Rome’s urban history inferred from Pb-contaminated waters trapped in its ancient harbor basins

Significance

Isotopic evidence showing that Rome’s lead water pipes were the primary source of lead pollution in the city’s runoff reveals the sedimentary profile of lead pollution in the harbor at Ostia to be a sensitive record of the growth of Rome’s water distribution system and hence, of the city itself. The introduction of this lead pipe network can now be dated to around the second century BC, testifying to a delay of about a century and a half between the introduction of Rome’s aqueduct system and the installation of a piped grid. The diachronic evolution of anthropogenic lead contamination is able to capture the main stages of ancient Rome's urbanization until its peak during the early high empire.

Abstract

Heavy metals from urban runoff preserved in sedimentary deposits record long-term economic and industrial development via the expansion and contraction of a city’s infrastructure. Lead concentrations and isotopic compositions measured in the sediments of the harbor of Ostia—Rome’s first harbor—show that lead pipes used in the water supply networks of Rome and Ostia were the only source of radiogenic Pb, which, in geologically young central Italy, is the hallmark of urban pollution. High-resolution geochemical, isotopic, and ¹⁴C analyses of a sedimentary core from Ostia harbor have allowed us to date the commissioning of Rome’s lead pipe water distribution system to around the second century BC, considerably later than Rome’s first aqueduct built in the late fourth century BC. Even more significantly, the isotopic record of Pb pollution proves to be an unparalleled proxy for tracking the urban development of ancient Rome over more than a millennium, providing a semiquantitative record of the water system’s initial expansion, its later neglect, probably during the civil wars of the first century BC, and its peaking in extent during the relative stability of the early high Imperial period. This core record fills the gap in the system’s history before the appearance of more detailed literary and inscriptional evidence from the late first century BC onward. It also preserves evidence of the changes in the dynamics of the Tiber River that accompanied the construction of Rome’s artificial port, Portus, during the first and second centuries AD.

Recent cases of lead contamination of drinking water in the Midwestern United States have highlighted the sensitive equilibrium between essential urban water supply, hydraulic infrastructures, and economic and public health (1). Lead pollution was already affecting the urban waters of great Roman cities two millennia ago (2–4), although to a lesser extent than suggested earlier (2, 5). Harbor sediment cores provide one of the best continuous records of human impact on the local environment. By combining isotopic analysis of lead in harbor sediments with evidence derived from archeological materials (e.g., lead pipes) and carbonates—known as travertine or sinter taken from aqueduct channels—recent studies have shown that the proportion of foreign lead in sediments constitutes a proxy for the expansion and contraction of the water supply system and therefore, of urban development over the lifetime of a city (2–4). Urban centers were particularly vulnerable to the interruption of their water supply network because of natural or human causes (6). Hiatuses in the imported Pb isotopic record of the Trajanic basin sediments at Portus (core TR14 in Fig. S1A), the maritime port of Imperial Rome, and those from the channel connecting Portus with the Tiber (Canale Romano) (core CN1 in Fig. S1A) reflect the sixth century AD Gothic Wars and ninth century Arab sack of Rome (2). The chronological record at Portus, however, only started with Trajan (AD 98–117), leaving the earlier history of Rome’s urbanization essentially undocumented, except for sparse archeological evidence and passing mentions in politically biased historical accounts. The same is the case for the development of Rome’s water distribution network before the reforms of Agrippa in the late first century BC (6), from which point on the admittedly problematic texts of Vitruvius (7) and Frontinus (Rome’s water commissioner century AD 100) (8) provide more information. The first Roman aqueduct, the Aqua Appia (late fourth century BC) (6, 9), appears well before any evidence of widespread use of lead water pipes (fistulae; in the late first century BC) (10), raising the question of how water was distributed.

Fig. S1.

(A) Location of ancient Rome’s harbor basins in the Tiber delta with the position of cores PO2 and PTXI-3 (orange circles) analyzed in this work and cores TR14 and CN1 analyzed by Delile et al. (2). (B) Map showing the archeological area of ancient Ostia with the location of core PO2 as well as the old and the new Tiber river courses. Both (A) and (B) modified from The Journal of Archaeological Science, with permission from Elsevier.

To explore these questions, we conducted a high-resolution analysis of major and trace element concentrations and Pb isotopic compositions of a 12-m-long sediment core (PO2 in Fig. S1 A and B) from Ostia. Ostia, today a featureless coastal plain near the Tiber River (Fig. S1B), was the first harbor to serve Rome (11). The time span encompassed by core PO2 is constrained by 15 radiocarbon dates covering the first millennium BC (Table S1) (12). Lead concentrations and isotopic compositions also were measured on sediments from the PTXI-3 core drilled at Portus in the basin of Claudius (AD 41–54) (Fig. S1A), the entry port of the Trajan basin, although at a coarser resolution than the PO2 core.

Table S1.

¹⁴C dates of cores PO2 and PTXI-3

Core	Depth (cm)	Material	14 C age B.P.	Calibrated 14C ages cal. BC/cal. AD (2σ)*
PO2	348–351	Wood	1.860 ± 30	AD 80–231
PO2	378	Wood	2.040 ± 25	158 BC–AD 24
PO2	400	Wood	2.040 ± 25	158 BC–AD 24
PO2	490	Charcoal	1.990 ± 25	43 BC–AD 62
PO2	490	Plant material	2.050 ± 25	163 BC–AD 16
PO2	526	Plant material	2.025 ± 25	96 BC–AD 51
PO2	526	Wood	2.050 ± 25	163 BC–AD 16
PO2	583–586	Plant material	2.235 ± 30	387–204 BC
PO2	603–606	Wood	2.160 ± 30	357–108 BC
PO2	700–707	Charcoal	2.185 ± 30	360–173 BC
PO2	815	Charcoal	2.350 ± 40	728–265 BC
PO2	851–854	Charcoal	2.125 ± 30	347–52 BC
PO2	869–872	Wood	2.165 ± 30	358–113 BC
PO2	1,050	Posidonia*	2.955 ± 25	835–736 BC
PO2	1,193	Wood	2.615 ± 30	827–774 BC
PTXI-3	203–206	Plant material*	2.150 ± 30	AD 115–310
PTXI-3	212–213	Posidonia*	2.375 ± 40	165 BC–AD 60
PTXI-3	265–266	Posidonia*	2.290 ± 30	35 BC–AD 135
PTXI-3	608–609	Charcoal	2.520 ± 30	795–540 BC
PTXI-3	621–622	Plant material	4.280 ± 30	3010–2870 BC

^*Ages were calibrated according to the IntCal13 radiocarbon calibration curve (31) using the Clam software (49). The age-depth modeling procedure was done with the Clam software to produce model dates comparable with those from Delile et al. (2). However, the Bayesian age-depth modeling software Bacon (56) provides similar model dates. cal, calibration.

