TMS ONLINE | TMS PUBLICATIONS | SITE MAP An Article from the May 2003JOM: A Hypertext-Enhanced Article

VISIT THE JOM COVER GALLERY JOM MENU

Feature: Archaeotechnology

Using Radiocarbon Dating to Establish the Age of Iron-Based Artifacts

Andrea C. Cook, John R. Southon, and Jeffrey Wadsworth

---

Dawn Ueda and Tom Brown next to the accelerator mass spectrometer at Lawrence Livermore Laboratory. (Photo courtesy of the Center for Accelerator Mass Spectrometry at Lawrence Livermore National Laboratory.)

Over the last 40 years, there has been a discernible increase in the number of scholars who have focused their research on early industrial organizations, a field of study that has come to be known as Archaeotechnology. Archaeologists have conducted fieldwork geared to the study of ancient technologies in a cultural context and have drawn on the laboratory analyses developed by materials scientists as one portion of their interpretive program. Papers for this department are solicited and/or reviewed by Michael Notis, a professor and director of the Archaeometallurgy Laboratory (www.Lehigh.edu/~inarcmet) at Lehigh University.

Figure a. Corroded iron from the Java Sea Wreck.

Figure b. Chinese Warring States arrowhead dating to about 400–200 B.C. Dawn Ueda and Tom Brown next to the accelerator mass spectrometer at Lawrence Livermore Laboratory. (Photo courtesy of the Center for Accelerator Mass Spectrometry at Lawrence Livermore National Laboratory.)

Figure c. A wrought-iron Roman cleaver.

Figure d. Large spear from Burkino Faso, Africa.

Figure e. Paperweight made by reworking iron from the Himeji Castle in Japan.

In this article, an overview is presented of the status of the radiocarbon dating of iron-based materials. Recent advances include simplification in sample preparation and reduction in sample size for accelerator mass spectrometry measurements, and the potential use of rust as a viable source of material for radiocarbon dating. Additionally, a summary is presented of all 63 previously published results for iron-based materials and 29 new results that have not been published previously. These materials range from low-carbon wrought irons to medium to very high-carbon steels and cast irons. Artifact dates range from several hundred years ago to several thousand years ago. Brief descriptions are given of some of these examined samples to illustrate issues and complexities that can arise in determining the age of iron-based carbon materials using radiocarbon dating.

INTRODUCTION

As a society, we are deeply interested in determining the age of things, from the history of our own universe to the details of how modern technological developments have taken place. A recent summary has been published¹ of techniques for dating that range from astronomical methods to cover time scales from the age of the universe (e.g., 13 billion years ago) to forensic entomology to determine when people died (e.g., times on the order of hours).

One well-known method for dating is based on the use of isotopic techniques. Included are reactions such as the uranium-to-lead transformation utilized for dates that range from 1 billion years to 4.5 billion years. Perhaps the best-known isotopic technique, however, is that of radiocarbon [e.g., carbon 14 (¹⁴C)] dating, which is used to cover time periods from several hundred years ago to about 50,000 years ago.

The present paper deals with an issue of great interest to materials scientists and archeologists—the dating of iron-based materials that contain carbon. The phrase “iron-based materials” is used to cover the three common groups of irons and steels: wrought irons, which are typically low carbon (e.g., less than about 0.05% carbon), steels (up to 2.1% carbon), and cast irons (over 2.1% carbon). In addition, however, the corrosion products or rust from these materials is included since they can also be used for dating in some cases. For the case of iron-based materials, the time span of interest is from the start of the Iron Age in the regions of interest (about 2000 B.C. or earlier) to several hundred years ago. The most appropriate method for this time span and group of materials is ¹⁴C dating. It is key to point out that the usefulness of the method of dating carbon in iron-based materials relies on the source of the carbon found in the materials (see sidebar).

RADIOCARBON DATING IRON-BASED MATERIALS

Techniques and Their Limitations

The concept of using radiocarbon dating to determine the age of carbon-containing materials was first proposed in the 1950s. For the case of iron-based materials, van der Merwe and Stuiver² first demonstrated that it was feasible to extract the carbon from different iron-based materials and use it to establish their age using radiocarbon dating. A total of 15 samples of iron-based materials were dated by beta counting at Yale University^2,3 using a dependable method to extract carbon from iron utilizing flow-through combustion in oxygen with cryogenic trapping of CO₂. These studies showed that in a wide range of cases, the carbon in iron-based materials could be extracted and reliably radiocarbon dated.

The Yale beta counter, however, required significant amounts of carbon compared to the amounts that were usually available from artifacts without consuming or damaging them. The amount of carbon required was 1g, equivalent to 50 g of a 2.0% carbon steel or cast iron, or 1,000 g of a 0.1% carbon iron, assuming 100% yields in the experimental process of extracting the carbon.

In the late 1980s, radiocarbon dating by accelerator mass spectrometry (AMS) became common. This new methodology required only 1 mg instead of 1 g of carbon. Cresswell⁴ miniaturized van der Merwe’s extraction technique and dated 12 different iron artifacts.^4–8 Sample sizes in these studies ranged from 3.4 g for an iron bloom (0.4% carbon) to 274 mg for a high-carbon (1.79% carbon) wootz steel.

