outlying group of weak yarns with excep- CV: coefficient of variation.
tionally high numbers of thick places. Redundant fiber
quality variables (related variables with jrj > 0.9) were (Table 6), were significantly longer ( p < 0.0005) than
also removed from this and all subsequent regressions other samples and had lower values of standard fine-
(e.g. Lw was used to indicate fiber length and all other ness (Hs; p < 0.0005). FM 9180 samples, which were
AFIS length measurements were removed) along with characterized by lower values of micronaire, Hs and
variables having regression coefficients with confidence low foreign matter contents, formed a cluster of
intervals encompassing zero (including harvest highly uniform yarns.
treatment).
After these modifications to the dataset, variations in
Yarn strength regression
fiber length accounted for most of the variation in PC1,
which explained 74% of the yarn evenness variance. Yarn tenacity was used to indicate yarn strength as it
FiberMax cultivars, which produced more even yarns correlated well with CSP (r ¼ 0.931). A PLS1 regression
Faulkner et al. 411

was performed in which yarn tenacity was regressed

against HVI, AFIS, harvest method and cultivar data.
Yarn work-to-break regression
After removing insignificant variables from the regres- Yarn work-to-break was correlated with yarn elonga-
sion, including harvest treatment, grouping by cultivar tion (r ¼ 0.838). A PLS1 regression was performed in
was prevalent in the score plots. PC1 explained 74% of which yarn work-to-break was regressed against HVI,
the tenacity variance and largely differentiated between AFIS, harvest method and cultivar data. In addition to
FiberMax and non-FiberMax cultivars. PC2 explained the seven immature samples removed in previous anal-
only 3% of the remaning yarn tenacity variance. yses, three samples with low work-to-break values (<
ST4554 and PHY 485 samples were less tenacious 300 cN-cm) were excluded from the analysis because
than most of the FiberMax samples. they lay outside the body of most data points and
Fiber length, fineness and reflectance were the most exerted high levereage on the regression analysis.
influential variables in PC1. Longer, more uniform After removal of insignificant variables from the
fibers produce more tenacious yarns because the slip- regression, including harvest treatment, the first two
ping resistance of fibers increases due to greater fiber- PCs explained 84% of the variance in yarn work-to-
to-fiber friction.18–20 Similar to the results of this study, break. Distinct clusters, grouped by cultivar, were obvi-
El Mogahzy21 reported that skein break factor (or CSP) ous in the regression plots and can be seen in boxplots
increased with increasing fiber length, length uniformity of the work-to-break data (Figure 10). Although the
and fiber strength but decreased with increasing Rd and FM 9058 and FM 9063 samples produced stronger
fiber fineness. El Mogahzy et al.22 found similar results yarns, the yarns required less total energy to break
for a different set of cottons but did not find significant than yarns produced by the PHY 485 and ST 4554
correlations between Rd and CSP. Ramey et al.18 found samples. An analysis of variance of fiber strength and
that fiber tenacity measured at 3.2 mm gage length elongation by cultivar revealed no significant differ-
explained more than 70% of the variation in observed ences between cultivars with regards to fiber strength,
yarn tenacity. Variances in fiber strength contributed to but PHY 485 and ST 4554 samples demonstrated sig-
variances in yarn tenacity, but fiber length was more nificantly better elongation than the FM 9058 and FM
prevalent for the samples in this study. 9063 samples (Figure 11; p < 0.0005).

Figure 10. Yarn work-to-break by cultivar.

412 Textile Research Journal 82(4)

Figure 11. Fiber strength (a) and elongation (b) by cultivar.