Results

Stratigraphy.

The stratigraphy of the PO2 core can be subdivided into three main sedimentary units: preharbor (unit A), harbor (unit B), and postharbor (unit C) (Fig. 1). (i) Unit A is represented by bedded and shelly gray sands with Posidonia, attesting to a depositional environment of deltaic progradation up to the middle of the fourth century BC (12). (ii) Unit B is further subdivided into two subunits, B1 and B2. Subunit B1 forms the lower harbor sequence characterized by compact dark gray silt suggestive of a quiet environment and lasting until the beginning of the second century BC. (iii) Subunit B2 forms the upper harbor sequence and contains levels of yellow sands brought in by repetitive floods of the Tiber, which occurred from the second century BC to the third century AD according to the ¹⁴C age-depth model (Fig. 1). (iv) Unit C’s yellow bedded silts from the third century AD onward represent floodplain deposits.

Fig. 1.

Stratigraphic log of core PO2 showing the age-depth model of the core constructed using the Clam software (49) on 15 radiocarbon dates (shown with black labels on the stratigraphic log). Additional details on the age-depth model and the ¹⁴C dates can be found in Fig. S4 and Table S1. The magnetic susceptibility values of the core along with the grain size 50 percentile (11), EF_Pb, and F1 (detrital influence) and F4 (seawater influence) of the factor analysis of major and trace element concentrations (the detailed distribution of the individual elements is shown in Fig. S2) are shown on the right-hand side. The gray shadings highlight the synchronicity between the highest EF_Pb values, detrital activity in the harbor basin (negative values of F1), the grain size 50 percentile, and magnetic susceptibility [centimeter, gram, second (CGS)].

Elemental Concentrations.

The lowest Pb values are consistent with the natural Pb background level of 22 ppm of the Tiber delta sediments (13, 14). The Pb enrichment factor (EF_Pb) records the relative enrichment of Pb relative to the natural environment (EF_Pb = 1 signifies no Pb excess). The evolution of EF_Pb in the harbor deposits (Fig. 1) goes from (i) values slightly above the natural Pb background level at the base of the stratigraphic section in the preharbor unit A located between 12- and 9-m core depth, with a mean EF_Pb of ∼2.2, to (ii) values highly above the natural Pb background in the harbor subunit B2 (between ∼6- and 3-m core depth; mean EF_Pb ∼ 3.8), with three peaks reaching EF_Pb values up to 9.

The stratigraphic record of Pb concentrations varies in concert with the presence of detrital material, magnetic susceptibility, and to a lesser extent, the median grain size (Fig. 1). The EF_Pb peaks in subunit B2 attest to a high-energy regime of fluvial activity (gray shadings in Fig. 1). Factor analysis of major and trace element abundances clearly identifies the siliciclastic component with terrigenous elements, such as the rare earth elements, K, Mn, and Ba, as the first factor (F1) (Fig. 1, Fig. S2, and Table S2). Lead concentrations of the bulk sediment are tightly associated with F1 (Fig. 1) (r = −0.66 between EF_Pb and F1), which emphasizes that the terrigenous fraction is the main carrier of this element. Factor 4 (F4) inferred from the factor analysis opposes Na to heavy metals, such as Pb, Sn, and Cd (Fig. 1, Fig. S2, and Table S2). The strong influence of Na within the stratigraphy reflects either the salinity of the harbor itself or the invasion of the sediments by the salt wedge. They are negatively correlated with the proportion of anthropogenic material carried by the Tiber. For both the Trajanic and Ostia harbors, the association of the ostracodal marine group with high sediment Na contents is clear evidence of a marine-dominated environment (11, 14, 15).

Fig. S2.

Factor analysis of elemental concentrations (34 elements) in the PO2 core for the two instructive factors. F1 and F4 are anticorrelated with detrital and marine influence, respectively.

Table S2.

F1 and F4 data of the factor analysis done on major and trace element abundances

Sample name	Depth (cm)	F1	F4
PO2-86	9	0.155	0.092
PO2-85	38	0.735	−0.598
PO2-84	67	0.737	−0.304
PO2-83	87	0.810	0.085
PO2-1	109	0.659	0.282
PO2-2	138	0.504	0.281
PO2-3	167	0.504	0.519
PO2_4	209	0.712	0.574
PO2_5	238	0.453	0.481
PO2-82	264	0.788	−0.379
PO2_6	267	0.578	0.508
PO2_7	282	0.553	0.545
PO2-81	288	0.348	0.186
PO2-8	294	0.587	1.687
PO2-80	299	0.648	0.126
PO2_9	303	0.483	0.777
PO2_10	315	0.277	2.529
PO2_11	324	−0.275	0.881
PO2_12	333	0.095	1.009
PO2_13	342	0.134	1.037
PO2-79	346	−0.097	−1.772
PO2_14	350	0.762	0.737
PO2_15	369	−6.348	0.518
PO2_16	381	−1.092	−0.315
PO2_17	393	−1.402	0.591
PO2_18	399	0.146	0.766
PO2_19	442	−0.162	0.976
PO2_20	454	−1.862	1.165
PO2-78	457	−0.810	−0.934
PO2-77	460	−0.595	0.643
PO2_21	463	−0.561	1.535
PO2_22	472	−0.721	3.176
PO2_23	481	−0.567	2.414
PO2_24	490	0.073	1.809
PO2_25	506	−0.132	0.401
PO2_26	512	−0.270	0.760
PO2_27	521	−0.195	0.708
PO2_28	530	−1.700	1.436
PO2-76	536	−1.696	−0.074
PO2_29	542	−0.585	1.415
PO2-75	548	−0.094	−0.178
PO2-74	554	0.139	−0.227
PO2-73	557	−0.101	−1.590
PO2_30	563	−0.136	−0.772
PO2_31	576	0.717	0.432
PO2_32	585	0.618	0.048
PO2_33	596	−0.037	−0.634
PO2_34	608	0.474	−0.210
PO2_35	620	0.838	0.234
PO2_36	632	1.259	0.465
PO2_37	645	0.693	−0.336
PO2_38	654	0.542	−0.371
PO2_39	666	0.791	0.247
PO2_40	681	0.780	−0.233
PO2_41	696	0.838	−0.220
PO2-72	699	0.184	−0.806
PO2_42	763	0.434	−0.266
PO2_43	772	−0.091	0.213
PO2_44	781	−0.099	−0.996
PO2_45	793	0.371	−0.483
PO2_46	805	0.648	−0.367
PO2_47	814	0.616	0.214
PO2_48	823	0.742	−0.161
PO2_49	832	0.820	−0.313
PO2_50	844	0.486	−0.437
PO2_51	859	1.048	0.384
PO2_52	871	0.863	0.708
PO2_53	880	0.735	−0.502
PO2_54	889	0.427	−0.479
PO2_55	910	0.354	−1.064
PO2_56	923	0.536	−0.038
PO2_57	938	−1.011	−0.964
PO2_58	948	−2.592	−1.902
PO2_59	959	0.110	−1.009
PO2_60	968	−0.300	−1.790
PO2_61	974	0.007	−0.559
PO2-62	985	−1.871	−1.645
PO2_62	994	0.122	−1.006
PO2-63	1,017	−0.855	−1.155
PO2_63	1,033	−0.392	−1.297
PO2-64	1,049	0.613	−0.893
PO2_64	1,082	−0.185	−1.196
PO2-65	1,108	−0.526	−0.739
PO2_65	1,145	0.128	−1.201
PO2-66	1,178	−0.034	−2.032
PO2_66	1,195	−0.305	−1.146

Lead Isotope Compositions.