In 2001, the present authors published⁹ a new carbon-extraction method for iron based on a sealed-tube combustion with CuO in quartz. This greatly simplified the previous technique and required only materials readily available in the standard AMS graphite-preparation laboratory: quartz tubes, CuO, vacuum lines, and a standard electric furnace capable of reaching 1,000°C. Unlike the previous techniques, no exotic gas-trapping equipment is required.

Thus, over the years, the sample-size requirement has been greatly reduced and the carbon-extraction procedure has been simplified. However, as has been mentioned, for a radiocarbon date on iron to be meaningful, the carbon extracted from the iron-based material must be from biomass contemporaneous with original manufacture. In addition to fossil fuels such as coal and coke, other carbon sources such as geological carbonates (e.g., limestone and siderite), shell, or old wood (which are all depleted in ¹⁴C) will cause artifacts to appear to be older than they are. Complications arising from the recycling of artifacts must also be considered. These limitations of the dating technique have been well summarized by van der Merwe³ and Cresswell.⁵

Radiocarbon Dating of Rust

A second interesting area concerns the use of rust for dating. If rust can be dated reliably, it opens up a large number of possibilities for dating iron artifacts. Investigators will not need to cut into valuable artifacts for clean metal, but will be able to use surface corrosion products. This potentially opens the way for dating precious samples such as the iron plate found in the Great Pyramid at Gizeh,^10,11 now at the British Museum. It may also be possible to date completely rusted artifacts, commonly found in waterlogged early Iron-Age sites in Europe and in underwater shipwrecks. Previous investigators had been careful to remove rust from iron prior to dating for fear that it adds contamination. A key issue though, is whether any of the original carbon remains within the matrix of rust and other corrosion products. If not, rust and similar materials are clearly of no interest for radiocarbon dating and should probably be removed since, at best, they can do no good. However, if original carbon is present, the corrosion products themselves may be appropriate targets for dating, subject to solving the potential contamination problems.

Most of the carbon in iron-based materials is in the form of the orthorhombic, crystalline iron carbide (Fe₃C) known as cementite. Morphologically, cementite appears either as spheroidized particles or as pearlite. For compositions exceeding the eutectoid level of about 0.8% carbon, the excess carbide often exists as massive plates of proeutectoid cementite. The thickness and sizes of all of these carbides can vary enormously, depending upon composition and heat-treatment history. For steels that have been quenched to form martensite (body-centered tetragonal structure), the carbon is essentially in solid solution in the iron up to the eutectoid composition, beyond which it too will usually be in the form of carbides.

Despite the complex range of possible amounts and morphologies of the cementite, the thermodynamic stability of iron carbide is significantly greater than that of iron. So, as iron rusts, the carbide phase will be more stable than the matrix and will remain behind. The question then becomes one of kinetics: How long will it take for the carbide to oxidize compared to the iron matrix? As long as the carbon remains in the rust, in whatever form, it will potentially be available for radiocarbon dating.

Although little appears to have been published on this subject, Knox¹² reported the detection of iron carbide in the remaining oxide from a corroded 2,800-year-old Iranian steel dagger.

Optical metallography explicitly showed the presence of “unaltered” iron carbides in the rust from this object. Also present in the microstructure were pseudomorphs of carbides, which had been reduced by corrosion to carbonaceous regions that were “probably an intimate mixture of oxide and amorphous carbon.” More recently, Notis¹³ has succeeded in making carbon maps in rusted old steel samples using electronprobe microanalysis techniques at Lehigh University and has observed good carbon images in the microstructures.

The present authors and van der Merwe¹⁴ have recently completed a study in this area. This work provides some evidence for the reliability of dating corrosion products from artifacts that have rusted in the air, in the ground, and under water, although it does not prove that all such samples can be successfully dated. Nonetheless, iron samples that had completely rusted produced plausible radiocarbon dates, but issues of contamination and post-depositional carbon exchange must be thoroughly tested in a variety of field settings before rust dating can be considered a validated technique. The present authors’ results indicate, however, that in at least some circumstances, the original carbon from iron-based materials is retained in rust and can be cleanly extracted and dated (see Figure 1a and Figure a). What the study does show, then, is that there is no a priori reason why the method should not work on rust. The work suggests that accurate radiocarbon dates may be obtainable with minimal material and with minimal risk to artifacts.

a		b
Figure 1. (a) Radiocarbon ages for clean and highly corroded metal: a comparison. Artifacts are designated as follows: (1) Saugus Ironworks, (2) Hopewell Furnace, (3) Redding Furnace, (4) Scottish Iron, (5) Szechwan China, and (6) Hunan China.		Figure 1. (b) A summary of iron artifacts verified by radiocarbon dating: Weight percent carbon vs. age in radiocarbon years before present.

OVERVIEW OF PREVIOUSLY PUBLISHED RESULTS

To the best of the authors’ knowledge, Table I lists all of the radiocarbon data ever published for iron and steel. This table provides sample identification, radiocarbon years Before Present [B.P.] (Note: This nomenclature is used in radiocarbon dating to avoid the variation introduced by calibration charts that convert radiocarbon years B.P. to calendar years by accounting for variations in the ¹⁴C levels through history^20–22) sample size, sample condition, date of presumed manufacture, calibrated date, method, and reference. Since 1968, a wide range of iron-based materials have been investigated (Figures a, b, c, d, and e). The determined ages range from relatively recent materials (350 B.P.) to materials from an era approaching the start of the Iron Age (4000–5000 B.P.). Materials range from very low-carbon wrought irons to cast irons (> 2.1% carbon). Sample sizes range from less than 0.05 g to more than 500 g. Sample conditions range from clean metal to rusty metal to very corroded metal. In only 11% (7/63) of the cases, complications arose such that the authors could not explain their data simply. The nuances of radiocarbon dating of iron-based materials will be explained by way of example in this paper.

