H43D-0391 1340h
Biogeochemistry of mercury in soils and sediments in a mining-impacted watershed, California
The East Davis Creek watershed, located in the California Coast Ranges, is host to historic mines that provided mercury for recovery of gold in the Sierra Nevada goldfields in the mid-to-late 1800s. Bedrock in this watershed includes marine sedimentary rock, serpentinite, and hydrothermally altered serpentinite. Cinnabar (HgS) found in the altered serpentinite is the primary ore mineral for mercury. We evaluated the hypothesis that mercury is sequestered in soil organic matter downstream from source areas, releasing a fraction as water-soluble methylmercury. Microbial biomass and the presence of sulfur-reducing bacteria implicated in mercury methylation were quantified using phospholipid fatty acid (PLFA) data. Methylation incubations were performed on soil and sediment inoculated with water from Davis Creek Reservoir and sealed in glass containers under an anoxic headspace for 21 days. Methylmercury was measured on extracts of the soils at the start and at the end of the incubation period. Two sources of mercury to stream sediments, a soil with an altered serpentinite parent and mine tailings, were incubated. Stream sediment, an overbank deposit soil and a wetland soil forming from these sediments were also incubated. The overbank deposit soil is periodically flooded. The wetland soil around the edge of Davis Creek Reservoir is perennially saturated with water. The altered serpentinite soil and mine tailings had the highest total mercury concentrations (170 and 150 ng Hg /g, respectively). Total mercury concentrations in stream sediments are low (Íš1 ng Hg/g), with higher mercury concentrations in the overbank (3 ng/g) and wetland soils (18 ng Hg/g). Mercury leached from altered serpentinite soils and mine tailings may be transported downstream and sequestered through sorption to organic matter in the overbank and wetland soils. PLFA biomarkers for {\it Desulfobacter} (10Me16:0) and {\it Desulfovibrio} (i17:1) were present in all incubated materials, with lower concentrations in mine tailings and stream sediment relative to the three soils examined. Methylmercury was initially present in greater concentrations in the overbank deposit (23 ng HgMe/g) soils. The elevated methyl mercury in the overbank deposit soil may be due to the greater biomass of sulfur reducing bacteria indicated by the 10Me16:0 and i17:1 biomarkers. During the 21-day incubation, methylmercury increased from 0.6 to 15 ng HgMe/g in the wetland soil concomitantly with sulfate decreasing from 130 to 7.0 mg SO$_{4}$$^{=}$/g. Methylmercury concentrations did not change appreciably in the other soils, although sulfate decreased from 19 to 2.0 mg SO$_{4}$$^{=}$/g in the overbank deposit soil. These data suggest that overbank deposits and wetland soils sequester mercury leached from upstream sources, with a fraction of this mercury released through microbial methylation.
H43D-0392 1340h
Geochemical Identification of Fresh Water Sources in Brackish Groundwater Mixtures; the Example of Lake Kinneret, Israel
Fresh waters that dilute brines are considered to have a negligible effect on the ion ratios of the resultant mixture. We show that the major element composition of the fresh end-member can be deduced from the chemical composition of the mixed waters. That composition, then, can be used to differentiate between different neighboring carbonate aquifers, which supply the water. This is demonstrated for the Fuliya and Tabgha saline springs, located on the northwest shore of Lake Kinneret (Sea of Galilee), Israel. At these springs, shallow meteoric fresh groundwater mixes with brines from deep aquifers. Seven saline springs and wells located at the Fuliya and Tabgha blocks were sampled over a year, and thirty two eastern Galilee fresh springs and wells were sampled as representatives of the fresh water end member. All samples were analyzed for major and minor ions. The saline spring data were used to construct mixing lines, followed by their extrapolation to low concentrations in order to derive the ion/chloride ratio characterizing the fresh component. We constructed ion/Cl vs. Cl curves; projection of the composition of fresh water on the calculated curve was used to identify a certain fresh water source as a possible end-member. Results indicate that the composition of the water feeding the Fuliya springs is different from that at Tabgha, reflecting interactions with different rocks in each basin. The major fresh water end-member diluting the Fuliya brines is characterized by high Mg/Cl and low Sr/Cl ratios, and is consistent with the composition of fresh groundwater in the dolomitic Cenomanian and Turonian aquifers widely exposed in the Fuliya drainage basin. The major fresh water end-member diluting the Tabgha brines, on the other hand, is characterized by low Mg/Cl and high Sr/Cl ratios, and is consistent with the composition of fresh groundwater in the chalky Eocene Timrat Fm. and Senonian outcrops. Although the chalky formations in the Tabgha drainage basin are exposed over only 20% of the area they contribute most of the solutes to the fresh water end-member. Rain flows over the chalky formations and then infiltrates into the Bar-Kokhba Eocene outcrops.
