Tracers for assessing exposure to environmental tobacco smoke: what are they tracing?

The effectiveness of various tracers for measurements of exposure to environmental tobacco smoke (ETS) as a complex chemical mixture is based on the physicochemical properties of four major organic components and their dynamic behavior in indoor environments. For the particulate matter (PM) component and the very volatile organic compounds, emission and ventilation rates are generally the most important processes controlling indoor concentrations and exposures of nonsmokers. For the volatile organic compounds (VOCs) and semivolatile organic compounds (SVOCs), sorption on and desorption from indoor surfaces are additional processes that influence exposures. Laboratory and modeling studies of the dynamic behavior of nicotine, an SVOC, and PM indicate that nicotine can be used to estimate PM exposures from ETS in indoor environments when certain criteria are met: (italic>a(/italic>) smoking occurs regularly in the environment, (italic>b(/italic>) the system is near quasi-steady state, and (italic>c(/italic>) sampling time is longer than the characteristic times for removal processes. Measurements in residential and workplace buildings also support the use of nicotine as a tracer for PM in ETS. Recent laboratory and field data indicate that the VOCs from ETS can be traced using compounds with similar physicochemical properties, such as 3-ethenylpyridine, pyrrole, or pyridine. The effectiveness of nicotine for estimating exposures to the VOCs and SVOCs has not been determined, although these constitute major mass fractions of ETS.

Tobacco smoke is a complex mixture of thousands of compounds, of which approximately 400 have been measured in both mainstream smoke (MS) and sidestream smoke (SS) (1,2). Environmental tobacco smoke (ETS) is composed primarily of SS, with lesser contributions from the exhaled MS. This complex mixture of particles, gas-, and vapor-phase components is rapidly diluted and dispersed after emission and undergoes changes in its physicochemical properties because of shifts in vapor-particle distributions, sorption and desorption of vapor-phase components on indoor surfaces, and chemical reactions. Only limited research has been conducted on these physiochemical processes, although they clearly influence the overall chemical composition of ETS and ETS exposures.
Epidemiologic studies implicate ETS exposures as a risk factor for a variety of adverse health effects in nonsmoking adults and in children (1,3,4). The nature of exposures to ETS as a complex mixture and the identification of the putative agents within this complex mixture have been addressed to only a limited extent. The epidemiologic studies generally rely on questionnaires to characterize exposures This article is based on a presentation at the Workshop on Environmental Tobacco Smoke Exposure Assessment held [12][13] September 1997 in Baltimore, Maryland. Manuscript received at EHP 2 July 1998; accepted 18  Telephone: (510) 486-7491. Fax: (510) 486-7202. E-mail: jmdaisey@lbl.gov Abbreviations used: 3-EP, 3-ethenylpyridine; ASHRAE, American Society of Heating, Refrigerating and Air Conditioning Engineers; ETS, environmental tobacco smoke; FPM, fluorescence of a methanol extract of ETS particulate matter; MMAD, mass median aerodynamic diameter; MS, mainstream smoke; NNK, 4-(methyl-nitrosaminoFl-(3-pyridyl)-1-butanone; NNN, Nnitrosonornicotine; RSP, respirable particulate matter; PAH, polycyclic aromatic hydrocarbon; PDCB, p-dichlorobenzene; PM, particulate matter; RSP, respirable particulate matter; SS, sidestream smoke; SVOCs, semivolatile organic compounds; UVPM, measure of the ultraviolet absorbance of a methanol extract of ETS particulate matter; VOCs, volatile organic compounds; VVOCs, very volatile organic compounds.
to ETS and the available exposure measurements generally are based on determination of one or more tracers of ETS such as nicotine, as it is not possible to quantify all the constituents. Thus, two related key scientific issues are important in making ETS exposure measurements: a) How effectively do tracers measure exposures to ETS as a whole complex chemical mixture of gases, vapors, and particles? b) Does the chemical composition of ETS in different indoor environments differ significantly because of dynamic processes that remove various components at different rates and in different proportions? This article considers recent field and chamber measurements of ETS tracers and research on the various dynamic processes that affect the composition of ETS. This growing body of work has significant implications for the use of tracers to estimate exposures to the entire ETS mixture.
