Accuracy of gas analysis in lung function laboratories.

Ambient particulate matter has been associated consistently with an increased risk for mortality largely due to cardiovascular diseases.1 Although the relative risk estimates from epidemiological studies are small, they apply to almost the entire population of the United States. Consequently, exposure to ambient particles produces considerable burden of disease, and its mitigation offers the benefit of improving life expectancy.2 Articles see pp 941 and 949 Over the past decade, research has substantiated the understanding of the pathophysiological mechanisms linking ambient particles to the cardiovascular system3,4 once it was noted that ambient air pollution elicits systemic inflammatory responses in the general population.5 An update of the American Heart Association statement on air pollution and cardiovascular disease3 is under way. Mechanisms considered for active and secondhand smoke as well as ambient air pollution are strikingly similar.4,6,7 They include progression of atherosclerotic plaques to vulnerable forms, prothrombotic states, endothelial dysfunction, and altered autonomic nervous system control (Figure). Increased systemic oxidative stress is considered the key mechanism responsible for most of these pathophysiological changes. Increased risks for cardiovascular disease in general and coronary artery disease in particular have been documented for active and secondhand smoke as well as ambient particulate matter. Deep venous thrombosis has been added to this list recently.8 Figure. Overview on pathomechanism linking ambient air pollution,4 secondhand smoke,7 and active smoking to acute coronary syndromes. Nevertheless, the public health relevance of particulate matter in the light of the smoking literature remains hotly debated. Smokers are exposed to considerably higher cumulative doses of particulate matter than the general nonsmoking population. Mortality due to low doses of ambient particles may be considered counterintuitive compared with doses of particles tolerated by smoking individuals. A systematic assessment of the exposure-response function ranging from low doses of inhaled particles …

Mycobacterium tuberculosis Senses Host-Derived Carbon Monoxide during Macrophage Infection

A gaseous emissions analysis of commercial aircraft engines during test-cell run

An audit of pulmonary function laboratories in the West Midlands

In this study we sought to determine whether an absorbent in which little carbon monoxide (CO) forms has a correspondingly small capacity to absorb carbon dioxide (CO(2)). Completely dried samples (600 g) of Baralyme (A), Drägersorb 800 (B), Drägersorb 800 Plus (C), Intersorb (D), Spherasorb (E), LoFloSorb (F), Superia (G), and Amsorb (H) were exposed to a flow of 0.5% (A-H; n = 4-5) and 4% isoflurane (F-H; n = 3) in pure oxygen at 5 L/min for 60 min. Downstream CO concentration, temperature, and isoflurane concentration were recorded every 60 s to calculate CO formation and isoflurane loss. The CO(2) absorption capacity of each brand was determined by passing 5.1% CO(2) in oxygen (flow, 250 mL/min) through untreated samples (30 g; n = 5) until the outlet CO(2) concentration reached 0.5%. CO formation was largest in absorbents containing potassium hydroxide (A and B) and negligible in absorbents not containing any alkali hydroxide (F-H). The outlet temperature correlated with CO formation, but the isoflurane loss did not. The duration of CO(2) absorption also did not correlate with CO formation. We conclude that absorbents that allow only very little CO formation are not necessarily poor CO(2) absorbents. In an in vitro study, carbon dioxide (CO(2)) absorption capacity and possible carbon monoxide (CO) formation were tested in different absorbent brands. Absorbents with very small CO formation are not necessarily poor CO(2) absorbents.

It has been reported that exhaled carbon monoxide (CO) concentrations and arterial carboxyhemoglobin (CO-Hb) concentration in blood may be increased in critically ill patients. However, there was no study that examined correlation among amount of CO in exhaled air, CO-Hb concentrations in erythrocytes, and bilirubin IXalpha (BR) in serum, i.e., the three major indexes of heme catabolism, within the same subject. Here, we examined CO concentrations in exhaled air, CO-Hb concentrations in arterial blood, and BR levels in serum in 29 critically ill patients. Measurements of exhaled CO, arterial CO-Hb, and serum total BR have been done in the intensive care unit. As control, exhaled CO concentration was also measured in eight healthy volunteers. A median exhaled CO concentration was significantly higher in critically ill patients compared with control. There was significant correlation between CO and CO-Hb and CO and total BR level. We also found CO concentrations correlated with indirect BR but not direct BR. Multivariate linear regression analysis for amount of exhaled CO concentrations also showed significant correlation with CO-Hb and total BR, despite the fact that respiratory variables of study subjects were markedly heterogeneous. We found no correlation among exhaled CO, patients' severity, and degree of inflammation, but we found a strong trend of a higher exhaled CO concentration in survivors than in nonsurvivors. These findings suggest there is an increased heme breakdown in critically ill patients and that exhaled CO concentration, arterial CO-Hb, and serum total BR concentrations may be useful markers in critically ill conditions.