Table I. Previously Published Radiocarbon Dates for Iron-Based Artifacts
Artifact Identification	14C B.P.	%C	Sample Size (g)	Presumed Manufacture	Calibrated Date	Notes	Reference
Radiocarbon Dates that Matched the Date of Presumed Manufacture
Saugus Ironworks, MA	350 ± 60	2.9	81.3	A.D. 1648–1678	A.D. 1600 ± 60	a, d	3
Hopewell Furnace, PA	200 ± 60	2	112.8	A.D. 1771–1845	A.D. 1750 ± 60	a, d	3
Redding Furnace, PA	180 ± 60	2.3	102.1	A.D. 1761	A.D. 1770 ± 60	a, d	3
Roman fort, Scotland	1850 ± 80	0.2	579.3	A.D. 83–87	A.D. 100 ± 80	a, d	3
Han Dynasty, Sian, China	2060 ± 80	3.2	102.1	221 B.C.–A.D. 220	110 ± 80 B.C.	a, d	3
Han Dynasty, Szechwan, China	2130 ± 100	2.6	32.4	221 B.C.–A.D. 220	180 ± 100 B.C.	a, d	3
Warring States Period, Honan, China	2380 ± 80	3.1	42.3	480–221 B.C.	430 ± 80 B.C.	a, d	3
La Tène I, Yugoslavia	2130 ± 60	0.66	216	400–180 B.C.	180 ± 60 B.C.	a, d	3
Sri Lankan wootz steel	980 ± 40	2.6	0.274	5th–13th cent. A.D.	A.D. 1012–1038	a, e	5, 6
ROM Luristan steel dagger	2940 ± 60	1.5	0.485	1st millennium B.C.	1137–992 B.C.	a, e	6
MIT Luristan steel dagger	2880 ± 60	0.7	1.44	1st millennium B.C.	1012–1038 B.C.	a, e	6
Japanese sword	880 ± 150	0.49	2.27	A.D. 1192–1573	A.D. 1021–1263	a, e	15, 16
Planing adze, China	1720 ± 160	3.6	0.93	Late Han or Jin dynasty	A.D. 119–457	a, e	15, 16
Ungwana, crucible and bloomery steel (Africa)	870 ± 100	0.9	0.347	8th–16th cent. A.D.	A.D. 990–1300	a, d	8
Ungwana, crucible steel (Africa)	530 ± 90	1.4	0.529	8th–16th cent. A.D.	A.D. 1290–1520	a, e	8
Ungwana, bloomery steel (Africa)	1210 ± 140	0.4	0.623	8th–16th cent. A.D.	A.D. 595–1030	a, e	8
Ungwana, crucible steel (Africa)	1360 ± 650	0.3	0.368	8th–16th cent. A.D.	785 B.C.–A.D. 1685	a, e	8
Galu, white cast iron (Africa)	740 ± 70	2	0.284	8th–16th cent. A.D.	A.D. 1170–1400	a, e	8
Hook from Horyuji Temple, Japan	1330 ± 110	0.18	4.53	Late 7th–early 8th cent.	A.D. 604–814	a, e	16
Himeji castle nail, small	373 ± 31	0.28	0.495	A.D. 1600	1440–1530 (55.7%),	a, e	9
					1550–1640 (39.7%)
Damascus knife	240 ± 19	2.13	0.119	A.D. 1650	1640–1670 (71%),	a, e	9
					1780–1800 (24.4%)
Cauldron, Java Sea wreck	930 ± 50	10.97	0.039	A.D. 1215–1405	A.D. 1000–1220	b, e	14
Nail, earthquake fault in Turkey	1620 ± 50	0.55	0.061	A.D. 250–420	A.D. 260–560	b, e	14
Italian armor (N-7)	570 ± 50	0.66	0.507	Late 15th cent. A.D.	A.D. 1300–1440	c, e	14
Italian armor plate (N-9)	510 ± 40	0.35	0.489	A.D. 1480	A.D. 1330–1450	c, e	14
Tie pin, Ipswich, MA (N-12)	230 ± 40	0.52	0.534	Late 17th cent. A.D.	A.D. 1530–1947	a, e	14
Denbigh, VA (N-20)	350 ± 40	0.41	0.488	17th cent. A.D.	A.D. 1440–1650	a, e	14
Radiocarbon Dates that Did Not Match the Date of Presumed Manufacture, but Were Easily Explainable by the Authors at the Time of Publication
Modern, coke-smelted cast iron	> 35000	3.1	74	Modern	NA	a, d, f	3
Fort Kiowa (?), SD	25000	3	97.5	A.D. 1870–1900	NA	a, d, f	3
Fort Berthold, ND	6700	2.7	83.3	A.D. 1845–1885	NA	a, d, f	3
Hunan Province, China	400 ± 60	1.9	97.8	4th–10th cent. A.D.	A.D. 1550 ± 60	a, d, g	3
