Comparison of formaldehyde measurements by Hantzsch, CRDS and DOAS in the SAPHIR chamber

Three instruments using different techniques measuring gaseous formaldehyde (HCHO) concentrations were compared in experiments in the atmospheric simulation chamber SAPHIR at Forschungszentrum Jülich. One instrument detected HCHO by using the wet-chemical Hantzsch reaction for efficient gas-phase stripping, chemical conversion and fluorescence measurement (AL4021, Aero Laser GmbH). An internal permeation HCHO source allows for daily calibrations. It was characterized 5 by sulfuric acid titration (overall accuracy 8.5 %). Measurements have a time resolution of 90 s with a limit of detection (3σ) of 0.3 ppbv. In addition, a new commercial instrument making use of cavity ring-down spectroscopy (CRDS) determined concentrations of HCHO, water, and methane (G2307, Picarro Inc.). The limit of detection (3σ) is specified as 0.3 ppbv for an integration time of 300 s and the accuracy is limited by the drift of the zero signal (manufacturer specification 1.5 ppbv). A custom-built, high-resolution laser differential optical absorption spectroscopy (DOAS) instrument provided HCHO mea10 surements with a limit of detection (3σ) of 0.9 ppbv and an accuracy of 6 % using an optical multiple reflection cell. The measurements were conducted from June to December 2019 in experiments in which either ambient air was flowed through the chamber or the photochemical degradation of organic compounds in synthetic air was investigated. Measured HCHO concentrations were up to 8 ppbv. Various mixtures of organic compounds, water vapour, nitrogen oxides, and ozone concentrations were present in these experiments. Results demonstrate the need to correct the baseline in the measurements of the 15 Hantzsch instrument to compensate for drifting background signals. Corrections were equivalent to HCHO mixing ratios in the range of 0.5 to 1.5 ppbv. The baseline of the CRDS instrument showed a linear dependence on the water-vapour mixing ratio with different slopes of (−11.20± 1.60) ppbv %−1 and (−0.72± 0.08) ppbv %−1 above and below 0.2 % water vapour mixing ratio, respectively. In addition, the intercept of these linear relationships drifted with time within the specification of the instrument (1.5 ppbv), but appeared to be equal for all water mixing ratios. Regular zero measurements are required to account 20 for the changes in the instrument zero. After correcting for the baselines of measurements by the Hantzsch and the CRDS instruments, a linear regression analysis of measurements from all three instruments in experiments with ambient air results in a good agreement with slopes between 0.93 and 1.07 with negligible intercepts (linear correlation coefficients R > 0.96). The new, small-sized CRDS instrument measures HCHO with a good precision and is accurate, if the instrument zero is taken into account. Therefore, it can provide accurate and calibration-free measurements like the DOAS instrument with a slightly 25 reduced precision compared to the Hantzsch instrument. 1 https://doi.org/10.5194/amt-2021-10 Preprint. Discussion started: 18 February 2021 c © Author(s) 2021. CC BY 4.0 License.

