I N D U S T R I A L A I R P O L L U T I O N BY M E T A L S M O N I T O R E D BY A SYNTHETIC
FABRIC:
A C A S E S T U D Y OF A C O M M U N I T Y
G. H. S M I T H , F. A. Y. G A I L E Y , and O. LL. L L O Y D
Environmental Epidemiology and Cancer Centre, Wolfson Institute of Occupational Health, Department of Community Medicine, University of Dundee, Scotland
(Received November 1, 1985) Abstract. To supplement an epidemiological investigation into respiratory cancer, a synthetic fabric called tak was used to study the deposition of atmospheric metal pollution within the town of Armadale, central Scotland. Pollution maps showed high concentrations of several metals in areas close to the town's steel foundry and in a second area in the north of the town. Through further statistical analyses, those metals were identified which had probably been emitted by the foundry, and temporal variations in metal deposition patterns were examined. The advantages of this method of low technology sampling, which include the low cost and a high density of sampling sites, are described.
1. Introduction
Armadale is a long-established industrial town, with a population of about 9000, in central Scotland. It is situated, 20 km from Edinburgh to the east and the same distance from Glasgow to the west, on a raised moorland which slopes down to the northeast (Lloyd et al., 1985a). Epidemiological studies of Armadale have revealed abnormalities including an association between mortality from respiratory cancer and atmospheric metal pollution (Lloyd, 1978; Lloyd et al., 1982; Lloyd et al., 1984; Lloyd et al., 1985a, b). To investigate this association further, more detailed and systematic surveys of air pollution were required. Because the cost of conventional equipment for sampling air pollution would have allowed measurements of pollution at only a few sites in the areas under study, it was decided to use less expensive methods of low technology sampling for collecting and measuring metal pollutants. In most surveys using low technology methods, biological monitors have been used (Goodman and Roberts, 1971). Although non-biological monitors (namely those which are completely man-made) have been used less frequently, they possess several advantages. One such monitor, tak, consists of coarse muslin-like cloth impregnated with a resinous material which remains sticky at ambient temperatures. Its loose mesh and adhesive properties facilitate the accumulation of metal particles (Figure 1). As a sampler of particulate pollution it has other advantages: there are no metabolic activities to alter mechanisms of metal uptake and retention; being a man-made material, tak is relatively homogeneous physically and chemically; the surface area of tak material can be easily calculated and, hence, so can metal deposition rates; its background concentrations of most metals are low; it retains metal particles by adherence over its entire exposed surface area, and its capacity for collecting metals is likely to be relatively high; its adhesive surface is relatively unaffected by conditions of high rainfall and wind speed. Environmental Monitoring and Assessment 8 (1987) 135-149. 9 1987 by D. Reidel Publishing Company.
136
G . H . SMITH ET AL.
Fig. 1.
A n electron micrograph of particles adhering to the tak's surface ( x 860).
Although tak has been used only infrequently in surveys of air pollution, the results of studies where it has been employed (Avon, Gloustershire and Somerset Environmental Monitoring Committee, 1977; Garbet, 1984; Jones et al., 1973; Short, 1984) justified its deployment as part of the research into the patterns of atmospheric metal pollution in Armadale. Finally, prior work in the Armadale area (Gailey and Lloyd, 1986) showed that this sampler was more reliable than other forms of low technology monitor including indigenous soils (Gailey and Lloyd, 1986). 2. Methods
The survey of air pollution in Armadale, in which tak was included, lasted for 17 months. This period, which started in May 1981, was divided into bimonthly batches (or exposures) apart from batch 4 which lasted for 3 months because bad weather delayed field work.
