Occupational Exposure to Cotton Dust in Cottonseed Oil Mills

Applied Occupational and Environmental Hygiene
Volume 17(2): 121–130, 2002
Copyright c○ 2002 Applied Industrial Hygiene
1047-322X/02 $12.00 + .00
Occupational Exposure to Cotton Dust in Cottonseed
Oil Mills
Semion Tabak, 1 David M. Broday, 1 Ilya Tabak, 2 and Gedalyahu Manor 1
1 Faculty of Agricultural Engineering, Technion-Israel Institute of Technology, Haifa, Israel;
2 Robert Ettinger Associate Consulting Engineers Inc., New York, New York
Air samples were collected at breathing height in the
hulling-separation department of a modern cottonseed oil
mill in Uzbekistan. The average elutriated mass concentration
measured by standard cotton dust samplers was
4.6 mg/m 3 , much lower than the average total dust concentration
measured by stationary personal samplers,
12.49 mg/m 3 , and by personal samplers attached to workers,
14.53 mg/m 3 . Differences in readings among the vertical elutriators,
stationary personal samplers, and roving personal
samplers are attributed to the distinct sampling nature and
dynamics of these samplers. The data suggest that most of
the dust consisted of particles larger than 15 µm, the particle
size cutoff of the vertical elutriator. Differences in readings
among stationary and roving personal samplers are statistically
significant, presumably representing biased sampling
by the roving personal samplers of regions characterized
by high dust concentration (due to machines malfunctioning),
the nonstatic nature of the sampling, and the interaction
between the sampler and the worker (the personal
cloud). Cotton dust concentrations in the hulling-separation
room were nonuniform, peaking in front of and between the
huller-separator pairs. The high total mass readings show
that workers were exposed to very high levels of nonthoracic
airborne dust, which upon inhalation tends to deposit
in the extrathoracic airways. The high elutriated mass concentrations
suggest that workers were exposed to respirable
cotton dust at levels higher than 1 mg/m 3 mean concentration,
the current Occupational Safety and Health Administration
(OSHA) Permissible Exposure Limit (PEL) for
cotton dust. Regressions between dust concentrations measured
by stationary vertical elutriators and by personal samplers
attached to workers serve for estimating the potential
occupational exposure to cotton dust of workers in the
hulling-separation room.
Keywords Dust Particles, Personal Samplers, Cottonseed Oil Mill
INTRODUCTION
Origin of Dust in Cottonseed Oil Mills
Cottonseed oil mills are an integrated part of the cotton industry,
contributing significantly to the agricultural economy
in cotton-growing countries. Cottonseed oil mills produce a
number of products such as oil for human consumption, hulls
and feed for livestock, and linters for the paper and cellulose
industries.
The cottonseed delivered to the mill is contaminated with
varying amounts of trash such as soil, pieces of boll hull, cotton
locks, large pieces of foreign materials (balls, rocks, wood),
loose meats, leaves, black seeds, and fine debris. All the foreign
matter must be separated from the seeds in the cleaning process.
The most effective way to remove the trash is by shakers and
pneumatic separators. (1,2) This process involves air flow that impinges
on the seeds as they are discharged from the feeder onto
the conveyor trays. The air is then emitted as dust jet both inside
and outside of the mill. Conveying, and in particular pneumoconveying,
contributes considerably to dust emanation in the mill
working environments as well. Dust is also generated from seed
delinting and hulling, and during the separation of kernels from
hulls. (1,3) Specifically, in preparation for pressing and oil extraction,
linters are removed from the seeds and the delintered seeds
are conveyed to the hulling-separation apparatus, where seed
embryos (meats) are separated from the seed coat (hull). Handling,
displacement, charging, and discharging of cottonseed
cakes also contribute to the emergence of dust during normal
processes in the cottonseed oil manufacturing industry.
Sometimes dust levels in cottonseed oil mills may be higher
than any accepted occupational health standard, posing a threat
to the workers’ well-being. Indeed, increased amounts of airborne
fine particles were reported in dust-laden air in cottonseed
oil mills, (4) with an average of ∼28% oil content. (5) Nonuniform
cotton dust composition and size distribution were found in
hulling-separation departments of cottonseed oil mills (6) , while
dust in delinting and baling work areas was rich in cellulosic
components, dust in the hulling area had higher protein content.
121

122 S. TABAK ET AL.
Occupational Exposure to Dust in Cottonseed Oil Mills
Occupational exposure to bioactive dust is oftentimes found
in the cotton, hemp, flax, and sisal industries. Susceptible workers
that are exposed via inhalation to bioactive agents such as
bacterial endotoxin may develop reversible acute lung disorders
and allergic reactions. (7,8) Repeated exposures to cotton
dust may lead to chronic bronchitis, emphysema, and byssinosis,
a disabling asthma-like respiratory disease which is characterized
by decrement in the forced expiratory volume in one
second (FEV1). (9) Controlling the dust level in the mill reduces
the risks of adverse health effects to workers, decreases
the worker discomfort, and helps maintain the plant
facilities. (10)
To prevent occupational exposure leading to byssinosis and
other disabling respiratory diseases, the textile and yarn manufacturing
industries, among other workplaces, are required to
comply with the occupational safety and health standards promulgated
under the Occupational Safety and Health Act of 1970.
