Downloaded from http://rsif.royalsocietypublishing.

org/ on March 29, 2015

J. R. Soc. Interface (2009) 6, S703–S714

doi:10.1098/rsif.2009.0388.focus
Published online 7 October 2009

Exhaled droplets due to talking

and coughing
Xiaojian Xie1,2, Yuguo Li1,*, Hequan Sun1 and Li Liu1
1
Department of Mechanical Engineering, University of Hong Kong, Pokfulam Road,
Hong Kong SAR, People’s Republic of China
2
Faculty of Power Engineering, Nanjing Normal University, 78 Bancang Road,
Nanjing, People’s Republic of China

Respiratory infections can be spread via ‘contact’ with droplets from expiratory activities such
as talking, coughing and sneezing, and also from aerosol-generating clinical procedures. Droplet
sizes predominately determine the times they can remain airborne, the possibility of spread of
infectious diseases and thus the strategies for controlling the infections. While significant incon-
sistencies exist between the existing measured data on respiratory droplets generated during
expiratory activities, a food dye was used in the mouth during measurements of large droplets,
which made the expiratory activities ‘unnatural’. We carried out a series of experiments using
glass slides and a microscope as well as an aerosol spectrometer to measure the number and size
of respiratory droplets produced from the mouth of healthy individuals during talking and
coughing with and without a food dye. The total mass of respiratory droplets was measured
using a mask, plastic bag with tissue and an electronic balance with a high precision. Consider-
able subject variability was observed and the average size of droplets captured using glass slides
and microscope was about 50–100 mm. Smaller droplets were also detected by the aerosol
spectrometer. More droplets seemed to be generated when a food dye was used.

Keywords: droplets; talking; coughing; airborne infection

1. INTRODUCTION Roberts 1967; Fairchild & Stamper 1987; Papineni &

Rosenthal 1997; Fennelly et al. 2004; Yang et al. 2007),
During human expiratory activities such as talking,
of which at least three detailed measured data on respirat-
laughing, coughing and sneezing, many droplets of
ory droplets exist (Duguid 1946; Loudon & Roberts 1967;
saliva and other secretions are expelled from the respir-
Papineni & Rosenthal 1997). However, there exist signifi-
atory tract (the mouth and nose). It is now known that
cant inconsistencies between the existing data, as
respiratory infections can be spread by these droplets
reviewed by Nicas et al. (2005) and Morawska (2006).
and their residues after evaporation (Garner 1996).
Difficulties in capturing and subsequent sizing of res-
Larger droplets may rapidly settle out of the air and
piratory droplets are mainly due to the small sizes and
thus contribute to disease transmission to individuals
rapid evaporation of droplets after release, and evapor-
in close proximity; smaller droplets may remain sus-
ation loss can not be neglected because of the typical
pended for a long time and contribute to disease
size range of 0–200 mm. There exist significant inconsis-
transmission over larger distances. Their sizes predomi-
tencies between the existing data using different
nately determine the times they can remain airborne
methods, which failed to consider the effect of evapor-
and thus the possibility of spread of infectious diseases
ation, which is critical. Different methods and
if these respiratory droplets contain infectious patho-
instruments employed are likely to have contributed to
gens (Wells 1934; Wan & Chao 2007; Xie et al. 2007).
the inconsistency in the findings. Moreover, we noticed
Moreover, the size distribution of such droplets influ-
that when the deposition method was applied, a dye
ences the type of microorganisms that may be carried
was used in the mouth to easily distinguish the droplets
as well as the strategies for controlling the infections.
(Duguid 1945, 1946; Loudon & Roberts 1967). The
Previous studies of hospital ventilation and infection
introduction of the dye may influence the formation of
control reveal a need for a more accurate measurement
secretion in the mouth and thus the number or size of
of respiratory droplets (Li et al. 2005; Qian et al. 2006).
droplets produced. Even though the subjects were told
In the literature, many researchers have investigated dro-
that they could swallow their saliva or spit it out and
plet generation during expiratory activities using different
try to make their expiratory activities as natural as poss-
methods (Jennison 1942; Duguid 1945, 1946; Loudon &
ible, they may still have had psychological effects that
could change saliva secretion. The number and size of
*Author for correspondence (liyg@hku.hk). droplets expelled without food dye were not studied.
One contribution of 10 to a Theme Supplement ‘Airborne We considered whether the introduction of a dye influ-
transmission of disease in hospitals’. ences the production of droplets during talking or

Received 2 September 2009

Accepted 4 September 2009 S703 This journal is q 2009 The Royal Society
 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

S704 Droplets due to talking and coughing X. Xie et al.