INDUSTRIAL AIR POLLUTION
BY M E T A L S BY A S Y N T H E T I C F A B R I C
137
6
A
m
5
4 3
0.5 km
Fig. 2. The location of the squares for the square map pattern; the concentrations of metals were calculated for the sites within the consecutiveperipheral areas of the squares. The tak samplers were made from rectangles (7 x 10 cm) of the synthetic fabric, tak rag (Gramos Chemicals Ltd). Each rectangle was wrapped around a plastic hair roller to allow the entire surface area of one side of the fabric to be exposed to the atmosphere, and was held in position by thread and elastic bands. The roller was suspended vertically from a plastic-covered metal loop by means of plastic string; (plastic materials were selected to minimize metal contamination of the tak). This device was fixed to a bamboo pole, at a height of 2 m above the ground, at 47 sites in Armadale; there was no roofing for the apparatus. The criteria for site selection have been described elsewhere (Yule and Lloyd, 1984, a, b). To prevent adherent particles from being dislodged from the surfaces of the tak when the samplers were collected after the period of exposure, each roller was put into a glass jar with a tapered neck so that the tak did not touch the sides of the bottle. In the laboratory, the tak roller was air-dried and stored sealed in the jar. Each sample was digested and analyzed by atomic absorbtion spectrophotometry for nine metals (Fe, Mn, Zn, Pb, Cu, Cr, Ni, Cd, and Co) generally as previously described for the other samplers (Yule and Lloyd, 1984a, c; Gailey and Lloyd, 1985). To illustrate the values at individual sites, the concentrations of the metals found in two or more batches were plotted on grid computer maps, constructed by means of the computer mapping package Camgrid (Yule and Lloyd, 1984a, c; Gailey and Lloyd, 1983a; Gailey and Lloyd, 1985). To show the probable flow of pollution, a contour map of the estimated concentrations of Fe, a recognized marker of pollution from the steel industry, was constructed by means of the statistical package GLIM, the values throughout the area being derived by means of a fourth order polynomial equation; the relative concentrations of a metal were denoted by letters of the alphabet, with letter A representing the lowest values. Finally, to allow the dispersion
138
G.H.
S M I T H ET A L .
of metal pollution from the foundry irrespective of wind direction to be analyzed, the map of Armadale was divided into six squares of different sizes, each centred on the foundry (Figure 2); the mean concentrations of the metals at the sites within the peripheral areas o f each square were calculated. Further details of the mapping techniques used have been published elsewhere (Yule and Lloyd, 1984a, b, c). The means and coefficients of variation for each metal in each batch were calculated. To discover if the mean values of the five metals (which had been detected frequently) differed significantly between batches and between sites and also between the six areas of the square map pattern, analysis of variance (ANOVA) was performed. The concentration gradients in the square map pattern were tested for linearity. Spearman's rank correlation coefficients were used to show those metals likely to have come from a common source or sources, to investigate the variability of that metal's concentrations (whether resulting from changes in output or in dispersal), and to indicate those metals mainly originating from the foundry. Lastly, associations were sought between the mean concentrations of each metal for the whole town in each batch and available metereological variables to indicate the probable influence of those variables on the metal concentrations.
3. Results
The grid maps of the mean concentrations at individual sites for the whole survey (Figures 3 to 7) showed that the higher ranges o f concentration of Fe, Mn, Pb and to a lesser extent Zn lay in a line across the foundry area in a W-SW/E-NE axis, with a protrusion to the north of the town. The northern area contained the only high concentration of Cu. The map of the mean estimated concentrations of Fe (Figure 8) similarly showed pollution to be maximal in the SE of the town, and flowing along the S W / N E axis and northwards. The pre-exposure values of the metals in Tak were approximately (in ~tg (100 cm2) - l): Fe, 100; Mn, 5; Zn, 75; Pb, 10; Cu, 3.8; Cr, 2.5. Concentrations of Fe, Mn and Zn above the background levels were detected in all batches, of Cu in batches 2, 3, and 7, of Pb in batches 7 and 8, and o f Cr in batch 8 only; Ni, Cd, and Co were found in no batch. Because the values for Cr were based on only one batch, its distribution was not studied. In all batches, Fe had the highest mean concentrations for the whole town, followed by Zn and Mn (Table I). For each metal, the ANOVA showed that the mean concentrations varied significantly between the eight batches. In each batch, the concentrations o f each metal varied widely between the 47 sites, as was shown by the coefficients of variation of all the mean values o f the batches. For the whole period o f the survey, the ANOVA also revealed that the mean values of Fe, Mn, Zn, and Cu at the individual sites differed significantly (/7< 0.005) while those of Pb (which came from only two batches) did not.