H43D-0393 1340h
Methane and nitrous oxide in submarine groundwater discharges: Toyama Bay, northwestern Japan
Submarine groundwater discharge (SGD) in coastal area plays important roles in global hydrologic cycles between land and ocean. Besides, SGD supplies chemical materials such as nutrient and organic carbons directly into the sub-surface ocean, so that it could be also important for discussing global geochemical cycles. The estimated global flux of SGD, however, is highly uncertain. Besides, little is clarified for the influence of the chemical input by SGD to ocean and the residence times of groundwater from recharge on land to discharge on seafloor. Oxic/anoxic condition of groundwater is one of the important parameters that characterize chemical compositions of groundwater. Methane and nitrous oxide in SGD fluids are sensitive tracers for the oxic/anoxic condition of groundwater. In this paper, we studied both SGD fluids and nearby groundwater on land using methane and nitrous oxide as tracers, in the coastal area of alluvial fans formed by Kurobe River and Katagai River in Toyama Pref., northwestern coast of Japan, where extensive distributions of SGD have been reported. Besides to their concentrations, we also determined stable isotopic compositions of methane ($\delta$$^{13}$C ) and nitrous oxide ($\delta$$^{15}$N and $\delta$$^{18}$O ) for additional tracers for discussing origins of the molecules. As for methane, we found that all of the SGD fluids contain less methane than natural waters in equilibrium with atmospheric methane. In proportion to the methane depletion, the $\delta$$^{13}$C values of methane in the SGD fluids vary from -50 $\permil$ almost comparable to atmospheric methane to highly $^{13}$C enriched value of -20 $\permil$. As for nitrous oxide, all of the SGD fluids contain more nitrous oxide than natural waters in equilibrium with the atmospheric nitrous oxide. Both the $\delta$$^{15}$N and $\delta$$^{18}$O values of nitrous oxide are lower than that of the atmospheric nitrous oxide. These results suggest that most of methane in the SGD fluids derived from atmosphere and a part of the methane is oxidized microbially in the course of groundwater flow through the aquifer, while nitrous oxide is produced via nitrification. On the other hand, we found that many of the land groundwaters contain more methane than the SGD fluids. The longer residence times of the land groundwaters than the SGD fluids must be responsible for the methane enrichment. When we categorized the groundwaters into oxic stage and anoxic stage based on the concentration of methane and nitrous oxide, we found that the oxic groundwaters are distributed along the Katagai River. We conclude that the SGD fluids must have flowed through the underground along the Katagai River.
H43D-0394 1340h
Hydrogeochemistry of formation water with implication to abnormal pressures in the Northern Songliao basin, China
Formation water is major component in sedimentary basin. The geochemistry properties of formation water are ought to be the final outcomes of basin evolution due to a number of hydrodynamic and thermodynamic factors. Therefore, hydrochemistry of formation water and its potential relationship with phenomena of abnormal pressures are of considerable interest and importance to study the generation, transport and accumulation of hydrocarbon in sedimentary basin. The Songliao basin is one of the largest Mesozoic continental basins in China related to rifting process, dominated by Cretaceous fluvial and lacustrine strata with volcanic and volcaniclastic rocks. The stratigraphic sequences can be classified into 6 hydrostratigraphic units from bottom to top: the Upper Jurassic Formation (J3), the Lower Cretaceous formation (K1), the Late Cretaceous Quantou Formation (K2q), Qingshankou (K2qn), Yaojia (K2y) Formation, Nenjiang (K2n) Formation. In this paper, the characteristics of the strata pressures, hydrogeology and hydrochemistry in the whole Songliao basin are investigated according to 2433 pressure data and 6270 water sample data collected in the different structural zones and geological formations in the northern Songliao basin. The pressure data shows that, not only the abnormal pressures exist in different formations and structural zones, but also the underpressure coexists with the overpressure in different structural zone within a single geological formation. Besides, the pressure coefficients change gradually from 1.02~1.30 in the western Central Depression to 1.05~0.5 in the eastern Central Depression, and then become 0.65~0.9 in the east Uplift zone, especially in the Quantou formation (K2q). The analysis of hydrochemistry data shows that the geochemistry of formation waters in the Songliao basin shows a distinct variation in different pressured systems. There are several significant vertical and horizontal zones of hydrochemistry characteristics in the whole basin, which have closed relationship with basin hydrodynamics, formation sequences and diagenesis reactions. The salinity of formation water ranges from 1.0 g/L to 20g/L in most formations. Water composition is dominated by water type of NaHCO3 but varies greatly in salinity and ion ratios in the different area. The origin of formation may be referred as the mixing of the evaporation and condensation of connate water, clay dehydration due to compaction, topography-driven meteoric water recharge as well as diagenesis alteration such as the dissolution/precipitation and albitization of plagioclase, and their impacts differ greatly in the different hydrochemical and abnormal pressured environment.