In the act, the National Institute for Occupational Safety and
Health (NIOSH) was authorized to develop and establish recommended
occupational safety and health standards, and to conduct
research and experimental programs for development and
improvement of criteria for occupational safety and health standards.
Specific to the cotton and yarn industries, employers are
required to: (1) limit the amount of respirable cotton dust in
the air by utilizing dust control measures (i.e. efficient ventilation,
enclosing the process equipment and placing it under
negative pressure); (2) provide provisions for medical monitoring
of workers; (3) supply the workers with personal protective
equipment; and (4) include administrative controls
(i.e. rules and policies such as the requirement to use face
masks).
In contrast with the National Ambient Air Quality Standards
(NAAQS), set by the Environmental Protection Agency (EPA)
for regulating ambient levels of PM10,PM2.5, and other criteria
pollutants, the NIOSH Recommended Exposure Limits (RELs)
and the OSHA PELs were set for the purpose of regulating
threshold levels of worker exposure to indoor (occupational and
residential) airborne contaminants. The current OSHA PEL for
cotton dust (raw) in the cotton, textile, and yarn manufacturing
industries range from 0.2 to 1 mg/m3 . These values represent
a time-weighted-average (TWA) concentration that must not be
exceeded during any 8-hour work shift in a 40-hour workweek.
The NIOSH REL for cotton dust (raw), representing the highest
allowable airborne concentration that is not expected to injure
workers, is set at 0.2 mg/m3 , based on TWA for up to 10 hour/day
during a 40-hour workweek.
The American Conference of Governmental Industrial Hygienists
(ACGIH ○R
) also set the threshold limit value (TLV ○R
)
for cotton dust at 0.2 mg/m3 . The TLV represents cotton dust
concentration to which it is believed that nearly all workers may
be exposed, day after day, without developing adverse health
effects. Nevertheless, evidence suggests that dust-related accelerated
decline in lung function among cotton textile workers
who smoke occurs even when the 0.2 mg/m 3 standard is strictly
imposed. (11,12)
Although for cotton dust the three indoor/occupational exposure
limits outlined above coincide, only the OSHA PEL has
an enforcement power as a legal standard. Therefore, it is noteworthy
that although OSHA PELs for cotton dust exist for specific
operations and industries (e.g., yarn manufacturing, cotton
washing, textile waste houses, slashing and weaving operations,
cotton waste recycling and garnetting), there is unfortunately
no specific PEL for cotton dust in cottonseed oil mills. Accordingly,
although workers in cottonseed oil mills (and other
nontextile cotton industries) breath dust similar in content and
size distribution to the dust found in the textile industry, (13)
and although such dust is known to induce respiratory diseases,
such as byssinosis, and lung disorders among susceptible
workers, cottonseed processing operations have no specificlegal
standard.
The OSHA recommended sampling procedure for determining
occupational exposure to lint-free respirable cotton dust utilizes
a vertical elutriator cotton dust sampler (or an equivalent),
with a particle size cutoff at approximately 15 µm aerodynamic
diameter when operating at a flow rate of 7.4 ± 0.2 L/min. (14)
Elutriated dust concentrations in cottonseed oil mills are expected
to be almost uniform over the whole working area, with
an average varying from 0.5 to 2.0 mg/m. (3,11) In fact, this assumption
is prerequisite if workers’ exposures to cotton dust is
to be surrogated by dust concentrations measured by a stationary
central vertical elutriator. Yet both Parnel et al. (3) and Matlock
et al. (15) reported nonuniform dust concentrations, with consistently
higher dust concentrations in the hulling-separation and
the cleaning areas.
The objectives of this work was to collect data on respirable
dust levels during normal process operations in a high-capacity
hulling-separation department of a modern cottonseed oil mill,
and to use these data for evaluating the recommended method
for assessing occupational exposure to cotton dust in cottonseed
oil mills. Consequently, airborne cotton dust was analyzed in
terms of (1) total and elutriated airborne mass, and (2) particle
size distribution. Elutriated concentrations collected by OSHA’s
recommended vertical elutriator method were correlated against
total mass concentrations measured by stationary and roving
personal samplers, in order to predict the workers’ exposure via
inhalation to cotton dust.
METHODS AND PROCEDURE
Data were collected during 1996 in the hulling-separation department
of a modern cottonseed oil mill in Uzbekistan, which
processed 1200 tons/day of cottonseed. A double hulling-separation
scheme was applied in the mill during the data acquisition.
The mill was equipped with new cottonseed cleaners, disk
hullers, and separators of kernel and hull fractions. The layout
of the second floor of the hulling-separation room, in which the
process equipment is located, is shown in Figure 1. Both the

FIGURE 1
The second floor plan of the hulling-separation room in the
mill. H1, H2, S1, and S2 are the hullers and shakers in the first
and second hulling-separation lines, respectively. V(1) and
SP(1) are the sampling locations of the vertical elutriators and
the stationary personal samplers, respectively. Dimensions
are in mm.
disk hullers and the shakers were significant dust sources; the
hullers (rotating at 1000–1200 rpm) as a result of air jet discharged
together with the hulled seeds, and the shakers due to
the hulled seed separation. The air jet issued from the hullers
increased the pressure in the enclosure over the shakers, significantly
boosting emission of dust to the working environment.
