Effects of Trash and Processing on Cotton Fiber Quality ...

Effects of Trash and Processing on Cotton Fiber Quality Measurements

by

João Paulo Saraiva Morais, M.Sc.

A Dissertation

In

Plant and Soil Science

Submitted to the Graduate Faculty

of Texas Tech University in

Partial Fulfillment of

the Requirements for

the Degree of

DOCTOR OF PHILOSOPHY

Approved

Eric F. Hequet

Chair of Committee

Brendan R. Kelly

Noureddine Abidi

Carol M. Kelly

John Wanjura

Mark Sheridan

Dean of the Graduate School

May 2020

Texas Tech University, João Paulo Saraiva Morais, May 2020

ii

ACKNOWLEDGMENTS

I would like to thank Dr. Eric Hequet for all his help, guidance, discussions about

logic and helping me to improve as a person and as a researcher. It is also a great honor to

be able to learn with Dr. Brendan Kelly, showing me new dimensions for my personal

and professional life.

I would like to express my gratefulness to my committee members, Dr.

Noureddine Abidi, Dr. Carol Kelly, and Dr. John Wanjura for their insights, comments,

guidance, and help. I want to thank all my fellow students who crossed their paths with

mine while I was learning and developing this research work, especially Abu Sayeed,

Addisu Ayele, Addisu Tesema, Arifa Sultana, Brooke Shumate, Deepika Mishra, Islam

Mahbubul, Jacob James, Rakib Hasan, Rohan Brown, Scott Baker, Suman Lamichhane,

Vikki Martin, and Zach Hinds. I want to show my gratitude to all staff and faculty in the

Plant and Soil Sciences Department, especially at the Fiber and Biopolymer Research

Institute. Without their help, it would not be possible for me to finish my studies.

I would like to thank Embrapa, Harold & Mary Dregne Graduate Program

Endowment, Todd & Kasey Thompson, Nancy Gonzales, and Cotton Incorporated for the

financial support.

I thank my wife, Ana Mônica for her help. My family and my wife’s families for

their support. My friends in Brazil and in Lubbock for helping me to think and rethink

about my past, present, and future. Finally, I want to give special thanks to GOD, because

with GOD all things are possible.


iii

TABLE OF CONTENTS

ACKNOWLEDGMENTS .................................................................................... ii

ABSTRACT .......................................................................................................... vi

LIST OF TABLES ............................................................................................. viii

LIST OF FIGURES ...............................................................................................x

1. LITERATURE REVIEW .................................................................................1

1.1 Economic importance of cotton ...................................................................1

1.2 Cotton physiology ........................................................................................2

1.3 Fiber development .......................................................................................3

1.3.1 Initiation ..............................................................................................4

1.3.2 Elongation ...........................................................................................4

1.3.3 Secondary cell wall thickening ...........................................................5

1.3.4 Maturation ...........................................................................................6

1.4 Cotton processing.........................................................................................7

1.4.1 Harvesting ...........................................................................................8

1.4.2 Ginning and lint cleaning ..................................................................10

1.5 Fiber quality assessment ............................................................................13

1.5.1 History of cotton classification .........................................................13

1.5.2 High volume instrument (HVI) ........................................................15

1.5.3 Advanced fiber information system (AFIS) ....................................18

1.5.4 Trash analyzer instruments ...............................................................19

1.5.4.1 Shirley analyzer .......................................................................20

1.5.4.2 Microdust and trash analyzer (MDTA) ...................................22

1.6 Final remarks .............................................................................................23

1.7 Bibliography ..............................................................................................25

2. EFFECTS OF NON-LINT MATERIAL ON HERITABILITY

ESTIMATES OF COTTON FIBER LENGTH PARAMETERS ...................35

2.1 Introduction ................................................................................................35

2.2 Material and methods .................................................................................39

2.2.1 Mating design and plant materials ....................................................39

2.2.2 Identifying parental material .............................................................39

2.2.3 Obtaining F2 seed ..............................................................................39

2.2.4 Field experiment ...............................................................................40

2.2.5 Harvesting, ginning, and processing .................................................41

2.2.6 Fiber quality testing ..........................................................................42


iv

2.2.7 Heritability estimates and statistics ...................................................43

2.3 Results and discussion ...............................................................................45

2.3.1 Sample type characteristics ...............................................................45

2.3.2 Parental material characteristics .......................................................46

2.3.3 Heritability estimates for HVI length measurements ........................49

2.3.4 AFIS fiber quality properties ............................................................52

2.3.4.1 AFIS length distributions .........................................................52

2.3.4.2 The 5% longer fibers by number and upper quartile

length by weight ...................................................................................52

2.3.4.3 Mean length by number ...........................................................55

2.3.4.4 Short fiber content by number .................................................56

2.4 Conclusions ................................................................................................57

2.5 Bibliography ..............................................................................................59

3. EFFECT OF THE SHIRLEY ANALYZER ON FIBER LENGTH

DISTRIBUTIONS ................................................................................................63

3.1 Introduction ................................................................................................61

3.1.1 Fiber length distribution and factors that can affect it ......................61

3.1.2 Instruments to measure fiber length distribution ..............................66

3.1.3 Shirley analyzer ................................................................................68


3.2.1 Germplasm development ..................................................................70

3.2.2 Field experiment ...............................................................................71

3.2.3 Mechanical processing ......................................................................71

3.2.4 Instruments to measure fiber length distribution ..............................73

3.2.5 Statistical analysis .............................................................................75

3.3 Results and discussion ...............................................................................76

3.3.1 Range of parameters .........................................................................76

3.3.2 Fiber length distributions ..................................................................79

3.3.3 Differences between the fiber length distributions ...........................82

3.4 Conclusions ................................................................................................88

3.5 Bibliography ..............................................................................................90

4. PROCESSING EFFECTS OF AFIS AND SHIRLEY ANALYZER

ON COTTON SAMPLES ...................................................................................94

4.1 Introduction ................................................................................................94


4.2.1 Cotton samples ..................................................................................97

4.2.2 Experimental procedure ....................................................................98

4.2.3 Statistical analysis ...........................................................................100


v

4.3 Results and discussion .............................................................................101

4.3.1 Descriptive statistics of AFIS fiber quality parameters ..................101

4.3.2 Processing effect in the fiber length distributions...........................104

4.4 Conclusions ..............................................................................................112

4.5 Bibliography ............................................................................................113

5 A METHOD TO IMPROVE COTTON FIBER LENGTH

MEASUREMENT FOR LABORATORY ANALYSIS .................................115

5.1 Introduction ..............................................................................................115

5.2 Method .....................................................................................................116

5.3 Comparison with other laboratory instruments and validation of the

method............................................................................................................117

5.3.1 Plant material ..................................................................................118

5.3.2 Cleaning and ginning ......................................................................118

5.3.3 Fiber quality testing ........................................................................120

5.3.3.1 Differences for trash and neps counts ....................................120

5.3.3.2 Differences for fiber length parameters .................................122

5.3.3.3 Differences for fiber length distributions ...............................123

5.4 Conclusions ..............................................................................................125

5.5 Bibliography ............................................................................................126

6 GENERAL CONCLUSIONS ........................................................................128


vi

ABSTRACT

Fiber quality improvement is of the utmost importance for all segments of the

cotton industry, i.e., cotton research, ginning, marketing, and textile processing. Farmers

receive premiums or discounts based on the quality of the cotton they produce. Textile

mills use fiber quality data to optimize their purchase decisions for their line of products.

Therefore, the precision and accuracy of fiber data must be as good as possible.

The experimental techniques used to measure fiber quality were typically

developed using clean cotton samples. Nevertheless, typical cotton samples from research

or commercial fields have a combination of lint and non-lint material, also known as

trash. The trash may impact the measurement of some fiber properties such as micronaire

and length. In Chapter 2 of this dissertation, I studied the impact of trash on the

measurement of cotton fiber length by HVI and AFIS. I observed that trash impacts the

length measurement precision. Therefore, researchers must keep their samples as clean as

possible for fiber quality analyses to be meaningful. Researchers may test samples with

native low trash content or may clean them with a type of mechanical processing. A

shortcoming of mechanically cleaning samples is that it may modify fiber properties and

in particular fiber length distribution.

Instruments typically used to process cotton samples may grip fibers between two

moving parts during the operation. If the fiber is strained with enough energy, it will

break. Therefore, instruments may change the fiber length distribution while cleaning the

sample. In Chapter 3, I analyzed the impact of a cleaning device, the Shirley analyzer, on

the fiber length distributions of a set of samples with large variability for fiber properties.

I concluded that there is an interaction cleaning x cotton, i.e., all cottons do not behave

the same way when submitted to the Shirley analyzer cleaning. I hypothesized that this

happens because this instrument may both break and remove fibers.

Researchers may use the Shirley analyzer for cotton cleaning experiments but not

for fiber breakage because this instrument may also remove fibers from a sample. It is

important to determine which instrument can be used for research on fiber breakage. In

Chapter 4, I compared the mechanical processing of the Shirley analyzer and the AFIS

(Advanced Fiber Information System) individualizer. I concluded that the AFIS


vii

individualizer is an instrument more suitable than the Shirley analyzer for cotton fiber

breakage studies because the AFIS may impact the fiber length distribution of a cotton

sample only by fiber breakage.

Treating cotton samples through a sequence of instruments adds a specific type of

processing effect to cotton samples. As a practical application of this observation, a

sample ginned and cleaned with industry-scale machinery may have a fiber length

distribution different from the distribution created by ginning this sample with a

laboratory-scale instrument. In Chapter 5, I studied how three different laboratory-scale

lint cleaners may impact the values of AFIS fiber length parameters of samples ginned

with a laboratory-scale gin. Furthermore, I analyzed which of these instruments can be

used to bring the values of the fiber length parameters to the same level than the samples

ginned with an industry-scale gin. I successfully defined a methodology to simulate

industry-scale ginning under laboratory-scale conditions.

This dissertation shows that trash and processing are factors that must be taken

into account when sending samples for fiber quality testing. Cotton researchers must be

aware of these factors when interpreting their results and making their decisions.


viii

LIST OF TABLES

2.1 Two-way analysis of variance used to calculate the heritability

estimates in this research ...........................................................................44

2.2 Average value for parameters related to trash and processing in

nine F2 samples, with three field replications, from three treatments

with different levels of trash content and mechanical processing .............45

2.3 Fiber quality properties of the hand picked parental varieties in

three field replications tested by HVI ........................................................47

2.4 Quality properties of the hand picked parental varieties in three

field replications tested by AFIS................................................................48

2.5 Heritability estimates for HVI length fiber properties in three

treatment of nine F2 samples, with three field replications, with

different levels of trash content and mechanical processing .....................49

2.6 Heritability estimates for AFIS length fiber properties in three

treatments of nine F2 samples, with three field replications, with

different levels of trash content and mechanical processing .....................54

3.1 Crossing scheme with the parent used to develop the populations in

this research ...............................................................................................70

3.2 Dynamic ranges between 95% and 5% quantiles for a set of 240 F3

lines tested by AFIS and HVI, with and without Shirley analyzer

processing ..................................................................................................77

3.3 Average values for fiber quality properties of 240 F3 samples with

and without Shirley analyzer processing ...................................................78

3.4 Explained variation per principal component from the principal

component analysis performed with the difference between the

flipped cumulative distributions with and without processing with

a Shirley analyzer of 240 samples .............................................................82

3.5 Fiber quality parameters for samples A and B, with and without

processing with a Shirley analyzer ............................................................85

3.6 Fiber quality parameters for samples C and D, with and without

processing with a Shirley analyzer ............................................................87

4.1 Parameters related to processing effect and trash content for

samples with different processing effects ................................................102

4.2 Fiber length parameters from the distributions for samples with

different processing effects ......................................................................103

4.3 Fiber length parameters related to the fineness/maturity complex

for samples with different processing effects ..........................................104

5.1 Average values of trash and neps parameters in a dataset of 20

genetic materials, processed with five different ginning approaches ......121


ix

5.2 Average values of fiber length parameters in a dataset of 20

genetic materials with a wide range of length properties, processed

with five different ginning approaches ....................................................122


x

LIST OF FIGURES

1.1 Developmental stages of cotton fibers, from initiation to

maturation (Lee et al., 2007) ........................................................................4

1.2 Students handpicking cotton (Zach Hinds, 2017) ........................................8

1.3 Cotton harvesting with a stripper machine (Addisu Ferede, 2018) .............9

1.4 Industrial-scale gin (A) with details for the gin stand (B) and lint

cleaner (C) in comparison to laboratory-scale gins (D) (Jacob

James, 2017) ..............................................................................................12

1.5 High volume instrument (HVI) (Addisu Ferede, 2019).............................16

1.6 Advanced Fiber Information System (AFIS) .............................................18

1.7 Shirley analyzer from Shirley Institute ......................................................21

1.8 Microdust and trash analyzer (MDTA)......................................................22

2.1 Fibrosamples in the High Volume Instrument (HVI). A cotton

sample is deposited into the fibrosampler basket (A). Fibers

protrude from the perforated curved wall (B). The comb is passed

externally on this surface, creating the fiber beard (C). The comb is

brushed on a carding surface on the fibrosampler (D) to remove

free fibers and to begin to parallelize the beard .........................................37

2.2 Flowchart from original crosses to field experiment .................................41

2.3 Flowchart from the field experiment to fiber quality analysis ...................43

2.4 AFIS length distributions for the analyzed samples. The black lines

represent the range among the different bins and the grey line

represents the average. Treatment I: samples with low trash content

and low mechanical stress; treatment II: samples with high trash

content and low mechanical stress; treatment III: samples with low

trash content and high mechanical stress ...................................................52

3.1 The scenarios that can occur during fiber processing. One fiber can

be gripped at two different points (A) or at one point (B or C).

Breakage may occur in scenario A, but not in B or C ...............................64

3.2 Length distribution by number of a cotton sample before and after

mechanical processing ...............................................................................66

3.3 Fiber length distributions of two cotton samples with the same

upper half mean length (30.0 mm) with and without processing.

Although the upper half mean length did not change for both

samples, sample A exhibits a smaller change in the whole

distribution than sample B .........................................................................68

3.4 Fiber length distribution (solid line) and the threshold (vertical

dashed line) for the length of fibers that can be gripped on both


xi

ends by the Shirley analyzer. Virtually all fibers in a sample can be

stressed .......................................................................................................69

3.5 Laboratory scale Compass 10-saw-gin, model MG1010 ...........................72

3.6 Shirley analyzer from Shirley Institute ......................................................73

3.7 Flowchart of the experimental procedure to evaluate the processing

effect ..........................................................................................................74

3.8 Conversion from the original AFIS length distribution of

subsamples with and without processing with a Shirley analyzer

(A) to the flipped first cumulative distributions (B) and to the

curve of differences between two subsamples (C) ....................................76

3.9 Examples of fiber length distributions for the same sample with

and without Shirley analyzer processing ...................................................80

3.10 PCA loadings for the bins of the difference between the flipped

cumulative distributions with and without processing with a

Shirley analyzer of 240 samples ................................................................83

3.11 Curves of the difference between the flipped first cumulative

distributions of subsamples with and without processing with a

Shirley analyzer (I), the flipped first cumulative distributions (II)

and the original length distributions (III) of two samples with

contrasting scores for PC1 .........................................................................84

3.12 Curves of the difference between the flipped first cumulative

distributions of subsamples with and without processing with a

Shirley analyzer (I), the flipped first cumulative distributions (II)

and the original length distributions (III) of two samples with

contrasting scores for PC2 .........................................................................86

4.1 A cotton sample is placed on the feed tray of the Shirley analyzer

(A). Cleaned fibers follow to the condenser (B), while rejected

fibers go to the trash tray (C) .....................................................................94

4.2 Parts of an Advanced Fiber Information System (AFIS). The

pinned perforated cylinder (A) spins at high speed and separates

fibers and trash. Fibers are sent through the plastic tube (B) to the

optical sensor (C) where they are tested. After testing, fibers and

trash are discarded in a single chamber (D) ...............................................96

4.3 Preparation of a sliver for AFIS testing. After weighing 0.50 g of

cotton (A), the sample is gently pulled (B) and rolled (C) to form a

sliver with a length of approximately 30 cm (D) .......................................98

4.4 Flowchart of the experimental procedure to compare the Shirley

analyzer and the AFIS processing effects ..................................................99

4.5 Scatterplot of scores for principal components 1 and 2 for samples

with different processing effects. Treatment X is the difference

between the fiber length distribution of a sample with one passage


xii

through the AFIS individualizer and the fiber length distribution of

a sample with one passage through the Shirley analyzer and one

passage through the AFIS individualizer. Treatment Y is the

difference between the fiber length distribution of a sample with

one passage through the AFIS individualizer and the fiber length

distribution of a sample with two passages through the AFIS

individualizer ...........................................................................................105

4.6 Curves of the flipped first cumulative distributions (I) and the

original length distributions (II) of sample A. Treatment X is the

difference between the fiber length distribution of the sample with

one passage through the AFIS individualizer (FL I) and the fiber

length distribution of the sample with one passage through the

Shirley analyzer and one passage through the AFIS individualizer

(FL II). Treatment Y is the difference between the fiber length

distribution of the sample with one passage through the AFIS

individualizer (FL I) and two passages through the AFIS

individualizer (FLIII) ...............................................................................107

4.7 Curves of the flipped first cumulative distributions (I) and the

original length distributions (II) of sample B. Treatment X is the








individualizer (FL III) ..............................................................................109

4.8 Curves of the the flipped first cumulative distributions (I) and the

original length distributions (II) of sample C. Treatment X is the








individualizer (FL III) ..............................................................................111

5.1 Flowchart of the validation method .........................................................119

5.2 Fiber length distributions by number of two samples from

industry-scale gin, laboratory-scale gin, laboratory-scale gin +

Shirley analyzer from Shirley institute, laboratory-scale gin +

Shirley analyzer MK2, and laboratory-scale gin + MDTA 3 ..................124


1

CHAPTER I

LITERATURE REVIEW

1.1 Economic importance of cotton

Cotton is the most important textile plant fiber in the world (Stewart et al. 2010).

Archeological evidence shows the use of cotton as textile fiber in Asia as string used to

connect copper beads around 6,000 years B.C. (Moulherat et al. 2002), in America as

yarn around 4,750 years B.C. (Dillehay et al. 2012), and as indigo dyed woven bags

around 4,250 B.C. (Splitstoser et al. 2016).

Cotton fiber is the common name for the elongated epidermal cell on the seed

surface from some plants of the Gossypium genus with economic importance (Larkin,

Brown, and Schiefelbein 2003; Fryxell 1971). The majority of the cotton produced and

traded in the world is upland cotton (Hovav et al. 2008; USDA 2018b). A total of 77

countries harvested cotton in 2017. The three major countries for harvested areas were

India (12,300,00 ha), the United States (US (4,593,000 ha), and China (3,400,000 ha).

These three countries are also the three major cotton producers, in the following order:

India (6,210,000 metric tons), China (5,990,000 metric tons), and the United States

(4,580,000 metric tons). On the international market, the three major cotton exporters, in

2017, were the United States (3,220,000 metric tons), Australia (958,000 metric tons),

and Brazil (914,000 metric tons), while the largest importers were Bangladesh (1,610,000

metric tons), Vietnam (1,460,000 metric tons), and China (1,110,000 metric tons) (USDA

2018b).

In 2014, a total of almost 5.7 million cotton futures contracts were traded on the

Intercontinental Exchange, the global price benchmark for this commodity. In these

contracts, around 131 million metric tons were traded with notional values of $220 billion

(Janzen, Smith, and Carter 2018).

In the US, cotton was grown in 17 states in 2017. The largest cotton-growing state

was Texas, with 51.96% of the harvested area and 44.07% of the national production

(USDA 2018a). The combined counties of Crosby, Floyd, Hale, Hockley, Lamb,


2

Lubbock, Lynn, and Terry represented 59.33% of the harvested area and 61.69% of the

cotton production in Texas (NCCA 2018).

Cotton lint is a naturally variable raw material that is transformed into yarn, a

relatively uniform industrial product (P. R. Lord 2003). Cotton lint is a complex

industrial product, affected by several factors, such as plant genetics, environment, and

lint preparation. All these factors and their interactions impact the within-sample fiber

properties and these properties will affect the yarn quality. The proper measurement of

fiber quality in a sample improves the prediction of yarn quality. Therefore, cotton fiber

quality researchers must understand how the aforementioned factors may introduce error

into the reported fiber quality to minimize this error and help other cotton researchers

with proper measurements of fiber quality.

1.2 Cotton physiology

The cotton plant has complex physiology. The wild ancestors of cotton were

perennial shrubs or small trees. The cotton stalk has an indeterminate growth, with

continuous production of leaves. Two buds are formed at the base of the leaves. The

axillary bud may develop a vegetative branch, while the extra-axillary bud can create a

fruiting branch (Stewart et al. 2010; Eaton 1955). The combination of the spatial

variability within the plant canopies and the temporal variability across the growing

season create a four-dimensional intricate plant morphology (Schaefer et al. 2017;

Clement et al. 2015; Stewart et al. 2010).

When adequate resources, such as light, water, and minerals, are provided to

cotton plants, they will keep growing and developing more flowers and fruits. Some

within-plant variation in the flowering pattern may occur among different types of

germplasm. Commonly, the lowest fruiting branch appears from the 7th to the 10th node,

but there are reports of fruiting branches as low as the 3rd or 4th node. Some early

varieties flower heavily during the early season, while other varieties flower during the

whole season (Eaton 1955; Quisenberry and Roark 1976).

Cotton varieties with early physiological maturity can be labeled as “determinate”

cotton while varieties without early physiological maturity can be labeled as


3

“indeterminate”. The early physiological maturity varieties tend to have a narrower

distribution of bolls along the plant in the season, with most of the flowers developing at

the beginning of the season. Most bolls that will be harvested at the end of the season are

set on lower positions on the plant. Late maturity cotton has a larger distribution of bolls

along the plant in the season. (Bednarz and Nichols 2005; Schaefer et al. 2017). Fiber

from bolls that are set later in the season will not have enough time to pass through all the

developmental stages before harvesting, resulting in a higher amount of fibers with poor

quality (Kothari et al. 2017, 2015; Ayele, Kelly, and Hequet 2018).

1.3 Fiber development

The diploid cotton species and G. hirsutum, species with economic importance,

present a layer of short and thick elongated epidermal cells on the surface of seeds,

similar to the hairs of the wild species of this genus (Wendel and Grover 2015; Stewart et

al. 2010). The textile economic importance for cotton arises from the layer of long fibers

denominated as lint. There are four main stages in cotton fiber development that result in

the lint formation: initiation, elongation, secondary wall biosynthesis, and maturation

(Figure 1-1) (Lee, Woodward, and Chen 2007; Stewart et al. 2010).


4

Figure 1-1 Developmental stages of cotton fibers, from initiation to maturation (Lee et

al., 2007).

1.3.1 Initiation

Initiation, the first stage of fiber development, happens on the day of anthesis

when flowers open for pollination. Some cells on the ovule surface begin to swell and

acquire a rounder shape because of isodiametric diffuse growth (Stiff and Haigler 2016).

These little bumps on the ovule are named initials. About 25%-33% of epidermal cells

will be converted to initials. The initials increase their diameter to about twice the size of

the other epidermal cells in just one day. These initials will later elongate and develop

into lint fibers (Ruan 2007; Stewart 1975).

In the ovules of the G. hirsutum species, not all initials protuberate at the same

time. In some ovules, the fibers begin to differentiate at the chalazal end. In other ovules,

initials appear as clusters. In a third pattern, the modified cells are widely spaced (Tiwari

and Wilkins 1995; Stewart 1975).

1.3.2 Elongation

In elongation, the second stage of development, the initials begin to expand

mostly in one dimension, increasing their length with a small change in the width (Stiff


5

and Haigler 2016). Although cotton fiber growth is not fully understood, fiber elongation

appears to occur with a mixed growth, with elements from tip growth or diffuse growth.

In tip growth, there is a concentration of cell organelles and calcium near the tip of the

growing cell, and the microtubes are arranged in bundles parallel to the main axis of the

cell. In diffuse growth, the cell organelles and calcium are distributed around the whole

cell, and the microtubes are arranged in helical bundles to the main axis of the cell

(Ruan 2007; Qin and Zhu 2011; Seagull 1993). In the cotton fiber elongation, there is a

high amount of calcium near the tip (like in the tip growth) and an angle between the

helical bundle of microtubes and the main axis of the cell (like in the diffuse growth) at

the same time (Qin and Zhu 2011; Stiff and Haigler 2016).

The maximum rate of elongation occurs between 6 and 12 days post-anthesis

(DPA), and the cell keeps the longitudinal growth until 20-30 DPA. At this time, the ratio

between fiber length and fiber diameter can range from 1,000 to 4,000, depending on the

variety (A. Basra and Saha 2000; Stiff and Haigler 2016; Ryser 2000).

Interactions between genetics and the environment will affect the total expansion

of the cell walls. Conditions less than optimum during fiber development, like droughts,

affects the expression of genes, leading to a reduction in the final elongation (Padmalatha

et al. 2012). The sub-optimum growth may impact the textile value of the fibers,

imparting the type of yarn that can be spun.

When elongation is coming to an end, a transition or winding layer appears in the

internal surface of this structure, preparing the fiber for the secondary cell wall thickening

(Y.-L Hsieh, Honik, and Hartzell 1995; Tuttle et al. 2015).

1.3.3 Secondary cell wall thickening

The secondary cell wall thickening is the third stage in cotton fiber development.

In this phase, the cotton fiber secondary cell wall (SCW) is deposited on the primary cell

wall. The SCW is not just a thicker layer of carbohydrates. There are important chemical

and physical differences between these two components of the cell wall.