H43D-0395 1340h
Groundwater study using drill holes in the Abukuma granitic province, NE Japan: the multi-isotopic approach to evaluate crack water stability
Chemical nature or origin of groundwater in cracks has been poor understood because of difficulties on collection of water samples preserving its natural conditions. Little is known on the quantification on stability and mean residence time of crack water also. We conducted a study for flow parameters and processes of groundwater in cracks by drilling two bore holes, and analysis using the multi-isotope approach. The drill sites have set in a granitic province called Abukuma in Fukushima prefecture, and drill holes 140m- and 180m-deep were made. In situ sampling of waters at 6 depths for each drill holes are done with the single and double packer methods. Chemical type of groundwater has a variety with depth, the shallower groundwater is categorized as Ca-HCO$_{3}$$^{-}$ type with slight NO$_{3}$ contamination whereas deeper groundwater has Na-HCO$_{3}$$^{-}$ type. Bicarbonate concentration becomes greater in deeper level suggesting that older groundwater is placed in deeper cracks. Stable isotope composition of water showed that all the sample water is of meteoric origin. However, the $__delta$D at deeper level has significantly low values (20$__permil$ lower than the present one) obviously indicating that the groundwater does not originate from the present one. The $__delta$$^{13}$C values of total dissolved carbon from two sites, Shirasawa and Miharu, show different profiles vertically. The $__delta$$^{13}$C profiles indicate that carbon at the Shirasawa site is derived from biogenic source, but that at the Miharu site is influenced from other sources, such as _gcrustal fluid_h upwelling from a deep geologic environment. The contribution of crustal fluid can be canceled using the carbon isotopic mass balance, and the $__delta$$^{14}$C value of crack water excluding deep source contribution was evaluated. The apparent $^{14}$C date calculated using the evaluated $__delta$$^{14}$C value is getting older to deeper depth at both sites. At the Miharu site, the tritium is detected even at depth of 180m, indicating that relatively young water has invaded into the crack water. As the vertical profiles of $__delta$D and $__delta$$^{18}$O show relatively small changes, the vertical mixing of crack water, shallow surface water and deep crustal fluid likely occurs. However, the result beyond the 10000 yrBP in the apparent $^{14}$C date suggests that the carbon mixing does not frequently occurred. As for the Shirasawa site, the deeper the crack water, the lower the $__delta$D and $__delta$$^{18}$O values are represented. Groundwater with very low $__delta$D and $__delta$$^{18}$O values is likely recharged in an ice age consistent with the apparent $^{14}$C date showing the older age of carbon in the crack water than 10000 or 20000 yrBP. The evidence shows that crack water can be trapped for a very long period even at a shallow depth (80-180m).