INDUSTRIAL
AIR POLLUTION
I ] l
I ] I I I I
:::::::::::::::: :::::::::::::::: ::::::::::::::.:
I ] J I
11 I I ] ]
:::::::':::::::: :::::::::::::::: :::::::::::::::: ::::::::::::.:::
II
................
~BB.~
NUMBER
BY METALS
...... 2~.i
OF
RESULTS
.9
3~I
IN
9
BY A S Y N T H E T I C
~19.7
~97.E
O
5
EI::ICH G R O U P
15
~
10
Fqq
[]
139
FABRIC
Ill llll
[]
:
~:
~:~:~-
iiiiiiii !!ii
iiii
iii!i!ii iiiii:ii
I
!ii!ii!i i!ili!!i
q
:.i:::.: : i : : j !!E:!!i
Fe
0
km !
0.4 11
13 Fig. 3.
The grid map of the mean concentrations of Fe collected in tak at 47 sites in Armadale. F = foundry. Units = ~tg (100 c m z)- ~ d a y - 1.
For the whole survey, high mean concentrations of metals (Table II) were found in the areas of squares close to the foundry (Figure 2). For Fe, the concentration gradient decreased discontinuously outwards. For Mn, the concentration gradient was more pronounced, the sole discontinuity resulting from the concentration in area 6 being slightly higher than in area 5. For Zn the gradient was reversed. There were no striking gradients for Pb or Cu. The A N O V A indicated that the concentrations in the squares differed significantly for Fe and Mn ( p < 0 . 0 0 5 ) and Cu (p < 0.05), but not for Zn and Pb. The concentration gradients for Fe, Mn, and Cu
140
G, H. SMITH ET AL. ::::::::::::::::::::::::::::::::
::::::: :::,-:::
I
: : : : : : : : : " " :
I
: : ......
~.9
NUMBER
i~i~:i!~i!iiiiiiii!!iiiiiii!i!!
:: :
iii~:. =======================
::"::
::::.i~i ." :.::~{~i~iii!i
t7.s
OF RESULTS 2e
IN
EACH
o
/~
-
Flq
[]
[]
[]
32,7
.... 2g:T
qO.
3
GROUP
7
q-
2
!i!i!!!i :i::i
i!ii
Mn
~1"
n Fig. 4.
0
km
0.4
~ - - - " 1
The grid map of the mean concentrations of Mn in tak at 47 sites in Armadale, F = foundry. Units =gtg (100 cm2) - l d a y - '
were highly significantly linear with distance from the foundry (p < 0.005); graphical analysis showed that these linear trends declined outwards from the steel foundry. The intercorrelations between the concentrations of each metal in each batch were positive, apart from the negative correlations found in batch 7 (Cu/Zn) and batch 8 (Pb/Zn, Fe/Zn, and Mn/Zn). The metal pairs which were significantly correlated varied between batches. In three of the four batches when only Fe, Mn, and Zn were found, these metals were significantly correlated with each other; in the exceptional
I N D U S T R I A L A I R P O L L U T I O N BY METALS BY A S Y N T H E T I C FABRIC
141
::::::::::::::::
i!!! N / / !
:::::::::':::
!!!i!:ii!:!ii::: $2 97
NUMBER
lOLl- 9 5
78.6
OF" R E S U L T S
IN
EACH
130.
t4
1 5 6 93
GROUP O
1
i:
iii
....
FF]
~
:::::::: . . . .
....
iiiiiiii !iiiiiil
:x::z: . . . .
!i::
....
::::::::
::;::::
:
ilil
:::
[]
!!!!
: i : j : :i"
Zn
0
km I~
0.4 |
n Fig. 5.
T h e grid m a p o f the m e a n concentrations o f Z n in tak at 47 sites i n A r m a d a l e . F = f o u n d r y . U n i t s = ~ t g (100 cm2) -~ day - l .
batch 4 (which was also exceptional in having lasted for three months), only M n / Z n was not significantly correlated. However, in the batches containing values of the other metals, there was less consistency in the metal pairs with significantly correlated values. The pairs of the metals with significantly correlated values in more than one batch were: M n / F e in eight batches; Fe/Zn in six batches; M n / Z n in four batches; and Cu/Zn, Fe/Pb and Fe/Cu in two batches each. Of the correlations between the mean concentrations of the metals for the whole survey calculated over 47 sites (Table III), all were positive except for Pb/Zn. Of
142
G. H. SMITH ET AL.
,,
IIII
9
I',
" " " " ',
"
"" ii
" ,, ' :::i!i:iiiii !iiiii:i'iii:i!i!i!'i:ill Ill II
O
~
II
7.6
3.8
NUMBER
OF
RESULTS
1B
11 . ~
IN 21
EACH ~
15.2
18.9
GROUP t4
2
1
i!ii ::::::::
::::::::
!i!i
ilii E~::: E~ ~
11:
71:1
""E~
....