In contrast, the hull beaters in the first floor of the hullingseparation
room were fully enclosed and under negative pressure,
thus efficiently preventing dust buildup and spread into
the working environment. All workers were supplied with personal
respirators but these were used only occasionally, mainly
when dust concentrations were extremely high and affected the
visibility.
Gravimetric Concentration Measurement
Vertical Elutriators
In accordance with the Occupational Safety and Health Standard,
(14) concentrations of lint-free respirable cotton dust were
determined by two vertical elutriator cotton dust samplers (Sierra
Instruments Inc., Carmel Valley, CA). The vertical elutriator
(VE) is a size-selective area sampler with a particle size cutoff
at 15 µm aerodynamic diameter when operating at a flow rate
of 7.4 ± 0.2 L/min. For sampling airborne dust at typical human
breathing height, the VEs were set with their inlets ∼1.6 m
above the floor. Dust samples were collected on 37 mm polyvinyl
chloride filters with a 5 µm pore size (SKC Inc., Eighty Four,
PA). Samples were gathered simultaneously by two vertical elutriators
in a total of seven locations (see Figure 1) adjacent to
the huller-separtor pairs, providing representative samples of air
to which the workers were exposed. Each sampling point was
sampled at random multiple times (see Table I).
OCCUPATIONAL EXPOSURE TO COTTON DUST 123
Personal Samplers
Sampling of total dust mass in the hulling-separation room
was carried out by stationary personal samplers (SPSs) sampling
at breathing height, and by personal samplers attached to workers
(RPSs) and sampling the breathing zone. Open-face type
stationary personal samplers were used, collecting samples on
37 mm polyvinyl chloride filters with a 5 µm pore size (SKC
Inc.). Total mass concentrations were measured simultaneously
by four stationary personal samplers in a total of 15 locations
within the hulling-separation room (see Figure 1), providing
representative samples of air at breathing height. Each sampling
point was sampled at random multiple times (see Table I).
IOM Inhalable Dust Samplers (SKC Inc.) with reusable filter
cassettes and sampling heads designed to better measure the
worker exposure to total airborne particulate matter (PM) were
used as roving samplers (RPSs). Each sampler was made of a
37 mm cylinder with a 15 mm circular opening shaped like a thin
lip protruding outward from one of its ends. The samplers were
attached to personal air pumps (Mine Safety Appliances Co.,
Pittsburgh, PA) that drew air at 2.0 L/min. Theoretically, such
air flow enables the IOM sampler to effectively collect particles
with a diameter up to 100 µm. (16) The sampler filter cassette held
a25mmfilter with a 5 µm pore size. The cassette and the filter
were weighed together before and after sampling. Therefore,
dust on both the filter and the inner walls of the cassette is
contained in the data, which typically include large particles that
are considered inhalable but nonrespirable (10 < d < 100 µm).
Sampling was performed by 6–8 samplers simultaneously. The
worker mobility during sampling was restricted to prescribed
locations and their neighborhoods (see Table I).
Weighing
In accordance with the OSHA guidelines, sampling lasted
8 hours during normal operating conditions and was repeated
16–42 times at each location. The calibration of all samplers
was checked frequently. An analytical balance with sensitivity
of 10 µg (Sartorius AG, Göttingen, Germany), located in
a dust-free air-conditioned room with temperature and relative
humidity set at 24 ◦ C and 45 percent, respectively, was used
for weighing. The filters were held in the weighing room for
24 hours prior to weighing. Three blank (control) filters as well
as a reference mass of approximately the same weight as the
filter were weighed together with the filters. The field blanks
were used to correct the weight of the field samples. (17,18) Dust
concentration was calculated by dividing the adjusted mass collected
on the filters by the volume of sampled air.
Particle Size Measurement
Dust samples for particle size analysis were collected with the
same sampling equipment used for gravimetric analysis. (19,20)
The most efficient filters for collecting ambient airborne particles
for subsequent size analysis are membrane filters made of
cellulose esters. (18) These filters are soluble in acetone, making
it possible to dissolve the filter and collect the deposits on

124
TABLE I
Cotton dust concentrations in the hulling-separation room at the sampling locations shown in Figure 1
Vertical elutriator Roving personal sampler
Sampler # of Mean SD Min. Max. # of Mean SD Min. Max.