H43D-0396 1340h
SF6 Tracer Release Study: A Contaminant Fate Study in Newtown Creek
Newtown Creek is a 5.5km creek that discharges into the East River, a 25km strait connecting Long Island Sound to the north and the New York Harbor to the south. Surface runoff dominates the freshwater input into the creek, for natural tributaries no longer exist. The areas directly adjacent to the creek are highly industrialized, and New York City's largest Water Pollution Control Plant (WPCP) discharges directly into creek. In August 2004, we injected sulfur hexafluoride (SF6) into Newtown creek to study the fate of oil seeping into the creek from an underground oil spill and the fate of nutrient rich effluent from the WPCP. We monitored SF6 in Newtown Creek, the East River, and the Upper Bay of New York Harbor for 7 consecutive days following the injection in order to investigate the spreading patterns and transport mechanics of waters exiting the creek, and to determine the ultimate fate of the contaminants/solutes originating in Newtown Creek. Dissolved oxygen (DO) measurements were collected simultaneously with SF6 measurements. A strong DO gradient exists in the creek, where waters in the upper reaches are anoxic. We use SF6 data to calculate mean residence times for Newtown Creek waters. SF6 was detected above background concentrations approximately 15km to the south of the creek at the Verrazano Bridge only 1 day after the tracer injection. By combining the movements of the SF6 distribution, the position of the oxygen gradient, and the residence time of Newtown Creek water, we can determine a lower boundary for oxygen consumption rates.
H43D-0397 1340h
Temporal Variations in $^{234}$U/$^{238}$U Activity Ratios in the Lower Mississippi River due to Changes in Source Tributary Discharges
The world's 25 largest river systems contribute nearly 50% of all freshwater to the global ocean and carry large quantities of dissolved trace metals annually. Trace metal concentrations in these systems show large variances on seasonal time scales. In order to constrain the causes of these variations, consistent sampling on sub-seasonal time intervals is essential. Here, we focus on the Mississippi River, the seventh largest river in the world in terms of freshwater discharge and the third largest in terms of drainage basin area. Biweekly sampling of the lower Mississippi River at New Orleans was performed from January 2003 to August 2004. Uranium concentrations and $^{234}$U/$^{238}$U activity ratios were measured for the dissolved component ($<$0.2 $\mu$m-fraction) of river water. Over the course of this study, dissolved U activity ratios spanned a range of about 25%, from 1.23 to 1.60. Dissolved U concentrations ranged from 0.28 to 1.06 ppb. The relationship between concentrations, activity ratios, and lower river discharge is complicated, and no clear pattern is observed on both biweekly and seasonal timescales. However, there does seem to be a relationship between the larger seasonal trends in the lower Mississippi River and variations in the discharge of its upstream tributaries. To constrain this relationship, we have sampled water from the Missouri River, the upper Mississippi River above the confluence with the Missouri, the Ohio River, and the Arkansas River in February, April, and August of 2004. For the upstream samples measured thus far, the highest dissolved uranium concentrations are observed for the Missouri River at 2.02 ppb, while the lowest are found in the Ohio River at 0.38 ppb. Dissolved $^{234}$U/$^{238}$U activity ratios are as unique for each tributary and vary from 1.36 in the Ohio River to 1.51 in the Missouri River. A preliminary mass balance analysis reveals that the lower river uranium activity ratios are controlled simply by the quantity and isotope signature of the waters discharged from the upstream tributaries. A discussion of the implications of this work for global ocean budgets of uranium will be presented.
H43D-0398 1340h
The Role of Plants in the 238U-234U Disequilibria of Stream Waters: The Example of the Strengbach Watershed (Vosges, France)
Recent TIMS or MC ICPMS analyses of U disequilibria in the dissolved load of stream and river waters have confirmed the potential of the U activity ratio in river waters as a specific tracer of chemical fluxes coming from rocks and soils (e.g., 1). These precise measurements have also outlined that occurrence of U activity ratios lower than one in dissolved load of river waters is not exceptional, especially at the scale of small watersheds. Such U values pose in turn the question of the real mechanisms controlling the supply of 234U-238U isotopes to the freshwaters. In order to address this question U activity ratios, Sr isotope ratios and the concentrations of major and some trace elements were analyzed in the different compartments of a small granitic watershed: the Strengbach environmental observatory (Vosges, France) ( http://ohge.u-strasbg.fr). In addition of the different streamwaters draining this watershed, the main horizons of weathering profiles, the associated soil solutions and the main tree species growing around were analyzed. The data confirm that the Strengbach stream water samples have generally U activity ratios lower than one and point out that soils solutions display a similar range of U and Sr variations. By contrast, tree and plant samples define different trends of variation in a plot of U activity ratios against Sr isotope ratios with, above all, U activity ratios systematically greater than one. These data show that trees and plants collected during this work cannot directly pomp their nutriments from soils and soil solutions sampled here, i.e. gravity solutions. They also suggest that, in this watershed, the plants, especially trees, play a central role in the weathering processes of rocks and minerals, and control a large part of the geochemical signature of the water samples collected on this watershed, including their U activity ratios. 1Riotte J and Chabaux F. (1999) GCA 63, 1263-1275.