Understanding the dynamic behavior of ETS in indoor environments is facilitated by grouping the compounds in ETS into several major components according to their physicochemical properties such as physical state, vapor pressure, and type of compound: 1) very volatile organic compounds (VVOCs), 2) volatile organic compounds (VOCs), 3) semivolatile organic compounds (SVOCs), 4) particulate matter (PM) and its organic compounds, and 5) gas-phase inorganic compounds. Examples of the compounds in each category are given in Table 1. In this article we focus on the first four components and consider the influence of physicochemical properties and dynamic processes on estimations of indoor air concentrations and exposures.

Physicochemical Properties and Dynamic Processes Affecting ETS Components and Tracers
Exposures to ETS in indoor environments depend on smoking and ventilation rates, the volume into which the smoke is emitted and dispersed, and several other Environmental Health Perspectives * Vol 107, Supplement 2 * May 1999 Thus, the concentration of any species, i, at any given time depends on the emission rate (first term of the equation) from SS and exhaled MS, the volume into which the ETS species is emitted (and convective air mixing), the rate of removal of the species by ventilation (second term), and the deposition and sorption and the desorption and 320 re-emission of species i from indoor surface j (third term).
Emission rates for the various constituents of ETS can be estimated as the product of the emission factors and the smoking rates in a given setting (5). Emission factors, mass of compound emitted per cigarette smoked, are important for estimating and predicting exposures to ETS over a wide range of smoking and ventilation rates and indoor conditions. Emission factors can also be used with measured concentrations of ETS tracers to estimate exposures to other ETS constituents when the tracer and constituents exhibit similar physicochemical behavior. In contrast to MS emission factors, ETS and SS emission factors generally exhibit relatively small variations across tobacco brands (6)(7)(8).
Volumes of the indoor spaces into which ETS is emitted can vary widely. The volume of a small office, for example, might be about 20 m3, whereas the volume of an industrial workplace might be of the order of several thousand cubic meters. Once emitted, ETS is diluted and mixed within the volume of the indoor space by convective mixing of indoor air. Recent experiments (9,10) have shown that the convective mixing of indoor air is generally quite rapid, e.g., minutes in room-size spaces, particularly if there is mechanical ventilation. Mixing times correlate well with the inverse of the cube root of the power input to mechanical ventilation in accordance with theoretical predictions (10). For example, mixing time was about 7 min for a typical power input for a supply air jet. This is quite fast compared to the typical times required for removal by ventilation or for the occurrence of most adverse health effects from ETS.
Outside air ventilation rates in office buildings typically range from about 0.3 to 3 air changes per hour (11,12 (14).
Indoor air concentrations of many ETS components are also influenced by several other dynamic processes including deposition, sorption, and desorption. The dynamic processes that affect the indoor concentrations of ETS components, the overall chemical composition of ETS, and the approximate time scales over which these occur are summarized in Table 2. The processes, which are most influential for each of the four components of ETS are discussed below in more detail.

Very Voladle Organic Compounds
The VVOCs have relatively high vapor pressures (-7-13 kPa). Their concentrations in indoor air are largely determined by emission rates, indoor volumes, and ventilation removal. In general, they are not very strongly sorbed on indoor surfaces. SS  (15) and Guerin et al. (2). Emission factors for some VVOCs have been measured in environmental chambers using both simulated ETS (diluted SS) (7,8) and ETS generated continuously by smokers (16). These measurements included formaldehyde, acetaldehyde, acrylonitrile, N,Ndimethylnitrosamine, and 1,3-butadiene, and were made using either the Kentucky reference cigarette 1R4F (University of Kentucky) or a representative sample of cigarettes from the marketplace.
Several chemically reactive VVOCs such as 1,3-butadiene and acrolein are present in high proportions in emitted SS. These species are likely to react irreversibly with indoor materials or with ozone or OH radicals in air (17,18) and thus be removed by this additional pathway. However, chemical reactions of ETS components have not been systematically investigated and the degree to which such reactions occur in indoor environments has not been determined.