6

The cellulose content in a cotton fiber increases at the end of elongation from

around 10% until the beginning of maturation to almost 95%. This change in the

proportion of cellulose is indicative that the cell wall thickening is caused by the

deposition of microfibrils composed of almost pure cellulose (Meinert and Delmer 1977;

Abidi, Hequet, and Cabrales 2010; Abidi, Cabrales, and Haigler 2014; Stewart et al.

2010). Nevertheless, there are variations of these microfibrils throughout the whole

developmental stage. For example, in the variety TM-1, the degree of polymerization

changes from 3,400 at 17 DPA to 19,400 at 46 DPA (Liyanage and Abidi 2018).

Microfibrils in the SCW are composed of para-crystalline and crystalline

cellulose, which is the allomorph I a monoclinic crystal composed of two parallel

cellulose chains (Saito et al. 2007; Liyanage and Abidi 2018; Y.-L. Hsieh, Hu, and

Nguyen 1997). During fiber thickening, the crystallinity index and crystallite size

increase, resulting in higher fiber strength and work-to-break (Liyanage and Abidi 2018;

Y.-L Hsieh, Honik, and Hartzell 1995; Hu and Hsieh 1996; You-Lo Hsieh, Hu, and

Wang 2000).

Another structural change during the SCW thickening happens in the angle for

spiraling microfibrils. Differences in the deposition angle create reversals and the

observed fiber orientation to the main axis change from “Z” to “S” direction or vice versa

(Flint 1950; Seagull 1993; Helmut Wakeham and Spicer 1951). Reversals are brittle

regions on the fiber, with a high amount of crystalline cellulose and internal stress (H.

Wakeham, Radhakrishnan, and Viswanathan 1959; Helmut Wakeham and Spicer 1955).

They are weak spots in the fiber and where fiber breakage may happen more frequently

(Helmut Wakeham and Spicer 1951, 1955).

At the end of the third developmental stage of the fiber cotton, the cells will begin

the maturation phase.

1.3.4 Maturation

The maturation is the less understood phase in cotton fiber development. In this

stage, around 45-60 DPA, cotton fibers begin a genetic program for cell death similar to


7

the one observed for the xylem tracheary element. The cytoplasmic materials dry out,

leaving a hollow lumen and the cell wall (Kim and Triplett 2001).

During the drying process, the overall cellulose crystallinity and the number of

crystallites in the secondary cell wall are not affected. However, these cellulose

crystallites are distorted by water loss, reducing the crystallite size. The size reduction

creates mechanical stress inside the cell wall, and the fiber collapses into a flattened

ribbon with a kidney bean shape. Further water loss results in convolution formation

(Haigler et al. 2012; Hu and Hsieh 2001).

Convolutions on the fiber surfaces are associated with previously cited reversals.

Cotton can present around ten convolutions per millimeter, reaching a total between 200

and 400 convolutions per fiber. Convolution angle may affect fiber strength and

elongation, with lower angles related to higher tenacity and lower strain. Convolution

formation is irreversible and they weaken the fiber, but they are essential for spinning

cotton fiber because they increase the fiber-fiber friction, preventing slippage. (David D

Fang 2018; Jacques J. Hebert 1975; Hearle and Sparrow 1979; Flint 1950; Cook 1984).

At the end of the maturation period, the cotton boll opens and exposes the cotton

fibers to the environment. Sunlight and usually low moisture air will contribute to

finishing the drying process. Once the bolls are open, they can be harvested. Farmers

defoliate fields that will be mechanically harvested when the percentage of open bolls is

between 60% and 65% (Stewart et al. 2010; M. van der Sluijs and Long 2016).

1.4 Cotton processing

In the cotton industry, lint typically passes through many processing steps to

convert fibers into finished textile products. Some of these steps are harvesting, ginning,

yarn spinning preparation, and yarn spinning.


8

1.4.1 Harvesting

Seedcotton is removed from the plants by harvesting. The most traditional way to

perform this operation is hand-picking (Figure 1-2). This method is labor-intensive and

skilled workers can pick around 13.6-45.4 kg/day (30-100 lb/day) of seedcotton. It was

the only way to harvest cotton until the development of the mechanical harvester at the

end of the 19th century and the beginning of the 20th century. Hand-picking is still used in

some regions, such as China (J. Campbell 1991; S. E. Hughs, Valco, and Williford 2008;

Wang et al. 2016). Sometimes in cotton research, hand-picking is the only way to harvest

material for the study. A breeder will not use an implement to harvest individual plants. If

researchers cannot afford for mechanical harvesters, they generally randomly pick bolls

from many plants in the experimental plot and take their decisions based on this

sampling.

Figure 1-2 Students handpicking cotton (Zach Hinds, 2017).


9

The first commercially successful mechanical harvester was the precursor of the

current spindle picker technology. In the picker system, rotating bars with a rough surface

twist the lint from the boll through the side of the plant. Another harvesting solution is

the stripper system. These devices use rotating brushes and paddle bats to strip out the

bolls from the stalk (Figure 1-3) (Faulkner et al., 2011a; Hughs et al., 2008; Peterson &

Kislev, 1986).

Figure 1-3 Cotton harvesting with a stripper machine (Addisu Ferede, 2018).

Material cleanliness and fiber maturity are important differences between the

harvesting procedures. Manual harvesting may have a better appearance than

mechanically-harvested cotton because non-lint material, also named as trash, can be

avoided by the pickers. This trash is made of residual leaves, bracts, stems, and other

types of contaminants (M. H. J. van der Sluijs and Hunter 2017). Pickers will also


10

typically avoid partially opened bolls, minimizing the number of immature fibers in the

harvest.

Spindle-picked cotton simulates the manual harvesting, avoiding partially opened

bolls. Nevertheless, the mechanical action of the implement will bring together a higher

content of trash than hand-picking. Stripper-harvest cotton strips the plants with brushes

and bats, increasing the captured trash together with seedcotton. Furthermore, partially

open bolls are generally captured by the stripper, increasing the number of immature

fibers in the harvest (Faulkner et al. 2011a,b). The additional amount of trash typically

results in the need for more cleaning. The mechanical processing may break more fibers

and degrade fiber quality in comparison to other harvesting methods.

After harvesting, seedcotton is usually compacted into rectangular or round

modules to achieve a density of 192.2 kg/m3 (12.00 lb/ft3) before hauling the material to a

gin (Faulkner et al., 2011a; Hughs et al., 2008; Muthamilselvan, Rangasamyt,

Ananthakrishnan, & Manian, 2007; Nelson, Misra, & Brashears, 2001). At the gin, the

lint will be extracted from the cottonseeds.

1.4.2 Ginning and lint cleaning

The mechanical separation of lint and cottonseed from the seedcotton is known as

ginning. In this process, the staple fibers are broken at the basal portion, near the fiber

elbow, constricted by surrounding cells in the seed coat epidermal layer (Fryxell 1963).

Similar to harvesting, ginning can be performed manually or machine-aided. It is

believed that the first separation process was based solely on hand pulling. This process

is so tedious and labor-intensive that machines with two rollers in a calender-like

appearance, known as churkhas, were created. At the end of the 18th century, machines

with saws and brushes were developed, such as the saw-gin created by Eli Whitney,

increasing lint removal speed from about 0.5 kg/day/worker to more than 22.0

kg/day/worker (Anthony and Mayfield 1994; Bailey 1994).

The current dominant ginning technologies are roller- and saw-ginning. Roller-

gins are an evolution of the churkha machines. In the 19th century, knives were added to


11

the device, close to rollers, helping to remove the lint from the seeds and in the 20th

century, the design was improved with the addition of rotatory knives. This ginning

system is commonly used for extra-long-staple (ELS) cotton because of roller-ginning

preserves fiber quality better than saw-ginning (Anthony 2000a; Anthony and Mayfield

1994; Delhom et al. 2017).

Saw-ginning is currently the most common process to remove lint from

cottonseeds. This is the dominant technology for upland cotton ginning. Since most of the

upland cotton is machine harvested, large pieces of trash are a common contaminant that

may impede lint extraction. In order to clean this type of trash, the harvest may pass

through a cleaning system named extractor. In this system, rotating bars, saws, and

brushes revolve harvested cotton to eliminate burrs and sticks (Anthony and Mayfield

1994).

After trash extraction, the seedcotton is presented to the gin stand. Cotton fibers

are torn or broken at the end closer to the seed by saw teeth and ribs at the gin stand.

Fibers are pulled out and forced to pass through openings between the ribs, while the

cottonseeds fall into a separate chamber. Brushes remove the extracted lint from the saws

and the fibers are collected. At high-speed and super-high-speed gin stands, more than

2,000 kg of lint can be extracted per hour. If the stand is overloaded, the fiber quality may

decrease and seeds can be damaged, contaminating the lint (Vigil et al. 1996; Anthony

2000a).

The final step before pressing the cotton bale is lint cleaning, a mechanical

process in which fibers are separated from contaminants such as leaf and grass particles

that might remain after the previous steps. There are two types of lint cleaner: flow-

through air or controlled-batt saw. In the flow-through air type, cotton from the gin stand

abruptly changes trajectory in a closed curve and loose foreign matter is expelled. In the

controlled-batt type, a batt of cotton is forced by compressing cylinders, the feed rollers,

against a very closely fitted feed bar or roller. In this lint cleaner type, gravity, airflow,

and scrubbing eliminate trash from cotton. Different variations and combinations of lint

cleaner types can be used to assure increased cleanliness (Anthony 2000b; Anthony and

Mayfield 1994; Robert and Blanchard 1997).


12

There are important differences between industrial-scale and laboratory-scale

ginning. Large industrial-scale gins typically are a combination of gin stand and lint

cleaners, while laboratory-scale devices are just miniaturized gin stands (Figure 1-4). The

speed in an industrial gin is higher than in laboratory-scale gins, working in a continuous

flow. A typical industrial setting has hundreds of saws and hundreds of kilowatts of

power consumption. Laboratory-scale gins typically operate at lower speeds. They are

suitable to produce small batches of lint. The gin stand has just 10 or 20 saws and the

power consumption is smaller (Anthony and Mayfield 1994; E. Hughs, Holt, and

Rutherford 2017). These differences between industrial- and laboratory-scale gins may

affect fiber quality profiles. Higher mechanical stress is applied to fibers in the industrial-

scale gin and this could, in theory, break more fibers than in laboratory-scale gins.

Figure 1-4 Industrial-scale gin (A) with details for the gin stand (B) and lint cleaner (C)

in comparison to laboratory-scale gins (D) (Jacob James, 2017).


13

After removing and cleaning the lint in an industrial-scale gin, the fibers are

pressed in bales to simplify the transport and management of the cotton. During bale

preparation, a sample of about 4 ounces (100 g) of lint is taken from both sides of the

bale. This sample is used for cotton classification and grading of the bale based on fiber

quality (Cotton Incorporated 2013). If the cotton is extracted for research purposes, it is

typically not compressed; a subsample is taken from the ginned batch and sent to fiber

quality assessment.

1.5 Fiber quality assessment

1.5.1 History of cotton classification

In the 18th century, cotton became an important cash crop in the United States.

Cotton quality was assessed solely on the basis of fiber length and geographical location.

The quality ranged from “Benders”, the highest quality possible, to “Upland”, short-

staple types. Spinners began to notice that some types of cotton had different properties

that could increase processing efficiency. Brokers and merchants began to market cotton

with several non-uniform systems, generating economical confusion between buyers and

sellers. The U.S. Cotton Futures Act of 1914 established a classification basis, with

standards for length and grade, a combination of color, preparation, and foreign matter in

the graded cotton. Cotton classers began to assess length pulling a bundle of fibers and

estimating the bundle length to the nearest 1/32 of an inch, while grade was assessed

based on a sample from a cotton bale by visual comparison to standards created by the

U.S. Government (May 2000; May and Lege 1999; Conant Jr. 1915).

Although there was a belief that “the classification of cotton is not, and cannot be,

an exact science” (Conant Jr. 1915), there was a clear understanding that “many different

fiber properties are concerned in quality” (USDA 1934). In a study relating fiber and yarn

quality, researchers noticed discrepancies between expected and observed results for

three types of cotton with similar staple length but different grades, acknowledging that

grade was not sufficient for a proper classification (USDA 1934).

Fiber strength was first graded based on a hand break of tufts of cotton. However,

experimentation showed that this system presented small relation to actual fiber or yarn


14

strength, probably with the cotton classer judging the strength uniformity. Researchers

suggested the use of a bundle strength, with or without a twist, to grade cotton. The

authors concluded that the Chandler method, using a wrapped bundle of fibers and

reporting strength per unit of cross-sectional area, was the best available method at that

time (Dewey and Goodloe 1913; May 2000; Richardson, Bailey Jr., and Conrad 1937).

In the 1930s, fineness or hair-weight per centimeter was already an important

fiber quality measurement for silk and wool, but little research was performed with

cotton. Early measurements of cotton fineness used ribbon width, cross-section, and

optical diffraction.

Cotton maturity, or wall thickness, was recognized as a factor for luster, neps

formation, and dye uptake. Proposed methods for maturity measurement included cell

wall thickness, the ratio of ribbon width to ribbon thickness, and fiber appearance after

mercerization. (Richardson, Bailey Jr., and Conrad 1937; Peirce and Lord 1939).

In the 1940s and 1950s, new instruments were developed to increase the accuracy

and expediency of cotton fiber quality measurements, improving cotton classification.

For example, the fibrogram was developed by Hertel and Zervigon in 1936 to analyze the

fiber length distribution in bundles of clean parallel fibers, such as a sliver or a halo of

combed fibers on a seed. Later on, the method was improved with the use of the

fibrosampler, a system of combs designed to generate beards of unbiased length samples

(Chu and Riley 1997; Hertel and Zervigon 1936; Hertel 1940).

The strength measurement was improved with the Pressley machine, a device six

to eight times faster to measure strength than the Chandler machine. In the Presley

machine, small bundles or ribbons of fibers are combed, clamped, and cut before

applying a constant load with a rolling weight on an inclined plan resulting in an index of

pounds per milligram (Shepherd 1943; Pressley 1942). A further improvement for

strength measurement was the creation of the Stelometer (STrength-ELOngation-

METER), where a bundle of fibers is removed from a fibrosampled beard, combed,

clamped, and cut before applying a constant elongation rate using a pendulum (Brown

1953b).


15

During the 1950s, Lord proposed the use of airflow permeability to study cotton

fineness/maturity. Initially, this measurement was believed to be related to the average

fineness (linear density) of a cotton sample. Further studies showed that the permeability

was inversely related to the square of the specific fiber surface, a product of maturity and

fineness, resulting in the development of the micronaire scale (E. Lord 1956b, 1956a,

1955, 1981). During the development of this technique, Lord used clean samples after

passing them through a Shirley Analyzer.

In the 1960s and 1970s, devices to measure micronaire, grade, length, and

strength were combined, resulting in the creation of the High Volume Instrument (HVI)

that is used today for cotton classification.

1.5.2 High volume instrument (HVI)

In January of 1969, Motion Control assembled automatic systems for length,

length uniformity, strength, micronaire, trash, and color at Texas Tech University. The

main objective of this new machine was to have a testing cycle of 10 seconds per

specimen. Although the original expectation was to measure between 75,000 and 100,000

bales using this fast-analysis system, a total of 230,000 bales were successfully classified

in 1969. From 1975 to 1979, the Agricultural Marketing Service used the new

“instrument test” to classify cotton from field trials in Lubbock, and first used the term

“High Volume Instrument”, or HVI (Sasser and Moore 1992).

The current HVI system consists of three different parts of a whole workstation

(Figure 1-5). Micronaire is measured in the first station. The sample is a plug of fibers

with a mass of 10.0 ± 0.5 g. This plug is inserted into a chamber and air pressure is

applied while the airflow is measured, resulting in the micronaire value (Cotton

Incorporated 2013; ASTM 2011).


16

Figure 1-5 High volume instrument (HVI) (Addisu Ferede, 2019).

The second station is used to measure samples for color and trash. Enough cotton

to fully cover a glass window is placed on a tray in this station. A black and white camera

measures the number of dark pixels in a sample through a window, calculating the

amount of foreign matter. This measurement is used to determine the leaf grade, an

ordinal scale from 1 to 8. At the same time, a colorimeter measures the reflectance (Rd)

and yellowness (+b) of the sample. Generally, higher reflectance and smaller yellowness

are indicative of better fiber quality (Cotton Incorporated 2013; ASTM 2012b).


17

In the third station, a cotton sample is measured for length and tensile properties,

using about the same amount of cotton utilized for trash/color measurements. The sample

is deposited in a bin. Mechanical pressure is applied to force tufts of fibers to protrude

through holes in the bin wall. A metallic comb is passed on the protruding fibers and it

takes a length-unbiased sample forming a fiber beard, based on the fibrosampler

principle. This fiber beard is brushed to straight it up and remove loose fibers. Then, the

beard is presented to the length/strength/elongation module.

The length measurement is based on the fibrograph principle, creating a graph

named fibrogram. The fiber beard passes through a photo-electric scanner. At 0.15 in.

(3.8 mm) away from the comb, the amount of transmitted light is attenuated to 100%

because of the fibers. The comb is moved away from the scanner and fewer fibers will

attenuate the light. Fibers still contributing to optical attenuation for a given length are

classified as a percentage of span length. A span length is used to calculate the upper half

mean length (UHML) and another span length is used for the mean length (ML). The

HVI reports the UHML and the uniformity index (UI), a percentage ratio between the

mean length and the upper half mean length (Morton and Hearle 2008; ASTM 2012a).

Strength and elongation are measured using the same beard utilized for length

measurement. Based on the optical amount, the beard is clamped between two bars with a

gauge length of 1/8 inch (3.175 mm). A constant load is applied to the clamps, the force

transducer is stretched, and a stress-strain curve is measured (Naylor et al. 2014; ASTM

2012b). From this curve, elongation may be directly reported as a percentage

(McCormick et al. 2019). Strength is calculated based on the breaking force and an

estimate of fiber mass that takes into account the micronaire reading and the optical

amount. Strength is reported as a value of gram-force per tex (Y. Mogahzy and Farag

2018).

The USDA uses the HVI as the “state-of-the-art methods and equipment to

provide the cotton industry with the best possible information on cotton quality for

marketing and processing.” All the cotton bales produced in the United States are

classified using this system. (Cotton Incorporated 2013). Nevertheless, for research

purposes, other tools are also commonly used for fiber quality measurement.


18

1.5.3 Advanced fiber information system (AFIS)

The Advanced Fiber Information System (AFIS) (Figure 1-6) provides

information about fiber length; fineness; maturity; trash content and trash size; and neps

content, neps type, and neps size.

Figure 1-6 Advanced Fiber Information System (AFIS).

A sliver of cotton fibers with a mass of 0.50 g is transported on a conveyor belt by

a feed roll to a 6.35 cm rotating pinned cylindrical wheel that spins at 7,500 rpm,

resulting in a linear velocity of 89.8 kilometers per hour. This speed is similar to the

speed of the licker-in of a card in an open-end spinning rotor. This perforated cylinder

has a radially inward airflow that improves the separation of fibers, neps, and trash

(Shofner and Shofner 1999).


19

Once fibers, neps, and trash are individualized, these entities are directed by

airflow to two different paths. Dust, trash, and heavy seedcoat fragments are forced to

one channel, while fibers, neps, and light seedcoat fragments are forced to another

channel. These different materials are analyzed in two different chambers (Shofner,

Baldwin, and Chu 1993).

In the chamber for light materials, two sources of infrared light (880 nm) are

combined with sensors that are used to determine the span time of light extinction. Since

the machine contains a sensor for downstream airflow velocity, it is possible to calculate

the length of the entity. Another sensor detects the scattered forward infrared light at an

angle of 40°, determining the volume of the entity. If this entity is a fiber, the information

is used to calculate maturity and fineness. If this entity is not a fiber, the information is

used to classify the entity as a nep or seedcoat fragment and its size (Shofner et al. 1995;

Shofner, Baldwin, and Chu 1993).

In the chamber for heavy materials, the light extinction and light scattering

sensors are used to measure the size of the entities. These entities are classified as trash,

neps, and seedcoat fragments based on the signal peak value, waveform above a

threshold, and duration of this waveform above a threshold (Shofner, Baldwin, and Chu

1993).

1.5.4 Trash analyzer instruments

Trash is the non-lint material that contaminates cotton lint. The term can be used

for undeveloped seeds, seedcoat fragments, particles of leaf, dust, sand, and stems

(ASTM 2013a). When cotton harvest transitioned from a labor-intensive hand-picking

system to a mechanical harvest approach, the mass of cotton picked per unit time

increased. Unfortunately, the amount of trash that was captured together with the harvest

also increased. The presence of trash reduces the processing performance, yarn yield, and

final quality of fabrics. Since the U.S. Cotton Futures Act, the foreign matter became a

component of the cotton grade as an important parameter for marketing cotton and to

understand cotton quality (S. E. Hughs, Valco, and Williford 2008; USDA 1934; May

2000).


20

Although information about trash in cotton can be acquired with HVI and AFIS,

these devices present some limitations. The HVI measures trash based solely on black

pixels detected on the glass window in the second workstation of the machine during the

analysis. This instrument cannot estimate the type of trash, nor detect trash particles

inside a sample. The AFIS separates trash particles from the cotton lint and analyzes all

the trash within a sample, but since the sample mass is low (0.50 g) and trash is not

uniformly distributed within-sample, the reported trash amount may not be representative

of the whole cotton sample (David D Fang 2018; Xu et al. 1997). Other devices were

created specifically to analyze trash in cotton samples and provide more information

about this issue for fiber quality assessment.

1.5.4.1 Shirley analyzer

The Shirley Analyzer is a machine that was created by the Shirley Institute,

currently British Textile Technology Group, to quantitatively measure trash content. The

Shirley Analyzer is used in the ASTM procedure D2812 for trash quantification in raw

cotton; partially processed cotton, such as a sliver; and even processing waste, such as

card waste (ASTM 2012d).

The classic Shirley machine (Figure 1-7) is an aeromechanical separator that

contains a feeding table, wired feeding roll, lickerin similar to a card, air baffles, fan, and

fiber condenser. Tufts of around 100 g of cotton are broken on the lickerin, and fibers are

carried by the airflow to the baffle and condenser. Since trash has a higher density than

fibers, it is dropped onto a tray, where it can be retrieved and weighed (Pfeiffenberger

1944; ASTM 2012d). Some fibers can also be dropped together with the trash.


21

Figure 1-7 Shirley analyzer from Shirley Institute.

The standard methodology to quantify trash in a cotton sample consists of passing

the same sample twice through the Shirley Analyzer (ASTM 2012d). One passage can

also be used, depending on the amount of tested lint, available time, and if the research

aims to minimize damage to the fibers (Shepherd 1961).

The Shirley analyzer provides a good estimate of the total amount of trash in a

cotton sample. However, the only qualitative information provided by the machine is the

amount of visible trash, heavy particles, and fragments of short fibers carried away by the

airflow, named invisible trash (Shepherd 1961).

An evolution of the Shirley analyzer is the lint and trash analyzer MK2, currently

manufactured by the company SDL Atlas. This instrument is smaller than the old model

and part of the invisible trash can be captured and weighted.


22

1.5.4.2 Microdust and trash analyzer (MDTA)

The Microdust and Trash Analyzer (MDTA) is another aeromechanical trash

separator (Figure 1-8). This machine can be used to analyze raw cotton or slivers. It

separates trash, dust, and fragments from a cotton sample. The manufacturer recommends

passing the sample twice through the cleaning devices to estimate the amount of trash and

propensity to clean the sample (Suessen 2008b; Weinans 2007).

Figure 1-8 Microdust and trash analyzer (MDTA).

The cotton sample with a mass between 10 and 20 g is placed on a feed conveyor

towards a feed roller and then to an opener roller. The test is performed when the opener

roller reaches the proper speed, the fiber channel reaches a vacuum of 3.5 mbar, and the

dust channel reaches a vacuum of 2.5 mbar. The sample is split into four parts: fibers,

trash, dust, and fiber fragments. A trash knife at 1.5 mm (0.06 in.) from the opener roller

will remove the trash. Fibers will be conducted to the fiber channel, and dust and


23

fragments will be conducted to the dust channel. One filter will capture dust and another

filter will capture fiber fragments (Suessen 2008b, 2008a; Weinans 2007).

1.6 Final remarks

Low-profit margins in spinning mills force spinners to carefully choose the raw

material that they purchase to fulfill their contract, which is typically not the best and

most expensive cotton available in the market (Mitra and Adhikary 2017). The type and

quality specification of a given yarn is defined by the demanded end-product and

available spinning technology. Spinners choose the raw material to manufacture yarn

with a given quality based on the available fiber quality for a cotton bale (Y. E. El

Mogahzy and Broughton 1992; Yang and Gordon 2017). Therefore, the fiber quality

analysis is of the utmost importance for spinners and they need trustworthy information

to purchase their bales.

Cotton breeders develop new varieties with improved fiber yield and quality to

fulfill the demand from spinners (B. T. Campbell et al. 2018; C. M. Kelly, Hequet, and

Dever 2013; Y. E. El Mogahzy, Broughton, and Lynch 1990). Breeders need precise and

accurate information about the fiber quality profile of each sample to properly select

materials that may result in future varieties. Unfortunately, some factors may reduce the

certainty of fiber quality analysis, resulting in errors that may make breeders keep entries

with poor performance or discard entries with good performance.

The capacity to successfully assess fiber quality profile is essential for cotton

research and industry. If cotton breeders can obtain better information about their

germplasm, the benefits will permeate throughout the cotton marketing. For example,

accurate information about fiber quality from breeders will help farmers to choose

varieties to plant and plan their expected revenues. Ginners must know the expected fiber

quality profile from the harvested crops to properly calibrate their machines and

minimize fiber damage during operation. Merchants need accurate information about

each bale to successfully market the cotton, linking producers and buyers. Spinners need

reliable data to blend cotton bales and fulfill their contracts for a given yarn type and end

product.