::::
k
!iiiiiii
t
Pb
0
........
km
OA
fl Fig, 6.
The grid map of the mean concentrations of Pb in tak at 47 sites in Armadale. F = f o u n d r y . Units = lag (100 c m 2)- 1 d a y - i.
these positive correlations, six were significant. Only Fe was significantly correlated with all other metals. The correlations between the concentrations of each metal at individual sites calculated over eight batches showed distinct differences between metals. For Fe, 25 of the possible 28 correlations between batches were positive and significant. For Mn, all 28 correlations were positive and significant. For Zn, only two of the 28 correlations were significant, and 10 correlations were negative. For Cu, there were no significant correlations between the three batches available. The only correlation of Pb was significant and negative.
143
I N D U S T R I A L AIR P O L L U T I O N BY METALS BY A SY N T H E T I C FABRIC
:::::::::::::::: :::::::7::::::::
ii
igl
NUMBER
OF
iiii m I~.8
7.2
RESULTS
IN 1
EACH
IW.W
18.2
GROUP
~
I
FR []
+
IIII I I I I
li I I
Cu
i~ n
Fig. 7.
Or." 0.4 k nl
, ~
The grid map of the mean concentrations of Cu in tak at 47 sites in Armadale. F = foundry. Units = ~tg (100 cm z)- l day- 1.
While for Fe and Mn the values in most of the batches were significantly correlated, the values for Zn in batches 4-7 were not significantly correlated with those in any other batches, and in the remaining batches they were significantly correlated with only one other batch. The values for Fe and Mn were inversely correlated with distance from the foundry (,o=0.02); those of Cu were similarly correlated (p
FABRIC:
A C A S E S T U D Y OF A C O M M U N I T Y
G. H. S M I T H , F. A. Y. G A I L E Y , and O. LL. L L O Y D
Environmental Epidemiology and Cancer Centre, Wolfson Institute of Occupational Health, Department of Community Medicine, University of Dundee, Scotland
(Received November 1, 1985) Abstract. To supplement an epidemiological investigation into respiratory cancer, a synthetic fabric called tak was used to study the deposition of atmospheric metal pollution within the town of Armadale, central Scotland. Pollution maps showed high concentrations of several metals in areas close to the town's steel foundry and in a second area in the north of the town. Through further statistical analyses, those metals were identified which had probably been emitted by the foundry, and temporal variations in metal deposition patterns were examined. The advantages of this method of low technology sampling, which include the low cost and a high density of sampling sites, are described.
1. Introduction
Armadale is a long-established industrial town, with a population of about 9000, in central Scotland. It is situated, 20 km from Edinburgh to the east and the same distance from Glasgow to the west, on a raised moorland which slopes down to the northeast (Lloyd et al., 1985a). Epidemiological studies of Armadale have revealed abnormalities including an association between mortality from respiratory cancer and atmospheric metal pollution (Lloyd, 1978; Lloyd et al., 1982; Lloyd et al., 1984; Lloyd et al., 1985a, b). To investigate this association further, more detailed and systematic surveys of air pollution were required. Because the cost of conventional equipment for sampling air pollution would have allowed measurements of pollution at only a few sites in the areas under study, it was decided to use less expensive methods of low technology sampling for collecting and measuring metal pollutants. In most surveys using low technology methods, biological monitors have been used (Goodman and Roberts, 1971). Although non-biological monitors (namely those which are completely man-made) have been used less frequently, they possess several advantages. One such monitor, tak, consists of coarse muslin-like cloth impregnated with a resinous material which remains sticky at ambient temperatures. Its loose mesh and adhesive properties facilitate the accumulation of metal particles (Figure 1). As a sampler of particulate pollution it has other advantages: there are no metabolic activities to alter mechanisms of metal uptake and retention; being a man-made material, tak is relatively homogeneous physically and chemically; the surface area of tak material can be easily calculated and, hence, so can metal deposition rates; its background concentrations of most metals are low; it retains metal particles by adherence over its entire exposed surface area, and its capacity for collecting metals is likely to be relatively high; its adhesive surface is relatively unaffected by conditions of high rainfall and wind speed. Environmental Monitoring and Assessment 8 (1987) 135-149. 9 1987 by D. Reidel Publishing Company.