Work-to-break is determined by integrating the area regardless of the cultivar, most of the variation in
under the displacement-force to break the curve up to work-to-break resulted from varations in fiber and
the point of rupture. For cotton this curve is mostly yarn elongation.
linear and can be approximated by the product of The results of yarn work-to-break analyses reveal
tenacity and elongation. Therefore, it is not surprising the importance of fiber elongation in producing yarns
that work-to-break was positively correlated to fiber that can withstand the forces placed on them during
elongation (r ¼ 0.841). Previous research has demon- weaving. However, fiber elongation is not currently
strated that the elongation of ring-spun yarns is most included in cotton classing reports due to difficulties
influenced by the elongation of the fibers comprising in calibrating this measurement between HVI
that yarn.23–27 Yarn elongation and work-to-break machines. Although it has been recognized that fiber
are important properties because they correlate well elongation affects yarn quality and weaving perfor-
with weaving efficiency.28 Modern air-jet looms require mance,29 presently, there is no method by which cus-
yarns to elongate more than 4% without breaking tomers can evaluate fiber elongation prior to purchase.
during typical opening and closing cycles, which typi- Bargeron31 attempted to predict the bundle elonga-
cally occur 3000 times per minute.29 Although yarn tion of cotton fibers using HVI measurements account-
tenacity is often used to judge weaving performance, ing for the flex in the columns connecting the stepper
work-to-break (or work of rupture) is a more appropri- motors and fiber clamps in HVI systems in order to
ate indicator of yarn performance because it describes develop a standard measurement protocol for fiber
the total energy required to break a yarn. elongation unaffected by fiber bundle strength.
As cotton fibers mature, the secondary wall thickens Benzina et al.29 reported that elongation results from
as highly crystalline cellulose is deposited inside the pri- a single HVI were repeatable, but values could not be
mary cell wall.30 The increasing thickness leads to reliably compared between HVI systems. The authors
greater fiber strength, but the crystalline structure of proposed developing calibration standards for fiber
the secondary wall makes the fibers more brittle, thus elongation to allow for inter-system comparison.
decreasing fiber elongation. This tradeoff between fiber Results of the present research point to the need for
strength and elongation explains why yarn work-to- continuing efforts to reliably characterize the breaking
break was positively correlated to fiber elongation but elongation of fibers as an additional cotton classing
negatively correlated to fiber maturity. paramter.
Although yarn work-to-break is approximated by
the product of tenacity and elongation, variations in
yarn work-to-break were more highly correlated to var-
Yarn hairiness regression
iations in yarn elongation than to variations in yarn A PLS1 regression was performed in which yarn hair-
strength parameters (Figure 8). Within the analyzed iness was regressed against HVI, AFIS, harvest method
samples, the CV for HVI strength was 3.9%, while and cultivar data. After removal of insignificant vari-
the CV for fiber elongation was 13%. Because there is ables from the regression, including harvest treatment,
little variation in fiber strength between samples, the regression explained a maximum of 59% of the
Faulkner et al. 413