The labile Pb can be broken down into three well-defined components. In Fig. 2, the ²⁰⁴Pb/²⁰⁶Pb vs. ²⁰⁸Pb/²⁰⁶Pb vs. ²⁰⁷Pb/²⁰⁶Pb 3D plot shows an apparently ternary mixture between geologically “recent” Pb (components α′ and α″) and “old” Pb (component β). The recent natural Pb is a mixture of volcanic Pb from the Alban Hills (component α′) and sedimentary Pb from the Mediterranean outflow water (component α″ present in local carbonate sediments) well-separated by different ²⁰⁸Pb/²⁰⁶Pb values (2). The exact isotope compositions of the natural components α′ and α″ (Table S3) are somewhat arbitrary, but their specific assignment does not affect the conclusions reached about the order of magnitude and relative variations of the anthropogenic component β (Table S3 lists the Pb isotope composition of β).

Fig. 2.

3D plot of ²⁰⁴Pb/²⁰⁶Pb vs. ²⁰⁸Pb/²⁰⁶Pb vs. ²⁰⁷Pb/²⁰⁶Pb measured on leachates from cores PO2 and PTXI-3. The track of the stippled blue drop lines suggests that the 3D dataset as a whole could be interpreted as a single alignment in ²⁰⁷Pb/²⁰⁶Pb–²⁰⁴Pb/²⁰⁶Pb space and therefore, as a two-component mixture. However, adding ²⁰⁸Pb/²⁰⁶Pb to the perspective shows that lead is a mixture of three separate components: α′, α″, and β. The three red lines connect components α′ and β, α″ and β, and α′ and α″. The α′ and α″ mixing line corresponds to unpolluted Tiber water and is composed of Mediterranean outflow water (α″) (50) and volcanic rocks from the Alban Hills (α′) (51, 52); β is the anthropogenic end member located near the fistulae from Rome (2), Naples (4), and Pompeii (26).

Table S3.

Pb isotope compositions, T_mod (million years), µ, κ, and proportions of the components β, α′, and α″ in the sediment leachates from the PO2 and PTXI-3 cores. Pb isotope compositions, T_mod, μ, and κ of the two natural components α′ and α″ and the anthropogenic component β are listed at the end of the table