location samples [mg/m3 ] [mg/m3 ] [mg/m3 ] [mg/m3 ] samples [mg/m3 ] [mg/m3 ] [mg/m3 ] [mg/m3 ]
V1 33 7.21 5.64 1.16 28.1 28 15.24 12.04 3.18 84.52
V2 42 3.83 4.12 0.62 8.62 24 14.87 13.12 2.96 86.35
V3 28 5.89 4.11 0.38 23.8 24 15.92 12.48 4.11 116.27
V4 36 1.61 1.23 0.1 4.18 31 5.46 5.11 0.46 26.95
V5 38 4.81 2.84 0.42 13 23 12.54 10.16 2.18 68.18
V6 38 3.25 2.14 0.18 16.3 28 11.06 8.96 1.84 74.25
V7 26 6.94 3.86 0.74 32.1 27 14.58 12.31 0.98 84.14
Weighted average 4.60 4.03
Stationary personal sampler
SP1 18 18.24 16.22 2.12 49.9 22 19.32 14.89 2.63 174.54
SP2 16 11.38 8.12 0.63 32.4 28 16.88 16.35 2.19 112.96
SP3 32 17.85 14.26 3.04 169 21 18.65 15.86 3.84 156.18
SP4 28 13.65 11.83 0.82 56.8 22 15.86 12.95 3.81 76.19
SP5 27 14.68 11.28 1.08 72.6 29 16.25 11.8 4.29 58.87
SP6 31 16.45 12.46 0.31 82.4 26 16.92 9.24 3.32 136.49
SP7 22 6.83 6.56 0.62 41.4 23 8.12 4.14 0.72 52.19
SP8 28 4.43 3.18 1.14 12.1 26 6.89 3.78 0.84 54.27
SP9 29 6.12 4.92 0.83 22.3 21 9.14 6.18 0.79 59.21
SP10 28 12.58 11.06 0.56 56.2 24 18.43 15.86 2.71 126.81
SP11 19 15.14 12.82 1.06 34.9 28 17.94 16.19 1.28 49.16
SP12 29 16.35 7.35 1.19 48.2 27 18.51 12.09 3.31 86.48
SP13 17 10.95 9.58 0.95 27.4 29 14.89 10.26 3.28 39.35
SP14 17 13.1 14.26 1.26 39.2 26 16.23 14.39 2.12 52.18
SP15 22 14.1 15.29 0.62 68.3 25 17.43 18.59 1.88 84.52
Weighted average 12.49 11.67 14.53 12.70

surfaces suitable for further examination. Mixed Cellulose Esters
(MCE) membrane filters with a 0.45 µm pore size (SKC,
Inc.) were used in the vertical elutriators for this purpose. Sampling
was carried on for 30–60 min to get an adequate deposition
of dust on the filters but to avoid overloading. Post exposure, the
filters were placed on dry microscope slides above an acetone
bath, the vapors of which rendered the filters transparent within
a few minutes. After drying for five minutes, the filters were
ready for further analysis.
Size measurements were performed with a computer vision
and image processing system (BarGold Electronics Ltd., Haifa,
Israel), which is a digital signal processor (DSP) video system
that captures images of particles on the transparent filter and
subsequently analyzes their size. The hardware consists of a
charge-coupled devices (CCD) camera with magnifying lenses
and a dedicated illumination source. The CCD is an array of photosites
(pixels) on a silicon substrate that record light as electrical
signal by converting photons to electrons. The sensitivity of a
standard CCD with ∼50 percent quantum efficiency (the number
of photons converted to electrons) is tenfold that of the fastest
film. A high performance stand-alone image processing system
was used for analyzing the particle images. This unit consists of
a proprietary software that outputs the particle size distribution
among other statistics.
RESULTS
Table I lists average dust concentrations measured in the second
floor of the hulling-separation room. The weighted average
elutriated concentration was 4.60 ± 4.03 mg/m 3 (n = 241, where
n is the number of filters analyzed). This concentration is more
than fourfold the OSHA PEL for cotton dust in the cotton textile
industry (1 mg/m 3 ), and is higher within more than an order
of magnitude than the recommended advisory exposure guidelines,
the NIOSH REL and the ACGIH TLV, for cotton dust
(0.2 mg/m 3 ). In fact, local average concentrations were much
higher than the action level of any yarn, cotton, and textile
operation (see Table I). Total mass concentrations measured
by the stationary personal samplers in adjacent locations averaged
12.49 ± 11.67 mg/m 3 (n = 363). Dust concentration measured
by the roving personal samplers averaged 14.53 ± 12.7
mg/m 3 (n = 562). The huge difference in airborne dust concentrations
measured by the total mass samplers and the vertical
elutriators resulted mainly from a considerable amount of short
nonrespirable airborne fibers (lint; see also Discussion section).
The highest elutriated concentrations were found in between
the huller-separator pairs (in locations V1, V3, V5, V7, see
Figure 1). The stationary personal samplers (SPSs), however, did
not give such distinctive results. Minimum readings by any measurement
technique were achieved in the alley between the two
rows of huller-separator pairs (in locations V4, SP7, SP8, and
SP9). This finding was predicted based on the huller-separator
design and the central feeding of the two huller rows, causing
dust emanation away from the symmetry plane in Figure 1.
Similarly, readings of personal samplers attached to operators
OCCUPATIONAL EXPOSURE TO COTTON DUST 125
spending more time near the locations V4, SP7, SP8, and SP9
were much lower (50% or less) than those of workers in other
locations (Table I). The high concentrations of cotton dust reported
here are in good agreement with the findings of Matlock
et al., (15) who found an average concentration of 13.2 mg/m 3
in the hulling and cleaning area, and of Parnell et al., (3) who
found that the hulling-separation room had consistently higher
dust concentrations than other regions in the mill.