Volatile Organic Compounds
VOCs generally are classified as those compounds having vapor pressures in the range of 0.01 to 10 kPa. VOCs are more sorptive than VVOCs. Emission factors have been determined for many VOCs in SS (15,19) and in smoker-generated and -simulated ETS (7,8,16,20). In addition to the influence of emission and ventilation rates on the indoor concentrations of these compounds, sorption on and desorption from indoor surfaces influence air concentrations.
Sorption and desorption of VOCs from specific indoor materials have been investigated to a limited extent. Most of these investigations were conducted in small chambers (<1 m3) with higher surface-tovolume ratios than rooms (21)(22)(23)(24). Thus, the rate constants cannot be easily extrapolated to room-size indoor environments. In general, however, both sorption and desorption appear to occur relatively rapidly, i.e., in minutes to hours. Tichenor et al. (23) investigated the sorption and desorption of ethylbenzene on carpet, painted wallboard, and ceiling tiles in small environmental chambers at concentrations of milligrams per cubic meter. The time required to reach equilibrium between sorbed and air concentrations for these materials was about 15 hr, with an initial very rapid sorption rate over the first few hours. Sorption on all three materials was well described by the linear portion of a Langmuir isotherm (25). After equilibrium had been reached between air and sorbed-phase ethylbenzene concentrations, fresh air was passed through the chamber to measure re-emission of the sorbed ethylbenzene. An initial rapid desorption occurred over the first few hours, followed by a longer period of tens of hours in which the ethylbenzene was more slowly emitted to yield very low air concentrations. Desorption from wallboard and ceiling tiles could be characterized using the Langmuir model (25). Re-emission from carpet deviated somewhat from this model. The data suggested that desorption from carpet is a more complex process that might involve diffusion from the solid phase (carpet backing) to the surface (fibers) followed by desorption from the surface into the room. Many other VOCs, such as benzene and toluene, would be expected to exhibit similar behavior.
There has been only limited investigation of the sorption and desorption of VOCs in room-size environmental chambers and buildings. In chamber experiments to measure emission factors for simulated ETS, Daisey et al. (7) reported evidence for the deposition and sorption of some of the higher molecular weight VOCs (phenol, the cresols, 3-ethenylpyridine (3-EP), and N-nitrosopyrrolidine) onto the stainless steel walls of the chamber. The rates of deposition in this 20 m3 environmental chamber were about 0.1 to 0.2/hr. Rates in other settings will differ depending on the size of the room, air mixing, and available sorption surfaces. Desorption rates were not determined in these experiments. Johnson et al. (26) investigated the sorption and desorption of a VOC, p-dichlorobenzene (PDCB), in an environmental chamber furnished with painted wallboard, carpet, and drapes and ventilated at an air exchange rate of 1 to 1.3/hr. The first set of three experiments was conducted with wallboard only; the next three experiments included carpet, and finally, three experiments were done with all three materials in the chamber. Within a few hours a steady-state concentration of PDCB was achieved due to sorption. After steady state was achieved, the emission source was removed and desorption monitored over the next 10 to 20 hr.
Desorption also occurred relatively rapidly over the first hour or two, with the air concentration reaching a level of about 10 to 15% of the original steady state. In general, the monitoring data agreed well with the model that included one sink for each of the materials. However, there was some evidence for a possible additional, slower sink for re-emission from the drapes.
Tichenor et al. (23) reported an experiment with PDCB emissions in a test house with an air exchange rate of 0.35/hr. The PDCB solid was placed in a closet in one of the three bedrooms for 11 days and air concentrations were measured on days 4, 6, 8, and 11. Air concentrations showed only small variations (<45% of the lowest measured concentration) among the three rooms or over time (< 30%) during this period. The PDCB source was removed on day 11 and air concentrations were measured on days 12, 14, 16, and 18. The air concentrations on day 12 dropped to about half those of day 11, and on day 14 to about one-fourth those of day 11, indicating re-emission of sorbed PDCB. The authors estimated that approximately 40% of the emitted PDCB had been sorbed on indoor surfaces. By day 18, air concentrations were about 10 to 15% of the original steady-state concentrations. Semivlatile Organic Compounds SVOCs generally are defined as those compounds with vapor pressures ranging from 10-2 to 104 kPa. Eatough et al. (14) summarized emission factors for many of the SVOCs in SS. Chortyk and Schlotzhauer (19) also reported SS emission factors for a number of SVOCs for 20 different cigarette brands with mainstream tar values ranging from 1 to 23 mg/cigarette. ETS emission factors for cigarette brands currently in the marketplace or for the Kentucky reference cigarette 1R4F have been determined (or can be inferred from measurements made in chambers) for nicotine (6,16,20) and some of the polycyclic aromatic hydrocarbons (PAHs) (27).