Gate from Myohouji Temple, Tokyo	38350 ± 2,300	3.23	3	A.D. 1866–1911	NA	a, e, f	17
Anchor dedicated to Isonomae shrine	29520 ± 1,300	Not avail.	Fragment	Not avail.	NA	a, e, f	17
Pail in Inari shrine	950 ± 100	Not avail.	Fragment	A.D. 1866	NA	a, e, h	17
Modern steel, 1.3%C	39140 ± 970	1.3	0.077	Modern	NA	a, e, f	9
Modern steel, 1.9%C	38330 ± 2870	1.9	0.053	Modern	NA	a, e, f	9
Italian armor (N-5)	1640 ± 50	0.2	0.844	A.D. 1400	NA	c, e, f	14
Italian sword (N-8)	4250 ± 50	0.11	0.841	16th cent. A.D.	NA	c, e, f	14
German armor (N-6)	2790 ± 50	0.04	0.618	A.D. 1550	NA	c, e, f	14
German armor (N-11)	2580 ± 40	0.9	0.58	Mid-16th cent. A.D.	NA	c, e, f	14
Axle Thimble, Fort Lower Brule, SD (N-15)	13420 ± 110	3.55	0.349	A.D. 1777–1778	NA	c, e, f	14
Fort Atkinson, WI (N-18)	460 ± 40	0.04	2.8	A.D. 1820–1827	NA	c, e, f	14
Williamsburg, VA (N-21)	730 ± 40	0.18	0.452	A.D. 1816–1817	NA	c, e, f	14
Eylon’s own sample A	473 ± 45	Not avail.	Not avail.	A.D.180 or A.D.1587	A.D. 1399–1474	a, e, i	18
Eylon’s own sample B	1210 ± 140	Not avail.	Not avail.	A.D.180 or A.D.1587	A.D. 568–1151	a, e, i	18
Radiocarbon Dates that Did Not Match the Date of Presumed Manufacture, but Were Not Easily Explainable by the Authors at the Time of Publication
Frobisher bloom #1	679 ± 133	0.051–0.127	30	A.D. 1576–1578	A.D. 1271 ± 133	a, d	19
Frobisher bloom #2	792 ± 107	0.048–0.061	30	A.D. 1576–1578	A.D. 1158 ± 107	a, d	19
Frobisher bloom #3, near surface	1340 ± 70	0.2	1.34	A.D. 1570s	A.D. 640–760	a, e	5, 6
Frobisher bloom #3, 2 cm in	550 ± 60	0.2	3.21	A.D. 1570s	A.D. 1307–1355	a, e	6
Frobisher bloom #3, 5 cm in	500 ± 60	0.2	3.4	A.D. 1570s	A.D. 1400–1442	a, e	6
Galu, bloomery steel (Africa)	1400 ± 240	0.3	0.556	8th-16th cent. A.D.	A.D. 125–1050	a, e	8
Galu, crucible steel (Africa)	1300 ± 70	1.7	0.355	8th-16th cent. A.D.	A.D. 630–890	a, e	8
Radiocarbon Dates for Iron-Based Materials That Were Dated More Than Once
Saugus Ironworks, MA	469 ± 144	3.73	10	A.D. 1648–1678	A.D. 1481 ± 144	a, d	19
Redding Furnace, PA	285 ± 145	3.98	10	A.D. 1761	A.D. 1665 ± 145	a, d	19
Modern, coke-smelted cast iron	39800 ± 3000	3.28	0.767	Modern	NA	c, e	14
Cast iron, Fort Kiowa, SD	26390 ± 550	3.56	0.107	A.D. 1870–1900	NA	c, e	14
Cast iron, Fort Berthold, ND	6610 ± 50	3.01	0.166	A.D. 1845–1885	NA	c, e	14
Cast iron, Sian, China	Did not graphitize	0.03	0.045	221 B.C.–A.D. 220	NA	c, e	14
Cast iron, Saugus MA	420 ± 40	3.63	0.119	A.D. 1648–1678	A.D. 1420–1630	c, e	14
Cast iron, Hopewell, PA	160 ± 40	4.08	0.204	A.D. 1771–1845	A.D. 1650–1950	c, e	14
Cast iron, Redding Furnace, PA	160 ± 40	3.83	0.233	A.D. 1761	A.D. 1650–1950	c, e	14
Bloomery iron, Scotland	1930 ± 50	0.13	0.932	A.D. 83–87	B.C. 40–A.D. 220	c, e	14
Cast iron, Hunan, China	340 ± 30	2.93	0.621	4th–10th cent. A.D.	A.D. 1450–1650	c, e	14
Cast iron, Szechwan China	1770 ± 610	0.01	0.337	250 B.C–250 A.D.	B.C. 1250–A.D. 1410	c, e	14
a—clean metal sample condition; b—extremely corroded sample condition; c— rusty metal sample condition; d—pre-AMS method; e—AMS method; f—coal; g—replica; h—reworking with coal; i—after A.D. 180