24

The objective of this dissertation is to present results related to techniques that can

improve the fiber quality characterization of cotton samples. Cotton contamination and

processing are discussed as factors that can impact the assessment of the true fiber quality

profile. The understanding of how these factors interfere with the fiber quality

measurements is necessary to allow researchers to isolate, eliminate, or minimize them,

avoiding errors and wrong conclusions, helping the progress of the cotton industry.


25

1.7 Bibliography

Abidi, Noureddine, Luis Cabrales, and Candace H. Haigler. “Changes in the Cell Wall

and Cellulose Content of Developing Cotton Fibers Investigated by FTIR

Spectroscopy.” Carbohydrate Polymers 100 (2014): 9–16.

http://dx.doi.org/10.1016/j.carbpol.2013.01.074.

Abidi, Noureddine, Eric Hequet, and Luis Cabrales. “Changes in Sugar Composition and

Cellulose Content during the Secondary Cell Wall Biogenesis in Cotton Fibers.”

Cellulose 17, no. 1 (February 21, 2010): 153–160. Accessed January 18, 2019.

http://link.springer.com/10.1007/s10570-009-9364-3.

Anthony, W.S. “Methods to Reduce Lint Cleaner Waste and Damage.” Transactions of

the ASAE 43, no. 2 (2000): 221–229.

———. “Postharvest Management of Fiber Quality.” In Cotton Fibers: Developmental

Biology, Quality Improvement, and Textile Processing, edited by A.S. Basra, 293–

338. 1st ed. Binghamton: The Haworth Press, 2000.

Anthony, W S, and William D Mayfield. Cotton Ginners Handbook. Springfield: United

States Department of Agriculture, 1994.

ASTM. “ASTM D1447-07 ‘Standard Test Method for Length and Length Uniformity of

Cotton Fibers by Photoelectric Measurement.’” 5. West Conshohocken: ASTM

International, 2012.

———. “ASTM D1448 ‘Standard Test Method for Micronaire Reading of Cotton

Fibers.’” 3. West Conshohocken: ASTM International, 2011.

———. “ASTM D5867-12 ‘Standard Test Methods for Measurement of Physical

Properties of Cotton Fibers by High Volume Instruments.’” 1–8. West

Conshohocken: ASTM International, 2012.

———. “ASTM Standard D123-13a, ‘Standard Terminology Relating to Textiles.’” In

ASTM, 1–68. West Conshohocken, 2013.

http://enterprise2.astm.org/DOWNLOAD/D123.389810-1.pdf.

———. “D1440-07 Standard Test Method for Length and Length Distribution of Cotton

Fibers (Array Method).” In Annual Book of ASTM Standards, 1–6. West


———. “D2812-07 Standard Test Method for Non-Lint Content of Cotton.” In Annual

Book of ASTM Standards, 5. West Conshohocken: ASTM International, 2012.

Ayele, Addissu G., Brendan R. Kelly, and Eric F. Hequet. “Evaluating Within-Plant

Variability of Cotton Fiber Length and Maturity.” Agronomy Journal 110, no. 1

(2018): 47–55.

https://dl.sciencesocieties.org/publications/aj/abstracts/0/0/agronj2017.06.0359.

Bailey, Ronald. “The Other Side of Slavery: Black Labor, Cotton, and Textile

Industrialization in Great Britain and the United States.” Agricultural History 68, no.

2 (1994): 35–50.

Basra, AS, and S Saha. “Growth Regulation of Cotton Fibers.” In Cotton Fibers:

Developmental Biology, Quality Improvement, and Textile Processing2, edited by


26

AS Basra, 47–64. 1st ed. Binghamton: The Haworth Press Inc., 2000.

Bednarz, Craig W., and Robert L. Nichols. “Phenological and Morphological

Components of Cotton Crop Maturity.” Crop Science 45, no. 4 (2005): 1497.

Accessed July 25, 2018. https://www.crops.org/publications/cs/abstracts/45/4/1497.

Brown, Hugh M. “Fiber Strength and Extensibility.” In The Cotton Research Clinic,

edited by National Cotton Council of America, 25-. Memphis, 1953.

Campbell, B.T., J. K. Dever, K. L. Hugie, and C. M. Kelly. “Cotton Fiber Improvement

through Breeding and Biotechnology.” In Cotton Fiber: Physics, Chemistry and

Biology, edited by D. Fang, 193–215. Cham: Springer International Publishing,

2018.

Campbell, John. “As ‘a Kind of Freeman’?: Slaves’ Market‐related Activities in the

South Carolina Upcountry, 1800–1860.” Slavery & Abolition 12, no. 1 (May 1991):

131–169. http://www.tandfonline.com/doi/abs/10.1080/01440399108575026.

Chu, Youe-Tsyr, and C. Roger Riley. “New Interpretation of the Fibrogram.” Textile

Research Journal 67, no. 12 (December 1997): 897–901.

http://journals.sagepub.com/doi/10.1177/004051759706701206.

Clement, J.D., G.A. Constable, W.N. Stiller, and S.M. Liu. “Early Generation Selection

Strategies for Breeding Better Combinations of Cotton Yield and Fibre Quality.”

Field Crops Research 172 (2015): 145–152.

Conant Jr., Luther. “The United States Cotton Futures Act.” The American Economic

Review 5, no. 1 (1915): 1–11. https://www.jstor.org/stable/71.

Cook, J Gordon. Handbook of Textile Fibres: Natural Fibres. 1st ed. Sawston:

Woodhead Publishing, 1984.

Cotton Incorporated. The Classification of Cotton. Edited by Cotton Incorporated. 1st ed.

Cary: Incorporated, Cotton, 2013.

http://www.cottoninc.com/fiber/quality/Classification-Of-Cotton/Classing-

booklet.pdf.

Delhom, Christopher D, Carlos B Armijo, S Ed Hughs, and C D Delhom. “High Quality

Yarns Produced via High-Speed Roller Ginning of Upland Cotton.” The Journal of

Cotton Science 21 (2017): 81–93.

Dewey, Lyester H., and Marie Goodloe. “The Strength of Textile Fibers.” In

Miscellaneous Papers, Circular 1:17–21. 1st ed. Washington, DC: U.S. Department

of Agriculture, Bureau of Plant Industry, 1913.

https://archive.org/details/miscellaneouspap128unit.

Dillehay, Tom D., Duccio Bonavia, Steven Goodbred, Mario Pino, Victor Vasquez,

Teresa Rosales Tham, William Conklin, et al. “Chronology, Mound-Building and

Environment at Huaca Prieta, Coastal Peru, from 13 700 to 4000 Years Ago.”

Antiquity 86, no. 331 (March 2, 2012): 48–70. Accessed February 7, 2019.

https://www.cambridge.org/core/product/identifier/S0003598X00062451/type/journ

al_article.

Eaton, F M. “Physiology of the Cotton Plant.” Annual Review of Plant Physiology 6, no.


27

1 (1955): 299–328.

http://www.annualreviews.org/doi/abs/10.1146/annurev.pp.06.060155.001503.

Fang, David D. Cotton Fiber: Physics, Chemistry and Biology. Edited by David D. Fang.

Cham: Springer International Publishing, 2018.

http://link.springer.com/10.1007/978-3-030-00871-0.

Faulkner, W. B., J. D. Wanjura, E. F. Hequet, R. K. Boman, B. W. Shaw, and C. B.

Parnell Jr. “Evaluation of Modern Cotton Harvest Systems on Irrigated Cotton:

Fiber Quality.” Applied Engineering in Agriculture 27, no. 4 (2011): 507–513.

Accessed May 1, 2018. https://naldc.nal.usda.gov/download/50027/PDF.

———. “Evaluation of Modern Cotton Harvest Systems on Irrigated Cotton: Harvester

Performance.” Applied Engineering in Agriculture 27, no. 4 (2011): 497–506.

Flint, E. A. “The Structure and Development of the Cotton Fibre.” Biological Reviews 25,

no. 4 (1950): 414–434.

Fryxell, Paul A. “Morphology of the Base of Seed Hairs of Gossypium I. Gross

Morphology.” Botanical Gazette 124, no. 3 (March 1963): 196–199. Accessed April

11, 2018. https://www.journals.uchicago.edu/doi/10.1086/336192.

———. “Phenetic Analysis and the Phylogeny of the Diploid Species of Gossypium L.

(Malvaceae).” Evolution 25, no. 3 (September 1971): 554–562.

http://doi.wiley.com/10.1111/j.1558-5646.1971.tb01916.x.

Haigler, Candace H, Lissete Betancur, Michael R Stiff, John R Tuttle, Seth Debolt, and

Ying Gu. “Cotton Fiber: A Powerful Single-Cell Model for Cell Wall and Cellulose

Research” 18, no. 18 (2012): 35–1. Accessed March 31, 2018.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3356883/pdf/fpls-03-00104.pdf.

Hearle, J. W. S., and J. T. Sparrow. “Mechanics of the Extension of Cotton Fibers. II.

Theoretical Modeling.” Journal of Applied Polymer Science 24, no. 8 (October 15,

1979): 1857–1874. http://doi.wiley.com/10.1002/app.1979.070240807.

Hebert, Jacques J. “Effect of Convolution Angle upon Cotton Fiber Strength.” Textile

Research Journal 45, no. 4 (1975): 356–357.


Hertel, K.L. “A Method of Fibre-Length Analysis Using the Fibrograph.” Textile

Research 10, no. 12 (October 8, 1940): 510–520.


Hertel, K.L., and M.G. Zervigon. “An Optical Method for the Length Analysis of Cotton

Fibres.” Textile Research 6, no. 7 (May 7, 1936): 331–339.


Hovav, Ran, Joshua A. Udall, Bhupendra Chaudhary, Einat Hovav, Lex Flagel, Guanjing

Hu, and Jonathan F. Wendel. “The Evolution of Spinnable Cotton Fiber Entailed

Prolonged Development and a Novel Metabolism.” PLoS Genetics 4, no. 2 (2008):

e25. Accessed March 24, 2018. http://dx.plos.org/10.1371/journal.pgen.0040025.

Hsieh, You-Lo, Xiao-Ping Hu, and Anjia Wang. “Single Fiber Strength Variations of

Developing Cotton Fibers—Strength and Structure of G. Hirsutum and G.


28

Barbedense.” Textile Research Journal 70, no. 8 (August 2, 2000): 682–690.

Accessed April 27, 2018.


Hsieh, You.-Lo., X.-P. Hu, and A. Nguyen. “Strength and Crystalline Structure of

Developing Acala Cotton.” Textile Research Journal 67, no. 7 (July 1, 1997): 529–

536. Accessed April 16, 2018.


Hsieh, You.-Lo, E Honik, and M M Hartzell. “A Developmental Study of Single Fiber

Strength: Greenhouse Grown SJ-2 Acala Cotton.” Textile Research Journal 65, no.

2 (1995): 101–112.

Hu, Xiao-Ping, and You-Lo Hsieh. “Crystalline Structure of Developing Cotton Fibers.”

Journal of Polymer Science Part B: Polymer Physics 34, no. 8 (June 1, 1996):

1451–1459. Accessed April 23, 2018.

http://doi.wiley.com/10.1002/%28SICI%291099-

0488%28199606%2934%3A8%3C1451%3A%3AAID-POLB8%3E3.0.CO%3B2-

V.

———. “Effects of Dehydration on the Crystalline Structure and Strength of Developing

Cotton Fibers.” Textile Research Journal 71, no. 3 (March 2, 2001): 231–239.

Accessed April 16, 2018.


Hughs, Ed, Greg Holt, and Ross Rutherford. “Saw Gin Stands.” The Journal of Cotton

Science 21 (2017): 60–69.

Hughs, S. E., T. D. Valco, and J. R. Williford. “100 Years of Cotton Production,

Harvesting, and Ginning Systems Engineering: 1907-2007.” Transactions of the

ASABE 51, no. 4 (2008): 1187–1198.

http://elibrary.asabe.org/abstract.asp??JID=3&AID=25234&CID=t2008&v=51&i=4

&T=1.

Janzen, Joseph P, Aaron Smith, and Colin A Carter. “Commodity Price Comovement and

Financial Speculation: The Case of Cotton.” American Journal of Agricultural

Economics 100, no. 1 (January 1, 2018): 264–285. Accessed March 24, 2018.

http://academic.oup.com/ajae/article/100/1/264/4283312.

Kelly, Carol M., Eric F. Hequet, and Jane K. Dever. “Breeding for Improved Yarn

Quality: Modifying Fiber Length Distribution.” Industrial Crops and Products 42

(March 2013): 386–396. Accessed December 8, 2016.

https://linkinghub.elsevier.com/retrieve/pii/S0926669012003408.

Kim, Hee Jin, and Barbara A Triplett. “Cotton Fiber Growth in Planta and in Vitro.

Models for Plant Cell Elongation and Cell Wall Biogenesis.” PLANT PHYSIOLOGY

127, no. 4 (December 1, 2001): 1361–1366.

http://www.plantphysiol.org/cgi/doi/10.1104/pp.010724.

Kothari, Neha, Jane Dever, Steve Hague, and Eric Hequet. “Evaluating Intraplant Cotton

Fiber Variability.” Crop Science 55, no. 2 (2015): 564. Accessed July 4, 2016.

https://dl.sciencesocieties.org/publications/cs/abstracts/55/2/564.


29

Kothari, Neha, Steve Hague, Lori Hinze, and Jane Dever. “Boll Sampling Protocols and

Their Impact on Measurements of Cotton Fiber Quality.” Industrial Crops and

Products 109, no. June (2017): 248–254.

http://dx.doi.org/10.1016/j.indcrop.2017.07.045.

Larkin, John C., Matt L. Brown, and John Schiefelbein. “How Do Cells Know What

They Want to Be When They Grow up? Lessons from Epidermal Patterning in

Arabidopsis.” Annual Review of Plant Biology 54, no. 1 (June 28, 2003): 403–430.

Accessed February 8, 2019.

http://www.annualreviews.org/doi/10.1146/annurev.arplant.54.031902.134823.

Lee, J. J., A. W. Woodward, and Z. J. Chen. “Gene Expression Changes and Early Events

in Cotton Fibre Development.” Annals of Botany 100, no. 7 (September 19, 2007):

1391–1401. Accessed October 19, 2018. https://academic.oup.com/aob/article-

lookup/doi/10.1093/aob/mcm232.

Liyanage, Sumedha, and Noureddine Abidi. “Molecular Weight and Organization of

Cellulose at Different Stages of Cotton Fiber Development.” Textile Research

Journal (January 23, 2018): 004051751775364. Accessed June 25, 2018.


Lord, E. “10—Air Flow through Plugs of Textile Fibres Part I—General Flow

Relations.” Journal of the Textile Institute Transactions 46, no. 3 (March 1955):

T191–T213.

http://www.tandfonline.com/doi/abs/10.1080/19447027.1955.10750307.

———. “2—Air Flow through Plugs of Textile Fibres: Part II. the Micronaire Test for

Cotton.” Journal of the Textile Institute Transactions 47, no. 1 (1956): T16–T47.

———. “51—Air Flow through Plugs of Textile Fibres Part III—The Causticaire Test

for Cotton.” Journal of the Textile Institute Transactions 47, no. 12 (December

1956): T635–T649.


———. The Origin and Assesment of Cotton Fibre Maturity. Washington, DC:

International Institute for Cotton, 1981.

Lord, P. R. Handbook of Yarn Production. Edited by P.R. Lord. 1st ed. Cambridge:


May, O. Lloyd. “Genetic Variation in Fiber Quality.” In Cotton Fibers: Developmental

Biology, Quality Improvement, and Textile Processing, edited by AS Basra, 183–

230. 1st ed. Binghamton: Haworth Press, 2000.

May, O. Lloyd, and K. E. Lege. “Development of the World Cotton Industry.” In Cotton:

Origin, History, Technology, and Production, edited by C. Wayne Smith and Joe

Tom Cothren, 65–97. New York: John Wiley & Sons, Ltd, 1999.

McCormick, Kolby M, João Paulo Saraiva Morais, Eric Hequet, and Brendan Kelly.

“Development of the Correction Procedure for High Volume Instrument Elongation

Measurement.” Textile Research Journal (February 11, 2019): 004051751982900.

Accessed May 16, 2019.



30

Meinert, Maureen C, and Deborah P Delmer. “Changes in Biochemical Composition of

the Cell Wall of the Cotton Fiber during Development.” Plant Physiology 59, no. 6

(June 1, 1977): 1088–1097. Accessed January 18, 2019.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC542513/pdf/plntphys00140-

0079.pdf.

Mitra, R. K., and B. K. Adhikary. “Determinants of Financial Performance: Empirical

Evidence from the Textile Sector in Bangladesh.” Journal of Accounting and

Finance 17, no. 8 (2017): 110–120.

http://eds.a.ebscohost.com/ehost/detail/detail?vid=0&sid=11d40ed8-3de5-4ffe-93ec-

cc60a0ab7bce%40sessionmgr4006&bdata=JnNpdGU9ZWhvc3QtbGl2ZQ%3D%3

D#AN=128937315&db=bth.

El Mogahzy, Yehia E., Roy Broughton, and W.K. Lynch. “A Statistical Approach for

Determining the Technological Value of Cotton Using HVI Fiber Properties.”

Textile Research Journal 60, no. 9 (September 2, 1990): 495–500.


El Mogahzy, Yehia E., and Roy M. Broughton. “‘Regressional’ Observations of HVI

Fiber Properties, Yarn Quality, and Processing Performance of Medium Staple

Cotton.” Textile Research Journal 62, no. 4 (April 1992): 218–226.


Mogahzy, Yehia, and Ramsis Farag. “7 - Tensile Properties of Cotton Fibers: Importance,

Research, and Limitations.” In Handbook of Properties of Textile and Technical

Fibres, edited by Anthony R. Bunsell, 223–273. 2nd ed. Cambridge: Woodhead

Publishing, 2018.

http://www.sciencedirect.com/science/article/pii/B9780081012727000079.

Morton, W.E., and J.W.S. Hearle. “Fibre Length.” In Physical Properties of Textile

Fibres, 134–162. 4th ed. Sawston: Elsevier, 2008.

https://linkinghub.elsevier.com/retrieve/pii/B9781845692209500044.

Moulherat, Christophe, Margareta Tengberg, Jérôme-F. Haquet, and Benoı̂t Mille. “First

Evidence of Cotton at Neolithic Mehrgarh, Pakistan: Analysis of Mineralized Fibres

from a Copper Bead.” Journal of Archaeological Science 29, no. 12 (December 1,

2002): 1393–1401. Accessed September 19, 2018.

https://www.sciencedirect.com/science/article/pii/S0305440301907794.

Muthamilselvan, M, K Rangasamyt, D Ananthakrishnan, and R Manian. “Mechanical

Picking of Cotton - A Review.” Agrk.Rev 28, no. 2 (2007): 118–126.

Naylor, Geoffrey RS, Christopher D Delhom, Xiaoliang Cui, Jean-Paul Gourlot, and

James Rodgers. “Understanding the Influence of Fiber Length on the High Volume

InstrumentTM Measurement of Cotton Fiber Strength.” Textile Research Journal 84,

no. 9 (June 16, 2014): 979–988. Accessed October 29, 2018.


NCCA. “Top County’s Production - Texas Upland Cotton.” National Cotton Council of

America. Last modified 2018. http://www.cotton.org/econ/cropinfo/cropdata/county-

db.cfm.


31

Nelson, J. M., S. K. Misra, and A. D. Brashears. “Cost Comparison of Alternative

Stripper and Picker Cotton Harvesting Systems.” Applied Engineering in

Agriculture (2001).

Padmalatha, Kethireddy Venkata, Gurusamy Dhandapani, Mogilicherla Kanakachari,

Saravanan Kumar, Abhishek Dass, Deepak Prabhakar Patil, Vijayalakshmi

Rajamani, et al. “Genome-Wide Transcriptomic Analysis of Cotton under Drought

Stress Reveal Significant down-Regulation of Genes and Pathways Involved in

Fibre Elongation and up-Regulation of Defense Responsive Genes.” Plant

Molecular Biology 78, no. 3 (February 7, 2012): 223–246. Accessed October 19,

2018. http://link.springer.com/10.1007/s11103-011-9857-y.

Peirce, F. T., and E. Lord. “13—The Fineness and Maturity of Cotton.” Journal of the

Textile Institute Transactions 30, no. 12 (December 1939): T173–T210.

http://www.tandfonline.com/doi/abs/10.1080/19447023908659334.

Peterson, Willis, and Yoav Kislev. “The Cotton Harvester in Retrospect: Labor

Displacement or Replacement?” The Journal of Economic History (1986).

Pfeiffenberger, George W. “The Shirley Analyzer.” Textile Research 14, no. 2 (February

1944): 50–54. http://journals.sagepub.com/doi/10.1177/004051754401400206.

Pressley, E. H. “A Cotton Fiber Strength Tester.” In ASTM Bullertin 118, 13–17.

Northampton: ASTM, 1942.

Qin, Yong-Mei, and Yu-Xian Zhu. “How Cotton Fibers Elongate: A Tale of Linear Cell-

Growth Mode.” Current Opinion in Plant Biology 14, no. 1 (February 1, 2011):

106–111. Accessed October 16, 2018.

https://www.sciencedirect.com/science/article/pii/S1369526610001299.

Quisenberry, E, and Bruce Roark. “Influence of Indeterminate Growth Habit on Yield

and Irrigation Water-Use Efficiency in Upland Cotton.” Crop Science 16, no. 6

(1976): 762–765.

Richardson, Howard B, T. L. W. Bailey Jr., and Carl M Conrad. Methods for the

Measurement of Certain Character Properties of Raw Cotton - Technical Bulletin

545. 1st ed. Washington, DC: USDA, 1937.

https://naldc.nal.usda.gov/download/CAT86200540/PDF.

Robert, Kearny Q., and Lawrence J. Blanchard. “Cotton Cleanability: Part I: Modeling

Fiber Breakage.” Textile Research Journal 67, no. 6 (June 1997): 417–427.


Ruan, Yong-Ling. “Rapid Cell Expansion and Cellulose Synthesis Regulated by

Plasmodesmata and Sugar: Insights from the Single-Celled Cotton Fibre.”

Functional Plant Biology 34, no. 1 (February 9, 2007): 1–10. Accessed October 19,

2018. http://www.publish.csiro.au/?paper=FP06234.

Ryser, U. “Cotton Fiber Initiation and Histodifferentiation.” In Cotton Fibers:

Developmental Biology, Quality Improvement, and Textile Processing, edited by AS

Basra, 1–46. 1st ed. Binghamton: The Haworth Press Inc., 2000.

Saito, Tsuguyuki, Satoshi Kimura, Yoshiharu Nishiyama, and Akira Isogai. “Cellulose


32

Nanofibers Prepared by TEMPO-Mediated Oxidation of Native Cellulose.”

Biomacromolecules 8, no. 8 (August 2007): 2485–2491. Accessed November 2,

2018. http://pubs.acs.org/doi/abs/10.1021/bm0703970.

Sasser, P. E., and J.F. Moore. “A Historical Perspective of High Volume Instrument

Developments in the U. S.” In Proceedings of the 21st International Cotton

Conference Bremen, edited by H Harig and S. A Heap, 21:21–25. Bremen:

Faserinstitut Bremen, 1992.

Schaefer, Curtis, Bob Nichols, Guy Collins, Jared Whitaker, Craig Bednarz, Chris Main,

and Glen Ritchie. “Cotton Maturity Determination through Vertical Mapping.” Crop

Science 57, no. 1 (August 2, 2017): 62. Accessed July 24, 2018.

https://dl.sciencesocieties.org/publications/cs/abstracts/57/1/62.

Seagull, Robert W. “Cytoskeletal Involvement in Cotton Fiber Growth and

Development.” Micron 24, no. 6 (January 1, 1993): 643–660. Accessed November

30, 2018.

https://www.sciencedirect.com/science/article/pii/096843289390042Y?via%3Dihub.

Shepherd, Jacob V. “Cotton Fiber Strength Study.” Textile Research 13, no. 9 (July 8,


Shofner, Frederick M., Joseph C. Baldwin, and Youe-Tsyr Chu. “Electro-Optical

Methods and Apparatus for High Speed, Multivariate Measurement of Individual

Entities in Fiber or Other Samples.” USA, 1993.

Shofner, Frederick M., Joseph C. Baldwin, Michael E. Galyon, and Youe-Tsyr Chu.

“Apparatus and Methods for Measurement and Classification of Generalized

Neplike Entities in Fiber Samples.” USA, 1995.

Shofner, Frederick M., and Christopher K. Shofner. “Aeromechanical Individualizer.”

USA, 1999.

van der Sluijs, M.H.J., and L. Hunter. “Cotton Contamination.” Textile Progress 49, no. 3

(July 3, 2017): 137–171.

https://www.tandfonline.com/doi/full/10.1080/00405167.2018.1437008.

van der Sluijs, M.H.J., and R.L. Long. “The Effect of Seed Cotton Moisture during

Harvesting on – Part 1 – Fiber Quality.” Textile Research Journal 86, no. 18

(November 22, 2016): 1925–1936. Accessed December 6, 2017.


Splitstoser, Jeffrey C., Tom D. Dillehay, Jan Wouters, and Ana Claro. “Early Pre-

Hispanic Use of Indigo Blue in Peru.” Science Advances 2, no. 9 (September 1,

2016): e1501623. Accessed February 7, 2019.

http://advances.sciencemag.org/lookup/doi/10.1126/sciadv.1501623.

Stewart, James McD. “Fiber Initiation on the Cotton Ovule (Gossypium Hirsutum).”