SVOCs are found in both the particle and vapor phases. Distribution depends on the temperature and age of the ETS and the partide concentration. The gas-particle distribution of nicotine has been measured by several investigators (20,(27)(28)(29); most of the nicotine was found in the vapor phase. Gundel et al. (30) measured the gas-particle phase distributions of some of the PAHs in ETS using a sampler designed to minimize changes in the vapor-particle partitioning during sampling. In general, as molecular weight increased, the fraction of PAH in the particulate phase increased. Liang and Pankow (29) determined gas-particle partition coefficients for a number of SVOCs in ETS based on the desorption of the compounds from particles Environmental Health Perspectives * Vol 107, Supplement 2 * May 1999 collected on a filter. The SVOCs included some alkanes (C16-C22), several PAHs, quinoline, isoquinoline, carbazole, and nicotine. As thermodynamic consideration predicts, vapor pressure and temperature are the major determinants of the partition coefficients for the alkanes, PAH, and carbazole. For nicotine, however, the pH of ETS is a major determinant of the partition coefficient.
N-Nitrosonornicotine (NNN) and 4-(methylnitrosamino)-1 -(3-pyridyl)-1butanone (NNK) are probably SVOCs also, based on their vapor pressures, but the gas-particle distribution for these and for other SVOCs have not yet been experimentally determined. Evaporative losses of VOCs and SVOCs from PM into the gas phase can also occur over time (31)(32)(33), but this appears to be a slower process than deposition to indoor surfaces.
For the SVOCs, sorption and desorption from indoor surfaces also can be significant processes that control the air concentrations. The dynamic behavior of nicotine is of particular interest, as it has so commonly been used as a tracer of ETS. Van Loy et al. (34)(35)(36) investigated the deposition and re-emission of vapor-phase nicotine in a 20-m3 stainless steel environmental chamber. Experiments were conducted first in an empty chamber in which all deposition losses were to the walls, ceiling, and floor, and then in the same chamber with a carpeted floor. Nicotine was repeatedly volatilized into the unventilated environmental chamber (the air exchange rate due to infiltration was 0.1 5/hr) and its decay over time was measured until a steady-state concentration was reached. This generally required 2 to 3 days following each injection. However, about 80 to 90% of the nicotine was deposited and sorbed on the surfaces within the first 1 to 2 hr after the nicotine was emitted into the chamber. The mass sorbed on the walls and carpet in equilibrium with the air was also determined. The carpet sorbed approximately 100 times more nicotine per square meter than did the stainless steel walls. After the fifth injection and decay process, the chamber was actively ventilated and resealed and the re-emission of sorbed nicotine into air monitored.
The sorption behavior of nicotine in the carpeted chamber was modeled using two different sorption dynamics models, one based on surface sorption and one based on sorbate diffusion in a homogeneous polymer (35). A deposition velocity of 4.5 m/hr and a re-emission rate constant of 0.0008/hr were estimated for carpet from these experiments (Equation 1). In the first 5 to 6 hr after emission of nicotine into the chamber, the surface-sorption model more accurately captured the dynamic behavior of the nicotine. Over the next several days, however, the air concentration of nicotine decreased more slowly in a nearly linear fashion. This suggests that there may be two sinks operating, one rapid and surface-dominated and a second slower sink controlled by diffusion into the polymer backing of the carpet. Once there was a large sorbed mass of nicotine, reduction of the air concentration through ventilation led to a relatively rapid re-emission of nicotine into air to re-establish the equilibrium concentration.
