Table 1. Precision as Determined by Analyses of Natural Water Samples ammonla-N
la 2a 3a 4a 5a 6a a
nltrate AAll
AAll
IKL
0.12 f 0.003 0.29 f 0.007 0.32 f 0.008 0.76 f 0.019 1.33 f 0.033 1.29 f 0.032
0.15 f 0.006 0.26 f 0.006 0.32 f 0.008 0.78 f 0.016 1.30 f 0.027 1.30 f 0.029
0.20 f 0.005 0.40 f 0,010 1.09 f 0.087 1.27 f 0.102 2.50 f 0.200 2.52 f 0.203
+ nltrlte-N
nitrlte-N
IKL
AAll
0.19 f 0.009 0.40 f 0.015 0.99 f 0.032 1.22 f 0.039 2.34 f 0.040 2.52 f 0,111
IKL
0.11 f 0.015 0.13 f 0.018 0.33 f 0.045 0.65 f 0.090 0.72 f 0.094 1.00 f 0.131
AAll
0.08 f 0,009 0.16 f 0,001 0.32 f 0.011 0.56 f 0.011 0.73 f 0.031 1.00 f 0.047
phosphate-P IKL
0.07 f 0.014 0.07 f 0.013 0.12 f 0.024 0.39 f 0.027 0.43 f 0.031 0.70 f 0.038
0.05 f 0.010 0.09 f 0.013 0.16 f 0.013 0.34 f 0.029 0.50 f 0.011 0.80 f 0.017
All results in mg/L & SD.
Table II. Analysis of Reference Standards a ammonla-N AAll
nltrate
+ nltrlte-N
phosphate-P
IRL
AAll
IKL
concn f 1 SD actual concn
0.232 f 0.010 0.23
Nutrient Reference Standard I 0.224 f 0.005 0.104 f 0.005 0.1 11 f 0.008 0.23 0.11 0.1 1
concn f 1 SD actual concn
1.64 f 0.026 1.59
1.63 f 0.021 1.59
a
AAll
f 0.006
0.066 f 0.006 0.052
0.193 f 0.006 0.19
0.186 f 0.005 0.19
0.058 0.052
Nutrient Reference Standard I1 0.398 f 0.010 0.370 f 0.01 1 0.38 0.38
IKL
All values in mg/L.
bias amounting to less than 1%is indicated by this statistic; however, the weighting of results at the detection limits of both instruments may be responsible for this observation. When the point (2.34, 2.35), which is not shown in Figure 3, is included, then the value for the correlation coefficient changes to 0.9998 with no significant change in the slope or y-intercept values. The results for the dissolved phosphate (PO*-P) analyses are shown in Figure 4. Again the slope was not different from 1.0; the y intercept was not different from 0.0; and the correlation coefficient was 0.995. The mean values f one standard deviation (mg/L) obtained for each analyte by both instruments during a &day period are listed in Table I. As evidenced by the data, the precision of the discrete analyzer is usually better than the precision of the segmented-flow analyzer. The main reason for this is probably the diminution of “carry-over” from sample to sample noted in the discrete analyzer. The precision of the segmented-flow system can be improved by increasing the wash-to-sample ratio; but in doing so, the rate of sample analysis is decreased. The results obtained for 11determinations during 2 weeks using the nutrient reference standards I and I1 provided by the U S . Environmental Protection Agency are shown in Table 11. These standards did not contain nitrite. The results that were obtained by both instruments show good agreement with the actual values.
Summary More than 30 natural water samples were analyzed, colorimetrically, for NHs-N, NO3 N02-N, N02-N, and Pod-P, by the Technicon AAII and the Coulter IKL analyzers. The former uses segmented-flow analysis, while the latter uses discrete analysis. The discrete analyzer provided a 50% increase in rate of analysis compared with the segmented-flow analyzer for the four determinations studied. Comparable accuracy and precision of the two analyzers were indicated by the results of the standard reference analyses, by the results of the six natural samples used for the precision study, and also by the regression statistics pertaining to the 39 unknown water samples.