instruments are also custom-built including the fibre laser needed for the excitation.
Wet chemistry methods are widely used to detect formaldehyde. Sampling with cartridges for derivatization with 2,4dinitrophenylhydrazine (DNPH) and subsequent offline analysis with high-performance liquid chromatography (HPLC) has a low limit of detection (40 pptv), but requires comparably high experimental effort (Winberry et al., 1999). Due to the long sampling time of typically 1 hour the time resolution is less than spectroscopic methods. Another online wet-chemical method 70 is based on the Hantzsch reaction, in which aqueous formaldehyde reacts with acetylacetone (Kelly and Fortune, 1994). The concentration of the product (3,5-diacetyl-1,4-dihydrolutidine) is then measured by fluorescence after excitation at 410 nm.
This type of instrument is nowadays commercially available by Aero Laser GmbH. A low limit of detection of 100 pptv is reached at a high time resolution of minutes. The disadvantage compared to spectroscopic methods are the need for regular maintenance work and calibrations with liquid and gaseous standards (see below) and the consumption of hazardous liquids. 75 Proton-transfer mass-spectrometry (PTR-MS) can also detect formaldehyde but due to the low proton affinity the sensitivity is not satisfying and the sensitivity is low and exhibits a strong dependence on the water vapour mixing ratio (Vlasenko et al., 2010;Warneke et al., 2011;Yuan et al., 2017). Therefore, formaldehyde concentrations have not become a standard measurement of PTR-MS instruments. A direct comparison measurement of instruments using different techniques, calibration and data evaluation procedures is a well approved way to evaluate the quality of the data and to identify possible instrumental 80 artefacts, inaccuracy of calibration procedures or instrumental interferences (Grossmann, 2003;Inomata et al., 2008;Warneke et al., 2011).
Several comparison exercises have been performed so far. Eleven Comparisons that were done before 2005 are summarized and discussed in Hak et al. (2005). They were performed during field and chamber experiments utilizing different techniques.
Most often an absorption spectrometer and Hantzsch instruments were involved. Hak et al. (2005) concluded that there is high 85 variability in the level of agreement between measurements of instruments without exhibiting a specific pattern. Most often calibration of instruments were assumed to be the likely reason for disagreement. The authors report the comparison of four different techniques (broadband DOAS, FTIR, Hantzsch, chromatography after cartridge sampling) during measurements at an urban site. Results showed agreement of measurements by DOAS, Hantzsch and FTIR instruments within 11 %, but also strong variations in the agreement between Hantzsch measurements and other instruments. Differences between Hantzsch instruments 90 were attributed to insufficiently working scrubbers that are used for zeroing of instruments and differences in the calibration results. For absorption instruments, Hak et al. (2005) pointed out that the use of different recommendations for (differential) absorption cross sections can lead to disagreement. In addition, the resolution of spectrometers needs to be correctly taken into account to determine the effective cross section.
Forschungszentrum Jülich measurements by DOAS, DNPH-HPLC, PTR-MS, and 2 Hantzsch instruments were compared for controlled atmospheric conditions in 3 experiments. Deviations between measurements of instruments were up to 50 %.
Several analytical problems were identified: (1) Measurements showed that the derivatization efficiency in the DNPH-HPLC method was significantly lowered in dry air. (2) Like in previous comparisons uncertainties in the zeroing led to a bias in the measurements of of the Hantzsch instrument. (3) Measurements by Hantzsch were lower in the presence of ozone (45 ppbv) 100 compared to DOAS measurements, which are likely less affected by ozone, if formaldehyde concentrations are high compared to ozone concentrations. However, deviations were variable among experiments, so that connection to ozone was uncertain.
Formaldehyde measurements by a LIF instruments were compared to Hantzsch measurements in the SAPHIR chamber in 2014 (Kaiser et al., 2014) and to a recently developed commercial TDLAS system by Aeris Technology in ambient air (Shutter et al., 2019). The agreement between measurements by the LIF and TDLAS instruments was better than 8 % for 105 formaldehyde mixing ratios higher than 1 ppbv. The comparison of LIF and Hantzsch instruments in SAPHIR allowed for a systematic investigation, if measurements were affected by ozone or water. No systematic deviations with the presence of water or ozone could be found, so that observations in (Wisthaler et al., 2008) with respect to a potential ozone interference in the measurements by Hantzsch could not be confirmed. Measurements between Hantzsch and LIF instruments agreed within the combined uncertainties of calibrations (13 %).