variation in yarn hairiness with two PCs. PC1 and PC2 the greatest bundle elongation-produced yarns, requir-
explained 56% and 3% of the variation in yarn hairi- ing more total energy to rupture. Fiber bundle elonga-
ness, respectively. Variations in fiber length by weight tion can be measured by HVI systems, but, due to a
(Lw; r ¼ 0.399; p ¼ 0.001) and length uniformity lack of calibration standards, the results are not com-
(r ¼ 0.563; p < 0.0005) explained most of the parable between systems.29 The present research points
accounted-for variation in yarn hairiness. The current to the need for a calibrated method for efficiently mea-
results support those of previous researchers that suring fiber elongation for both breeding selection and
increasing fiber length reduces yarn hairiness.32–34 for predicting a cotton’s performance during spinning
Viswanathan et al.33 found that fiber fineness had the and textile production.
greatest effect on yarn hairiness, followed by fiber
length, but Zhu and Ethridge34 found no correlation Acknowledgements
between fiber fineness and yarn hairiness. The authors would like to gratefully acknowledge Brad
Some hairiness is desirable as it leads to comfort and Palmer, Craig Heinrich, James Brown, Rickey Bearden,
coverage in fabrics and it increases the velocity and Steve Verett and Kelly Kettner for providing cooperation,
insertion rate of wefts during weaving due to higher cotton fields and harvesting equipment for this project.
friction between yarns and the air surface.35,36
However, excessive hairiness affects the appearance of Disclaimer
yarns in a fabric and increases the tension placed on Mention of trade names or commercial products in this man-
yarns during the weaving process, which in turn affects uscript is solely for the purpose of providing specific informa-
both ends down and fabric properties. Adanur and tion and does not imply recommendation or endorsement by
Jing37 reported that the maximum filling tension the US Department of Agriculture.
placed on a warp yarn increased between 53% (for an
average of 35 picks) and 145% (for a single pick) as Funding
hairiness increased from 1750 to 2500 cnt/1000 m
This work was supported by the Texas Department of
during 3/1 twill weaving. This increase in filling tension Agriculture Food and Fibers Research Grant Program
led to increased fabric air permeability, which led to [Grant Number FF-a-0809-02], Cotton Incorporated, Texas
greater fabric shrink after laundering and a decrease State Support Committee – Cotton Incorporated, and the
in fabric tear strength.36 While the observed changes Plains Cotton Growers – Plains Cotton Improvement
in filling tension resulted from a complex interaction Program, as well as an equipment donation from John
of yarn twist, count and hairiness, it suggests that exces- Deere – Des Moines Works.
sive increases in yarn hairiness lead to reductions in
yarn performance and fabric quality, and that varia- Conflict of interest statement
tions in yarn hairiness were not predicted well by var- None declared.
iations in measured fiber characteristics for the
analyzed samples.
References
1. National Cotton Council. ‘Country Statistics’, http://
Conclusions www.cotton.org (2010, accessed June 2010).
2. United States Census Bureau. Current industrial reports: yarn
Although differences in yarn quality were evident production (MA313F). Washington, DC: US Department of
between yarns formed from picker- and stripper- Commerce, 2005.
harvested cottons,15 based on the 76 cotton samples 3. United States Census Bureau. Current industrial reports:
analyzed in this study, differences in yarn quality result- textiles (MQ313A). Washington, DC: US Department
ing from harvest method were sufficiently captured by of Commerce, 2010.
differences in fiber quality parameters measured by the 4. Plastina A. Open-end versus ring spun cotton yarns.
HVI and AFIS such that harvest method was not a Washington, DC: International Cotton Advisory
significant factor in any of the final yarn quality Committee, 2009.
models analyzed. Variations in yarn evenness were lar- 5. International Cotton Advisory Committee (ICAC). World
gely explained by variations in fiber length such that cotton trade, September 2010: a report from the Secretariat of
the International Cotton Advisory Committee. Washington,
FiberMax cultivars, which in general had longer
DC: ICAC, 2010.
fibers, formed more even yarns than non-FiberMax 6. United States Department of Agriculture - National
samples. Yarn strength was most influenced by varia- Agricultural Statistics Service. ‘US Cotton Production.
tions in fiber length, fineness and reflectance. Yarn National Agricultural Statistics Database’, http://
work-to-break was substantially affected by fiber www.nass.usda.gov (2010, accessed June 2010).
bundle elongation, which is not currently included in 7. Cotton Incorporated. EFSÕ USCROPTM. Ver. 4.0.0.
cotton classing reports. Those samples demonstrating Cary, NC: Cotton Incorporated, 2009.
414 Textile Research Journal 82(4)