136
G . H . SMITH ET AL.
Fig. 1.
A n electron micrograph of particles adhering to the tak's surface ( x 860).
Although tak has been used only infrequently in surveys of air pollution, the results of studies where it has been employed (Avon, Gloustershire and Somerset Environmental Monitoring Committee, 1977; Garbet, 1984; Jones et al., 1973; Short, 1984) justified its deployment as part of the research into the patterns of atmospheric metal pollution in Armadale. Finally, prior work in the Armadale area (Gailey and Lloyd, 1986) showed that this sampler was more reliable than other forms of low technology monitor including indigenous soils (Gailey and Lloyd, 1986). 2. Methods
The survey of air pollution in Armadale, in which tak was included, lasted for 17 months. This period, which started in May 1981, was divided into bimonthly batches (or exposures) apart from batch 4 which lasted for 3 months because bad weather delayed field work.
INDUSTRIAL AIR POLLUTION
BY M E T A L S BY A S Y N T H E T I C F A B R I C
137
6
A
m
5
4 3
0.5 km
Fig. 2. The location of the squares for the square map pattern; the concentrations of metals were calculated for the sites within the consecutiveperipheral areas of the squares. The tak samplers were made from rectangles (7 x 10 cm) of the synthetic fabric, tak rag (Gramos Chemicals Ltd). Each rectangle was wrapped around a plastic hair roller to allow the entire surface area of one side of the fabric to be exposed to the atmosphere, and was held in position by thread and elastic bands. The roller was suspended vertically from a plastic-covered metal loop by means of plastic string; (plastic materials were selected to minimize metal contamination of the tak). This device was fixed to a bamboo pole, at a height of 2 m above the ground, at 47 sites in Armadale; there was no roofing for the apparatus. The criteria for site selection have been described elsewhere (Yule and Lloyd, 1984, a, b). To prevent adherent particles from being dislodged from the surfaces of the tak when the samplers were collected after the period of exposure, each roller was put into a glass jar with a tapered neck so that the tak did not touch the sides of the bottle. In the laboratory, the tak roller was air-dried and stored sealed in the jar. Each sample was digested and analyzed by atomic absorbtion spectrophotometry for nine metals (Fe, Mn, Zn, Pb, Cu, Cr, Ni, Cd, and Co) generally as previously described for the other samplers (Yule and Lloyd, 1984a, c; Gailey and Lloyd, 1985). To illustrate the values at individual sites, the concentrations of the metals found in two or more batches were plotted on grid computer maps, constructed by means of the computer mapping package Camgrid (Yule and Lloyd, 1984a, c; Gailey and Lloyd, 1983a; Gailey and Lloyd, 1985). To show the probable flow of pollution, a contour map of the estimated concentrations of Fe, a recognized marker of pollution from the steel industry, was constructed by means of the statistical package GLIM, the values throughout the area being derived by means of a fourth order polynomial equation; the relative concentrations of a metal were denoted by letters of the alphabet, with letter A representing the lowest values. Finally, to allow the dispersion
138
G.H.
S M I T H ET A L .
of metal pollution from the foundry irrespective of wind direction to be analyzed, the map of Armadale was divided into six squares of different sizes, each centred on the foundry (Figure 2); the mean concentrations of the metals at the sites within the peripheral areas o f each square were calculated. Further details of the mapping techniques used have been published elsewhere (Yule and Lloyd, 1984a, b, c). The means and coefficients of variation for each metal in each batch were calculated. To discover if the mean values of the five metals (which had been detected frequently) differed significantly between batches and between sites and also between the six areas of the square map pattern, analysis of variance (ANOVA) was performed. The concentration gradients in the square map pattern were tested for linearity. Spearman's rank correlation coefficients were used to show those metals likely to have come from a common source or sources, to investigate the variability of that metal's concentrations (whether resulting from changes in output or in dispersal), and to indicate those metals mainly originating from the foundry. Lastly, associations were sought between the mean concentrations of each metal for the whole town in each batch and available metereological variables to indicate the probable influence of those variables on the metal concentrations.