110
Reports of instrument comparisons concluded that the measurement of formaldehyde remains challenging specifically for atmospheric concentrations in the low ppbv range. Calibration and instrumental zeros were identified as the major source of systematic uncertainties in the data. In this work, formaldehyde measurements from three different instruments are compared in experiments in the SAPHIR chamber located at Forschungszentrum Jülich. One instrument made use of the Hantzsch technique that was also applied in most of the previous comparisons. A high-resolution DOAS system (308 nm) for the detection 115 of hydroxyl radical concentrations provided also formaldehyde concentrations. In addition, formaldehyde was detected by a recently available commercial CRDS instrument from Picarro Inc.. Experiments included the investigation of the photochemical oxidation of specific organic compounds as well as experiments, in which ambient air was flowed through the chamber. Therefore, these experiments gave the chance to investigate the performance of the instruments in controlled conditions that allowed the systematic variation of parameters and for realistic ambient conditions and concentrations over a long period of 120 time.
2 Experimental procedure 2.1 Atmospheric simulation chamber SAPHIR Measurements were performed from June to December 2019 in experiments in the outdoor atmospheric simulation chamber SAPHIR at Forschungszentrum Jülich, Germany. Detailed descriptions of the chamber and its properties can be found 125 elsewhere (Rohrer et al., 2005). The SAPHIR chamber consists of a cylindrical, double-wall Teflon (FEP) film with a length and diameter of 18 m and 5 m, respectively, and an effective volume of 270 m 3 . The film is mounted in a metal frame with a shutter regulating the penetration of natural sunlight. Two fans ensure mixing of trace gases within 2 minutes. Transmittance of solar radiation through the Teflon film of the chamber is regularly determined by actinometric experiments (Bohn and Zilken, 2005). The volume between the 130 two Teflon films is permanently flushed with pure nitrogen (Linde, purity > 99.9999 %) and the chamber is held under slightly increased atmospheric pressure (∼ 30 Pa) to prevent any contamination from ambient air. Air that is consumed by instruments or small leakages is permanently replenished by dry synthetic air. This leads to a dilution of trace gases by 3 to 5 % per hour.
The Teflon film releases small amounts of nitrous acid (HONO) and small hydrocarbons such as formaldehyde and acetaldehyde when it is exposed to solar radiation. The source strength depends on temperature, illumination and humidity (Rohrer 135 et al., 2005). For formaldehyde, the source strength is approximately (20 ± 10) pptv min −1 (30 % RH at 298 K) for clear sky summer conditions (j(NO 2 ) ≈ 5 × 10 −3 s −1 ). This value was determined in experiments in this work, when the chamber only contained humidified clean synthetic air that was exposed to sunlight. Similar values were obtained in the past.
Some of the experiments carried out in 2019 and included in this work were part of the Jülich Atmospheric Chemistry Project (JULIAC), which was designed to investigate the seasonal and diurnal variation of atmospheric trace gases, radicals, 140 and particles in air influenced by anthropogenic and biogenic emissions. Ambient air was continuously sampled for four weeks in each season of the year through an inlet line made of Silconert coated stainless steel that was mounted at a 50 m high tower.
Large particles (> 10 µm) were removed in a cyclone before the air was transported into the chamber. A blower compressed the air before entering the chamber (pressure difference: 15 hPa). Only a fraction (250 to 260 m 3 h −1 ) of the total flow (660 m 3 h −1 ) controlled by a three-way valve flowed into the chamber. This flow rate results in a residence time of air in the 145 chamber of approximately 1 h. Temperature, pressure, relative humidity, and solar radiation were constantly monitored inside and outside of the SAPHIR chamber. During strong wind or heavy rain the louvre system of SAPHIR was closed to prevent damage of the Teflon film.
In addition to measurements in ambient air (JULIAC), data from experiments are included in the analysis, in which the photooxidation of anthropogenic and biogenic volatile organic compounds (e.g. isoprenoids, terpenes and derivatives of acetone) 150 were investigated under controlled conditions. Before the start of a typical experiment, the chamber was flushed with a mixture of ultra-pure nitrogen and oxygen (Linde, purity > 99.99990 %) to remove any remaining trace gases or contaminations. In most of the experiments, the chamber air was humidified by evaporating Milli-Q water that was flushed into the chamber with a high flow of synthetic air. Organic or inorganic compounds (e.g. alkenes, nitrogen oxides, ozone) or particles were injected to compose various conditions, for which the oxidation and degradation of organic compounds were investigated either in the 155 dark or under solar radiation. Emissions from up to 6 trees housed in a plant-chamber (Hohaus et al., 2016) were occasionally transferred into the chamber, in order to study their photo-oxidation. Reference experiments under similar conditions and with no injections of the analytes provide background data in order to determine the strength of chamber sources e.g. for HCHO.