American Journal of Botany 62, no. 7 (August 1975): 723. Accessed October 19,

2018. http://doi.wiley.com/10.1002/j.1537-2197.1975.tb14105.x.

Stewart, James McD., Derrick M. Oosterhuis, James J. Heitholt, and Jackson R. Mauney.

Physiology of Cotton. Edited by James McD. Stewart, Derrick M. Oosterhuis, James


33

J. Heitholt, and Jackson R. Mauney. 1st ed. Dordrecht: Springer Netherlands, 2010.


Stiff , Michael R., and Candace H. Haigler. “Cotton Fiber Tips Have Diverse

Morphologies and Show Evidence of Apical Cell Wall Synthesis.” Scientific

Reports 6, no. 1 (September 15, 2016): 27883. Accessed October 19, 2018.

http://www.nature.com/articles/srep27883.

Suessen. “Original Instructions MDTA 3 Microdust-Trash Analyser - Maintenance.” In

Original Instructions MDTA 3 Microdust-Trash Analyser, 22. Suessen: Suessen,

2008.

———. “Original Instructions MDTA 3 Microdust-Trash Analyser - Operating

Instructions.” In Original Instructions MDTA 3 Microdust-Trash Analyser, 14.

Suessen: Suessen, 2008.

Tiwari, Suresh C., and Thea A. Wilkins. “Cotton ( Gossypium Hirsutum ) Seed

Trichomes Expand via Diffuse Growing Mechanism.” Canadian Journal of Botany

73, no. 5 (May 1995): 746–757. http://www.nrcresearchpress.com/doi/10.1139/b95-

081.

Tuttle, John R., Gyoungju Nah, Mary V. Duke, Danny C. Alexander, Xueying Guan,

Qingxin Song, Z. Jeffrey Chen, Brian E. Scheffler, and Candace H. Haigler.

“Metabolomic and Transcriptomic Insights into How Cotton Fiber Transitions to

Secondary Wall Synthesis, Represses Lignification, and Prolongs Elongation.” BMC

Genomics 16, no. 1 (December 27, 2015): 477. Accessed April 23, 2018.

http://www.biomedcentral.com/1471-2164/16/477.

USDA. A Study of the Raw Cotton and the Yarn and Sheeting Manufactured from Three

Grades of American Upland Cotton. Technical Bulletin 406. 1st ed. Washington,

DC: USDA, 1934.

———. “Cotton and Wool Yearbook.” Last modified 2018.

https://www.ers.usda.gov/data-products/cotton-wool-and-textile-data/cotton-and-

wool-yearbook/.

———. “Production, Supply and Distribution.” Last modified 2018. Accessed March 10,

2018. https://apps.fas.usda.gov/psdonline/app/index.html#/app/home.

Vigil, Eugene L, W Stanley Anthony, Eugene Columbus, Eric Erbe, and William P

Wergin. “Fine Structural Aspects of Cotton Fiber Attachment to the Seed Coat:

Morphological Factors Affecting Saw Ginning of Lint Cotton.” International

Journal of Plant Sciences 157, no. 1571 (1996): 92–102.

http://www.jstor.org/stable/2474994.

Wakeham, H., T. Radhakrishnan, and G.S. Viswanathan. “X-Ray Diffraction Patterns of

Cotton at and between Reversals.” Textile Research Journal 29, no. 5 (May 2,


Wakeham, Helmut, and Nancy Spicer. “The Strength and Weakness of Cotton Fibers.”

Textile Research Journal 21, no. 4 (April 2, 1951): 187–194.



34

———. “The Strength and Weakness of Cotton Fibers Part II: Reversal Distribution and

Breaking Properties.” Textile Research Journal 25, no. 7 (July 2, 1955): 585–591.


Wang, Xiangru, Yuru Hou, Mingwei Du, Dongyong Xu, Huaiyu Lu, Xiaoli Tian, and

Zhaohu Li. “Effect of Planting Date and Plant Density on Cotton Traits as Relating

to Mechanical Harvesting in the Yellow River Valley Region of China.” Field

Crops Research 198 (November 1, 2016): 112–121. Accessed April 5, 2018.

https://www.sciencedirect.com/science/article/pii/S037842901630315X.

Weinans, Anja. “MDTA 3 - Microdust and Trash Analyser for the Cost Effective

Selection and Combination of Raw Material.” Spinnovation. Süßen, 2007.

Wendel, Jonathan F., and Corrinne E. Grover. “Taxonomy and Evolution of the Cotton

Genus, Gossypium.” In Cotton, edited by D.D. Fang and R.G. Percy, 25–44. 2nd ed.

Madison: American Society of Agronomy, 2015.

https://dl.sciencesocieties.org/publications/books/abstracts/agronomymonogra/agron

monogr57/agronmonogr57.2013.0020.

Xu, B., Chaoying Fang, Robin Huang, and Michael D. Watson. “Chromatic Image

Analysis for Cotton Trash and Color Measurements.” Textile Research Journal 67,

no. 12 (December 1997): 881–890.


Yang, Shouren, and Stuart Gordon. “Accurate Prediction of Cotton Ring-Spun Yarn

Quality from High-Volume Instrument and Mill Processing Data.” Textile Research

Journal 87, no. 9 (June 16, 2017): 1025–1039. Accessed March 8, 2019.



35

CHAPTER II

EFFECTS OF NON-LINT MATERIAL ON HERITABILITY

ESTIMATES OF COTTON FIBER LENGTH PARAMETERS

2.1 Introduction

Cotton fiber yield and quality are both essential elements for the producers while

fiber quality is of the utmost importance to market the crop and transform it into a textile

product. Farmers need varieties with high yield while spinning mills need cotton with a

fiber quality profile that allows them to spin a yarn of a given quality. This demand leads

cotton breeders to target the improvement of both fiber yield and quality (Ethridge and

Hudson 1998; Stewart et al. 2010).

Fiber quality is a complex trait. There are many interrelated fiber quality

parameters critical to the spinning industry. This myriad of parameters and their

interrelationships cannot be summarized as a single variable. Different spinning

technologies and end products will need fibers with a different set of quality properties.

For example, fiber length is essential for ring spinning, but the fibers must also be strong

and mature, or they will break during processing (Elhawary 2014; Kelly and Hequet

2018).

Cotton breeders usually work with seed cotton harvested by hand from individual

plants or harvested by research-type machines (e.g., stripper harvesters without field

cleaners) from rows to make their selections while cotton producers harvest with

commercial-type machines (e.g., cotton stripers with field cleaners). These differences in

harvesting methods can result in differences in non-lint content; such as pieces of leaves,

bracts, and stem (Faulkner et al. 2011a,b). Additionally, the type of non-lint material can

vary from one sample to the next (van der Sluijs and Hunter 2017). The non-lint content

is generally referred to as trash.

Trash content has the potential to cause problems with fiber quality measurements

(Peirce and Lord 1939; Liu and Delhom 2018). For example, the micronaire, a

combination of maturity and fineness, is based on the measurement of airflow through a

plug of lint of a known weight in a chamber of known dimension (Lord 1955, 1956).


36

Excessive trash content in the sample may create air channels through the plug of fibers,

resulting in an unreliable micronaire reading. Furthermore, since the mass of trash in the

sample is unknown, the true lint mass is unknown (Lord 1956; Fryer et al. 1996).

Excessive sample trash may cause problems with other High Volume Instrument

(HVI) measurements as well. For example, the micronaire reading is used in combination

with the optical amount of the fiber beard being broken to estimate the sample mass

during bundle strength measurement. An imperfect micronaire measurement will affect

the estimate of the beard mass and, therefore, the strength measurement (Keskin et al.

2001; Naylor et al. 2014). Thus, trash can affect the reliability of HVI measurements

directly and indirectly.

Fiber length is often the focus of trait improvement efforts in cotton because it is a

critical fiber quality property in spinning, and it is heritable (Elhawary 2014; Campbell et

al. 2018). The HVI measures fiber length using the fibrogram principle (Delhom et al.

2018). In the fibrogram principle, a sample is deposited into a fibrosampler and pressed

against a perforated curved wall, forcing tufts of fibers to protrude through openings

(Figures 2-1A and B). A metallic comb passes over the external surface of the

fibrosampler, allowing fibers to be randomly captured by the comb teeth, forming a fiber

beard (Figure 2-1C). The beard is brushed against the opposed surface of the

fibrosampler, on a carding surface that removes most of the fibers that are not caught in

the comb and begins to parallelize the beard (Figure 2-1D). Finally, a brush straightens

and removes loose fibers from the beard before exposing the fibers to a light scanner that

generates a fibrogram. In the fibrograph method, the attenuation of light is measured over

the randomly caught beard of fibers, producing a curve representing the optical amount

versus displacement. The optical amount is assumed to be proportional to the weight of

the fiber beard at each point. If this assumption is true, the fibrogram would correspond

to a fiber length distribution by weight. The HVI reports the upper half mean length

(UHML) and the uniformity index (UI), a percentage ratio between mean length (ML)

and UHML (Chu and Riley 1997; ASTM 2013; Delhom et al. 2018). However, this

method was developed on clean samples and the effects of trash on this method are not

well-documented (Hertel and Zervigon 1936; Hertel 1940).


37

Figure 2-1 Fibrosampler in the High Volume Instrument (HVI). A cotton sample is

deposited into the fibrosampler basket (A). Fibers protrude from the perforated curved

wall (B). The comb is passed externally on this surface, creating the fiber beard (C). The

comb is brushed on a carding surface on the fibrosampler (D) to remove free fibers and to

begin to parallelize the beard.

While the HVI is the dominant marketing tool for cotton classification, the

Advanced Fiber Information System (AFIS) is commonly used in spinning mills and

research applications. The AFIS measures properties of individual fibers and creates by-

number distributions related to fiber quality attributes, such as length (Shofner et al.

1995). Researchers typically compare treatments based on parameters extracted from

these distributions; such as short fiber content by number (SFCn) (the percentage of

fibers with a length equal or smaller than 12.5 mm), the mean fiber length by number

(Ln), and the 5% longer fibers by number (L5%) (B. R. Kelly and Hequet 2018; C. M.

Kelly, Hequet, and Dever 2013). The AFIS also generates fiber distributions by-weight.

Several parameters can be derived from these distributions; such as the mean fiber length

by weight (Lw) and the upper quartile length by weight (UQLw). However, the by-

weight measurements are calculated from the length by-number distribution assuming a

Md. Abu Sayeed, 2017

Khawar Arain, 2019 Khawar Arain, 2019

A B

C D


38

constant fiber linear density across length groups (Delhom, Kelly, and Martin 2018; Krifa

2006).

The AFIS fiber quality parameters are based on different fiber quality

measurement principles than HVI. Samples are presented to the AFIS fiber individualizer

as a hand-shaped sliver. The sliver is submitted to the action of a perforated pinned

cylinder that individualizes fibers and removes trash particles. An airflow separates trash

and fibers, carrying them to two different sensors. The optical sensors will then measure

fibers, fiber entanglements, i.e., neps, and trash particles. Not all fibers that are presented

to the sensor are used to create the reported fiber length distribution. There are electronic

filters that reject the result from some of the fibers if the fiber has, for example, a

calculated speed faster than the airflow or a signal that cannot be fully distinguished from

a nep signal (Shofner 1985; Shofner and Shofner 1999; Shofner et al. 1995; C. M. Kelly,

Hequet, and Dever 2013).

The two systems, HVI and AFIS, can provide fiber length distributions but they

are not equivalent. Cottonseed has a native fiber length distribution. During ginning,

some fibers may be broken or removed from the sample. If the lint is then processed with

a lint cleaner, more breakage and removal may happen, increasing the difference between

the native fiber length distribution and the post-processing distribution (P. Bel-Berger et

al. 1991). The HVI fibrogram is a tool to measure and report the native fiber distribution

modified by processing at the gin. The AFIS has a built-in cleaning device that may also

break fibers (Shofner and Shofner 1999). Therefore, results from the AFIS report on the

native length distribution modified by the effect of processing at the gin and the effect of

AFIS processing. In addition, the HVI cannot measure the very short fibers as it begins to

scan the beard at 3.81 mm from the comb.

Finally, trash particles may be present in the sample while the beard is formed

prior to the HVI length measurement or in the AFIS sliver. We hypothesize that the

presence of trash particles in the samples may compromise the quality of the length

measurement and screening decisions in breeding programs. Thus, breeders may need to

manage the trash content in their samples. However, it is often not practical to apply

industrial-scale processes to research material. Therefore, we devised an experiment to


39

evaluate three trash management strategies and the effects that these strategies have on

fiber quality length heritability estimates.

2.2 Material and methods

2.2.1 Mating design and plant materials

A North Carolina mating design II (NCII) was chosen to perform this experiment.

This mating design was selected because it provides the information needed to meet our

objective without requiring as many resources as a full diallel (Fristche-Neto, Akdemir,

and Jannink 2018). The heritability estimates used in this research were calculated based

on the parental lines and the F2 generation. (Hill, Becker, and Tigerstedt 1998; Comstock

and Robinson 1952; Tang et al. 1996).

2.2.2 Identifying parental material

A set of six cultivars were chosen for this experiment based on a range of upper

half mean lengths (UHML) as measured by HVI. The selected cultivars were split into

maternal and paternal sets of three varieties each. Each parental set is composed of

cultivars representing low, medium, and high UHML (Boman, Kelley, and Ashbrook

2010; Kelley and Keys 2015).

2.2.3 Obtaining F2 seed

The parental seeds were planted 2.54 cm deep in perforated plastic pots of 30 L

with 25 L of Sungro Metro-Mix® 360 at the Texas Tech University greenhouse. Each pot

was fertilized with 32 g of Osmocote® Classic Fertilizer 14-14-14 and received 0.5 g of

Mantra® 1 G. Four seeds were planted per pot and thinned to two seedlings after three

weeks. Each pot was weekly watered as needed until water began to drop from the

bottom. Plants were sprayed with a rotation of insecticides to minimize acquired

resistance. The used insecticides were Avid® 0.15EC, Distance®, Discuss®, and Pylon®.

Closed flowers from the maternal set of plants were emasculated in the afternoon of the

day before anthesis. Stigma and style were sprayed with tap water and covered with a


40

piece of paper straw to protect them and avoid undesired outcrossing. In the morning of

anthesis, the piece of paper straw was removed. Open flowers from the paternal set were

picked and their pollen was sprinkled on the stigma of emasculated flowers. After

crossing, the pollinated flowers were covered again with the piece of paper straw.

Seedcotton was hand-picked from each boll in order to obtain F1 seed. The seed

cotton was ginned with a laboratory-scale roller gin (Dennis Manufacturer, Athens, TX),

delinted with a research mechanical delinter at Texas A&M AgriLife Research &

Extension Center, Lubbock TX, and planted by hand at the Texas Tech University

research farm in Lubbock, Texas, on a loam soil with drip irrigation and following the

recommendation for irrigated cotton production in the region (Ayele, Kelly, and Hequet

2018). A total of 75 F1 seeds for each of the nine crosses were planted in unreplicated

7.6-meter long plots. Three seeds were planted per hill and the hills were spaced 30 cm

apart. Seedlings were thinned to one plant per hill after three weeks. The hybrid plants

were self-pollinated using clips to prevent flower opening. Seedcotton was hand-picked

from each boll in order to obtain F2 seeds. The seedcotton was ginned with a laboratory-

scale roller gin (Dennis Manufacturer, Athens, TX), delinted with sulfuric acid 20% w/w

in a delinting system at Texas Tech University, and prepared for machine planting.

2.2.4 Field experiment

The heritability experiment was planted at the Texas Tech University Research

farm, Lubbock, Texas, in a completely randomized block design with three field

replications. Each plot was a single row 6.1-meter long with a density of 10 seeds per

meter to simulate commercial planting (Figure 2-2). The seeds were planted on a loam

soil with drip irrigation and following the recommendation for irrigated cotton production

in the region (Ayele, Kelly, and Hequet 2018). During the experiment, the accumulated

rainfall was 272 mm and growing degree days (GDD15.6) were 1308 units (Snowden et al.

2013).


41

Figure 2-2 Flowchart from original crosses to field experiment.

2.2.5 Harvesting, ginning, and processing

A 50-boll-sample was hand-picked from each F2 plot from across boll positions to

obtain samples that are about identical to typical breeding samples, i.e., very low trash

content due to hand harvesting. The remaining seed cotton from each plot was

mechanically harvested with a cotton stripper with no field cleaner to obtain samples with

a higher amount of non-lint content. The mechanically harvested samples were pre-

cleaned with a burr and stick extractor in an Imperial III Lummus gin to remove sticks

and other large pieces of foreign material that could impede laboratory-scale ginning. The

two types of samples were conditioned at 21 ± 1°C and relative humidity of 55 ± 5% for

seven days before ginning with a laboratory-scale Compass 10-saw gin (Model

MG1010), with no seed cotton cleaner and no lint cleaner.

Subsamples of lint from the trashy samples were mechanically processed with a

Shirley analyzer laboratory-scale lint cleaner after ginning. This device was chosen

because it mimics the cleaning action and mechanical stress that is applied during

industrial processing of cotton (Pfeiffenberger 1944). These samples represent a scenario


42

where trash particles are not managed in the field but instead removed in the laboratory

prior to fiber quality testing.

2.2.6 Fiber quality testing

The three different types of lint samples were conditioned at 21 ± 1°C and 65 ±

2% of relative humidity (ASTM 2016) and tested by HVI following a 4-4-10 protocol,

which consists of four measurements of micronaire, four measurements of trash/color,

and ten measurements of length/strength/elongation. The samples were also tested by the

Advanced Fiber Information System (AFIS) following a protocol of five replications of

3,000 fibers (Figure 2-3).


43

Figure 2-3 Flowchart from the field experiment to fiber quality analysis.

2.2.7 Heritability estimates and statistics

Mean values for fiber quality parameters (leaf grade, visible foreign matter, trash

count, dust count, nep count, HVI upper half mean length, HVI mean length, AFIS short

fiber content by number, AFIS mean length, AFIS upper quartile length by number, and

AFIS 5% longer fibers by number) in the F2 populations were analyzed using analysis of

variance (ANOVA) and honest significant difference (HSD) Tukey’s test at 5% of

significance.


44

Heritability estimates for fiber quality parameters were performed based on the

variance components obtained from the half-sibling experimental design (Hill, Becker,

and Tigerstedt 1998; Comstock and Robinson 1948). The NCII can be analyzed as a two-

way ANOVA factorial design in which one factor is the maternal component while the

other factor is the paternal component. Following this procedure, the parental

components are related to the additive variance component of heritability (Va), the

interaction between the components is related to the non-additive variance component of

heritability (Vd), and the error is related to environmental variance (Ve) (Table 2-1).

Table 2-1 Two-way analysis of variance used to calculate the heritability estimates in

this research.

Item of variance Degrees of freedom Component of variance

Males 2 Va1

Females 2 Va2

Males x Females 4 Vd

Within plots 18 Ve

Total 26 Vp

The broad sense heritability estimate was calculated as the fraction of all

components of genetic variance over the total variance: H2 = (Va1 + Va2 + Vd) / (Va1 + Va2

+ Vd + Ve). The narrow sense heritability estimate was calculate as the fraction of the

additive variance component and the total variance: h2 = (Va1 + Va2) / (Va1 + Va2 + Vd +

Ve). Broad sense (H2) and narrow sense heritability (h2) estimates were calculated for

parameters obtained from HVI and AFIS.


45

2.3 Results and discussion

2.3.1 Sample type characteristics

A total of three types of samples with different trash contents were obtained from

each plot following this protocol; low trash content with low mechanical processing

(treatment I), high trash content with low mechanical processing (treatment II), and low

trash content with high mechanical processing (treatment III). The HVI measurement of

trash is based on image analysis of the sample surface and is reported on an ordinal scale

from 1 to 8 (Cotton Incorporated 2013). The treatment I samples have an average HVI

leaf grade of 1.1, while the treatment II has an intermediate value of 4.3. Cleaning

treatment II samples resulted in treatment III samples, with an average leaf grade of 1.0

(Table 2-2).

Table 2-2 Average value for parameters related to trash and processing in nine F2

samples, with three field replications, from three treatments with different levels of trash

content and mechanical processing.

Treatment* Leaf

(no unit)

VFM**

(%)

Trash/g

(count)

Dust/g

(count)

Neps/g

(count)

Treatment I 1.1 ± 0.1a 0.30 ± 0.03a 12 ± 2a 64 ± 7a 106 ± 3a

Treatment II 4.3 ± 0.2b 2.20 ± 0.14b 132 ± 9b 420 ± 27b 155 ± 3b

Treatment III 1.0 ± 0.1a 0.20 ± 0.03a 13 ± 1a 54 ± 5a 180 ± 4c

*Average value for three field replications ± standard error. Means in the column

followed by the same letter are not statistically different by HSD Tukey’s test at 5% of

significance. Treatment I: samples with low trash content and low mechanical stress;

treatment II: samples with high trash content and low mechanical stress; treatment III:

samples with low trash content and high mechanical stress.

**Visible Foreign Matter

The AFIS provides a different method for measuring the trash content. It is based

on the aeromechanical separation of trash particles from the lint (Shofner and Shofner

1999). The trend observed with HVI holds true with the AFIS assessment of visible

foreign matter (VFM), trash, and dust for the three types of samples. The samples with


46

lower leaf grade, treatments I and III, have the lowest levels of non-lint content according

to AFIS.

Fiber entanglements, called neps, can create imperfections in spun yarns (M. H.

van der Sluijs and Hunter 2016). The highest level of neps/g in this experiment was

obtained with the laboratory cleaned samples. The mechanical processing used to clean

the samples also has the potential to reduce fiber quality by breaking and entangling

fibers, increasing the measured number of neps.

2.3.2 Parental material characteristics

The parents were selected based on reported data for their HVI upper half mean

length (UHML) to create a large range for length properties. The HVI data for the hand

picked parents show a range of 3.04 mm for UHML and ML, and 1.87% for UI (Table 2-

3). Compared to the global average, these values have a range of -4.54% to 6.50% for

UHML, -5.45% to 7.92% for the mean length (ML), and -0.94 to 1.32% for UI. UI is the

ratio of ML to UHML (ASTM 2013). The observed range of variability for UI is

narrower than the observed range for UHML and ML. Without a large range for a

property, the experimental variability is low and the heritability measurements are

negatively impacted. Therefore, we calculated heritability estimates only for UHML and

ML.

The AFIS data exhibit a range of -4.70% to 8.06% for the 5% longer fibers

(L5%), -4.52% to 7.83% for the mean length by number (Ln), -13.76% to 14.28% for

short fiber content by number (SFCn), and -4.22% to 7.30% for the upper quartile length

by number (UQLw) (Table 2-4). The observed ranges for AFIS length parameters are as

good as the ranges observed for HVI UHML and ML, and they were used in our

analyses.


47

Table 2-3 Fiber quality properties of the hand picked parental varieties in three field replications tested by HVI.

HVI

Parameter*

Upper half mean

length

(mm)

Mean length

(mm)

Uniformity index

(%)

Micronaire

(no unit)

Strength

(kN·m·kg-1)

Elongation

(%)

Maternal set

Cultivar A 26.25 ± 0.29 21.50 ± 0.25 81.90 ± 0.12 4.90 ± 0.08 273.0 ± 4.9 7.03 ± 0.12

Cultivar B 26.59 ± 0.39 22.08 ± 0.42 83.03 ± 0.24 5.24 ± 0.10 291.4 ± 1.7 7.27 ± 0.03

Cultivar C 29.29 ± 0.78 24.54 ± 0.92 83.77 ± 0.55 4.57 ± 0.03 307.3 ± 4.9 5.97 ± 0.09

Paternal set

Cultivar X 27.43 ± 0.44 22.72 ± 0.51 82.80 ± 0.32 4.93 ± 0.05 287.7 ± 1.9 8.10 ± 0.06

Cultivar Y 28.19 ± 0.51 23.12 ± 0.58 82.00 ± 0.40 4.84 ± 0.03 274.0 ± 1.7 5.77 ± 0.07

Cultivar Z 27.26 ± 0.64 22.51 ± 0.54 82.57 ± 0.13 4.87 ± 0.05 304.1 ± 6.4 6.30 ± 0.10

Global

average 27.50 ± 0.27 22.74 ± 0.26 82.68 ± 0.19 4.89 ± 0.05 289.6 ± 3.5 6.74 ± 0.20

*HVI: High Volume Instrument, average value for three field replications, four laboratory readings for micronaire, and 10 readings for

other parameters ± standard error.


48

Table 2-4 Quality properties of the hand picked parental varieties in three field replications tested by AFIS.

AFIS

Parameter*

5% longer fibers

by number

(mm)

Mean length by

number

(mm)

Short fiber

content by

number

(%)

Upper quartile

length by weight

(mm)

Fineness

(mtex)

Maturity ratio

(no unit)

Maternal set

Cultivar A 31.33 ± 0.34 19.64 ± 0.22 21.2 ± 0.8 27.60 ± 0.31 192 ± 1 0.92 ± 0.02

Cultivar B 30.99 ± 0.44 20.32 ± 0.51 17.7 ± 1.6 27.43 ± 0.39 192 ± 1 0.96 ± 0.02

Cultivar C 35.14 ± 0.72 22.18 ± 0.59 16.3 ± 0.9 30.73 ± 0.64 173 ± 1 0.96 ± 0.02

Paternal set

Cultivar X 32.17 ± 0.34 20.83 ± 0.64 17.3 ± 2.0 28.36 ± 0.34 187 ± 1 0.92 ± 0.02

Cultivar Y 33.70 ± 0.47 20.40 ± 0.37 21.6 ± 1.0 29.72 ± 0.51 180 ± 1 0.95 ± 0.01

Cultivar Z 31.83 ± 0.45 20.07 ± 0.39 19.5 ± 1.5 28.02 ± 0.31 179 ± 1 0.95 ± 0.02

Global

average 32.52 ± 0.39 20.57 ± 0.25 18.9 ± 0.6 28.64 ± 0.32 174 ± 2 0.94 ± 0.07

*AFIS: Advanced Fiber Information System, average value for three field replications and five laboratory readings of 3,000 fibers ±

standard error.