Introduction The carbon monoxide (CO) in hydrogen fuel is known to degrade the fuel cell performance [1-4]. The allowable concentration of CO is 0.2 ppm in the quality standards for hydrogen fuel of fuel cell vehicles [5]. The allowable concentration of impurities would be revised in accordance with the development of the fuel cell components, such as reduction of platinum loadings and thinning of the membrane. In order to discuss the allowable concentration of CO, the understanding of the performance degradation mechanism by CO and the evaluation in conditions close to the actual environment are required. In our previous work, we showed that the low anode platinum loadings would decrease the fuel cell performance significantly at the CO concentration as low as 1 ppm in the hydrogen [4]. But the CO adsorption behavior of near the allowable concentration remain incompletely understood. In this study, the adsorption behavior of low concentration CO on the polymer electrolyte fuel cell was investigated by gas analysis at the anode outlet. Experimental Commercial Pt/C catalyst (TEC10E50E, Tanaka Kikinzoku Kogyo) and electrolyte membrane (Nafion NR211, DuPont) were used for the single cell tests. The platinum loading on the anode / cathode was set at 0.1/ 0.3 mg cm-2, respectively. The MEAs were assembled into a JARI standard cell (25 cm2 of electrode area). The cell temperature was 60ºC, and the dew point temperatures of the anode and the cathode were 47ºC and 40ºC , respectively. The anode gas was hydrogen mixed with CO (0.2 - 1.0 ppm), and the cathode gas was purified air. The stoichiometry of fuel and air was 1.4 and 2.5, respectively. The CO and carbon dioxide (CO2) at the anode outlet concentrations were analyzed by a gas chromatograph with a pulse discharged helium ion detector. Results and Discussion The cell voltage at 1000 mA cm-2 and the CO and CO2 exhaust rate at the anode outlet (R CO, out and R CO2, out, respectively) during the 0.4 ppm of CO exposure test were shown in Fig. 1. The cell voltage started to drop after 5 hours and became stable after 25 hours. The exhaust rate of CO and CO2 rose when the cell voltage dropped, and became stable after 25 hours. The sum of CO and CO2 exhaust rate after 30 hours was nearly equal to the CO supply rate (R CO, in). The effect of CO concentration was investigated at the ranges from 0.2 to 1.0 ppm. The amount of CO adsorption was calculated from the results of the CO and CO2 measurement. Firstly, the CO adsorption rate (q CO) can be determined from the molar balance of carbon in the gas phase at the inlet and the outlet, that is q CO = R CO, in - (R CO, out + R CO2, out). Then, the amount of CO adsorption was obtained by time integration of q CO. The amount of CO adsorption during the CO exposure tests were shown in Fig. 2. At the beginning, the amount of CO adsorption increased proportional to time, and the slope correspond approximately to the CO supply rate. This indicates that most of the CO species were adsorbed on the anode. Then, the amount of CO adsorption were come to steady regardless of the cumulative CO supply amount. The saturated adsorption of CO was increased with the increase of CO concentration at cell inlet. The increase of saturated adsorption of CO will cause the voltage degradation of the cell. From these results, the adsorption behavior of CO near the allowable concentration was analyzed. The relationship between the voltage drop and the amount of CO adsorption will be discussed at the meeting. Acknowledgements This work was partly supported by the New Energy and Industrial Technology Development Organization (NEDO).