HCHO detection by the wet chemical method using the Hantzsch reaction
One of the instruments for the detection of HCHO in this work is a commercial instrument (AL4021, Aero-Laser GmbH) that For the experiments in this work, the instrument was placed in an air-conditioned sea container next to the chamber. Parts of the instrument are temperature-stabilized: the stripping coil at 283 K, the reaction volume at 341 K and the fluorimeter at 175 308 K. This ensures an approximate constant reaction efficiency and fluorescence detection sensitivity (Rurack and Spieles, 2011;Resch-Genger and Rurack, 2013). The tubing of the peristaltic pumps was exchanged every two weeks due to a degradation of tubes and in order to prevent occlusions or strong loss of pumping performance, which cause changes in the instrument sensitivity of up to 10 % within this time span (Section 3.1). To prevent chemical degradation and ageing, all chemicals were stored in a box at a controlled temperature of 277 K. Chemicals needed to be refilled every one to two weeks. Stopping the 180 liquid flow or switching the instrument off for regular tubing exchange, cleaning or maintenance required an additional warmup time (30 to 120 min) for the instrument to be in an thermal equilibrium and to stabilize the flow rates of liquids in order to achieve a stable response signal. The instrument shows a significant zero signal (S 0 ) that needs to be subtracted from the total measured signal (S), in order to derive the signal caused by HCHO. Automatized zero measurements are performed for about 20 minutes every two to three 185 hours by removing HCHO from the sampled air by means of a scrubber consisting of a cylindrical glass, which is filled with a brown rod-shaped material, which -according to the manufactures -provides reliable and sufficient HCHO scavenging.
The internal software of the instrument takes the last zero measurement to evaluate the current measurement.
The HCHO mixing ratio in the sampled gas (flow rate F gas ) is derived from the instrument sensitivity E Cal . The signals are normalized to the inverse of the flow rate of the liquid solution (F liq ). Additional parameters like the molar volume V mol under 190 standard ambient temperature and pressure (SATP) conditions (298.15 K; 1 bar), the molar mass of formaldehyde M HCHO and the stripping efficiency α are necessary to calculate the HCHO mixing ratio from the measured signal: The instrument sensitivity (E Cal ) is determined in daily calibration measurements using an internal, temperature-controlled (T = 318 K) HCHO permeation source, which provides a constant mass flow of HCHO ( * m perm ). The sensitivity can be 195 calculated from the liquid flow rate F liq and the measured signal S Cal during calibration using Equation 1: Because the instrument can also measure liquid formaldehyde concentrations, the permeation source strength ( * strength calibration. The reproducibility of this procedure was 2 %. The uncertainty of the titration method is mainly due to the uncertainty of 3 % in the liquid flow measurements. Therefore, the combined accuracy of the Hantzsch method in this work is 8.5 % (Table 1).

Cavity ring-down spectroscopy (CRDS)
The second instrument for the detection of HCHO in this work uses Cavity Ring-Down Spectroscopy (CRDS) for the simulta- inside the cavity that are monitored (Barry et al., 2002;Hoffnagle et al., 2017).

225
Data are internally averaged to 1 Hz by the instrument and are averaged to 60 s time resolution for further analysis to improve the precision of data. The instrumental precision and accuracy are reduced, if the humidity rapidly changes, because of the overlapping absorptions lines of water and HCHO. Peak shapes also change with changes of temperature and pressure and can impact the result of the peak fitting procedure (Picarro Inc. personal communication). Therefore, the cavity is pressureand temperature-stabilized (Picarro Inc.).

Differential optical absorption spectroscopy (DOAS)
A high-resolution laser differential optical absorption spectroscopy (DOAS) instrument provided absolute HCHO measurements (Dorn et al., 1995;Schlosser et al., 2007Schlosser et al., , 2009. Light from a sub-picosecond pulsed, frequency-doubled dye laser provides UV radiation around 308.04 nm with a bandwidth of 0.41 nm. The dye laser is synchronously pumped by a frequency doubled model-locked Nd:YAG laser. The light passes through the central axis of the chamber in a White cell type 235 multi-reflection cell whose mirrors are installed at each end of the chamber in a distance of 20 m resulting in a total absorption path length of (2240 ± 2) m. The light intensity transmitted through the chamber is spectrally analysed by a high resolution Echelle grating spectrograph (∆λ = 2.7 pm, λ/∆λ = 114 000, f = 1.5 m) and detected by a linear photodiode array detector.
The DOAS method relies on the separation of narrow absorption lines of specific absorbers from light attenuation that 240 does not vary much with wavelength. In the evaluation, the wavelength-dependent intensity of the transmitted light is fitted to a polynomial to account for broadband extinctions and a superposition of differential absorption spectra from specific absorbers. In order to calculate absorber concentrations, Lambert-Beer's law can be applied, but instead of using the total absorption cross section, the differential absorption cross section (σ dif f ) needs to be used. Similar to the analysis of the transmitted spectrum, also the differential absorption cross section only contains the narrow absorption lines. The differential 245 absorption cross section for formaldehyde is 9.39 × 10 −21 cm 2 at the maximum of the absorption in the wavelength region used in the DOAS instrument here (308.1034 nm). The value of the absorption cross section was derived from the cross- The slope of a regression analysis between both measurements were used to scale the differential absorption cross section that 250 was applied in the evaluation procedure of the DOAS signal. It is worth noting that measurements used in the cross calibration are not part of comparison in this work. In addition, the Hantzsch instrument that delivered HCHO concentrations between 2011 and 2018 was different from the one used in this work. Therefore, DOAS measurements can be regarded as independent from Hantzsch measurements in the comparison here.
High selectivity is achieved by the high resolution of the measured spectrum that allows separation of overlapping narrow- Formaldehyde measurements by DOAS have a time resolution of 135 s with a stated 1 σ precision of 0.5 ppbv (Brauers et al., 2007). The detection limit is mainly limited by the residual structures in the spectra. The combined accuracy of 6 % is basically derived from the uncertainty of the absorption cross section, which is a function of ambient pressure and temperature (Cantrell et al., 1990) and additionally depends on the instrumental response function of the detection system.