Material	Sample name	Depth (cm)	206Pb/204Pb	207Pb/204Pb	208Pb/204Pb	Tmod (Ma)	µ	к	fα′	fα″	fβ
Core PO2 leachate	PO2-86	9	18.567	15.668	38.709	178.2	9.752	3.964	0.17	0.34	0.49
Core PO2 leachate	PO2-85	38	18.732	15.680	38.878	75.9	9.759	3.950	0.19	0.64	0.17
Core PO2 leachate	PO2-84	67	18.777	15.683	38.919	47.5	9.760	3.944	0.17	0.75	0.08
Core PO2 leachate	PO2-83	87	18.735	15.679	38.884	71.9	9.756	3.950	0.22	0.63	0.15
Core PO2 leachate	PO2-1	109	18.723	15.680	38.864	82.0	9.759	3.949	0.16	0.65	0.19
Core PO2 leachate	PO2-2	138	18.728	15.675	38.866	72.2	9.749	3.945	0.18	0.66	0.16
Core PO2 leachate	PO2-3	167	18.731	15.676	38.868	71.6	9.751	3.945	0.17	0.66	0.16
Core PO2 leachate	PO2_4	209	18.732	15.682	38.876	77.6	9.762	3.949	0.17	0.66	0.17
Core PO2 leachate	PO2_5	238	18.675	15.672	38.815	107.3	9.750	3.952	0.18	0.55	0.27
Core PO2 leachate	PO2-82	264	18.740	15.680	38.890	69.7	9.757	3.950	0.22	0.63	0.14
Core PO2 leachate	PO2_6	267	18.741	15.680	38.883	69.9	9.759	3.947	0.17	0.68	0.15
Core PO2 leachate	PO2_7	282	18.741	15.677	38.877	65.5	9.752	3.943	0.16	0.70	0.14
Core PO2 leachate	PO2-81	288	18.673	15.672	38.823	107.4	9.748	3.956	0.25	0.49	0.26
Core PO2 leachate	PO2-8	294	18.582	15.667	38.732	166.5	9.748	3.966	0.24	0.31	0.45
Core PO2 leachate	PO2-80	299	18.656	15.675	38.815	123.7	9.756	3.963	0.27	0.43	0.31
Core PO2 leachate	PO2_9	303	18.626	15.670	38.777	139.2	9.750	3.962	0.23	0.41	0.36
Core PO2 leachate	PO2_10	315	18.633	15.670	38.780	134.1	9.749	3.960	0.22	0.43	0.35
Core PO2 leachate	PO2_11	324	18.604	15.666	38.757	150.5	9.745	3.965	0.27	0.34	0.40
Core PO2 leachate	PO2_12	333	18.657	15.672	38.821	119.2	9.750	3.964	0.32	0.39	0.29
Core PO2 leachate	PO2_13	342	18.654	15.673	38.825	123.0	9.753	3.968	0.35	0.36	0.29
Core PO2 leachate	PO2-79	346	18.723	15.676	38.897	77.6	9.752	3.962	0.38	0.47	0.15
Core PO2 leachate	PO2_14	350	18.785	15.682	38.911	40.9	9.758	3.936	0.09	0.84	0.07
Core PO2 leachate	PO2_15	369	18.761	15.687	39.028	63.6	9.769	3.999	0.85	0.13	0.03
Core PO2 leachate	PO2_16	381	18.721	15.683	38.970	87.6	9.766	3.996	0.75	0.13	0.12
Core PO2 leachate	PO2_17	393	18.674	15.677	38.921	114.1	9.759	3.999	0.76	0.03	0.20
Core PO2 leachate	PO2_18	399	18.636	15.674	38.843	137.0	9.757	3.986	0.53	0.16	0.31
Core PO2 leachate	PO2_19	442	18.702	15.678	38.880	94.6	9.757	3.966	0.38	0.42	0.20
Core PO2 leachate	PO2_20	454	18.716	15.684	38.944	91.9	9.768	3.987	0.62	0.23	0.15
Core PO2 leachate	PO2-78	457	18.669	15.679	38.860	118.7	9.762	3.976	0.42	0.31	0.26
Core PO2 leachate	PO2-77	460	18.595	15.671	38.770	161.8	9.754	3.976	0.35	0.24	0.41
Core PO2 leachate	PO2_21	463	18.574	15.665	38.738	170.2	9.746	3.973	0.32	0.23	0.45
Core PO2 leachate	PO2_22	472	18.517	15.658	38.654	201.8	9.738	3.966	0.19	0.24	0.57
Core PO2 leachate	PO2_23	481	18.482	15.653	38.610	220.6	9.731	3.965	0.16	0.20	0.64
Core PO2 leachate	PO2_24	490	18.568	15.661	38.716	168.6	9.737	3.965	0.26	0.28	0.46
Core PO2 leachate	PO2_25	506	18.651	15.676	38.814	128.6	9.759	3.966	0.27	0.41	0.32
Core PO2 leachate	PO2_26	512	18.641	15.670	38.808	128.9	9.749	3.967	0.34	0.35	0.32
Core PO2 leachate	PO2_27	521	18.687	15.674	38.869	100.6	9.751	3.969	0.42	0.36	0.21
Core PO2 leachate	PO2_28	530	18.675	15.680	38.889	115.9	9.764	3.985	0.55	0.22	0.24
Core PO2 leachate	PO2-76	536	18.585	15.673	38.774	171.8	9.760	3.984	0.41	0.16	0.43
Core PO2 leachate	PO2_29	542	18.570	15.671	38.758	179.6	9.757	3.985	0.41	0.13	0.46
Core PO2 leachate	PO2-75	548	18.632	15.678	38.835	144.2	9.765	3.985	0.47	0.19	0.33
Core PO2 leachate	PO2-74	554	18.710	15.687	38.935	100.1	9.774	3.987	0.58	0.25	0.18
Core PO2 leachate	PO2-73	557	18.751	15.690	38.973	74.7	9.776	3.982	0.56	0.34	0.10
Core PO2 leachate	PO2_30	563	18.745	15.686	38.956	73.7	9.769	3.977	0.52	0.37	0.11
Core PO2 leachate	PO2_31	576	18.764	15.684	38.948	57.8	9.763	3.963	0.40	0.52	0.08
Core PO2 leachate	PO2_32	585	18.787	15.686	38.944	43.8	9.765	3.949	0.24	0.70	0.06
Core PO2 leachate	PO2_33	596	18.803	15.683	38.937	29.2	9.758	3.937	0.14	0.83	0.03
Core PO2 leachate	PO2_34	608	18.791	15.683	38.951	38.0	9.760	3.950	0.28	0.68	0.04
Core PO2 leachate	PO2_35	620	18.801	15.685	38.951	33.0	9.763	3.944	0.21	0.76	0.03
Core PO2 leachate	PO2_36	632	18.803	15.688	38.935	35.4	9.768	3.937	0.08	0.86	0.05
Core PO2 leachate	PO2_37	645	18.783	15.685	38.945	46.0	9.764	3.952	0.27	0.67	0.06
Core PO2 leachate	PO2_38	654	18.780	15.685	38.966	48.1	9.764	3.962	0.40	0.55	0.05
Core PO2 leachate	PO2_39	666	18.791	15.692	38.978	48.3	9.776	3.963	0.37	0.58	0.05
Core PO2 leachate	PO2_40	681	18.783	15.683	38.941	43.8	9.760	3.950	0.27	0.68	0.06
Core PO2 leachate	PO2_41	696	18.800	15.690	38.957	39.5	9.772	3.948	0.21	0.75	0.04
Core PO2 leachate	PO2-72	699	18.754	15.683	38.930	63.8	9.763	3.960	0.35	0.55	0.10
Core PO2 leachate	PO2_42	763	18.780	15.684	38.957	46.0	9.761	3.958	0.37	0.59	0.05
Core PO2 leachate	PO2_43	772	18.780	15.686	38.970	49.0	9.766	3.964	0.42	0.54	0.05
Core PO2 leachate	PO2_44	781	18.776	15.689	38.982	56.2	9.772	3.972	0.49	0.45	0.05
Core PO2 leachate	PO2_45	793	18.785	15.685	38.945	43.8	9.763	3.951	0.27	0.68	0.06
Core PO2 leachate	PO2_46	805	18.789	15.683	38.947	38.5	9.759	3.948	0.27	0.69	0.04
Core PO2 leachate	PO2_47	814	18.793	15.686	38.939	40.5	9.765	3.944	0.18	0.77	0.06
Core PO2 leachate	PO2_48	823	18.780	15.681	38.941	43.4	9.757	3.951	0.29	0.65	0.05
Core PO2 leachate	PO2_49	832	18.782	15.685	38.933	46.5	9.764	3.947	0.21	0.72	0.07
Core PO2 leachate	PO2_50	844	18.791	15.685	38.945	39.9	9.763	3.947	0.23	0.72	0.05
Core PO2 leachate	PO2_51	859	18.808	15.680	38.920	21.4	9.751	3.927	0.05	0.93	0.03
Core PO2 leachate	PO2_52	871	18.747	15.663	38.883	44.4	9.725	3.940	0.27	0.65	0.08
Core PO2 leachate	PO2_53	880	18.783	15.683	38.940	43.3	9.760	3.949	0.26	0.68	0.06
Core PO2 leachate	PO2_54	889	18.768	15.683	38.936	54.4	9.762	3.955	0.31	0.61	0.08
Core PO2 leachate	PO2_55	910	18.789	15.688	38.955	44.7	9.769	3.953	0.28	0.67	0.05
Core PO2 leachate	PO2_56	923	18.769	15.685	38.975	56.0	9.766	3.972	0.51	0.43	0.06
Core PO2 leachate	PO2_57	938	18.771	15.685	39.000	53.7	9.764	3.981	0.65	0.32	0.03
Core PO2 leachate	PO2_58	948	18.771	15.686	39.010	56.5	9.768	3.986	0.69	0.28	0.03
Core PO2 leachate	PO2_59	959	18.786	15.687	38.991	45.4	9.766	3.970	0.51	0.47	0.02
Core PO2 leachate	PO2_60	968	18.784	15.685	38.991	44.8	9.764	3.970	0.53	0.45	0.02
Core PO2 leachate	PO2_61	974	18.785	15.685	38.986	43.7	9.763	3.968	0.50	0.48	0.02
Core PO2 leachate	PO2-62	985	18.772	15.684	38.987	52.4	9.763	3.975	0.58	0.38	0.04
Core PO2 leachate	PO2_62	994	18.792	15.688	38.989	42.3	9.768	3.966	0.45	0.53	0.02
Core PO2 leachate	PO2-63	1,017	18.779	15.683	38.984	45.5	9.759	3.970	0.54	0.44	0.02
Core PO2 leachate	PO2_63	1,033	18.781	15.684	38.987	46.4	9.763	3.970	0.53	0.44	0.03
Core PO2 leachate	PO2-64	1,049	18.816	15.684	38.954	21.1	9.760	3.937	0.15	0.84	0.00
Core PO2 leachate	PO2_64	1,082	18.779	15.683	38.989	46.3	9.760	3.972	0.56	0.42	0.02
Core PO2 leachate	PO2-65	1,108	18.784	15.687	38.998	47.0	9.767	3.974	0.56	0.42	0.02
Core PO2 leachate	PO2_65	1,145	18.797	15.687	38.992	37.7	9.766	3.964	0.45	0.54	0.01
Core PO2 leachate	PO2-66	1,178	18.793	15.684	38.985	36.7	9.760	3.963	0.46	0.53	0.01
Core PO2 leachate	PO2_66	1,195	18.780	15.684	38.994	46.5	9.762	3.974	0.57	0.41	0.02
Core PTXI-3 leachate	PTXI-3_12	136	18.630	15.670	38.801	136.2	9.749	3.970	0.36	0.31	0.34
Core PTXI-3 leachate	PTXI-3_5	169	18.615	15.670	38.780	146.5	9.750	3.969	0.31	0.31	0.37
Core PTXI-3 leachate	PTXI-3_10	362	18.472	15.656	38.599	231.4	9.739	3.966	0.13	0.20	0.68
Core PTXI-3 leachate	PTXI-3_3	382	18.414	15.655	38.557	271.5	9.744	3.980	0.19	0.02	0.79
Core PTXI-3 leachate	PTXI-3_8	495	18.774	15.681	38.982	47.3	9.757	3.971	0.56	0.41	0.03
Core PTXI-3 leachate	PTXI-3_9	576	18.776	15.686	38.996	51.4	9.766	3.977	0.59	0.38	0.03
Core PTXI-3 leachate	PTXI-3_11	611	18.765	15.683	38.991	55.7	9.761	3.980	0.64	0.32	0.04
Core PTXI-3 leachate	PTXI-3_7	633	18.790	15.684	38.978	39.5	9.762	3.961	0.43	0.55	0.02
Core PTXI-3 leachate	PTXI-3_2	657	18.775	15.686	38.996	52.2	9.766	3.978	0.60	0.37	0.03
Core PTXI-3 leachate	PTXI-3_6	731	18.773	15.683	38.929	50.2	9.761	3.949	0.25	0.67	0.08
Core PTXI-3 leachate	PTXI-3_1	736	18.781	15.685	38.989	47.4	9.764	3.971	0.53	0.44	0.03
Core PTXI-3 leachate	PTXI-3_4	880	18.784	15.684	38.986	44.0	9.762	3.968	0.51	0.47	0.02
End member α′	—	—	18.762	15.685	39.053	60.4	9.764	4.008	1.00	0.00	0.00
End member α″	—	—	18.832	15.687	38.945	13.3	9.764	3.926	0.00	1.00	0.00
End member β	—	—	18.315	15.641	38.416	323.7	9.727	3.969	0.00	0.00	1.00