coughing. We demonstrated that droplet stain marks another dust monitor away from the box to monitor
even without dye could be distinguished on clean glass the particle concentration in the clean room. A sche-
slides under a microscope, hence we could measure the matic diagram and a photo of the experimental setup
naturally expelled droplets. are shown in figure 1.
The studies reported here were undertaken to deter- All the subjects were from our research group, non-
mine the number and size of respiratory droplets emitted smoking and around 20– 40 years old. We studied two
by different healthy people during talking and coughing. expiratory activities, i.e. talking and coughing. During
Experiments both with and without food dye were each talking experiment, the subject counted from 1
conducted. To determine the number and size, subjects to 100 loudly and slowly, and 56 slides would be used.
were told to speak and cough into a small chamber As for each coughing experiment, 20 coughs were
which was constructed according to the design by previous made into the box and 60 slides would be used. Usually
investigators. Large droplets produced were recovered 10 pieces of WSP were used in each experiment, four on
under a microscope from the glass slides placed inside the ground and two on each of the other three walls.
the chamber by the method of impaction and settling. The placement of these slides and WSP (in black) is
The concentration of small droplets was monitored by a shown in figure 2.
portable dust monitor by the method of air sampling. As Experiments were conducted with the following
to the measurements of total mass, respiratory droplets procedures.
were collected with a surgical face mask and plastic bag
with tissue inside, which were weighed before and after (i) Before each experiment, all the slides were
the collection with a high-precision electronic balance. cleaned first with 75 per cent alcohol and then
with distilled water, rubbed with cloth to make
them dry and clean and stored inside the slide
2. MATERIAL AND METHODS boxes to partition them off. The air-tight box
was cleaned and then put into the clean room
2.1. Measurements of droplet sizes and numbers
in which the HEPA system was always on.
Experiments were undertaken to determine the number (ii) The glass slides and WSPs were loaded into the
and size of respiratory droplets emitted by a group of box, and the dust monitor was turned on inside
healthy people during talking and coughing. A small and outside the box to monitor the particle
air-tight box was constructed with the same dimensions concentration. After about 2 h, the particle con-
as the design of Loudon & Roberts (1967), which was centration would be very low compared with
366 mm (14.4 inches) by 508 mm (20 inches) by that outside the room.
305 mm (12 inches) inside. In the following text, we (iii) The subject was asked to go into the clean room,
refer to the six walls as the front wall, back wall, left sit on the chair before the box and wait for
wall, right wall, ceiling and ground. The box was made 20 min. As the indoor particle concentration
of stainless steel except for the ceiling which was made would increase due to door opening and subject
of Perspex. At about two-thirds of the height of the entering, it took some time for the HEPA
front wall, an entry hole of 102 mm (4 inches) in diameter system to dilute the particle concentration.
was cut for the purpose of respiratory droplet expulsion (iv) The subject carried out the planned expiratory
from a subject’s mouth. A detachable air-tight door activity into the box through the entry hole. As
was screwed onto the entry hole. The left and right soon as this was completed, the subject screwed
walls were detachable. When droplets were expelled the entry hole to seal the box. The subject was
into the box from the entry hole, large ones would asked to wait for another 20 min before going
quickly settle or impact on the surfaces nearby and out for particle concentration measurement
leave stain marks, while small ones would totally evapor- inside the box, because opening the door would
ate into droplet nuclei, which could remain suspended in influence the indoor particle concentration.
the air. To collect the large droplets, microscope glass (v) After another 2 h, the researchers could go inside
slides and strips of water-sensitive paper (WSP) both the room, open the box and collect the glass
of standard size of 76 by 26 mm were attached to four slides and WSP cards, which would be labelled
walls of the box prior to each test, i.e. the back wall, left as to position. The box was then cleared for
wall, right wall and ground. WSP is a special slide strip the next experiment.
made of specially coated yellow paper that turns blue
when exposed to water droplets. These strips can give us To test the effect of food dye on droplet generation,
a quick and visual indication of droplet size and density. experiments were conducted. Food dye powder (Lemon
Small droplets or droplet nuclei suspended in the air Yellow Powder H1794) was dissolved in distilled water
were measured by a 16-channel dust monitor (Grimm and mass concentration was prepared at about 2.5 per
1.108, Germany), which could provide real-time size cent. Before expiratory activities, the subject was
measurements of particles from 0.3 to 20 mm. The asked to gargle using the food dye solution two or
sampling flow rate was 1.2 l min21 and reproducibility three times to stain the saliva. The subject could spit
was +2 per cent. A small hole was drilled in the out the excessive saliva if he/she felt that there was
centre of left and right walls to insert the air inlet of too much saliva in the mouth. During talking or cough-
the dust monitor. The box was placed in a clean room ing, the subject could swallow or spit the excessive
(3 m  3 m  3 m) equipped with a high-efficiency par- saliva if he/she felt there was too much saliva in the
ticulate air (HEPA) filtration system. We also put mouth, in order to make the expiratory activity as

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

Droplets due to talking and coughing X. Xie et al. S705

(a) (b)
HEPA
fan

3m
3m
3m

Figure 1. (a) Schematic diagram and (b) photo of the experimental setup.

B42 B41 R44 R41

right (R)

B12 B11
back (B, for talking) R14 R11

B43 B41 G44 G41

ground (G)

B13 B11 G14 G11

back (B, for coughing)
L14 L11
R
B m
5m
30
left (L)
366 mm

508 mm G L44 L41

Figure 2. Schematic diagram of slide placement in the test box.

natural as possible. After certain times of counting or obtained by subtracting the concentration outside
coughing, the subject was asked to gargle using the the box from that inside the box.
food dye solution again.
These slides were completely scanned under a
2.2. Measurements of total mass of droplets
microscope (Leica DM1000, Leica Microsystems,
Germany) visually. We distinguished droplet stain It is known that droplets evaporate quickly after they
marks by morphology. Usually, a ‘ring’ could be are produced. To accurately measure the total mass of
observed surrounding the dried residue of a droplet. droplets produced during talking and coughing, evapor-
Every droplet stain mark was photographed with a ation should be avoided as much as possible. In our
high-resolution CCD camera (Leica DFC320, Leica experiments, an air-tight plastic bag with tissue inside
Microsystems) connected to the microscope. The was used to collect the droplets. A subject put his/her
sizes of these stain marks were analysed using an mouth inside the bag and finished the requested expira-
image-processing software developed in the labora- tory manoeuvres, i.e. counting from 1 to 100 or
tory. The particle concentration in different size coughing 20 times. During this process, condensation
ranges produced during expiratory activities could be of exhaled breath would occur. So we also used a

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

S706 Droplets due to talking and coughing X. Xie et al.