Typical size distributions of cotton dust particulate matter
generated during normal operations in the hulling-separation
room are listed in Table II. About 30–60 percent (by count) of the
measured particles were < 2 µm, and ∼70–90 percent had a diameter
< 8 µm. Figures 2a and 2b depict two typical examples of
FIGURE 2
Size distribution of dust particles collected by vertical elutriator
in sampling locations V2 (a) and V5 (b). The area of each bar
is proportional to the sample number fraction in that size bin.

126 S. TABAK ET AL.
TABLE II
Size distribution of cotton dust particles collected by the vertical elutriator.
Data are given in terms of frequency [%] and standardized frequency [% µm −1 ]
Bin size
Sampling 0.5–1 1–2 2–4 4–8 8–16 16–32 32–64 64–128
point [µm] [µm] [µm] [µm] [µm] [µm] [µm] [µm]
Mean
size
[µm]
[%] 15.4 44.2 20 11.2 5.1 2.65 1.45
V1 3.99
[% µm −1 ] 30.8 44.2 10 2.8 0.6 0.2 0.05
[%] 16.7 32.2 27.8 16.7 2.2 2.2 1.7 0.5
V2 4.53
[% µm −1 ] 33.4 32.2 13.9 4.2 0.3 0.15 0.05 0.01
[%] 29.7 23 23 11.5 5.5 5.5 1.8
V3 4.79
[% µm −1 ] 59.4 23 11.5 2.9 0.7 0.34 0.06
[%] 13.4 19.1 34.3 13.4 8.6 7.62 2.9 0.95
V4 4.39
[% µm −1 ] 26.8 19.1 17.2 3.4 1.1 0.5 0.1 0.01
[%] 16.3 23.1 21 11.6 9.6 6.7 5.8
V5 7.34
[% µm −1 ] 32.6 23.1 10.5 2.9 1.2 0.4 0.2
[%] 19.2 20 36.9 16.2 4.6 2.3 0.8
V6 3.08
[% µm −1 ] 38.4 20 18.5 4.1 0.6 0.14 0.03
[%] 14.1 22.8 37 19.7 3.2 2.4 0.8
V7 4.47
[% µm −1 ] 28.2 22.8 18.5 4.9 0.4 0.15 0.01
the particle number distribution in the hulling-separation room
in terms of the ratio between the particle fraction accommodated
in each bin (%) and the average particle size in the bin
(µm). This presentation makes the area of any bar in the histogram
(“the area under the curve”) proportional to the fraction
of particles measured in that size range. With the exception
of size distributions measured in the locations V1 and V3, the
former due to a relatively high reading in the second bin (1–
2 µm) and the latter due to a relatively high reading in the first
bin (0.5–1 µm), the shape of the cotton dust aerosol is generally
preserved. Therefore, it seems that the size distribution of
the coarse fraction (> 0.5 µm) of dust in the mill is independent
of the total amount of dust suspended in the air and
of the sampling point. This result is supported by the fact that,
excluding the location V5 for being an outlier, the particle mean
size varied relatively little among the different sampling
points, d ¯=
4.21 ± 0.61 µm (i.e. coefficient of variability of
14.5%).
Figures 2a and 2b show a decrease in the particle number
with the increase in size. Due to lack of size measurements in
the range d < 0.5 µm, it is impossible to conclude whether the
aerosol has an exponential shape, which typically results from
mechanical shaping processes such as cutting, slicing, grinding,
and pounding, (21) or a trimmed lognormal shape, which
frequently describes the shape of an ambient aged aerosol. (22)
Regardless, a large fraction of the dust emanated during normal
operations in the cottonseed oil mill is respirable. Due to its biogenic
composition it has the potential to irritate the respiratory
airways, and therefore this dust is considered an occupational
hazard. From an engineering perspective, this imposes stringent
requirements on the solutions that should be employed to control
dust emanation and workplace PM levels. Namely, suction
of air and maintaining of negative pressure in the enclosures of
the technological equipment, and effective removal of fine dust
particles from the air are the core requirements from any air
cleaning system.

DISCUSSION
Our results show a considerable variation in dust levels in
the hulling-separation room, with the highest readings obtained
in between huller-separator units on the same row. This is in
accordance with the findings of Jones et al. (11) that average dust
concentrations in cottonseed oil mills were nonuniform, ranging
from 0.3 to 7.6 mg/m3 . Clearly, though, the spatial concentration
variation is specific to the arrangement of the technological
equipment in the mill, and therefore has to be examined in any
particular mill before conclusions can be drawn.
The inhomogeneity of cotton dust concentrations in the
hulling-separation room suggests that stationary samplers per
se may be inadequate for assessing the potential occupational
exposure of workers (i.e. the concentrations in contact with
humans (23,24) in cottonseed oil mills. Specifically, to secure a
continuous and profitable operation of the mill machinery, at
least some operators may be more prone to work in areas characterized
by a potential for process failure. Such critical operations
include the cottonseed cake handling: seed charging onto and
discharge from the feeder, smooth running of the pneumoconveyors,
and efficient delinting and hulling of the seeds. Yet these
processes, which frequently require machinery maintenance,
are also the ones associated with the major dust emissions in
the mill.