The anthropogenic origin of the old Pb that dominates subunit B2 (Figs. 3 and 4) can be shown by converting the Pb isotope compositions into Pb model ages T_mod, ²³⁸U/²⁰⁴Pb (µ), and ²³²Th/²³⁸U (κ) ratios (Table S3) using the equations given by Albarède et al. (16). The advantages of this representation over that based on raw Pb isotope ratios have been shown in a number of geological and (geo)archeological contexts (2–4, 16–19). In the present case, Peninsular Italy is a geologically young mountain range (T_mod < 30 Ma); the presence of Hercynian Pb (T_mod > 200 Ma) in Ostia sediments, therefore, unambiguously signals contamination of local waters by foreign Pb, likely lead artifacts (Figs. 2 and 3). Quantitative breakdown of Pb isotopes into natural and anthropogenic components as proposed by Delile et al. (2) essentially reproduces the contamination patterns visible in the T_mod record of Ostia.

Fig. 3.

Plot of T_mod vs. κ for leachates from cores PO2 (unit C in green, subunit B2 in yellow, subunit B1 in purple, and unit A in blue), PTXI-3 (Claudian harbor), TR14 (Trajanic harbor), and CN1 (Canale Romano deposits) and fistulae from Rome, Naples, and Pompeii (2, 4, 26) (the same plot for the raw Pb isotope ratios is in Fig. S3B).

Fig. 4.

Down core variations of EF_Pb, T_mod, f_β, µ, and κ. The red dotted line marks the first appearance of anthropogenic Pb pollution recorded in core PO2 from around 200 BC according to the age-depth model. The gray bands show the main declining trends of anthropogenic lead pollution discussed in this study (the same plot for the raw Pb isotope ratios is in Fig. S3A).

To characterize the evolution of the anthropogenic Pb signal, we use another pollution index f_β, which is based on the proportion of the anthropogenic β component in the Ostia sediments (equation details are in Materials and Methods). Beyond some minor trends, the EF_Pb, T_mod, f_β (Fig. 4), and raw Pb isotope ratio records (Fig. S3A) largely correlate, showing that the anthropogenic component does not fluctuate randomly but displays robust peaks and troughs.

Fig. S3.

(A) Down core variations of ²⁰⁶Pb/²⁰⁷Pb, ²⁰⁸Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, and ²⁰⁶Pb/²⁰⁴Pb. The red dotted line marks the first appearance of anthropogenic Pb pollution recorded in core PO2 from around 200 BC according to the age-depth model. The gray bands show the main declining trends of anthropogenic lead pollution discussed in this study. (B) Plots of ²⁰⁴Pb/²⁰⁶Pb vs. ²⁰⁸Pb/²⁰⁶Pb for leachates from cores PO2 (unit C in green, subunit B2 in yellow, subunit B1 in purple, unit A in blue), PTXI-3 (Claudian harbor), TR14 (Trajanic harbor), and CN1 (Canale Romano deposits) and fistulae from Rome, Naples, and Pompeii (2, 4, 26).

Discussion

Natural and Anthropogenic Pb Sources.

In Figs. 2 and 3, the preharbor and early harbor samples from Ostia (core PO2) and Portus (cores TR14, CN1, and PTXI-3) plot along the mixing line between the two local subcomponents α′ and α″. The sandy preharbor sediments (unit A of core PO2 and most of core PTXI-3) show consistently higher values of ²⁰⁸Pb/²⁰⁶Pb (Fig. S3B) and κ (Fig. 3) than the harbor sediments highly dominated by clays and silts (harbor subunit B1). This can be explained by changes in mineral sorting processes resulting from varying hydrodynamic levels. According to Garçon et al. (20), some heavy minerals concentrated in coarse sediments are extremely radiogenic, allowing sandy sediments to reach higher ²⁰⁸Pb/²⁰⁶Pb values (20) and thus, higher values of κ (r of ²⁰⁸Pb/²⁰⁶Pb vs. κ ∼ 0.7).