mask to collect the droplets. Subjects wore a surgical talking experiments without the food dye was 323
face mask and then counted or coughed. The mask or during the process of counting from 1 to 100 for three
plastic bag with tissue was weighed before and immedi- times, i.e. 108 if counting from 1 to 100 only once.
ately after the collection with an analytical balance Averages of 315 and 273 were found during the process
with an accuracy of 0.1 mg (Shimadzu AUW 220, of counting from 1 to 100 when food dye and food dye
Japan). The total mass of droplets collected using with sugar solution were used, respectively. For 20
these two methods could then be estimated and coughs, the average value was only 108. Figure 5
compared. shows the number percentages of droplet stain marks
observed on different slides during talking and cough-
ing, which were calculated from the data of all
3. RESULTS experiments. We found that most of the droplets were
deposited on the ground. For talking experiments,
3.1. Results of respiratory droplet sizes
93.7 per cent of the large droplets were deposited on
and numbers
the ground. Among the slides on the ground, 57.4 per
Three male subjects and four female subjects partici- cent of the droplets deposited on the slides in the first
pated in the measurements of droplet sizes, of which row (about 0.1 m away from the mouth) and 27.4 per
three male subjects and two female subjects were all cent deposited on the slides in the second row (about
Chinese-speaking healthy adults and two female sub- 0.2 m away). Almost 90 per cent fell within a distance
jects (F3 and F4) were English-speaking healthy of 0.3 m. For coughing experiments, 80.9 per cent of
adults. Experiments with and without food dye were the droplets were deposited on the ground, which
carried out for talking, while only experiments without were almost evenly distributed on the four rows. Fifteen
food dye were done for coughing. The food dye solution per cent of the droplets could reach the back wall, which
has a taste like sea water. In order to make the subjects is more than 0.5 m away. This deposition pattern agrees
feel not so bad to have the dye in their mouth, we added well with that of droplet spots on WSP and also that in
sugar in three trials of the experiments with food dye. In experiments of Loudon & Roberts (1967). Compared
experiments with the food dye used, the subjects slowly with talking, droplets from coughing dispersed longer
counted aloud from 1 to 100 into the box. In the exper- and in a lager area.
iment without food dye, the subject slowly counted The images of droplet stain marks were analysed
aloud from 1 to 100 into the box three times. That using our image-processing software, and the sizes of
was done because in the test not so many droplets the droplet stain marks could be calculated. As we
were observed from the strips of WSP after one trial know, the shapes of these stain marks will not be totally
of talking. For each coughing experiment, the subjects circular. Here we choose an area-equivalent diameter to
were asked to cough 20 times. represent the size of the droplet stain mark. We also
Particle concentrations both inside and outside the need to determine the relation between the sizes of
box were monitored by dust monitors during the exper- the droplets while in their original spherical state and
iments. Figure 3 shows the particle concentration the sizes of the stain marks that the droplets leave on
history during one trial of coughing. When the door evaporation after impinging and flattening upon a
was opened, the particle concentration inside the slide. Based on the work of Duguid (1946), we assumed
room but outside the test box would first increase and that the stain marks left on glass slides were about
then decrease quickly because the HEPA filter in the three times the diameter of the original droplets.
room was on, which could remove particles. Before the We also carried out simple experiments to confirm
subject’s expiratory activities, particle concentration this ratio.
inside the air-tight box would not be higher than out- We also assumed that the droplet number distri-
side the air-tight box. Figure 3 shows that when the bution on each wall was the same as that on the slides
subject started to cough into the box, a higher concen- attached to this wall. Then the total numbers of dro-
tration inside the box was monitored, which means plets inside the box were estimated from the numbers
small droplets/particles (more than 0.5 mm) were of droplet stain marks counted on the slides. The
generated. However, only in some experiments was an total numbers of droplets produced by each subject in
obvious higher concentration inside the box observed different size intervals during talking (counting from 1
than that outside the box for some size channels to 100) and coughing (20 times) are summarized in
during the time period of talking or coughing. From table 2. Having these data, we also calculated the per-
those data, it was very difficult to calculate the num- centage of droplets in different diameter ranges and
bers of small droplets and droplet nuclei produced. cumulative percentage of droplets less than the stated
Thus, only results of the measured deposited droplets diameter ranges produced in different sets of exper-
obtained from glass slides were reported subsequently. iments, which are summarized in table 3. For talking
Droplet stain marks on the slides were scanned under experiments, almost all the droplets recovered were
a microscope and images were taken, one example of less than 500 mm. When food dye with sugar was
which is shown in figure 4a (no dye) and figure 4b used, a small fraction of droplets less than 5 mm were
(with dye). Droplet spots on WSP strips are shown in recovered. It is obvious that more droplets were pro-
figure 4c. The numbers of droplet stain marks observed duced when the food dye solution was used. During
on the slides in different experiments are summarized in the process of counting from 1 to 100, the number of
table 1. A great individual variability was observed. droplets produced without a food dye (average: 760)
The average number of droplets found in seven subjects’ was only about one-third of the droplets produced

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

Droplets due to talking and coughing X. Xie et al. S707

3.0
door open

particle concentration ( × 107 count m–3)

2.5

2.0
outside the box

1.5
door open

1.0
inside the box

0.5

start coughing
0
11:00 11:15 11:30 11:45 12:00 12:15 12:30 12:45 13:00 13:15 13:30 13:45 14:00
time

Figure 3. Particle concentration history inside and outside the box during the entire process of one trial of coughing experiment
conducted on 25 October 2006. Filled circle, diameter greater than 0.50 mm (outside the box); open diamond, diameter greater
than 0.50 mm (inside the box).