The importance of telomere biology in human disease is increasingly recognized and, in parallel, use of telomere length (TL) measures is proliferating in epidemiological and clinical studies. Such studies measure leukocyte TL (LTL) using several methodological approaches. Shorter LTL is associated with atherosclerosis1 and all-cause mortality.2 Given the increasingly recognized role of TL in human ageing and its related diseases, it is essential to know more about the reliability and validity of TL measurement methods, their comparability and which method is optimal for a specific epidemiological/clinical setting. In an effort to address this knowledge gap, Martin-Ruiz et al. (MR)3 studied the reliability of TL measurement techniques. They compared the popular qPCR method with the labour-intensive Southern blots (SBs) and single telomere length analysis (STELA). MR concluded that ‘neither technique nor laboratory had strong influence on result variation’, and that ‘Southern blotting and qPCR are similar in their reproducibility’. Unfortunately, for the following reasons we believe that for epidemiological studies neither conclusion is justified by the data. Reliability of LTL Most DNA samples (10/12) used by MR were obtained from human placenta, cell cultures and cancer cells. However, the inter-assay reliability of LTL is the pertinent parameter for epidemiological studies. MR included only two DNA samples from leukocytes and, because these were added in the second round of the study, they could not be used to measure inter-assay reliability of LTL. TL results for human placenta, cultured and cancer cells cannot be automatically generalized to LTL reliability, which is the primary concern of epidemiologists. Note also that MR used pooled leukocyte samples of multiple donors, and effects of pooling on assay reliability can therefore not be excluded. A previous comparison of LTL reliability has been done for the SB and the qPCR methods in a study4 cited by MR. The study reported a clear difference in inter-assay coefficient of variation (CV) between SB = 1.74% and qPCR = 6.54%, using 50 leukocyte DNA samples from individual donors. Moreover, Steenstrup et al.5 investigated whether LTL elongation in longitudinal studies can be attributed to measurement error vs a real biological phenomenon. They found little evidence for LTL elongation over and above the effects expected from measurement error. At the same time, the available data indicated a substantially larger proportion of individuals with an apparent LTL elongation in qPCR-based studies when compared with SB-based studies. In our view, the most parsimonious explanation for this finding is the higher measurement error of the qPCR method. MR observed that rank correlations between measurements obtained in different laboratories and with different methods were high, reflecting similar rank orders of the observations. Due to the inclusion of different cell types, the range of TLs in this study (4.7-9.2 kb) is much higher, however, than the age group-specific range (about 3 kb by direct SBs within age groups) used in most epidemiological studies of LTL. This will have inflated the rank correlation beyond what is relevant for LTL in epidemiological studies considerably, contributing to the erroneous conclusion that the SB and qPCR methods yielded similar results. Sample size and composition MR used 12 samples. These were measured by two laboratories using SBs, one laboratory using STELA and seven laboratories using qPCR. As both the number of samples and the number of laboratories using techniques other than qPCR were low, the statistical tests used by MR to infer no difference in reliability between methods are underpowered and consequently of limited value. We are thus left puzzled by the authors’ claim of > 95% power to detect the difference previously reported between inter-assay CVs for LTL using SBs and qPCR in 50 leukocyte DNA samples.4 MR provide no details of their calculation in support of this statement, nor on the exact difference between inter-assay CVs for which they calculated their statistical power. Furthermore, the authors combined the two SB and one STELA laboratories for comparisons of inter-laboratory CV across methods. We see little scientific justification for this choice, which in effect leaves one with no information specific to either the SB or STELA technique. For the two leukocyte samples, the inter-laboratory CVs were 6.2% and 6.5% for the SB/STELA laboratories vs 22.2% and 22.2% for the qPCR laboratories (samples K and L, Table 2, in erratum MR)6. These results, albeit from a tiny sample size, are consistent with higher measurement error of the qPCR over SB/STELA based-methods. This is not specific for the leukocyte samples; overall the inter-laboratory CVs were substantially higher when using qPCR (P = 0.001 according to MR). Finally, for the crucial analyses of the inter-assay and intra-assay CVs, the total number of DNA samples was restricted to 5 and 3, respectively, and none of these were from leukocytes. CV as a measure of reliability A characteristic of the CV is its dependence on the mean, and hence the implicit assumption when using the CV is heteroscedasticity, i.e. that the variance is proportional to the mean. We examined whether this assumption holds in the results presented by MR. Figure 1 suggests that it holds for SB. There is a negligible correlation between mean and CV, which is not surprising given the logarithmic nature of molecular size ladders on gels.7 By contrast, Figure 1 suggests that it does not hold for qPCR. There is a strong negative correlation between average TL and CV, which implies that the error made in qPCR-based TL measurements is not proportional to the mean, but instead is closer to a constant (assay-specific) value. Such a finding undermines the CV as a reliability measure for qPCR-based TL studies. Instead we recommend using the intra-class correlation coefficient, which yields an informative estimate, provided that the ‘test’ samples are similarly distributed as the samples in the investigated population. Figure 1. Coefficient of variation (CV%) between laboratories for SB/STELA vs qPCR plotted against telomere length. Telomere length was standardized per laboratory, dividing the results for all samples by the value obtained for sample G. The X-axis displays the ... Figure 1 also illustrates the larger range of values obtained with qPCR when compared with SB. MR suggest that the larger ‘dynamic range’ obtained with qPCR compensates for the lower precision of the method. However, when CV values are calculated for SB laboratories alone (i.e. ignoring the STELA results), the inter-laboratory CV is in fact over 40% higher for the qPCR laboratories (paired t-test, t = 2.39, df = 18, P < 0.025). Therefore, the larger range in TL values obtained using qPCR compared with SBs was more likely to be caused by a lower precision of qPCR, rather than compensating for it. DNA quality MR reported that they assessed DNA quality (purity and integrity) by ‘UV spectroscopy and agarose gel electrophoresis’, which is not typical of epidemiological studies that use the qPCR-based method. This may be critical if qPCR-based results are influenced by DNA integrity, which cannot be ruled out, as intact amplifiable target sequences are essential for reliable and valid results.8 Therefore, it is important to demonstrate in impartial studies that DNA integrity does not affect the T/S ratio results. Conclusions We see little evidence in MR that the reliabilities of SB and qPCR in measuring TL are equivalent. The number of laboratories performing SBs and STELA in their study was very small, as was the number of samples examined. Furthermore, only two of the 12 samples were from human leukocytes, the standard cell type used in epidemiological studies, and the inter-assay reliability of LTL was not measured. The qPCR does have the advantage over SB and other methods in that it costs less and requires fewer resources, but at the expense of measurement reliability. This implies that to demonstrate the same effect statistically, a larger sample size is needed when using qPCR in comparison with using SB/STELA. It is informative therefore to examine the consequences of lower reliability (higher CVs) for the actual sample sizes required. The following example might serve to contextualize the impact of inter-assay CVs on required sample sizes. On average, women’s LTL is longer by 0.15 kb than men’s LTL. As shown in Figure 2, to detect this difference with 90% power, with an increase in inter-assay CV from 2 to 20%, the required sample size increases by approximately six-fold. Figure 2. Effect of inter-assay coefficient of variation (CV%) on sample size required for a statistical power of 0.9. Shown on the left axis are the multiples of the sample size needed compared with CV = 0% (i.e. perfect reliability). The required ... The paper by MR and this commentary highlight an issue that is of great importance to the future of telomere epidemiology. As proposed in the pages of this journal 5 years ago,9 large-scale epidemiological studies, based on measurements of LTL using both SB and qPCR in laboratories experienced in these techniques, are urgently needed to resolve matters related to ‘noise’ and to assess how the two methods compare in capturing the associations of LTL with a host of human traits. Without such comparison, we fear that the claim by MR that SB and qPCR are equally reliable methods to measure LTL may result in suboptimal choices of methods, thereby wasting precious resources. Conflict of interest: None declared.