265
3 Results and discussion

Stability of the instrument zero and sensitivity of the Hantzsch instrument
The instrumental zero of the Hantzsch instrument exhibits fluctuations and drifts. This is likely caused by temperature changes of parts of the instrument that are not temperature-controlled. Therefore, the Hantzsch instrument determines the instrument zero from regular, automatized measurements, in which formaldehyde is removed from the sampled air. HCHO concentrations 270 automatically provided by the instrument are calculated according to Equation 1. By default the last zero measurement for S 0 is used until the next zero measurement is done. As a consequence, the time series of HCHO concentrations can show artificial jumps after a new zero reading has been taken, if the zero value has changed (Fig. 1).
In order to smooth the fluctuations caused by changes in the zero signal, data in this work were reprocessed by applying a linear interpolation of the zero measurements before and after the actual HCHO measurement. In addition, the time interval 275 between two zero measurements was reduced from 4 hours to 2 hours after the significant changes in the instrument zero had been noticed. Figure 1 shows that changes of the instrument zero can be as high as 0.2 mV, which is equivalent to HCHO mixing ratios of up to 1.2 ppbv. The exact value depends on the current sensitivity of the instrument. By applying more frequent zero measurements and interpolating between zero measurements, the uncertainty in the HCHO measurements could be significantly reduced by a at least a factor of 10, so that the accuracy of measurements due to the uncertainty in the zero is 280 well below 100 pptv.
The Hantzsch instrument was in operation at the SAPHIR chamber for nearly half a year. During experiment episodes, calibration measurements were done once a day. This allows analysing the stability of the instrument's sensitivity. Figure 2 shows the deviation from the mean sensitivity for 116 calibration measurements. The mean value of the sensitivity of the instrument was 75 L mV µg −1 . The 1 σ reproducibility of the calibration measurements was 5 % and has an accuracy of 285 8.5 % ( Table 1). The record of calibration measurements indicates a good long-term stability of the instrument's sensitivity.
Nevertheless, day-to-day changes are most likely due to real changes of the sensitivity, because the sensitivity is expected to decrease with ageing of the tubing. This is also clearly seen in the increase of the sensitivity after each exchange of the tubing (Fig. 2). Therefore, the calibration value that was measured close to the time of the actual measurement was applied for the evaluation of measurements in this work.