H43D-0399 1340h
The Role Of Hydrologic Pathways And Biogenic Methane In The Sarita Wetland, St. Paul, Minnesota
Biogenic methane in wetlands and wetland sediments has been studied for its role in the carbon cycle, atmospheric chemistry and global warming, but the interaction of wetland pore-waters with groundwater and the dynamics of dissolved methane in groundwater are unclear. We investigated the role of groundwater recharge and advection in the distribution of dissolved methane, and the mechanism by which methane produced in the surface could be carried into the groundwater system. We present isotopic measurements of biogenic methane and dissolved inorganic carbon (DIC) in the Sarita Wetland, on the St. Paul Campus of the University of Minnesota, and also in six monitoring wells located in its watershed. Carbon isotopic values range between -58.4 and -10.6 permil in methane and between -14.1 and 0.8 permil in DIC. Results show that there is a major component of acetate fermentation in the production of the biogenic methane found in the wetland and in the groundwater. Hydrologic pathways are controlling the distribution of methane in the groundwater. As methane concentration decreases, the residual methane becomes more 13C-enriched. These results indicate that groundwater flow paths constitute a key element in the allocation and oxidation of dissolved methane. The degree of methane oxidation increases as methane moves in the flow path. This is an important observation since carbon allocation to groundwater as methane has not generally been taken into account for carbon budgets.
H43D-0400 1340h
Monitoring and Modeling the Fluctuations in Apparent Groundwater Age During a 30-Day Pumping Test.
Recent research shows that dispersion due to geologic heterogeneity can cause large (10's to 100's of yrs) variations in actual groundwater age within individual samples drawn from a well, even if well bore mixing is not significant. We hypothesize that the presence of such large ranges in groundwater age may cause the mean apparent age as estimated from environmental tracers such as CFC's, SF&_{6}$, and $^{3}$H-$^{3}$He to drift measurably during long-term, continuous pumping. This hypothesis was confirmed by 3-D numerical experiments wherein variation in groundwater ages under high-rate ($\sim$0.06 m$^{3}$/s; 1,000 gpm) long-term pumping was modeled using backward-time random walk particle tracking techniques combined with geostatistical simulations of hydrofacies heterogeneity. Results indicate that the age distribution within a water sample and the mean apparent age implied by environmental tracers is strongly influenced by historical atmospheric concentrations of environmental tracers and subsurface heterogeneity. As a partial implementation of this same experiment in the field, an abandoned well was pumped at a low rate ($\sim$0.005 m$^{3}$/s; 75 gpm) during 53 days. Water samples were collected from the top and bottom of a 25 foot well screen at 12 hour intervals for the first 30 days and were analyzed for CFC's, SF&_{6}$, and $^{3}$H-$^{3}$He. The measured tracer ages indicate that 1) CFC-11 apparent ages increased with time within the first five days of pumping and then remained constant for the remainder of the pump test; 2) trends in CFC-12, CFC-113 and SF&_{6}$ indicate a discrepancy in apparent ages with CFC-11, 3) water reaching the top interval is younger than water reaching the bottom interval. Gas samples were collected from the unsaturated zone to investigate possible CFC contamination and tracer concentration spatial variations in the pumping well recharge zone. Potential effects of heterogeneity and local CFC contamination on the monitoring results will be discussed.