Dynamic changes in the vapor-phase concentrations of several PAHs in ETS have also been examined over a 3-hr period in a 30-m3 environmental chamber under static conditions (no ventilation) (33). Deposition losses of the vapor-phase PAH were highly dependent on vapor pressure-more rapid losses were observed for the higher molecular weight PAH. For example, phenanthrene, pyrene, and benz[a]anthracene decayed faster in their gas-phase concentrations than the more volatile species such as naphthalene and its methyl derivatives. The effect of these differential deposition rates was to change the relative proportions of the different vapor-phase PAH in ETS and therefore the overall chemical composition of ETS.

Particulate Matter from EMS
Exposures to the PM component of ETS have been a major focus of much of the research on ETS. ETS emission factors for PM have been reported by a number of investigators. These range over about a factor of two, from an average of 17 mg/cigarette reported by Leaderer and Hammond (6) for 10 U.S. cigarette brands to 8 mg/ cigarette reported by Daisey et al. (7) for simulated ETS (diluted and aged SS) from six commercial U.S. brands. Differences in brands smoked accounted for only about 15 to 20% of the variability (6,7). Two other factors can contribute to differences in the reported emission factors: whether the ETS included exhaled MS and experimental protocol differences that result in dynamic changes in the ETS. The higher PM emission factor reported by Leaderer and Hammond (6) was based on measurements of ETS generated continuously by smokers in a ventilated chamber with high rates of recirculated air. This "fresh" ETS induded exhaled MS as well as diluted SS. Exhaled MS accounts for 15% of ETS PM (37), which would account for some of the differences among reported emission factors. The experiments of Daisey et al. (7) involved emission, dilution, and mixing of the SS from three cigarettes into a chamber over a period of approximately 30 min. The chamber was not ventilated (infiltration was approximately 0.15/hr). The simulated ETS was held in the chamber for an additional 4 hr, with samples collected over this period, during which time there were probably volatilization losses of SVOCs and water from the PM.
Emission factors for some of the organic compounds in PM were summarized by Eatough et al. (14) and by Guerin (40). The particulate PAH in ETS exhibited a bimodal distribution. In general, the highest concentrations were associated with particles with MMADs less than 0.1 pm, with a second mode with a MMAD of about 0.6 pm. These results suggest that lung deposition patterns and amounts differ for the PAH particulate benzo[a]pyrene and ETS particle mass.
Emission and ventilation rates generally have the largest influence on the indoor concentrations of PM. There are also small depositional losses of PM to indoor surfaces. For PM, the third term in Equation 1 can be considered the difference between the deposition removal term and the reemission term (-zero). Xu et al. (39) reported deposition losses of PM in ETS to the walls and surfaces of a chamber at a rate of about 0.01 to 0.05/hr. They estimated that 10 hr after one cigarette is smoked, at an air exchange rate of 0.03 /hr, 22% of the partides by mass are deposited Environmental Health Perspectives * Vol 107, Supplement 2 * May 1999 on indoor surfaces. At 0.5 air changes per hour, 6% is deposited. There also may be losses of SVOCs from the PM over time due to removal of vapor-phase SVOCs from air by deposition to surfaces. As the vapor-phase SVOCs are deposited to surfaces, particulate-phase SVOCs volatilize from particles to restore equilibrium between the two phases.

Estimating Exposures to ETS and Its Components
Because it is prohibitively expensive to measure all or even most of the constituents of ETS in field surveys of ETS exposure, exposure typically has been quantified using two approaches: a) mass balance modeling based on measured emission rates and the physical characteristics of the indoor environment (Equation 1), or b) measurements of one or several tracers ETS to infer concentrations of other components or of ETS as a whole. Such tracers must be unique to tobacco smoke, have similar emission rates for different cigarette brands, and have proportions consistent with those of the species they are used to trace (1). To meet this last criterion, some researchers have suggested that the tracers should be similar to the ETS species they are used to trace (20,41,42).

T I re to ETS Partculate Matter
Much of the focus on ETS exposures has been on the PM component. Nonvolatile tracers of PM include UVPM, a measure of the ultraviolet absorbance of the methanol extract of particulate matter at 325 nm (43), FPM, the fluorescence of a methanol extract of PM (44), solanesol (31,45), and scopoletin (46). The latter two tracers are high molecular weight compounds believed to be specific to tobacco smoke.