The intersection of two additional mixing lines—between α′ and α″ and the third anthropogenic component β—reveals that β is Hercynian in age (T_mod ∼ 325 Ma) (Fig. 3) and characterized by high ²⁰⁴Pb/²⁰⁶Pb (∼0.0546) (Fig. S3B). Sediments located on the mixing line α′-β correspond to Pb-contaminated particles that were transported in turbulent conditions, explaining their presence exclusively in the sandy subunit B2. In contrast, sediments falling along the α″-β trend were deposited under quieter hydrodynamic conditions, typically the silty floodplain deposits of unit C. Nevertheless, the whole of the Portus (cores TR14, CN1, and PTXI-3) and Ostia (core PO2) sediments (n = 177) affected by lead pollution converges toward the same distinct source, or mix of sources, of imported lead.

In Roman times, a substantial number of ships’ hulls were sheathed with large millimeter-thick lead plates to protect their submerged portions against fouling and corrosion (21). Lead plates, anchors, and sounding lead weights are, however, unlikely to have been significant contributors to the anthropogenic Pb signal of Ostia sediments. The reasons for this are several. First, in seawater, lead passivation by the deposition of a film of chlorides and carbonates is fast and efficient (22). Second, because the earliest wreck with a hull sheathed in lead discovered so far dates to the midfourth century BC (23) and the practice had reached a peak by the end of the fourth century BC (21), if the sheathing of ship’s hulls was a significant lead pollution source, we would expect an excess of lead to be recorded in the functional harbor unit B1, which covers the fourth and third centuries BC. Such lead pollution does not occur, however, until the harbor unit B2, which was deposited at the end of the use of the harbor (11, 15), from around the second century BC. In the same way, the practice of lead sheathing of ship’s hulls was no longer significant from the middle of the first century AD (21), while Pb pollution of Rome’s harbor water column lasts until the ninth century AD (2). Third, and most importantly, the anthropogenic component β is clearly homogenous (Figs. 2 and 3) (2). This is inconsistent with local contamination by vessels covered in lead plates of a variety of origins but consistent with distant sources well-mixed during the journey from Rome down the Tiber. Remarkably, the isotope abundances of component β match those of fistulae from the Roman urban water supply system (Figs. 2 and 3). The persistence of contamination for well over a millennium (200 BC to AD 800) and the uniqueness of its isotopic composition argue against random pollution. They rather suggest that a mechanism existed upstream from Ostia that efficiently mixed the isotopic compositions of all sources of anthropogenic Pb contributing to the anthropogenic signal so visible in the harbor sediments. Judging from the 20th century mean flow (230 m³/s) (24), around 3% (7 m³/s) of the Tiber’s water was running through Rome’s aqueducts at the peak of the Roman Empire (8, 25), a significant fraction of which passed through fistulae. Lead from Rome fistulae, therefore, seems the only acceptable source of contamination in Ostia sediments.

It is not clear whether the Pb ores derive from one or more mining districts, but recycling and salvaging practices of lead implemented by Romans (6, 10, 26) favor a mixture from several Hercynian sources (Germany, Spain, England, France). Such recycling would have contributed significantly to the stability of the isotopic signature of the imported Pb component over the long period observed. Additionally, the conspicuous isotopic stability, in turn, argues for long-term stability in the lead trade networks between Rome and the western half of its empire.

Control Factors of Pb Content.

Changes in two main processes control the variation in the Pb contents of the PO2 core: (i) the input of Pb in the hydrological system upstream of the core and (ii) the transport of that Pb to the PO2 site. The major change in Pb inputs is the construction of the piped water distribution system in Rome. The imported Pb of Rome’s pipes (with older Hercynian T_mod values) is progressively dissolved and transported down the Tiber to be deposited in the B2 and C units, except during disruptions to flow in the piped system occasioned by damage or neglect during political unrest, epidemics, natural disasters, etc. F1 and F4 correlate with two different transport processes: flooding by contaminated river water and dilution by uncontaminated seawater.

The correlation of EF_Pb with particle size as well as with the magnetic susceptibility and F1 shows that Pb excesses in the bulk sediment trapped in the harbor basin of Ostia were associated with Tiber flooding episodes that acted as a transport vector of Pb pollution. In other words, a period of increased river discharge (e.g., flooding) resulted in deposition of larger amounts of larger particles and hence, deposition of more Pb.

In subunit B2, imported Pb is present in the labile rather than the bulk fraction. It is correlated with F4 (r = 0.6) but not with F1 (r = 0.09). F4 is dominated by heavy metals, such as Sn and Cd, which, like T_mod, signal anthropogenic influence and which are opposed to Na (Fig. S2). Between 250 and 500 cm, several striking drops in Pb contamination levels occur (²⁰⁶Pb/²⁰⁷Pb in Fig. S3A), which consistently are accompanied by steep marine (F4) peaks (Fig. 1). It, therefore, seems that sudden, uncontaminated seawater inputs into the harbor basin invariably resulted in a decrease in Pb pollution, probably because of a dilution effect. Dilution by an uncontaminated source of water is commonly observed in both ancient harbors [for example, Sidon (27)] and fluvial environments [such as the Ruhr River, a tributary to the Rhine (28), the Caima River (Portugal) (29), and the Belle Fourche River (United States) (30)].

A Chronology of Lead Pollution and Roman Urbanism.

Fig. S4 reports errors on calendar ages from 200 BC to AD 300 using the values by Reimer et al. (31) and shows that ¹⁴C chronology is particularly imprecise for the time interval covering the Roman Republic and Empire. For a given value of ¹⁴C age, the range and, occasionally, the multiplicity of possible calendar ages in this interval are large, and uncertainties cannot be unambiguously represented by a single-error interval. Only those Pb isotope fluctuations that lasted in excess of age uncertainties are discussed here. With this caveat in mind, four main periods can nevertheless be distinguished from Fig. 4.

Fig. S4.

Illustration of the uncertainties on the ¹⁴C dates with ±2 sigma uncertainty (or 95% confidence) intervals. B.P. ages, calendar years, and uncertainties are from the IntCal13 radiocarbon age calibration (31). The white bars emphasize how large the uncertainties can be for some dates. Because the ¹⁴C age vs. calendar age curve is not single-valued everywhere (one x value may correspond to more than one y value), the magnitude of the 2 sigma (95% confidence) intervals on calendar ages varies and, occasionally, may break into discontinuous segments.

Uncontaminated environment (eighth/ninth to second century BC).