49

2.3.3 Heritability estimates for HVI length measurements

The UHML and ML have the highest broad and narrow sense heritability

estimates when measured on hand-picked samples, with low trash content and low

mechanical processing. The lowest heritability estimates were obtained on bulk harvested

samples, with high trash and low mechanical processing (Tables 2-1, 2-2 and 2-5).

Table 2-5 Heritability estimates for HVI length fiber properties in three treatments of

nine F2 samples, with three field replications, with different levels of trash content and

mechanical processing.

HVI values

Parameter Upper half mean length

(mm)

Mean length (mm)

Treatment*

Treatment I 27.9 ± 0.2a 23.2 ± 0.2a

Treatment II 27.9 ± 0.2a 23.0 ± 0.2a

Treatment III 27.5 ± 0.2a 22.4 ± 0.1b

Broad sense heritability (H2)

Parameter Upper half mean length Mean length

Treatment

Treatment I 0.701 0.650

Treatment II 0.564 0.513

Treatment III 0.625 0.538

Narrow sense heritability (h2)

Parameter Upper half mean length Mean length

Treatment

Treatment I 0.622 0.579

Treatment II 0.540 0.478

Treatment III 0.588 0.538







50

The treatment I samples have a narrow sense heritability estimate of 0.622, while

treatment II samples have a value of 0.540. Similarly, the heritability estimates of ML for

treatment I samples are 0.579 and 0.478 for treatment II samples. The samples in both

treatments have the same genetic background and were managed the same way in the

same field. Therefore, the reduced estimated heritability is reflecting the effects of

harvesting and cleaning methods.

Heritability is defined as “the proportion of the observed variation in a progeny

that is inherited”. Breeders are more likely to improve a trait with higher heritability

estimate using typical breeding methods based on phenotype (Acquaah 2012) The

heritability estimates are based on the trait, the population, and the environment.

Modifications in the population or environment will impact the heritability estimate for

the trait (Hill, Becker, and Tigerstedt 1998; Acquaah 2012).

The UHML and strength are traits with high heritability. Analyzing data from 13

experiments, Campbell et al. (2018) observed that the narrow sense heritability for

UHML ranged from 0.00 to 0.70, with an average value of 0.30. This is an indication that

the data used in this experiment are of good quality.

Seedcotton samples have a native length distribution. The treatment I samples

have low trash content and low mechanical process, possibly resembling more the native

length distribution. Therefore, the treatment I samples may serve as a reference treatment

for the effects that trash and processing may have on other sample types.

There is a reduction in the heritability estimates for UHML and ML from the

treatment I to treatment II. Trash may affect fiber sampling in the HVI, creating

differences in the heritability estimates. As explained earlier, trash particles may result in

fiber quality measurement bias. In the HVI, a comb takes fibers from the tufts of lint

protruding through the orifices of the sampling basket, forming a fiber beard. Trash can

interfere with this sampling mechanism and beard formation. Since trash is not uniformly

distributed within the sample, the population of fibers from one beard to another may be

different for the same sample. The inflated within-sample variation imparted by this

environmental variation leads to reduced calculated heritability for treatment II samples.


51

The heritability estimates are based on the genetic component of the trait of

interest, the variability imparted by the environment, and the interaction between them.

We observed numerical differences among treatments for all analyzed fiber properties

with the hand picked boll samples providing the highest estimates and the machine

harvested samples the lowest. The only difference between both treatments is the trash

content. This confirms our hypothesis that trash content may negatively impact cotton

fiber length measurements.

Researchers may not always measure fiber properties on samples with native low

trash content. One alternative to managing trash is to clean samples in the laboratory

before fiber quality assessment. It results in samples with low trash content and high

mechanical stress, the treatment III samples. Treatment I and III samples have the same

leaf grade, dust, and trash content (Tables 2-2 and 2-5).

Cleaning samples with a Shirley analyzer, treatment III, resulted in heritability

estimates for UHML and ML with values between the calculated values for treatments I

and III (Table 2-5). The Shirley analyzer is a device created to simulate the lickerin of a

card. In a lickerin, mechanical stress is applied to remove non-lint content from the

samples (Pfeiffenberger 1944; ASTM 2012d). This mechanism can break fibers, modify

the native fiber length distribution, and entangle fibers creating neps. Cleaning samples

with a Shirley analyzer removes the trash effect but adds a processing effect. There is no

statistical difference at 95% of confidence for the UHML between treatment I UHML

(27.9 ± 0.4 mm) and treatment III UHML (27.5 ± 0.3 mm). Nevertheless, there is a

statistical difference at 95% of confidence between treatment I ML (23.2 ± 0.4 mm) and

treatment III ML (22.4 ± 0.3 mm).

The shortening of ML is evidence for processing effect. The heritability estimate

numerically improved from treatment II to treatment III. We hypothesize that the

improvement happens because fewer trash particles in the lint samples result in smaller

within-sample variation among HVI combs.


52

2.3.4 AFIS fiber quality properties

2.3.4.1 AFIS length distributions

The AFIS provides length distributions by number. There are differences in the

length distributions among the three treatments that will likely impact the means and

heritability estimates for the AFIS length parameters (Figure 2-4). As previously

discussed, differences are related to the effects of harvesting and cleaning methods. The

effects will impact the trash content and processing effect in the samples.

Figure 2-4 AFIS length distributions for the analyzed samples. The black lines represent

the range among the different bins and the grey line represents the average. Treatment I:

samples with low trash content and low mechanical stress; treatment II: samples with

high trash content and low mechanical stress; treatment III: samples with low trash

content and high mechanical stress.

2.3.4.2 The 5% longer fibers by number and upper quartile length by weight

The treatment I samples exhibit the highest numeric value of broad (0.691) and

narrow sense (0.607) heritability estimates for the 5% longer fiber by number (L5%)

while treatment II samples have a lower numerical value of broad (0.574) and narrow

sense (0.572) heritability estimates for the L5%, and treatment III samples have the

lowest numerical value of broad (0.542) and narrow sense (0.517) heritability estimates


53

for the L5%. There are no statistical differences among the L5% mean values for the

three treatments. Although the average measurements for L5% are not affected by trash

or processing, the heritability measurements, based on variance, are impacted by trash

and processing (Table 2-6).

The AFIS has a built-in fiber individualizer that is also a cleaning device. This

perforated pinned cylinder applies mechanical processing on the tufts of fibers, separating

trash and individual fibers. We hypothesize that some long fibers may entangle with trash

in treatment II samples. When the fibers are individualized, a fraction of the long fibers

entangled with trash may break or get removed with the trash or be rejected when

presented to the length sensor. If degradation occurs on the length of the population of

fibers related to the longer fibers, the heritability may be reduced.


54

Table 2-6 Heritability estimates for AFIS length fiber properties in three treatments of

nine F2 samples, with three field replications, with different levels of trash content and

mechanical processing.

AFIS values

Parameter 5% longer

fibers by

number

(mm)

Upper quartile

length by

weight

(mm)

Mean

length by

number

(mm)

Short fiber

content by

number

(%)

Treatment*

Treatment I 33.2 ± 0.3a 29.1 ± 0.2a 21.0 ± 0.2a 18.53 ± 0.51a

Treatment II 32.8 ± 0.2a 28.7 ± 0.2a 19.9 ± 0.1b 22.26 ± 0.40b

Treatment III 32.6 ± 0.2a 28.5 ± 0.2a 19.6 ± 0.1b 23.07 ± 0.38b

Broad sense heritability (H2)

Parameter 5% longer

fibers by

number

Upper quartile

length by

weight

Mean

length by

number

Short fiber

content by

number

Treatment

Treatment I 0.691 0.662 0.471 0.535

Treatment II 0.574 0.622 0.642 0.779

Treatment III 0.542 0.562 0.446 0.610

Narrow sense heritability (h2)

Parameter 5% longer

fibers by

number

Upper quartile

length by

weight

Mean

length by

number

Short fiber

content by

number

Treatment

Treatment I 0.607 0.603 0.427 0.413

Treatment II 0.572 0.607 0.617 0.694

Treatment III 0.517 0.540 0.530 0.464







55

Treatment III samples were cleaned with a Shirley analyzer and this processing

impacts the heritability measurements. Cleaning samples removes trash and adds

processing effect. The longer fibers may be expelled together with trash or broken into

shorter fibers or separated from trash but fragilized during cleaning. Some of the

fragilized fibers from treatment III samples may break when they are processed by the

AFIS individualizer. We hypothesize that this additional degradation of the longer fibers

used to calculate the L5% for treatment III samples resulted in the calculation of a lower

heritability measurement.

The upper quartile length by weight (UQLw) measures longer fibers in the

samples, but shorter than the population of fibers used to calculated the L5%. The AFIS

does not measure the weight of individual fibers. The reported distribution by-weight is

calculated based on the distribution by-number and an assumption of constant linear

density to all measured fibers. We hypothesize that the UQLw is following a similar

trend to L5%, and the factors influencing the heritability estimates for L5% will also

impact the UQLw measurement calculations.

2.3.4.3 Mean length by number

Treatment I presented the highest average mean length by number (Ln) (21.0 mm)

and the lowest numeric value for narrow sense estimate (0.427) (Table 2-6). Treatment II

has an average mean length shorter than treatment I (19.9 mm) and the highest numerical

value for broad (0.642) and narrow (0.617) sense heritability estimates. Our hypothesis to

explain this trend is similar to what was discussed for the HVI UHML. Treatment I

presents the highest variation around the Ln (Figure 2-4). This variation is the closest as

possible as it can be measured in the native length distribution. As previously discussed,

trash particles may entangle with longer fibers. These trash particles are not uniformly

distributed within the samples. We hypothesize that some of these longer fibers will

break when processed by the AFIS individualizer. The breakage will not uniformly

happen in the different samples because the trash particles are not uniformly distributed.

The number of shorter fibers will increase and the number of longer fibers will decrease,

reducing the within-sample variation around the region related to Ln (Figure 2-4),


56

inflating the heritability estimate calculated for treatment II samples to values above the

reference value calculated for treatment I. If this holds true, even if the calculated narrow

sense heritability estimate for treatment II is higher than the value for treatment I, this is a

value artificially inflated.

Treatment III samples were cleaned to remove the trash. The numeric value for

the Ln narrow sense heritability estimate increases to a value (0.530) between the

calculated estimates for treatments I and II. We hypothesize that fiber breakage may have

occurred during cleaning or some fibers were fragilized and broke at the AFIS fiber

individualizer. The additional breakage increases the variability for shorter fibers,

numerically increasing within-sample variation and decreasing the heritability estimates.

The variation added by processing is different from the variation observed in treatment I

(Figure 2-4 and Table 2-6).

2.3.4.4 Short fiber content by number

Treatment I samples present the lowest average short fiber content by number

(SFCn) (18.53%) and lowest broad (0.535) and narrow (0.413) sense heritability

estimates. The SFCn value for treatment II samples (22.26%) is statistically higher than

the value for treatment I. Treatment II samples have the highest numerical value for

broad (0.779) and narrow (0.694) sense heritability estimates for SFCn. As previously

discussed, we hypothesize that breakage may occur during fiber individualization

because of trash entanglement with fibers. The breakage increases the number of shorter

fibers in the samples, reducing within-sample variation and artificially inflating the

calculated heritability estimate for SFCn.

The broad (0.610) and narrow (0.464) sense heritability estimates for the

treatment III samples are an intermediate numerical value between the calculated

heritability estimates for treatments I and II. As previously hypothesized, fiber breakage

and fragilization during cleaning may have added variation to the fiber length

distribution. If fragilized fibers break during fiber individualization, the within-sample

variation increases and the heritability measurement is reduced.


57

2.4 Conclusions

Cotton researchers may obtain samples with different trash contents because of

crop management, harvest, and even post-harvest processing such as ginning. Variability

in the trash content may be also influenced by traits such as the shape of leaves and

bracts, and the pubescence of leaves and stems. We observed by this research that trash is

a factor in fiber quality assessment that degrades the quality of HVI and AFIS length

measurements.

Samples with high trash content (treatment II) presented the lowest calculated

heritability estimates for HVI length parameters. We hypothesize that trash particles

impact HVI length measurements by entangling with the fibers, creating variation among

fiber beards. This variation created by trash particles artificially increases variation in

fiber length, obfuscating the true fiber length measurements. Cleaning samples (treatment

III) reduces the trash content and increases the calculated heritability estimates.

Nevertheless, this is the heritability of the native length distribution and the processing

effect. The most appropriate trash management to minimize impacts on the fiber length

heritability estimates is using samples with native low trash content and low processing

effect (treatment I). If breeders can only obtain samples with high trash content, they

need to increase the selection pressure to overcome the lower heritability estimates or

they need to clean their samples before testing. This strategy may result in higher costs

because breeders may need to test a larger number of entries than they would have if they

used clean samples.

The AFIS has a built-in cleaning device, the AFIS fiber individualizer, which

removes trash and ensures that clean fibers are delivered to the sensor. This device may

create different impacts on fiber quality analysis. We observed by the results of our

research that AFIS testing is also impacted by trash and processing. We hypothesize that

trash particles may entangle with longer fibers. These fibers may be broken during

individualization or may be carried away with the trash or their signal may be rejected at

the fiber sensor. The degraded fiber length distribution results in length heritability

estimates with values different from samples with native low trash content. If trashy

samples are cleaned to remove trash, processing effect is added. The fiber length

variation will be different from the native length distribution. Once again, samples with


58

native low trash content and low processing are expected to provide research results

closer to the native length distribution (treatment I).

This research proved the impact of trash on the fiber length distribution analysis.

Further research will study the processing effect and clarify the impact of this factor in

cotton fiber length quality research.


59

2.5 Bibliography

Acquaah, George. Principles of Plant Genetics and Breeding. Edited by George

Acquaah. 2nd ed. West Sussex: Wiley-Blackwell, 2012.

ASTM. “ASTM D1776/D1776M-16 Standard Practice for Conditioning and Testing

Textiles.” 1–5. West Conshohocken: ASTM International, 2016.

———. “ASTM Standard D123-13a, ‘Standard Terminology Relating to Textiles.’” In

ASTM, 1–68. West Conshohocken, 2013.

http://enterprise2.astm.org/DOWNLOAD/D123.389810-1.pdf.





(2018): 47–55.


Bel-Berger, P., E. P. Columbus, C. K. Bragg, and K. Q. Robert. “Effects of Mechanical

Cleaning on Cotton Fibers: Part I: Ginning.” Textile Research Journal 61, no. 2

(1991): 83–88.

Boman, Randy, Mark Kelley, and Chris Ashbrook. Systems Agronomic and Economic

Evaluation of Cotton Varieties in the Texas High Plains. Lubbock: Texas A&M

Agrilife Extension, 2010.

Cai, Yiyun, Xiaoliang Cui, James Rodgers, Devron Thibodeaux, Vikki Martin, Mike

Watson, and Su-Seng Pang. “A Comparative Study of the Effects of Cotton Fiber

Length Parameters on Modeling Yarn Properties.” Textile Research Journal 83, no.

9 (June 8, 2013): 961–970. Accessed October 4, 2018.





2018.

Chu, Youe-Tsyr, and C. Roger Riley. “New Interpretation of the Fibrogram.” Textile

Research Journal 67, no. 12 (December 1997): 897–901.


Comstock, R. E., and H. F. Robinson. “Estimation of Average Dominance of Genes.” In

Heterosis, edited by J. W. Gowen, 494–516. Iowa State College Press, 1952.

http://www.biodiversitylibrary.org/bibliography/4498.

———. “The Components of Genetic Variance in Populations of Biparental Progenies

and Their Use in Estimating the Average Degree of Dominance.” In Biometrics,

4:254–266, 1948.



60



booklet.pdf.

Delhom, Christopher D., Brendan Kelly, and Vikki Martin. “Physical Properties of

Cotton Fiber and Their Measurement.” In Cotton Fiber: Physics, Chemistry and

Biology, edited by D Fang, 41–73. Cham: Springer International Publishing, 2018.

Elhawary, I.A. “Fiber to Yarn: Staple-Yarn Spinning.” In Textile and Fashion -

Materials, Design, and Technology, edited by Rose Sinclair, 191–211. Sawston:


Ethridge, D, and D Hudson. “Cotton Market Price Information: How It Affects the

Industry.” Journal of Cotton Science 2, no. 1 (1998): 68–76.

https://www.scopus.com/inward/record.uri?eid=2-s2.0-

3142777967&partnerID=40&md5=4f0cf0120e5ceb0ff3b47f953b55171c.

Faulkner, W. B., J. D. Wanjura, E. F. Hequet, R. K. Boman, B. W. Shaw, and C. B.

Parnell Jr. “Evaluation of Modern Cotton Harvest Systems on Irrigated Cotton:

Fiber Quality.” Applied Engineering in Agriculture 27, no. 4 (2011): 507–513.

Accessed May 1, 2018. https://naldc.nal.usda.gov/download/50027/PDF.

———. “Evaluation of Modern Cotton Harvest Systems on Irrigated Cotton: Harvester

Performance.” Applied Engineering in Agriculture 27, no. 4 (2011): 497–506.

Fristche-Neto, Roberto, Deniz Akdemir, and Jean-Luc Jannink. “Accuracy of Genomic

Selection to Predict Maize Single-Crosses Obtained through Different Mating

Designs.” Theoretical and Applied Genetics 131, no. 5 (May 14, 2018): 1153–1162.

Accessed June 24, 2019. http://link.springer.com/10.1007/s00122-018-3068-8.

Fryer, L.F., Jon P. Rust, and P.R. Lord. “Effects of Cotton Fiber Blending and Processing

on HVI Measurements—Part I.” Textile Research Journal 66, no. 6 (June 2, 1996):

349–357. Accessed July 3, 2019.


Hertel, K.L. “A Method of Fibre-Length Analysis Using the Fibrograph.” Textile

Research 10, no. 12 (October 8, 1940): 510–520.





Hill, J, H C Becker, and P M A Tigerstedt. Quantitative and Ecological Aspects of Plant

Breeding. 1st ed. Dordrecht: Springer Science+Business Media, 1998.

http://www.thomson.com.

Kelley, Mark, and Kristie Keys. Systems Agronomic and Economic Evaluation of Cotton

Varieties in the Texas High Plains. Lubbock: Texas A&M Agrilife Extension, 2015.

Kelly, Brendan R, and Eric F Hequet. “Variation in the Advanced Fiber Information

System Cotton Fiber Length-by-Number Distribution Captured by High Volume


61

Instrument Fiber Length Parameters.” Textile Research Journal 88, no. 7 (April






Keskin, M., R. B. Dodd, Y. J. Han, and A. Khalilian. “Comparison of Different Types of

Light Sources for Optical Cotton Mass Measurement.” Transactions of the ASAE 44,

no. 3 (2001): 715–720.

Krifa, Mourad. “Fiber Length Distribution in Cotton Processing: Dominant Features and

Interaction Effects.” Textile Research Journal 76, no. 5 (May 2, 2006): 426–435.

Accessed March 8, 2019.


Liu, Youngliang, and Christopher Delhom. “Effect of Instrumental Leaf Grade on HVI

Micronaire Measurement in Commercial Cotton Bales.” The Journal of Cotton

Science 22 (2018): 136–141. http://www.cotton.org/journal/2018-

22/2/upload/JCS22-136.pdf.

Lord, E. “10—Air Flow through Plugs of Textile Fibres Part I—General Flow

Relations.” Journal of the Textile Institute Transactions 46, no. 3 (March 1955):

T191–T213.


———. “2—Air Flow through Plugs of Textile Fibres: Part II. the Micronaire Test for

Cotton.” Journal of the Textile Institute Transactions 47, no. 1 (1956): T16–T47.

Naylor, Geoffrey RS, Christopher D Delhom, Xiaoliang Cui, Jean-Paul Gourlot, and

James Rodgers. “Understanding the Influence of Fiber Length on the High Volume

InstrumentTM Measurement of Cotton Fiber Strength.” Textile Research Journal 84,

no. 9 (June 16, 2014): 979–988. Accessed October 29, 2018.


Peirce, F. T., and E. Lord. “13—The Fineness and Maturity of Cotton.” Journal of the

Textile Institute Transactions 30, no. 12 (December 1939): T173–T210.




Shofner, Frederick M. “Apparatus and Methods for Aeromechanical and Electrodynamic

Release and Separation of Foreign Matter from Fiber.” USA, 1985.





USA, 1999.


62

van der Sluijs, M.H.J., and L. Hunter. “A Review on the Formation, Causes,

Measurement, Implications and Reduction of Neps during Cotton Processing.”

Textile Progress 48, no. 4 (2016): 221–323.

———. “Cotton Contamination.” Textile Progress 49, no. 3 (July 3, 2017): 137–171.

https://www.tandfonline.com/doi/full/10.1080/00405167.2018.1437008.

Snowden, Chase, Glen Ritchie, Justin Cave, Wayne Keeling, and Nithya Rajan.

“Multiple Irrigation Levels Affect Boll Distribution, Yield, and Fiber Micronaire in

Cotton.” Agronomy Journal (2013).

Stewart, James McD., Derrick M. Oosterhuis, James J. Heitholt, and Jackson R. Mauney.

Physiology of Cotton. Edited by James McD. Stewart, Derrick M. Oosterhuis, James

J. Heitholt, and Jackson R. Mauney. 1st ed. Dordrecht: Springer Netherlands, 2010.


Tang, Bing, J. N. Jenkins, C. E. Watson, J. C. McCarty, and R. G. Creech. “Evaluation of

Genetic Variances, Heritabilities, and Correlations for Yield and Fiber Traits among

Cotton F2 Hybrid Populations.” Euphytica 91, no. 3 (1996): 315–322. Accessed July

5, 2017. http://link.springer.com/10.1007/BF00033093.


63

CHAPTER III

EFFECT OF THE SHIRLEY ANALYZER ON FIBER LENGTH

DISTRIBUTIONS

3.1 Introduction

3.1.1 Fiber length distribution and factors that can affect it

Cotton lint is a natural product with inherent variability among fibers, even on a

single seed. This variability results in within-sample distributions of values for fiber

properties, such as length, maturity, and fineness (Wakeham 1955; Hequet et al. 2006;

Benzina et al. 2007; Kelly et al. 2013).

The length distribution of the lint from a cotton seed before ginning is called the

native fiber length distribution (Hertel and Zervigon 1936; Wakeham 1955). This native

length distribution is a combination of genetics and the environment. Fibers are removed

from the cotton seeds during ginning, resulting in a length distribution that is a

combination of the native length distributions and the fiber breakage resulting from

mechanical handling at the gin. Furthermore, if post-ginning processing is applied to a

lint sample, such as lint cleaning, additional fiber breakage may occur (Bel et al. 1991).

Breakage modifies the fiber length distribution, increasing the number of shorter fibers.

Cleaning can also change the length distribution because of fiber removal (Mogahzy and

Farag 2018). Therefore, the more processing is applied to a sample, the more different is

the sample distribution in comparison to the native distribution.

During processing, cotton fibers are moved by airflow or friction with the surface

of moving parts or other fibers. More power applied to the system results in faster

processing speeds. If there are two points where a fiber can be gripped, and if that fiber

cannot withstand the applied energy, the fiber will break (Figure 3-1). Even if the fibers

are not broken by this mechanical stress, they may be stretched, increasing the propensity

to break in subsequent steps (Mogahzy et al. 1990, 1998; Mogahzy and Farag 2018).


64

Figure 3-1 The scenarios that can occur during fiber processing. One fiber can be

gripped at two different points (A) or at one point (B or C). Breakage may occur in

scenario A, but not in B or C.

The mechanical resistance of fibers is affected by several factors. Immaturity is

the most common cause of weak spots in the fibers because of the reduced cellulose

deposition in the secondary cell wall or the lack of cellulose organization. These weak

spots require less energy to be broken. Therefore, if the energy added with processing is

high enough to break the weak spots, the fiber will break. (Y.-L Hsieh, Honik, and

Hartzell 1995; Hu and Hsieh 1996; Helmut Wakeham 1956; Meredith 1946).

Structural flaws in the secondary cell wall, like pits, pores, and voids, can also

make fibers more fragile to mechanical processing (Flint 1950; Magne, Portas, and

Wakeham 1947). These microfractures are defects that can weaken the secondary cell

wall. These flaws may reduce cellulose content and organization, making the breakage

more likely to occur.

High slenderness, the ratio between fiber length and diameter, can also increase

the propensity to break (Dever, Gannaway, and Baker 1988). Considering two fibers with

the same maturity but different diameters, the thinner the fiber, less cellulose will be

present in the secondary cell wall. A fully mature fiber with a small diameter will not

withstand the same energy than another fully mature fiber with a larger diameter.

Furthermore, the longer the fiber, the higher the probability of occurrence of weak spots

that can result in fiber breakage (Brown 1953a).


65

Reversals are parts of the fiber with high crystallinity, and breakage is more

common at reversals than between them (Helmut Wakeham and Spicer 1951; Harzallah,

Benzina, and Drean 2010; H. Wakeham, Radhakrishnan, and Viswanathan 1959). The

high content of crystalline cellulose reduces the elongation in these specific parts of the

fiber. The elevated brittleness at reversals makes these regions more prone to break.

A high coefficient of friction can increase the resulting force applied to a fiber,

increasing the propensity to break (Hosseinali and Thomasson 2018). These authors

hypothesize that fibers with high coefficients of friction need to receive more energy to

be pulled out of the tufts of fibers. The higher energy may result in more breakage and

increase of the short fiber content.