(a) (b) (c)

Figure 4. Droplet stain mark: (a) on the slide surface when no food dye was used; (b) on the slide surface when food dye was used;
(c) on a WSP.

when a food dye was used (average: 2273). In all the mouth. On the one hand, they would stimulate the
coughing experiments, no food dye was used and secretion of saliva. On the other hand, droplet evapor-
many droplets larger than 500 mm were observed. No ation rate decreases if food dye and sugar were added.
droplets less than 5 mm were detected. An average of For all the coughing experiments, no food dye was
800 droplets could be observed from 20 coughs. used. About 2.5 per cent of the droplets were less
The cumulative percentages of large droplets than 20 mm and 1.4 per cent less than 10 mm. Only 20
expelled during talking and coughing are shown in per cent of the droplets were less than 50 mm and 64
figure 6. For talking experiments, when food dye with per cent of the droplets less than 100 mm. Compared
sugar was used, about 15 per cent of the droplets were with talking, a higher percentage of droplets larger
less than 10 mm, 52 per cent of the droplets less than than 500 mm were observed. The difference in droplet
50 mm and 80 per cent of the droplets less than size distribution may lie in the difference of droplet gen-
100 mm. When food dye was used, about 5 per cent of eration mechanism between talking and coughing.
the droplets were less than 20 mm, 49 per cent of the Size distribution information can be presented in
droplets less than 50 mm and 83 per cent of the droplets many forms. Here we also divided the percentage of dro-
less than 100 mm. When no food dye was used, only plets in each interval by the width of that interval and
about 3 per cent of the droplets were less than 20 mm, plotted the figure of percentage mm21 versus droplet
37 per cent of the droplets less than 50 mm and 82 per diameter (figure 7), which includes the results in the
cent of the droplets less than 100 mm. More small dro- studies of Duguid (1946; labelled as ‘T-Duguid’ and
plets were recovered when food dye was used, ‘C-Duguid’) and Loudon & Roberts (1967; labelled as
especially when sugar was added. This may be because ‘T-L&R’ and ‘C-L&R’). In this figure, only the data
of the introduction of a food dye and sugar into the of droplets sampled by the surface deposition method

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

S708 Droplets due to talking and coughing X. Xie et al.

30
B42 B41 R44 R41

25 right (R)
B12 B11
back (B, for talking) R14 R11
20 B43 B41 G44 G41
percentage (%)

ground (G)
15
B13 B11 G14 G11
back (B, for coughing) L14 L11
R
10 B
left (L)
L

G L44 L41
5

0
G11
G21
G31
G41
G12
G22
G32
G42
G13
G23
G33
G43
G14
G24
G34
G44

R11
R21
R31
R41
R12
R22
R32
R42
R13
R23
R33
R43
R14
R24
R34
R44
B11
B21
B31
B41
B12
B22
B32
B42
B13
B23
B33
B43
L11
L21
L31
L41
L12
L22
L32
L42
L13
L23
L33
L43
L14
L24
L34
L44 slide name

Figure 5. Number percentages of droplet stain marks observed on different slides. Unfilled bar, talking; filled bar, coughing.

Table 1. Total number of droplet stain marks observed on all slides (56 slides used for talking and 60 slides used for coughing)
in different experiments.

talking (counting from 1 to 100) coughing (20 times)

food dye with

no food dye (three repeats)a food dye sugar no food dye

subject M1 M2 M3 F1 F2 F3 F4 M1 F1 M1 M3 F1 M1 M2 M3 F1
number 456 1173 41 96 395 40 58 518 113 303 335 183 169 133 38 91
average 323 (108 for each talking) 315 273 (for each 108 (for each coughing)
(for each talking)
talking)
a
These data were for subjects who counted from 1 to 100 for three times, while the rest of the experiments were done with
one time talking (counting from 1 to 100 only once) or one time coughing (20 coughs) only.

were considered in the percentage calculation. The figure 6. However, there is no great difference between
medium droplet size in each interval was used as the the size distributions of droplets produced by the differ-
characteristic droplet diameter. This droplet size distri- ent types of expiratory activities. As to the effect of food
bution curve was the graphical representation of the dye, because the introduction of food dye slows down
frequency function or probability density function. the droplet evaporation, the peak of the frequency func-
From figure 7a, we could find that the peak diameter tion moves towards larger droplet diameter when no
lay between 35 and 50 mm in current talking exper- food dye was used, especially when compared with
iments without food dye, between 30 and 45 mm in Duguid’s (1946) results. This large difference may also
current talking experiments with food dye, between 15 be explained by the different droplet collection methods
and 25 mm in Duguid’s (1946) study. In the study of used. Droplets were collected inside a box in the current
Loudon & Roberts (1967), droplets of 8 and 100 mm study, while in Duguid’s (1946) study droplet spray was
in diameter almost have the same frequency. directed at a slide in front of the mouth. Droplets
Figure 7b shows the results of coughing. The common- experienced different evaporation times during their
est diameter lay between 35 and 100 mm in current aerial transport.
coughing experiments without food dye, between 15
and 25 mm in Duguid’s (1946) study and between 22
3.2. Total mass of droplets
and 73 mm in Loudon & Roberts (1967). We notice
that a larger proportion of droplets between 5 and The total mass of droplets collected using surgical face
20 mm could be generated during talking than coughing mask and plastic bag with tissue inside is shown in
when no food dye was used, which also can be seen in table 4. Considerable subject variability was observed,

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

Droplets due to talking and coughing X. Xie et al. S709

Table 2. Estimated total numbers of droplets in different diameter ranges emitted during talking or coughing (M, male subject;
F, female subject; the sizes of droplets used the values at sampling positions).

each coughing
each talking (counting from 1 to 100) (20 times)

no food dye food dye food dye with sugar no food dye
size range
(mm) M1 M2 M3 F1 F2 F3 F4 M1 F1 M1 M3 F1 M1 M2 M3 F1