8. Hequet EF and Abidi N. Spinning performances of West 23. Fiori LA, Sands JE, Little HW and Grant JN. Effect of
Texas upland cotton. In: Proceedings of the 2009 Beltwide cotton fiber bundle break elongation and other fiber
Cotton Conferences, Memphis, TN: National Cotton properties on the properties of a coarse and a medium
Council, 2009. singles yarn. Textil Res J 1956; 26: 553–564.
9. McCreight DJ, Feil RW, Booterbaugh JH and Backe EE. 24. Virgin WP and Wakeham H. Cotton quality and fiber
Short staple yarn manufacturing. Durham, NC: Carolina properties, part IV. Textil Res J 1956; 26: 177–191.
Academic Press, 1997. 25. Willium TW and Phillips J. The effects of fiber bundle
10. Krifa M, Gourlot JP and Drean JY. Effect of seed coat elongation of medium staple cottons on processing per-
fragments on yarn strength: dependence on fiber quality. formance and yarn properties. Textil Res J 1966; 36:
Textil Res J 2001; 71: 981–986. 1004–1012.
11. Krifa M and Ethridge D. Texas plains cotton perfor- 26. Bogdan JF. The prediction of cotton yarn strengths.
mance in high value-added ring spinning applications: a Textil Res J 1967; 37: 536–537.
progress report. Textil Top 2007; 1: 2–8. 27. Majumdar PK and Majumdar A. Predicting the breaking
12. Krifa M. The quality of Texas plains cotton: recent elongation of ring spun cotton yarns using mathematical,
achievements and current challenges. In: Proceedings of statistical, and artificial neural network models. Textil
the 2011 Beltwide Cotton Conferences, Memphis, TN: Res J 2004; 74: 652–655.
National Cotton Council, 2011. 28. Aggarwal SK. A model to estimate the breaking elonga-
13. Faulkner WB, Wanjura JD, Boman RK, Shaw BW and tion of high twist ring spun cotton yarns – part I: deriva-
Parnell CB. Evaluation of modern cotton harvest systems tion of the model for yarns from single cotton varieties.
on irrigated cotton: harvester performance. Appl Eng Textil Res J 1989; 59: 691–695.
Agric 2011; 27: 497–506. 29. Benzina H, Hequet E, Abidi N, Gannaway J, Drean JY
14. Faulkner WB, Wanjura JD, Hequet EF, Boman RK, and Harzallah O. Using fiber elongation to improve
Shaw BW and Parnell CB. Evaluation of modern genetic screening in cotton breeding programs. Textil
cotton harvest systems on irrigated cotton: fiber quality. Res J 2007; 77: 770–778.
Appl Eng Agric 2011; 27: 507–513. 30. Wakelyn PJ, Edwards JV, Bertoniere NR, et al. Cotton
15. Faulkner WB, Wanjura JD, Hequet EF, Boman RK, fiber chemistry and technology. Boca Raton, FL: CRC
Shaw BW and Parnell CB. Evaluation of modern Press, 2007.
cotton harvest systems on irrigated cotton: yarn quality. 31. Bargeron JD. Cotton elongation measurement using High
Appl Eng Agric 2011; 27: 523–532. Volume Instruments. Trans ASABE 1998; 41: 1583–1587.
16. ASTM. ASTM D1425/D1425M: standard test method 32. Barella A and Manich AM. The influence of the spinning
for unevenness of textile strands using capacitance testing process, yarn linear density, and fibre properties on the
equipment. ASTM Standards. West Conshohocken, PA: hairiness of ring-spun and rotor-spun yarns. J Textil Inst
ASTM, 2009. 1988; 79: 189–197.
17. Esbensen KH. Multivariate data analysis – in practice, 5th 33. Viswanathan G, Munshi VG, Ukidve AV and Chandran
ed. Woodbridge, NJ: CAMO Process AS, 2004. K. A critical evaluation of the relationship between fiber
18. Ramey HH, Lawson R and Worley S. Relationship of quality parameters and hairiness of cotton yarns. Textil
cotton fiber properties to yarn tenacity. Textil Res J Res J 1989; 59: 707–711.
1977; 47: 685–691. 34. Zhu R and Ethridge MD. Predicting hairiness for ring
19. Graham JS and Taylor RA. Development of a method to and rotor spun yarns and analyzing the impact of fiber
measure cotton spinnability. Textil Res J 1978; 48: properties. Textil Res J 1997; 67: 694–698.
286–292. 35. Turel T, Bakhtiyarov S and Adanur S. Effects of air and
20. Üreyan ME and Kadoğlu H. Regressional estimation of yarn characteristics in air-jet filling insertion. Part I: air
ring cotton yarn properties from HVI fiber properties. velocity and air pressure measurements. Textil Res J
Textil Res J 2006; 76: 360–366. 2004; 74: 592–597.
21. El Mogahzy YE. Selecting cotton fiber properties for fit- 36. Adanur S and Turel T. Effects of air and yarn character-
ting reliable equations to HVI data. Textil Res J 1988; 58: istics in air-jet filling insertion. Part II: yarn velocity mea-
392–397. surements with a profiled reed. Textil Res J 2004; 74:
22. El Mogahzy YE, Broughton R and Lynch WK. 657–661.
Statistical approach for determining the technological 37. Adanur S and Jing Q. Property analysis of denim fabrics
value of cotton using HVI fiber properties. Textil Res J made on air-jet weaving machine. Part II: effects of ten-
1990; 60: 495–500. sion on fabric properties. Textil Res J 2008; 78: 10–20.