When fibers are broken, the obvious consequence is a reduction in the number of

longer fibers and an increase in the number of shorter fibers (Figure 3-2). However, a

modification in the fiber length distribution is not the only consequence of processing.

The number of neps may also increase with more processing. Breakage increases the

number of immature shorter fibers in a sample. This type of fiber is flexible and can

easily tangle with other fibers, creating neps (Pearson 1933; M. H. van der Sluijs and

Hunter 2016; J. J. Hebert, Boylston, and Thibodeaux 1988). An increase in the neps

number will degrade yarn strength and evenness, and increase the number of ends down

(M. H. van der Sluijs and Hunter 2016).


66

Figure 3-2 Length distribution by number of a cotton sample before and after mechanical

processing.

A higher number of shorter fibers will reduce uniformity, strength, and elongation

of drawn slivers, rovings, and yarns (Backe 1986; Patricia Bel-Berger and Von Hoven

1997; Helmut Wakeham 1955). Some authors hypothesize that short fibers cause all these

negative impacts because they are less aligned than longer fibers to the main axis of the

yarn (Childers and Baker 1978; Ülkü, Özipek, and Acar 1995). Other authors indicate

that shorter fibers do not have the same frictional force than longer fibers, resulting in

fiber slippage and poor yarn quality (Shorter 1957; Hosseinali and Thomasson 2018)

As previously mentioned, processing may alter the distributions by fiber breakage

and fiber removal. During cleaning, trash may entangle with lint. When the trash is

rejected by the cleaning device, some fibers may also be removed and be considered as a

waste (Morey et al. 1976; Pfeiffenberger 1944; Ramey and Beaton 1989). Since fibers are

removed from the fiber population, the fiber length distribution may be altered.

0

2

4

6

8

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Without processing With processing


67

3.1.2 Instruments to measure fiber length distribution

The standard instrument to measure the fiber length distribution is the Sutter-

Webb fiber comb sorter (ASTM 2012c). An operator gently brushes a sample against a

set of combs, sorts the fibers by length, and acquires information related to the fiber

length distribution. Although this is the standard method to measure fiber length

distribution, it is slow (Richardson, Bailey Jr., and Conrad 1937). Other instruments were

created to measure length and length variability.

The High Volume Instrument (HVI) is the most common machine used to classify

cotton quality. The HVI uses the fibrograph principle to measure a beard of fibers in a

comb (Delhom, Kelly, and Martin 2018). The HVI reports two fiber length parameters,

upper half mean length (UHML) and uniformity index (UI), the percent ratio between

mean length (ML) and UHML (Cotton Incorporated 2013; ASTM 2013b). The higher the

UI, the more uniform a cotton sample is expected to be.

The Advanced Fiber Information System (AFIS) is another device to measure

fiber length. A fiber sample of 500 mg is pulled into a hand-drawn sliver. The sliver is

fed to the AFIS fiber individualizer, a perforated spiked cylinder that separates trash and

individualized fibers. These fibers are carried by an airflow towards an optical sensor.

The optical sensor measures the time of the flight and the speed of the airflow,

calculating the length of individual fibers and creating a distribution by number (Shofner,

Baldwin, and Chu 1993; Shofner and Shofner 1999; Shofner et al. 1995).

Fiber length distribution modifications caused by processing may reflect in the

HVI length parameters. For example, if breakage happens, the UHML may be reduced.

However, this behavior will not occur for all cotton samples, and the analysis of the

whole distribution is needed to understand the processing effect. Figure 3-3 presents the

AFIS length distributions by number of two samples with the same UHML of 30.0 mm

with and without mechanical processing. Observing just the UHML, both samples appear

to have similar behaviors during processing. But when the whole fiber length distribution

is taken into account, the samples are drastically different. Sample A length distribution

is not affected much by processing while sample B is affected.


68

Figure 3-3 Fiber length distributions of two cotton samples with the same upper half

mean length (30.0 mm) with and without processing. Although the upper half mean

length did not change for both samples, sample A exhibits a smaller change in the whole

distribution than sample B.

3.1.3 Shirley analyzer

The Shirley analyzer is the standard instrument to determine non-lint content,

named trash, in a cotton sample (ASTM 2012d). There is a good correlation between the

trash content and combined card and picker waste (Pfeiffenberger 1944). A sample is

placed on a feeding tray and it is carried by a feed roller towards the lickerin, a spiked

cylinder. Removal of trash occurs when the lickerin breaks tufts of fiber. The mixture of

fibers and trash are beaten against a cleaning knife, aiding the separation between fibers

and trash. An airflow carries the light released fibers to a condenser but this airflow

cannot carry heavy trash particles. Trash falls into a tray and the fibers are condensed in a

chamber.

In order to clean the samples, mechanical processing is applied to the fibers. This

process results in fiber-fiber friction and fiber-metal friction. During cleaning, the fibers

pass through gaps between parts of the machine. The gaps between the feed roller and the

lickerin and between the lickerin and the cleaning knife are 0.1 mm of distance. All fibers

that are at least 0.1 mm long can potentially be gripped and stretched by the machine

(Figure 3-4) (ASTM 2012d).

0

4

8

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample A


0

4

8

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample B



69

Figure 3-4 Fiber length distribution (solid line) and the threshold (vertical dashed line)

for the length of fibers that can be gripped on both ends by the Shirley analyzer. Virtually

all fibers in a sample can be stressed.

In the Shirley analyzer, fibers can entangle with trash and be carried out to the

trash tray, resulting in fiber removal from the sample (Morey et al. 1976; ASTM 2012d).

The processing effect from a Shirley analyzer involves a combination between breakage

and rejection of fibers, altering the fiber length distribution.

The standard procedure to determine non-lint content is to pass the sample twice

through the Shirley analyzer (ASTM 2012d). There is evidence that one passage though

the Shirley analyzer may increase accuracy when the goal of the research is the lint itself,

not the trash (Shepherd 1961).

In this chapter, I will present an experimental procedure that was performed on a

set of cotton samples with diverse genetics. The aim of this experiment is to quantify the

Shirley analyzer processing impact on the fiber length distribution of each sample

measured by AFIS. If this experiment is successful, the procedure can be used to predict

modifications caused by processing in fiber length distribution of samples in future

research projects.

0

2

4

6

8

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)


70

3.2 Materials and methods

3.2.1 Germplasm development

A set of four mutated lines and three commercial varieties were crossed at Texas

Tech University, resulting in 12 populations of F1 seeds. The original seven parents were

chosen based on HVI UHML, aiming to create populations with variability for length

(Table 3-1).

Table 3-1 Crossing scheme with the parents used to develop the populations in this

research.

Paternal commercial varieties

Variety 1 Variety 2 Variety 3

Maternal

mutated

lines

1-1295 F1-1 F1-2 F1-3

1-136 F1-4 F1-5 F1-6

2-1 F1-7 F1-8 F1-9

2-738 F1-10 F1-11 F1-12

Line 1-1295 is a mutated line from TAM 94L-25 with higher UHML (32.00 mm); line 1-

136 is a mutated line from TAM 94L-25 with lower UHML (27.43 mm); Line 2-1 is a

mutated line from Acala 1517-99 with higher UHML (30.99 mm); line 2-738 is a

mutated line from Acala 1517-99 with lower UHML (27.94 mm); variety 1 is a

commercial variety with lower UHML (27.43 mm); variety 2 is a commercial variety

with average UHML (28.19 mm); variety 3 is a commercial variety with higher UHML

(30.23 mm).

The F1 seeds were planted and the 12 populations of hybrids were self-pollinated

to obtain F2 seeds. These F2 seeds were planted at Texas Tech Research farm in spring

2015, on loam soil with subsurface irrigation, following the recommended practices for


71

Lubbock county (Ayele, Kelly, and Hequet 2018). The season had an accumulated

rainfall of 470 mm and a total number of 1414 GDD15.6 units (Snowden et al. 2013).

A set of 720 F2 plants were selected based on agronomic performance. From this,

a subset of 240 plants were selected based on AFIS length distribution. The seeds were

delinted with sulfuric acid 20% and prepared for planting in 2016.

3.2.2 Field experiment

The set of 240 cotton F3 lines were planted at the Texas Tech University Lubbock

TX in the spring of 2016. The seeds (45 seeds per plot) were planted on loam soil with a

four-row cone planter. The depth was ~25.4 mm deep in single-row plots with subsurface

irrigation.

The rows were 4.6 m long, 1.0 m wide, and the alleys were 0.9 m long. There was

no replication due to the limited amount of seeds available. The accumulated rain

precipitation was 192 mm, and the total number of GDD15.6 thermal units was 1480 units

(Snowden et al. 2013). In fall, the experiment was stripper harvested with a John Deere

implement without field cleaner.

The harvest from each plot was placed in mesh bags. The bags were stored at the

Texas Tech University Fiber and Biopolymer Research Institute (FBRI) in Lubbock TX

at 21.0 ± 1.0°C and the relative humidity of 55 ± 5% for at least seven days before

ginning.

3.2.3 Mechanical processing

Seedcotton from each experimental plot was cleaned by hand to remove sticks

and burrs that could prevent good quality ginning. The cleaned samples were ginned

using a Compass 10-saw gin (Model MG1010) to obtain at least 200 g of lint (Figure 3-5.


72

Figure 3-5 Laboratory scale Compass 10-saw-gin, model MG1010.

A mass of approximately 200 g of cotton lint was obtained from each sample. The

lint was conditioned for at least two days at the standard atmospheric conditions of 21 ±

1°C and relative humidity of 65 ± 2% (ASTM 2016) before dividing each sample into

two subsamples of 100 g. One of the subsamples was kept as is, and the other subsample

was passed once through a Shirley analyzer from the Shirley Institute to add a processing

effect similar to lint cleaning at the gin (Figure 3-6).


73

Figure 3-6 Shirley analyzer from Shirley Institute.

Subsamples with and without processing by the Shirley analyzer were kept in the

standard atmosphere conditions for at least 72 hours prior to be tested for fiber quality

properties.


Subsamples with and without Shirley analyzer processing were tested by HVI

following a 4-4-10 protocol, which consists of four measurements of micronaire, four

measurements of trash/color, and ten measurements of length/strength/elongation. The


74

samples were also tested with the AFIS following a protocol of five replications of 3,000

fibers (Figure 3-7).

Figure 3-7 Flowchart of the experimental procedure to evaluate the processing effect.


75

3.2.5 Statistical analysis

Descriptive statistics were performed to evaluate the range of fiber properties

among samples with and without Shirley analyzer processing. A pair-wise t-test at 5% of

significance was performed between the two groups of samples (with and without Shirley

analyzer processing) for fiber property parameters.

Fiber length distributions obtained from the AFIS are frequency histograms

distributed over 40 length bins (Figure 3-8A). Each length bin has a range of 1.59 mm (B.

R. Kelly and Hequet 2018). The distributions were converted to the flipped first

cumulative distribution, resulting in length as the dependent variable (Figure 3-8B. The

difference between the processed and unprocessed distributions for each sample was

calculated, resulting in a difference histogram (Figure 3-8C). To summarize the variation

of the differences, the difference histograms were centered and scaled before performing

a principal component analysis (PCA) on the 240 samples.


76

Figure 3-8 Conversion from the original AFIS length distribution of subsamples with and

without processing with a Shirley analyzer (A) to the flipped first cumulative

distributions (B) and to the curve of differences between two subsamples (C).

Then, a correlation analysis was performed between HVI (Strength, elongation,

reflectance, yellowness) and non-length parameters from AFIS (immature fiber content,

maturity ration, fineness, standard fineness) to the principal components.


3.3.1 Range of parameters

The dynamic range over the 240 samples, i.e., the ratio between the difference of

the 95% and 5% quartiles and the mean, exhibits variation for all analyzed parameters

(Table 3-2). The parameters with the highest variation for both groups were SFCn and

IFC, while MR has the lowest variation for both groups.

0

2

4

6

8

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

A


0

25

50

0 50 100

Len

gth

(m

m)

Percentage (%)

B


-1.0

-0.5

0.0

0.5

1.0

0 50 100

Len

gth

dif

fere

nce

(m

m)

Percentage (%)

C


77

Table 3-2 Dynamic ranges between 95% and 5% quantiles for a set of 240 F3 lines tested

by AFIS and HVI, with and without Shirley analyzer processing.

Without Shirley analyzer

processing

With Shirley analyzer

processing

Parameter 5%

Quantile

95%

Quantile

Dynamic

Range (%)

5%

Quantile

95%

Quantile

Dynamic

Range (%)

Short fiber

content (%)

17.8 29.7 51.7 18.5 28.2 42.7

Mean length

by number

(mm)

19.8 23.9 18.8 20.1 23.6 16.0

Upper quartile

length by

weight (mm)

30.0 35.6 16.9 30.0 35.3 16.1

5% longer

fibers by

number (mm)

34.0 40.4 17.1 34.0 39.9 15.8

Fineness

(mtex)

149 178 17.9 147 175 17.5

Immature fiber

content (%)

4.0 7.3 61.1 4.3 7.5 55.2

Maturity ratio

(no unit)

0.87 0.96 9.9 0.85 0.94 10.0

Strength

(kN·m·kg-1)

283.5 336.4 17.1 281.6 333.5 17.0

Elongation (%) 5.6 8.8 45.7 6.1 8.9 38.3


78

Processing samples with the Shirley analyzer resulted in statistical differences for

all the analyzed parameters, except SFCn and Ln (Table 3-3). The reduction in the length

of the longer fibers, as demonstrated by UQLw and L5%, indicates that this type of

processing may be breaking fibers. The increase in the immature fiber content and

reduction in the fineness and maturity ratio is evidence that not all the fibers are breaking,

but the immature fibers.

Table 3-3 Average values for fiber quality properties of 240 F3 samples with and without

Shirley analyzer processing.

Parameter Without Shirley

analyzer processing

With Shirley

analyzer processing

Short fiber content (%) 23.0 22.7

Mean length by number (mm) 21.8 21.8

Upper quartile length by weight (mm)* 33.1 33.0

5% longer fibers by number (mm)* 37.4 37.5

Fineness (mtex)* 162 160

Immature fiber content (%)* 5.5 5.8

Maturity ratio (no unit)* 0.91 0.90

Strength (kN·m·kg-1)* 308.9 306.1

Elongation (%)* 7.0 7.3

*Parameters statistically different at 95% of confidence.

Immature fibers have a smaller strength than mature fibers (Y.-L. Hsieh, Hu, and

Nguyen 1997). Processing with the Shirley analyzer may break some fragile fibers in the

sample, increasing the within-sample number of short immature fibers. When immature

longer fibers are broken and the fragments stay in the sample, it is likely that the number


79

of shorter fibers in the fiber length distribution will also increase, affecting length

parameters extracted from the distribution of sampled fibers.

A reduction in the number of longer fibers should happen together with an

increase in the number of shorter fibers. Nevertheless, no difference was observed in the

SFCn, and Ln means. The next step in the research was to evaluate the whole fiber

length distributions, not just parameters extracted from the distributions to find an

explanation for this fact.

3.3.2 Fiber length distributions

The AFIS length distribution is a histogram of relative frequency per bin of 1.59

mm. The observation of the distributions of each sample with and without Shirley

analyzer processing reveals that some samples change more than others after processing

(Figure 3-9).


80

Figure 3-9 Examples of fiber length distributions for the same sample with and without

Shirley analyzer processing.

Some samples with little or no modification caused by processing have a local

peak of very short fibers (bin with a mean value of 2.3 mm) and a second peak towards

the longer fibers (higher length than the mean length). This case represents a sample with

a good length distribution for ring spinning, with low within-sample variation, longer

fibers, and low modification to the fiber length profile caused by processing (Figure 3-

9A).

Other samples with little or no modification caused by processing may only have

one peak at 2.3 mm and no second peak, which signals a very large number of broken

0

6

12

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

A


0

6

12

0 10 20 30 40 50

Per

centa

ge

Length (mm)

B


0

6

12

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

C


0

6

12

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

D



81

fibers. In the example of Figure 3-9B, the sample has a maturity ratio of only 0.82 and an

immature fiber content of 8.6%. The peak at 1.59 mm indicates a higher amount of fiber

fragments in comparison to the example of Figure 3-9A, indicating that this sample may

not perform well in ring spinning. Indeed, the high amount of shorter fibers may create

imperfections, reduce yarn strength, and increase the number of end-breaks during

spinning (Thibodeaux et al. 2008; Helmut Wakeham 1955).

A third type of sample has a degradation in the fiber length profile after the

Shirley analyzer processing (Figure 3-9C). Samples that exhibit large differences before

and after processing reveal a possible source of bias in breeding programs. For example,

when breeders are making individual plant selections, they typically use a laboratory-

scale gin with no post-ginning processing. Therefore, they select germplasm based on this

information only. If the breeder unknowingly develops a cultivar based on the selection

of an entry with a fiber length profile that degrades with post-ginning processing, it could

lead to a product with poor spinning performance, i.e., high short fiber content after

processing that was not revealed by testing non-post-processed samples.

In the dataset of 240 samples, there are samples whose fiber length profile

improved with processing (Figure 3-9D). In the three most extreme cases, the samples

presented a high immature fiber content (above 8.0%), low maturity ratio (below 0.87),

and high visible foreign matter (above 7.50%) with no processing. I hypothesize that the

improvement in the fiber quality profile happens when the sample has many immature

fibers. During processing, some of the immature fibers may stick with the trash and be

expelled from the sample into the trash tray of the Shirley analyzer. If this holds true, the

efficiency of the Shirley analyzer to study breakage would be compromised. The

processing effect is sample-dependent. Therefore, fiber breakage may be more important

for some samples and fiber removal for other samples. In this case, the researcher cannot

isolate the contribution of these two effects.

Even if it is not possible to isolate fiber breakage and fiber removal after

processing, it is possible to study how the processing alters fiber length distributions. The

next step in the research was to quantify and understand how the Shirley analyzer

processing impacts the variation in fiber length distributions.


82

3.3.3 Differences between the fiber length distributions

A total of 99.33% of variation from the principal component analysis (PCA) of

the 240 difference histograms can be explained with three principal components (Table 3-

4). The first principal component (PC) is more influenced by loadings resulting from the

first half of the distribution. The second PC is more affected by bins in two areas of the

curve, at the beginning and the end of the distribution. The third PC has higher loadings

in three regions of the curve, in the beginning, middle, and end of the distribution (Figure

3-10). Because of parsimony, I decided to analyze how the two first PCs, which explain

97.84% of the variation, are related to modifications in the fiber length distribution

caused by Shirley analyzer processing.

Table 3-4 Explained variation per principal component from the principal component

analysis performed with the difference between the flipped cumulative distributions with

and without processing with a Shirley analyzer of 240 samples.

Principal component Explained variability (%) Cumulative explained variability (%)

PC1 91.92 91.92

PC2 5.92 97.84

PC3 1.49 99.33


83

Figure 3-10 PCA loadings for the bins of the difference between the flipped cumulative

distributions with and without processing with a Shirley analyzer of 240 samples.

I analyzed the biplots of the scores between PC1 and PC2, keeping the PC2 score

as constant as possible and observing two samples with contrasting PC1 scores. Sample

A has a PC1 score of -0.478 and a PC2 score of -0.011. Sample B has a PC1 score of

0.547 and a PC2 score of -0.021. Comparing the graphs of the differences (Figure 3-11),

sample A has a graph with only negative values for length difference, while sample B has

only positive values for length differences. In the graphs of the flipped cumulative

distributions, there is no crossover between the curves of both samples. This is an

indication of changes in amplitude for the curves. Comparing the original fiber length

distributions, sample A presents an improvement of the length distribution, while sample

B presents a degradation.

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0 30 60 90

Lo

adin

gs

Percentage (%)

PC1 PC2 PC3


84

Figure 3-11 Curves of the difference between the flipped first cumulative distributions of

subsamples with and without processing with a Shirley analyzer (I), the flipped first

cumulative distributions (II) and the original length distributions (III) of two samples

with contrasting scores for PC1.

The SFCn decreased for sample A while Ln and L5% increased (Table 3-5).

Modifications for sample B are in the opposite direction (Table 3-5). Parameters related

to the maturity/fineness complex are also altered. The IFC decreased for sample A while

MR and H increased. Sample B presented modifications in the opposite direction.

-4.0

-2.0

0.0

2.0

4.0

0 30 60 90

Len

gth

dif

fere

nce

(m

m)

Percentage (%)

Sample A (I)

-4.0

-2.0

0.0

2.0

4.0

0 30 60 90

Len

gth

dif

fere

nce

(m

m)

Percentage (%)

Sample B (I)

0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample A (II)


0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample B (II)


0

3

6

9

0 10 20 30 40 50Per

centa

ge

(%)

Length (mm)

Sample A (III)


0

3

6

9

0 10 20 30 40 50Per

centa

ge

(%)

Length (mm)

Sample B (III)



85

Table 3-5 Fiber quality parameters for samples A and B, with and without processing

with a Shirley analyzer.

Samples Sample A Sample B

Parameters Without

processing

With

processing

Without

processing

With

processing

Short fiber content by

number (%)

24.4 19.2 24.2 30.0

Mean length by number

(mm)

21.6 23.6 22.6 20.6

5% longer fibers by

number (mm)

37.6 39.1 38.9 38.1

Immature fiber content

(%)

6.5 5.5 5.2 6.1

Maturity ratio

(no unit)

0.89 0.90 0.92 0.90

Fineness (mtex) 161 164 176 171

The alterations in the length distribution and quality parameters are indications

that the Shirley analyzer processing removes fibers from sample A, especially short

immature fibers. Longer and more mature fibers are left in the population, resulting in an

improvement in the distribution. Sample B was altered by the Shirley analyzer processing

in a different way. The overall reduction in fiber length with an increase of finer and

immature fibers are indications that breakage is the main factor. The observed changes

for length distributions are in accordance with the expected consequences of fiber

breakage.

The IFC has the highest coefficient of correlation with the PC1 (-0.563). Although

the value is moderate, it is additional evidence that maturity is an important factor related

to the variation captured by PC1.

A similar analysis was performed for PC2. Sample C has a PC1 score of 0.284

and a PC2 score of 0.121 and sample D has a PC1 score of -0.256 and a PC2 of -0.284.

Comparing the graphs of the differences, both samples present a crossover on the x-axis

(Figure 3-12). Sample C has positive values up to 69.2% and sample D has negative

values up to 59.0%. This difference is related to a crossover between the flipped


86

cumulative distributions for the subsamples with and without processing. The difference

is also related to changes in the shape of the original distributions. After Shirley analyzer

processing, sample C lost the peak at 27 mm while sample D developed a peak at 27 mm.

Figure 3-12 Curves of the difference between the flipped first cumulative distributions of

subsamples with and without processing with a Shirley analyzer (I), the flipped first

cumulative distributions (II) and the original length distributions (III) of two samples

with contrasting scores for PC2.

The changes in the whole fiber length distributions are more complex for PC2

than for PC1. The Ln for sample C decreased, while SFCn and L5% increased.

-4.0

-2.0

0.0

2.0

4.0

0 30 60 90

Len

gth

dif

fere

nce

(m

m)

Percentage (%)

Sample C (I)

-4.0

-2.0

0.0

2.0

4.0

0 30 60 90

Len

gth

dif

fere

nce

(m

m)

Percentage (%)

Sample D (I)

0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample C (II)


0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample D (II)


0

3

6

9

0 10 20 30 40 50

Len

gth

(m

m)

Length (mm)

Sample C (III)


0

3

6

9

0 10 20 30 40 50Per

centa

ge

(%)

Length (mm)

Sample D (III)



87

Modifications for sample D presented the opposite direction. The IFC increased for

sample C, while the MR and H decreased. The IFC and MR decreased for sample D,

while the H decreased (Table 3-6).

Table 3-6 Fiber quality parameters for samples C and D, with and without processing

with a Shirley analyzer.

Samples Sample C Sample D

Parameters Without

processing

With

processing

Without

processing

With

processing

Short fiber content by

number (%)

19.6 24.0 24.1 18.7

Mean length by number

(mm)

22.1 21.3 22.1 22.6

5% longer fibers by

number (mm)

36.6 37.1 38.9 36.3


(%)

4.5 6.0 5.3 4.7

Maturity ratio

(no unit)

0.91 0.90 0.93 0.92

Fineness (mtex) 170 161 161 174

Modifications in the length for sample C indicate that some fibers are broken,

increasing the SFC and reducing the Ln. Breakage is also related to the reduction of MR

and H while increasing the IFC. However, some of the process-generated fibers are also

removed, increasing L5%. In sample D, longer fibers are broken, increasing the number

of average length fibers, but the number of shorter fibers is not increasing. Reduction in

the IFC and increment of H are compatible with fiber removal.

There is no significant coefficient of correlation among the analyzed fiber quality

parameters and the PC2. I have two hypotheses for this fact. My first hypothesis is that a

combination of parameters is responsible for the variation captured by PC2. My second

hypothesis is that another parameter, not commonly reported by HVI neither AFIS, can

be related to the variation captured by PC2.


88

Shirley analyzer processing has two main factors that modify a fiber length

distribution: fiber breakage and fiber removal. I hypothesize that when one factor is

dominant, the amplitude of the fiber length distribution is affected in certain areas, such

as in the middle length fibers region, and the PC1 captures this modification. If both

factors impact processing, the shape of the length distribution will be affected in certain

areas, such as the shorter length fiber region, and the PC2 captures this modification.

3.4 Conclusions

The Shirley analyzer applies mechanical stress similar to the stress of a card to

remove trash. The small feeler gauge between the pieces of this machine may virtually

allow all the fibers in the sample to be gripped and stressed during processing, causing

fiber breakage. The card-like action of the Shirley analyzer may also remove some fibers

with the trash from the population.