A valid question is the significance of the small difference
between the readings of the stationary personal samplers (SPSs)
and those of the roving samplers (RPSs). We assume that the data
collected by the SPS (designated by X) and the RPS (Y) consist
of independent random samples drawn from two populations
that are not necessarily identical, since Y represents a population
of airborne PM that potentially interacted with the worker (i.e.,
were affected by its mere presence or were biasedly sampled as
a result of the worker’s task). However, testing the hypothesis
H0:µx = µy when σ 2 x = σ 2 y (the Behrens-Fisher problem) is
one of the acknowledged unsolved problems in statistics, which
is worked out only by approximation methods. (25) Yet when
σ 2 x = σ 2 y = σ 2 accepting or rejecting H0 follows a standard
two-sample t-test. Thus, in order to be able to compare the two
samples (SPS vs. RPS) the test of H0:σ 2 x = σ 2 y should precede
the test of H0:µx = µy. To test the hypothesis H0:σ 2 x = σ 2 y versus
H1:σ 2 x = σ 2 y we employ a standard F-test for X ∼ N(µx,σ2 x ) and
Y ∼ N(µy,σ2 y ); H0 is accepted at the α level of significance if
F1−α/2,m−1,n−1 < S2 y
S 2 x
OCCUPATIONAL EXPOSURE TO COTTON DUST 127
< Fα/2,m−1,n−1. [1]
Taking the level of significance at α = 2% and using the data in
Table I H0 cannot be rejected, and we therefore accept the null
assumption that the SPSs and the RPSs drew air samples (at proximate
locations) from populations having identical variance. In
fact, since Sy/Sx ∼ = 1 and the sample size is large (Table I), the
central limit theorem implies that the t-test is robust enough to
withstand small deviations from the null assumption. (26)
Next, we test the hypotheses H0:µx = µy versus H1:µx =
µy, employing a standard t-test for two independent random
samples: X ∼ N(µx,σ 2 ) and Y ∼ N(µy,σ 2 ). Defining the
pooled variance,
S 2 p = (n − 1)S2 x + (m − 1)S2 y
, [2]
n + m − 2
where n and m are the sample size of SPS and RPS, respectively,
and
T = ¯X − ¯Y − (µx − µy)
, [3]
Sp
1
n
+ 1
m
H0 is accepted at the α level of significance if −tα/2,n+m−2 <
T < tα/2,n + m−2. Our data show that at 5 percent level of significance
H0 should be rejected, suggesting that the SPSs and the
RPSs sample different populations. The power of this calculation
(associated with making a Type II error—accepting a false H0)
is 0.7, i.e. β = 30%. It is noteworthy that, rigorously, both the F
and t tests are suitable for normal populations rather than to lognormal
(or other) populations, which usually better represent
the size distribution of airborne PM. Nevertheless, the probabilistic
behavior of these robust tests for n, m ≫ 1 is usually
minimally affected by the non-normality of the two populations
being sampled. (25)
A major fraction of the elutriated cotton dust found in the
hulling-separation room consisted of respirable particles (Table
II). For quantifying potential exposure of workers in the hullingseparation
department to respirable cotton dust, a correlation
between the average concentration of cotton dust at breathing
height, as measured by the stationary VEs, and the concentration
actually in contact with the workers, as measured by the
RPSs, is required. To accomplish this task, the relationships
among the VE, SPS, and RPS measurements were sought. A
regression analysis between elutriated concentrations (the independent
variable) and SPS total mass readings (the dependent
variable) in adjacent locations is shown in Figure 3.
For low airborne dust loading (< 15 mg/m 3 ) a linear regression
between the VE and SPS readings seems to be adequate
(R 2 = 0.9972). Differences between the sampler readings result
from the 15 µm cutoff size of the VE cotton dust sampler,
which readings therefore do not include most of the airborne
lint. Yet due to their shape these short fibers may still be
inhalable, (27) as evident from the high concentrations measured
by the RPSs (see Table I). For high airborne dust loading (total
mass concentration ∼15 mg/m 3 ) there is no correlation between
the VE and SPS readings. While the total suspended mass varies
very little, elutriated concentrations almost doubled from < 4to
> 7 mg/m 3 .
We suggest that the variability in VE readings can be attributed
to changes in the humidity content of the particulate
matter. The humidity content of the seeds, which in order to be
accepted to the mill should not exceed 12 percent, directly affects

128 S. TABAK ET AL.
FIGURE 3
Regression of total mass concentrations measured by
stationary personal samplers (SPS) against elutriated dust
concentrations (VE) measure simultaneously in adjacent
locations. The filled squares are data points from Table I, the
horizontal line is an eye guide only.
the size distribution of dust released during hulling-separation–
oil extraction processes. First, the higher the water content of
the dust, the more adhesive the particles are; at high concentrations
this causes the mode of the size distribution to shift toward
larger particles. Second, low humidity of the cottonseeds may
lead to higher dust concentrations, characterized by a varying
coarse fraction of particles >15 µm.