0–5 0 0 0 0 0 12 0 0 0 66 92 0 0 0 0 0
5–10 0 0 5 0 0 6 7 24 0 303 309 115 0 44 0 0
10 –15 2 11 24 0 0 0 2 14 0 158 208 138 0 7 0 7
15 –20 12 35 11 0 9 7 0 165 14 82 108 79 7 15 0 0
20 –25 14 86 13 7 28 0 4 230 28 87 72 72 0 0 0 0
25 –30 28 154 7 12 32 0 7 280 43 115 93 43 7 44 36 21
30 –35 40 187 0 2 58 3 7 345 43 122 86 57 7 28 0 14
35 –40 65 239 4 0 79 0 0 302 36 72 93 43 7 85 0 14
40 –45 84 229 0 0 65 2 9 338 50 72 57 43 42 71 28 50
45 –50 50 246 0 9 65 2 9 259 43 152 86 57 7 50 14 21
50 –75 271 854 16 57 236 20 31 763 237 230 446 216 218 281 57 158
75 –100 256 369 7 62 147 7 19 420 159 299 316 180 253 180 100 172
100 –150 180 233 7 48 103 29 24 335 100 251 259 161 387 63 21 129
150 –200 54 58 2 14 56 6 14 146 28 121 36 28 145 43 7 28
200 –250 15 23 0 4 25 2 0 74 21 61 28 53 66 13 8 21
250 –300 9 14 2 2 7 2 2 7 7 0 36 7 17 0 0
300 –350 4 4 2 2 2 2 15 0 92 30 58 20 13
350 –400 7 4 4 2 7 0 8 7 0
400 –450 0 2 2 0 0 17 0
450 –500 0 2 14 8 10 0
500 –1000 3 14 69 8
1000–1500 7
total 1091 2749 100 225 918 100 135 3738 809 2213 2425 1322 1331 952 271 648
average 760 2273 1986 800

Table 3. Percentage of droplets in different diameter ranges emitted during talking or coughing (at sampling position).

coughing coughing
talking (counting from 1 to 100) (20 times) talking (counting from 1 to 100) (20 times)

no food food dye no food size no food food dye

size range dye (T- food dye with sugar dye (C-N) range dye (T-N) food dye with sugar no food dye
(mm) N) (%) (T-D) (%) (T-S) (%) (%) (mm) (%) (T-D) (%) (T-S) (%) (C-N) (%)

0–5 0.2 0 2.7 0 ,5 0.2 0 2.7 0

5–10 0.3 0.5 12.2 1.4 ,10 0.6 0.5 14.8 1.4
10 –15 0.7 0.3 8.5 0.4 ,15 1.3 0.8 23.3 1.8
15 –20 1.4 3.9 4.5 0.7 ,20 2.7 4.8 27.8 2.5
20 –25 2.9 5.7 3.9 0 ,25 5.5 10.4 31.7 2.5
25 –30 4.5 7.1 4.2 3.4 ,30 10.1 17.6 35.9 5.9
30 –35 5.6 8.5 4.4 1.5 ,35 15.6 26.1 40.4 7.4
35 –40 7.3 7.4 3.5 3.3 ,40 22.9 33.5 43.8 10.7
40 –45 7.3 8.5 2.9 6.0 ,45 30.2 42.0 46.7 16.7
45 –50 7.2 6.6 4.9 2.9 ,50 37.4 48.7 51.7 19.6
50 –75 27.9 22.0 15.0 22.3 ,75 65.3 70.7 66.6 41.8
75 –100 16.3 12.7 13.3 22.0 ,100 81.6 83.4 80.0 63.9
100 –150 11.7 9.6 11.3 18.7 ,150 93.4 93.0 91.2 82.6
150 –200 3.8 3.8 3.1 7.0 ,200 97.2 96.8 94.3 89.6
200 –250 1.3 2.1 2.4 3.4 ,250 98.5 98.9 96.7 92.9
250 –300 0.7 0.3 0.7 0.5 ,300 99.2 99.2 97.4 93.5
300 –350 0.3 0.3 2.0 2.8 ,350 99.5 99.5 99.5 96.3
350 –400 0.3 0.2 0.1 0.2 ,400 99.8 99.7 99.6 96.5
400 –450 0.1 0 0 0.5 ,450 99.9 99.7 99.6 97.1
450 –500 0 0.3 0.1 0.3 ,500 99.9 100 99.8 97.4
500 –1000 0.1 0 0.2 2.4 ,1000 100 100 100 99.8
1000–1500 0 0 0 0.2 ,1500 100 100 100 100

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

S710 Droplets due to talking and coughing X. Xie et al.

100

cumulative percentage (%)

90 T-S
80
T-D
70
T-N
60
C-N
50
40
30
20
10
0
<10 <20 <50 <100 <200 <500 <1000
droplet diameter (µm)

Figure 6. Cumulative percentage of droplets less than the stated diameter produced by talking and coughing (T-S, talking
experiment with food dye with sugar used; T-D, talking experiment with food dye used; T-N, talking experiment without
food dye; C-N, coughing experiment without food dye).

(a) 2.5 (b)

2.0
percentage µm–1

1.5

1.0

0.5

0
1 10 100 1000 1 10 100 1000
droplet diameter (µm) droplet diameter (µm)

Figure 7. Percentage mm21 versus droplet diameter detected on the sampling slides: (a) talking (T-N, talking experiment with-
out food dye (filled square); T-D, talking experiment with food dye used (filled star); T-Duguid, talking experiment in Duguid
(1946) (open circle); T-L&R, talking experiment in Loudon & Roberts (1967) (open triangle)); (b) coughing (C-N, coughing
experiment without food dye (filled square); C-Duguid, coughing experiment in Duguid (1946) (open circle); C-L&R, coughing
experiment in Loudon & Roberts (1967) (open triangle)). (Note: see the discussion in the text on the calculation of the vertical
axis value.)