3. Results
The grid maps of the mean concentrations at individual sites for the whole survey (Figures 3 to 7) showed that the higher ranges o f concentration of Fe, Mn, Pb and to a lesser extent Zn lay in a line across the foundry area in a W-SW/E-NE axis, with a protrusion to the north of the town. The northern area contained the only high concentration of Cu. The map of the mean estimated concentrations of Fe (Figure 8) similarly showed pollution to be maximal in the SE of the town, and flowing along the S W / N E axis and northwards. The pre-exposure values of the metals in Tak were approximately (in ~tg (100 cm2) - l): Fe, 100; Mn, 5; Zn, 75; Pb, 10; Cu, 3.8; Cr, 2.5. Concentrations of Fe, Mn and Zn above the background levels were detected in all batches, of Cu in batches 2, 3, and 7, of Pb in batches 7 and 8, and o f Cr in batch 8 only; Ni, Cd, and Co were found in no batch. Because the values for Cr were based on only one batch, its distribution was not studied. In all batches, Fe had the highest mean concentrations for the whole town, followed by Zn and Mn (Table I). For each metal, the ANOVA showed that the mean concentrations varied significantly between the eight batches. In each batch, the concentrations o f each metal varied widely between the 47 sites, as was shown by the coefficients of variation of all the mean values o f the batches. For the whole period o f the survey, the ANOVA also revealed that the mean values of Fe, Mn, Zn, and Cu at the individual sites differed significantly (/7< 0.005) while those of Pb (which came from only two batches) did not.
INDUSTRIAL
AIR POLLUTION
I ] l
I ] I I I I
:::::::::::::::: :::::::::::::::: ::::::::::::::.:
I ] J I
11 I I ] ]
:::::::':::::::: :::::::::::::::: :::::::::::::::: ::::::::::::.:::
II
................
~BB.~
NUMBER
BY METALS
...... 2~.i
OF
RESULTS
.9
3~I
IN
9
BY A S Y N T H E T I C
~19.7
~97.E
O
5
EI::ICH G R O U P
15
~
10
Fqq
[]
139
FABRIC
Ill llll
[]
:
~:
~:~:~-
iiiiiiii !!ii
iiii
iii!i!ii iiiii:ii
I
!ii!ii!i i!ili!!i
q
:.i:::.: : i : : j !!E:!!i
Fe
0
km !
0.4 11
13 Fig. 3.
The grid map of the mean concentrations of Fe collected in tak at 47 sites in Armadale. F = foundry. Units = ~tg (100 c m z)- ~ d a y - 1.
For the whole survey, high mean concentrations of metals (Table II) were found in the areas of squares close to the foundry (Figure 2). For Fe, the concentration gradient decreased discontinuously outwards. For Mn, the concentration gradient was more pronounced, the sole discontinuity resulting from the concentration in area 6 being slightly higher than in area 5. For Zn the gradient was reversed. There were no striking gradients for Pb or Cu. The A N O V A indicated that the concentrations in the squares differed significantly for Fe and Mn ( p < 0 . 0 0 5 ) and Cu (p < 0.05), but not for Zn and Pb. The concentration gradients for Fe, Mn, and Cu
140
G, H. SMITH ET AL. ::::::::::::::::::::::::::::::::
::::::: :::,-:::
I
: : : : : : : : : " " :
I
: : ......
~.9
NUMBER
i~i~:i!~i!iiiiiiii!!iiiiiii!i!!
:: :
iii~:. =======================
::"::
::::.i~i ." :.::~{~i~iii!i
t7.s
OF RESULTS 2e
IN
EACH
o
/~
-
Flq
[]
[]
[]
32,7
.... 2g:T
qO.
3
GROUP
7
q-
2
!i!i!!!i :i::i
i!ii
Mn
~1"
n Fig. 4.