+
Literature Cited (1) Fed. Regist. 1976,41,52780-5. (2) Skougstad, M., Fishman, M., Friedman, L., Erdmann, D., and Duncan, S., Eds. “Methods For Determination of Inorganic Sub-
stances in Water and Fluvial Sediments”; Techniques of Water Resources Investigations of the United States Geological Survey, 1979;Book 5, Chapter Al. (3) “Methods Chemical Analysis of Water and Washes”; Environmental Monitoring and Support Laboratory: Cincinnati, OH 1978; NTIS No. PB 297 686. (4) Hale, D. R. Am. Lab. (Fairfield, Conn.) 1979,11,117-30. (5) “Coulter Industrial Kem-0-Lab Reference Manual”; Coulter Electronics, Inc.: Hialeah, FL 33010.
Received for review August 26,1980. Accepted February 25,1981.
Automobile Traffic and Lung Cancer. An Update on Blumer’s Report Lincoln Polissar” and Homer Warner, Jr. Fred Hutchinson Cancer Research Center, 1124 Columbia Street, Seattle, Washington 98104
In widely circulated reports (I,2), Blumer et al., studied the cancer mortality of a small Swiss town and calculated that residents living on a busy highway during 1958-70 had a cancer death rate 9 times greater than the rate for residents of a newer section of town with much less automobile traffic ( 3 ) .They tentatively linked the higher rate to automobile
exhaust, including PAH, which was found in a high concentration in the soil near the highway ( 1 ) . Because of the obviously significant public health implications of such a large relative risk, we believed that testing of Blumer’s results was essential. To do this, we examined traffic volume at the residences of
0013-936X/81/0915-07 13$01.25/0 @ 1981 American Chemical Society
Volume 15, Number 6, June 1981 713
A study carried out in Switzerland showed a high cancer risk for persons heavily exposed to automobile exhaust. An attempt to corroborate these findings was carried out. Exposure to automobile exhaust was imputed from the average daily traffic flow (ADT) a t the residence of 664 lung cancer cases
and 666 nonmelanotic skin cancer controls living in Seattle. There was a nonsignificant positive risk of lung cancer for female cases living on busy streets but no excess risk for males. The analysis included factors to control for the effects of age, sex, socioeconomic status, and other variables.
lung cancer cases and nonmelanotic skin cancer controls. Lung cancer was chosen as the malignancy with the greatest likelihood of correlation with exposure to automobile exhaust. We expected the milder nonmelanotic skin cancer to show little, if any, correlation. We hypothesized that the lung cancer cases would have been exposed to a higher volume of traffic than the skin cancer controls. The cases were identified from the records of the Cancer Surveillance System, a population-based tumor registry which covers 2.4 million people in the Seattle area; over 98% of the new cancer cases in the reporting area are registered. We chose a systematic random sampling of 373 males and 291 females from among the 1328 lung cancer cases diagnosed in Seattle during 1974-77. For the controls, we drew a systematic random sample of 373 males and 293 females from an incidence survey of 2229 nonmelanotic skin cancer cases diagnosed in the greater Seattle area during 1977-78. Exposure to automobile exhaust was estimated by the average daily traffic (ADT) flow (vehicles per 24 h) a t the residence of each case and control at the time of diagnosis. We used traffic flow figures from a 1977 map prepared by the City of Seattle Engineering Department. Although the diagnosis dates spanned a 5-yr period, we found that the traffic flow maps differed little from year to year. Using the Mantel and Haenszel method for pooling odds ratios ( 4 ) ,we stratified by age, by a socioeconomic (SES) indicator for the census tract, and by the geographic district of Seattle. The district variable controlled for a geographic gradient in the ratio of cases to controls. The mean size of the nine districts studied was 56 000 people. We did not stratify by race, since over 95% of the cases and the controls were white. In sharp contrast to Blumer ( 3 ) ,we found no significant associations between traffic volume and cancer risk (Table I). Our sample size provided a 95% probability that a pooled odds ratio of 1.6 or larger (indicating a 1.6-fold increase in lung cancer risk for persons living on streets with a t least 5000 ADT) would have been significant a t the 5% level for both sexes combined. The pooled odds ratios for several definitions of high vs. low exposure are all fairly close to 1.0 (with one exception), and all have significance levels greater than 5%. The strongest result, a nonsignificant odds ratio of 1.97, implies that women living on streets with a t least 15 000 vehicles passing per day are a t almost twice the risk of devel-
oping lung cancer as women who live on streets with