The oxidation of carbon with the simultaneous oxidation of silicon, manganese, and iron of liquid alloys by carbon dioxide in nitrogen and the absorption of oxygen by the alloys from the gas were studied using 1-g liquid iron droplets levitated in a stream of the gas at 1575 °C to 1715 °C. Oxidation of carbon was favored over oxidation of silicon and manganese when cast iron (3.35 pct C, 2.0 pct Si, 0.36 pct Mn, and 0.05 pct S) reacted with CO2/N2 gas at 1635 °C. An increase in the flow rate of CO2/N2 gas increased the decarburization rate of cast iron. The rate of carbon oxidation by this gas mixture was found to be independent of temperature and alloying element concentrations (in the range of silicon = 0 to 2.0 pct manganese = 0 to 0.36 pct and sulfur = 0 to 0.5 pct) within the temperature range of the present study. Based on the results of a kinetic analysis, diffusion of CO2 in the boundary layer of the gas phase was found to be the rate-limiting step for the reactions during the earlier period of the reaction when the contents of carbon, silicon, and manganese are higher. However, the limiting step changed to diffusion of the elements in the metal phase during the middle period of the reaction and then to the diffusion of CO in the gas phase during the later period of the reaction when the content of the elements in the metal were relatively low. For the simultaneous oxidation reactions of several elements in the metal, however, the diffusion of CO2 in the gas phase is the primary limiting step of the reaction rate for the oxidation of carbon during the later period of reaction.