Figure 4 depicts the relationship between the SPS and the
RPS readings. Since similar samplers were used for stationary
and the roving sampling, and since the sampler mode of
operation and the sampling height were identical, it is a common
error to seek a linear regression between the two data
sets. The positive intercept of such a regression line is often
taken to represent the personal cloud. However, the best linear
regression line fitting our data is characterized by only
R 2 = 0.86, whereas higher order polynomial regressions are
characterized by higher R 2 values (see Figure 4). It seems that
the different dynamics of SPS and RPS sampling, which can
be attributed mainly to the distinct sampling kinematics and
to the associated flow fields prevailing at the samplers openings,
gives rise to a nonlinear relationship even among like
samplers.
Figure 5 reveals that the nonlinear relationships presented
in Figures 3 and 4 lead to a third-order polynomial regression
curve between concentrations measured by VE according to the
OSHA guidelines and readings obtained by personal samplers
attached to workers. The nature of the nonlinear relationship
arises from (1) different cutoff size of the samplers, (2) different
dynamics of the samplers, and (3) different conditions at the
sampler inlets that reflect on the characteristics of the particle
influx.
FIGURE 4
Regression analysis between average dust concentrations
measured by stationary personal samplers (SPS) and personal
samplers (RPS) attached to workers who were working in
nearby locations in the hulling-separation room. Data (filled
squares) were collected simultaneously during normal
operation conditions.
Using the empirical expression given in Figure 5, occupational
exposure of workers in cottonseed oil mills can be predicted
from measurements made by stationary vertical elutriators.
Although implementation of this procedure is not
economically sound (due to the higher cost of VEs vs. personal
samplers), predictions based on elutriated concentrations may be
more reliable for estimating the average exposure to lint-free respirable
cotton dust of the whole shift. Elutriated concentrations
FIGURE 5
Surrogate to cotton dust occupational exposure of workers in
the hulling-separation room. The solid line is the best
empirical prediction of respirable particulate matter from
stationary sampling of airborne PM by vertical elutriator.

are also more convenient and easy to obtain by an industrial
hygienist. Nevertheless, without correcting the stationary VEs
readings using the third-order polynomial regression curve presented
in Figure 5, predictions of worker’s exposure to airborne
cotton dust in cottonseed oil mills (and possibly other cotton,
textile, and yarn manufacturing industries) are significantly
underestimated.
There is an ongoing debate whether endotoxin is the primary
cause of respiratory diseases. However, respiratory disorders
undoubtedly appear to be more pronounced in people exposed
to higher concentrations of airborne endotoxin. 6−8,27,28 Indeed,
endotoxin is a more specific index of exposure to cotton dust,
showing a much stronger dose-response relationship, (28) than
airborne dust concentrations obtained by gravimetric methods.
Therefore, to predict occupational exposure more accurately it
will be helpful if endotoxin could be correlated against elutriated
concentrations. This will render endotoxin as a quantitative,
rather than a qualitative, surrogate marker for the yet unknown
pathogenic components of cotton dust. If such a correlation between
endotoxin and VE readings can be found, our results combined
with this correlation may provide a way of predicting the
inhaled endotoxin dose.
CONCLUSIONS
Nonuniform airborne dust concentrations were found at
breathing height in the hulling-separation department of a cottonseed
oil mill. A major fraction of the dust (wt %) consisted of
particles larger than 15 µm. Differences in readings of the vertical
elutriators, stationary personal samplers, and roving personal
samplers are statistically significant, and are attributed to the distinct
nature of the sampler dynamics during sampling. The high
total mass readings show that workers were exposed to very high
levels of nonthoracic airborne dust, which upon inhalation tends
to deposit in the extrathoracic airways. The high elutriated mass
concentrations suggest that workers were exposed to respirable
cotton dust levels higher than the 1 mg/m 3 OSHA PEL for cotton
dust. Regression analysis was performed on data obtained by
the stationary vertical elutriators (independent variable) and data
gathered by personal samplers attached to workers (dependent
variable). The empirical expression obtained can be used for estimating
the potential occupational exposure to cotton dust of
workers in the hulling-separation room of cottonseed oil mills.
For a cottonseed batch characterized by uniform endotoxin content,
it is reasonable to assume that the higher the level of dust
the higher the workers’ exposure to potential respiratory disease
inducing bioactive agents.
ACKNOWLEDGMENTS
Semion Tabak gratefully acknowledges support from the
Center for Absorption in Science, Israeli Ministry of Immigrant
Absorption. The authors would like to thank the general
manager of the cottonseed oil mill for allowing the mill to be
sampled.
OCCUPATIONAL EXPOSURE TO COTTON DUST 129
REFERENCES
1. Bailey, A.E.; Hilditch, T.P.; Longenecker, H.E.; Markley, K.S.:
Cottonseed and Cottonseed Products. Interscience Publishers, Inc.,
New York (1948).
2. Morey, P.R.: Bract and Leaf in the Process Stream in Cottonseed
Oil Mills. Am Indus Hyg Assoc J 42:244–246 (1981).
3. Parnel, C.B.; Gracy, J.A.; Clark, C.P.; et al.: Engineering Controls
to Lower Dust Level in Cottonseed Oil Mills. Transactions of the
ASAE 25(1):170–174 (1982).
4. Smith, K.J.: Status of Cottonseed Oil Mills Compliance with Environmental
and Safety Regulations. Oil Mill Gazetteer 79(11):36–
38 (1975).