OVERVIEW OF NEW RESULTS

Table II lists all of the unpublished radiocarbon data for iron known by the authors. Most of the data is the authors’, but nine of the data points come from their predecessor, Richard Cresswell.²³ Together, 29 new data points have been generated for iron-based materials that fall into two distinct categories: radiocarbon dates that matched the dates of presumed manufacture, and radiocarbon dates that did not match the dates of presumed manufacture. In only one case (the nose ring from Burkina Faso, Africa), the authors were not able to provide a simple explanation for the radiocarbon date obtained. All pertinent information is provided in Table II. Figure 1b summarizes all of the iron artifacts which have ever been ratified by radiocarbon dating.

Brief sample descriptions and commentary are provided in the following.

Gibson axe: What is believed to be a pick-axe point was found during the 12th season in Nippur, Iraq on the floor of a temple in Area WA (sample ID 12 N 380). The floor dates to the 19th century B.C., but it is possible that the axe was intrusive from a 13th century B.C. level above. The axe was completely corroded. Direct radiocarbon dating on the rust showed the date of the axe to be in accordance with the date of presumed manufacture: 1900 B.C. To date, this is the oldest piece of iron-based material to be radiocarbon dated. It suggests that corrosion products contain original carbon that can be extracted and reliably radiocarbon dated.

Wrought-iron cleaver, Roman: This wrought-iron cleaver blade with handle (Figure c) was purchased from Edgar L. Owen at a special auction of Roman-period iron tools (www.edgarlowen.com). The piece was reported to be from the Roman or late Roman period. The radiocarbon date obtained for the cleaver was in accordance with the date of presumed manufacture.

Wrought-iron nails, Roman: A large number of wrought-iron nails were recovered in the 1960s from the legionary Roman fortress at Inchtuthil (which stood only from A.D. 83–87). The nails are reported to have been made in the nearby forest of Dean at Beauport Park, East Sussex, Britain in one of the largest iron refineries in the history of the Roman Empire. The nails were purchased through a catalog by Harlan J. Berk, Ltd. (www.harlanjberk.com). The radiocarbon date obtained from the Inchtuthil nail was in accordance with the date of presumed manufacture.

Spear blade, Israel: What is believed to be a spear blade reportedly from Israel, is dated at about 1000 B.C. It was purchased from Alex G. Malloy, and had a radiocarbon date in accordance with the date of presumed manufacture.

Roman-period arrowhead: A Roman-period arrowhead from the late Roman or Crusader period was also purchased from Alex G. Malloy. The radiocarbon date obtained for the arrowhead was in accordance with the date of presumed manufacture.

Burkina Faso, African artifacts (small spear, large spear, and nose ring): These iron-based artifacts were recovered during an archeological investigation of various mound complexes in Burkina Faso, Africa. Each iron artifact was associated with a specific mound layer that was radiocarbon dated independently using charcoal remains. The radiocarbon dates obtained for the two spears (the larger one is shown in Figure d) were in accordance with the date of the layer in which they were found; however, the date for the nose ring was not. The authors can offer no simple explanation for the date of the nose ring. Perhaps it was reworked using coal or was somehow contaminated with a petroleum-based oil (or other product).

Japanese folded steel: This steel was made in A.D. 1995 for use in the traditional Japanese sword-making industry. It was smelted in small batches using only modern charcoal. The radiocarbon date obtained for the steel was in accordance with the date of presumed manufacture. This piece of steel is the youngest iron-based material to ever be radiocarbon dated.

Japanese tanto tang: Some years ago, an old knife was given to a Japanese swordmaker (Yoshino Yoshihara) to be reforged and used to refurbish and repair other blades from the same period. The tang (back end of the knife) is inscribed with the date A.D. 1539. The swordmaker dismantled the tanto, keeping the blade for repair work and giving the tang as a gift. The radiocarbon date obtained for the tanto tang was in accordance with the date of presumed manufacture.

Himeji castle artifacts (pinch dog, large nail, small bracket, medium nail, and reforged nail): There have been fortifications in Himeji since A.D. 1333. Today’s castle was built in A.D. 1580 by Toyotomi Hideyoshi and enlarged some years later (A.D. 1600–1609) by Ikeda Terumasa for the Tokugawa Shogunate. The five-story castle has been home to 48 successive lords. While undergoing restoration about 35 years ago, some iron-based materials were removed from the castle in Japan.

The authors were able to obtain four pieces from the castle, plus a nail and a paperweight (see Figure e) that had been reforged (mildly, using low heat) from castle materials.

In radiocarbon dating these items, three pieces were found to be in accordance with the original building of the castle (pinch dog, large nail, small bracket) and the other (medium nail) to be in accordance with a later remodel. The reforged nail appeared to be almost 2,000 years old, and, therefore, must have been reheated using either coal or a mixture of charcoal and coal. Since coal was used, the date of original manufacture for this nail cannot be determined using radiocarbon dating.

Nikko Shrine, large bracket: The Shrine in Nikko was constructed as a memorial to the warlord, Tokugawa Ieyasu, whose shogunate ruled Japan for 250 years. Tokugawa Ieyasu was laid to rest among Nikko’s towering cedars in 1617 A.D., but it was his grandson, Tokugawa Iemitsu, who commenced work in 1634 A.D. on the shrine that can be seen today. The original shrine was completely rebuilt in 1818 A.D. The authors obtained what appears to be an iron brace, possibly from one of the large doorways in the shrine. It was assumed to originate with the 1634–1636 A.D. construction. The radiocarbon date, however, accords better with the 1818 A.D. reconstruction.