Water vapour dependent offset in HCHO measurements by Picarro CRDS
Formaldehyde measurements by the CRDS instrument appeared to have a bias. An offset was noticed in measurements in synthetic air in the clean chamber, when no formaldehyde was present (Fig. 3). This offset is variable with time and depends    on the presence of water vapour. As seen in Fig. 3, a significant offset is also measured by the instrument in clean dry air.
The offset in dry air shows relatively small changes of typically less than 0.1 ppbv from day to day. In addition to the offset 295 in dry air, changes in humidity causes additional variations of the instrument offset of up to 2 ppbv. The exact value scales linearly with the water vapour mixing ratio. The offset in dry air can be derived by linearly extrapolating measurements without formaldehyde to dry conditions. This could be done for experiments here, when the clean air in the chamber was humidified.  Figure 5 shows the water vapour dependence of the bias in the CRDS measurements that adds to the offset in dry air. In 305 order to make data of different days comparable, all measurements were corrected for the variability of the day-specific offset in dry air. Again, data from the experiments between 26 June to 03 August 2019 in the SAPHIR chamber were taken, when no HCHO was present. Measurements at water vapour mixing ratios lower than 0.2 % separate from the measurements at water vapour mixing ratios higher than 0.2 %. Apparently, the intercepts for dry conditions are the same for both groups of data on a particular day. Both subsets of data exhibit a bias that decreases linearly with increasing water vapour mixing ratio. However, 310 slopes of the linear relationship for lower water vapour mixing ratios are approximately a factor of 15 higher compared to that for higher water vapour mixing ratios with a value of (-11.20 ± 1.60) ppbv % −1 for water vapour mixing ratios lower than 0.2 % and a value of (-0.72 ± 0.08) ppbv % −1 for water vapour mixing ratios higher than 0.2 %. The water vapour dependence is similar for all data. Therefore slopes were determined from linear fits using all data. These values combined with the day-specific intercept were used to correct all CRDS HCHO data in this work. The observed changes in the offset is caused by the data processing algorithm in the instrument, which takes into account the spectral overlap of water and formaldehyde infra-red absorption lines (Picarro Inc. personal communication). A very strong water line interferes with the formaldehyde absorption. Above 0.2 % water mixing ratio, the water vapour absorption line is strong enough, so that its contribution can be accounted for in a fit of absorption lines in which amplitudes and line widths are free parameters. In contrast, below 0.2 % the signal-to-noise is too poor for an independent fit of the line width. Therefore, 320 a fixed value for the line width which was derived from a spectrum that was acquired at very low water concentration is used. This procedure has been improved in new versions of the instrumental software (version 1.6.015 implemented in the instrument used here) (Picarro Inc. personal communication). Therefore, the correction described here might not be applicable for all Picarrao HCHO instruments.

Precision of measurements 325
In order to analyze the precision of measurements, the Allan deviation was calculated from measurements in air that did not contain formaldehyde (Fig. 6). The DOAS instrument did not provide a sufficiently high number of data points to perform an Allan deviation analysis. Corrections of data from the Hantzsch and CRDS instruments were applied as described above.
The Allan deviations for the CRDS and Hantzsch measurements result in 1 σ precisions of 0.08 and 0.014 ppbv, respectively, at an averaging time of 2 min. The Hantzsch instrument provides an overall better signal-to-noise ratio and therefore that are not fully taken into account as seen in Fig. 4, but also short-term variability in the instrument sensitivity. In contrast, CRDS data follow White noise up to an averaging time of 2 hours. For example, the Allan deviation is approximately 50 pptv for an integration time of 5 minutes consistent with the typical precision that is specified for the instrument (Picarro Inc.).

Comparison of measurements
From June 2019 to December 2019 numerous experiments were performed in the SAPHIR chamber. In photochemical experi-340 ments studying the photo-degradation of organic compounds, the chamber was operated with synthetic air, in which trace gases were injected. During the JULIAC campaign, the chamber was filled with ambient air. The diversity of experiments allowed for measuring HCHO for a wide range of conditions regarding temperature, relative humidity, ozone, nitrogen oxides, and methane concentrations (Table 2).
Regular flushing of the chamber with pure synthetic air provided a solid instrumental zero for most instruments and specific 345 oxidation experiments provided high levels of nitric oxides and ozone up to peak values of 60 ppbv and 600 ppbv, respectively.
For instance, formaldehyde was monitored during the photo-oxidation of cyclic monoterpenes such as limonene, carene, αpinene, β-pinene, isoprene and alkenes. In these batch experiments levels of nitrogen oxides and ozone were variable, in order to influence chemical oxidation and degradation process of aliphatic hydrocarbons and therefore the formation of formaldehyde.
The comparison of all formaldehyde measurements by the three instruments described above are shown as time series in Fig.   350 7 and as correlation plot in Fig. 8. Hantzsch measurements were interrupted for regular instrumental maintenance or calibration.
Each exchange of peristaltic tubes or a power off for maintenance required a longer time until the measurements became again   The statistical errors of each of the fit parameters of the linear regressions are lower than 0.01 due to the small errors of single 370 data points. All deviations of the slopes from unity are within the combined accuracies of instruments (Table 1). No systematic deviations in the measurements are identified and no corrections other than those described above need to be applied.