Vapor-phase nicotine (and its metabolite, cotinine) generally has been the most widely used tracer for ETS PM and ETS as a whole (6,(47)(48)(49). Vapor-phase nicotine is a particularly attractive ETS tracer because it is relatively easy and inexpensive to measure and it meets the general criteria for a tracer. However, its suitability as a marker has been questioned by some researchers because it exhibits decay patterns different from those of many other ETS constituents. Chamber experiments indicate that 80 to 90% of the nicotine in freshly emitted SS is deposited to surfaces within a few hours of emission (20,34,35,42). In contrast, respirable particulate matter (RSP) from ETS is removed from indoor environments largely by ventilation, with some smaller losses through deposition to indoor surfaces (38,39). Thus, the ratios of RSP to nicotine in ETS vary substantially with changes in time and ventilation rates (42), at least over short periods.
In contrast to the findings in shortterm experiments in environmental chamber studies, Leaderer and Hammond (6) found reasonably good linear correlations between 1-week measurements of RSP and nicotine (r2 =0.71) and between nicotine and the number of cigarettes smoked (r2= 0.67) in field measurements made in 96 homes. The slope of the regression line, 10.8, was consistent with the ratio of RSP to nicotine measured for ETS in an environmental chamber. These results suggest that nicotine measured over 24 hr or more can provide a reasonable estimate of exposure to ETS particles. The results also suggest a possible quasi-equilibrium for sorption and desorption of nicotine over longer time periods and indicate that nicotine measurements made over such periods might provide reasonable estimates of exposures to other physicochemical components if we had a quantitative understanding of the dynamic behavior of those components relative to nicotine.
Van Loy et al. (36) recently examined the apparent discrepancy between shortterm measurements in environmental chambers and the longer term nicotine and RSP measurements made in residential buildings. They used new models to examine the effects of reversible sorption on nicotine's suitability as an ETS PM marker. The dynamic behaviors of nicotine and PM from ETS were modeled for both the environmental chamber experiments described by Nelson et al. (42) and for 24-hr average concentrations in a 500-m3 house in which smoking occurs regularly for 16 hr/day. Results showed that the apparently contradictory observations could be reconciled by taking into account the sampling times and the dynamic behavior of nicotine, including its sorption and re-emission under different environmental conditions. Specifically, in indoor environments in which smoking occurs on a fairly regular basis for an extended period, for example, the 96 homes in which Leaderer and Hammond made measurements, the sorbed mass of nicotine becomes large relative to the mass emitted by a single cigarette, and re-emission from indoor surfaces becomes significant relative to direct emission. Furthermore, the time intervals during which air concentrations are elevated (immediately after nicotine emission) are quite short relative to the sampling period. Thus, the average air concentration of nicotine remains relatively constant and the ratio of nicotine to PM is observed to be relatively constant. The environmental chamber experiments of Nelson et al. (42) and others are examples of conditions under which the ratio of nicotine to PM changes rapidly because of the rapid deposition to surfaces that occurs immediately after emission. The shorter measurement intervals spanned periods in which the air concentration of nicotine changed dramatically.
Evidence from field measurements supports the idea of indoor reservoirs of sorbed nicotine that are reemitted into air when smoking has stopped. Vaughan and Hammond (50) measured average weekly nicotine concentrations in offices of smokers that ranged from 4 to 24 pg/m3 before a smoking ban. Seven weeks after the ban, nicotine was still measured in the air but at concentrations ranging from 0.1 to 0.5 pg/m3. This suggests a large remaining reservoir of sorbed nicotine.