The processing effect is sample- and machine-dependent. Some samples may

have more fiber breakage than fiber removal and other samples may show the opposite. I

observed that these two factors with the Shirley analyzer processing cannot be separated

using only one passage of the sample through this machine, followed by the fiber length

analysis with an AFIS.

A future step in this research is the study of how the fiber length changes in a set

of samples after multiple passages through the Shirley analyzer. In a sample, the fibers

may break with the initial passage until a limit is reached, and further modifications will

be caused only by the removal of fibers. This is a scenario that may occur in samples with

low trash and low maturity. In other samples, all the removable fibers may be eliminated

with the initial passage until a limit is reached and further modifications will be caused

only by fiber breakage. This is a scenario that may happen in samples with high trash

content and high maturity.

In this study, I defined a mathematical space where the difference between fiber

length flipped cumulative distributions can be compared, using principal component

analysis. I also defined the meaning of the first two principal components in regard to

modifications in the fiber length distributions and explained how two factors of


89

processing, fiber breakage and removal, affect the distributions. The first principal

component is related to amplitude differences caused when only fiber breakage or only

fiber removal is dominant. The second principal component is related to shape

differences resulting from a combined effect of fiber breakage and fiber removal.


90

3.5 Bibliography



———. “ASTM Standard D7139-09 Standard Terminology for Cotton Fibers.” In

ASTM, 5. West Conshohocken: ASTM International, 2013.

———. “D1440-07 Standard Test Method for Length and Length Distribution of Cotton

Fibers (Array Method).” In Annual Book of ASTM Standards, 1–6. West






(2018): 47–55.


Backe, Everett E. “Effect of Short Fiber Content in Cotton on Plant Performance and

Quality.” Textile Research Journal 56, no. 2 (1986): 112–115.



(1991): 83–88.

Bel-Berger, Patricia, and Terri Von Hoven. “Effects of Mechanical Cleaning on Cotton

Fibers: Part III: Effects of Card Wire Condition on White Specks.” Textile Research

Journal 67, no. 2 (1997): 857–865.

Benzina, H., E. Hequet, N. Abidi, J. Gannaway, J. Y. Drean, and O. Harzallah. “Using

Fiber Elongation to Improve Genetic Screening in Cotton Breeding Programs.”

Textile Research Journal 77, no. 10 (2007): 770–778.

Brown, Hugh M. “Correlation of Yarn Strength with Fiber Strength Measured at

Different Gauge Lengths.” In The Cotton Research Clinic, edited by National

Cotton Council of America, 25–31. Memphis: National Cotton Council of America,

1953.

Childers, Roy E, and Roy V Baker. “Effect of Moisture Conditioning on Ginning

Performance and Fiber Quality of High Plains Cotton.” American Society of

Agricultural and Biological Engineers 21, no. 2 (1978): 0379–0384.




booklet.pdf.

Delhom, Christopher D., Brendan Kelly, and Vikki Martin. “Physical Properties of

Cotton Fiber and Their Measurement.” In Cotton Fiber: Physics, Chemistry and

Biology, edited by D Fang, 41–73. Cham: Springer International Publishing, 2018.


91

Dever, J. K., J. R. Gannaway, and R. V. Baker. “Influence of Cotton Fiber Strength and

Fineness on Fiber Damage during Lint Cleaning.” Textile Research Journal 58, no.

8 (1988): 433–438.

Flint, E. A. “The Structure and Development of the Cotton Fibre.” Biological Reviews 25,

no. 4 (1950): 414–434.

Harzallah, O., H. Benzina, and J-Y. Drean. “Physical and Mechanical Properties of

Cotton Fibers: Single-Fiber Failure.” Textile Research Journal 80, no. 11 (July 6,

2010): 1093–1102. Accessed April 16, 2018.


Hebert, J. J., E. K. Boylston, and D. P. Thibodeaux. “Anatomy of a Nep.” Textile


Hequet, Eric F., Bobby Wyatt, Noureddine Abidi, and Devron P. Thibodeaux. “Creation

of a Set of Reference Material for Cotton Fiber Maturity Measurements.” Textile






Hosseinali, Farzad, and J. Alex Thomasson. “Variability of Fiber Friction among Cotton

Varieties: Influence of Salient Fiber Physical Metrics.” Tribology International 127

(November 1, 2018): 433–445. Accessed March 8, 2019.

https://www.sciencedirect.com/science/article/pii/S0301679X18303219.

Hsieh, You.-Lo., X.-P. Hu, and A. Nguyen. “Strength and Crystalline Structure of

Developing Acala Cotton.” Textile Research Journal 67, no. 7 (July 1, 1997): 529–

536. Accessed April 16, 2018.


Hsieh, You.-Lo, E Honik, and M M Hartzell. “A Developmental Study of Single Fiber

Strength: Greenhouse Grown SJ-2 Acala Cotton.” Textile Research Journal 65, no.

2 (1995): 101–112.

Hu, Xiao-Ping, and You-Lo Hsieh. “Crystalline Structure of Developing Cotton Fibers.”

Journal of Polymer Science Part B: Polymer Physics 34, no. 8 (June 1, 1996):

1451–1459. Accessed April 23, 2018.

http://doi.wiley.com/10.1002/%28SICI%291099-

0488%28199606%2934%3A8%3C1451%3A%3AAID-POLB8%3E3.0.CO%3B2-

V.







92




Magne, Frank C., H. J. Portas, and Helmut Wakeham. “A Calorimetric Investigation of

Moisture in Textile Fibers.” Journal of the American Chemical Society 69, no. 8

(1947): 1896–1902.

Meredith, R. “13-Molecular Orientation and the Tensile Properties of Cotton Fibres.”

Journal of the Textile Institute Transactions 37, no. 9 (1946): T205–T218.

Mogahzy, Yehia E. El, Roy M. Broughton Jr, and Hong Guo. “Evaluating Staple Fiber

Processing Propensity Part I: Processing Propensity of Cotton Fibers.” Textile


Mogahzy, Yehia E. El, Roy Broughton, and W.K. Lynch. “A Statistical Approach for

Determining the Technological Value of Cotton Using HVI Fiber Properties.”

Textile Research Journal 60, no. 9 (September 2, 1990): 495–500.


Mogahzy, Yehia, and Ramsis Farag. “7 - Tensile Properties of Cotton Fibers: Importance,

Research, and Limitations.” In Handbook of Properties of Textile and Technical

Fibres, edited by Anthony R. Bunsell, 223–273. 2nd ed. Cambridge: Woodhead

Publishing, 2018.

http://www.sciencedirect.com/science/article/pii/B9780081012727000079.

Morey, P. R., R. M. Bethea, P. J. Wakelyn, I. W. Kirk, and M. T. Kopetzky. “Botanical

Trash Present in Cotton before and after Saw-Type Lint Cleaning.” American

Industrial Hygiene Association Journal 37, no. 6 (1976).

Pearson, Norma L. Technical Bulletin No. 396 - Neps and Similar Imperfections in

Cotton. Washington, DC: United States Department of Agriculture, 1933.

https://books.googleusercontent.com/books/content?req=AKW5QaeFjgCBSRij-

4VXgIOBccPwYSlKr7Vxkf_7DwwZGVQEgCQBiHjrNoEBNv3lwx4jziZHhyTxF9

zj3jkSm8HR2qQjbvYhSU9kn1-

kP_WZSMMlp9nR9gT62LITSQc_KWLk8pEktfE7rsJkjpWSh5ZxztQQ3p1i8-

mrsDHD1VtFOQ6ie2ud_784h3tfkxt4VNcrFyNvP.



Ramey, Harmon H., and Paul G. Beaton. “Relationships between Short Fiber Content and

HVI Fiber Length Uniformity.” Textile Research Journal 59, no. 2 (February 2,


Richardson, Howard B, T. L. W. Bailey Jr., and Carl M Conrad. Methods for the

Measurement of Certain Character Properties of Raw Cotton - Technical Bulletin

545. 1st ed. Washington, DC: USDA, 1937.

https://naldc.nal.usda.gov/download/CAT86200540/PDF.

Shepherd, Jacob V. “Non-Lint Content Measurement by the One-Pass Shirley Analyzer


93

Procedure.” Textile Research Journal 31, no. 1 (1961): 75–78.








USA, 1999.

Shorter, S. A. “9—The Elements of a Unified Theory of Yarn Structure and Strength.”

Journal of the Textile Institute Transactions 48, no. 4 (April 1957): T99–T108.


van der Sluijs, M.H.J., and L. Hunter. “A Review on the Formation, Causes,

Measurement, Implications and Reduction of Neps during Cotton Processing.”

Textile Progress 48, no. 4 (2016): 221–323.




Thibodeaux, D, H Senter, J L Knowlton, D Mcalister, and X Cui. “The Impact of Short

Fiber Content on the Quality of Cotton Ring Spun Yarn.” The Journal of Cotton

Science 12, no. 4 (2008): 368–377.

Ülkü, S., B. Özipek, and M. Acar. “Effects of Opening Roller Speed on the Fiber and

Yarn Properties in Open-End Friction Spinning.” Textile Research Journal (1995).

Wakeham, H., T. Radhakrishnan, and G.S. Viswanathan. “X-Ray Diffraction Patterns of

Cotton at and between Reversals.” Textile Research Journal 29, no. 5 (May 2,


Wakeham, Helmut. “Cotton Fiber Length Distribution— an Important Quality Factor.”

Textile Research Journal 25, no. 5 (May 2, 1955): 422–429.


———. “Cotton Quality and Fiber Properties Part VI: Comparison of a Matched Pair of

Raingrown and Irrigated Cottons.” Textile Research Journal 26, no. 12 (December

2, 1956): 925–936. http://journals.sagepub.com/doi/10.1177/004051755602601203.

Wakeham, Helmut, and Nancy Spicer. “The Strength and Weakness of Cotton Fibers.”

Textile Research Journal 21, no. 4 (April 2, 1951): 187–194.



94

CHAPTER IV

PROCESSING EFFECTS OF AFIS AND SHIRLEY ANALYZER ON

COTTON SAMPLES

4.1 Introduction

The Shirley analyzer has a card-like design (Pfeiffenberger 1944; ASTM 2012d).

When a sample is processed by a card, the fibers have two different fates. The card may

reject some fibers with trash or may send them to form a card web. A similar process

happens with the samples processed with the Shirley analyzer. There are two chambers in

the Shirley analyzer where the fibers can be sent (Figure 4-1). If the fibers are not

removed from the process, they pass through the condenser (Figure 4-1B) and go to a

chamber in the back of the instrument. If the fibers are removed together with the trash,

they fall in a tray (Figure 4-1C).

Figure 4-1 A cotton sample is placed on the feed tray of the Shirley analyzer (A).

Cleaned fibers follow to the condenser (B), while rejected fibers go to the trash tray (C).

The fiber removal is not the only factor that can act during the Shirley analyzer

processing. The fibers are also gripped and strained by the moving parts of the

instrument. It may lead to some fiber breakage.


95

Fiber breakage is a problem for the cotton industry. Some degree of fiber

breakage is expected during any industrial processing. The broken fibers will modify the

fiber length distribution, reducing the yarn mechanical properties, and increasing the

number of yarn imperfections (Helmut Wakeham 1955; Thibodeaux et al. 2008; Tallant

et al. 1961).

The creation of predictive models for yarn spinning based on fiber properties is an

important tool for spinners (Yang and Gordon 2017; Cai et al. 2013). The prediction

power of these models can be reduced if there are differences between the properties of

the measured population of fibers and the properties of the population of fibers that end

up in the yarn. Therefore, the understanding of the breakage phenomenon is an important

topic for cotton research (Tallant, Fiori, and Landstreet 1960; Tallant, Fiori, and

Legendre 1959)

The Shirley analyzer may not be a good instrument if researchers need to isolate

and study cotton fiber breakage. In the previous chapter, I demonstrated that in a diverse

set of samples, the Shirley analyzer processing effect may cause different types of

modifications of the fiber length distributions for different samples. These modifications

may result from fiber breakage, fiber removal, or both. The interaction of both factors in

the same dataset may hinder researchers to draw proper conclusions. An alternative

approach to solving or minimizing this problem is to use a device that does not remove

fibers during processing, limiting the fiber length modifications to the fiber breakage.

The Advanced Fiber Information System (AFIS) separates trash and

individualizes fibers with a perforated pinned cylinder (Figure 4-2A) before sending

(Figure 4-2B) individual fibers to the optical sensor (Figure 4-2C) to be tested. This

cylinder is named AFIS individualizer. The AFIS individualizer has a diameter of 6.35

cm and spins at 7,500 rpm, resulting in a linear velocity of around 25 m/s (Shofner and

Shofner 1999; Shofner, Baldwin, and Chu 1993; Shofner et al. 1995). For a cotton fiber

of 25 mm and a linear density of 170 mtex, this velocity results in an added energy of

approximately 1.3 mJ. This energy may stress and break fibers before the fibers are tested

by the optical sensors.


96

After testing, all individualized fibers and trash are sent to a single chamber

(Figure 4-2D). This design eliminates or minimizes fiber loss during processing. If the

machine is stopped after testing a sample, the fibers can be recovered from the chamber

and passed again through the AFIS. In this case, only one type of processing stress is

applied, and modifications in the fiber length distribution may be limited to fiber

breakage. Therefore, the AFIS may be considered not only as a measuring instrument but

also as a processing device for cotton fiber breakage research.

Figure 4-2 Parts of an Advanced Fiber Information System (AFIS). The pinned

perforated cylinder (A) spins at high speed and separates fibers and trash. Fibers are sent

through the plastic tube (B) to the optical sensor (C) where they are tested. After testing,

fibers and trash are discarded in a single chamber (D).

The objective of this chapter is to compare and understand the differences caused

by the processing effect of the Shirley analyzer and the AFIS on a set of cotton samples.

If modifications in the fiber length distributions of samples are caused only by fiber


97

breakage in the AFIS, this instrument can become a candidate tool for further research

focusing only on fiber breakage.

4.2 Material and methods

4.2.1 Cotton samples

In the previous chapter, I studied a set of 240 cotton samples with diverse fiber

quality properties. I defined a mathematical space using principal component analysis and

quantified the variation in the two first principal axes. For the experiment in this chapter,

I selected a subset of 27 samples from the set of 240 samples. The selection was made to

have samples with a range of values in the scores for the first two principal axes of the

defined mathematical space.

A mass of approximately 100 g of raw cotton lint was conditioned for at least two

days at the standard atmospheric conditions of 21 ± 1°C and relative humidity of 65 ± 2%

(ASTM 2016). For each raw sample, subsamples of 0.50 g were taken to hand-draw

slivers (Figure 4-3).


98

Figure 4-3 Preparation of a sliver for AFIS testing. After weighing 0.50 g of cotton (A),

the sample is gently pulled (B) and rolled (C) to form a sliver with a length of

approximately 30 cm (D).

4.2.2 Experimental procedure

The processing effect of each instrument was determined using three different

experimental procedures (Figure 4-4). After characterizing the processing effect for the

Shirley analyzer and the AFIS, the processing effects of these instruments were

compared.


99

Figure 4-4 Flowchart of the experimental procedure to compare the Shirley analyzer and

the AFIS processing effects.

To determine the Shirley analyzer effect (Figure 4-4, treatment X), five

subsamples of 0.50 g per sample were initially drawn into five slivers, and they were

tested with the AFIS for 3,000 fibers per sliver (Fiber length distribution I). The rest of

the sample was passed once through the Shirley analyzer from the Shirley Institute. After

applying the processing effect, each sample of processed lint was tested again by the

AFIS. The same protocol of five replications of 3,000 fibers (Fiber length distribution II)

was used. The instrument reported for each sample a fiber length distribution based on

the relative frequency histogram of 15,000 tested fibers into 40 bins of 1.59 mm.

The AFIS processing effect (Figure 4-4, treatment Y) was calculated using two

passages through the AFIS. Initially, five subsamples of 0.50 g per sample were drawn

into five slivers, and they were tested with the AFIS for 3,000 fibers per sliver (Fiber


100

length distribution I). The lint was carefully recovered from the trash chamber to avoid

bringing back the trash that was separated by the AFIS individualizer. The chamber was

vacuumed to remove the trash before the next passage. The recovered lint was used to

prepare three new subsamples of 0.50 g. These subsamples were tested with the AFIS for

3,000 fibers each per sliver per sample (Fiber length distribution III). The instrument

reported a fiber length distribution based on 15,000 tested fibers for each sample in the

first passage and based on 9,000 tested fibers for each sample in the second passage.

4.2.3 Statistical analysis

The differences among the fiber quality parameters nep count, dust count, trash

count, visible foreign matter (VFM), the 5% longer fibers by number (L5%), mean length

by number (Ln), short fiber content by number (SFCn), fineness, immature fiber content

(IFC) and maturity ratio (MR) for fiber length distributions I, II, and III were analyzed

using the ANOVA (analysis of variance) procedure with the software JMP Pro 14.0. If

needed, percentage parameters were converted to the arcsine of the square root of the

percentage before analysis, and count parameters were converted to the logarithm basis

10 before analysis. If there were statistical differences among the samples for a

parameter, a Tukey’s HSD (honestly significant difference) at 5% of significance was

performed for that parameter.

The fiber length distributions were converted to the flipped cumulative

distribution to turn the length into the dependent variable. The difference between fiber

length distribution I and II (Figure 4-4) for each sample was calculated for each bin to

determine the Shirley analyzer effect (treatment X) and the difference between

distributions I and III (Figure 4-4) for each sample was calculated for each bin to

determine the AFIS effect (treatment Y). This analysis resulted in difference histograms

as obtained in the previous chapter (Figure 3-8C). To summarize the variation of the

differences, the difference histograms were centered and scaled before performing a

principal component analysis (PCA) on the 54 datapoints.


101


4.3.1 Descriptive statistics of AFIS fiber quality parameters

The three experimental procedures cleaned and stressed the samples (Table 4-1).

The raw samples, which were stressed only once with the AFIS individualizer, have the

highest amount of trash and the lowest number of neps (fiber length distribution I). One

passage through the Shirley analyzer and one passage through the AFIS individualizer

(fiber length distribution II) decreased the trash parameters to the lowest values and

increased the number of neps. Two passages through the AFIS individualizer (fiber

length distribution III) resulted in the highest number of neps and intermediate value for

trash parameters.

There is an explanation for the intermediate values for trash parameters in the

fiber length distribution III. The Shirley analyzer is a cleaning machine, and the trash

does not follow to the same chamber where cleaned fibers are sent, resulting in the lowest

values for trash parameters. Trash and fibers go to the same chamber in the AFIS. When

the fibers are recovered for the second passage, some trash comes together.

The AFIS individualizer is a high-speed device. The increase in the number of

neps is expected. The Shirley analyzer also stresses the fibers and increase the number of

neps. It was not possible to determine if the Shirley analyzer creates fewer neps than the

AFIS individualizer or if the Shirley analyzer creates and eliminates neps together with

the trash.


102

Table 4-1 Parameters related to processing effect and trash content for samples with

different processing effects.

Nep/g (count) Visible foreign

matter (%)

Dust/g (count) Trash/g (count)

Fiber length

distribution I 209a 9.80c 2041c 603c

Fiber length

distribution II 271b 0.48a 118a 31a

Fiber length

distribution III 524c 1.13b 199b 71b

Means followed by the same letter in the column are not statistically different at 95% of

confidence.

Fiber length distribution I: samples passed once through the AFIS. Fiber length

distribution II: samples passed once through the Shirley analyzer and once through the

AFIS. Fiber length distribution III: samples passed twice through the AFIS.

The different processing effects cleaned the samples and altered the number of

neps. The instruments also modified parameters extracted from the fiber length

distribution (Table 4-2). Samples with only one passage through the AFIS individualizer

(fiber length distribution I) presented a short fiber content by number (SFCn) and a mean

length by number (Ln) statistically different from the samples with two passages through

the AFIS. The AFIS individualizer may break fibers that are gripped between the feed

roller and the spinning spiked cylinder (B. Kelly 2014). Some fibers do not break in the

first passage but they may be stressed with the processing. The design of the AFIS

minimizes the rejection of fibers and when the fibers are recovered in the chamber, the

fragilized fibers are also present. When a sample is passed through the AFIS for the

second time, the fragilized fibers from the previous passage may break, reducing the

mean length by number and increasing the short fiber content by number.

Processing with one passage through the Shirley analyzer and one passage

through the AFIS individualizer (fiber length distribution II) does not create a statistical

difference for the mean length by number and the short fiber content by number in

comparison to the procedure with only one passage through the AFIS individualizer


103

(Table 4-2). I hypothesize that the Shirley analyzer fiber breakage effect reduces the

mean length by number and increases the short fiber content by number in a sample.

Some fibers that do not break with the Shirley analyzer processing may be fragilized and

they would break when processed with the AFIS individualizer. Nevertheless, the Shirley

analyzer fiber removal effect eliminates some of the shortened and fragilized fibers from

the sample, minimizing the fiber breakage effect. When the sample processed with the

Shirley analyzer is processed with the AFIS, the difference observed for the mean length

by number and short fiber content by number in comparison to a sample with only the

AFIS processing is not statistically significant.

Table 4-2 Fiber length parameters from the distributions for samples with different

processing effects.

Short fiber content

by number (%)

Mean length by

number (mm)

5% longer fibers by

number (mm)

Fiber length

distribution I 28.1a 20.5a 36.8

Fiber length

distribution II 29.4ab 20.1a 36.6

Fiber length

distribution III 32.1b 19.0b 35.7

Means followed by the same letter in the column are not statistically different at 95% of

confidence.




The 5% longer fibers by number (L5%) exhibits a trend of reduction from fiber

length distribution I to III, but this difference is not statistically significant (p-value =

0.059). Fibers contributing to calculate the L5% are sturdy fibers, more mature, and more

resistant to breakage than fibers that contribute to the Ln (Hosseinali 2012; Dever,

Gannaway, and Baker 1988; Kothari et al. 2017). Therefore, less breakage will happen in

the longer fibers than with the average length fibers.


104

Although there are differences in the fiber length parameters, no statistical

differences among the treatments were observed for parameters related to the complex

fineness/maturity (Table 4-3). There is a trend to increase the immature fiber content

(IFC) and decrease the maturity ratio (MR) from the fiber length distribution I to III. This

is expected because the original output from AFIS is a report of tested fibers by number.

If the immature fibers are breaking, their number is increasing, and this fact may reflect

on the parameters related to maturity.

Table 4-3 Fiber length parameters related to the fineness/maturity complex for samples

with different processing effects.

Fineness

(mtex)


(%)

Maturity ratio

(no unit)

Fiber length

distribution I 164 5.9 0.90

Fiber length

distribution II 162 6.3 0.89

Fiber length

distribution III 165 6.5 0.88




4.3.2 Processing effect in the fiber length distributions

The processing effect has different impacts on the samples. The difference

between the fiber length distributions of a sample passed once through the AFIS

individualizer (fiber length distribution I) and passed once through the Shirley analyzer

and then through the AFIS individualizer (fiber length distribution II), the treatment X

(Figure 4-5), presented a similar behavior to what was observed in the previous chapter.

Within the data set, there are samples with positive or negative values for the PC1 and

PC2 scores (Figure 4-5). The spread of score values for the difference calculated between

fiber length distribution I and fiber length distribution II samples is similar to the spread


105

observed in the previous chapter and it may confirm our hypothesis that two effects are

happening. In the case of the Shirley analyzer, it appears that the processing effect has a

contribution from the fiber breakage and fiber removal.

A different situation happens with the samples for the comparison based on the

difference between the fiber length distributions of a sample passed once through the

AFIS (fiber length distribution I) and passed twice through the AFIS (fiber length

distribution III), the treatment Y (Figure 4-5). Samples from this treatment resulted in

only positive values for PC1 (Figure 4-5). The grouping of all points with AFIS

processing effect in one quadrant of the mathematical space is evidence that only one

effect is happening. In the case of the AFIS individualizer, it appears that the processing

effect has a contribution only from the fiber breakage.

Figure 4-5 Scatterplot of scores for principal components 1 and 2 for samples with

different processings effects. Treatment X is the difference between the fiber length

distribution of a sample with one passage through the AFIS individualizer and the fiber

length distribution of a sample with one passage through the Shirley analyzer and one

passage through the AFIS individualizer. Treatment Y is the difference between the fiber

length distribution of a sample with one passage through the AFIS individualizer and the

fiber length distribution of a sample with two passages through the AFIS individualizer.

-0.6

-0.3

0.0

0.3

-0.4 0.0 0.4 0.8

PC

2

PC1

Treatment X Treatment Y


106

I checked the fiber length distributions and the differences in these distributions. I

will present some examples of the modifications observed in this set of samples. Sample

A is a sample with a similar modification caused by both treatments. This sample has a

score value of 0.461 for PC1 and -0.163 for PC2 in treatment X (Shirley analyzer effect)

and 0.456 for PC1 and -0.148 for PC2 in treatment Y (AFIS effect). Based on the results

from the previous chapter, the positive value for PC1 is related to the reduction in the

amplitude of the fiber length distribution, reduction of the mean length by number, and

increase of the short fiber content by number. The difference observed in the fiber length

distributions for both treatments show a reduction of the fiber length distributions

amplitudes (Figure 4-6). There is no crossover between the flipped cumulative fiber

length distributions. It seems that the breakage factor has a similar impact on the fiber

length distributions for this sample independently of the instrument used.


107


Figure 3-6 Curves of the flipped first cumulative distributions (I) and the original length

distributions (II) of sample A. Treatment X is the difference between the fiber length

distribution of the sample with one passage through the AFIS individualizer (FL I) and

the fiber length distribution of the sample with one passage through the Shirley analyzer

and one passage through the AFIS individualizer (FL II). Treatment Y is the difference

between the fiber length distribution of the sample with one passage through the AFIS

individualizer (FL I) and two passages through the AFIS individualizer (FL III).

Not all the samples presented the type of modification observed in the sample A.