Air pollution is a growing concern in rural areas where agricultural production can be reduced by it. This article analyses data obtained as part of a research project. The aim of this study is to understand the influence of atmospheric pressure, air temperature, air relative humidity, longitude and latitude of the location, and indoor and outdoor environment on local rural workplace diversity of air pollutants such as carbon monoxide (CO) and suspended particulate matter (SPM), as well as the contribution of these variables to changes in such air pollutants. The focus is on four topics: motivation, innovation and creativity, leadership, and social responsibility. Furthermore, this study developed an artificial neural network (ANN) model to predict CO and SPM concentrations in the air based on data collected from the mentioned inputs. The related sensors were assembled on an Arduino Mega 2560 board to form a field-portable device to detect air pollutants and meteorological parameters. The sensors included an MQ7 sensor for CO concentration measurement, a Sharp GP2Y1010AU0F dust sensor for SPM concentration measurement, a DHT11 sensor for air temperature and air relative humidity measurement, and a BMP180 sensor for air pressure measurements. The longitude and latitude of the location were measured using a smartphone. Measurements were conducted from 20 December 2021 to 16 July 2022. Results showed that the overall average outdoor CO and SPM concentrations were 10.97 ppm and 231.14 μg/m3 air, respectively. The overall average indoor concentrations were 12.21 ppm and 233.91 μg/m3 air for CO and SPM, respectively. Results showed that the ANN model demonstrated acceptable performance in predicting CO and SPM in both the training and testing phases, exhibiting a coefficient of determination (R2) of 0.575, a root mean square error (RMSE) of 1.490 ppm, and a mean absolute error (MAE) of 0.994 ppm for CO concentrations when applying the testing dataset. For SPM concentrations, the R2, RMSE, and MAE using the test dataset were 0.497, 30.301 μg/m3 air, and 23.889 μg/m3 air, respectively. The most influential input variable was air pressure, with contribution rates of 22.88% and 22.82% in predicting CO and SPM concentrations, respectively. The acceptable performance of the developed ANN model provides potential advances in air quality management and agricultural planning, enabling a more accurate and informed decision-making process regarding air pollution. The results of short-term estimation of CO and SPM concentrations suggest that the accuracy of the ANN model needs to be improved through more comprehensive data collection or advanced machine learning algorithms to improve the prediction results of these two air pollutants. Moreover, as even lower cost devices can predict CO and SPM concentrations, this study could lead to the development some kind of virtual sensor, as other air pollutants can be estimated from measurements of particulate matters.

Generally, larger cities are characterized by traffic congestion, which is associated with higher concentrations of pollution, including Carbon Monoxide (CO) pollution. However, this convention requires empirical support on the basis of accurate and reliable measurements. In addition, the assessment of the effect of CO on the autonomic nervous system (ANS), as measured by heart rate variability (HRV), has yielded conflicting results. A majority of the (few) studies on the topic have shown that increases in CO concentration of up to about 10 parts per million (ppm) are associated with a decrease in stress and risk to health in subjects. Beyond the hypothesis postulating city size as a determinant of increased CO concentration, the hypothesis proposing a causal link between CO concentration and HRV balance also requires empirical support. This article compares CO concentrations in a large metropolis with those in a small town, analyzing the relationship between CO and the HRV responses of young women in terms of city size. Four different types of environments were compared, taking into account mediating variables. The study participants spent 35 min in selected environments (a city center, a residential environment, a park, and a home) wearing Polar devices to measure HRV, and portable devices to measure noise thermal load and CO. The average concentrations of CO in each environment were calculated, along with the time distribution of the CO concentration, and the regression slopes between the concentrations of CO and the ANS balance, as measured by the low frequency power/high frequency power ratio (LF/HF) expressed as an HRV index. The results show that, regardless of size, the cities measured were all characterized by low levels of CO, far below the maximal accepted threshold standards, and that urban residents were exposed to these concentrations for less than half of the daytime hours. Furthermore, in contrast to the common view, larger cities do not necessarily accumulate higher concentrations of CO compared to small cities, regardless of the level of transport congestion. This study confirms the findings of the majority of the other studies on the subject, which showed a decrease in stress (as measured by HRV) as a result of an increase in CO concentrations below 7 ppm. Finally, following the assessment of the differential contribution attributed to the different environmental factors, it appears that noise, thermal load, and congestion all contribute more to a higher level of HRV balance than CO. This finding highlights the importance of a multivariable approach to the study, and a remediation of the effect of environmental factors on stress in urban environments.