Table 4. Total mass of droplets collected using surgical face mask and plastic bag with tissue inside.

weight (mg)

activity M1 M2 M3 M4 M5 M6 M7 F1 F2 average (mg)

talking (counting from 1 to 100)

mask 3.7 41.8 61.3 — — — — 15.7 1 18.7
5.5 12.6 7.7
bag 69.4 48.8 121.7 — — — — 113.7 66.2 79.4
38.6 151.5
coughing (20 times)
mask 44.9 30.4 17.5 — — — — 15.1 4.6 22.9
31.5
16.2
bag 154.5 87.8 85.8 62.7 67.4 41.8 91.9 41.4 55.4 85.0

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

Droplets due to talking and coughing X. Xie et al. S711

which coincides with that in droplet size distribution 200

measurements. An average of 22.9 mg of fluid was

droplet size on the sampling slide (µm)

obtained during 20 coughs using the surgical face
mask method and 85 mg of fluid using a plastic bag 160
with tissue. And averages of 18.7 and 79.4 mg of fluid
were measured using mask and plastic bag, respectively,
during counting to 100. 120

80
4. DISCUSSION
The droplet numbers and sizes presented above were
40
obtained on glass slides at different sampling positions.
From the droplet origin (mouth) to the sampling pos-
ition, a droplet would evaporate and its size would
shrink. So it is not enough to know the size distribution 0 40 80 120 160 200
of droplets detected on sampling slides, which is not droplet size at the origin (µm)
the real size distribution of droplets generated during
expiratory activities. We need to know the droplet size Figure 8. Relationship between droplet size at the origin and
at the origin. How much one droplet loses water is predo- droplet size on the sampling slide. Black line, ground wall of
minantly determined by the time it takes to fly from the the box; dashed line, Y ¼ X.
droplet origin (mouth) to the sampling position, which
we could call the ‘residence time’. Because of the limit-
ation of the experimental design, we could not know respectively. We can see the size shift. At the origin,
the exact value of residence time for each droplet. To more droplets fall into a size range of 50– 100 mm. For
roughly estimate the droplet size change, we assume talking, only less than 10 per cent of the droplets were
that the residence time equals the time for the droplet less than 50 mm and more than 50 per cent of the dro-
freely falling to the same height of sampling position. plets were in the size range of 50– 75 mm when no
During the experiments, we also recorded the air temp- food dye was used. And for coughing, about 7 per
erature and relative humidity inside the box. The cent of the droplets were less than 50 mm and more
average temperature was 288C and relative humidity than 30 per cent of the droplets in the size range
was 70 per cent. Using the evaporation model for a of 50– 75 mm. We could also plot the figure of
freely falling pure liquid droplet described in Xie et al. percentage mm21 versus droplet diameter (figure 9)
(2007), we could roughly back-calculate the droplet size using the estimated data of droplet sizes at the origin.
at the origin. The peak diameter lay between 45 and 75 mm in current
Figure 8 shows the relationship between droplet size talking and coughing experiments. It indicated that
at the origin and droplet size on the sampling slide using the current deposition method only large droplets
when the sampling slide is placed on the ground of generated from the expiratory activities could be
the box. Because of droplet evaporation, the droplet sampled. A methodology that can cover the whole size
size at the sampling position is smaller than the droplet range is needed.
size at the origin. That is why in figure 8 the black line In the literature, Duguid (1945, 1946) found that the
is below the dashed line. When released from the same average number of expelled droplets was 250 by count-
height, larger droplets would quickly reach the ground ing aloud from 1 to 100. And an average of 1764
and their size would not change too much, while smaller droplets were obtained by Loudon & Roberts (1967)
droplets would evaporate quickly during the relatively for the same expiratory activity, in which 1652 droplets
slow falling process. From figure 8, we can see that if were recovered from droplet stain marks. Food dye was
the droplet size at the origin is less than 47 mm, the dro- used in both studies. In our experiments, we obtained
plet would dry out before it reaches the ground and an average of 760 droplets for talking without food
could not form a droplet stain mark on the glass slide. dye (2273 when food dye was used) using glass slides.
When the droplet size at the origin is larger than The average number of droplets produced during talk-
80 mm, droplet size changes very little. ing when food dye was used did not differ greatly
Figure 5 shows that almost all the droplets were from that noted by Loudon & Roberts (1967), but
detected on the ground. Among the droplets detected was higher than that recorded by Duguid (1946), even
on the box walls, more than 65 per cent of the droplets when no food dye was used. This may be because of
produced by talking are smaller than 75 mm. More than the vigour and loudness of the talking as pointed out
40 per cent of the droplets produced during coughing by Loudon & Roberts (1967) or the different types of
are smaller than 75 mm. According to figure 8, these food dye that perhaps could influence the secretion
droplets should have larger sizes at the origin, i.e. the of saliva. In our experiments, five subjects are Chinese
mouth. Table 5 summarizes the measured size distri- and only two subjects are native English speakers.
bution of droplets detected at the sampling positions The latter two non-Chinese subjects produced relatively
and the estimated size distribution of droplets at the fewer droplets than all other subjects. Difference in pro-
origin. The percentages and cumulative percentages of nunciation may induce the number difference of droplet
droplets in different diameter ranges were calculated, generation. Inouye (2003) suggested that the efficiency

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

S712 Droplets due to talking and coughing X. Xie et al.

(a) 3.0 (b)

percentage µm–1 2.5

2.0

1.5

1.0

0.5

0
1 10 100 1000 1 10 100 1000
droplet diameter (µm) droplet diameter (µm)

Figure 9. Percentage mm21 versus droplet diameter estimated at the origin: (a) talking (T-N, talking experiment without food
dye (filled square); T-D, talking experiment with food dye used (filled star); T-Duguid, talking experiment in Duguid (1946)
(open circle); T-L&R, talking experiment in Loudon & Roberts (1967) (open triangle)); (b) coughing (C-N, coughing exper-
iment without food dye (filled square); C-Duguid, coughing experiment in Duguid (1946) (open circle); C-L&R, coughing
experiment in Loudon & Roberts (1967) (open triangle)). (Note: see the discussion in the text on the calculation of the vertical
axis value.)