Basque nail, Labrador coast, Canada: A large nail was recovered from Red Bay on the Labrador coast, Canada, associated with a Basque whaling station based there in the mid-1560s. A large number of iron artifacts have been recovered,²⁴ some imported, others made on site. The age of this nail accords well with the postulated origin and was probably brought to the site from Europe.

Fishbourne nail, Sussex, United Kingdom: Shortly after the Roman invasion of Britain in the first century, a large villa was erected at Fishbourne (just west of Chichester, Sussex, United Kingdom). This burned down in the late 3rd century and was disused until Saxon times when it became a cemetery. In medieval times it was razed, put under crop, and so remained until the 19th century.²⁵ The nail was recovered from a large spoil heap presumed to be of Roman origin. The date, however, accords better with Saxon origin.

Modern bloom: A modern bloom was smelted in A.D. 1986 at the University of Toronto using charcoal derived from young saplings. The measured activity agrees within precision with the expected atmospheric concentration as measured at northern hemisphere atmosphere stations.²⁶ The bloom is therefore in accordance with the date of presumed manufacture.

Roman iron, Colona Antonina, Italy: A wrought-iron railing bar was recovered by La Sapienza University during the A.D. 1985 restoration of the Colona Antonina (Marcus Aurelius Column) in Rome. It was replaced with a titanium railing. A segment of the iron railing was given to the University of Dayton to determine through carbon dating and metallography if it was installed during the original A.D. 180 construction, or when it was restored and “Christianized” in A.D. 1589.

D. Eylon¹⁸ was the first to radiocarbon date the railing and concluded that it must have been installed in A.D. 1589 because the radiocarbon ages obtained (see Table I), while not concurrent with A.D. 1589, were not old enough to be original. Eylon, seeking clarification, sent the authors a segment of the column to radiocarbon date. The authors sectioned the bar into three pieces and radiocarbon dated each piece separately. Three different radiocarbon dates were obtained. In this case, it is clear that radiocarbon cannot be used to obtain the date of original manufacture.

This sample exemplifies another category of iron-based materials for which radiocarbon dating will be unable to determine the date of original manufacture. The railing was clearly shown by metallography¹⁸ to be inhomogeneous and made of a composite of many different strips of iron welded together. The individual strips could have been manufactured at any time using any method.

World War II steel, Marin, California: Steel believed to be from World War II was collected by one of the authors (A. Cook) from an abandoned military battery in Marin, California. While the exact manufacturing process is unknown, it was thought to have been mass produced with coal near the start of World War II. The radiocarbon date obtained for the iron-based material was in accordance with at least partial manufacture by coal.

Indian trade axes, Ontario, Canada: Cahiague Ball: Axes, thought to have been associated with European-Native American trading in the 19th century, were recovered from Cahiague sites 26712a, 26712b, 26697, and 26698 and Ball 2046 inner and outer sites near Ontario, Canada. A complete axe head was recovered from the Ball site, while only pieces of axes were recovered at the Cahiague site. Radiocarbon dating showed the pieces of axes to have inconsistent dates that were only explained following examination of the complete axe head.

Metallurgical examination revealed the axe to be a composite piece of welded strips of iron. Pieces of the inner and outer strips were separated and analyzed, giving quite distinct radiocarbon signatures. These axes were clearly the result of significant re-working of iron from different origins, many being much younger than hoped, and including strips of iron from coal-fired furnaces. As with the Colona Antonina railing, radiocarbon dating cannot determine the manufacture date for these objects.