The recent personal monitoring measurements reported by Jenkins et al. (49) also indicate that nicotine is a suitable tracer for the PM from ETS. These investigators measured various tracers of ETS for 24-hr periods for four groups: smokers working in environments in which smoking was permitted, smokers working in nonsmoking environments, nonsmokers working in workplaces in which smoking was permitted, and nonsmokers working in nonsmoking workplaces. Figure  Environmental Health Perspectives * Vol 107, Supplement 2 * May 1999 regression line between RSP and nicotine for 96 residences in New York State (6). Jenkins et al. (49) also measured the ETS particulate tracer solanesol. The correlation coefficient for the regression of solanesol with RSP for the 12 data points from groups 1, 2, and 3 (for which all data were above the limit of detection) is 0.92, and the slope of the regression line, which should be equivalent to the ratio of solanesol to RSP from ETS, is 0.018. Benner et al. (31) reported that the concentration of solanesol in ETS particulate matter is 22.2 mol/g or 0.017 g/g of PM. This agrees well with the ratio implied by the data ofJenkins et al. (49) and supports the use of solanesol as a tracer of PM from ETS. A similar regression of concentrations of scopoletin versus RSP for the Jenkins et al. data yields a regression line with more scatter (r2=0.90)-probably because of the difficulty of the scopoletin analysis-and a slope of 0.69 ng scopoletin per pg RSP. Risner (46) reported an emission factor of 111 ng scopoletin/cigarette that combined with an RSP emission factor of 17 mg/cigarette (6) implies a concentration of 0.0065 ng scopoletin/pg PM from ETS. The reason for this discrepancy is not known.

Estimatig E ures to VOCs and SVOCs from ETS
VOCs and SVOCs constitute a major proportion of the organic mass of ETS, and many of the compounds in these fractions are known to be biologically active as carcinogens in animals. Gas-phase ETS also has been implicated in coronary heart disease (51). However, efforts to trace exposures to these constituents have been much more limited than those for PM. There have been only a few investigations of the contributions of ETS to VOC exposures in offices. Comparisons of the concentrations of selected VOCs in smoking and nonsmoking offices have been reported by Bayer and Black (52) and by Proctor et al. (53). They found little difference in VOC concentrations that could be attributed to smoking. However, estimating the ETS contribution to VOCs is complicated by the differences among buildings in the relative contributions of such sources and in ventilation rates. Thus, it is unlikely that any consistent differences would be observed, even when smokers contributed to the indoor concentrations of these compounds. Therefore, ETS tracers or mathematical models based on the mass balance principle are needed to provide estimates of the ETS-attributable exposures to VOCs. Daisey et al. (54) did this for four office building scenarios based on a mass balance model and emission factors for VOCs in SS available in 1990. Table 3 presents the modeled concentrations for one VVOC and three VOCs for the two of the cases considered in Daisey's paper (54) but with ETS emission factors (7). For comparison Table 3 also indudes the concentrations of these compounds found in office buildings in which smoking was allowed (52,53).
Several VOCs unique to ETS have been examined as possible tracers of other VOCs in ETS. Heavner et al. (55) investigated the use of 3-EP as a tracer of VOCs in ETS in residential settings. This compound was not found in homes of nonsmokers (55) and there was a significant correlation between 3-EP and smoking activity in homes of smokers. Heavner et al. (55) used source apportionment based on 3-EP and   Unfortunately, Hodgson et al. (56) did not examine the relationships between nicotine and the VOCs in these office buildings. However, linear regression analysis of the concentrations of nicotine and 3-EP tracers reported by Jenkins et al. (49) can provide some information. Figure 2 shows the high correlation between these two tracers. The slope of the linear regression line, which is the ratio of 3-EP to nicotine, is 0.4. This is considerably lower than the ratio of 0.7, which can be calculated from environmental chamber measurements (7). This difference in the ratio may be due to the difference between the There are many other VOCs and SVOCs that are of interest, for example, the N-nitrosamines. Presently we lack experimental evidence that they can be traced effectively by nicotine or even by a VOC tracer such as 3-EP or pyrrole. Simultaneous measurements of multiple VOCs, SVOCs, and the various tracers in environmental chambers under more realistic conditions (with furnishings, repeated cycles of smoking, etc.) and in real buildings are needed to determine the relative proportions of these components to ETS tracers. Such experiments are also needed to evaluate the accuracy of various ETS tracers in tracing exposures to the entire ETS mixture.