H43D-0401 1340h
Weathering in and Calcium Losses From Semi-Arid Agricultural Landscapes: Insight From Strontium Isotope Ratios
The strontium isotope ratio ($^{87}$Sr/$^{86}$Sr) has been used in a number of recent studies of calcium cycling in forested ecosystems. In this research $^{87}$Sr/$^{86}$Sr was used to investigate weathering and, specifically, seasonal variation of calcium loss in drainage from semi-arid, agricultural landscapes in the Palouse Region of Washington State, USA. The Palouse is dominated by rolling loess hills. The soils are silt-loam Mollisols and the predominant origin of the loess substrate is continental crust. Tile drains are widely used to improve drainage of lower-slope fields. $^{87}$Sr/$^{86}$Sr of tile drainage, soil water, stream water and precipitation water were measured by multiple-collector Inductively Coupled Plasma Mass Spectrometry. $^{87}$Sr/$^{86}$Sr of precipitation waters exhibited considerable variation (0.708 to 0.713). $^{87}$Sr/$^{86}$Sr of the other water samples ranged from 0.707 to 0.708. These values are not close to continental crust values (0.716) but are more similar to basalts (0.702 to 0.707). $^{87}$Sr/$^{86}$Sr of tile drainage appeared to be negatively correlated with discharge during the rainy season. Thus it appears that sources of dissolved calcium in drainage vary seasonally. Results of this work may help predict the soil-acidification effects of heavy loading of these systems with ammonia-N fertilizers.
H43D-0402 1340h
Stable Cl And O Isotope Ratios Of Anthropogenic And Natural Perchlorates
Perchlorate (ClO$_{4}$$^{-}$) in aqueous systems, even in low concentrations, is recognized to have potential human health risks. The drinking and irrigation water supplies of millions of people in the U.S. have recently been found to be contaminated with perchlorate, and this problem continues to become even more widespread. Perchlorate, as a highly soluble and relatively inert anion, tends to persist over long time periods and its removal by conventional water treatment technologies is difficult and expensive. Many known sources of perchlorate contamination are anthropogenic, resulting from its extensive use as an oxidizer component in solid propellants for missiles, rockets, and fireworks. However, certain fertilizers derived from Chilean nitrate evaporate deposits are known to contain a low percentage of perchlorate ($<$0.5%) that may contaminate groundwater. New isotopic evidence provides insights on the possible natural sources of perchlorate in surface and ground waters. Stable isotope ratios of Cl and O can now be used to determine whether the source of perchlorates in a given area is natural or anthropogenic. Microbial perchlorate reduction has a large ($\sim$15 per mil) kinetic isotope effect, and this may be used to identify whether natural attenuation of perchlorate is occurring. Anthropogenic perchlorate salts in milligram amounts are readily analyzed for $^{37}$Cl and $^{18}$O isotopes. Extracting an isotopically measurable amount of perchlorate from natural waters, which are usually within ppb range of concentrations, is a challenge. But with the use of a new class of highly-selective bifunctional anion exchange resins, recovery of trace amounts of perchlorate for accurate isotopic analysis has been demonstrated. Isotopic characterization is being conducted on anthropogenic perchlorate reagents, natural perchlorate-bearing salt deposits, and perchlorate-bearing groundwaters. Significant and consistent isotopic differences in both the Cl and O isotope ratios between anthropogenic and natural perchlorate sources have been identified. Anthropogenic perchlorate has \delta$^{37}$Cl values between -3.1 and +1.3, and \delta$^{18}$O values between -24.7 and -16.1. In contrast, naturally-occurring perchlorate analyzed thus far has \delta$^{37}$Cl values between -14.5 and -11.8, and \delta$^{18}$O values between -9.3 and -4.2, as well as a significant $^{17}$O excess not seen in anthropogenic perchlorate. These distinct isotopic characteristics may indicate an atmospheric origin for natural perchlorate. Perchlorate extracted from groundwater samples can be clearly identified in terms of source. Stable isotope forensics will provide a powerful tool for understanding perchlorate occurrences and contamination in the environment.
H43D-0403 1340h
Seasonal Variation Of C/N and Delta 13C Of Particulate Organic Matters From Three Lakes in Taiwan
Surface waters from three lakes were sampled and measured every one to two months from 2003 to 2004. Shuanglian Pond (470 m in altitude) and Lake of Plum Blossom (50 m) are located at the northern Taiwan, whereas Longluantan (15 m) is situated in the southern part. To compare the differences of organic carbon isotopes and C/N among these lakes, filtered particulate organic matters (POM) were studied. POM delta 13C and C/N showed variations which corresponded to seasonal change during the monitoring period. The values of O13C ranged from -21.11 to -28.65 (concentrated between -22.45 and -26.77), while C/N ranged from 5.1 to 9.34. POM in the three lakes recorded heavier delta 13C in 2003 than in 2004, and showed lighter trend to the end of the year. However, the C/N of POM in these lakes revealed a more variable signal between 2003 and 2004.