Table II. Previously Unpublished Radiocarbon Dates for Iron-Based Materials
Artifact Identification	14C B.P.	%C	Sample Size (g)	Presumed Manufacture	Calibrated Date	Notes
Radiocarbon Dates that Matched the Date of Presumed Manufacture
Gibson axe	3740 ± 60	0.41	0.078	1900 B.C.	B.C. 2330-1960	b, d, f
Wrought-iron cleaver, Roman	1880 ± 40	0.38	0.539	27 B.C.–A.D. 395	A.D. 30–240	a, d, g
Wrought-iron nails, Roman	2090 ± 50	0.52	0.675	A.D. 85	B.C. 350–A.D. 30	a, d, g
Spear blade, Israel	2270 ± 50	0.4	0.504	1000 B.C.	B.C. 410–200	a, d, g
Roman period arrowhead	1130 ± 50	0.11	0.971	A.D. 1096–1272	A.D. 770–1020	a, d, g
Burkina Faso, Africa, small spear	740 ± 40	0.59	0.471	A.D. 1042–1379	A.D. 1220–1380	a, d, h
Burkina Faso, Africa, large spear	570 ± 30	0.3	0.545	A.D. 1163–1393	A.D. 1300–1430	a, d, h
Japanese folded steel	>Modern	0.55	0.125	A.D. 1995	After A.D. 1950	a, d, i
Japanese tanto tang	490 ± 40	0.44	0.347	A.D. 1539	A.D. 1330–1480	a, d, i
Himeji Castle, pinch dog	390 ± 40	0.35	0.185	A.D. 1580–1610	A.D. 1430–1640	a, d, i
Himeji Castle, large nail	350 ± 40	0.01	2.4	A.D. 1580–1610	A.D. 1450–1640	a, d, i
Himeji Castle, medium nail	180 ± 40	0.26	1.081	A.D. 1580–1610	A.D. 1640–1950	a, d, i
Himeji Castle, small bracket	290 ± 50	0.12	1.4	A.D. 1580–1610	A.D. 1470–1800	c, d, i
Nikko Shrine, large bracket	210 ± 50	0.13	0.813	A.D. 1634–1636, A.D. 1818	A.D. 1530–1950	c, d, i
Basque nail, Labrador coast, Canada	530 ± 70	0.1	11.28	Mid-1560s A.D.	A.D. 1320–1440	a, e, j
Fishbourne nail, Sussex, U.K.	1070 ± 50	0.35	2.46	A.D. 1–present	A.D. 890–1010	a, e, j
Modern bloom	>Modern	0.14	2.72	Modern, A.D. 1986	After A.D. 1950	a, e, k
Radiocarbon Dates that Did Not Match the Date of Presumed Manufacture and are Discussed by the Present Authors
Himeji Castle, reforged nail	1890 ± 40	0.22	0.742	A.D. 1580–1610	NA	a, d, i, n
Burkina Faso Africa, nose ring	1830 ± 50	0.13	0.411	A.D. 1292–1453	NA	a, d, h, o
WWII steel, Fort SF	21890 ± 60	0.1	0.2	Early 1940s AD	NA	a, d, l, p
Roman iron, Colona Antonina, Italy	2200 ± 40	0.05	1.778	A.D. 180 or A.D. 1587	NA	a, d, m, q
Roman iron, Colona Ant., Italy C	1470 ± 130	0.03	0.426	A.D. 180 or A.D. 1587	NA	a, d, m. q
Roman iron, Colona Ant., Italy D	1980 ± 110	0.05	0.438	A.D. 180 or A.D. 1587	NA	a, d, m, q
Cahiague 26712a, axe head, Ontario, Canada	1567 ± 137	0.04	4.1	Pre-19th cent. A.D.	A.D. 339–632	a, e, j, r
Cahiague 26712b, axe head	1830 ± 70	0.04	6.28	Pre-19th cent. A.D.	A.D. 86–250	a, e, j, r
Cahiague 26697, axe head	1506 ± 410	0.06	5.25	Pre-19th cent. A.D.	A.D. 73–956	a, e, j, r
Cahiague 26698, axe head	530 ± 80	0.14	4.79	Pre-19th cent. A.D.	A.D. 1385–1439	a, e, j, r
Ball 2046: inner, axe head	>Modern	0.23	3.17	Pre-19th cent. A.D.	After A.D. 1950	a, e, j, r
Ball 2046: outer, axe head, Ontario, Canada	3900 ± 180	0.11	1.59	Pre-19th cent. A.D.	2613–2137 B.C.	a, e, j, r
a—clean metal sample condition; b—extremely corroded sample condition; c— rusty metal sample condition; d—14C dated by Cook et al.; e—14C dated by Cresswell; f—donated by Gibson, University of Chicago; g—donated by Pense, Lehigh University; h—donated by Holl, University of Michigan; i —donated by Kapp, University of San Francisco; Yoshihara, Japan; j —donor not stated in unpublished manuscript; k—donated by the University of Toronto; l—donated by Cook, High Tech High; m—donated by Eylon, University of Dayton; n—reforged using coal; o—no explanation; p—made with coal; q—inhomogeneous; r—reworked

CONCLUSIONS

To date (including this paper), a total of 92 different radiocarbon measurements have been published on iron-based materials. The determined ages range from very recent materials (1995 A.D.) to materials approaching the commonly accepted era for the start of the Iron Age (4000–5000 B.P.). Materials range from very low-carbon wrought irons (0.01% carbon) to cast irons (> 2.1% carbon). Sample sizes range from less than 0.05 g to more than 500 g. Sample conditions range from clean metal to very corroded metal and rust.

In principle, then, there is not a period in Iron-Age history that cannot be investigated using radiocarbon dating. As long as assumptions hold (the iron-based material is manufactured using only contemporaneous charcoal—no old wood, no reworking, no coal, no limestone flux), the radiocarbon dating of iron-based materials has been shown to be very reliable.

Not surprisingly, however, there are iron-based materials that are not suited for accurate dating by radiocarbon. These include materials manufactured using coal (e.g., most modern steel made after 1800 A.D.), materials reheated using coal (e.g., reforged Himeji castle nails and possibly the European armor and Frobisher bloom), and materials that are composites (Indian trade axes, the iron railing from the Colona Antonina). Although radiocarbon dating cannot be used to determine the age of these materials, it may, however, yield valuable insight into the manufacturing processes that were used to create these materials.

The first warning that an artifact is unsuitable for dating by radiocarbon is when multiple samples are run and the dates obtained are widely variable. Reworked samples, especially those reworked with coal, are often inhomogeneous with respect to the age of the carbon in the metal due to variations in absorption. Clearly, under these conditions, the iron-based material is not suitable for dating by radiocarbon. Fortunately, such cases are usually quite obvious from preliminary results.