From the bottom of the core to −563 cm (units A and B1), anthropogenic lead is not visible, even in harbor subunit B1, which dates from the foundation of Ostia in the fourth or early third century BC (32) to the beginning of the second century BC. Fig. 4 shows that contamination is not detected until the second century BC. This underscores once more that Pb pollution derives primarily from the dissolution of fistulae (Figs. 2 and 3). Lead fistulae would not be expected at Ostia in this period, since Ostia relied on wells for its water supply until the first half of the first century AD (33). The water of the Aqua Appia and Anio Vetus aqueducts at Rome, built in the late fourth century and early third century BC, respectively, were distributed by a lead-free system of masonry channels or terracotta or wooden pipes, of which only few have been found. It seems likely that these aqueducts supplied only a small number of focal points in the city, perhaps centrally located public fountains. Such a minimalist system was a far cry from that of Imperial Rome, which supplied hundreds of baths and private residences through a complex network of fistulae (10).

Manifestation of anthropogenic Pb (basal part of subunit B2).

The next period (Fig. 4 and Fig. S3) is characterized by the first rise of anthropogenic Pb excesses in the sandy harbor subunit B2. The age-depth models of cores PO2 (Fig. 1) and PTXI-3 (Fig. S5 and Table S1) both suggest that Pb contamination began around the second century BC.

Fig. S5.

Stratigraphic log of core PTXI-3 showing its stratigraphic description, ¹⁴C dates, the age-depth model of the core constructed with the Clam software (49), EF_Pb, f_β, the geological parameters, and the Pb isotopic compositions.

The ¹⁴C age-depth model receives some confirmation from scattered mentions in textual sources. The first dated stamps on lead pipes do not appear at Rome until 11 BC (10) and even later at Ostia (AD 37–41) (33–35). Fistulae seem to have been in use in water systems at the beginning of the first century BC in Rome (36), however, and by the late second century BC at nearby Alatri (10). Rome’s system of public water basins and private connections was probably in operation by 184 BC, but lead pipes are not mentioned (9, 37). Already, Cato the Elder (234–149 BC) (38) used the word fistula as meaning “pipe.”

The high-resolution Pb isotopic characterization of core PO2 provides continuous and detailed insight into water supply and urbanization upstream. After the first appearance of anthropogenic Pb in the harbor sediments of Ostia at −563 cm, the trends of decreasing ²⁰⁶Pb/²⁰⁷Pb, ²⁰⁸Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁶Pb/²⁰⁴Pb (Fig. S3A), and µ (Fig. 4) with time as well as increasing EF_Pb, T_mod, and f_β (Fig. 4) argue for an overall increase of the imported Pb component. Like at other Roman cities (4, 34), this trend reflects the increase in the geographic extent and/or density of the lead pipe system as a result of urban development. At Rome, it is consistent with the repair and expansion of the water supply system that accompanied the commissioning of the Aqua Marcia (late 140s BC) and Aqua Tepula (125 BC) aqueducts (6).

Then, a sudden drop in EF_Pb, T_mod, and f_β and a sharp spike in ²⁰⁶Pb/²⁰⁷Pb, ²⁰⁸Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁶Pb/²⁰⁴Pb, and µ take place between −536- and −506-cm core depth (bottom gray bands in Fig. 4 and Fig. S3A), corresponding to the first century BC or early first century AD. After the Aqua Tepula, a century of rampant unrest and outright civil war prevented aqueduct construction and hampered maintenance (6). The Ostia PO2 core thus provides the first evidence of the scale of the contemporaneous reduction in flows in Rome’s lead pipe distribution system—of the order of 50%—resulting in decreased inputs of lead-contaminated water into the Tiber. Octavian’s (Augustus’s) progressive defeat of his rivals during the 30s BC allowed his future son-in-law, Agrippa, to take control of Rome’s water supply by 33 BC (6). Over the next 30 years, they repaired and extended the existing aqueduct and fistulae system as well as built an unprecedented three new aqueducts (6, 10), leading to renewed increase in Pb pollution of the Tiber river. This sequence of events is recorded as a sharp decrease in ²⁰⁶Pb/²⁰⁷Pb, ²⁰⁸Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁶Pb/²⁰⁴Pb (Fig. S3A), and µ (Fig. 4) accompanied by a major spike in EF_Pb, T_mod, and f_β (Fig. 4) between −506- and −472-cm core depth. While the correlation between politics, stability, and lead isotopic composition is clear, the precise mechanism behind this correlation is undoubtedly complex and requires additional study.

Variability in pollution level (late first century BC/early first century AD to post-AD 250).

The next chapter in the PO2 core stretches from −472 to −294 cm, covering the Roman Imperial period. The end of this phase falls outside the dated portion of the core. The overall trend shows continuing but declining Pb pollution from a peak around the beginning of this period. This continued Pb pollution is consistent with the Pb isotopic compositions of sediments deposited during the High Roman Empire in the Claudian (core PTXI-3) (Fig. S5) and Trajanic basins (core TR14) (2) at Portus as well as the canal leading to them (core CN1) (2), where T_mod values are ∼125–200 Ma and κ values are ∼3.96–3.98 (2), both indicative of imported lead. These sustained levels of anthropogenic Pb in multiple cores confirm the picture from textual and archeological data that the aqueduct and lead pipe distribution system was generally maintained until at least the midthird century AD (10, 39). Reduced dissolution of Pb because of insulation of the pipes by a coating of a limestone deposit known as travertine (8) is unlikely to be a factor in this decline. Coating of many ancient aqueducts, cisterns, fountains, and pools, including those at Rome and Pompeii, by centimeter-thick travertine deposits takes place as a result of carbon dioxide degassing and warming (40). Aqueducts, cisterns, fountains, and pools are well-ventilated, but in a pressurized piped system, like the ones that distributed water at Rome and Ostia, the water fills the entire pipe, inhibiting degassing and travertine deposition. Pipe damage—mentioned by two legal sources at Rome (6, 41)—probably led to regular replacement, thereby exposing fresh, uncoated lead for dissolution. The relative lack of travertine in pressurized pipes is borne out by our field observations at Rome (2) and around the Bay of Naples (42, 43), where almost all parts of the system are coated in travertine, except the pressurized pipes. Likewise, at Patara in Turkey (40), the pressurized inverted siphon pipes have much thinner travertine deposits than surrounding areas of the aqueduct. The travertine deposits from Pompeii (42) and the nymphaeum of Trajan at Ephesus (44) show elevated levels of lead that was clearly derived from the lead pipes and fittings, despite the widespread presence of travertine deposits in the systems. At Ostia, the situation is more complex: early lead contamination of travertine disappeared. The authors posit a change in water source (45).