Table 5. Percentage of droplets in different diameter ranges emitted at the origin (i.e. mouth, estimated by using a simple
evaporation model) during talking or coughing.

talking (counting from 1 to 100) coughing (20 times)

no food dye (T-N) food dye (T-D) food dye with sugar (T-S) no food dye (C-N)

size range sampling mouth sampling mouth sampling mouth sampling mouth
(mm) position (%) origin (%) position (%) origin (%) position (%) origin (%) position (%) origin (%)

1–2 0 0 0 0 0 0 0 0
2–4 0.1 0 0 0 0.9 0 0 0
4–8 0.3 0.1 0.4 0 7.4 0 1.1 0
8–16 1.1 0.3 0.3 0 15.8 0.3 0.9 0
16–24 3.3 0.1 6.6 0 6.9 0.1 0.7 0.3
24–32 7.6 0.2 11.1 0 5.7 0 4.0 0
32–40 10.4 0.2 12.3 0 6.6 0.4 4.0 0.5
40–50 14.4 8.9 16.6 14.2 8.1 34.2 8.9 6.2
50–75 27.5 51.1 22.8 52.5 14.6 28.4 22.2 30.8
75–100 16.5 19.3 12.5 15.2 14.0 15.4 22.0 23.4
100– 125 7.7 8.3 6.3 6.3 6.6 7.3 12.3 15.0
125– 150 4.2 4.6 3.9 4.4 4.9 5.0 6.4 5.9
150– 200 3.9 4.0 4.0 4.0 3.4 3.6 7.0 7.4
200– 250 1.3 1.3 2.1 2.3 2.8 2.8 3.4 3.4
250– 500 1.6 1.6 1.1 1.1 2.2 2.2 4.5 4.5
500– 1000 0.1 0.1 0.4 0.4 2.4 2.4
1000–2000 0.2 0.2

of transmission of severe acute respiratory syndrome were recovered from droplet stain marks. Food dye
(SARS) by talking might be affected by the spoken was used in both studies. On average, we only found
language. The aspiration pronunciation system in 40 droplets in one natural cough without food dye,
different languages is different, and aspiration could which is far less than those found by Duguid (1946)
generate droplets. More studies and more samples are and Loudon & Roberts (1967). As mentioned in
needed to support this hypothesis. Loudon & Roberts (1967), many factors have effects on
As to coughing, an average of 5000 droplets by a the numbers of droplets, such as the amount of secretion
cough with mouth initially closed was reported in present in the mouth and its location, and the placement
Duguid (1945, 1946), and 464 by one ‘natural’ cough and movement of lips, tongue and teeth during the
recorded in Loudon & Roberts (1967), of which 237 cough. Based on the results of talking, the usage of

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

Droplets due to talking and coughing X. Xie et al. S713

Table 6. Total mass of droplets calculated using measured droplet number and size data during talking or coughing (S, the total mass of droplets detected at sampling position; O, the
food dye would be a reason to explain the above differ-

(0.58)
ence. Moreover, in Duguid’s (1946) study, the health

0.55
648
F1
status of test subjects was not described. The subjects
were healthy both in the study of Loudon & Roberts

(0.14)
(1967) and in the current study. As we know, it may

0.13
be difficult for healthy people to produce violent

271
M3
coughs. The violence of coughs would be different for
patients with respiratory diseases and thus has an

(1.66)
coughing (20 times)
effect on droplet generation, as well as more secretions

1.62
952
M2
of fluids on airway surfaces and higher frequency of
coughing (Papineni & Rosenthal 1997). All these

no food dye

(20.56)
factors are difficult to control and to quantify.

(5.74)
In the measurements of total droplet mass gener-

20.44
1331

5.68
800
M1
ated during talking and coughing, the average value
obtained with a surgical mask was much less than

(1.43)
that when a plastic bag with a tissue was used. This
could be explained by the condensation of water

1322

1.36
F1
vapour in the exhaled breath when a plastic bag was
used. In the measurements of droplet size distribution,

food dye with sugar

(3.71)
we obtained the size data of each droplet captured at

2425

3.56
the sampling position. We also back-calculated the

M3
corresponding droplet size at the origin. Having

(3.80)

(2.98)
these two droplet diameters, and by assuming the den-
sity of these droplets as the same as pure water

2213

1986
3.67

2.86
M1
droplets, we could estimate the mass of each droplet
both at sampling position and at the origin. Adding

(0.56)
the mass of all the droplets captured together, we

0.52
have the total droplet mass data for each experiment

809
F1
which are summarized in table 6. ‘S’ means the total
mass of droplets detected at sampling position and food dye

(3.43)

(2.00)
‘O’ means the total mass of droplets at the origin.

3738

2273
3.23

1.88
Compared with the estimated droplet mass data in M1
table 6, the measured results using surgical face

(0.12)
mask were much larger, even if we doubled the density
of the droplets.
0.11
135
F4

It is difficult to capture all the droplets produced

during expiratory activities using the methods in the
(0.15)

current study and previous studies. If we have the size

0.15
100

distribution, the total number of droplets produced

F3

may be estimated from the total mass measurement.

(1.02)

In the literature, we only found one study that

measured the total mass. Zhu et al. (2006) reported
0.97
918
F2

that an average of 6.7 mg of saliva per cough was col-

lected on the mask. In the current study, an average
(0.50)

of 1.1 mg of fluid was obtained per cough using the sur-

0.49

gical face mask method and 4.2 mg of fluid using plastic

225
F1
talking (counting from 1 to 100)

bag with tissue, less than 6.7 mg reported by Zhu et al.

(0.30)

(2006). And averages of 18.7 and 79.4 mg of fluid were

measured using mask and plastic bag, respectively,
0.29
100
M3

during counting from 1 to 100. Actually, both methods

have their limitations. The mask and plastic bag could
total mass of droplets at the origin).