5. Tabak, S.; Chernenko, T.; Glushenkova, A.: Lipid Composition of
Oil Component in Dust. J Chem Nat Comp 3:142–143 (1989).
6. Rylander, R.; Morrey, P.; Bethea, R.: Endotoxin Content in Cottonseed
Oil Mill Dust. Am Indus Hyg Assoc J 44(5):330–332 (1983).
7. Smid, T.; Heederick, D.; Houba, R.; et al.: Dust and Endotoxin-
Related Respiratory Effects in the Animal Feed Industry. Am Rev
Respir Dis 146:1474–1479 (1992).
8. Sigsgaard, T.; Pedersen, O.F.; Juul, S.; et al.: Respiratory Disorders
and Atopy in Cotton, Wool, and Other Textile Mill Workers in
Denmark. Am J Indus Med 22:163–184 (1992).
9. Glindmeyer, H.V.; Lefant, J.J.; Jones, R.N.; et al.: Cotton Dust and
Across-Shift Change in FEV1 as Predictors to Annual Change in
FEV1. Am J Respir Crit Care Med 149:584–590 (1994).
10. Lear, D.H.: Controlling Dust in Cottonseed and Soybean Mills. Oil
Mill Gazetteer 89(12):36–40,42 (1985).
11. Jones, R.N.; Carr, J.; Glindmeyer, H.; et al.: Respiratory Health and
Dust Levels in Cottonseed Oil Mills. Thorax 32:281–286 (1977).
12. Glindmeyer, H.W.; Lefant, J.J.; Jones, R.N.; et al.: Exposure-
Related Declines in Lung Function of Cotton Textile Workers:
Relationship to Current Workplace Standards. Am Rev Respir Dis
144:675–683 (1991).
13. Zey, J.N.; Piacitelli, G.M.; Gones, W.G.: Characterization of Airborne
Cotton Dust Concentrations in the Non-Textile Cotton Industry.
Am Indus Hyg Assoc J 45(8):538–546 (1984).
14. 43 Fed. Reg. 27351. Occupational Safety and Health Administration:
Occupational Exposure to Cotton Dust; Final Mandatory
Occupational Safety and Health Standards. Codified at 29 CFR
1910.1043 (1978), and 50 Fed. Reg. 51120. Occupational Safety
and Health Administration: Occupational Exposure to Cotton Dust;
Final Rule. Codified at 29 CFR 1910.1043 (1985).
15. Matlock, S.W.; Wiederhold, L.R.; Parnel, C.B.: Particle Sizing of
Dust Found in Cottonseed Oil Mill. Transactions of the ASAE
19(5):970–976 (1976).
16. Mark, D.; Vincent, J.H.: A New Personal Sampler for Airborne
Total Dust in Workplaces. Ann Occup Hyg 30:89–102 (1986).
17. Hinds, W.C.: Aerosol Technology. Properties, Behavior, and Measurement
of Airborne Particles. John Wiley & Sons, Inc., New York
(1999).
18. Allen, T.: Particle Size Measurement. Chapman and Hall, London
(1990).
19. American Society of Heating, Refrigeration, and Air Conditioning
Engineers: Handbook of Fundamentals of Heating, Refrigeration,
and Air Conditioning. ASHRAE Inc., Atlanta (1997).
20. American Conference of Governmental Industrial Hygienists:
Advances in Air Sampling. ACGIH, Cincinnati, OH (1988).
21. Kragelsky, I.V.; Dobychin, M.N.; Kombalov, V.S.: Friction and
Wear Calculation Methods. Pergamon Press, Oxford, UK (1982).

130 S. TABAK ET AL.
22. Environmental Protection Agency: Air Quality Criteria for Particulate
Matter, Vol. I. EPA Pub. No. EPA/600/P-95/001bF. EPA,
Washington, DC (1996).
23. Lioy, P.: Assessing Total Human Exposure to Contaminants. Environ
Sci Technol 24(7):938–945 (1990).
24. Georgopoulos, P.G.; Lioy, P.: Conceptual and Theoretical
Aspects of Human Exposure and Dose Assessment.
J Expos Anal Environ Epidemiol 4(3):253–285
(1994).
25. Larsen, R.J.; Mark, M.L.: An Introduction to Mathematical Statistics
and Its Applications. Prentice-Hall, Englewood Cliffs, NJ
(1986).
26. Johnson, R.A.; Bhattacharyya, G.K.: Statistics Principles and
Methods. John Wiley & Sons, Inc., New York (1992).
27. Broday, D.M.; Fichman, M.; Shapiro, M.; et al.: Motion of
Spheroidal Particles in Viscous Vertical Shear Flows. Phys Fluids
10:86–100 (1998).
28. Rylander, R; Haglind, P.: Exposure of Cotton Workers in an Experimental
Cardroom with Reference to Airborne Endotoxins. Environ
Health Perspect 66:83–86 (1986).
29. Castellan, R.M.; Olenchock, S.A.; Kinsley, K.B.; et al.: Inhaled
Endotoxin and Decreased Spirometric Values: An Exposure–
Response Relation for Cotton Dust. N Eng J Med 317:605–610
(1987).