Carbon monoxide (CO) and carbon dioxide (CO2) emissions arising from vehicles and combustion processes in motor parks predisposes to adverse health outcomes and associated health risks. There is dearth of studies and data on air quality in motor parks in Nigeria, hence, the need to assess the levels of CO and CO2 and their relationship with meteorological parameters in three major motor parks in Ibadan, Nigeria. A cross-sectional comparative design was adopted for this study. Akinyele Motor Park (AMP) in Akinyele Local Government Area (LGA), Iwo Road Motor Park (IMP) in Ibadan North East LGA and New Garage Motor Park (NMP) in Ibadan South West LGA were purposively selected and levels of CO and CO2 were monitored for 2 months with appropriate gas meters, meteorological parameters were also monitored using Ventus W155 wireless weather station. Values obtained were compared with WHO and ASHRAE guideline limits. Descriptive and inferential statistics were used for data analysis at p=0.05. CO concentrations (ppm) for AMP, IMP and NMP ranged from 2.0-106.0, 2.0 – 83.0 and 2.0 – 90.0 respectively while the mean CO2 concentrations (ppm) were 395.4 ± 30.5, 356.3 ± 57.1 and 388.1 ± 42.1 respectively. There was a significant positive correlation between CO and CO2 (r=0.258, p=0.000) and also with four meteorological parameters; temperature (r=0.164), rainfall (r= 0.105), heat index (r= 0.134) and dew point (r= 0.127) (p<0.05). A positive correlation was also found between CO2 and four meteorological parameters; temperature (r= 0.276, p=0.000), rainfall (r=0.125), heat index (r= 0.232, p=0.003) and dew point (r=0.028). For the three motor parks, CO concentrations (ppm) were 80% higher than WHO guideline of 9 ppm for 8 hour monitoring. Mean CO2 concentrations (ppm) were within ASHRAE guideline limit of 400 ppm. The study showed that meteorology has influence on CO and CO2 concentrations and motor park users are exposed to high levels of CO. Routine monitoring of CO and CO2 is recommended in order to ensure these emissions do not exceed guideline limits.

Low correlation between household carbon monoxide and particulate matter concentrations from biomass-related pollution in three resource-poor settings

Air pollution has long been and is still a problem for humans and the environment, a priority in countries with high vehicle rates. The contribution of gas released reaches 60-70%. Motor vehicles produce one of the pollutants, namely Carbon Monoxide (CO) gas. This research was conducted to determine the level of CO concentration results from sampling using impinger and prediction models using Caline4 software. The results of the concentration between direct measurements with the Caline4 model were then compared. This research was conducted in three streets for three working days namely Jl.Urip Sumoharjo, Jl.Talasalapang, and Jl.Nusantara. The methodology used is direct measurement using impinger, calculating vehicle volume at each measurement point, and analyzing CO concentrations using Caline4. The results of sampling showed the highest CO concentration on Jl.Urip Sumoharjo at 5.80 ppm, Jl.Talasalapang at 1.06 ppm, and Jl.Nusantara 1.15 ppm. The highest estimation results of CO concentrations with Caline4 on Jl.Urip Sumoharjo were 5.7 ppm, Jl.Talasalapang by 1.1 ppm, and Jl. Nusantara was 1.4 ppm. High or low CO concentration value depends on vehicles volume, for instance, CO concentration increases with increasing vehicles volume. In addition, it is also depends on the meteorological factor, such as, the faster the wind increases, the faster the pollutants will increase. Then compare the results of CO impinger concentration and Caline4 using the t-Test to see the difference of the two CO concentrations. Comparisons were made using the t-Test to meet the t-stat < t-critical concluded that there was no significant difference between the two CO concentrations.

Forsøg på rekonstruktion af en fortidig jernudvindingsproces