(1.53)

not be fitted to the face perfectly. There would be gaps

2749

1.41

and droplets may have escaped from these gaps during

M2
no food dye

coughing. Droplet evaporation also occurs in the pro-

cess. Saliva on the lips may touch the mask or plastic
(1.00)

(0.63)

bag. Condensation of water vapour in the exhaled

1091

0.94

0.59
760
M1

breath occurs in the plastic bag. Droplets may be

re-inhaled in the plastic bag. All these factors would
weight (mg)

influence the results.

weight (mg)

However, as seen in tables 4 and 6, we found that the

number

number
S (O)

S (O)

total mass measured from experiments was much larger

average

average
droplet

than that calculated from measurements of droplet

numbers and sizes. This could be explained as both

J. R. Soc. Interface (2009)

 Downloaded from http://rsif.royalsocietypublishing.org/ on March 29, 2015

S714 Droplets due to talking and coughing X. Xie et al.

experiments have limitations. In the experiments of Garner, J. S. 1996 Guideline for isolation precautions in hos-
measuring droplet numbers and sizes, only part of the pitals. The Hospital Infection Control Practices Advisory
droplets expelled was captured, which was indicated Committee. Infect. Control Hosp. Epidemiol. 17, 53–80.
from the size range covered, and in the experiments of (doi:10.1086/647190)
Inouye, S. 2003 SARS transmission: language and droplet pro-
measuring droplet total mass, maybe more respiratory
duction. Lancet 362, 170. (doi:10.1016/S0140-
droplets were involved. 6736(03)13874-3)
Jennison, M. W. 1942 Atomizing of mouth and nose secretions
into the air as revealed by high speed photograph. Aero-
5. CONCLUSIONS
biology 17, 106– 128.
This study demonstrates the feasibility of measuring the Li, Y. G., Huang, X., Yu, I. T. S., Wong, T. W. & Qian, H.
respiratory droplets produced during talking and cough- 2005 Role of air distribution in SARS transmission
ing without a dye, by which expiratory activities are during the largest nosocomial outbreak in Hong Kong.
natural compared with the subject’s behaviour when a Indoor Air 15, 83 –95. (doi:10.1111/j.1600-0668.2004.
00317.x)
food dye is used. The glass slide method shows consider-
Loudon, R. G. & Roberts, R. M. 1967 Droplet expulsion from
able promise, although scanning and analysing the the respiratory tract. Am. Rev. Resp. Dis. 95, 435–442.
droplet stain marks are very time consuming. The Morawska, L. 2006 Droplet fate in indoor environments, or
study also supports the belief that talking and coughing can we prevent the spread of infection? Indoor Air 16,
play important roles in the generation of respiratory dro- 335–347. (doi:10.1111/j.1600-0668.2006.00432.x)
plets and provides more information about respiratory Nicas, M., Nazaroff, W. W. & Hubbard, A. 2005 Toward under-
droplets produced by healthy subjects. More droplets standing the risk of secondary airborne infection: emission
were generated when food dye was used. There was no of respirable pathogens. J. Occup. Environ. Hyg. 2,
great difference about the size distributions of droplets 143–154. (doi:10.1080/15459620590918466)
produced when food dye was used or not used, nor Papineni, R. S. & Rosenthal, F. S. 1997 The size distribution
of droplets in the exhaled breath of healthy human sub-
between talking and coughing. More small droplets
jects. J. Aerosol Med. 10, 105–116. (doi:10.1089/jam.
were produced in the more violent activity of coughing. 1997.10.105)
Ethical approval for the experimental study was obtained Qian, H., Li, Y., Nielsen, P. V., Hyldgaard, C. E., Wong,
from the Institutional Review Board of the University of T. W. & Chwang, A. T. Y. 2006 Dispersion of exhaled dro-
Hong Kong/Hospital Authority Hong Kong West Cluster. plet nuclei in a two-bed hospital ward with three different
ventilation systems. Indoor Air 16, 111–128. (doi:10.1111/
The work was supported by a grant from the Research Grants
j.1600-0668.2005.00407.x)
Council of the Hong Kong Special Administrative Region,
Wan, M. P. & Chao, C. Y. H. 2007 Transport characteristics
China ( project no. HKU 7150/06). We thank post graduate
of expiratory droplets and droplet nuclei in indoor environ-
student volunteers at the Department of Mechanical
ments with different ventilation airflow patterns.
Engineering for participating in the tests.
J. Biomech. Eng. 129, 341–353. (doi:10.1115/1.2720911)
Wells, W. F. 1934 On air-borne infection. Study II. Droplets
REFERENCES and droplet nuclei. Am. J. Hyg. 20, 611– 618.
Xie, X. J., Li, Y. G., Chwang, A. T. Y., Ho, P. L. & Seto, W.
Duguid, J. F. 1945 The numbers and the sites of origin of the H. 2007 How far droplets can move in indoor environ-
droplets expelled during expiratory activities. Edinburgh ments—revisiting the Wells evaporation–falling curve.
Med. J. 52, 385 –401. Indoor Air 17, 211–225. (doi:10.1111/j.1600-0668.2007.
Duguid, J. F. 1946 The size and the duration of air-carriage of 00469.x)
respiratory droplets and droplet-nuclei. J. Hyg. 4, 471–480. Yang, S., Lee, G. W. M., Chen, C. M., Wu, C. C. & Yu, K. P.
Fairchild, C. I. & Stamper, J. F. 1987 Particle concentration 2007 The size and concentration of droplets generated by
in exhaled breath. Am. Ind. Hyg. Assoc. J. 48, 948 –949. coughing in human subjects. J. Aerosol Med. 20, 484–
Fennelly, K. P., Martyny, J. W., Fulton, K. E., Orme, I. M., 494. (doi:10.1089/jam.2007.0610)
Cave, D. M. & Heifets, L. B. 2004 Cough-generated aero- Zhu, S. W., Kato, S. & Yang, J. H. 2006 Study on transport
sols of Mycobacterium tuberculosis: a new method to characteristics of saliva droplets produced by coughing in
study infectiousness. Am. J. Resp. Crit. Care Med. 169, a calm indoor environment. Build. Environ. 41, 1691–
604 –609. (doi:10.1164/rccm.200308-1101OC) 1702. (doi:10.1016/j.buildenv.2005.06.024)

J. R. Soc. Interface (2009)