Implications of Existing Data on the Chemical Composition of ETS
It has generally been assumed that the overall chemical composition of ETS is similar to that of diluted SS. There are sufficient data to indicate that this assumption is incorrect for at least some of the VOCs and SVOCs in ETS that are sorbed on and desorbed from indoor surfaces. For example, the emission factor for nicotine in SS is about 5 mg/cigarette (7,57), whereas the emission factor measured in environmental chambers in which dynamic processes occur is only about 1 mg/cigarette. It is interesting to note that the emission factor calculated from the measurements by Eatough et al. (20) of gas-phase nicotine in a Teflonwalled environmental chamber is more consistent with that of SS, presumably because nicotine does not sorb on Teflon. There have been only a few experiments on sorption and desorption of VOCs and SVOCs in environmental chambers furnished to simulate real indoor environments (58). Such experiments are needed to better characterize the relative proportions of these compounds in ETS and their dynamic behavior and to evaluate the effectiveness of various ETS tracers. These experiments can also provide some insight into one of the questions raised in the introduction, i.e., "Does the chemical composition of ETS in different indoor environments differ significantly because of dynamic processes that remove different components at different rates and in different proportions?"

Summary and Conclusions
This paper has considered the dynamic behavior of four major physicochemical components of ETS and recent experimental evidence on the effectiveness of various tracers for measuring exposures to ETS as a whole and to major components of ETS in indoor environments. For PM and the unreactive VVOC components, emission and ventilation rates are probably the most important processes controlling indoor concentrations and exposures of nonsmokers. For the VOCs and SVOCs, sorption on and desorption from indoor surfaces are also significant processes controlling the indoor concentrations. There is evidence that the distributions of gas-and sorbedphase compounds in these two groups are functions of vapor pressure and molecular weight, with a higher proportion of higher molecular weight compounds sorbed on surfaces. However, the behavior of these ETS compounds has not been sufficiently investigated to allow their concentrations to be predicted with any confidence.
Field measurements in residential and office environments (6,49) indicate that nicotine is a good tracer for PM. Laboratory and modeling studies suggest that nicotine measurements can be used to estimate PM from ETS in indoor environments when the following criteria are met: a) smoking occurs regularly in the environment, b) the system is almost quasi-steady state, and c) sampling time is significantly longer than the characteristic times for removal processes.
When these criteria are not met, for example, for certain individual residences or offices, nicotine is probably a less reliable quantitative indicator of PM exposures from ETS, although how biased a measure it provides in such circumstances remains to be determined.
The field measurements of Jenkins et al. (49) combined with laboratory measurements support the usefulness of a particulate tracer such as solanesol for tracing PM from ETS. Because there is reason to believe that the dynamic behavior of the VVOCs is dominated by the same factors that dominate PM, tracers for PM may also be good tracers for unreactive VVOCs from ETS.
VOCs and SVOCs also constitute a significant proportion of total ETS mass. Recent laboratory and field data (56) provide evidence that VOCs from ETS can be traced using compounds within this component such as 3-EP, pyrrole, or pyridine, which have similar physicochemical properties. Source apportionment of VOCs based on field measurements of 3-EP and the ratio of 3-EP to the VOCs indicates that ETS can be a significant contributor to the total VOCs in office smoking rooms (56). To date, no systematic investigations have been made of the effectiveness of any of the usual ETS tracers for estimating exposures to the SVOCs in ETS.
Nicotine may also prove to be a useful tracer for the VOCs and SVOCs from ETS if the previously mentioned criteria are met. However, there are significant differences between the ratios of 3-EP and nicotine measured in the field and in an unfurnished environmental chamber. This is hypothesized to be because of differences in the sorptive materials in these two environments. If this hypothesis applies to other VOCs and SVOCs, the ratios of the compounds to nicotine (and other tracers) would have to be determined in furnished environmental chambers using more realistic smoking patterns and ventilation rates.
It is clear from existing data that freshly generated and diluted SS is not chemically identical to ETS, but we do not yet have quantitative information on the relative proportions of many of the VOCs and SVOCs in ETS. Carefully designed experiments should be conducted in environmental chambers furnished to better simulate real indoor environments to provide this information. Such experiments should also provide much needed experimental evidence that nicotine can be used to trace the VVOCs, VOCs, and SVOCs in ETS under some conditions.