In only a very small fraction of the cases in which iron-based materials have been dated can the data not be readily interpreted. In most cases, when the radiocarbon date did not match the date expected, it was possible to deduce something about the manufacturing process. The Colona Antonina and the Indian trade axes were most likely reworked composites of iron scraps with different coal/charcoal origins. The Frobisher blooms seem to have undergone multiple reworking attempts over time. The European armor investigated may have been replicas, or perhaps coal was used in the process of making European armor earlier than was previously thought. On the other hand, the authors do not understand why the nose ring from Burkina Faso appears to be so old. There is no simple explanation. Either the sample is contaminated or perhaps old carbon was somehow used in the manufacturing process. Either way, radiocarbon is not likely to help in determining the exact date of manufacture for these objects. As discussed, nuances and complications still exist in interpreting the radiocarbon dates from iron-based materials; however, many aspects are becoming better understood as more samples are examined.

It is important to note that with the large reduction in required sample size and the recent discovery that rust (at least in some cases) can be reliably dated, the door has opened for investigations on the very earliest iron, dating from 1000 B.C. and earlier. It is of no surprise that this period contains the fewest data, and the present results suggest there is no a priori reason to believe that items from this time period cannot be dated reliably with radiocarbon.

ACKNOWLEDGEMENTS

The authors thank the Lawrence Livermore National Laboratory support staff; Brian Frantz and Paula Zermeno for making graphite and sealing 9 mm quartz tubes; Rob Robinson for photographing the artifacts; John Knezovich for providing the images of the Center for Accelerator Mass Spectrometry; and Bob Vallier for diamond saw cutting and metallography. We are also grateful to Kirk Bertsche of Fermilab and David Loyd of Angelo State University for assistance in developing the sealed-tube combustion method. Mike Notis at Lehigh University enlightened us regarding work both past and current on the detection of carbides in rust. Richard Cresswell donated his previously unpublished results on iron. We also thank all those who have donated iron samples —McGuire Gibson, University of Chicago; Alan Pense, Lehigh University; Augustin Holl, University of Michigan; Leon Kapp, University of San Francisco; Yoshindo Yoshihara, swordmaker in Japan; and Daniel Eylon, University of Dayton. This work was performed under the auspices of the U.S. Department of Energy by the University of California Lawrence Livermore National Laboratory under Contract No. W-7405-Eng-48.

References

1. C. Zimmer, National Geographic, 200 (3) (2001), pp. 78–101.
2. N.J. van der Merwe and M. Stuiver, Current Anthropology, 9 (1) (1968), pp. 48–53.
3. N.J. van der Merwe, The Carbon-14 Dating of Iron (Chicago, IL: University of Chicago Press, 1969).
4. R.G. Cresswell (M.Sc. thesis, University of Toronto, 1987).
5. R.G. Cresswell, Historical Metallurgy, 25 (1991), pp. 76–85.
6. R.G. Cresswell, Radiocarbon, ed. A. Long and R.S. Kra, 34 (3) (1991), pp. 898–905.
7. R.G. Cresswell, Extended abstract (paper #41) (Paper presented at the Sixth Australasian Archaeometry Conference, Sidney, Australia, 10–13 February 1997).
8. C.M. Kusimba, D.J. Killick, and R.G. Cresswell, Society, Culture, and Technology in Africa, 11 (supp.) (1994), pp. 63–77.
9. A.C. Cook, J. Wadsworth, and J.R. Southon, Radiocarbon, 43 (2A) (2001), pp. 221–227.
10. El Gayar, El Sayed, and M.P. Jones, Historical Metallurgy, 23 (2) (1989), pp. 75–83.
11. J. Wadsworth and D.L. Lesuer, Journal of Materials Characterization, 45 (2001), pp. 289–313.
12. R. Knox, Archaeometry, 6 (1963), pp. 43–45.
13. M. Notis, personal communication (9 May 2001).
14. A.C. Cook et al., Journal of Archaeological Science, 30 (2003), pp. 95–101.
15. K. Igaki et al., Proceedings of the Japan Academy, 70 (B) (1994), pp. 4–9.
16. T. Nakamura, M. Hirasawa, and K. Igaki, Radiocarbon, 37 (2) (1995), pp. 629–636.
17. K. Yoshida, Bulletin of the National Museum of Japanese History (in Japanese), 38 (1992), pp. 171–198.
18. D. Eylon, BUMA V Messages from the History of Metals to the Future Metal Age, ed. Gyu-Ho Kim, Kyung-Woo Yi, and Hyung-Tai Kang (Seoul, Korea: Korean Institute of Metals and Materials, 2002), pp. 161–169.
19. E.V. Sayre et al., American Chemical Society Symposium Series, 176 (1982), pp. 441–451.
20. M. Stuiver and H.A. Polach, Radiocarbon, 19 (20) (1977), pp. 355–363.
21. M. Stuiver and P.J. Reimer, Radiocarbon, 35 (21) (1993), pp. 215–230.
22. M. Stuiver et al., Radiocarbon, 40 (3) (1998), pp. 1041–1083.
23. R.G. Cresswell, personal communication (1 May 2002).
24. J.A. Tuck and R. Grenier, Scientific American, 245 (1981), pp. 180–188.
25. B.W. Cunliffe, Fishbourne: A Roman Palace and Gardens (London: Thames & Hudson, 1971).
26. I. Levin et al., Radiocarbon after Four Decades: An Interdisciplinary Perspective, ed. R.E. Taylor, A. Long, and R.S. Kra (New York: Springer-Verlag, 1992), pp. 503–518.

For more information, contact A.C. Cook, High Tech High, 2861 Womble Road, San Diego, California; (619) 243-5033; fax (619) 243-5050; e-mail acook@hightechhigh.org.