Handheld XRF to Study Ancient Roman Drainage System (1BCE to 1CE) Lead Pipe & Connective Keys near Orvieto, Italy

INNOV-X Academic Grant Series
Saint Anselm College, Chemistry & Classics Depts Manchester, NH USA

Focus Conservation Non-Destructive Testing In-situ Metal Measurements PMI of Sn, Fe, & Pb Ancient Alloy Recycling Introduction Handheld XRF is a valuable tool for in-situ metals measurements at archaeological excavations. The non-destructive analyses are performed in minutes and can provide valuable information to aid in the assessment of a site - where to dig next, the relationships among loci, the homogeneity of materials, etc. A handheld XRF was recently used at an archaeological site near Orvieto, Italy to better understand the metal content and elemental distributions in lead pipes and keys.

Sections of lead pipe and connective keys in close proximity to a water supply and drainage system dated between 1 BCE to 1 CE were discovered at the site. The quantity, condition, and context of the lead pipes and keys are rare archaeological finds but would be difficult to sample and near politically and bureaucratically impossible to transport away from the site. In-situ analyses on the lead pipes were therefore required and handheld XRF was an ideal method for the work.

"…This supports previous theory that Roman pipes were often the product of recycled lead from other sources. The Sn remained relatively constant for the length of the pipe while Fe and Pb varied inversely to one another over the length of the pipe. …"

Experimental

XRF Specifications
An Innov-X Systems Alpha Series portable XRF was used for all data collection. The analyzer has a Ag anode x-ray tube excitation source and Si Pin diode detector. The instrument was operated in "Analytical" mode, which uses a 40 kV excitation. Power is provided by rechargeable Li-ion batteries. Control of the instrument and data storage is through a Hewlett Packard iPAQ personal digital assistant. Fluorescence signal was collected for thirty seconds for each analysis.

Data Collection
All sampling locations were cleaned with a trowel and/or brush to clear surface contamination prior to data collection. Five replicates were collected at 10 cm intervals along the length of each pipe section. Twenty replicates were collected at random locations on each key.

Conclusions

Portable XRF and statistical analysis were demonstrated as valuable site tools for samples that can not be removed for chemical analyses in a laboratory. Evaluation of material content and homogeneity provided the researchers with evidence toward production of the pipes.

Data and Results

Key Comparisons
Five water pipe keys were found at the excavation site, four of which were connected as one pipe system (labeled as Keys 1-4) with the fifth (labeled as Key 5) separate and further from the drainage system. Analyses were made at the same time from similar raw material sources. Five analyses were taken on keys 1, 2, 3, and 4 on June 2nd and twenty analyses were taken on keys 1, 3, 4, and 5 on June 20th. Fe, Sn, and Pb were detected in the samples. An analysis of variance on the three elements across two days was performed with no statistical difference found between days. The data were therefore pooled as one set.

Sn was found to be consistent for all of the keys while levels of Fe and Pb showed differences both in content and homogeneity. A plot of % Fe content versus % Pb content in each key is shown as Figure 1. The error bars represent the 95% confidence intervals for each sample set. As can be seen in the figure, there is significant overlap of the Fe and Pb contents in Keys 1, 2, and 4. This leads one to suppose that these three keys were made at the same time from the similar raw material sources. Homogeneity indicated as the error bar sizes are similar for Keys 2 and 4, providing further support for their coincidental production. Key 3 shows no overlap with 1, 2, and 4 and was most likely made separately. Key 5 shows some overlap with the other four keys but not enough to make conclusions as to its origins.

Metal Content in Pipe Segments
A plot of % Fe content versus % Pb content in each segment on each analysis day was created. This graph is shown as Figure 2. The error bars represent the 95% confidence intervals for each sample. The data are clearly grouped by analysis location. There is no overlap between the in situ analyses and the analyses performed after the pipes were removed from the site. This shows the elemental data for the segments varied depending on the location of the analysis with iron decreasing after cleaning. This leads one to surmise that soil contamination may be the iron source. The mean, standard deviation, and 95% confidence interval for Pb, Sn, and Fe in each pipe segment was calculated and are provided in Table 1.

Pipe Homogeneity Across Length
Four water pipes were found at the excavation site. Three of the segments (A-C) were connected as one pipe system with a fourth pipe segment D in a separate area. Analyses were performed along the length of the pipes to characterize content and homogeneity. Pipes were first analyzed in situ and then again after their removal and cleaning. Twenty analyses were taken on each pipe segment on each day. Fe, Sn, and Pb were the significantly detected elements. Figure 3 is a plot of Fe, Sn, and Pb over the length of pipe segment normalized by subtracting the first measurement from all the measurements. This graph is representative of the trends found for all the pipe segments in that the pipes are not homogeneous over their length. This supports previous theory that Roman pipes were often the product of recycled lead from other sources. The Sn remained relatively constant for the length of the pipe while Fe and Pb varied inversely to one another over the length of the pipe.

Innov-X Academic Grant Awarded to Principal Investigator, Dr. Mary Kate Donais. She and her colleague, Dr. David George, are professors at St. Anselm College in Manchester, NH, USA.