The smooth declining trend in core PO2 is punctuated by two additional dips in Pb pollution, the second and third in the core (middle and top gray bands in Fig. 4 and Fig. S3A, respectively). The second decrease occurs from −470 to −440 cm (sometime between the late first century BC and the first century AD), and the third occurs from −370 to −350 cm (sometime from the first century to the early third century AD). Instability of the Tiber’s flow regime is excluded as a cause for these drops because of the lack of correlation between f_β and F1 (r = 0.2 in subunit B2). These decreases coincide with two strong marine (F4) peaks (Fig. 1), suggesting that the most likely cause is the diversion of the contaminated Tiber water away from Ostia’s harbor. Manmade links between the Tiber and the sea upstream of Ostia were constructed in this period: the northern canal, the Canale Romano, and the Claudian basin at Portus (midfirst century AD) followed by the Trajanic basin and probably, the northeastern canal (early second century AD) (46). These reductions in Tiber outflow would have initially and mechanically (i) brought less anthropogenic Pb and (ii) allowed more uncontaminated seawater to enter the harbor of Ostia, diluting the reduced Tiber-derived Pb. This would also explain why clear evidence of these anthropogenic Pb drops is not seen in the TR14 and CN1 cores (2). We cannot rule out, however, that the second and third drops in Pb pollution may be caused by damage or neglect of the water distribution system.

After the second drop (−470 to −440 cm), the renewed rise in Pb excesses shows that the Pb concentration in the water recovered to a level only slightly below the previous peak, while the values of f_β recovered to levels consistent with imported lead being the source (Fig. 4). After the third drop passed (∼−350 cm), the lead pollution in the harbor water column returned to approximately two-thirds (f_β ∼ 0.3–0.5) (Fig. 4) of that of the major spike, with a surge of older Pb in leachates (∼125–175 Ma) and a dramatic drop in the ²⁰⁶Pb/²⁰⁷Pb, ²⁰⁸Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁶Pb/²⁰⁴Pb (Fig. S3A), and µ (Fig. 4) derived from leaching of the fistulae. This last and durable Pb pollution phase recorded in the upper part of the PO2 core shows that maintenance of a somewhat reduced water system continued.

Decrease of the pollution level (post-AD 250).

The last period, postdating AD 250, is characterized by a decrease in imported lead (f_β ∼ 0.2 in unit C) (Fig. 4). This decrease is also observed at the same time in the TR14 and CN1 cores at Portus (2) and is not associated with a strong period of marine influence (F4) (Fig. 1). It thus represents a contraction of the effective water distribution system at Rome, likely related to the increase in instability beginning in the third century AD. Indeed, after the midthird century AD, no more aqueducts were built, and maintenance was on a smaller scale, while no pipe stamps can be dated to the period from the midthird to midfourth century AD (10). This period of receding Pb contamination corresponds to the apparent decline of Pb and Ag mining (47) and of overall economic activity in the Roman Empire (48).

Materials and Methods

Major and trace element concentrations were determined by complete digestion of the bulk sediment, whereas Pb isotope compositions were measured only on the most labile fraction to emphasize the anthropogenic contribution. A large selection of major and trace element concentrations was measured for the 12-m-long Ostia core PO2 every 14 cm (86 samples) (Dataset S1), while only Pb concentrations were analyzed for 12 samples from core PTXI-3. All samples from both cores were additionally measured for their Pb isotope compositions. SI Text has analytical details.

The proportion f_β of the β component in the leachate is calculated by least squares using three sets of equations: for example,

$${\left(\frac{{}^{204}\text{Pb}}{{}^{206}\text{Pb}}\right)}_{\mathrm{\alpha }\prime f\mathrm{\alpha }\prime }+{\left(\frac{{}^{204}\text{Pb}}{{}^{206}\text{Pb}}\right)}_{\mathrm{\alpha }″f\mathrm{\alpha }″}+{\left(\frac{{}^{204}\text{Pb}}{{}^{206}\text{Pb}}\right)}_{\mathrm{\beta }f\mathrm{\beta }}={\left(\frac{{}^{204}\text{Pb}}{{}^{206}\text{Pb}}\right)}_{\text{leach}},$$

with similar equations for the ²⁰⁷Pb/²⁰⁶Pb and ²⁰⁸Pb/²⁰⁶Pb ratios. In these equations, f is the proportion of ²⁰⁶Pb assigned to each component. A closure equation ensuring that all of the f values sum to one must be added.

SI Text

Major and Trace Element Concentrations.

After sieving at 63 µm, aliquots of 100 mg sediment were weighed into screw-top Savillex beakers and dissolved in a clean laboratory in laminar flow hoods using a 3:1:0.5 mixture of concentrated double-distilled HF, HNO₃, and HClO₄. The samples were left to attack at 120 °C to 130 °C for 48 h and then evaporated to dryness. Perchlorates were converted to chlorides by drying down with 6 M distilled HCl. The samples were redissolved in 2 mL concentrated distilled HNO₃, from which ∼10% aliquots were further diluted to 2% HNO₃ and internal standards [10 ppm Sc for inductively coupled plasma atomic emission spectroscopy (ICP-AES) and 2 ppb In for quadrupole inductively coupled plasma mass spectrometry (Q-ICP-MS)] were added. Major elements were analyzed by ICP-AES (ICAP 6000) and trace elements were analyzed by Q-ICP-MS (Agilent 7500 CX) at the Ecole Normale Supérieure de Lyon. The upper limit of blank contribution is negligible for major elements and <2% of the sample content for trace elements. The data are listed in Dataset S1.

Lead Isotope Compositions.

The same method as described in Delile (13) and Delile et al. (2–4) was used. After sieving at 63 µm, representative aliquots of 500 mg sediment were weighed out into screw-top Savillex beakers and leached with hot Suprapur chloroform to separate the labile, or anthropogenic, component of the Pb. A second leaching step was done with hot dilute double-distilled HBr. The two leachates were combined and evaporated to dryness. Lead from the leachates was then separated on anion exchange columns using distilled 1 M HBr to elute the sample matrix and distilled 6 M HCl to elute the Pb. The sample residues after leaching were not analyzed in this work because the isotopic composition of the natural Pb background had already been determined previously (2, 13). Lead isotope compositions were measured by multicollector ICP-MS (Nu Plasma 500 HR) at the Ecole Normale Supérieure de Lyon using thallium doping, sample standard bracketing (53, 54), and the values for National Institute of Standards and Technology (NIST) standard 981 of Eisele et al. (55). The total procedural Pb blank was <20 pg. The external reproducibility of the reported Pb isotope ratios as estimated from the repeated (every two samples) runs of NIST 981 is 100–200 ppm (or 0.01–0.02%) for ratios based on 204 (²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁸Pb/²⁰⁴Pb) and 50 ppm (or 0.005%) for ²⁰⁷Pb/²⁰⁶Pb, ²⁰⁸Pb/²⁰⁶Pb, and ²⁰⁷Pb/²⁰⁸Pb. The Pb isotope data are listed in Table S3.