0
km
0.4
~ - - - " 1
The grid map of the mean concentrations of Mn in tak at 47 sites in Armadale, F = foundry. Units =gtg (100 cm2) - l d a y - '
were highly significantly linear with distance from the foundry (p < 0.005); graphical analysis showed that these linear trends declined outwards from the steel foundry. The intercorrelations between the concentrations of each metal in each batch were positive, apart from the negative correlations found in batch 7 (Cu/Zn) and batch 8 (Pb/Zn, Fe/Zn, and Mn/Zn). The metal pairs which were significantly correlated varied between batches. In three of the four batches when only Fe, Mn, and Zn were found, these metals were significantly correlated with each other; in the exceptional
I N D U S T R I A L A I R P O L L U T I O N BY METALS BY A S Y N T H E T I C FABRIC
141
::::::::::::::::
i!!! N / / !
:::::::::':::
!!!i!:ii!:!ii::: $2 97
NUMBER
lOLl- 9 5
78.6
OF" R E S U L T S
IN
EACH
130.
t4
1 5 6 93
GROUP O
1
i:
iii
....
FF]
~
:::::::: . . . .
....
iiiiiiii !iiiiiil
:x::z: . . . .
!i::
....
::::::::
::;::::
:
ilil
:::
[]
!!!!
: i : j : :i"
Zn
0
km I~
0.4 |
n Fig. 5.
T h e grid m a p o f the m e a n concentrations o f Z n in tak at 47 sites i n A r m a d a l e . F = f o u n d r y . U n i t s = ~ t g (100 cm2) -~ day - l .
batch 4 (which was also exceptional in having lasted for three months), only M n / Z n was not significantly correlated. However, in the batches containing values of the other metals, there was less consistency in the metal pairs with significantly correlated values. The pairs of the metals with significantly correlated values in more than one batch were: M n / F e in eight batches; Fe/Zn in six batches; M n / Z n in four batches; and Cu/Zn, Fe/Pb and Fe/Cu in two batches each. Of the correlations between the mean concentrations of the metals for the whole survey calculated over 47 sites (Table III), all were positive except for Pb/Zn. Of
142
G. H. SMITH ET AL.
,,
IIII
9
I',
" " " " ',
"
"" ii
" ,, ' :::i!i:iiiii !iiiii:i'iii:i!i!i!'i:ill Ill II
O
~
II
7.6
3.8
NUMBER
OF
RESULTS
1B
11 . ~
IN 21
EACH ~
15.2
18.9
GROUP t4
2
1
i!ii ::::::::
::::::::
!i!i
ilii E~::: E~ ~
11:
71:1
""E~
....
::::
k
!iiiiiii
t
Pb
0
........
km
OA
fl Fig, 6.
The grid map of the mean concentrations of Pb in tak at 47 sites in Armadale. F = f o u n d r y . Units = lag (100 c m 2)- 1 d a y - i.
these positive correlations, six were significant. Only Fe was significantly correlated with all other metals. The correlations between the concentrations of each metal at individual sites calculated over eight batches showed distinct differences between metals. For Fe, 25 of the possible 28 correlations between batches were positive and significant. For Mn, all 28 correlations were positive and significant. For Zn, only two of the 28 correlations were significant, and 10 correlations were negative. For Cu, there were no significant correlations between the three batches available. The only correlation of Pb was significant and negative.
143
I N D U S T R I A L AIR P O L L U T I O N BY METALS BY A SY N T H E T I C FABRIC
:::::::::::::::: :::::::7::::::::
ii
igl
NUMBER
OF
iiii m I~.8
7.2
RESULTS
IN 1
EACH
IW.W
18.2
GROUP
~
I
FR []
+
IIII I I I I
li I I
Cu
i~ n
Fig. 7.
Or." 0.4 k nl
, ~
The grid map of the mean concentrations of Cu in tak at 47 sites in Armadale. F = foundry. Units = ~tg (100 cm z)- l day- 1.
While for Fe and Mn the values in most of the batches were significantly correlated, the values for Zn in batches 4-7 were not significantly correlated with those in any other batches, and in the remaining batches they were significantly correlated with only one other batch. The values for Fe and Mn were inversely correlated with distance from the foundry (,o=0.02); those of Cu were similarly correlated (p