There are samples with one type of modification for one processing effect and another

type of modification for another processing effect. Sample B has a PC1 score value of -

0.258 and a PC2 score value of 0.056 for treatment X. Based on the results from the

previous chapter, the negative score for PC1 is related to an increment in the amplitude of

the shape for the distribution. This increase in the amplitude results in higher mean length

0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample A (I)

FL I FL II

0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample A (I)

FL I FL III

0

3

6

9

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample A (II)

FL I FL II

0

3

6

9

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample A (II)

FL I FL III


108

by number and smaller short fiber content by number. Nevertheless, this same sample has

a reduction in the amplitude of the fiber length distributions in the treatment Y (Figure 4-

7). The PC1 score for this sample and this treatment is a positive value of 0.380 and the

PC2 score is 0.012. The mean length by number decreased and the short fiber content by

number increased.

My explanation for the meaning of this difference is that the Shirley analyzer

removed some shorter fibers of the population. When the sample with a modified

population was tested with the AFIS, the fiber length distribution exhibited an increase in

the amplitude. This effect did not happen in the AFIS because fibers are not removed to a

separate chamber. Without a fiber removal effect, the differences between the fiber length

distributions are a consequence of the fiber breakage effect only.


109



distributions (II) of sample B. Treatment X is the difference between the fiber length






Sample C is an example of the third type of difference between the fiber length

distributions obtained with different experimental procedures. This sample does not

exhibit a large difference between the distributions for treatment X. It seems that the

distribution is slightly affected by fiber breakage and fiber removal after one passage

through the Shirley analyzer (Figure 4-8). The PC1 score value is -0.067 and the PC2

score value is -0.107. The immature fiber content (IFC) is 8.1%, and the maturity ratio

0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample B (I)

FL I FL II

0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample B (I)

FL I FL III

0

3

6

9

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample B (II)

FL I FL II

0

3

6

9

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample B (II)

AFIS AFIS+AFIS


110

(MR) is 0.80 for the sample with just one passage through the AFIS. The IFC for sample

C is higher than the mean IFC value for this set samples, while the MR for sample C is

smaller than the men MR for this set of samples (Table 4-3). I hypothesize that the small

modification happens because all the fragile fibers were already broken during ginning

and the Shirley analyzer cannot break them more.

Sample C has a different behavior when processed twice through the AFIS

individualizer. There is a higher modification in the amplitude of the distribution, with a

PC1 score value of 0.682. The shape is also more altered than it was by the Shirley

analyzer processing effect, with a PC2 score value of -0.080. I hypothesize that the

observed breakage impact of the AFIS individualizer for this sample is more aggressive

than the Shirley analyzer. Weak spots in the fibers that were not broken with the Shirley

analyzer can be broken with the AFIS individualizer, reducing the fiber length

distribution amplitude and reducing the original small peak around 30 mm.


111



distributions (II) of sample C. Treatment X is the difference between the fiber length






0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample C (I)

FL I FL II

0

10

20

30

40

0 30 60 90

Len

gth

(m

m)

Percentage (%)

Sample C (I)

FL I FL III

0

4

8

12

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample C (II)

FL I FL II

0

4

8

12

0 10 20 30 40 50

Per

centa

ge

(%)

Length (mm)

Sample C (II)

FL I FL III


112

4.4 Conclusions

The Shirley analyzer has a card-like design. It is the device employed in the

standard procedure to determine non-lint content in cotton. The processing effect from

the Shirley analyzer resembles the industrial processing in a card. A shortcoming of this

design is that, like in industrial processing, fibers may be removed together with the

trash.

The AFIS fiber individualizer has a different design that eliminates or minimizes

fiber removal from a sample during processing. Reducing the fiber removal factor, the

processing effect from the AFIS individualizer is limited to the breakage factor.

Furthermore, the results indicate that the fiber length distributions of samples processed

by the AFIS individualizer are more degraded than the distributions of samples processed

by the Shirley analyzer. Samples with a higher impact of the fiber removal when treated

with the Shirley analyzer effect do not exhibit this behavior when treated with the AFIS

effect, but only the fiber breakage. Samples with apparent no fiber length modification

caused by the Shirley analyzer effect exhibit the typical modification caused by the fiber

breakage effect when treated the AFIS effect.

Based on these results, the AFIS is a better tool for future researches focusing on

fiber breakage than the Shirley analyzer. This result does not indicate that the Shirley

analyzer is not useful for fiber quality studies. The Shirley analyzer is the reference

instrument to determine non-lint content in cotton samples. The card-like effect simulated

by the Shirley analyzer may also be useful in future research projects related to

commercial ginning and fiber quality properties.

In the next chapter of this dissertation, I will compare the Shirley analyzer

processing with the processing of other laboratory-scale cleaning devices and how these

different instruments can simulate the fiber quality properties of industry-scale ginning.


113

4.5 Bibliography










Dever, J. K., J. R. Gannaway, and R. V. Baker. “Influence of Cotton Fiber Strength and

Fineness on Fiber Damage during Lint Cleaning.” Textile Research Journal 58, no.

8 (1988): 433–438.

Hosseinali, Farzad. “Investigation on the Tensile Properties of Individual Cotton Fibers.”

Texas Tech University, 2012.

Kelly, Brendan. “Multivariate Analysis of Fiber Properties and Their Relation to Yarn

Properties.” Texas Tech University, 2014.

Kothari, Neha, Steve Hague, Lori Hinze, and Jane Dever. “Boll Sampling Protocols and

Their Impact on Measurements of Cotton Fiber Quality.” Industrial Crops and

Products 109, no. June (2017): 248–254.

http://dx.doi.org/10.1016/j.indcrop.2017.07.045.










USA, 1999.

Tallant, John D., Louis A. Fiori, David M. Alberson, and Walter E. Chapman. “The

Effect of the Short Fibers in a Cotton on Its Processing Efficiency and Product

Quality Part III: Pilot-Scale Processing of Yarns.” Textile Research Journal 31, no.

10 (October 2, 1961): 866–872.


Tallant, John D., Louis A. Fiori, and Charles B. Landstreet. “The Effect of the Short

Fibers in a Cotton on Its Processing Efficiency and Product Quality Part II: Yarns

Made by Miniature Spinning Techniques from Differentially Ginned Cotton.”

Textile Research Journal 30, no. 10 (October 2, 1960): 792–795.



114

Tallant, John D., Louis A. Fiori, and Dorothy C. Legendre. “The Effect of the Short

Fibers in a Cotton on Its Processing Efficiency and Product Quality Part I. Affecting

the Short Fiber Content by the Addition of Cut Cotton Fibers.” Textile Research

Journal 29, no. 9 (1959): 687–695.

Thibodeaux, D, H Senter, J L Knowlton, D Mcalister, and X Cui. “The Impact of Short

Fiber Content on the Quality of Cotton Ring Spun Yarn.” The Journal of Cotton

Science 12, no. 4 (2008): 368–377.

Wakeham, Helmut. “Cotton Fiber Length Distribution— an Important Quality Factor.”

Textile Research Journal 25, no. 5 (May 2, 1955): 422–429.


Yang, Shouren, and Stuart Gordon. “Accurate Prediction of Cotton Ring-Spun Yarn

Quality from High-Volume Instrument and Mill Processing Data.” Textile Research

Journal 87, no. 9 (June 16, 2017): 1025–1039. Accessed March 8, 2019.



115

CHAPTER V

A METHOD TO IMPROVE COTTON FIBER LENGTH

MEASUREMENT FOR LABORATORY ANALYSIS

5.1 Introduction

Length and length uniformity index are fiber quality parameters used in marketing

cotton (Cotton Incorporated 2013). These two parameters are the most important

properties for ring spinning, the spinning technology with the greatest market share in the

world (Cai et al. 2013; Elhawary 2014; Klein 2014), and they are affected by genetics,

environment, and processing conditions (P. Bel-Berger et al. 1991; B. T. Campbell et al.

2018). Spinners use these length parameters to configure their machines and predict yarn

properties using the information provided with each purchased bale of cotton lint.

Farmers process their harvested upland cotton in industry-scale saw gins. Many

improvements have happened in this industry since the saw gin was invented by Eli

Whitney (Whitney 1794; House of Representatives 1812). A typical modern industry-

scale gin is composed of many parts to process tons of cotton fiber per hour, including

pre and post ginning cleaning processes not proposed in the original patent (E. Hughs,

Holt, and Rutherford 2017). Seedcotton cleaners eliminate many sticks, burrs, leaves, and

some dust that is harvested with the cotton before cotton goes to the gin stand. At the gin

stand, saws separate lint and cottonseed. Ginned lint is processed with lint cleaners to

mechanically remove part of the non-lint content, including plant particulates called trash

that have made it through the pre-cleaning and ginning process, before the fibers are

pressed into bales (P. Bel-Berger et al. 1991; Anthony and Mayfield 1994).

Industry-scale gins require hundreds of pounds of seedcotton to function

correctly, making industry-scale ginning impractical in most research applications. For

example, if cotton breeders are selecting individual plants, they cannot process the

harvest from one plant with an industry-scale gin. They must use laboratory-scale gins.

Laboratory-scale gins are typically very basic gin saws and brushes, minimized for small

scale use. These gins typically have no pre-cleaning or lint cleaning and lack uniform

feed controls that industry-scale gins have. Another difference between laboratory- and

industry-scale gins is the amount of used power. An industry-scale gin operates at 0.38-


116

0.82 kW/saw, while a laboratory-scale may run at 0.04 kW/saw (E. Hughs, Holt, and

Rutherford 2017; Testex 2019).

Fibers can break when aggressive mechanical forces are applied during the

cleaning processes used in the ginning industry. Factors such as the smaller throughput,

slower and less consistent feed speeds, fewer cleaning stages, and lower levels of power

applied to process the fibers during laboratory-scale ginning may affect the fiber quality

profile. This results in differences within the length profile of a sample that was

processed in an industry-scale gin. If these differences are statistically significant,

decisions taken on data from laboratory-scale gins may not reflect real-world

performance of a particular cotton sample in an industry-scale gin. However, if additional

treatment could be applied to laboratory-scale ginned fibers, the differences between the

two ginning methods may be reduced. The objective of the method described in section

5.2 is to identify an instrument that can be used after laboratory-scale ginning to reduce

the differences in the fiber length quality profiles between laboratory- and industry-scale

ginning.

5.2 Method

1. Harvest seedcotton using a cotton stripper harvester.

2. Store the harvest at the desired atmospheric conditions, such as 21 ± 1°C and

relative humidity of 55 ± 5% for at least seven days after which equilibrium with

the ambient conditions is assumed. This will minimize the risk to introduce

environmental variation not related to the field experiment.

3. Clean the seedcotton by hand to remove large sticks and burrs that can damage or

hinder the laboratory-scale gin.

4. Feed a laboratory-scale saw-gin between 10 and 20 saws with 4-6 g of seedcotton

per saw per minute. This feed rate has been found to be optimal for the processing

of samples in an efficient and consistent manner.

5. Condition the lint at standard atmospheric conditions for cotton fiber

classification and testing, 21 ± 1°C and relative humidity of 65 ± 2% (ASTM


117

2016) for at least two days after which equilibrium with the ambient conditions is

assumed.

6. Verify the settings of the microdust-trash analyzer 3 (MDTA 3). If necessary,

adjust the gauges between the feed roller and feed plate, the feed plate and

opening roller, and the opening roller and trash knife to 0.1 mm.

7. Turn on the MDTA 3 and adjust the vacuum to 3.5 mbar in the fiber channel and

2.5 mbar in the dust channel.

8. Evenly spread 4-5 g of conditioned lint on the whole surface of the conveyor belt

of the MDTA 3 and press the fibers at the end of the feed belt with the feed roller.

9. Close the rotor box and feed the lint sample into the machine.

10. Recover the sliver of processed fibers from the rotor ring box and spread it evenly

on a flat surface.

11. Sample small portions of processed lint from several parts of the sliver up to 0.5

g.

12. Repeat steps 8-11 for each laboratory replication from each sample that will be

tested for fiber length quality.

13. Use each sampled mass of cleaned lint to hand-draw a uniform sliver of fibers

with a length of 25-30 cm.

14. Feed the slivers to an Advanced Fiber Information System (AFIS) Pro 2 testing

3,000 fibers per sliver.

15. Retrieve the report.

5.3 Comparison with other laboratory instruments and validation of the method

The objective of the proposed method is to help cotton researchers to simulate the

fiber length quality profile of cotton ginned in an industry-scale gin with a laboratory-

scale gin. This would prevent researchers from making conclusions that cannot be

reproduced under industrial conditions.

The MDTA 3 is not the only available laboratory-scale lint cleaner. Shirley

analyzers from the Shirley Institute and SDL Atlas are tools frequently used to determine

the trash in cotton samples (ASTM 2012d) that could also be used to post-process lint.


118

To evaluate the proposed method, we performed an experiment comparing the

fiber length quality profile of cotton samples ginned in an industry-scale gin with samples

ginned in a laboratory-scale gin, with and without post-processing with different types of

laboratory-scale lint cleaners.

5.3.1 Plant material

A set of 20 entries with a wide range of fiber length properties was used. These

entries were planted at Texas Tech Research Farm, Lubbock, TX, in crop year 2018. The

double-row plots were 6.1 meters long, with a density of 10 seeds per meter to simulate

regional commercial planting practices. The seeds were planted on loam soil. Drip

irrigation was applied, and the recommendations for irrigated cotton production in the

region were followed (Ayele, Kelly, and Hequet 2018). During the experiment, the

accumulated rainfall was 272 mm and the growing degree days (GDD15.6) were 1308

units (Snowden et al. 2013).

Each plot was harvested into mesh bags with a John Deere stripper harvester with

no field cleaner, resulting in seedcotton samples with a high amount of sticks and burrs.

The seedcotton was conditioned for seven days at 21 ± 1°C and relative humidity of 55 ±

5%.

5.3.2 Cleaning and ginning

After bringing all seedcotton samples to the same atmospheric conditions, lint

from each sample was processed with five different methods. Initially, each sample was

divided into two subsamples. One subsample was passed through the seedcotton cleaners,

gin stand, and lint cleaners of an industry-scale Imperial III Lummus microgin to obtain a

representative industry-scale lint sample. Another subsample was passed through the

seedcotton cleaners of the Imperial III Lummus microgin and recovered before the gin

stand. The cleaned seedcotton was then ginned with a tabletop laboratory-scale gin

(Dennis Manufacturer, Athens, TX) to obtain a representative laboratory-scale lint

sample.


119

The laboratory-scale lint was divided into four subsamples of 100 g. The first

subsample was sent to analysis (Method I). The second subsample was processed through

a Shirley analyzer from the Shirley Institute (Method II). The third subsample was passed

through a Shirley analyzer MK2 from SDL Atlas (Method III). Both Shirley analyzers

were set up following the standard protocol for non-lint content measurement (ASTM

2012d). The fourth subsample was cleaned with an MDTA 3 from Suessen (Method IV)

(Figure 5-1).

Figure 5-1 Flowchart of the validation method.


120


We used an Advanced Fiber Information System (AFIS) Pro 2 (Uster) with five

laboratory replications of 3,000 fibers per sample to assess the length fiber quality profile

of the 20 entries and five types of lint per entry. The AFIS is a sensitive instrument used

to analyze fiber length parameters of individual fibers (Ayele, Kelly, and Hequet 2018; B.

R. Kelly and Hequet 2018). The AFIS has a pinned cylinder that individualizes fibers and

trash. It creates fiber length distributions with 40 bins of 1.27 mm after testing 3,000

fibers. Both the entire length distribution and length parameters extracted from the

distribution, such as short fiber content by number (SFCn, fibers with length ≤ 12.7 mm),

mean length (Ln), and length of the 5% longer fibers by number (L5%) are reported by

this instrument (Shofner et al. 1995; Shofner and Shofner 1999). Statistical differences

were measured using a paired t-test at 95% of confidence.

5.3.3.1 Differences for trash and neps counts

Statistical differences are observed among the treatments for trash and neps

(entanglements of fibers) (Table 5-1). Lint from laboratory-scale ginning (Method I)

exhibits a smaller number of neps and a higher content of trash, with greater values for

visible foreign matter, trash per gram, and dust per gram. The presence of a lint cleaner in

the industry-scale ginning (Reference method) may be a reason for these differences

(Mangialardi 1996). Indeed, it is well documented that aggressive mechanical handling

from industrial lint cleaners not only removes trash particles but also tends to create fiber

entanglements, i.e., neps, entanglements of fibers that reduce yarn spinning efficiency

and yarn quality (J. J. Hebert, Mangialardi, and Ramey 1986; Long et al. 2010).


121

Table 5-1 Average values of trash and neps parameters in a dataset of 20 genetic

materials, processed with five different ginning approaches.

Lint type Neps per

gram

(count)

Visible

Foreign

Matter (%)

Trash per

gram

(count)

Dust per

gram

(count)

Reference method 353 3.11 156 427

Method I 196* 3.71* 185* 501*

Method II 256*ª 0.28* 16* 50*

Method III 257*ª 0.33* 21* 49*

Method IV 258*ª 0.78* 43* 128*

*Treatments statistically different from industry-scale lint at 5% of significance.

ªTreatments statistically different from laboratory-scale lint at 5% of significance.

Reference method: industry-scale ginning; Method I: laboratory-scale ginning; Method

II: laboratory-scale ginning + Shirley analyzer from Shirley Institute; Method III:

laboratory-scale ginning + Shirley analyzer MK2; Method IV: laboratory-scale ginning +

MDTA-3 from Suessen.

Applying a laboratory-scale lint cleaner to the lint from the regular laboratory-

scale gin reduces the trash content in the samples and increases the number of neps. The

numbers of neps are statistically different from the regular laboratory-scale lint, but they

are still lower than the industry-scale lint (Table 5-1).

The differences are indications that although additional processing simulates

industrial processing, it is not a perfect replacement. The next step in the research was to

determine how each laboratory-scale lint cleaner altered fiber length parameters and

distributions in the samples.


122

5.3.3.2 Differences for fiber length parameters

There are statistical differences between industry- (Reference method) and

laboratory-scale (Method I) lint samples for short fiber content by number (SFCn), which

is defined as fibers less than or equal to 12.7 mm, and mean length by number (Ln)

(Table 5-2). However, longer fibers as represented by the average length of the 5% longer

fibers by number in the measured sample (L5%), are not statistically different between

these samples (B. R. Kelly and Hequet 2018). Cotton samples generated by Method I

represent a common scenario in cotton fiber length research. Nevertheless, these samples

are different from the Reference method, the expected outcome in the cotton ginning

industry.

Table 5-2 Average values of fiber length parameters in a dataset of 20 genetic materials

with a wide range of length properties, processed with five different ginning approaches.

Lint type Short fiber content

by number (%)

Mean length by

number (mm)

5% longer fibers by

number (mm)

Reference method 30.3 18.8 34.0

Method I 26.8* 19.6* 34.3

Method II 27.8* 19.3* 34.3

Method III 32.3* 18.0* 33.5*

Method IV 29.2 19.0 34.0

*Treatments statistically different from industry-scale lint at 5% of significance.

Reference method: industry-scale ginning; Method I: laboratory-scale ginning; Method

II: laboratory-scale ginning + Shirley analyzer from Shirley Institute; Method III:

laboratory-scale ginning + Shirley analyzer MK2; Method IV: laboratory-scale ginning +

MDTA-3 from Suessen.

While the Shirley analyzer from the Shirley Institute (Method II) increases the

SFCn and decreases the Ln, it is not enough to eliminate the statistical difference with the

industry-scale lint type. When this instrument processes a sample, it may break and

remove some fibers from the sample, altering the measured length parameters.


123

Lint processed with the Shirley analyzer MK2 (Method III) has a different fiber

quality profile in comparison with the lint from industry-scale ginning for all the three

analyzed parameters. This machine was developed to perform the same processing as the

Shirley analyzer from the Shirley Institute. Nevertheless, there are design differences

between both instruments. The Shirley analyzer MK2 increases sample SFCn and reduces

Ln and L5% in comparison to industry-scale lint type samples. Fibers are likely to be

broken and kept in the sample, increasing the SFCn. Since the overall fiber length was

reduced, L5% and Ln were decreased.

Processing industry-scale lint type with the MDTA-3 (Method IV) eliminates the

statistical differences between industry- and laboratory-scale lint types among the

samples in the dataset. This treatment is the proposed method, and it could be used as a

proxy to simulate the fiber length profile from industry-scale ginning.

5.3.3.3 Differences for fiber length distributions

Fiber length parameters are extracted from distributions. The following examples

are fiber length distributions of two genetic materials from the data set, KH-13-155-02

and TAM A106-16 (Figure 5-2).


124

Figure 5-2 Fiber length distributions by number of two samples from industry-scale gin,

laboratory-scale gin, laboratory-scale gin + Shirley analyzer from Shirley institute,

laboratory-scale gin + Shirley analyzer MK2, and laboratory-scale gin + MDTA 3.

The Shirley analyzer from the Shirley Institute processes the samples and creates

an intermediate distribution between industry- and laboratory-scale ginned lint types. The

Shirley analyzer MK2 seems to have over-processed the samples and shortened the

overall length of the fibers, increasing the differences in the distribution in comparison

with industry-scale ginned lint. The MDTA-3 seems to simulate the industry-scale cotton

fiber length profile better than the other tested machines.


125

5.4 Conclusions

Industry-scale and laboratory-scale ginning have features that create differences

in fiber length distributions. The differences can be minimized with additional processing

for the laboratory-scale ginned lint. We tested the processing effect of three different

instruments and verified that the MDTA 3 is the best instrument among the three tested to

minimize length differences with industry-scale ginning.


126

5.5 Bibliography

Anthony, W S, and William D Mayfield. Cotton Ginners Handbook. Springfield: United

States Department of Agriculture, 1994.

ASTM. “D1776/D1776M-16 Standard Practice for Conditioning and Testing Textiles.”

In Annual Book of ASTM Standards, 1–5. West Conshohocken: ASTM

International, 2016.


Book of ASTM Standards, 1–5. West Conshohocken: ASTM International, 2012.



(2018): 47–55.




(1991): 83–88.









2018.




booklet.pdf.

Elhawary, I.A. “Fiber to Yarn: Staple-Yarn Spinning.” In Textile and Fashion -

Materials, Design, and Technology, edited by Rose Sinclair, 191–211. Sawston:


Hebert, J. J., G. Mangialardi, and H.H. Ramey. “Neps in Cotton Processing.” Textile

Research Journal 56, no. 2 (February 1, 1986): 108–111. Accessed November 30,

2016. http://trj.sagepub.com/cgi/doi/10.1177/004051758605600206.

House of Representatives. “Petition of Eli Whitney Requesting the Renewal of His Patent

on the Cotton Gin.” Last modified 1812. Accessed August 22, 2019.

https://catalog.archives.gov/id/306631.

Hughs, Ed, Greg Holt, and Ross Rutherford. “Saw Gin Stands.” The Journal of Cotton

Science 21 (2017): 60–69.


127





Klein, Werner. The Rieter Manual of Spinning. Vol. 1. Winterthur: Rieter, 2014.

Long, Robert L., Michael P. Bange, Stuart G. Gordon, Marinus H.J. van der Sluijs,

Geoffrey R.S. Naylor, and Greg A. Constable. “Fiber Quality and Textile

Performance of Some Australian Cotton Genotypes.” Crop Science 50, no. 4 (2010):

1509–1518.

Mangialardi, G. J. “Lint Cleaning Options to Preserve Fiber Quality at Gins.” Applied

Engineering in Agriculture 12, no. 5 (1996): 555–562.





USA, 1999.




Testex. “Saw Gin Portable TB510A.” Last modified 2019. Accessed August 22, 2019.

https://www.testextextile.com/product/lab-saw-cotton-gin-tb510a/.

Whitney, Eli. “Cotton Gin.” Last modified 1794. Accessed August 22, 2019.

https://catalog.archives.gov/id/102278457.


128

CHAPTER VI

GENERAL CONCLUSIONS

In my research, I observed that trash and processing may impact the measurement

of cotton fiber length. There is evidence that the trash content affects the formation of the

fiber beard in the HVI comb, increasing the experimental error for the reported fiber

length parameters. There is also evidence that trash content affects the sampling of the

longer fibers with the AFIS, altering the analyzed population of fibers by this instrument.

Therefore, researchers must send samples with the lowest native trash content as possible

for fiber testing to avoid these problems. If researchers cannot send clean samples,

mechanically cleaning the samples before testing them could be an alternative approach

to reduce the trash impact on fiber measurements. A shortcoming of this approach is that

cleaning may alter the fiber length distribution and that there is an interaction mechanical

cleaning x cotton, i.e., all cottons do not behave the same way when submitted to

mechanical cleaning.

The processing effect may impact the fiber properties of a sample and this effect

changes based on the type of instrument used. For example, the Shirley analyzer may

modify the fiber length distribution in a sample by removing or breaking fibers, while the

AFIS individualizer may modify this distribution by breaking fibers, i.e., no fiber

removal.

It is possible to simulate the processing effect of an instrument with a

combination of instruments. I verified that the values for AFIS fiber length parameters

measured on a set of samples ginned with a laboratory scale gin and cleaned with the

MDTA-3 are at the same level than samples ginned with an industrial scale gin.

This research led to other questions that must be investigated in the future. For

example, can a higher number of replications reduce the error caused by the presence of

trash for both HVI and AFIS? Which fiber quality properties are more important to

explain how the fiber length distribution is modified by a lint cleaning device? What is

the impact of trash and processing on the yarn quality prediction? Answering these

questions may improve the quality of the measured fiber properties, helping cotton

researchers and the cotton industry.

Effects of Trash and Processing on Cotton Fiber Quality ...

Documents

Transcript of Effects of Trash and Processing on Cotton Fiber Quality ...